Framework for structuring information for

Framework for structuring information for environmental management of industrial systems

RAUL CARLSON

Department of Computer Science and Engineering Research group Industrial Environmental Informatics - IMI CHALMERS UNIVERSITY OF TECHNOLOGY AND GOTEBORG UNIVERSITY Göteborg, Sweden, 2006

Framework for structuring information for environmental management of industrial systems RAUL CARLSON ISBN 91-7291-765-2 © RAUL CARLSON, 2006-03-28 Doktorsavhandlingar vid Chalmers tekniska högskola Ny serie nr 2447 ISSN 0346-718X Department Of Computer Science And Engineering Industrial Environmental Informatics - IMI Chalmers University of Technology SE-412 96 Göteborg Sweden http://www.imi.chalmers.se Telephone +46-(0)31-772 10 00 Göteborg, Sweden, 2006

Why did brains evolve to start with? The answer lies in the value of information, which brains have been designed to process. Pinker S., How the mind works

For by the fruit the tree is known. The New Testament, Matthew 12:33

Framework for structuring information for environmental management of industrial systems Raul Carlson Department of Computer Science and Engineering Research group: Industrial Environmental Informatics Chalmers University of Technology

This thesis addresses information structuring for reducing cost, facilitate understandability, increase relevance and improve quality of environmental information. The research is based in experience, practical results and syntheses from three different but related research and development projects conducted between 1994 and 2004. The projects have involved different industrial and academic partners in Sweden and other EU countries, and have also had active interactions with the international standardisation effort within ISO (International Organization for Standardization). To actively take environmental responsibility for designing, producing or purchasing a product it is necessary to have information about consequences from the decisions. Environmental impacts from industrial products, activities and systems may be gradual, diffuse, long term and complex. They originate not only from extraction of natural resources, emissions and waste generation, but also from transportation, storage, use and end of life phases. The complexity makes it difficult to foresee the total environmental impact from industrial systems. There are methods and tools available to environmentally assess environmental performance of the life cycle of products and services, but it is still difficult to acquire the information needed to perform the assessment. It is also difficult to present the results from the assessments in a clear and logic way. Hence, it is difficult for consumers or professional decision makers to take into regard environmental consequences from their decisions. Due to the increasing global population and accelerating economies of the developing worlds, it is probable that the needs for more effective and efficient information handling for different global responsibility issues will increase. The material presented concerns solving problems of environmental management of different industrial systems, using techniques for information structuring. The techniques include a combination of linguistic analysis, relational database modelling, system architecture design, and general description of data aggregation and data quality. Significant and immediately useful results from the research work are 1) the outline of a methodological framework for building environmental information structures, 2) the importance of environmental indicators to build functioning environmental information systems and 3) the different practically useful and partially integrated prototype information systems for LCA and DfE resulting from the three research projects. The doctoral studies are financed by the Swedish competence centre CPM (Center for environmental assessment of Product and Material systems), which is a joint research forum including Swedish industry and Chalmers university of technology, supported by the Swedish government through VINNOVA (Swedish Governmental Agency for Innovation Systems). Keywords: industrial environmental information systems, information structuring, sustainable development, database, data format, Life cycle assessment (LCA), Design for Environment (DfE), characterisation, weighting, indicator i

ii

Publications The thesis is based on the following published articles: I. Carlson R., Löfgren G., Steen B., Tillman A-M., LCI Data Modelling and a Database Design; Published in The International Journal of Life Cycle Assessment, Vol. 3, No.2, pp. 106-113, 1998 II. Bengtsson M., Carlson R., Molander S, Steen B., An Approach for Handling Geographical Information in Life Cycle Assessment Using a Relational Database; Published in Journal of Hazardous Materials, vol. 61, pp. 67-75, 1998 III. Carlson R., Pålsson A-C., Industrial environmental information management for technical systems, Journal of Cleaner Production, 9, pp 429-435, 2001 IV. Carlson R., Forsberg P., Dewulf W., Ander Å., Spykman G., A Full Design for Environment (DfE) Data Model, PDT Europe 2001, April 24th-26th, 2001, pp. 129-135, 2001 V. Tivander J, Carlson R, Erixon M, Pålsson A-C., OMNIITOX Concept Model Supports Characterisation Modelling for Life Cycle Impact Assessment International Journal of Life Cycle Assessment, Vol. 9, No. 5, pp. 289-294, 2004 VI. Carlson R., Erixon M., Forsberg P., Pålsson A-C., System for Integrated Business Environmental Information Management; Advances in Environmental Research, 5/4, pp. 369-375, 2001 VII. Carlson R., Learning from management of LCA data, Journal of Life Cycle Assessment, Japan, Vol. 1, No. 2, pp 102-111, 2005

iii

Examples of other material that the author has published in this area: I. Pålsson A-C., Carlson R., Maintaining Data Quality within Industrial Environmental Information Systems, 12th International Symposium Computer Science for Environmental Protection, Bremen; Band 1/Volume 1 p. 252-265, 1998 II. Carlson R., Environmental Performance Indicators; INSIGHT, Vol 5 Issue 2, pp. 22-23, The International Council on Systems Engineering (INCOSE), 2002 III. Häggström S., Pålsson A-C., Carlson R., Policy controlled environmental management work, LCM 2005, 2nd International Conference on Life Cycle Management, Barcelona, September 5-7, 2005 IV. Flemström M., Erlandsson M., Carlson R., Standards and tools for environmental design and supply chain management in railway industry, The Sixth International Conference on EcoBalance, Oct. 25th - 27th, Tsukuba, Japan, 2004 V. Flemström K., Erixon M., Erlandsson M., Häggström S., Tivander J., Carlson R., Wiklund C., Riise E., Ågren U., Implementation of integrated environmental information systems for sustainable development, LCM 2005, 2nd International Conference on Life Cycle Management, Barcelona, September 5-7, 2005 VI. Carlson R., Erixon M., Erlandsson M., Flemström K., Häggström S., Tivander J., Establishing common product life cycle data, LCM 2005, 2nd International Conference on Life Cycle Management, Barcelona, September 5-7, 2005

iv

Table of contents 1

2 3

4

5

6

Introduction............................................................................................................1 1.1 The research project.......................................................................................1 1.2 Information and environmental management ................................................1 1.3 Sustainable development ...............................................................................2 1.4 The task of information structuring ...............................................................3 State of the art ........................................................................................................5 Concepts and terms ................................................................................................9 3.1 General concepts, terms and distinctions.......................................................9 3.2 Concepts of industrial systems.....................................................................10 3.2.1 Scale and productivity..........................................................................10 3.2.2 Organisational intelligence and information........................................10 3.2.3 The life cycle of industrial information systems..................................10 3.3 Concepts and terms of environmental management ....................................11 3.3.1 Sustainable development .....................................................................11 3.3.2 Environmental management ................................................................12 3.3.3 Life cycle assessment...........................................................................13 3.3.4 Design for environment .......................................................................15 3.4 Concepts of informatics ...............................................................................15 3.4.1 Basic concepts......................................................................................15 3.4.2 Information structuring ........................................................................17 3.4.3 The OSI reference model, vertical information structuring.................17 3.4.4 Information structuring as structuring of ideas ....................................19 3.5 Concepts of environmental informatics .......................................................20 3.5.1 Environmental information..................................................................20 3.5.2 Example of environmental information structuring.............................21 3.5.3 Example of an environmental information system ..............................24 Methodological approach.....................................................................................27 4.1 The aims.......................................................................................................27 4.2 The interdisciplinary starting-point..............................................................27 4.3 The research project.....................................................................................28 The applied research work ...................................................................................29 5.1 Environmental life cycle assessment ...........................................................31 5.1.1 Brief introduction to the project...........................................................31 5.1.2 The life cycle assessment information structuring...............................32 5.2 Environmental design of products ...............................................................43 5.2.1 The RAVEL project.............................................................................43 5.2.2 The DfE information structuring .........................................................43 5.3 Environmental characterization modelling ..................................................51 5.3.1 The OMNIITOX project ......................................................................51 5.3.2 The LCA characterisation information structuring..............................52 Distillation of results from the applied research ..................................................55 6.1 Integrated industrial environmental information management systems ......55 6.2 Beyond concept modelling ..........................................................................57 6.3 Need for indicators.......................................................................................58 6.4 General principles from experiences ...........................................................58 6.4.1 Economy of the information system life cycle ....................................58 v

6.4.2 Cognitive sciences ...............................................................................59 6.4.3 Physical reality.....................................................................................60 6.4.4 Quality..................................................................................................60 6.4.5 Work procedure ...................................................................................62 7 Synthesizing a methodological framework..........................................................63 7.1 General need for methodological guidelines ...............................................63 7.2 Outlining the framework..............................................................................65 7.2.1 Principle 1: The economic life cycle ...................................................66 7.2.2 Principle 2: The cognitive science perspective....................................67 7.2.3 Principle 3: The physical reality ..........................................................68 7.2.4 Principle 4: Quality ..............................................................................69 7.2.5 General application of the framework .................................................69 8 Conclusions..........................................................................................................73 8.1 Quality of the research .................................................................................73 8.2 Practical results ............................................................................................74 8.3 Final conclusion about the research project.................................................75 9 Recommended future research.............................................................................77 9.1 Refinement of the framework ......................................................................77 9.1.1 Economic life cycle of environmental information systems................77 9.1.2 Cognitive science .................................................................................77 9.1.3 Establishment of the physical reference ..............................................78 9.1.4 Quality dimensions ..............................................................................78 9.1.5 Generalising the framework to encompass sustainable development..79 9.2 Research economic and practical consequences of improved environmental information structures ..............................................................................................79 9.3 Indicator information system .......................................................................79 9.4 Development of methods and tools, and integration of methods and tools for improved information structuring ............................................................................80 References....................................................................................................................81 Acknowledgements......................................................................................................89

vi

Framework for Structuring Information for Environmental Management of Industrial Systems - Raul Carlson

1 Introduction 1.1 The research project The research presented in this thesis has aimed at addressing, and trying to solve two problems: • Develop information structures for databases, communication files, reports and software for environmental management of industrial systems. • Synthesize a general methodology or framework for how to develop such information structures. This thesis will present how the problems have been approached and solved, by presenting the interdisciplinary toolbox, the methodological approach to solving the problem, three information structures that describe aspects of environmental management of industrial systems, and an interdisciplinary methodological framework. In addition, the thesis will highlight the insight that indicators are economically crucial when establishing information systems for environmental management of industrial systems. The studies and the writing of this thesis have been financed and performed within the ten year governmentally financed Swedish national competence centre CPM1, which is a joint research forum, equally financed by the Swedish government through VINNOVA2, Chalmers University of Technology and Swedish industry3. The work has intended to strengthen the theoretical foundation of Swedish work with developing and integrating environmental product life cycle responsibility in industrial applications, and specifically the unique emerging new research area of industrial environmental informatics, developed out from the CPM research.

1.2 Information and environmental management Few immediate physical relationships with Earth’s natural environment are easily understood from the viewpoint of being within the industrial society. Societies of hunters, collectors and farmers have a much more direct relationship with the physical environmental consequences of their actions. Instead awareness about environmental consequences from actions within the industrial society must come only from second hand information about how energy, resources, emissions and waste are related to all the goods and services from which industrialised people benefit. Such second hand information needs to be both understandable to people, as well as correct with regards to the environmental aspects it addresses. Time has proven the business relevance of environmental responsibility, and there are no indications of a change in the future. On the contrary; business relevance may be exemplified by commercial stakeholders and participants in, for example, development of international standards for environmental management and tools (ISO 1

CPM: Center for environmental assessment of Product and Material systems, Chalmers University of Technology, Sweden, Göteborg 2 VINNOVA: Swedish Governmental Agency for Innovation Systems 3 Since the start of CPM in 1996 the following different companies have participated: ABB, Akzo Nobel, Assi Domän, Avesta Sheffield, Bombardier, Duni, Electrolux, Ericsson, IKEA, ITT Flygt, MoDo, Norsk Hydro, Perstorp, SAAB Automobiles, SCA, Stora Enso, SwedPower (Vattenfall), Telia Sonera (Telia), Tetra Pak, AB Volvo, Volvo Cars

1


14001, 2004), Environmental Product Declarations (GEDnet, 2005) (ISO, 2002) and the World Business Council for Sustainable Development (WBCSD, 2005). Environmental management of industrial systems is exercised by different people with different responsibilities, different competencies and with different motivations. Depending on their responsibilities and what roles they play, they need different environmental information. The information needs to be understandable, relevant to the responsibility and have a suitable degree of complexity with regards to how much time can be spent on the information. Because environmental realities change, information systems need to be designed for efficient updating of data and quality review. Unless data are updated, the data will not reflect the changes in reality. And without information quality management, data are not taken seriously. This thesis shows how environmental information systems have been built with these practical requirements in mind. The information structures are based on different theories and techniques for data modelling, and the resulting information systems have proven to be stable, functional, understandable, relevant and in line with industrial economics. During the work a lack of general theoretical guidance for the work has been noticed. To improve this, general methodological principles have been identified and a general framework has been developed to support both ongoing and future work.

1.3 Sustainable development In 1987, the World Commission on Environment and Development (the Brundtland Commission), agreed to define sustainable development as follows: Sustainable development is development that meets the needs of the present without compromising the ability of future generations to meet their own needs. Since the Brundtland commission the significance of sustainability and the urge for sustainable development was emphasized and globally accepted in 1992 at the world summit in Rio de Janeiro, as Agenda 21 (United Nations website, 2005) and in 2002 at the world summit in Johannesburg. Sustainable development means that the way that mankind uses resources, structures global society and develops new technology needs to be improved. This implies a necessity to assess the consequences of our current and intended behaviour and actions. Such consequences include depletion of natural resources, economic inequality and global spreading of diseases. These are global issues, and they are strongly catalysed and amplified from the current globalization of production, trade and economy. Hence, sustainable development is tightly linked with global information sharing and global trade. Chapter 40 of Agenda 21 describes the needs for data, information, experience and knowledge and also calls for development of indicators for monitoring progress towards sustainable development. This need may in general be described as a need for information structuring and information systems by • National and industrial leaders to develop and maintain sustainabilityoriented policies. • Citizens to guide everyday lifestyle and political decisions. • Business people to support their actions and goals, and to guide their business towards sustainability.

2


But still, sustainability is considered to hold a relatively low priority in most people’s everyday lives. This paradox needs to be taken into account when designing solutions for provision of information for environmental management of industrial systems. Complex environmental relationships need to be formulated and structured into information that makes sense in the situation, with reasonable cost and effort. Information about the environmental consequences resulting from a change in the design of a product needs to be followed by information about the economic and social consequences from that same change. Information about the environmental burden from the production of a product needs to be compared to the environmental burdens from the raw material extraction, the use phase, and the end of life phase. The environmental impacts associated with different chemicals need to be presented comparably.

1.4 The task of information structuring This thesis presents an approach to structuring information in such a way that it efficiently and effectively helps people who take responsibility for environmental management to relate their actions to their environmental consequences. This includes formally relating information from different disciplines into interdisciplinary information structures, identifying information items to include with data structures, identifying logical relationships and flexibility between different related applications, identifying needs and possibilities for nomenclatures and distinguishing between numerical information, classifiers, and free descriptive texts. The practical cases presented in this thesis provide examples of these aspects of environmental information structuring. There are many examples pointing to a critical need for a good framework and principles for developing good information structures for industrial environmental management and intelligence. A quote from Professor Sterner (Sterner, 2003) may further emphasize the relevance of the scope of this thesis: Information plays a special role in policymaking, and in fact, information provision can be considered an instrument in its own right. On a general level, all policy depends on information; that is, policymakers must understand the technology and ecology of the issues under consideration. Many databases with valuable environmental data exist, but those will not suffice. In the report Establishing common primary data for environmental overview of product life cycles (Carlson et al, 2005) (Naturvårdsverket, 2005) the availability of environmental data is assessed, and it shows that the existing environmental databases have been developed for specific applications, such as emission control, chemical risk management or environmental reporting, but all new applications for environmental supply chain management, life cycle responsibility, scrapping manuals and design for environment in all sectors will require both more data and adaptations to new compatible nomenclatures and meta data. New databases will have to be developed for specific tasks, for shared or public applications or for environmental intelligence and management in commercial businesses. Guidance about how to develop structures for environmental information is needed. Structuring of information is to a large extent practical interdisciplinary work. To do this work one needs to deal with the language and the logic of the future users of the information, and one needs to try to understand how the future users see the world. This thesis formulates some of the techniques, theories and procedures applied. 3


It also outlines a practical framework for structuring of environmental information, based on much practical experience and theoretical reasoning.

4


2 State of the art The interdisciplinary scientific terrain navigated by this thesis is not very well investigated in literature, but there are authors in the literature pointing both at the needs and towards parts of the solutions. But in spite of the fact that the needs are quite clear both to industry and to society it has not been possible to find literature that addresses the problem in any similar way. When the work started in the mid 1990s life cycle assessment (LCA) was about to become interesting for both academic and industrial applications. Hence, the needs were then concretely expressed as needs for LCA databases and data exchange formats (Ekvall et al., 1992) (Grisel et al, 1997). The results of trying to meet these needs were a number of competing public LCA data formats and databases (Singhofen et al, 1996) (Carlson et al, 1995) and the initialisation of an ISO standardisation that started in 1998 (ISO, 2002). The discussion throughout this work with structuring information has been confused, and has consisted either of listings of important data items identified by LCA experts (SPOLD, 1997) on the one hand, and on the other hand discussions of analysis of concept modelling and industrial application of LCA (Carlson, Löfgren, Steen, Tillman,1998) supported by computing science theories. For example, the discussions during the international standardisation work aimed at including both environmental management experts from the technical committee responsible for developing the environmental management standards ISO 14000-series (ISO, 2004) and the information systems experts from the technical committee working with standardizations product data structuring (ISO, 1998) (ISO, 2004). This bridging did not work out, to some degree because environmental experts mistrusted data modelling experts, and the vice versa. The resulting ISO document, Environmental management – Life cycle assessment – Data documentation format ISO/TS 14048 (ISO, 2002) is a document at the peak of formalism from the viewpoint of environmental expertise, but which was criticized for being vague and ambiguous by information systems experts within the CASCADE project (CASCADE, 2006). The conclusion is that there is a wide gap between the domain of experts in environmental management methods and tools and the information systems experts. The problems that may arise from this communication gap between information systems experts and environmental experts, is that many of the environmental information systems that are needed for sustainable development and environmental management are built either by environmental experts without formal knowledge of the basics of information structuring, or by information systems experts based on an oversimplified view of the complexities of the field of environmental management and sciences. Today the needs for environmental product information have gone beyond the LCA methodology, and instead encompass all environmental aspects of products and production. These needs are expressed at the European level (Nuij, Rentsch, Ryder, 2005), national level (IVL, 2002) (Kemikalieinspektionen, 2004) (Naturvårdsverket, 2005) and in international forums (UNEP/SETAC, 2006). Considering that it is likely that issues of sustainable development will never be simply based on consensus (Hopwood, 2005) it is also likely that environmental information will remain based on ambiguous and complex facts that cannot easily be simplified. In his book The skeptical environmentalist Lomborg (Lomborg, 2001) presents an analysis of occasions when environmentalists seem to have pushed for their 5


environmental cause at the expense of a minimal formalism and accuracy. Similar tendencies were noticed during the above mentioned development of the SPOLD format for LCA data. Consistency with LCA practice was more important than consistency with, for example, statistics describing the quantitative data and a formally structured format. These examples from the literature, as well as experience, point to a need to facilitate discussions about the quality of environmental information structuring. This is not intended to concern computing science as such, since there is already a substantial literature available on that subject (Cornwell, 1990) (Elmasri, Navathe, 1994). For example, many systems, methods and tools exists to support business intelligence by acquiring, producing, interpreting and structuring business relevant information (Raisinghani, 2004) to design for example the database and communication formats (Hawryskiewysz, 1994). But there definitely is a need for a practically based framework that can facilitate a formal discussion about good practice when structuring information for environmental management, and this framework needs to be formulated from the viewpoint of the environmental sciences. Currently this discussion is mainly held between environmental experts and software or database vendors (DG Environment and DG JRC, 2006) (ecoinvent, 2006) (Pré, 2006). It is difficult to elevate the discussion to consider, for example, the fact that data need to be well-structured because in the long run the costs for data will far exceed the costs for all other parts of the information system. Figure 1 is intended to provide a simple picture to describe the fact that the structuring of data is important, since the costs for data are important in relation to the other costs of an information system. The picture is taken from a small handbook about product data structuring in industry (Celander, 1995). Education Integration Software Hardware

Data

Figure 1. Schematic sketch of typical relative costs for creation of data and other costs for an information system (after Celander, 1995) Areas represent costs.

Attempts to combine environmental information structures and structured data handling to reduce costs and enhance quality have been made in different industrial sectors, such as within the Swedish paper and pulp sector (Pålsson et al, 2005), the Swedish iron and steel sector (Axelsson et al, 2004), and the international automotive manufacturers (IMDS, 2005) and another is currently being initiated within the Swedish building and construction sector (BASTA, 2006). The paper and pulp sector and the iron and steel sector approaches were based on one general principle and had the same aim. The idea was to use one integrated environmental information system to generate all of the various environmental reports within a company. The automotive and building sectors approach instead aims at sharing materials and 6


components data between all suppliers and manufacturers in the two industrial sectors. These four projects and the two different approaches have not had any common approach or idea of how to structure the information, and there is no common way to judge whether the structures used are good or bad from the environmental viewpoint. The background of the paper and pulp approach is described in section 5.1 of this thesis. The basis was the PHASETS model, presented in Figure 26. The iron and steel approach was inspired by this same model and initiated and performed a project within that sector, with the same overall aim as the paper and pulp sector project. The automotive industry system is intended as an exclusive system, currently outsourced to the company EDS4 that also developed the system. The Swedish building and construction sector system BASTA is based on similar principles but has nothing else in common with the IMDS. When paper and pulp, iron and steel, automotive and building and construction industries need to share data, or when the have the same sub-suppliers, the question of information structuring will arise as a compatibility and quality question. Hence, the current societal and industrial needs for environmental information structures for different information systems, together with the apparent gap between environmental experts and computing experts, motivates the research behind this thesis. It should be stressed that the thesis is not intended to address a lack in either of the domains of environmental science or computing science, but is intended to bridge the gap between these two scientific areas. Hopefully the practical experience in this strikingly unexplored interface between fields at least provides something new to both computing science and environmental sciences. Due to the kinship between information structuring and linguistics the focus is on the wording and language of the environmental experts, and on formal weaknesses introduced by the necessary breadth of this interdisciplinary field.

4

http://www.eds.com

7


3 Concepts and terms Concepts and terms relevant to this interdisciplinary thesis relate to the different competence areas of informatics, environmental management and some general aspects of industrial systems. The chapter is divided into five sections. The first section defines the meaning in this context of some terms that are generally used throughout the text. The next three handle the different competence areas, and the fifth one clarifies the view of environmental informatics in the context of this thesis. Each section is intended to present concepts from the different disciplines or competence areas involved.

3.1 General concepts, terms and distinctions Concepts and terms with potentially ambiguous meanings will be used in different contexts throughout this thesis. Their meaning in the context of this thesis is defined here: DATA, INFORMATION AND FACTS • Data: refers to raw figures, letters or other symbols that are not yet subject to interpretation. • Information: any form of data that is interpreted. • Fact: information that describes some aspect of an ontology. DATA FORMAT • Format (data f., data documentation f., database f.): The term relates to how the data fields are structured in, for example a computer file, a questionnaire, or a relational database. CONCEPTUAL MODEL VERSUS CONCEPT MODEL • Both concept models and conceptual models are presented and discussed in this thesis. Conceptual models provide a simplified (conceptual) view of items in the real world, their constitutions and different physical, chemical, and biological relationships. Concept models describe the concepts and terms of a language and how these concepts and terms are logically related to each other. For example, Figure 4 provides a conceptual model of the life cycle of a product, from the viewpoint of how it impacts the natural environment. Figure 9 provides a concept model of basically the same thing. It illustrates the importance of the three concepts Technical System, Nature System and Social System, as well as their interrelationship. Note: It should be stressed here that this distinction is not made throughout the literature. In many publications there is no distinction at all between the meaning of conceptual and concept. ONTOLOGY • Ontology is the study or the theory of being or existence, and about how to describe reality and how to acquire knowledge about reality. In this thesis ontology represents an address to a physical reality, an address to the real world, as opposed to, for example, consensus, conventions or fiction. 9


3.2 Concepts of industrial systems 3.2.1 Scale and productivity Concentration of production capacity into factories and work force into cities are two important factors behind the success of industrial society. These factors lead to many synergetic advantages of scale both due to increasing cost-efficiency and by larger, more concentrated and more homogeneous consumer markets. These elements of industrial systems are considered in this work by specifically addressing information systems that support large scale production systems, large markets, distributed supply chains and business competition. Large numbers and homogeneous markets stabilize supply chains and market demands. But the stability is different for different organizations and businesses, and also change over time. For an organization to be aware and respond to change it needs both intelligence and information strategies and systems.

3.2.2 Organisational intelligence and information Organisations handle information for intelligence and management (Frankelius, 2001). Intelligence aims at building up the knowledge within the organisation, so that the organisation can plan and react in line with this. Management needs to control and develop the production capacities into best available productivity. To be successful organizations need to maintain and update their knowledge, and adapt to a fragmented world, spontaneously react to unexpected facts, adapt to changed circumstances, take decisions based on uncertainties, be creative and flexible, and be able to handle generalizations. At the same time organisations need to be productive. Tasks, activities and functions need to be merged and integrated, spontaneity needs to be replaced with standards and formal routines, information systems need to be stable and trustworthy, work tasks need to be rational and efficient, new ideas need to be turned into operative functions, structures need to be cemented and specialized tasks need be developed. Successful organisations balance between knowledge maintenance and the productivity. Since environmental management concerns change in the world outside businesses, environmental information systems need to be designed for efficient updating of data when knowledge changes, as well as for effective quality review. Unless data are updated they will not reflect knowledge about the changes in the external reality. And without effective quality review, the data will not be taken seriously. The focus of this thesis lies on productivity, with respect to the fact that a sound business needs both perspectives.

3.2.3 The life cycle of industrial information systems Information systems are developed, exist, and remain within organisations during changes and development over time. Sometimes the development of the information system leads a change, and at other times the information systems need to be adapted to changes with other causes. For the knowledge oriented viewpoint (section 3.2.2) such information system life cycles are short. Dirk Vriens proposes an intelligence cycle of four stages to structure the process of intelligence (Vriens, 2004): 1. Direction: The organization determines the information requirements i.e. what aspects of the world to collect data about. 2. Collection: Determine what data sources to use, as well as actually collecting the data. 10


3. Analysis: The actual production of intelligence. 4. Dissemination: Forward results from intelligence to strategic decisionmakers. This cycle describes how organisations should look after their organisational environment, e.g. for competitors and other market relevant changes. The intelligence process also develops the organisation into core business areas and into its strategic directions and domains. But as the strategic domain becomes stabilized and familiar to the organization, the intelligence system needs to be established as a productivity information system (section 3.2.2), with more clearly defined information items and structures. This organisational maturity cycle leads to development and change in information systems. During the life cycle of an information system functionality may be added, removed or redeveloped. Such changes may be the result of new requirements from the users, new requirements on the data, additions and removals of surrounding information systems, or due to changes in the business environment. Regardless of the reasons, any information system that is built for a real business environment must be designed for such future changes during its life cycle. In particular, the data in the system needs to be separable from choices of implementation on specific software or hardware platforms (see also Figure 6, section 3.4.1).

3.3 Concepts and terms of environmental management 3.3.1 Sustainable development Sustainable development is a globally accepted agenda, the overall agenda for environmental management. The large scales of the industrial systems have been successful at producing commodities and economy to many people in a relatively short time. But in parallel to this successful development social and environmental drawbacks have increased as well. Exhaustion of natural resources, spreading of manmade substances and global social inequality may lead to the fact that poverty, health problems and wars are the main legacy of our industrial society. In 1987, the World Commission on Environment and Development (the Brundtland Commission) said Sustainable development is development that meets the needs of the present without compromising the ability of future generations to meet their own needs. The significance of sustainability and the urge for sustainable development has again been emphasized and globally accepted both in 1992 at the world summit in Rio de Janeiro and in 2002 at the world summit in Johannesburg.

11


Figure 2. The target dimensions of Sustainable development: Environment, Economy and Society.

Sustainable development is often described as three circles for the target dimensions of Environment, Economy and Society (see Figure 2). It illustrates that economic, social and environmental processes are interlinked. Any public or private actions must take this into account. Also, sustainable development goes beyond environmental issues. In order to satisfy our material and immaterial needs, economic growth and stability is needed. Hence, sustainable development implies that changes are needed in the economic and social systems, to reduce consumption of the environment and resources, while maintaining economy and social well-being, global relationships included. In conclusion, sustainable development is intended to bring about long-term improvements for the majority of the people of today and the future. This thesis focuses on the environmental dimension of sustainable development, with full awareness of that the different dimensions are closely interlinked, in time as well as globally.

3.3.2 Environmental management Environmental Management is addressed here in the broad sense of sustainable development. Management of the environment is a responsibility for any person, at his or her position in society and with his or her role as citizen, professional or consumer. Due to the large scales of the industrial systems (section 3.2.1), personal choices of individual actions may either be regarded as vain and pointless if they are driven by individual beliefs and convictions, since the individual actions does not make much difference, or as firm and powerful if they are performed in concert with a common agenda, since many individual contributions in the same direction combines into a strong force. This thesis considers sustainable development to be a common agenda for a concerted action, and hence considers individual decisions as controlling the total outcome of today’s industrial society into the future. Figure 3 represents how such individual decisions are taken throughout the society. Individual decisions are taken by the controller, on the basis of how the controller appreciates the current status, the short time goal, and the vision of the long term goal. The model in figure 3 is a cybernetic control system (Wiener, 1991), which implies that all the variables change dynamically. The vision of the sustainable society changes, the short term goals changes, the current status changes, and even the controller and the controlled system change over time. 12


Current status

Goal

Control

Controlled system Controlled system

Sustainability performance

Sustainable industrial Controller society Figure 3. A conceptual cybernetic control model describing environmental management. All variables changes and the information therefore need to be relevantly updated.

A control system intended to control a real system relies on the fact that the controller has sufficiently correct information, so that the controlled system, the current status and the short term goal are appropriately updated when changes happen due to control actions or external changes. This means that the information needs to be updated at all times. Methods, routines and processes of environmental management have been standardized within ISO, as the ISO 14000 series. The 14001 standard (ISO, 2004) specifically prescribes how to control the environmental performance of individual organizations in a continuously improving process. In terms of Figure 3 the ISO 14001 standard considers the controlled system to be, for example, a company or a production facility. The current status and goal are updated on, for example, a yearly basis to produce environmental reports. Each individual person in the company is responsible for taking environmental decisions in their everyday work, ranging from purchase decisions, design choices, waste handling, etc. In support of environmental management there are many different methods and tools (Carlson et al, 2005). In the following subchapter the methodologies of environmental life cycle assessment (LCA) and design for environment (DfE) will be introduced.

3.3.3 Life cycle assessment Life cycle assessment (LCA) is a method by which to study and assess the significant environmental impacts from the full life cycle of a product or a service, including all significant matter and energy needed. The life cycle is assessed upstream from resource extraction and downstream to waste management, including recycling or reuse. Figure 4 shows a conceptual model of the different items and relationships addressed in LCA. An LCA is performed by a subjective individual, who decides which environmental issues to consider in the natural environment, which upstream and downstream causal activities to include with the assessment, which resource use and emissions (elementary flows) to include, and on the transparency and overall quality of the resulting study. An LCA is performed by deciding on a product or the function for which the assessment should be performed. Examples of a product is ‘kg of steel’, and an example of a service is ‘ton*km of passenger transportation’. From the basis of the product or service all significant upstream and downstream processes are identified and linked into a flowchart, like the upper half of Figure 4. Identification of what is significant is quite difficult, and may be supported by principal policy decisions for the whole study, by specific knowledge or by ad hoc investigations during the work. 13


With a foundation in the flowchart, data for all significant elementary flows are described for each process. Each elementary flow is assigned to the category of environmental change to which it contributes, and the environmental consequences of each elementary flow are modelled as characterization models. Characterisation will be presented in more detail in the following.

Figure 4. A conceptual model of the scope of the methodology of life cycle assessment (LCA) (after Carlson, Pålsson, 1998. The Subjective mind is added to stress the importance of choices during an LCA study).

Characterisation is an LCA term that describes characterisation of environmental impact on specified selected environmental issues, i.e. environmental indicators (such as increased number of cases of cancer or depletion of oil resources) from specific environmental loads from elementary flows (such as emission of freon or extraction of crude oil). Characterisation models that describe these relationships are used in LCA studies, but they are generally modelled by specialised experts with expertise in different environmental impact modelling. Figure 5 represents how a load enters a natural system and how its impact is represented by impact on the indicator. When describing this impact by a model, the nature is represented by a framework of different properties, for example temperature and concentration of H+-ions, and mechanisms, such as leaching, dispersion, permeation and biochemical reaction, which describe how the indicator is impacted by the load.

Figure 5. Conceptual model of the system of a characterisation model. This picture is partly borrowed from (Tivander, Carlson, Erixon, Pålsson, 2004), but with addition of the figure representing the subjective mind of LCA.

14


LCA also includes the steps of weighting and interpretation. Weighting means to systematically prioritize between many different environmental changes, such as ozone depletion, acidification, depletion of natural resources and human health effects. Interpretation is the way to interpret the result of the LCA study. LCA may be performed in different ways, but there are four international standards that establish an international consensus about how to perform LCA: • The framework is described in Environmental management - Life cycle assessment - Principles and framework ISO 14040 (ISO, 1997). • The way to perform inventory of all material and energy flows of the life cycle is described in Environmental management - Life cycle assessment - Goal and scope definition and inventory analysis ISO 14041 (ISO, 1998). This stage of LCA is called LCI, life cycle inventory • The way to perform environmental impact assessment is described in Environmental management - Life cycle assessment - Goal and scope definition and inventory analysis ISO 14042 (ISO, 2000). • How to perform interpretation of results is described in Environmental management - Life cycle assessment - Life cycle interpretation ISO 14043 (ISO, 2000). The ISO 14040 standard was updated, and a new version released by ISO, in 2006. The standards ISO 14041-14043 will be replaced by a new standard ISO 14044, also in 2006.

3.3.4 Design for environment Design for Environment (DfE) is based on the idea that a significant part of the environmental impact from a product originates from choices made during the design of the product. The later in the design process one considers environmental issues, the more difficult and the more expensive will it be to introduce any changes. Hence, environmental benefits and improvements should be considered as early as possible during product design. Introduction and management of DfE is the responsibility of environmental management, and as such it has been noticed as a technical report in the ISO 14000 series of standards ISO/TR 14062, Guidelines to Integrating Environmental Aspects into Product Design and Development (ISO, 2002).

3.4 Concepts of informatics 3.4.1 Basic concepts The section about informatics concepts is introduced by presenting how two dictionaries describe informatics (The original dictionary texts are edited): Wikipedia5: Informatics is the study of information. It is related to database, ontology and software engineering, and is primarily concerned with the structuring, creation, management, storage, retrieval, dissemination and transfer of information. Informatics also includes studying the application of information in organizations, on its usage and the interaction between people, organizations and information systems. 5

http://en.wikipedia.org/wiki/Informatics, 2006-02-28

15


Oxford English Dictionary6: Informatics is the discipline of science which investigates the structure and properties (not specific content) of scientific information, as well as the regularities of scientific information activity, its theory, history, methodology and organization. The problem falls into two parts: the preparation of decisions, which is a matter of informatics, and the making of the decisions themselves, which is a matter of ‘politics’.

Figure 6. Basic functions and concepts of an information system. (Carlson, 1994)

Figure 6 shows by analogy the fundamental parts of an information system based on a database. The analogy is based on a library, with a librarian that speaks a foreign language (SQL, Structured Query Language, an ANSI/ISO standardised database language), a translator, information handling and manipulating tools and the user or controller of the information system. The arrows represent information exchange between the different parts. It should be understood that by using network technology, such as the Internet, the arrows can exchange data over very long physical distances. Key issues for turning this information system into a functioning system are: 1. The correct understanding of how the user perceives the information presented to him or her. 2. The correct interpretation of all the information and logic of all the different data to store in the database. 3. The correct storage, retrieval and updating of information in the database and presentation of data into user interfaces and reports. 4. Correct data manipulation with regards to all intentions by the user and his or her intended domain of applications.

6

http://dictionary.oed.com/cgi/entry/50116495?single=1&query_type=word&queryword= informatics%A8&first=1&max_to_show=10, 2006-02-28

16


Key issue 1 and 2 are the central themes of this thesis, since the meaning and the logic of the information is of core interest for information structuring. Both of these aspects of information define the structure of the database and the communication between different parts of the information system. Examples of database schemas that prescribe the structure of a relational database are presented in figure A1 and A2 in Annex A. Key issue 3 is the central theme when designing the architecture (section 3.4.2) and the functionality of the information system and when programming the functionality. Key issue 4 is the interdisciplinary responsibility when developing the entire information system for, for example the purpose of life cycle assessment (LCA) or design for environment (DfE), to understand how LCA or DfE is performed.

3.4.2 Information structuring Information structuring addresses three major structuring types: • Concept modelling: The result of an analysis of concepts, terms and the logic of the language of the application domain, such as the language used to perform a life cycle assessment (LCA) or to support design for environment (DfE). An example will be presented in section 3.5.2. • Vertical information structuring: A model of all consequences of an information request. Models how a specific piece of information is consecutively aggregated into a resulting report, or how analysis of ideas are synthesised into an information system. Examples are presented in sections 3.4.3 and 3.4.4. • Information system architecture: Descriptions of how an information system is modularised by separated components, and how these components communicate. Describes also the boundary of the scope of the information system. Figure 36 in section 5.2 and figure 37 in section 5.3 provide examples. This thesis also addresses conceptual modelling (see section 3.1 for definition) and structuring of quality aspects of information (section 5.1).

3.4.3 The OSI reference model, vertical information structuring The OSI7 reference model is presented here as an example of a widely used vertical information structure. It has inspired some of the work presented in this thesis. The OSI model describes how to consecutively aggregate data from the lowest physical layer of computer networks to the user interfaces of the applications, through increasingly meaningful layers. In the reverse direction it shows how information from applications is disaggregated into standardised signals on computer networks. The OSI reference model was developed within ISO8 to address problems of network incompatibility, and it is used to understand how information travels throughout a network. It explains how packets travel through the various layers between computers on networks, even if the sender and destination computers have different types of network media. The model consists of seven numbered layers, each 7 8

OSI: Open Systems Interconnection Reference Model ISO: International Organization for Standardization

17


of which illustrates a particular network function. See figure 7. This vertical dividing into layers provides a number of advantages, such as that: • It breaks complex computer network communication into small, manageable tasks. • The division makes computer network communication easier to understand. • It allows computers from different vendors to communicate. • It separates tasks, and thereby prevents changes in one task from affecting other tasks. 7 Application

7 Application

6 Presentation

6 Presentation

5 Session

5 Session

4 Transport

4 Transport

3 Network

3 Network

2 Data Link

2 Data Link

1 Physical

Physical network

1 Physical

Figure 7. The OSI reference model describes how information is aggregated/disaggregated from/to physical network signals into/from application specific data, like file exchange protocols.

The specific functions of each layer in the OSI reference model are (NetworQuest, 2006): 1. Physical Layer: Handles the bits, connectors, voltages and data rates i.e. it deals with the physical medium like the cable and sends the data in forms of 0s and 1s. 2. Data Link Layer: Responsible for reliable data transfer of data through the media. Understands the physical address of the network device and helps in flow control and error correction during data transfer. 3. Network Layer: Provides reliable data transfer. Deals with the logical Internet protocol address. 4. Transport Layer: Responsible for transportation issues between computers. Responsible for fault detection and recovery of flow control. 5. Session Layer: Responsible for the virtual circuits for data transfer between computers. 6. Presentation layer: Responsible for decrypting data, and formats the data as sent by the sender. 7. Application Layer: Provides the network services to the applications, like e-mail, file transfer or remote monitoring of another computer.

18


Computer networking can be developed, established and understood without the OSI reference model, but it makes networking simpler to understand. Therefore compatibility and development of new compatible functionality gets easier as well.

3.4.4 Information structuring as structuring of ideas Concepts and terms are the prime constituents of information systems. Examples of concepts or terms are Greenhouse gas or Mass. There is no clear difference between concepts and terms. Generally a term is a prime item while a concept can be expressed by a combination of terms and other concepts. In some applications Greenhouse gas may be a concept and in other applications it may be a term. Concepts and terms mediate content and provide meaning to data in, for example, databases, files and reports. Content is mediated by letting a container of data represent a concept or a term. An example of such a container may be a data field in a database, named Mass, which contains the data kg. In this case the data item kg is mediated by its data field Mass. Meaning is provided to a data item when a user can make sense of the data from relating the concepts or terms into his or her context. Information structuring starts with identifying the concepts and terms for which an information system should hold content and provide meaning. ideas

language user A

user B

Direction of design process

Functionality of information system

concept model

data model

information system shared and commonly understood information

copyright Raul Carlson, 2000

Figure 8. The figure describes how information systems are intended to mediate communication of ideas, and how the design and the use of information systems are related with the language and the ideas.

The vertical information structuring model presented in Figure 8 is partly inspired by the OSI reference model presented in the previous section. The model simultaneously presents two important aspects of information structuring: • Concept modelling: How to design the data model for an information system based on a database like the one presented in figure 6. 19


•

Functional system architecture: The information system is ultimately intended to mediate communication of ideas between its users.

Figure 8 illustrates that an information systems is intended to mediate communication of ideas. Here idea means a mental state with a specific meaning that can be formulated into a message. Exchange of ideas between people is naturally done by language. If users A and B are separated in space or time they need to communicate their message over some medium that transports or stores their language, in the form of books, e-mail or databases. The lack of direct contact introduces risks for misinterpretation. This risk increases if the two users have different competence, are in different situations, and if the information they exchange is not supplied with any explanatory additional information. The risk decreases if the users communicate about a specific issue that is common between the two, such as environmental management, LCA or DfE or if they take action to add extra explanatory text to their messages. Design of information systems is accomplished by acquiring the meaning of the language of the field that the users intend to communicate about, and developing concept models of this language. It is stressed in this thesis that the concept modelling is facilitated by ensuring that the conceptual models are shared between the different intended users and the information structuring expertise. The resulting concept models are translated into the data model of reports, communication files and databases of the information system (section 3.5.2). The result is that users can mediate ideas and communicate meaning, even if A and B never meet. Concept models and data models are intended to aid the design of databases, communication files, questionnaires, reports, forms or web pages for use in different control systems, like the one described by figure 3.

3.5 Concepts of environmental informatics 3.5.1 Environmental information In this thesis environmental information has the purpose of supporting environmental management of different systems of industrial society (section 3.2.1), in the context of sustainable development (section 3.3.1). This means that information about the natural environment is always considered in relation to how it is impacted by activities in the industrial society, and in relation to how people or groups of people relate to these impacts. Figure 9 is a high level concept model that describes how information from three different disciplinary areas together are needed to form a unit of interdisciplinary environmental information: • Technical System: Represents information that describes industrial processes and systems of such processes i.e. data from e.g. engineering disciplines. • Natural System: Represents information that describes nature i.e. data from e.g. medicine, biology, or ecological disciplines. • Social System: Represents information about the non-physical aspects of human beings i.e. data from e.g. economy or humanities disciplines. The diamonds between the three disciplinary boxes represents the interdisciplinary relations of environmental information: 20


• • •

TN: Represents the relationships between causes in the Technical system that results in impacts in the Nature System. SN: Relates both to the fact that people or groups of people in the Social system notice impacts in the Nature System, and that people prioritize or give different weights to the different impacts that they notice. TS: Represents that people or groups of people in the Social system benefit from the goods produced by the Technical System of the industrial society.

TS

Social Social System System

Technical Technical System System

TN

SN

Nature System System

Figure 9 Three types of information that are related into environmental information. This concept model was developed by the author as a high-level entity-relationship model describing the SPINE (Carlson, Löfgren, Steen, 1995) information structure.

The concept model in Figure 9 was developed by the author of this thesis to serve as an overview of the SPINE information structure (Carlson, Löfgren, Steen, 1995), and as a pedagogic tool to describe the nature and structure of environmental product data, including e.g. the life cycle perspective.

3.5.2 Example of environmental information structuring The following example describes in detail how concept modelling is performed on environmental information, using relational modelling by analysing functional dependencies9 of concepts and terms used by the experts in the field. The example here is taken from the LCA impact assessment model presented in section 5.1, and is based on an analysis of the then ongoing standardization of LCA impact assessment within ISO (ISO, 2000). Thanks to the efforts made in the ISO working group, much of the work associated with identifying and defining important concepts was already done. But some freedom for interpretation still remained. Important concepts from ISO 14042 for this example are Impact category, Category indicator (life cycle impact category indicator) and Weighting factor. According to the standard, the different concepts are defined in the following ways: • An impact category is a class representing environmental issues of concern to which LCI results may be assigned. • A category indicator is a quantifiable representation of an impact category. 9

A functional dependency is when one concept or term uniquely determines another concept or term. This is written A → B, and means the same as saying "B is functionally dependent on A."

21


•

•

Weighting is defined as converting and possibly aggregating indicator results across impact categories using numerical factors based on valuechoices. Weighting factor is not formally defined in the standard, but is described as being used to convert the indicator results or normalized results in some way. The standard also defines the concept of Category endpoint, as an attribute or aspect of natural environment, human health or resources, identifying an environmental issue of concern.

During the concept modelling it was not possible to practically or logically distinguish between the two concepts Category endpoint and Category indicator. Examples of endpoints could also represent indicators and vice versa. Therefore only Category indicator is represented in the model. It may be interpreted as representing both of the two concepts in the ISO standard. It was also not possible to individually distinguish between the two concepts Impact category and Category indicator. Examples of Category indicators could also represent Impact categories and vice versa. Semantic analysis led to the conclusion that the two concepts are defined in relation to each other, giving the result that Category indicator is an observed, quantified and measured environmental change, while Impact category is only observed as a class of environmental changes. The Impact category is defined in terms of ‘issues of concern’, which means that the selection or choice of Impact categories are chosen by a subjective mind (see also Figures 4 and 5). During the concept modelling it was concluded that a Category indicator or an Impact category needs to be identified in terms of the underlying reasoning. This introduced the new concept Impact indication principle, a concept not explicit in the standard. But it was implied by the definitions of the other concepts. The resulting concept model is presented in Figure 10. The arrows represent functional dependencies, and the model may be read as: • The underlying reasoning of the Impact indication principle needs to be defined and described before a Category indicator can be introduced. • An Impact category first needs to be defined as a Category indicator, before it can become an Impact category. • Each Category indicator may be a subclass of an Impact category. ImpactCategory

CategoryIndicator

ImpactIndicationPrinciple Figure 10. The concept model of Category indicators and Impact categories, as interpreted from the ISO standard ISO 14042:2002 Environmental management — Life cycle assessment — Life cycle impact assessment (see section 3.3.3).

In the standard ISO 14042 weighting is defined as the relative weights across Category indicators or Impact Categories. The standard also requests that all weighting methods used shall be documented to provide transparency. Figure 11 presents the concept model of weighting.

22


CategoryIndicator

WeightingFactor

WeightingMethod

Figure 11. The concept model of weighting, as interpreted from the ISO standard ISO 14042:2002 Environmental management - Life cycle assessment - Life cycle impact assessment(see section 3.3.3).

The concept model in Figure 11 may be read as: • Any Weighting factor must be assigned to Category indicator10. • Any Weighting factors are based on a transparently documented Weighting method. The combination of Figures 10 and 11 is straightforward, and is presented in Figure 23 in section 5.1.2.2. The result from the concept modelling is used to implement a data model, for example a relational database schema. Each concept presented above then comes to represent a table in that schema, and each table holds a number of fields or columns for different data. In the relational database schema the arrows are represented by foreign keys11, which establish the logical relationships between the different tables of the database.

Figure 12. The relational database schema resulting from the concept models presented in figures 11 and 12, which is designed as an interpretation of concepts from the ISO standard ISO 14042:2002 Environmental management - Life cycle assessment - Life cycle impact assessment.

10

To simplify notations in the concept models, the author has hesitantly but deliberately avoided cardinality of dependencies and relationships. Here it may be noted that weighting factor to category indicator should have a 2, n:m relationship, since weighting must concern the relative weight between at lest two category indicators. Weighting of one indicator makes no sense. This also implies that weighting factor is not an attribute of a category indicator, but an attribute of a selected and defined set of category indicators. 11 A foreign key is a set of selected fields in a relational database table, which matches the fields that uniquely identifies the rows of another table. The foreign key establishes the logical relations between the information in the relational database tables.

23


Figure 12 presents the relational database schema resulting from the concept models in Figures 10 and 11. Other examples of database schemas are presented in Figure A1 and A2 in Annex A. The concept models of each of these three types of information systems will be presented in Chapter 5, and claims about generalized applicability of the model are presented in Chapter 6.

3.5.3 Example of an environmental information system Figure 13 shows an example of an environmental relational database system built for industrial applications and scientific research (see also section 5.1). The system should be considered as a component in an environmental management system, like the conceptual model of the environmental management system presented in Figure 3.

Figure 13. An environmental database system, consisting of both the database file and the necessary administrative functions. See also section 5.1.

The database system in Figure 13 is built around a relational database file (SPINE@CPM) (Carlson, Pålsson, 1998). The file is structured with all fields and logical relationships dictated by the information structure, like the relational database schema presented in Figure 12 in the previous section, or the examples in Figures A.1 and A.2 in Annex A. Data acquisition routines to build up the contents of the database are formalised, to ensure that information is provided with appropriate quality in the right form and with the right semantics according to the concept model. A data review process assesses all acquired information so that it fulfils the quality requirements. The quality review is essential to environmental databases, since environmental data generally cannot be verified (Carlson, Pålsson, 1998). Users access the information in the form that data are published with the intended meaning as prescribed by the concept model, and with the intended and reviewed quality. The technical administration of the information system is considered as being part of the information system, responsible for maintaining the information structure. Environmental management and sustainable development facts and priorities are continuously changing due to new knowledge, increased public awareness and new policies. Therefore a functioning environmental database system needs a continuous relationship with research and development activities, and to be designed to receive and respond to user requirements. 24


Since environmental issues are not core business of most industries, the environmental information system also needs to be simple to use. Environmental issues do not come up on the daily agenda in the ordinary business life, and the issues are not discussed very often. They therefore also drop out of the mind of the intended users of the information systems. Therefore it is important if the information system itself supports and maintains the education of environmental issues within the organisation each time it is used. The work with specifying the information structure of this information system is further described in section 5.1, where also the information system is presented in more detail.

25


4 Methodological approach 4.1 The aims This research has aimed at solving two problems: • Develop information structures for databases, communication files, reports and software for environmental management of industrial systems. • Synthesize a general methodology or framework for developing such information structures. In addition to its practical nature, the applied research also identifies the important relationship between information systems and indicators.

4.2 The interdisciplinary starting-point The present work started out from a master’s degree in engineering physics with specialization in environmental science, statistics and computing science. The master’s thesis work was an interdisciplinary study, pertaining to structuring a relational database for life cycle assessment (Carlson, 1994), and it resulted in the initial input to the work described in this thesis. The work with that master thesis was based on the education from engineering physics, previous experience from practical work with production quality management in industry and the author’s personal interest in industrial total quality management (Shingo, 1984), cybernetic control theories (Wiener, 1991), and on theories of mass communication (DeFleur, BallRokeach, 1989) and cognitive psychology (Ellis, Hunt, 1989). These different interests and competencies were combined into an ad hoc methodological approach for structuring information and building information systems for industrial environmental management, consisting of a blend of classical physics, cognitive science, computing science and quality management. • Classical physics: The perspective of classical physics implies that environmental issues are based on a scientific study of matter and motion. The information systems are established from identification of simple physical properties, interactions, processes, and laws. The industrial environmental systems analytical approach is to study the natural or material world and phenomena. • Cognitive sciences: Information is regarded from a human viewpoint and terms such as meaning, semantics, context etc. are handled from the perspective of cognitive sciences. • Computing sciences: Methods and techniques from the area of computing science have been applied for data modelling, information systems design, etc. • Quality management: Ideas of total quality management (TQM) are built in to industrial environmental information systems. It is considered that information systems and information management methods should be designed with both short- and long-term total quality in focus, and that such principles should also be economically sound for all sorts of organizations in the industrial society. This interdisciplinary blend, and the different knowledge and principles behind the concepts and terms introduced in chapter 3 were the methodological starting points of the research work presented in this thesis. 27


4.3 The research project The research behind this thesis began with 6 to 10 separate projects running between 1994 and 2004, and financed by the Nordic Council of Ministers12, VINNOVA13, the European commission’s 5:th framework14 and 6:th framework15 programmes, different large Swedish and European industrial companies16, and by Chalmers University of Technology. The author of the thesis is employed by Chalmers University of Technology and has been project manager, subproject leader or technical expert with responsibility to design, develop, implement or maintain information structures and information systems in the different projects. Having been responsible at different levels for the outcome of these projects, it has been necessary for the author to formulate practical principles for the development of information structures for environmental management of different industrial systems. The methodological framework presented in chapter 6 is based on this practical experience. The aim formulated in section 4.1 and the interdisciplinary starting-point presented in section 4.2 have been the leading themes through all of these separate projects. In large, the work has consisted of building three different information systems aimed at supporting different environmental management methods and tools. Experience has been paired with theoretical studies aimed at understanding the underlying principles of information structuring for environmental management. Results have been reported in both scientific articles and in reports, and in the form of functioning information systems. The three different information systems are for: • Life cycle assessment (LCA) (see sections 3.2.3 and 5.1), • Design for Environment (DfE) (see sections 3.2.4 and 5.2), and • LCA characterization modelling (see sections 3.2.3 and 5.3). Considering the practical nature of the development of the information structures and the actual testing and implementation of the information systems the research work may resemble technical development, but with respect to that most of the work has been pioneer work, and the fact that the overall aims and theoretical foundations have been deliberate, and has been consciously updated and reformulated during the work, the projects might instead rather be categorized as action research, according to Checkland (Checkland, 1998).

12

The NEP-project: Nordic project on Environmentally sound Product development (see section 5.1) VINNOVA: Swedish Governmental Agency for Innovation Systems 14 The RAVEL-project (see section 5.2) 15 The OMNIITOX-project (see section 5.3) 16 ABB, AB Volvo, Akzo Nobel, Alstom, Antonio Puig S.A., Assi Domän, Avesta Sheffield, Bombardier, Deutsche Bahn, DSB, Duni, Electrolux, Ericsson, Holmen, IKEA, ITT Flygt, M-real (MoDo), Norsk Hydro, P&G Europé, Perstorp, Rolls Royce, SAAB Automobiles, SCA, SJ, Stora Enso, SwedPower (Vattenfall), Södra, Telia,Volvo Cars, Woodville Polymer Engineering 13

28


5

The applied research work

This chapter presents the applied research on which the reasoning of this thesis is based. The work has been constituted by developing information structures for three separate and different information systems in support of environmental management methods and tools. During the projects additional effort has been taken to make the information systems compatible by striving to have them share some of the same data structures. In the presentations of the concept models in this chapter this compatibility will be pointed out, together with reasoning regarding how it was done. The compatibility between information structures for different methods and tools is one of the benefits of having had a number of related projects over a long time. Through this it has been possible to learn how to integrate different industrial environmental information systems with each other, which in practice also means that the different methods and tools for industrial environmental management have been integrated with each other, concisely and cost-efficiently. In Annex C all concept models are listed together.

29


5.1 Environmental life cycle assessment ARTICLES THAT PRESENT THIS PROJECT: I. Carlson R., Löfgren G., Steen B., Tillman A-M., LCI Data Modelling and a Database Design; Published in The International Journal of Life Cycle Assessment, Vol. 3, No.2, pp. 106-113, 1998 II. Bengtsson M., Carlson R., Molander S, Steen B., An Approach for Handling Geographical Information in Life Cycle Assessment Using a Relational Database; Published in Journal of Hazardous Materials, vol. 61, pp. 67-75, 1998 III. Carlson R., Pålsson A-C., Industrial environmental information management for technical systems, Journal of Cleaner Production, 9, pp 429-435, 2001

5.1.1 Brief introduction to the project In 1993 LCA practitioners in Swedish academia and industry were in need of a good information structure for LCA databases and LCA data exchange. The work started with modelling an LCI database format for storage and exchange of LCI data. This first pilot work was performed as a master’s thesis project performed by the author (Carlson, 1994). It produced the information structure presented in figure A.1 in Annex A. This work fed directly into a governmentally supported Nordic project17 with participants from academia, industry and research institutes. The project produced a specification of a relational database format named SPINE18 (Carlson, Löfgren, Steen, 1995), presented in figure A.2 in Annex A. With the SPINE format as a foundation, the Swedish national LCI database was established by CPM (see section 1.1). The establishment of the database started in May 1996. During the following months and years the activities in the project spun off several new research, development, testing and implementation projects (Carlson, Pålsson, 1998). Before users could add or retrieve data at the common database much work was needed to describe the practical implications of the information structure of the SPINE format. The work also initiated deepened discussions about data quality of LCI data, as well as initiated international standardization of a format for LCA data. To meet the need for a complete LCA data format, rather than only an LCI data format (see section 3.3.3), a complementary format for impact assessment was developed between 1997 and 1998. The work was done in parallel with the finalisation of the ISO standard for LCA impact assessment (ISO, 2000) (see also section 3.5.2). Today there is an ISO technical specification, Environmental management – Life cycle assessment - Data documentation format ISO/TS 14048 (ISO, 2002) partly based on many of the experiences from the development and industrial testing of the SPINE format. New tests are still in 2006 being performed in current CPM projects. In the following sections practical details of the work will be described. 17 18

The NEP project, financed by the Nordic Council of Ministers SPINE: Sustainable Product Information Network for the Environment

31


5.1.2 The life cycle assessment information structuring 5.1.2.1 The ad hoc approach Development of the SPINE format was divided into three distinct methodological stages. The first stage was performed largely on the basis of analysis of the concepts, meaning and logic of the language and conceptual models used by LCA experts, acquired through interviews and literature (Carlson, 1994), and based on the database design principles described by Elmasri and Navathe (Elmasri, Navathe, 1994). This first stage was a limited and well-defined task, performed intensively in a few months. The result was a data model in the form of a relational database structure. During the second stage the resulting data model was analysed from the perspective of e.g. symmetric properties19, redundant structures20 and model analogies21. The work was an isolated technical task performed by modelling experts. During the project many detailed decisions were taken regarding the resulting structure. The work resulted in a simpler model than after the first stage, with fewer concepts and a clearer logic (Carlson, Löfgren, Steen, 1995). The third stage was to interpret, test and practically use the resulting model, including its simplifications and to develop an understanding of how the model was related to the intended physical reality of LCA in practice (Carlson, Pålsson, 1998). This stage was clearly defined in the beginning, as the actual implementation of a database for use in industry and academia. But the work has then been stretched over many years, in parallel with that the number of users and stakeholders of LCA information structuring has increased. This stage has been a strong driver of the overall research quests of this thesis. As already described above, the division into three stages was not strictly intentional at the time. At a large scale it was propelled by the practical establishment of the Swedish national LCI database within CPM and by the international ISO standardisation of the LCI data format ISO/TS 14048. And the work was also seen as a case study of the interdisciplinary theories that originally inspired the approach of the work.

5.1.2.2 The stages of information structuring STAGE 1, CATCHING THE CONCEPTS, TERMS AND LOGIC The first stage of the work was strongly guided by a physicist’s scientific view, the strict school-book principles for modelling databases (Elmasri, Navathe, 1994), together with a layman’s interest in cognitive science and experiences from industrial quality management (see also section 4.2). Figure 14 shows an early conceptual overview of LCA, as interpreted for the initial concept modelling. An LCA study is performed by modelling the isolation of a Technical system from its physical environment, and to investigate its physical exchange (Flows: Resources from nature, Emissions) with the natural Environmental system. The Technical system consists of included linked technical systems. All physical flows of the studied system originate from these included technical systems. An LCA information structure needs to be designed to describe these items, their 19

Example, identifying if similarities between different concepts may be interpreted as different types, sorts or instantiations of the same higher level concept. 20 Example, identifying whether the same concept or terms has been identified with different names. 21 Example, considering whether life cycle models resemble electric circuit models or organizational charts.

32


relationships, properties and this methodology, in a flexible way, so that information remains transparent during an LCA study. Environmental system

Technical system being studied

Other technical systems

Flows: Resources from nature

Waste

Emissions

Products

Figure 14. The conceptual model of the relationships between the systems of interest in an LCA (Carlson, Löfgren, Steen, Tillman,1998).

From interviews with LCA experts and from literature studies, it was evident that data quality was a very important aspect of LCA. All parts of LCA, from data collection to review and interpretation of results were subject to critique and scrutiny. The concept of LCA data quality had a complex and vague structure, with general references to undefined statistics and large faith in peer review. The first attempt to support data quality through information structuring was made by attempting to restrict reuse of data. An LCA is produced from data acquired in some earlier situation. And those data often again originates from even earlier LCA studies etc., in a long chronological sequence, so that an LCA may be said to ‘consist of’ earlier LCA data (see Figure 15). The idea with restricting reuse of data was intended to insist that users should be aware of when data were reused over and over again, to minimise the risk of unlimited error propagation. This approach was soon abandoned in favour of facilitating data review by the aid of the data transparency, enabled by advanced documentation possibilities. LCA data quality issues were also included by structuring for statistical parameters like standard deviation on the numerical data.

33


Figure 15. Model describing how error propagates when originally retrieved data (black circle) are reused and aggregated in LCA studies over and over again (white circles) (Carlson, 1994).

STAGE 2, CONCEPT MODELLING The second stage of the structuring of information for LCA data was guided by focusing on the economic life cycle of a future LCA information system (section 3.2.3). It was expected that data acquisition would need to be coordinated with other environmental reporting, such as governmental reporting or reporting for environmental management systems. The focus was on producing a small model without redundancy22 and without unnecessary complexity23. This was done to facilitate future data quality management (Liker, 2004), and to prevent from making errors (Shimbun, 1988) (Nikkan, 1988). These decisions were intended to integrate continuous improvement already in the design of the concept model of the intended information system. Modelling The Concepts Of Life Cycle Inventory During the second stage potentially inefficient modelling in LCA was recognized. Transports and processes were handled as different entities by LCA practitioners, but they were described almost identically. The same was true for resources, emissions, waste and products. In LCA calculations these concepts were treated differently and were therefore also modelled differently in available software and data sheets. To rationalise and to produce a minimal model these examples were noted as candidates for being merged into more general concepts, like ‘General process’ and ‘In- or outflow’. In the following the different modelling decisions and distinctions will be presented in the form of concept models, as was introduced by the example in section 3.5.2. Activity

Flow

Figure 16. The concept model of the concepts Activity and Flow

The concept that represents general processes was named Activity and the concept that represents in- or outflow was named Flow. The concept model in Figure 22

In information structures redundancy means that the same data is stored in more than one place, which introduces the risk of missing to update at both places and to therefore leave the system in an incorrect state. 23 Information structures with high complexity may have many dependencies, and may therefore be difficult to read and interpret. Complexity may introduce errors when programming or handling data.

34


16 states that Activity has Flow and that an Activity must have been defined before a Flow can be created. Flow cannot exist independent of Activity. The information used to describe Flow and Activity will be partly explained by the rest of the concept model, and will also be presented as an example after the presentation of the model.

ObjectOfStudy

Inventory

Activity

Componentship

Figure 17. Concept model describing the technical system studied in LCA.

An Activity is a model of some real world occurrence, such as a production plant, a transport route, or a partial life cycle of a product. The concept of this real world occurrence was given the name ObjectOfStudy (see figure 17). One may produce different models of any such ObjectOfStudy. The concept of such model was given the name Inventory. The concept model states that first one needs to define an ObjectOfStudy, and then one can define an Inventory (the dotted arrow24) onto which one may define an Activity. The three concepts together constitute a technical system as presented in Figure 14. The fact that technical systems in themselves include technical systems is modelled by the concept Componentship. One arrow relationship between Componentship and Activity refers to that one Activity is the host of the other, and the other arrow refers to that the other Activity is included with the host25. An Activity can have an unlimited number of other Activity within itself. Activity

Componentship

Flow

FlowConnection


Figure 14 shows that any two or more Activity within an Activity may be connected with each other via an in and an out Flow. The concept model in Figure 18 shows that this connecting concept is named FlowConnection. A FlowConnection is defined by the two Flow that are connected and the two Componentship-related Activity that are connected by these Flow. Any two Activity cannot be connected, only those that are included by the same host Activity.

24

Unfortunately SPINE is not modelled with the dotted arrow, but with the arrow between Activity and ObjectOfStudy. This is an error that was made during the modelling discussions. 25 The two arrows could have been denoted by cardinality Componentship Activity relationship of 2n:0, m, which would mean that Componentship to Activity relations always needs to be paired, from the perspective of Componentship, but not from the perspective of Activity. The author decided to exclude cardinality notations, and to instead use the simpler notations of only logical functional dependencies.

35


QMetaData

Activity

FlowType

Flow

FlowProperty

Substance

SubstanceProperty

Figure 19. Model of the concept Flow.

Figure 19 shows that the concept Flow is defined in terms of whether it is an emission, a natural resource, a product or some other relevant type of Flow. The different types of Flow are listed in the concept FlowType. Flow is also defined in terms of the matter which flows. The matter is described in the concept Substance. For the purpose of environmental assessment of a specific flow it is sometimes valuable to know more about the substance, like its temperature, price or other properties. The concept model allows some properties to be stably related to the Substance, such as density or other chemical or physical properties. The concept of stable substance properties is named SubstanceProperties, and the concept of properties related to a specific flow of an Activity is named FlowProperty. As described in the previous section, the model was designed to support LCA data quality management by facilitating data review. The concept QMetaData was introduced for this purpose. It allows for detailed qualitative meta data for the quantitative data of each Flow, as well as for all quantitative data of an Activity.

36


Figure 20. Screenshots of the software tool SPINE@CPM Data Tool, developed by the author for a relational database implementation of the concept model described here.

Figure 20 shows how some of the data based on the concept model described here is presented to users through a software developed by the author of this thesis. The different data fields that constitute the concepts ObjectOfStudy, Activity and General and Specific QMetaData are partly shown in the figure. Annex B presents the architecture and user interfaces of the Swedish national LCI database network based on this information structure. Modelling The Concepts Of Impact Assessment The modelling of a information structure for impact assessment data was performed when the LCA impact assessment standard ISO 14042 was being developed within ISO (ISO, 2000). The draft concepts and the logic were clarified in the ISO consensus process, and it became possible to construct a good information structure based on this (see also the example in section 3.5.2). The following text and figures presents the resulting concept model. The part which describes indicators and weighting is presented as the detailed example in section 3.5.2, and is therefore not again described here. A full and detailed description of all concept models are provided in the report Documentation of environmental impact assessment, compatible with SPINE and ISO/TS 14048 (Carlson, Pålsson, 2002).

37


Activity

Flow

CharacterisationMethod

CharacterisationParameterType

CharacterisationParameter

CategoryIndicator

Figure 21. Concept model of LCA characterization.

Figure 21 presents the concept model of characterisation in LCA. Note that the concepts Activity and Flow were already defined in the previous section, and that Category indicator was defined in section 3.5.2. A characterisation parameter describes the quantitative impact from a flow on a category indicator. Since this quantitative impact is strongly dependent of how the environmental cause-effect chain has been modelled, a characterisation method needs to be documented for each characterisation parameter. A characterisation parameter is most often defined as a single value that should be multiplied with the quantity of the flow. But other mathematical forms are allowed by the model. The characterization parameter may be made up from calculation of more than one quantitative value, to form a nonlinear relationship that is defined by the method. Each such value is named in the list CharacterisationParameterType.

Activity

Substance

Flow

SubstanceProperty XOR Aspect

CharacterisationParameter Figure 22. Model of concept of Aspect.

The concept model of Figure 22 was designed later than in the project described here. But it is introduced here since it is most naturally explained in this context. The concept model of characterization presented in Figure 21 states that a characterization parameter describes the impact from a flow on a category indicator. The concept Aspect generalizes this into that a characterization parameter may also describe the impact from a substance, without assigning it to a Flow of an Activity. This is suitable when for example assessing the environmental properties of materials of a product, or when making assessments of risks from use of substances. This will be presented in more detail when describing the concept model for design for environment in section 5.2.

38


Activity

Substance

Flow


ImpactAssessmentMethod



ImpactCategory


CategoryIndicator

ImpactIndicationPrinciple

WeightingFactor

WeightingMethod

Figure 23. Concept model of LCA impact assessment.

In Figure 23 all partial models of LCA impact assessment are put together into one complete concept model for LCA impact assessment. This model is unique, in that there is no other known complete concept model for LCA impact assessment that transparently follows all steps of the LCA impact assessment standard ISO 14042. ImpactAssessmentMethod combines weighting factors and characterization parameters in a systematic way to result in a useful package of impact assessment data for use in LCA studies. The concept model in Figure 23 clearly shows how category indicator is needed to determine the characterisation parameter, and how the characterisation parameter in turn determines which flows or substances that are significant. It is economically crucial that it is only those flows and substances about which it is meaningful to collect data. If there is no identifiable relationship between how a flow or a substance contributes to the environmental impact of a category indicator, it is uneconomical to collect data about these flows or substances. This insight hints that LCA database build-up is uneconomical, unless the intended impact assessment methods are at least vaguely decided to begin with. The model has been implemented as relational databases and works well. It has been successfully implemented and tested in for example the software system WWLCAW (World Wide LCA Workshop, 2006) since the year 2000, from which the picture in Figure 24 is taken.

39


Figure 24. The software system WWLCAW (World Wide LCA Workshop) LCA impact assessment web documentation tool. The snapshot shows a characterisation model for Benzene impact on crop, from the EPS method (Steen, 1999).

Figure 24 shows a screenshot from the web tool WWLCAW, which is built on a relational database implementation of a combination of both the concept model for LCA impact assessment described here, as well as on the LCA inventory part described in the previous sections26.

STAGE 3, CONCEPTUAL AND INTERDISCIPLINARY ONTOLOGY MODELLING During the third stage of the information structuring, the concept models were tested together with users in practical situations, to ensure that the concept models actually describe the same view of the reality as the methodology of LCA. The conclusion was that the concept models largely satisfied the user requirements. The fact that LCA data documentation was now expressed as different data fields in software and at web pages gave mixed feelings. The users of LCA data who acquired data from the databases highly appreciated the orderly structured and easily reviewed documentation, but the LCA practitioners who entered data into the systems became aware of their actual lack of documented data and considered the new data fields as a new obstacle for data collection. This asymmetry between data users 26

The software was designed by the author of this thesis, but was programmed by Marcus Carlson between 1999 and 2000.

40


and documenters was, and still is, a consequence of the fact that LCA training and discussion focuses on the methods to assess the collected data, rather than on collection and documentation of the data. The high-level concept model of environmental information presented in Figure 9 in section 3.5.1 was developed to provide users with ‘the big picture’ of the information structure. The model was a simplification of the overall LCA data model presented in article II of this thesis (Bengtsson, 1997) When applying the concept model in software and databases, both industrial and academic users were addressing the quality of the resulting data. To understand the different questions and requirements from the users the structure of the dimensions of data quality of Figure 25 was produced. It shows that LCA data quality is a manydimensional question, which cannot be handled by the information structure itself. Reliability

precision credibility

transparency competence

data communication Accessibility

openness (after data acquisition) semantic

Relevance

general (suited for LCA) specific (suited for the application)

Figure 25. Dimensions of LCA data quality (Pålsson, 1999).

Figure 25 presents a structuring of the dimensions of LCA data quality, or information quality, developed to facilitate discussions and for practical and educational purposes. The three basic terms, Reliability, Accessibility and Relevance, were chosen quite arbitrarily on the basis of personal communication (Steen, 1996). To be practically useful for the task, the interpretation of the three terms were reformulated in terms that were meaningful for LCA data. • Reliability is partly a matter of numerical precision27 and partly a matter of credibility about that precision. • Accessibility regards whether a data user can access the data, such as whether the data are in such a form that they can be communicated to the data user, so that data are not hidden from the user for secrecy reasons or so that data are not written in a language or jargon that the user is not familiar with. • Relevance regards whether the data are relevant for the type of questions that the user has, and in specific whether data can exactly answer the question of the user. 27

It may be more accurate to instead of Precision here instead spell Accuracy. But for practical reasons the terminology is here consistent with the terminology in the reference (Pålsson, 1999).

41


A fuller interpretation of Figure 25 for the purpose of LCA data is given in (Pålsson, 1999). This structuring substantially enhanced the efficiency of the discussions and the practical handling of LCA information quality, but it still left the question of precision unresolved. Paper III of this thesis (Carlson, Pålsson, 1998) presents a model inspired by a combination of the ontology of classical physics, the OSI model presented in section 3.4.3 and general industrial quality management principles. The model was given the name PHASETS (PHASEs in the design of a model of a Technical System). PHASETS describes the phases needed for acquiring any numerical data item of a technical system model to be used in, for example LCA. In analogy with the OSI model PHASETS starts with the physical world at the bottom. Quantitative data originates from an explicit source, such as a measurement at a production site, a modelling tool or from an estimation made by an expert. At each consecutively higher layer the data is aggregated and reported upwards until it is used in an application for an environmental management purpose. PHASETS is presented in Figure 26.

Information aggregation

55 Communicating Communicating information information between between different different contexts contexts 44 Aggregating Aggregating models models of of technical technical systems systems 33 Synthesizing Synthesizing aa model model of of aa technical technical system system 22 Forming Forming aa frequency frequency function function 3,7 g NO from from aa set set of of sample sample values values

2

11 Sampling Sampling an an individual individual value value

3,7 g NO2

00 Defining Defining an an entity entity for for aa selected selected parameter parameter

Figure 26. The PHASETS model for establishing a physical reference between LCA and the ontology of the real world (Carlson, Pålsson, 1998).

PHASETS is a conceptual model of the relationship between an LCA study and its underlying data, or the relationship between a model of system and the analytical data on which it is based. In analogy with the OSI reference model PHASETS is also a reference model for environmental data management. By utilising the reference model one may in principle achieve arbitrarily high precision and transparency of the environmental data. The relevant cost for the quality sets the limits28. The PHASETS model was later generalised into the PHASES model (PHASES: PHASEs in the design of a model of a System) (Carlson, Pålsson, 2000). PHASES in similar ways describe the data quality handling of data about also the nature system and social system (compare with Figure 9). 28

The actual relevant costs are still to be investigated, but it may be expected to be based on risk assessments, the precautionary principle, and on willingness to pay.

42


5.2 Environmental design of products ARTICLES THAT PRESENT THIS PROJECT: IV. Carlson R., Forsberg P., Dewulf W., Ander Å., Spykman G., A Full Design for Environment (DfE) Data Model, PDT Europe 2001, April 24th-26th, 2001, pp. 129-135, 2001

5.2.1 The RAVEL project The RAVEL29 project (Dewulf et al, 2001) aimed at developing a methodology and a workbench that facilitates integration of environmental issues during the design of rail vehicles, such as trains, subway vehicles and trams. In the project plan it was stated that the methodology and the workbench should improve eco-efficiency30 of rail vehicles by 25%. This quantitative goal introduced an important challenge for the project, and it forced the participants to actively think about how to define and collect quantitative data. The huge number of materials and components needed to produce a rail vehicle also forced the participants to think about how to structure the quantitative data and other relevant information in a practical way. With the experiences from having developed the LCA information structure described in section 5.1 the work could this time be performed much more straightforwardly.

5.2.2 The DfE information structuring STAGE 1, CATCHING THE CONCEPTS, TERMS AND LOGIC A first RAVEL concept analysis and concept synthesis was performed during an intensive three day workshop at Chalmers University of Technology, where all core expertise available in the RAVEL project attended actively. The result was a draft information structure that was discussed with the attendees at the end of the third day (see Figure 27). In the rear view mirror one can see that the information structure after these three early days of intense and focussed work already included most of the important concepts and knowledge, which later was developed in more detail in the project.

29

RAVEL = RAil VEHicLe eco-efficient design. The Brite-Euram project RAVEL was founded by the European Commission. Partners and subcontractors include rail operators (Swedish SJ and Danish DSB), a rail vehicle manufacturer (Bombardier Transportation), a supplier (Woodville Polymers), universities (KU Leuven, Chalmers University of Technology, and Linköping University), and consultants (ABB Corporate Research, GEP and Traintech Engineering). The duration was from November 1998 to October 2001. 30 Eco-efficiency is a combined environmental and economic sustainability concept, meaning that a product, process or service is optimally better from both the environmental and the economic perspective.

43


4 3

Project target Knowledge system

1

2 Project

5 Target setting

Product

Design

Environment

Indicator

Figure 27. First concept model for the RAVEL information structure (Carlson, 1999).

During the workshop five major distinguishable and interdependent types of information were identified; the design project (1), the physical information that describe the material and the components of the rail vehicle in terms of the environmental properties (2), the background of guidelines and other built up knowledge (3), the environmental targets for the design (4), and the actual methods and tools needed to assess the environmental performance in terms of eco-efficiency (5). During the analysis of the initial concept model it was concluded that environmental requirements should be handled as any other rail vehicle design criteria. They should be formulated by e.g. project management, customers or by strategists at the marketing department. This conclusion also led to the understanding that the environmental performance assessment of the product should exactly match how other ordinary design criteria were expressed. It was understood that environmental assessments with unclear stakeholders and targets would not be operative, and that such would introduce extra costs to implement, maintain and use. The implication of this understanding was that the RAVEL project directed efforts to develop methodology for environmental performance indicators (EPI) of rail vehicles, including the information structures and algorithms necessary to calculate such indicators. It was understood that in the same way as other design requirements may vary over time, the EPIs would also come to vary over time. Therefore, not only the data, but also the algorithms were intended to be stored in the RAVEL information structure as data. In addition two conceptually different types assessment of environmental performance assessments were identified, property based assessment and life cycle based assessment. This means that one could perform an assessment of the physical product, either based on properties of the product structure or its materials and components (Figures 28), or based on life cycle assessments of the full product or its materials or components (Figure 29)

44


Property X

Property X

Property X

Property X

Property X

Property X

Figure 28. A conceptual component structure where each included component has a set of properties declared (Gernez, 2000).

The difference between the conceptual models presented in Figures 28 and 29 suggests that to assess the environmental performance of the product structure in Figure 28, one would need only to traverse the product structure and read a quantitative or qualitative property-value for each material and component. To assess the product structure in Figure 29, for each materials or component in the product structure one need to perform a full life cycle assessment and traverse the full LCA information structure (see subchapter 5.1 and compare also with Figure 4).

Figure 29. Conceptual representation of a material or component as being described in terms of its life cycle. This figure is a combination of the figures 4 and 28.

The performance assessment method conceptually presented in Figure 29 requires well-established supply chain relationships for it to be meaningful. The costs for acquiring LCA data for each component manually without well-established routines are unrealistic. However, the idea of relating components with supply chain based LCA studies came from the intended scope of application of the RAVEL workbench. See Figure 30.

Figure 30. The RAVEL methodology was intended for environmental design requirements and information exchange throughout the entire supply chain.

45


The RAVEL methodology was intended for environmental management of the design during the entire supply chain, from initial call for tender to final delivery of complete rail-vehicle. This should be facilitated by exchange of design requirements and information throughout the entire supply chain. The RAVEL workbench should be made available to all stakeholders throughout the supply chain.

STAGE 2,CONCEPT MODELLING The concept model for design for environment in the RAVEL project was based in the same principles as the LCA concept model presented in section 5.1, as well as, when appropriate, some of the same concepts. When this is the case, this will be clearly explained, and denoted in the Figures 31 to 35. The LCA concepts are written in italic and are underlined, like Substance, while the new RAVEL concepts are written in straight fonts, like Project. A full presentation of all concepts and terms used in the RAVEL information structure is available in The RAVEL Information Platform (Carlson, Forsberg, 2000). Project

Design

Substance

SubstanceProperty

RAVELEntity

Composition

CompositionProperty

EntityProperty

Figure 31. The concept model that describe a product design.

Figure 31 presents the concept model of a product design. The concept for the physical product is named RAVELEntity, and it is an exact copy of the LCA concept Substance (see figure 19 in section 5.1.2.2), like EntityProperty is an exact copy of the LCA concept SubstanceProperty. Figures 28 and 29 describe that products are analysed as component structures. The concept Composition and CompositionProperty are introduced to handle this. Composition arranges RAVELEntities in a hierarchical structure, and CompositionProperty makes it possible to specify certain properties that are valid for the composition but not for each constituent. An example is that some materials are forbidden, except for in specific applications or when safely enclosed within specific encapsulations. To keep track of the different products and component structures in the database of the RAVEL information system, each RAVELEntity may be addressed as a Design within a Project.

46


Project

ProjectTarget DesignScore

Design

RAVELEntity Figure 32. Model of concept of design targets and environmental performance results (score).

The concept model in Figure 32 introduces the two concepts ProjectTarget and DesignScore. The distinction of having the target as referring to the project and the score to relate to the design is important, since it implies that it is the responsibility of those who set up the project to also take responsibility for setting the targets, while the designers and other experts in the project have the responsibility to achieve scores that should be tested against the target. This distinction is one of the methodologically most important outcomes from the RAVEL project, since it implies that a design project to begin with needs to have an explicit eco-efficiency performance target. The consequence is that the rail-vehicle customer, i.e. the operator of rail vehicles, needs to know how to set eco-efficiency targets on the design of the vehicle. The RAVEL project developed a complete methodology for this purpose, which is based on standardised agreements on environmental performance indicators (EPI) and standardised data format and materials database between railway operators and railway manufacturers. The responsibility for developing this standardised agreement today lies in a so-called ecoprocurement board31, mutually maintained by the rail vehicle manufacturers through UNIFE32 and the rail operators UIC33. Figure 33 presents in detail how the concepts of scores and targets are handled in the RAVEL concept model.

31

Eco-procurement board website: http://www.railway-procurement.org 2006-03-01 UNIFE, the Union of the European Railway Industries 33 UIC, the International Union of Railways 32

47


Aspect



CategoryIndicator

Project

WeightingFactor

Function/Algorithm


Design

Composition

RAVELEntity

CompositionProperty

EntityProperty

Figure 33. Concept model of how environmental performance is interpreted and calculated.

Figure 33 shows that both DesignScore and ProjectTarget refer to the LCA concepts CategoryIndicator. This means that the targets and the score are interpreted as indicators, in the same sense as LCA category indicators. Figure 22 in section 5.1 presented the concept Aspect. In Figure 32 the concept Aspect relates Design with an associated environmental impact on CategoryIndicator through the concept of CharaterisationParameter. The latter is the same LCA concept as was introduced in figure 21. The box Function/Algorithm shows that the actual value for the DesignScore is calculated from the EntityProperty of each RAVELEntity in the product structure. The calculation is based on the ImpactAssessmentMethod specified by the ProjectTarget. This integration of assessment of environmental design performance and LCA impact assessment has proven to be powerful beyond product design, and has after the RAVEL project were finished also been successfully used for environmental management systems (Policy controlled environmental management work) (Carlson, Häggström, Pålsson, 2004). Design DesignEntityScenario

Activity

Flow

Substance

RAVELEntity

Figure 34. Concept model of the environmental properties of a material or component, in terms of its life cycle assessment. See also figure 29.

The concept model in Figure 34 describes the combination of the LCA concept model and the RAVEL concept model. Each component of the product is regarded as a substance that flows out from an activity in an LCA flowchart. The concept DesignEntityScenario describe complete LCIs of the RAVELEntity. This concept have not been implemented in any software, but has proven to have a strong pedagogic 48


value, when discussing LCA in relation to DfE. Product design analyses based on the method has been successfully tested and assessed in a project with Toshiba Corporation in Japan, between 2005 and 2006. Aspect


CategoryIndicator

Project

ProjectTarget


WeightingFactor

Function/Algorithm

DesignScore

Design DesignEntityScenario

Activity

Flow

Substance

SubstanceProperty

RAVELEntity

Composition

CompositionProperty

EntityProperty

Figure 35. The full concept model of RAVEL design for environment.

Figure 35 presents the combined concept model of all concepts presented in this chapter. It should be noted how the RAVEL concept model is focused around the concepts Substance and CategoryIndicator from the LCA concept model. By having developed the concept model in the described way the relationships between environmental indicators, the product design and the environmental data become logical. It makes it possible to, for example, plan the information acquisition for databases of the information system very precisely, so that only the requested information and data are acquired to the database. It needs to be stressed that these are not all the concepts in the RAVEL information platform. The presented selection is intended to give a detailed overview of how the RAVEL methodology was developed from the concept models, and how quantitative environmental assessments and knowledge about the industrial design situation was interpreted into the model. A full presentation of the information structure is provided in The RAVEL Information Platform (Carlson, Forsberg, 2000) In the following the RAVEL information system architecture and prototype will be briefly presented.

49


STAGE 3, INFORMATION STRUCTURE AND INFORMATION SYSTEM DEMO PROTOTYPE RAVEL Workbench

Agent

RAVEL Knowledge system

Design project members

Project management

RAVEL Function dictionary Target setting

LCC

Agent

Design

Measure design performance

''Stripper' Functions for measurement of performance indicators

Picasso External information source containing LCC data (process chains/aggregated systems with e.g. costs)

''Stripper' LCI/A

''Stripper'

External information source containing LCI data (process chains/aggregated systems with e.g. resource use, emissions, waste generation)

RAVEL basic database

RAVEL Data dictionary

External information source containing product design data, e.g. CAD, PDM, …

Figure 36. Overview of the scope of the RAVEL workbench (Dewulf et al, 2001).

Figure 36 describes the intended overall RAVEL workbench architecture (Dewulf et al, 2001). This architecture was designed by the author of this thesis to clarify the scope of the work and the role of the methodology in its organisational environment. It was intended to help illustrate many of the concepts of the RAVEL information structure. For example, the functions needed to calculate the design score are data stored in the RAVEL architecture since the environmental performance indicators (EPI) are flexibly defined. The principle of having flexible functions for calculating EPIs is implemented as the heart of the RAVEL methodology; when defining an EPI one shall not only specify the data needed for calculating the score, but also in detail specify the exact algorithm of the function that calculates it.

50


5.3 Environmental characterization modelling ARTICLES THAT PRESENT THIS PROJECT: V. Tivander J, Carlson R, Erixon M, Pålsson A-C., OMNIITOX Concept Model Supports Characterisation Modelling for Life Cycle Impact Assessment International Journal of Life Cycle Assessment, Vol. 9, No. 5, pp. 289-294, 2004

5.3.1 The OMNIITOX project Many types of loads from elementary flows have quite simple mechanisms from the viewpoint of LCA methodology, such as depletion of global resources (like oil) or the simple mechanistic models of global warming potential (IPCC, 2005). Somewhat less simple are the models that describe, for example, acidification34 or eutrophication35, since it is necessary to also consider the soil and water chemistry in the natural system where the impact happens. Chemical properties are different depending on location, and this is a fact that needs to be taken into account when assessing the environmental impact by characterisation models. There are large variances, ranging from highly negative effects to highly positive. Modelling toxic impact on indicators is even more complex. In addition to taking into account the properties of the load and of the natural system, one also must consider e.g. dose of the substance and sensitivity of the organisms, population or ecosystem. It is a challenge to include such very specific mechanisms into the often globally smeared steady-state generality of the LCA methodology. But this was the goal of the OMNIITOX36 project (Molander et al, 2004). The OMNIITOX project aimed at developing characterisation models to produce characterization data for toxic substances for LCA studies and at developing an information system for such models and data. The layout of the information system presented in Figure 37 was sketched by the author of this thesis in the beginning of the project. The author was responsible for the planning and the management of the information system build-up in the OMNIITOX project37.

34

Acidification: ongoing decrease in pH in soil and water, with environmental consequences both on the chemical properties of the soil and the water and on the biological reactions. 35 Eutrophication: the enrichment of an ecosystem with chemical nutrients, leading to environmental consequences such as promoting plant growth, change species composition. 36 OMNIITOX: Operational Models aNd Information tools for Industrial applications of eco/TOXicological impact assessments. EU project within the "Competitive and Sustainable Growth"Programme. 2001-2204. Intended to facilitate decision-making regarding potentially hazardous compounds by improving methods and developing information tools necessary for Life Cycle Assessment (LCA) and (Environmental) Risk Assessment (E)RA. Contractors: AB Volvo (Volvo), Sweden, The Procter & Gamble Company (P&G Europe), Belgium, Stora Enso OY (Stora Enso), Sweden, Antonio Puig S.A. (PUIG), Spain, European Chemicals Bureau (JRC-ECB), Joint Research Centre, Italy, Randa Group S.A. (Randa), Spain, Technical University of Denmark (IPT-DTU), Denmark, Leiden University (UL-CML), Netherlands, University of Stuttgart (USTUTT), Germany, Ecole Polytechnique Federale de Lausanne (EPFL), Switzerland 37

The practical implementation and detailed structuring of the information structure was performed by the systems developers M. Sc. Johan Tivander and M. Sc. Klas Geiron and the research engineers M. Sc. Maria Erixon and M. Sc. Karolina Flemström, all working in the research group IMI at Chalmers University of Technology

51


LCA-user in industry, academy, etc.

LCA-tool (not in project-scope)

The OMNIITOX-IS Easy-to-use tool to get characterisation-data

(scope and system architecture)

Experts’ tool to review data

Dynamic characterisationmodel

Tool for modelers to add dynamic characterisationmodels to the OMNIITOX Database

Different flexible modellers’ tools, e.g. MS Excel, software, etc. (not in the scope of the informationsystem).

Codes:

OMNIITOX Database

Import and quality-check (manually, to be automated as far as possible)

ERAKnowledge base (an OMNIITOX ERA Internet-portal)

R. Carlson, IMI, Chalmers, 2002

External databases with data for OMNIITOX, e.g. ECB, PhysProp, etc.

User interface Internal technics

Selected data sources for OMNIITOX Information System

General information sources with OMNIITOX-relevant information

Out of scope

Figure 37 OMNIITOX information system architecture overview.

5.3.2 The LCA characterisation information structuring STAGE 1, CATCHING THE CONCEPTS, TERMS AND LOGIC Project planning for the OMNIITOX information system was based on the experiences from the LCA and DfE information systems presented in section 5.1 and 5.2, respectively. In the plan much time was allocated for the beginning of the project, the development of the information structure. Work tasks included an early workshop for the concept modelling, and follow-up time for discussing and testing the developed concept model among the LCA, risk and toxicology experts. The first workshop produced a result similar to the first workshop results during the modelling of the information structure in RAVEL project (see Figure 27 and subchapter 5.2). Many of the core concepts and terms had been identified, and a draft structure was outlined. Figure 38 presents a conceptual model of characterization as it was understood during this first stage. Compare also with Figure 5 in section 3.3.3. After having produced the first important results from the basic conceptual model and the basic concepts, the actual concept modelling took much longer. This introduced a difficulty, but also provided better understanding of the process of concept modelling.

52


Figure 38. Conceptual model using core OMNIITOX concepts (Tivander, Carlson, Erixon, Pålsson, 2004).

STAGE 2, CONCEPT MODELLING In the OMNIITOX information system the characterisation factor (compare Figure 21 in section 5.1.2.2) is calculated from underlying laboratory and model data. The details around those underlying data and the calculations are not handled here, since they are considered outside the scope of the expertise of the author of this thesis. The author mainly contributed with the concept modelling, and Figure 39 is the high level view of the OMNIITOX concept model of a toxicity characterization model. NatureFramework Load

Mechanism

Indicator

Figure 39. High level view of the core of the OMNIITOX concept model.

Figure 39 indicates that an environmental Mechanism is defined by the Load that causes an impact on the Indicator. The NatureFramework concept describes relevant properties of the natural system where the toxic substance is released and where the indicator is indicated, and in which the mechanism is valid. The concept of Mechanism holds the parameters of the characterisation model specified by the experts in the OMNIITOX project. This concept model is conceptually similar to the characterisation model presented in Figure 21, since the characterisation model has the same semantic meaning, including both Load (Aspect in Figure 22) and Indicator (CategoryIndicator in Figure 21). But it is still different since the characterisation factor itself does not have a concept in the model but is instead the result from calculations on data from the other concepts. Hence, the two concept models are compatible and not redundant, and they may reside in for example the same database system. The concept model of Figure 21 is more efficient when performing full LCA studies, since it holds a quantitative data value, while the OMNIITOX concept model in Figure 39 is a unique efficient tool when producing characterisation parameters considering different variables, such as different substances, different regions, or different time spans. One benefit from concept modelling the information structure for the characterisation modelling is that it is easy to get an overview of the relationship

53


between the different information needs. It is especially important to notice that it is not meaningful to collect any data before the indicators are defined.

54


6 Distillation of results from the applied research ARTICLES PRESENTED IN THIS SECTION: VI. Carlson R., Erixon M., Forsberg P., Pålsson A-C., System for Integrated Business Environmental Information Management; Advances in Environmental Research, 5/4, pp. 369-375, 2001 VII. Carlson R., Learning from management of LCA data, Journal of Life Cycle Assessment, Japan, Vol. 1, No. 2, pp 102-111, 2005

6.1 Integrated industrial environmental information management systems The different concept models described in Chapter 5 influenced each other consecutively during the years of developments, and thereby also share some of the same concepts. This has led to the fact that the information systems for LCA, DfE and detailed characterization modelling in principle are compatible with each other. One practical implication of this is that they can be integrated with each other into common integrated environmental information systems for environmental management of industrial systems. The different tools can share and exchange common information and data so that, for example, LCA studies of products and components can be seamlessly used in design processes and so that characterization models produced by different experts are instantly put into use for LCA studies. And thanks to the modelling and technology choices the results are also enabled for integration in industrial business organizations. The conceptual model of LCA presented in Figure 14 in section 5.1.2.2 works also as a conceptual model of the responsibilities of an environmental management system, according to for example ISO 14001. This similarity was noted early and scientific articles (Carlson, Pålsson, 1998) (Carlson, Pålsson, 2001) and industrial projects (Svending, 2003) (Carlson, Häggström, Pålsson, 2004) (Pålsson et al, 2005) have explored how the information structures for LCA match the need for information structures in environmental management systems. Figure 4 presented in section 3.3.3 describes a conceptual model of LCA. The model was developed during the third stage, when the information structure should be understood and tested by the users. It did not exist before, and was created to aid with establishing a common understanding of the ideas behind the LCA methodology, the LCA information structures and its relationships with the responsibilities and information needs of an environmental management system. The conclusion drawn here is that since the different methodologies have the same conceptual models the matching concept models can also be made to match and work for the information needs for both methodologies.

55


Retrieving reports Aggregated systems analysis Systems analysis

5) Communication and reporting 4) Aggregation of activities 3) Relate quantitative data to activities

Statistical analysis

2) Aggregation of quantitative data

Entering data

1) Registering of quantitative value

Entering documentation

0) Definition of quantitative entity

Database(-s)

Figure 40. Schematic illustration of the IBEIM system modularisation

Figure 40 presents a figure from article VI, in which the PHASETS model presented in section 5.1.2.2 and Figure 26 has been combined with a database like, for example, the environmental database presented in Figure 13 in section 3.5.3. The figure presents how this system is integrated with the appropriate organisational activities needed to produce a report of the environmental production performance, of some kind. The figure states that each phase in the PHASETS model represents a task that needs to be done to produce and compile that report. The integrated system proposes that these phases are linked through the common environmental database. There would be many benefits from such an integration of environmental information handling: • The data may be accessed by many different users, in different applications. • When new data are needed for new types of reports, the overall infrastructure is already established. • Transparency is established in the form of traceability. • Since each data handling task is clearly defined, it is also easy to apply improved quality control management systems on the data handling, hence achieving better data quality in any dimension (see also Figure 25). • Since each data handling task is well defined it is also possible to get an overview of cost drivers and cost-efficiency improvement potentials. The method of the model has been tested in industry (Pålsson, 2005), and two major weaknesses were noted: 1. Even though the runtime costs are evidently lower, there are initial investment costs which exceed the ordinary budgets of environmental management and therefore would need project financing. Investments are needed.

56


2. Companies seem to consider the system as too transparent, which is feared to maybe lead to stricter legal requirements on their environmental reporting. Therefore, in spite of cost-savings there is a contradictory lack of motivation. It is possible that the weaknesses are temporary, and that they will be overcome when costs for information handling for different responsibility and sustainability issues increase. The integration approach is also tested in practice between 2004 and 2006 in the CPM project IMPRESS (Implementation of Integrated Environmental Information Systems)38. The project simultaneously tests integration of different methods and tools together with integration of information management tools with organisational tasks. With currently 6 months left of the project period when this text is written, it seems as if the approach is conceptually feasible, and pedagogically effective. Conceptually feasible here means that an integrated prototype is being built, and that the specification seems very promising at the moment. Pedagogically effective here means that the work with integrating environmental methods and tools by integrating the information seems to be a very good way to understand how the different concepts of environmental management are linked to each other, concepts like environmental policy, environmental indicators, characterisation, auditing, performance measurements etc. The IMPRESS project will yield results which are valuable for exploitation and interesting for further industrial and academic research.

6.2 Beyond concept modelling During these projects much has been learned about not only concept modelling and information structuring for environmental management, but also about information system design, development, implementation and maintenance for environmental management of industrial systems. Some important practical experiences are: • An information structure for external responsibilities like LCA should be well-built and comply with industrial technology and quality standards. This facilitates integration with industrial data systems and (environmental) management systems. • Environmental performance indicators (EPIs) and database strategy are two sides of the same thing; there is a direct relationship between environmental indicators and choice of contents in an environmental database for environmental performance measurements. This is one of the individually most important results from this research project. When producing a new database one needs to solve interesting problems, which might lead farther than was intended by the actual database development. From the viewpoint of the new research field of industrial environmental informatics, the intensive work at the cross section between industry and academy during the establishment of the Swedish national LCA database was an ideal spot for evolution of new ideas. Development, rejection and redevelopment of tools, methodologies and a new scientific approach have been quick and have resulted in much new knowledge 38

More information: http://www.imi.chalmers.se/impress.htm

57


and understanding of how to establish information structures for environmental management of industrial systems.

6.3 Need for indicators Having developed the information structures for LCA by concept modelling, one resulting product is a set of pedagogical conceptual models of the entire field of LCA and industrial life cycle responsibility. Other results are the concept models that function as maps of how to design the information systems, including how to approach the strategy of populating the information systems with content, starting with the indicators. For example, the information structure for DfE (Figure 35 in section 5.2) provided important insights about indicators, or environmental performance indicators (EPI). In the concept model it can be seen how the different information is related. Both the project target and the design score are expressed in terms of indicators, and all environmental information about the design, its physical composition and their properties are defined on the basis of the algorithms needed to calculate the score. One can therefore see that environmental data should only be collected when they are needed to calculate such a design score. Hence, one must decide how the calculations are performed, and which material data to have in the database, before implementing any software functionality. The applied research work presented in this chapter clearly indicates the necessity of anchoring an information system in indicators for environmental management of any industrial systems. This ensures that resources for building up databases are not wasted on data that is not requested, and it ensures that only those functionalities that are needed are implemented in the information system. It also ensures that users understand and appreciates the output from the information system, since it was requested to begin with. Collecting data without clear indicator objectives is natural for building up knowledge and the strategic intelligence of an organisation, but not for the information systems that shall guide immediate decisions in a productive manner (see also section 3.2.2).

6.4 General principles from experiences 6.4.1 Economy of the information system life cycle The experiences from over ten years of developing, maintaining and implementing the SPINE format for LCA information (section 5.1) show that the information structure works well. It is easy to develop quite advanced software for different types of LCA studies. Some improvements have been made over the years, but the format has mainly proven to be ahead of its time. But time is catching up. The ISO technical specification ISO/TS 14048:2002 Environmental management – Life cycle assessment - Data documentation format (ISO, 2002) was developed with many influences from the concept model developed for the SPINE format. The format has been an ISO Technical Specification (ISO/TS) since 2001, and in September 2005 it was again accepted as an ISO/TS until 2008. The RAVEL information structure presented in section 5.2 also has shown strengths by the time. The fact that RAVEL and REPID39 (REPID web-page, 2005) are on the agenda of the industry indicates the expected business values of the work 39

REPID: Rail sector framework and tools for standardising and improving usability of Environmental Performance Indicators and Data formats

58


done. The software has been redesigned in different versions for different purposes, first in the RAVEL follow-up project REPID, by the consultancy firm Semcon, and then within different CPM projects at Chalmers University of Technology. The REPID project successfully finalised a materials database, software for exchange of data between CAD (Computer Aided Design) tools and the REPID DfE assessment tool. In addition, the international rail industry eco-procurement board was set up to standardise indicators for the rail industry. Many resources have also been invested in real industrial standardisation efforts. Hence, the RAVEL results are practiced in the rail industry. The RAVEL information structure and methodology is also currently (20052006) being tested practically outside the rail industry, in a CPM project together with the companies ITT Flygt and IKEA. There is also a separate project between the Industrial Environmental Informatics (IMI) group at Chalmers as one partner and the Power and Industrial Systems R&D Center at Toshiba Corporation as the other partner. In this project the RAVEL EPI methodology will be benchmarked to Toshiba’s eco-efficiency method Factor T (Toshiba website on Factor T, 2006). Results are due in the first half of 2006. During this applied research long term economical thinking has been a basis of the concept modelling and the information system design. Considering long term economy rather than quick development is experienced as one important principle for succeeding with building information structures for environmental information.

6.4.2 Cognitive sciences Database modelling is necessarily based on availability of concepts, terms and logic of the intended user domain, and of the procedures and logic of the work processes described and performed by the users. When these are identified, database design is a straightforward engineering task. However, literature in data modelling provides little guidance about how to actually acquire the concepts. There are practical recommendations to use techniques like prototyping, interviews and workshops with different types of concept modelling. These are valuable techniques, but without a theoretical understanding they may seem subjective, and may be arbitrarily applied and interpreted. One of the benefits from designing information structures using concept models and functional dependencies is that the databases and information become easy to adapt to new applications in the same or similar domains. If the concepts and their relations are correct in the first place, they will remain correct for a wide range of related applications. For example: • LCA are equally valuable for environmental management systems, • environmental product design of rail vehicles are equally valuable for any other product category, • including toxicity in LCA is equally valuable and useful for including any other environmental characterization modelling in LCA. It was noted quite soon when applying the LCA information structure in industry that users learn quickly how to think when applying the structure, both for advanced LCA studies and for different simplifying LCA tools. Similar experiences were noted when developing the RAVEL methodology and the RAVEL information structure. Users and other stakeholders considered them easy to understand and easy to apply. However, during the concept modelling in the OMNIITOX project, the term indicator was introduced as a concept. But the term indicator did not occur in this 59


way in the language of the toxicity and environmental risk experts. They did not make any apparent references to any such physical item. In fact, they argued against the idea of the concept Indicator. Therefore the concept model and the information system were not fully accepted by the intended users, and one may therefore say that the OMNIITOX concept model failed its intended purpose. Experiences from this applied research seem to suggest that one principle for succeeding with information structuring for environmental management of industrial systems, is that the intended users and stakeholders perceives the information structure as a tool that matches their own idea of the reality they work with. The relationship between the languages used to describe and understand the conceptual models of the user application, the meaning of the concept models and the definition and understanding of (environmental performance, condition or category) indicators are evident in this work. Therefore the relationship with the disciplines of cognitive sciences was extracted as a basic principle for information structuring in the interdisciplinary domain of environmental science for environmental management. The perspective of cognitive science needs to be applied to ensure user relevance.

6.4.3 Physical reality When applying the LCA information structure it was noticed that there was a lack of a concise description of how LCA data relate to empirical data. With PHASETS (Figure 26 in section 5.1.2.2) it is in principle possible to relate an LCA study to physical reality of empirical data. The RAVEL DfE method and its information structure connect the intentions of environmental policy with the decisions where they can be made operative and where they matters. The RAVEL information structure thereby comes to represent a control system like the one described in Figure 3. It controls the physical outcome of the design process. The OMNIITOX concept model can handle any environmental characterisation model, regardless of whether it addresses toxic impact, acidification or eutrophication. It will be useful for modelling any local, regional or global impact associated with chemicals from activities in the industrial society. It is designed to flexibly handle very detailed or simple characterisation models. But the information structure is not used outside the work group where it was developed. Time is still needed before it is accepted by the intended users. Characterisation modelling in general can be modelled in similar ways. The actual data needed for toxicology modelling is not the same as the data needed for, for example, modelling acidification or eutrophication, but the detailed modelling modules of the OMNIITOX information system can be replaced. Since the information structure has proven to work practically for complex LCA characterisation modelling it will most likely be taken up in future industrial applications and in methodological research or development projects. Hence, experience from this applied research indicates that identifying the physical reality, the ontology, is an important practical principle. The relationship between the information and its corresponding ontology can turn information systems into control systems, like the one presented in Figure 3 in section 3.3.2.

6.4.4 Quality As was presented in section 4.2 total quality management (TQM) was one part of the methodological starting point of this applied research work. This gave rise to an effort to identify and implement significant quality dimensions into the results. Due to the physical nature of environmental management special considerations have been 60


given to the quality of data, and due to the industrial applications specific consideration has been given to quality of information systems. LCA is conceptually simple to understand, but it is difficult to perform due to complex information requirements. Therefore, both PHASETS and SPINE (section 5.1) aid users with systematic information handling of the documentation, quantitative data and calculated results. PHASETS was designed partly to support data quality management, and it has been used both explicitly and implicitly in different work. PHASETS was tested and implemented in the EU project CASCADE (CASCADE, 2006), as model for a procedural guideline for how to produce LCA quality data (Weidema et al, 2001). It was also tested for industrial environmental quality management in a project between CPM and the Swedish paper and pulp industry, between 1999 and 2001. That project led to development of a practical method for operative environmental information quality management (Pålsson et al, 2005), and also to the major content of a licentiate thesis on the subject (Svending, 2003).

Figure 41. Snapshot of the RAVEL prototype workbench.

The RAVEL information structure serves as an example illustrating the importance of quality of information systems for industrial applications. The RAVEL information structure (Carlson, Forsberg, 2000) was proven meaningful and practically efficient when producing the prototype of the workbench within the RAVEL project. The information structure enabled the software developers40 to produce a stable prototype software system with all major functionalities implemented. The concept model provided the necessary semantic support and the 40

The author of this thesis was leader of the system development sub-project of the RAVEL project, and was also designer of the system architecture (see Figure 36), but the programming was performed by Dr Dag Ravemark at ABB, Sweden and Ing. Gerold Spykman at GEP consultants, Germany.

61


logic for the systems development. Figure 41 provides a picture of the RAVEL prototype workbench built on the RAVEL information structure The success from developing concept models and information system with a specific emphasis on dimensions of quality, suggests that quality is an important principle when structuring information for environmental management of industrial systems.

6.4.5 Work procedure Section 5.1.2.2 is structured according to the natural work stages applied when producing the LCA information structure. The different stages were stressed in the description because each was practically distinguishable from the other and because the work process produced successful results. Also, the important failure in the process of the synthesis of the OMNIITOX concept model strongly suggested a need for a practical guideline for the work processes. Because a long time passed between the first and second versions of the OMNIITOX concept model, the many experts who had contributed with concepts and terms in the early workshop both forgot their contributions and lost track of the process. They lost interest and faith in the concept model, and they did not adopt the concept model as theirs. They regarded it as something of interest only to the information system developers. Considering the fact that the functionality of the information system successfully matches all needs and expectations of LCA, toxicity or environmental risk experts, it is very likely that the idea Indicator could easily have become a highly appreciated concept; maybe just with another name. If just the work process had been more explicit this could have been found out. It was evident that the concept model itself was not sufficient to succeed, but that it is equally necessary that the model is also adopted by its indented users. The delay with the concept model had different reasons, both project organisational and methodological, which are not elaborated here. But much was due to the lack of an explicit work process. The experience is that the same process of three stages worked when developing the information structures for LCA and for DfE, and that the result failed when not applying the same process. Therefore these experiences suggest that the process should include all the three stages, and that they should be quite closely linked in time.

62


7 Synthesizing a methodological framework 7.1 General need for methodological guidelines It has taken many years to build the different information structures presented in Chapter 5. The LCI part of the SPINE structure took about three years to develop, from initial concept analysis that started in 1993 to actual user acceptance and practical deployment in late 1996. The impact assessment part of SPINE took about three years as well, from concept analysis in 1997 to first implementations and tests with different impact assessment methods in 2000. The RAVEL and the OMNIITOX information structures each also took three years to build, test and implement. During the work much time has been spent on describing and explaining the importance of information structuring. The case studies have been scientific research projects, and therefore time for discussions has been in line with the intentions. But when building future operative information systems for, for example, governmental sustainable development or for integrated environmental information systems for globalised corporations the information structuring processes need to be more efficient and effective. Hence, there is a general need for methodological guidelines for effective and efficient operative build-up of information systems for environmental management and sustainable development. General database design principles (Cornwell, 1990) (Elmasri, Navathe, 1989) do not supply enough guidance for the practical work of identifying the concepts and terms of sustainable development. Hence, in spite of the venerable intention to support global sustainable development, many environmental information systems are instead individually suboptimal, incompatible with each other, difficult for users to comprehend, and uneconomical. Examples of information systems and structures that are much used and important, but that may also serve to exemplify a lack of common principles are IUCLID (International Uniform ChemicaL Information Database) (ESIS: European chemical Substances Information System, 2005) maintained at European Chemicals Bureau (ECB) at the Joint Research Center (JRC) (ECB’s website, 2005), the automobile industry’s materials database International Material Data System (IMDS) (International Material Data System, 2005), and the SPOLD data format (Singhofen et al, 1996) for life cycle assessment (LCA) data. These three systems represent different types of issues that may be addressed from the viewpoint of a need for a common framework. However, it needs to be admitted that each of the systems are here criticised from a quite narrow viewpoint. A closer assessment of each system than what has been possible here may reveal that this specific viewpoint is not applicable on each specific information system. The following reasoning anyway exemplifies how a methodological framework could be applied, if the information systems were governed by a common agenda and a policy that aimed towards sustainable development. The IUCLID database contains toxic risk information for the European chemicals regulation system. According to an analysis by Erixon and Flemström (Erixon, Flemström, 2005) the IUCLID database lacks a consistent structure and clear concepts. Review and maintenance is difficult and data acquisition is very difficult. These weaknesses are potentially critical if the same lack of principles is applied when moving into the new chemical regulation system REACH (Registration, Evaluation and Authorisation of CHemicals) (European Commission, 2001). The difficulties with quality review and data accessibility may lead to severe problems 63


concerning chemical risk and public health and security. This exemplifies the fact that environmental information structures without concept models and consistency may be dangerous. The IMDS is set up to facilitate the recycling of old cars in the future. All suppliers of the participating automotive industry41 enter material data about the components that they supply to the cars. This is done to facilitate future recycling of the vehicles. The possibility is hypothetical, but if the automotive industry plans to adopt sustainable development by taking life cycle responsibility of their supply chains, the IMDS system would have been better designed if the life cycle costs were considered from the beginning. Substantial costs are needed to develop the system towards life cycle responsibility. The SPOLD format has stood as a model for many questionnaires, databases and software formats. The SPOLD format was developed without any concept analysis or concept modelling. The format is largely an unstructured collection of words that prescribe some important elements of LCA and issues important to some specific applications of LCA. There are no principles for structuring or interpretation of the elements, and many elements are ill-defined (Erixon, Ågren, 1998). Today much proactive environmental work is based on life cycle responsibility, while many of the important databases are based on the SPOLD format. This might change as the ISO/TS 14048 Data documentation format (ISO, 2002) is disseminated. Other examples of the need for a framework for information structuring are apparent when developing different simplified LCA approaches. Most simplified LCA tools relate to how easy it is to perform the LCA calculations, but considering that the major problem with LCA is to acquire data and to interpret the results it could be that the simplifications should have been made on data acquisition and on simplifying interpretation of results instead. Examples of the latter are so called environmental communication tools, where simple interface replaces both factual and meaningful. Simple graphical representation replaces simplicity in interpretation of actual complexity. A curious example is the BASF tool (Saling, 2002) presented as explicitly not being based on strategies for information handling or information improvements.42 By not understanding the relationship between the value of the tool for credible communication and its underlying information, the representative of the company and the tool vehemently declined any plans for data quality or data improvements. Obviously the representative was not aware of the value of having a strategy for continuous quality improvement of the environmental information, and of the necessity for having a credible information structure behind that simple user interface. This serves to suggest that a publicly understood methodological framework for information structuring is needed.

41

BMW, DaimlerChrysler, Fiat, Ford, Fuji Heavy Industries, General Motors, Hyundai, Isuzu, Mazda, Mitsubishi, Nissan, Nissan Diesel, Porsche, Suzuki, Toyota, Volkswagen, Volvo 42 Andreas Kicherer, representing the BASF eco-efficiency communication method and tool at Chalmers University of Technology in Göteborg 9th of November 2004.

64


7.2 Outlining the framework

Stages

Framework

Economic life cycle Principles

In: • Application • Environmental science • Computer science

1. Analyse concepts 2. Synthesise concept model 3. Establish ontology

Cognitive sciences Physical reality

Out: Information structure for meaningful facts for industrial environmental management and intelligence

Quality Limitations Environmental aspects Cybernetic control systems Economic restrictions apply

Figure 42. The framework of principles and work stages for developing information structures for environmental management of industrial systems.

Figure 42 provides an outline of the methodological framework developed by synthesising experiences from the applied research presented in Chapter 5. The framework rests on four practically derived principles (see sections 6.4.16.4.4): • Economic life cycle: the information structure shall be developed, maintained and used, as well as redesigned etc. for long term economic rationale. • Cognitive sciences: the information system shall be built with the viewpoint of users as information processors and with perception guided by mental constructs. • Physical reality: establish a reference with the physical world, and establish a logic of the model of the information structure that imitates the logic of the physical world as it is perceived by the intended users, i.e. the conceptual model. • Quality: identify quality dimensions needed by the information structure to define a pragmatic trade-off between different quality dimensions and economics. The framework also prescribes a practical work-process of three stages (see section 6.4.5): 1. Analysis of the concepts used by the users with environmental expertise. 2. Synthesis of a logical, economically rational, and physically meaningful concept model (draft information structure). 3. Iterative feedback of results with the intended user domain, to establish coherence between the meaning of the concept model and the ontology of the users, communicated by concept models, conceptual models and practical use of the information structure. Variants of these three stages are part of most data modelling principles, but in combination with the four principles above they are anchored in relevant theories 65


within the application domain of environmental management and sustainable development. The framework is built on the conclusions from the practical experience that successful information structures for environmental management of industrial systems are developed from concept models. These concept models are synthesized from the language of the intended users or representative experts when they describe the physical reality that they study, or the conceptual models that they use in their work. This technique establishes the information structure on the ontology of the application domain, and thereby enables control systems that connect management activities with changes in the physical reality. When designing an information structure in this way, an interdisciplinary blend of competency is needed, including knowledge and skill in environmental sciences and in software engineering43 (the “In”-side of figure 42). The framework of principles can support an interdisciplinary group of competencies to efficiently and effectively develop information structures for environmental databases, reporting formats, questionnaires or data exchange files. The results from having applied this framework (the “Out”-side of figure 42) will have a number of benefits not achieved with ad hoc approaches, such as: • The information structures will be aligned with the economic significance of the environmental aspects in the organization in which they are to be used. • The data and information that environmental information systems need as input and produce as results make sense to the intended human users (i.e. not only appropriate for computer interpretation). • The environmental information based on these principles will as closely as possible represent the natural environment in the real world. In other words, information in the information system will reflect facts of the physical world. • It will be possible for owners and users of the resulting information system to decide on a sufficient and arbitrary information quality, based on conscious understanding of the relevant quality dimensions of the information handled in the system. Quality management will be built in to the design of the fundament of the information systems. • The information will support decisions by communicating with its users in terms of well-defined and understood indicators, which can match the PHASETS model (Figure 26 section 5.1.2.2) from top to bottom, or from control signal to currents status of the cybernetic control model as presented in Figure 3 in section 3.3.2. The following sections will give a more detailed presentation of each principle of the outlined framework.

7.2.1 Principle 1: The economic life cycle Environmental work is a necessary part of business, hence needs to be scrutinised for economic rationality and efficiency improvements over its life cycle. Data acquisition is a major cost driver of industrial environmental information systems (Sterner, 2003). Aspects of the information structure that bring down costs for data acquisition improve the economic life cycle of information systems. 43

Data modeling and structuring, such as relational database modeling and object modeling.

66


Mismatch between user expectations and the actual information produced by the information system are identified in many different studies, such as (Erixon, Ågren, 1998) (Johnson, 2003) (Medin, 2001) (Taprantzi, 2001). Information systems that do not produce what they are aimed to produce are not economical. Hence, one aspect of the economic life cycle of the information system is that the information structure shall improve the process of matching information to user expectations. Economy of the life cycle of environmental information systems need to be built in to the information systems through proactive analysis of cost drivers and potentials for efficiency improvements. Different approaches for data acquisition for general business intelligence are described in (Raisinghani, 2004) and experiences with environmental information are provided in (Carlson et al, 2005) (Carlson, Dewulf, Karlson, 2001). It is necessary to acquire knowledge about possible future use scenarios of the environmental information system (Epstein, 1996) (Eriksson, 1989) (Hawryszkiewycz, 1994) and implicit future requirements on the information structures. It is also important early-on to seek to identify the core indicators that are useful to the users, in the form of a logic concept model that are significant with regards to the conceptual models of the user domain. This will both reduce costs for data acquisition and for user training.

7.2.2 Principle 2: The cognitive science perspective Environmental issues are known to people as information about, for example, environmental risks (KemI, 1999), potential environmental damage (Goedkoop, 2000), or physical environmental impact models (IPCC, 2005). Environmental facts are communicated through information media and are presented in the form of figures, pictures or analogies. The scientific field of the environmental sciences is still indistinct, with significant complexity, paradox, logical gaps, overlapping meaning, and poorly defined concepts. In terms of Kuhn this may be because of a lack of welldefined paradigms (Kuhn, 1992) or, in terms of de Mey because of a lack of mental constructs (de Mey, 1992). There are many ways in which environmental data and information therefore might be misinterpreted (Lomborg, 2001) (Johnson, 2003), which may lead to environmental mistakes such as underestimating risks44, 45. Misinterpretation is an issue not only of information itself, but of both information and of how the mind handles information through its mental models. To understand and correctly use and value the techniques for concept analysis, this research has shown that one needs to consult theories of linguistics and mental models and images, that is, cognitive science (Gardner, 1987). Those theories provide an explanation about the relationship between the language of concepts and terms and the ontology of the user domain. Findings from the area of cognitive science provide guidance. One cognitive science perspective sees the human brain as an information processor. Perception has 44

The Swiss-based company ABB bought the USA-based company Combustion Engineering, and hence became legally responsible for the workers who had been exposed to asbestos in the production at that company. The fact that there had been asbestos in the products and the production was know to ABB during the purchase. But this fact was misinterpreted as insignificant (ABB website, 2005). 45 The Swedish companies Banverket and Skanska were responsible for drilling a tunnel through a Hallandsås (a ridge of stones and rocks since the ice age), and a) they drained lakes and streams from water, and b) used a strongly toxic seal to try to stop the drainage. The work is classified as a major environmental catastrophe due to the irreparable damages to nature (Tunneling and Trenchless Construction, 2003).

67


a limited bandwidth. Short term memory has a limited capacity (Stillings, 1998). In many cases it takes more time to process more information. Information is meaningful only if it can be associated with previous concepts in the mind. Information is interpreted through mental representations in the mind of the receiver (Sobel, 2001), the mental handling of language (deFleur, 1989), drawing conclusions and problem solving (Reisberg, 2001), and pattern recognition, attention and information processing capacity (Ellis, 1989). The cognitive perspective emphasizes that in almost all areas of human endeavour, perception and communication are possible only on the basis of mental models. They account for selectivity in information processing and directionality in action (Mey, 1992). These insights from the interdisciplinary field of cognitive science explain some of the observations made in the practical work presented in Chapter 5. Changing values of the same well-understood indicator, are easier to comprehend than new information about new concepts. And concepts that are understood in terms of the conceptual model of a world are easier to understand than arbitrary concepts and terms. Linguistic analysis of the semantics of concepts (Linell, 2003) (Dirven, 2004) is practically related to theories and algorithms of the normal forms of the relational database schemas (Elmasri, Navathe, 1994). A table or a column in relational database terminology is basically a concept in the linguistics terminology. Causal logic between concepts and terms in the language of the users are directly interpreted into functional dependencies in the relational theories (see also section 3.5.2 and the concept models in chapter 5).

7.2.3 Principle 3: The physical reality Environmental responsibility concerns the real world, e.g. Earth’s natural resources, the physical conditions for life, chemical interference of biological productivity, urban air quality and growing waste dumps. But the ways in which different experts deal with this physical reality may be exemplified with a toxic heavy metal like cadmium. The different conceptual models that represent the physical reality look different depending on discipline and field of expertise. It is dealt with by environmental risk analysts (Molak, 1997) in the form of statistical models, by empirical environmental toxicologist through laboratory tests (Johnson, 2003), by life cycle practitioners as a characterisation factor (Flemström, Carlson, Erixon, 2004) (section 5.1), by an environmental manager as an important restriction policy (Waddell), and by the environmental coordinator of a design project as an environmental performance indicator (EPI) (section 5.3). Figure 3 in section 3.3.2 indicates that a control system is meaningful if it provides the controller with facts that represent real environmental performance and physical environmental impacts from the controlled system. To ensure and maintain relevant information about the reality, the information needs to be updated frequently, and the information system needs to facilitate traceability. Unfortunately little guidance for how to relate environmental management systems and tools with reality is provided in such resources as the ISO 14000 series of standards (Pålsson, Flemström, 2004). Hence, neither LCA nor EMS describes how to reference a physical reality with the information. The PHASETS model presented in section 5.1.2.2, and its generalisation PHASES (Carlson, Pålsson, 2000) are examples of how to use an information structure to relate an interdiscipline to an ontology. The author of this thesis has not 68


found any complementary or alternative models for relating system model data with empirical data and analytical methods.

7.2.4 Principle 4: Quality Feedback from environmental impact is often substantially delayed from the decisions that caused them, often years, tens of years or even longer. This means that users of environmental information will not in time see whether a specific action results in the consequence described by the information or not, so that any correcting actions can be made if the information was misleading. In situations where decisions ar based on information correct information will lead to correct actions and misleading information will lead to incorrect actions. Therefore quality needs to be taken more seriously when structuring environmental information. Since most environmental information is associated with uncertainty, lax handling of environmental information quality creates not only the risk of leading to wrong decisions, but also the risk of wrecking the credibility of all environmental information (Lomborg, 2001). People might come to ignore whatever environmental information they are presented. The quality dimensions described in section 5.1 and Figure 25, together with the PHASETS model in Figure 26 suggest the overall perspective on information quality as provided by the industrial management systems of Total Quality Management (TQM). These management systems are formulated in, for example, the ISO 9000 quality management standards (ISO, 2000), the six sigma quality management systems (Breyfogle, 2003) and the Toyota production technology (Liker, 2004). An interpretation into management of quality of information was described by Swindells in 1995 (Swindells, 1995). Similar quality management principles have also been applied for the Swedish national LCA database since 1997 (see also section 3.5.2). A research work on these quality aspects of environmental information systems is performed in parallel with this work, by the colleague46 of the author.

7.2.5 General application of the framework The practical experience from constructing the information structures as described in Chapter 5 show that there is a natural but indistinct division of tasks into three stages during the development. Experience shows that to succeed with establishing the concept model as a representation of the conceptual model of the physical reality for the users each of the three stages need to be traversed. Experience also shows that during the different stages different emphasis should be given to the four principles described in sections 7.2.1-7.2.4. The work should not seek to maximize the output in relation to any of the principles, but should seek an optimal operating point at which all four principles are considered in a balanced way. Figure 43 provides a rough graphical representation of how the emphasis is given to the different principles at the different stages of the information structuring.

46

Pålsson, A-C, Chalmers University of Technology, [email protected]

69


COGNITIVE SCIENCE STAGE 1

STAGE 2 ECONOMIC LIFE CYCLE

QUALITY STAGE 3

PHYSICAL WORLD

Figure 43. Applying the four principles of the framework during the different stages of the information structuring.

7.2.5.1 Stage 1. Analysis of concepts Stage 1. Analysis of concepts used by the users in the environmental expertise: The analysis of concepts is performed with strong emphasis on Principle 2, Cognitive science. The language of the users, which terms they use to describe their domain and their conceptual models, and how they prioritize and logically structure the concepts and terms. Quality issues, Principle 4, are also emphasised so that concepts and terms specifically addressing sensitivity, credibility and reliability are analysed with specific interest.

7.2.5.2 Stage 2. Synthesis of concept model Stage 2. Synthesis of a logical, economically rational, physically meaningful concept model (draft information structure): The synthesis of a concept model is performed with strong emphasis on both quality issues, Principle 4, and the economical life cycle of the information system, Principle 1. Quality aspects need to be considered with regard to how to facilitate correct use of the information structure, and how to prevent incorrect use and errors. Economic life cycle aspects are considered by analysis of the context of the intended future information system. New skills are likely to set new requirements on simpler use of the structure. New organisational restructuring are likely to lead to integrated information structures and data exchange. New tools and methods are likely to be developed on top of the information structure to make use of the same information structure and data.

7.2.5.3 Stage 3. Establish concept model with users Stage 3. Iterative feedback of results with the intended user domain, to establish coherence between the meaning of the concept model and the ontology of the users: The ontology of the users is established by strong emphasis on the relationship between the concept models, Principle 2, and the real world, Principle 3. All concepts, terms, logic, functions, process models, etc. are related with physically or otherwise 70


scientifically verified items, causalities, physical mechanisms etc. in the real world. Different conceptual models are tested, developed and discussed together with the users. When ambiguity or confusion is observed further analysis of the users’ concepts and terms needs to be performed. The resulting information structure should represent a causal structure of a physical real world as mutually understood conceptual models. A causal structure here means that the causal logic of the information structure should as closely as possible represent the causal physics of the conceptual model. Such causal structures are exemplified in the concept models presented in Chapter 5. For example, Figure 19 in section 5.1.2.2 states that a Flow cannot exist without an Activity, and that the substance that flows first needs to exist as a Substance. These are simple facts in the conceptual reality of LCA, and they need to be mirrored equally simple and factual in the concept model and information structure. Also, data should represent observable facts about observable states in the real world, the ontology intended by the users.

71


8 Conclusions 8.1 Quality of the research The fact that the results from this research are consistent and coherent suggests that the initial methodological starting point presented in Chapter 4 was quite relevant. The information systems based on the information structures work well, and much has been disseminated to ISO and to other practical use in industry. The methodological framework has been updated and corrected during the work. The approach to this work has been benchmarked and reflected upon in different ways over the years, and on two occasions this has been done through interview processes. In the fall of the year 2000 the author together with a colleague47 visited a number of researchers and experts in Australia, Japan and Korea, to share the experiences accrued up to that point. During these visits a handful of experts in the different countries were interviewed with regards to the then newly-developed representation of environmental management in sustainable development, shown in Figure 3 in section 3.3.2. In 2004 a number of CPM company representatives were interviewed from the viewpoint of the cognitive science approach. The results from these two series of interviews gave indications that the approach to environmental information structuring as described in this thesis was indeed both relevant and unique. The interviews concerning whether environmental information was part of a cybernetic control or feedback system (Figure 3 section 3.3.2) were interesting. Only one of the interviewees had before the interviews reflected over this. But all interviewees considered it very helpful to change their environmental viewpoint into this utilitarian perspective. Also, all interviewees noticed that they were lacking a clear view of some of the important aspect of the feedback system. All became aware of an obvious lack of feedback to decision makers during the interviews. It is not claimed that this interview result is scientifically significant (being anecdotal), but it is used here to argue that the framework at least has a pedagogic and relevant starting point in the model described by Figure 3. The interview series within the CPM companies on the subject of cognitive science was intended to give a preliminary indication as to whether cognitive science could provide input to the everyday work of environmental experts in the CPM member companies. The intention was to conduct a series of deeper studies based on these initial interviews, but changes beyond the control of the author in the focus and scoping of this research and this thesis halted the continuation of this work. A simple report and a conference article were produced (Carlson, Pålsson, 2004) (Carlson, Pålsson, 2004). The interviews were held in an informal way, and it is therefore not possible to draw explicit scientific conclusions from them. However, a few consistencies between the different interviewees in the different companies can be noted. The interviewees responded that improved understanding of how to structure environmental messages is relevant in all companies. This is especially important since environmental experts do not share language and conceptual models with representatives of the core business of the companies. Analysing environmental work from the viewpoint of how the world is perceived and described provides new and valuable questions and answers to environmental management work. The interview results hence indicated that beyond 47

Pålsson, A-C, Chalmers University of Technology, [email protected]

73


utilising the cognitive science toolbox for structuring database formats and reports the cognitive science perspective may also have a role in the general development of environmental management of industry.

8.2 Practical results The information structuring projects described in Chapter 5 exemplify how an interdisciplinary framework, like the one presented in Chapter 7, may be applicable to practical problems of environmental information handling. The synthesis of a framework requires the existence of such practical problems, since otherwise none of the principles would have been meaningful. The successful development of these information structures has led to the result that users of the information systems apply the concept models as part of their natural language. The concepts of the models are used both to describe the work itself and the tools. The relation between the design requirements and the product data has become part of the practical method, in terms of environmental performance indicators (see section 5.2). In contrast, in the case of the OMNIITOX method, the gap between the information structure modellers on the one hand and the toxicology, risk and LCA experts on the other hand led to the result that the language of the OMNIITOX information structure is not applied by its intended users (see section 5.3). In addition, the RAVEL project in particular showed that the framework may result in information structures that mediate relationships between the market, the business agreement, and the product design process, as well as with purchasing activities and supply chain management (see Figure 30 in section 5.2.1). These same experiences unfortunately also show that the framework has the capability to overshoot the target. Industry, commercial forces and the legal systems are as yet way behind the capabilities of all the information structures presented here, (i.e. the SPINE, PHASETS, PHASES, RAVEL and OMNIITOX information structures). Some practical benefits from modelling concepts drawn from conceptual models are that the information structures can then be used simultaneously for many different applications. This results from the fact that the area of environmental sciences and environmental management addresses basically the same conceptual model of how the industrial society interacts with natural environment, but with different tools and methods. By producing concept models of how mankind relates with the environment, rather than of the methods and tools, the information becomes a model of the physical world, independent of the methods and tools. In this way the same information can be viewed, manipulated and assessed in many different ways. Multiple use of the same information is probably more cost-efficient than collecting the data many times for different purposes48. One further important consequence of having developed the models using concept modelling techniques is that the role of indicators is evident. Information systems should be built for flexible choices of indicators, and indicators must be defined before any data are collected. It is obviously more cost-efficient to only collect the data that are requested. The concept models also result in easily reviewed models, where the general data quality weaknesses may be assessed, and plans for data quality management may be developed.

48

It needs to be stressed that the validity of this statement may be logically argued for, but that its validity has not been empirically proven.

74


Hence, the practical cases presented in Chapter 5 indicate that the resulting outlined methodological framework has the potential of producing practically useful results. The four principles and the three stages are crucial for information structuring. It is essential that this framework becomes commonly known, so that requirements on environmental information can be made to match the actual state of the art concerning accessibility, transparency, relevance and precision (see Figure 25 in section 5.1.2.2). The state of art is far better than what is required in most environmental reporting systems, but this fact is not yet well-known, neither by the environmental experts who compile the information nor by the information receivers in governmental, non-governmental or business organisations. There is a need to further develop practical guidelines and pedagogical material.

8.3 Final conclusion about the research project The research presented in this thesis addressed two problems: information structures for databases, communication files, reports and software for environmental management of industrial systems, and a synthesized general methodology or framework for developing such information structures: • Chapter 5 presents how the practical results have been achieved by applying concept modelling in practical information systems development. The work has produced practically useful results. • Chapter 7 presents a methodological framework for how to develop information structures for environmental management of industrial systems. This framework is synthesised from the practical research work described in Chapter 5. • An additional crucial result from the research is the insight that environmental information systems for environmental management of industrial systems need to be based on indicators, and that the indicators should be defined before functionality is implemented in the information system and before any information is collected to the databases. This is a fundamental economic result of the research. • The overall result from this work may contribute to the overall improvement of the quality of tools and methods that handle or produce information for the purpose of environmental management of industrial systems. The author sees an urgent need for a support in structuring and making available environmental information for environmental management of industrial systems, and suggests the framework proposed in Chapter 7 as a way to meet this need.

75


9 Recommended future research This thesis finalises an initial formulation of a methodological framework for structuring information for environmental management of industrial systems. But more important is that it also establishes a new foundation for research intended to facilitate sustainable development of our increasingly globalised industrial society. The need for such a new research foundation partly comes from the evolution and development of new concepts in environmental sciences and sustainable management practices, and partly from the increasing demand for data and information for these sciences and management practices. The practical framework elaborated in Chapter 7 establishes methods for developing effective and economically efficient information systems for sustainable development and social responsibility. The vision is to help different stakeholders to acquire the correct understanding of the state of the environment, sustainability and social responsibility From this thesis a number of major paths for further research can be chosen: • Refinement of the framework. • Evaluation of economic and practical consequences of improved environmental information structures. • Development of methods and tools, and integration of methods and tools for improved information structuring • Assessment of information systems based on indicators

9.1 Refinement of the framework One future research path is to refine the framework of practical guides to apply when structuring environmental information. This may be needed when developing information systems for decision support, for reports, for databases, for software or for graphical presentations. The four principles of the outlined framework presented in Chapter 7 may be further developed in interdisciplinary scientific relationships, between different disciplinary scholars and in the context of relevant applications in industrial society.

9.1.1 Economic life cycle of environmental information systems Studies and assessments of scenarios and models of the economic life cycle of existing and future environmental information structures and information systems are needed. Weaknesses, strengths, needs and experiences should be compiled into practical guidelines. Such guidelines would facilitate integration of the economic considerations in new work, and it would be of help for benchmarking and evaluation of results. Today many resources are spent because knowledge and guidelines about the economical life cycles of environmental information systems is missing.

9.1.2 Cognitive science To further develop the understanding of cognitive science in the context of the framework, the disciplines of linguistics and cognitive psychology as applied to environmental decision making need to be further investigated. It seems specifically important to study cognitive models for mental imaging and constructs to better understand how to develop concept models and conceptual models. There are 77


valuable parts of the toolbox of the linguist that are needed when describing meaning and logical relationships between concepts and terms. Relevant elements of that toolbox need to be identified for applications in environmental informatics. Improved practical understanding of cognitive psychology is needed to better understand the human limitations on handling environmental information. Environmental issues are based on complex relationships that include scientific difficulties and uncertainties as well as ethical elements. In addition environmental aspects are out of focus in most decision situations. How could information be structured to optimally fit the way the mind works? Principle 3 of the framework states that the reference to the real world is important for information systems. It would therefore be valuable to better understand how the human mind apprehends reality. It is interesting to understand how people prioritize reality when presented as complex information in comparison to simpler concepts. If, for example, there are common principal concepts that all human minds share it could be interesting to establish primary environmental ontologies based on such common concepts. Through such knowledge it could be possible to identify new strategies for handling complex information about environmental systems. It would be valuable to find methods that are both intuitively comprehended by people as well as based in an ontology with a simple physical reference.

9.1.3 Establishment of the physical reference When developing an information system, the understanding about the reference to the physical reality is established by involving experts that represents the intended user, such as toxicologists for toxicity information systems, LCA experts for LCA information systems etc. This is a sensitive dimension of the development, since the concept modelling performed by informatics specialists partly interferes with the work of the experts on the user side, and also often requires a certain measure of active criticism of the language and the formalism of the subject of the experts. Research is needed to better understand the social difficulties of this interdisciplinary work, to avoid widening disciplinary gaps instead of building interdisciplinary bridges. Improved understanding of the social aspects of the interdisciplinary collaboration will become especially important when developing information systems for social and economic aspects of sustainable development and social responsibility. Practical work with identifying and describing the different ontologies of the different relevant disciplines would have a high practical value. This would much facilitate both interpretation and application of data and results of different environmental assessment tools.

9.1.4 Quality dimensions The understanding of environmental information and data quality is of major importance. The issues of quality are not mainly in the regions of precision, which is well-handled by commonplace measurement technique, and may be guided by for example the PHASETS models (see section 5.1.2.2). The problem rather lies in quality management – e.g. how to make economic trade-off for sufficient quality for data, and how to overcome barriers of business secrecy. The latter is a major difficulty in, for example, the chemical industry. Recipes are both the core of the business and the root of the environmental problem. The quality issues must be dealt with if meaningful environmental information will also become facts that actually tell us something about our physical environment. 78


In parallel to the research behind this thesis, other research financed by CPM is being performed on the quality aspects of industrial environmental information systems. Questions regarding these aspects are therefore mainly left to that work.

9.1.5 Generalising the framework to encompass sustainable development The principles of the framework presented in Chapter 7 were both applied and developed during the different applied research projects presented in Chapter 5. The framework should not be limited to the environmental aspects of sustainable development. Currently environmental aspects constitute the only scope yet probed, analysed and experienced, but future work may show how the framework develops in the context of the new application domains of sustainability. The fact that everything changes is an argument for the conclusion that information systems need to be designed for efficient quality review and updating of data. Unless data are updated, they will not reflect changes in reality. Without information quality management, data will not be worth taking seriously. The information system needs to be developed for the needed flexibility within the control system, but it still needs to be designed as if the control situation was stable.

9.2 Research economic and practical consequences of improved environmental information structures This thesis shows how information structuring helps with improving availability and transparency of environmental information. It strives to show how this can be done in practice and by examples from case studies (Chapter 5). But since information and data gaps are built into environmental methods and tools (Pålsson, Flemström, 2004) the methodological consequences from the proposed improvements need to be investigated. Organisational and societal scenario assessments of costs, benefits, risks and avoided risks from improved environmental data need to be studied. Such consequence analysis could be used to aid political discussions and guide how to prioritize investment decisions aimed at developing environmental information systems for organisations and societies.

9.3 Indicator information system Information systems need to be established on understandable and physically relevant indicators for which information can be acquired and continuously updated. Research programmes should be established to set up a number of such key information systems. They need to be based both on models with physical ontology, like PHASETS and PHASES, and on cybernetic control models like the one shown in Figure 3. Many interdisciplinary research projects may be needed to establish such information systems so that they equally well support product design, professional and consumer purchases, business management and regional and international governance. The basic tools for the design and structuring of such information systems are provided through this thesis. Further theoretical and applied research and development are needed.

79


9.4 Development of methods and tools, and integration of methods and tools for improved information structuring A rational person might assume that cost reductions and quality improvements should drive organisations to develop economic environmental information systems based on effective and efficient information structures. But there are significant barriers. Research should be performed to identify and understand these barriers. It is recommended that this is done in parallel with practical development of real information systems, since removing barriers in theory might be very different from removing barriers in an operative industrial system. Such research may be performed with one or a few companies as case study objects. A handful of relevant environmental indicators should be selected for the study. The indicators should be chosen so that they collectively express the main environmental performance of the company. Establish information channels needed to acquire the information to quantify the indicators. Set up a quality management system for the information channels. Decide upon or develop the information structures for all the data needed. Start to acquire and collect the necessary data. While performing this data acquisition, observe when problems occur. Examples of important problems include difficulties with understanding the data, difficulties with secrecy barriers, or difficulties with interpreting the data for quantifying the indicator. Each such identified problem should then be analysed in detail, and further research and development should be addressed to find ways to overcome the problems. The result from such a detailed study should provide a basis for finding focus areas for future research. It should provide guidance on how to prioritise deepened research into the four areas of economic life cycle, cognitive sciences, physical reality and quality.

80


References ABB website, A brief history 1987- 2003, http://www.abb.com/global/abbzh/abbzh252.nsf/ 0/253055e0c2ca1ae0c125670f004a493c?OpenDocument, 2005-10-31 Axelsson U., Kumlin A-S., Olshammar M., Nydahl P., Arvidsson E-M., Olofsson C., Ström A., Thorsén E., Östman K., Axelsson H., Ekdahl Å., Strukturerad miljödatahantering inom järn- och stålindustri –Etapp 2; Miljöinformationssystem, Oktober 2004, IVL Swedish Environmental Research Institute, IVL report nr B1596, 2004 BASTA website, www.bastaonline.com, 2006-03-16 Bengtsson M., Carlson R., Molander S, Steen B., An Approach for Handling Geographical Information in Life Cycle Assessment Using a Relational Database, Journal of Hazardous Materials vol. 61 (1998) p. 67-75; Elsevier Science; Shannon, 1998 Breyfogle F. W., Implementing Six Sigma: Smarter Solutions Using Statistical Methods, Second Edition, John Wiley & Sons, 2003 Carlson R., Design and implementation of a database for use in the life cycle inventory stage of environmental life cycle assessments, Master thesis in Engineering Physics at Department of Computing Science, Chalmers University of Technology, Göteborg, Sweden, 1994 Carlson R., Erixon M., Tivander J., Flemström K., Häggström S., Erlandsson M., Establishing common primary data for environmental overview of product life cycles, Swedish EPA-report 5523, Stockholm, Sweden, 2005 Carlson R., Dewulf W., Karlson L., Basic Needs for Environmental Assessment of Rail Vehicles Chapter 7 in Dewulf W., Duflou J., Ander Å, ed. Integrating EcoEfficiency in Rail Vehicle Design, Leuven University Press, Leuven, Belgium, 2001 Carlson R., Forsberg P., The RAVEL Information Platform Model, RAVEL project report, Chalmers University of Technology, Göteborg, Sweden, 2000 Carlson R., Häggström S., Pålsson A-C., Policy controlled environmental management work - Final report, CPM-report 2004:10, Chalmers University of Technology, Göteborg, Sweden, 2004 Carlson R., Löfgren G., Steen B., SPINE, A Relation Database Structure for Life Cycle Assessment, Swedish Environmental Research Institute, IVL-Report B 1227, Göteborg, Sweden, 1995 Carlson R., Pålsson A-C., Establishment of CPM's LCA Database, Appendix V, CPM Report 1998:3, Chalmers University of Technology, Göteborg, Sweden, 1998

81


Carlson R., Erixon M., Forsberg P., Pålsson A-C., System for Integrated Business Environmental Information Management, Advances in Environmental Research, 5/4 (2001), p. 369-375, 2001 Carlson R., Pålsson A-C., Industrial environmental information management for technical systems, Journal of Cleaner Production, 9 (2001) 429-435, 2001 Carlson R., Löfgren G., Steen B., Tillman A-M.; LCI Data Modelling and a Database Design; Published in The International Journal of Life Cycle Assessment Vol. 3. No.2. pp. 106-113, 1998 Carlson R., Pålsson A-C., Maintaining Data Quality within Industrial Environmental Information Systems, 12th International Symposium 'Computer Science for Environmental Protection', p. 252-265, Band 1/Volume 1, Germany, Bremen, 1998 Carlson R., Pålsson A-C., Documentation of environmental impact assessment, compatible with SPINE and ISO/TS 14048, IMI-report 2002:1 (second edition), Chalmers University of Technology, Göteborg, Sweden, 2006 Carlson R., Pålsson A-C., Mind the environment, CPM-report 2004:4, Chalmers University of Technology, Göteborg, Sweden, 2004 Carlson R., Pålsson A-C., PHASES Information models for industrial environmental control, CPM-report 2000:4, Chalmers University of Technology, Göteborg, Sweden, 2000 Carlson R., Pålsson A-C., Understanding and environmental reporting, Presented at The Sixth International Conference on EcoBalance, Oct. 25th - 27th, Tsukuba, Japan, 2004 CASCADE, http://192.107.71.126/cascade, 2006-03-16 Celander L., Introduktion till STEP, IVF-skrift 93805, Institutet för Verkstadsteknisk Forskning, Mölndal, Sweden, 1995 Checkland P., Holwell S., Action Research: Its Nature and Validity, Systemic Practice and Action Research, Vol. 11, No. 1, 1998 Cornwell D. Computer Systems Modelling & development, A disciplined approach, Studentlitteratur, Bromley : Chartwell-Bratt, Lund 1990 ecoinvent, Data exchange format (EcoSpold), ecoinvent documents and technical specifications, http://www.ecoinvent.ch/en/index.htm, 2005-10-29 DeFleur M., Ball-Rokeach S., Theories of mass communication, fifth edition, Longman, White Plains, New York, USA, 1989 Dewulf W. (Ed.) et al, Integrating Eco-Efficiency in Rail Vehicle Design, Leuven University Press, Leuven, Belgium, 2001 82


DG Environment and DG JRC website on European Platform on Life Cycle Assessment (LCA), http://europa.eu.int/comm/environment/ipp/lca.htm, 2006-03-16 Dirven R., Verspoor M., editors, Cognitive Exploration of Language and Linguistics, Second Revised Edition, chapters 2 and 3, in Cognitive Linguistics in Practice, Vol 1., John Benjamins Publishing Company, Amsterdam/Philadelphia, 2004 EcoInvent, http://www.ecoinvent.ch/en/index.htm, 2006-03-16 Ekvall T., Baumann B., Svensson G., Rydberg T., Tillman A-M., Life Cycle Assessment: Pilot study in inventory methodology and data bases, in Product Life Cycle Assessment, - principles and methodology, Nordic Councils of Ministers, Nord, 1992 European Chemical Bureau website: http://ecb.jrc.it, 2005-10-29 Ellis H. C., Hunt R.R. Fundamentals of human Memory and Cognition, fourth edition, Wm. C. Brown Publishers, Dubuque, Iowa, USA, 1989. Elmasri R., Navathe S. B., Fundamentals of database systems, Benjamin/Cummings, Redwood City, California, USA, 1994 Epstein M. J., Measuring Corporate Environmental Performance – Best Practices for Costing and Managing an Effective Environmental Strategy, Institute of Management Accountants, Certified Management Account Program, McGraw-Hill, USA, 1996 Eriksson B. A. Systemering – från informationsbehov till informationssystem, Studentlitteratur, Lund, 1989 Erixon M., Flemström K., Data transfer from IUCLID to OMNIITOX - Contribution to Work-package 9 of the OMNIITOX Project, OMNIITOX project internal report, draft of 2005-01-27 Erixon M., Ågren S., An Assessment of the SPOLD format, CPM Report 1998:5, CPM, Chalmers University of Technology, 1998 ESIS: European chemical Substances Information System: http://ecb.jrc.it/esis/esis.php?PGM=hpv&DEPUIS=autre, 2005-10-29 European Commission (2001), Green Paper on Integrated Product Policy, http://europa.eu.int/comm/environment/ipp, 2005-10-29 Flemström K., Carlson R., Erixon M. Relationships between Life Cycle Assessment and Risk Assessment – Potentials and Obstacles, Swedish EPA report: Naturvårdsverket Rapport 5379, Sweden, 2004 Flemström K., Erixon M., Erlandsson M., Häggström S., Tivander J., Carlson R., Wiklund C., Riise E., Ågren U., Implementation of integrated environmental information systems for sustainable development, LCM 2005, 2nd International Conference on Life Cycle Management, Barcelona, September 5-7, 2005 83


Frankelius P., Omvärldsanalys, Liber ekonomi, Liber AB, Malmö, Sweden, 2001 Gardner H., The mind’s new science – A history of the cognitive revolution, Perseus Books Group: BasicBooks, 1987 GEDnet, Global Type III Environmental Product Declarations Network: Facilitating environmental information exchange, http://www.gednet.org, 2005-11-04 Gernez L. ed., Software Requirement Document, chapter 3 System architecture for the Ravel workbench, RAVEL project deliverable, Leuven, Belgium, 2000 Goedkoop M., Spriensma R., The Eco-indicator 99 - A damage oriented method for Life Cycle Impact Assessment Methodology Report, Second edition, April 17, 2000 Griesel L., de Beaufort A., Wriesberg N., Coelho-Schwirtz V.,Glavind M., van Dam A., Carlson R., Schmidt A., Schmidt W.-P., van den Berg N., Meyer U., Bretz R.,. Pajula T, McKeown P., Hochfeld C., Kreissig J.; Databases and softwares in: LifeCycle Assessment: State-of-the-Art and Research Priorities - Results of LCANET; a Concerted Action in the Environment and Climate Programme (DGXII); Helias A. Udo de Haes; Nicoline Wriesberg (editors); LCA Documents Vol 1. 1997 Hallandsås – A Saga Which Runs And Runs, Tunnelling & Trenchless Construction, World Tunneling, Nordic Focus, June, 2003, http://www.ttcmag.net/tunnel/archive/2003/jun/Hallands%20s-a%20saga.pdf, 200510-31 Hawryszkiewycz I. T., Introduction to Systems Analysis and Design, Prentice Hall, University of Technology, Sydney, 1994 Hopwood B., Mellor M., O'Brien G., Sustainable development: mapping different approaches, Sustainable Development Volume 13, Issue 1 , Pages 38 – 52, 2005 Intergovernmental Panel on Climate Change (IPCC), The IPCC Data Distribution Center, Climate Scenario Data, http://ipcc-ddc.cru.uea.ac.uk/ddc_climscen.html, 2005-10-30 Intergovernmental Panel on Climate Change (IPCC), The IPCC Data Distribution Center, Emissions Scenarios, http://ipccddc.cru.uea.ac.uk/sres/ddc_sres_emissions.html, 2005-11-12 IMDS: International Material Data System: http://www.mdsystem.com, 2005-10-29 ISO 10303:1998, Product Data Representation and Exchange (multi-part standard), International Organization of Standardization, 1998 ISO 15926-1:2004, Integration of life-cycle data for process plants including oil and gas production facilities (multi-part standard), International Organization of Standardization, 2004

84


ISO 14001:2004, Environmental management systems – Requirements with guidance for use, International Organization of Standardization, 2004 ISO 14025:2002, Environmental labels and declarations - Type III environmental declarations , International Organization of Standardization, 2002 ISO 14040:1997, Environmental management - Life cycle assessment - Principles and framework, International Organization of Standardization, 1997 ISO 14041:1998, Environmental management - Life cycle assessment - Goal and scope definition and inventory analysis, International Standards Organisation, 1998 ISO 14042:2000, Environmental management - Life cycle assessment - Life Cycle Impact Assessment, International Organization of Standardization, 2000 ISO 14043:2000, Environmental management - Life cycle assessment - Life cycle interpretation, International Standards Organisation, 2000 ISO/TS 14048:2002, Environmental management – Life cycle assessment – Data documentation format, International Standards Organisation, 2002 ISO/TR 14062:2002, Guidelines to Integrating Environmental Aspects into Product Design and Development, International Standards Organisation, 2002 ISO 9000:2000, Quality management systems - Fundamentals and vocabulary, International Standards Organisation, 2000 Johnson B. B., Chess C., Gibson G., Communicating Status and Trends in Environmental Quality: Reactions of Legislative Staff, Reporters, Activists, and Citizens, Environmental Assessment and Risk Analysis Element, Research Project Summary, State of New Jersey, Division of Science, Research and Technology, New Jersey, USA, 2003 Kemikalieinspektionen, Information om varors innehåll av farliga kemiska ämnen, Rapport nr 6/04, Stockholm, Sweden, 2004 Kemikalieinspektionen, Product information – a guide to be consulted in the work to classify and label chemical products and when preparing safety data sheets, The Swedish national chemical inspectorates, Stockholm, Sweden, 1999 Kuhn T., De vetenskapliga revolutionernas struktur, Thales, Stockholm, 1992 Lomborg B., The skeptical environmentalist: measuring the real state of the world, Cambridge University Press, Cambridge, United Kingdom, 2001 Liker J. K., The Toyota Way: 14 Management Principles from the World's Greatest Manufacturer, McGraw-Hill, 2004 Linell P., Människans språk, Gleerups utbildning AB, Klippan, Sweden, 2003 85


Medin S., Byström A-K., Larsson K., Förenklad LCA-baserad information – En intervjuundersökning bland slutkunder – ’Livscykelanalys – Vete fasen vad det kan vara’, CPM-report 2001:11, CPM, Chalmers University of Technology, 2001 Mey M. de, The cognitive paradigm: an integrated understanding of scientific Development, edition 1992, University of Chicago, Chicago, USA, 1992 Molak V., Ed., Fundamentals of Risk Analysis and Risk Management, CRC Press, Inc., Lewis Publishers, USA, 1997 Molander S. et al, OMNIITOX - Operational Life-Cycle Impact Assessment Models and Information Tools for Practitioners, pp 282-288, International Journal of Life Cycle Assessment, 9 5, 2004 Naturvårdsverket, Information om produkters miljöbelastning, Swedish EPA report 5526, Stockholm, Sweden 2005 Naturvårdsverket, Kunskap om produkters miljöpåverkan: tillgång, behov och uppbyggnad av livscykeldata, Swedish EPA Report 5229, Stockholm, Sweden, 2002 NetworQuest website, http://library.thinkquest.org/04oct/01910/osi.htm, 2006-02-21 Nikkan Kogyo Shimbun, Ltd.: Poka-Yoke: Improving Product Quality By Preventing Defects, Productivity Press, 1988 Nuij R., Rentsch C. Ryder B., Informal European IPP Network, Workshop on Product Information, Workshop Report, Bern, 17-18 January 2005, February 2005 Pré Consultants website, http://www.pre.nl/ecoinvent/default.htm, 2006-03-16 Pålsson A-C., Introduction and guide to LCA data documentation using the CPM documentation criteria and SPINE format, pp 6, Centre for Environmental Assessment of Product and Material Systems, CPM Report 1999:1, CPM, Chalmers University of Technology, 1999 Pålsson A-C., Flemström K., Gap analysis of the documents in the ISO 14000-series with regard to quality management of environmental data and information, CPMreport 2004:3, CPM, Chalmers University of Technology, 2004 Pålsson A-C, Enqvist A., Karlsson G., Loviken G., Möller Å, Nilseng AB, Nilsson C., Olsson L., Svending O. Methodology for handling forest industry environmental data – Manual, CPM-report 2005:2, CPM, Chalmers University of Technology, 2005 Raisinghani M.,Business Intelligence in the Digital Economy —Opportunities, Limitations, and Risks, Idea Group Publishing (an imprint of Idea Group Inc.), Hershey, Pennsylvania, USA, 2004 Ravemark Dag (2003), State of the art study of LCA and LCC tools, Report for the DANTES project accessible at http://www.dantes.info/Publications/publications.html 86


European Commission, REACH, White paper – Strategy for a future chemicals policy, European Commission, 2001 http://europa.eu.int/comm/environment/chemicals/reach.htm, 2005-10-29 Reisberg D., Cognition: exploring the science of mind, Second edition, pp 377-484, Reed College, W.W. Norton & Company, New York, London, 2001 REPID web-page http://www.railway-procurement.org/repid.htm, 2005-11-21 Saling P., Eco-efficiency Analysis by BASF: The Method, pp 203-218, 7 International Journal of LCA (4), 2002 Shingo S., Den nya japanska produktionsfilosofin Shigeo Shingo, bearb. för svenska förhållanden av Lars O. Sødahl, Svenska managementgruppen, Lidingö, Sweden, 1984 Singhofen A. et al., Life Cycle Inventory Data: Development of a common format, The International Journal of Life Cycle Assessment, 1996 1(3):171-178. Sobel C. P., The cognitive sciences – An interdisciplinary approach, pp 44-73, New College of Hofstra University, Mayfield Publishing Company, Mountain View, California, USA, 2001 SPOLD, the SPOLD format published in 1997, http://www.lcanet.com/spold/publ/Fileform.zip, 2006-03-16 Steen B., A Systematic Approach to Environ- mental Priority Strategies in Product Development (EPS) Version 2000 - General System Characteristics, CPM-report 1999:4, CPM, Chalmers University of Technology, 1999 Steen B., Environmental Systems Analysis, Chalmers University of Technology, Göteborg, Sweden, personal communication, 1996 Sterner T., Policy Instruments for Environmental and Natural Resource Management, Time, and Space, The World Bank & Sida, RFF (Resources For the Future) Press, USA, Washington DC, 2003 Stillings N. A., et al., Cognitive science – An Introduction 2nd edition, A Bradford Book, the MIT Press, Cambridge, Massachusetts, USA, 1998 Svending O., Industrial Management of Environmental Data - Suggested procedures for internal allocation based on stakeholder needs, Licentiate thesis, ESA report 2003:14, Environmental Systems Analysis, Chalmers University of Technology, Göteborg 2003 Swindells N., Managing the Quality of Information Products, pp 35-37, Managing Information, Aslib (The Association for Information Management), London, UK, 1995

87


Taprantzi A., A Systematic Approach for Acquiring Industrial Data and Information for Environmental Applications - Procedures for modelling the communication of environmental inquiries within an organisational structure, Department of Earth Sciences, Applied Environmental Impact Assessment, Uppsala University, Uppsala, Sweden, 2001 Tivander J, Carlson R, Erixon M, Pålsson A-C; OMNIITOX Concept Model Supports Characterisation Modelling for Life Cycle Impact Assessment Intl. Journal of LCA 9(5), 2004 Toshiba website on Factor T, http://www.toshiba.co.jp/env/en/products/factor.htm, 2006-02-26 UNEP/SETAC Life-Cycle Initiative, http://lcinitiative.unep.fr, 2006-03-23 United Nations website Agenda 21: http://www.un.org/esa/sustdev/documents/agenda21, 2005-10-29 Vriens D., Information and Communications Technology for Competitive Intelligence, Chapter 1: The Role of Information and Communication Technology in Competitive Intelligence, University of Nijmegen, The Netherlands, Idea Group Publishing, Pennsylvania, USA, 2004 Waddell C., Defining Sustainable Development: A Case Study in Environmental Communication, pp 7-11 Vol 11 in ATTW Contemporary Studies in Technical Communication, Coppola, N. W., Karis B., ed. Technical communication, deliberative rhetoric, and environmental discourse: Connections and Directions, Ablex Publishing Corporation, Stamford, Connecticut, 2000 Weidema B. P., Cappellaro F., Carlson R., Notten P., Pålsson A-C., Patyk A., Regalini E., Sacchetto F., Scalbi S., Procedural Guideline for Collection, Treatment, and Quality Documentation of LCA Data. Document LC-TG-23-001 of the CASCADE project, Copenhagen, Denmark, 2001 Wiener N.,Cybernetics: or Control and Communication in the Animal and the Machine, 2:nd edition, The M.I.T. press, Cambridge, Massachusetts, USA, 1991 http://en.wikipedia.org/wiki/OSI_model, 2005-11-15 World Business Council for Sustainable Development, http://www.wbcsd.org, 200511-04 World Wide LCA Workshop, http://workshop.imi.chalmers.se, 2006-03-13

88


Acknowledgements This work could not have been made without others. I therefore owe gratitude to many people. Some have supported me out of pure generosity, love or friendship, others as part of a mutual development or for search for solutions and answers in the same terrain as I have, yet others by providing me an interesting viewpoint, crucial criticism, reality contact or positive feedback. First of all I want to mention all you colleagues in the VINNOVA competence centre CPM. The open doors to the everyday life of industrial environmental work within many different companies, the uncountable relevance checks and the continuous generation of new practical problems offer a horn of plenty for a researcher. Without it this thesis would not have contained meaningful text. Reality drives evolution. I also wish to thank all of you who added on top of your full workdays the responsibility to help me finalise this thesis, my supervisor Professor Steve Frysinger, my examiner Professor Jan Smith, and Professor Oliver Lindqvist and Dr Dan Strömberg. Without the important contributions and support from each one of you, this thesis would not have been produced at all. Your contributions were necessary. I also first want to thank all of my colleagues in the research group IMI, who provided confidence, loyalty and hard work while I stealthed to write a thesis that describes the theoretical framework for our work. A specific thanks to Ann-Christin who have shared almost 10 years of workdays in this new field of industrial environmental informatics. It has been much hard work. Hard work makes a difference. But the largest gratitude I owe to my family, and especially to my two daughters Viktoria and Antonia who have always been there for me. I love you! To all of you mentioned here, and to all of you others who I have much to thank for but who I have not mentioned here, Thank you!

Raul Carlson, 27th of March 2006

89


This page is intentionally blank.

90

Figure A.1 The first LCI information structure, an entity-relationship diagram of the database. Carlson R., ”Design and implementation of a database for use in the life cycle inventory stage of environmental life cycle assessments”, Master thesis in Engineering Physics at Department of Computing Science, Chalmers University of Technology, Göteborg, Sweden, 1994

Annex A

A.1

A.2

Figure A.2 The second LC! information structure, a relationaral database schema of the database. Carlson R., Löfgren G., Steen B., ”SPINE, A Relation Database Structure for Life Cycle Assessment”, Swedish Environmental Research Institute, IVL-Report B 1227, Göteborg, Sweden, 1995

Annex B Information structure in information systems The concept model developed for LCA information was implemented into the SPINE relational database structure, used as basis for the development of the Swedish national LCA database. Figure 20 in section 5.1.2.2 presents a screenshot from the software tool SPINE@CPM Data Tool that was developed entirely by the author of this thesis for this project. It was used for documenting data, sharing, reporting and collecting data to the Swedish national LCA SPINE@CPM. Figure B.1 shows how the data collection and database system was set up, using the SPINE@CPM Data Tool and a web-server for data publishing.

_ Y

sending data

]

Z

data review

[ X

W

entering data

data publishing i

SPINE@CPM Data Tool

SPINE@CPM Data Tool

SPINE@CPM Data Tool

data acquisition

SPINE@CPM Gateway

public database

local SPINE

\

^

Dynamic web publishing

SPINE@CPM Master

Figure B.1. The SPINE@CPM distributed database system. (Carlson, Pålsson, 1998)

Figure B.1 shows a model of the distributed SPINE@CPM database system: 0. Data enters the system by being manually typed in to a local SPINE database by the use of SPINE@CPM Data Tool. This can be done anywhere at the CPM companies, at Chalmers or by any other data supplier. 1. The local copy of the SPINE database is supplied freely with the software. 2. Data is sent to CPM by attaching a whole database file in an e-mail. If only a few data sets are to be published from a larger database, these data can be moved from the larger database into a smaller SPINE database using the SPINE@CPM Data Tool. 3. The database sent to CPM is reviewed using the SPINE@CPM Data Tool. 4. When the review has accepted the data, it is moved into the SPINE@CPM Gateway database, a local copy of the SPINE database, used for refinishing and late reviews, before data is sent to the SPINE@CPM Master database. 5. Data entered into the SPINE@CPM Master database has been reviewed, and can be used for publishing. Any data in this database is sure to hold all SPINE@CPM data documentation and quality requirements. 6. When publishing data, data is copied from the SPINE@CPM Master database into a local SPINE database, using the SPINE@CPM Data Tool. B.1

7. The local public database will hold a subset of the total SPINE@CPM data set, and is published through Internet. a) There are a number of ways in which the Internet-published data may reach the data users' databases. The strength of the system presented her lie in its modularity. The SPINE@CPM Data Tool is standalone software. The local database shipped with the software can be copied and shipped to any other SPINE@CPM Data Tool user. And any data set from one such database can be moved to any other database, ensuring full data communication capabilities. This information system was established in the beginning of 1997 and has been in practice in Swedish industry since then. In the following a number of screenshots from the SPINE@CPM web database will be presented and explained.

Figure B.2. Search result from the SPINE@CPM database.

Figure B.2 shows how the database content is listed on a web page. The user gets an indication about what each data set represents, from the name of the object of study, the type of process assigned to the activity and the name of the flow that is classified as Product or service.

B.2

Figure B.3. Head of data sheet that present the technical system for which the data is valid. Compare ObjectOfStudy in concept model in figure 16 in section 5.1.2.2.

Figure B.3 shows the head of a data sheet as it appears on the web page. At the top of the page the user can navigate to the different parts of the document with hypertext links. The document structure starts with presenting a selection of data fields grouped as administrative information and then identifying information to increasingly technical levels.

Figure B.4. List of flow data in the SPINE@CPM database. Compare concept model of Flow in figure 19 and SPINE@CPM Data Tool screenshot in figure 20 in section 5.1.2.2.

Figure B.4 shows the list of flow data in the data sheet. The small document icons to the left of the table are hyperlinks that open up the specific QMetaData documentation for each flow. Compare with the concept model of Flow in figure 19 in section 5.1.2.2. The fields that specify a flow denotes the direction of the flow, its B.3

type (emission, resource, product etc.), the substance (compare figure 19 in section 5.1.2.2), the quantity (including statistical information if available) and its unit, a typing of the environment impacted, and information about the geographic location of the occurrence.

Figure B.5. An LCA flowchart presented on the SPINE@CPM database. Compare the concept of Componentship in figure 18 in section 5.1.2.2.

Figure B.5 presents the part of the data sheet that presents a flowchart if the technical system is based on documentation of underlying technical systems. Compare also figure 14 in section 5.1.2.2 that describe how it was intended that a technical system should consist of underlying technical systems, figure 16 that describe the Componentship concept and figure 18 in section 5.1.2.2 that describe the concept that describes how the flows are connected. The full data sheet is presented on the following pages.

B.4

Contents

Administrative SPINE

Technical System

System Boundaries

Flow Data

About Inventory

Flow Chart This data set transparently reported, including a flowchart where each process individually described

Administrative Back to Contents Finished

Y

Date Completed

1999-01-25

Copyright Availability

Public

Technical System Back to Contents Name

Production of CAN fertiliser

Functional Unit (see also Functional Unit Explanation)

1 kg of CAN fertiliser

ProcessType

Cradle to gate

Site

Europe

Sector

Materials and components

Owner

Europe

Function

The production route of CAN fertiliser produced at Hydro Agri AB in Landskrona can be seen in the aggregated activity window. For information about each separate production step, please see each included dataset. Emissions from transports, energy consumption and production of steam, district heat and electricity have been included in the system by using information and emission factors from the database in LCAiT 3.0. LCAiT 3.0 is a computer programme created by CIT Ekologik in Göteborg for practitioners of life cycle assessments. Production/consumption of steam is assumed to replace/be produced by combustion of oil (efficiency of 0.90). Oil has been chosen as fuel source, as it in terms of emissions lies between coal and natural gas. Included transports and assumptions made regarding transports are described below (transports cannot be seen in the aggregated activity window). The ammonia used in production of fertilisers in Sweden is purchased at the spot market in the Baltic Sea. This ammonia mostly originates from Russia and Poland. The ammonia has been assumed to origin from the area around Moscow. It has been assumed that the ammonia is transported by train from Moscow to the harbour in Ventspils in Latvia (1000 km, train, diesel) and then transported by boat to Landskrona (400 km, ship, bulk carrier). Dolomite used at Hydro Agri AB in Landskrona comes from Hammerfall in Northern Norway. It is transported by truck (5 km, medium truck, rural, full) to a harbour close to Bodø, from where it is transported by boat (1800 km, ship, bulk carrier) to Landskrona.

Flowchart Click on flowchart to open each data set description

B.5

System Back to Contents Functional Unit Explanation (see also: Functional Unit)

Suprasalpeter, N 28 (27,6 % N) Calcium ammonium nitrate (CAN)

Nature Boundary

The system starts with extraction of raw materials and ends at the outgoing fertiliser factory gate. Emissions from transports, production of steam, district heat, electricity and combustion of energy resources have been included by using information and emission factors from the database in LCAiT 3.0. LCAiT 3.0 is a computer programme created by CIT Ekologik in Göteborg for practitioners of life cycle assessments. Production/consumption of steam is assumed to replace/be produced by combustion of oil (efficiency of 0.90). Oil has been chosen as fuel source, as it in terms of emissions lies between coal and natural gas.

Time Boundary

The time boundary varies for each included dataset; from 1996 to 1998 and in some cases unspecified (see each dataset for further information).

Geographical Boundary

Europe.

Other Boundaries

Coatings, micronutrient and small additives have not been taken into account. Packaging of fertiliser has also been left outside the system boundary. Production and waste treatment of catalysts and production of capital goods are not included.

Allocations

See each included dataset.

Systems Expansions

Production/consumption of steam is assumed to replace/be produced by combustion of oil (efficiency of 0.90). Oil has been chosen as fuel source, as it in terms of emissions lies between coal and natural gas.

Flow Data General Activity QMetaData Flow Table and Specific QMetaData General Activity QMetaData Back to Contents DateConceived DataType

Derived, unspecified

Represents Method

Data are gathered from the official environmental report distributed by Hydro Agri AB in Landskrona, from personal communication with people working there and also from literature and reports on production of fertilisers. For further information, please see the facts under ’’Method’’ and ’’LiteratureRef’’ in each included dataset.

B.6

Emissions from combustion of the energy demanded in each included process are not included in the separate datasets, but are included in this aggregated dataset. Emissions and use of resources due to transports are also included in this dataset. The transports are discribed under ’’function’’. Data for emissions related to the consumption of energy and transports have been taken from the database in LCAiT 3.0. LCAiT 3.0 is a computer programme produced by CIT Ekologik in Göteborg for practitioners of life cycle assessments. Production/consumption of steam is assumed to replace/be produced by combustion of oil (efficiency of 0.90). Oil has been chosen as fuel source, as it in terms of emissions lies between coal and natural gas. Literature Reference

Besides references used in each included dataset, the database in LCAiT 3.0 has been used to provide data for use of resources and emissions from production and combustion of fuel and from transports. LCAiT 3.0 is a computer programme created by CIT Ekologik in Göteborg for practitioners of life cycle assessments. Please see information in each included dataset for further information.

Notes

Emissions from consumption of energy and transports are included in the outputs. Included transports are described under ’’function’’. Flow Table and Specific Meta Data Back to Flow Data Back to Contents

QMetaData Direction FlowType Substance

Quantity

Min

Max

SDev Unit

Environment Geography

Input

Natural resource

Dolomite

0.2

kg

Ground

Input

Natural resource

Hard coal

1.34

MJ

Ground

Input

Natural resource

Natural gas

10.7

MJ

Ground

Input

Natural resource

Oil, heavy fuel

1.53

MJ

Ground

Input

Refined resource

Diesel

0.253

MJ

Technosphere

Input

Refined resource

District heat

-2.06

MJ

Technosphere Sweden

Input

Refined resource

Electricity

0.102

MJ

Technosphere Europe

Input

Refined resource

Electricity

0.709

MJ

Technosphere Sweden

Input

Refined resource

Fuel oil, ship (2-stroke)

0.0258

MJ

Technosphere

Output

Emission

Acetaldehyde

0.0000107

g

Air

Output

Emission

Acetylene

0.000578

g

Air

Output

Emission

Al

0.000253

g

Air

Output

Emission

Aldehydes

0.00000179

g

Air

Output

Emission

Alkanes

0.00126

g

Air

Output

Emission

Alkenes

0.000614

g

Air

Output

Emission

Aromates

0.000233

g

Air

Output

Emission

Aromates

0.00000825

g

Water

Output

Emission

As

0.00000523

g

Water

Output

Emission

As

0.0000916

g

Air

Output

Emission

B

0.00014

g

Air

Output

Emission

Be

0.00000938

g

Air

Output

Emission

Benzene

0.00606

g

Air

B.7

Output

Emission

Benzo(a)pyrene

0.00000017

g

Air

Output

Emission

BOD

0.00000498

g

Water

Output

Emission

Butane

0.00742

g

Air

Output

Emission

Ca

0.000123

g

Air

Output

Emission

Cd

0.00000253

g

Water

Output

Emission

Cd

0.0000734

g

Air

Output

Emission

CH4

0.869

g

Air

Output

Emission

Cl-

2.26

g

Water

Output

Emission

CN-

0.0000125

g

Air

Output

Emission

CN-

0.00000157

g

Water

Output

Emission

Co

0.000176

g

Air

Output

Emission

Co

0.0000004

g

Water

Output

Emission

CO

-0.0975

g

Air

Output

Emission

CO2

918

g

Air

Output

Emission

COD

0.00000126

g

Water

Output

Emission

Cr

0.00000493

g

Water

Output

Emission

Cr

0.00058

g

Air

Output

Emission

Cr3+

0.0000329

g

Water

Output

Emission

Cr3+

-0.000065

g

Air

Output

Emission

Cu

-0.0000000124

g

Water

Output

Emission

Cu

0.000486

g

Air

Output

Emission

Dioxin

0.000000000209

g

Air

Output

Emission

Dissolved solids

0.0435

g

Water

Output

Emission

DOC

0.000000000000319

g

Water

Output

Emission

Ethane

0.00174

g

Air

Output

Emission

Ethene

0.00348

g

Air

Output

Emission

F-

0.000676

g

Water

Output

Emission

Fe

0.000276

g

Air

Output

Emission

Fe

0.00579

g

Water

Output

Emission

Formaldehyde

0.00184

g

Air

Output

Emission

F-tot

g

Air

Output

Emission

H+

0.000035

g

Water

Output

Emission

H2S

0.0000106

g

Air

Output

Emission

H2S

0.0000000514

g

Water

Output

Emission

Hazardous waste

80.7

g

Technosphere

Output

Emission

HC

0.0000245

g

Water

Output

Emission

HCl

0.0627

g

Air

Output

Emission

Heavy metals

3.37E-19

g

Air

Output

Emission

HF

0.00331

g

Air

Output

Emission

Hg

0.00000421

g

Air

Output

Emission

Highly active rad ac waste

0.00321

g

Technosphere

B.8

Output

Emission

Metals

0.00000117

g

Air

Output

Emission

Metals

0.00000583

g

Water

Output

Emission

Mn

0.0000485

g

Water

Output

Emission

Mn

0.000509

g

Air

Output

Emission

Mo

0.00014

g

Air

Output

Emission

N2O

Output

Emission

Na

Output

Emission

Output

4.6

g

Air

0.00115

g

Air

NH3

0.2

g

Air

Emission

NH3

0.0000000461

g

Water

Output

Emission

Ni

0.00169

g

Air

Output

Emission

Ni

0.0000198

g

Water

Output

Emission

NMVOC

0.367

g

Air

Output

Emission

NO2

0.77

g

Air

Output

Emission

NOx

1.56

g

Air

Output

Emission

N-tot

0.669

g

Water

Output

Emission

Oil

0.0424

g

Water

Output

Emission

Organics

0.00000357

g

Air

Output

Emission

Organics

Output

Emission

PAH

Output

Emission

Particulates

Output

Emission

Output Output

0.0376

g

Water

0.000108

g

Air

0.23

g

Air

Pb

0.000019

g

Water

Emission

Pb

0.00041

g

Air

Emission

Pentane

0.0127

g

Air

Output

Emission

Phenol

7.98E-15

g

Water

Output

Emission

Phosphate

0.0000156

g

Water

Output

Emission

PO43-

0.000127

g

Water

Output

Emission

Propane

0.00331

g

Air

Output

Emission

Propene

0.000568

g

Air

Output

Emission

P-tot

0.000000929

g

Water

Output

Emission

Radioactive emissions

-6340

kBq

Water

Output

Emission

Radioactive emissions

-675000

kBq

Air

Output

Emission

Radioactive waste

0.00134

g

Technosphere

Output

Emission

Rn-222

Bq

Air

Output

Emission

Salt

0.0173

g

Water

Output

Emission

Sb

0.00000000561

g

Water

Output

Emission

Sb

0.0000567

g

Air

Output

Emission

Se

0.0000653

g

Air

Output

Emission

Sn

-0.00000176

g

Water

Output

Emission

Sn

0.0000000319

g

Air

Output

Emission

SO2

1.36

g

Air

Output

Emission

SO3

g

Air

5840

About Inventory Scope Data Back to Contents Scope Back to About Inventory Back to Contents Publication

Davis J, Haglund C (1999). ’’Life Cycle Inventory (LCI) of Fertiliser Production - Fertiliser Products Used in Sweden and Western Europe’’. SIK report no. 654. The Swedish Institute for Food and Biotechnology (SIK). Göteborg, Sweden.

B.9

Data documented by: Jennifer Davis, SIK (The Swedish Institute for Food and Biotechnology). Documentation reviewed by: Ann-Christin Pålsson, CPM, Chalmers University of Technology Intended User

The data are intended to be used by people working with life cycle assessments of food production systems.

General Purpose

To generate an inventory of emissions and use of resources due to the production of fertilisers used in Sweden.

Detailed Purpose

The purpose was not to compare the production of different fertilisers with each other, but to generate a thorough inventory of emissions and use of resources due to the production of different mineral fertilisers. The data are intended to constitute a useful basis of input information in life cycle assessments of food production systems.

Commissioner

SIK AB, The Swedish Institute for Food and Biotechnology Box 5401 SE-402 29 Göteborg Sweden .

Practitioner

Davis, Jennifer and Caroline HaglundSIK AB Box 5401 402 29 Göteborg Sweden.

Reviewer Data Back to About Inventory Back to Contents Applicability

The data are applicable for production of CAN fertiliser produced and used in Sweden. It can be assumed that production of CAN in Sweden is representative for production of all CAN that is used in Sweden, but it is important to be aware of the fact that there may exist imports of CAN into Sweden.

About Data

Data are gathered from the official environmental report distributed by Hydro Agri AB in Landskrona, from personal comminication with people working there and also from literature and reports on fertiliser production. There are only two sites in Sweden that produce CAN fertiliser and the site in Landskrona is representative for Swedish production of CAN. Emissions from transports, combustion of energy carriers and production of steam, district heat and electricity have been included by using information and emission factors from the database in LCAiT 3.0. LCAiT 3.0 is a computer programme created by CIT Ekologik in Göteborg for practitioners of life cycle assessments.

Notes

Internal review of the report was performed by: Olle Ramnäs, CTH (Chalmers University of Technology), Berit Mattsson and Magnus Stadig, SIK (The Swedish Institute for Food and Biotechnology).

SPINE Data Report formatted as Hypertext Markup Language Created by SPINE@CPM Data Tool

Copyright IMI - Industrial Environmental Informatics, Chalmers University of Technology, 1997

B.10

SPINE http://www.imi.chalmers.se

Annex C Concept models This annex provides an overview of the concept models presented in chapter 4.

Flow

Activity

Figure 16. The concept model of the concepts Activity and Flow

ObjectOfStudy

Inventory

Activity

Componentship

Figure 17. Concept model describing the technical system studied in LCA. Activity

Flow

Componentship

FlowConnection


QMetaData

Activity

Flow

FlowType

Substance

FlowProperty

SubstanceProperty

Figure 19. Model of the concept Flow. Activity

Flow




CategoryIndicator

Figure 21. Concept model of LCA characterization.

C.1

Activity

Substance

Flow



Figure 22. Model of concept of Aspect.

Activity

Substance

Flow

SubstanceProperty XOR

Aspect




ImpactCategory


CategoryIndicator

WeightingFactor

ImpactIndicationPrinciple

WeightingMethod

Figure 23. Concept model of LCA impact assessment. Project

Design

Substance

SubstanceProperty

RAVELEntity

Composition

CompositionProperty

EntityProperty

Figure 31. The concept model that describe a product design.

C.2

Project


Design

RAVELEntity Figure 32. Model of concept of design targets and environmental performance results (score). Aspect



CategoryIndicator

Project

WeightingFactor

Function/Algorithm


Design

Composition

RAVELEntity

CompositionProperty

EntityProperty

Figure 33. Concept model of how environmental performance is interpreted and calculated.


Activity

Flow

Substance

RAVELEntity

Figure 34. Concept model of the environmental properties of a material or component, in terms of its life cycle assessment. See also figure 29.

C.3

Aspect


CategoryIndicator

Project


WeightingFactor

ProjectTarget

Function/Algorithm

DesignScore


Activity

Flow

RAVELEntity

Substance

SubstanceProperty

Composition

EntityProperty

Figure 35. The full concept model of RAVEL design for environment.

NatureFramework Load

Mechanism

Indicator

Figure 39. High level view of the core of the OMNIITOX concept model.

C.4

CompositionProperty

Carlson R., Löfgren G., Steen B., Tillman A-M.; "LCI Data Modelling and a Database Design" Published in The International Journal of Life Cycle Assessment Vol. 3. No.2. pp. 106-113; 1998

Abstract A large scale operative data format for transparent storage, administration and retrieval of environmental life cycle inventory (LCI) data has been implemented by applying data modelling and database design. Key concepts in the design are 'activity' and 'flow': An activity is a technical system, such as a process or a transport, or an aggregate of different processes or transports. A flow is any matter entering or leaving an activity, such as natural resources, energyware, raw material, emission, waste or products. Any numerical data set on an activity can be thoroughly described by supplying meta data. Meta data fields are prepared for a wide set of commonly known LCA-data aspects, such as descriptions of data acquisition methods, system boundary conditions and relevant dates.

Keywords:

LCA, LCI, database, model, conceptual, formalise, meta data, computable data, transparent, systems analysis, activity, flow

1(14)


LCI Data Modelling and a Database Design Raul Carlson, Anne-Marie Tillman, Bengt Steen; Chalmers University of Technology, Göteborg, Sweden Göran Löfgren, Nordic Port AB, Göteborg, Sweden

Introduction This paper presents a flexible life cycle inventory (LCI) data model, which has been mapped and implemented into a relational database table structure named SPINE (Sustainable Product Information Network for the Environment) [1]. The data model supports full LCI data transparency, both in regards of references to data in the same database and in regards of references to data sources external to the database. The model is methodologically flexible since it supports any LCI methodology, and it is technically flexible since it is not restricted to being implemented as a relational database, but can be implemented using any existing database technology. The data model is also useful as a design aid when designing calculating LCI and life cycle assessment (LCA) software tools, or when modelling and designing LCI data communication formats. The choice of implementing SPINE as a relational database is in itself a flexible choice, since a relational database can be used in many technically and practically different ways. Therefore SPINE can be used as a storage medium for LCI data and LCI reports or it can be used as a data storage format for LCI or LCA software or it can be used as both, alternately or simultaneously. SPINE is the merged outcome of two separate projects. The aim of both these projects was to develop a database which could be used both as a data storage medium for LCI data and as a data storage format for LCA software [2][3]. The first of these two projects, a master thesis, focused mainly at finding a data model which satisfied the general and specific needs formulated by the LCI methodology and which at the same time was correct from the viewpoint of computing science and database technology. The other project, Nordic Project on Environmentally Sound Product Development 1 (NEP) [1] (as described below), focused on reaching an industrial consensus in regards of a data model and of the required contents of an LCI database, indiscriminate to economical or industrial sector. Concerning data modelling, the latter project focused on feasibility and applicability. During both projects, LCA-practitioners and system developers worked closely together in a continuous dialogue, exchanging their respective expertise. This cross fertilisation of competence was an important contributor to the outcome of the two projects and to their final merging into SPINE. The project performed as a master thesis, was a co-operation between the departments Technical Environmental Planning and Computing Science at Chalmers University of Technology, Göteborg (1994) [4]. The industrial consensus project was a sub-project within the NEP project, aiming at finding a common LCA database format for Scandinavia (19941995) [1].

1

Twenty-two large companies and institutes in Scandinavia participated in NEP. 2(14)


Related works LCI databases are important in practical work with LCA. This because it is difficult to find suitable data, and when data is found it often originates from different sources, has different formats, and must be interpreted and reformatted into a form that suits LCA. This makes data retrieval both an expensive and a time consuming part of LCA. Databases are expected to solve many of these problems, and, consequently, a great deal of work has already been done in this area, as may be exemplified by the great interest that has been shown towards the SPOLD data format work, as well as towards a great many earlier works: • Major database collections of LCI data: 1994: Swiss database for LCI on energy systems [5] 1996: European Database for Corrugated Board Life Cycle Studies [6] • Serious approaches towards recommendations and requirements for LCA databases: 1993: General principles for life cycle assessment databases, I. Bousted, UK [7] 1994: Criteria for a national database for life cycle assessment, Sweden [8] • Projects aimed at finding a co-ordinated common view of LCA data: 1996: An approach to creating a common format for Life-Cycle Inventory Data, SPOLD [9] • Major approaches towards computer based databases for storing easily retrieved LCA data: 1991: IDEA- An International Database for Ecoprofile Analysis [10] 1994: The Swedish Pulp and Paper Research Institute builds a database for LCA in the forest industry [11] • Databases as files within LCA calculation software: Examples: LCA Inventory Tool [3] [12] and SimaPro More approaches to LCA databases may be found. But it should be stressed that none of these works have had a general data modelling approach towards the task. Therefore these solutions typically are either methodologically imprecise (Boustedt’s General principles… or Ekvall’s Criteria…) or are they too rigid (typical for implemented systems like LCA Inventory Tool and SimaPro or early standardisation approaches like SPOLD’s data format).

The approach The intention was to develop a database which transparently should store LCI information on any technical system, of any size or type, such as single machines or entire production networks following a product from cradle to grave. It should be possible to store information on both single technical systems, as well as on models of technical systems assembled from information on other technical systems in the same database. The latter should enable storage of fully transparent reports by literally including the contents of the references and their interpretation within the report of any specific study. This should give that in addition to reusing information on simple original data sets, also entire models of technical systems could be reused. This should be practically useful for information on systems supplying, for example, electrical power, waste treatment and waste water treatment. It was intended that the database should support any LCI approach or methodology, e.g. LCI:s with different approaches to system boundaries or allocation methods, and it should be general in terms of database technology. In addition to this, it should be possible to use the same database as a data storage format for different LCA software, as well as for general databases for LCI data.

3(14)


This approach gave the following criteria for the database: 1

The database must be based on a general model in order to conceptually define the term ‘technical system’.

2

In order to enable full data transparency, the database must allow for descriptive information about each technical system and about any quantitative data in the database. It should be possible to describe every data entity for which ambiguity may be raised.

3

To make the result generally available and independent of any specific database technology, a conceptual data model of the information needed to describe LCI data was needed. It should be possible to implement this model into any database system. 4 The intended generality regarding the use of the database implies requirements on the choice of implementation: the database should be based on generally accepted computer based database technology.

SPINE complies to these criteria by a series of solutions: In the following section (Modelling Transparent LCI Data) a conceptual systems model2 will be described, which enables SPINE to store any size and any type of a technical system. In the subsequent section (LCI Data) is given a description of the data needed to describe these systems. By having made a conceptual model of LCI data implies that the database solution will be independent of any specific database system. The requirement for the database to be useful both as a storage format for LCA software and for large LCI databases suggested the use of a relational database system. This because they are commercially available and because the competence on this technology is widely spread.

Modelling Transparent LCI Data When data is used by and communicated within a large group of users, for their work and for their decisions, there is a need for a common understanding of the data. Not only must the data format be understood collectively, but also must all relevant and implicit understanding of the data be made explicit, and be supplied together with the data. Explanations cannot be adhered orally, and cannot be implicitly understood from the familiarity with the contextual situation in which the data originally evolved. Data modelling is a means for formalising data shared within a group of information users. The purpose of the data modelling process is to retrieve all relevant types of information on a certain topic, and to seek to find an effective and efficient image (model) of that information. The result is aimed to allow any user within the group all the information which can be retrieved from within the entire group and which this user needs to fulfil his or her responsibility within the group. That is: anyone who needs information can reach it, while it can be secured from anyone who should do not need it. When modelling SPINE, the modelling topic was LCI, as a part of LCA, and the group of users were practitioners, reviewers and commissioners of LCA.

The Concept of Activity

2

The conceptual model gives an interpretation to the concepts and their relations. 4(14)


In LCA the detailed operations of a process are of limited interest. The items of main interest in LCA are the inputs to and outputs from the processes. In this respect, transport processes, transforming processes and other processes have enough similar characteristics to be modelled as one common concept, which we call activity. For most processes the inputs and the outputs are dependent, linearly or not. The data model, as presently implemented, supports only linear relations between inputs and outputs, although the conceptual model supports any dependencies. Transports, however, differ from most other activities, in that the inputs and outputs also depends on another parameter the transport distance. Thus, it was found useful to differ between activities with outputs depending on inputs only, and those where an additional parameter is needed to calculate the inputs and outputs. Figure 1 graphically exemplifies the modelling transformation from different types of processes into activities.

Extraction Process Transport Transport Use Transport Manufacturing Process Process

Transport Waste Treatment Process

Extraction Process

Extraction Process

Transport Transport Use Manufacturing Process Transport Process

Transport

Waste Treatment Process

Extraction Process

activity

product flow

Figure 1. Graphic representation of the modelling of processes and transports into activities.

Most processes can be disaggregated into sub-processes. For example, a process for making steel sheet (figure 2) can be disaggregated into the process of making steel bars from iron ore, rolling the bars into thick sheets, and then rolling the thick sheets into thin sheets. The steel making process can then be further disaggregated into a primary steel making process and a casting process. A disaggregation like this can be generally applied to any complex process.

5(14)


Emissions

Iron ore Coal

Thin steel sheets Steel sheet production

etc...

Iron ore

etc...

Emissions

Coal

Oils

Steel bars

etc...

Iron ore

Steel bands

Rolling 1

etc...

etc...

Emissions

Coal

Emissions Thin steel sheet

Thick sheets

Steel Making

Rolling 2

etc...

etc...

etc...

Emissions

Steel Steel Making

etc...

Emissions

Steel bars Casting

etc...

etc...

etc...

Figure 2. Disaggregating a process into components.

In LCA, a process disaggregation is used in the goal definition stage, or during the inventory stage: The initial technical system, supplying a product or a functional unit, is divided into a raw material extraction activity, a manufacturing activity, a use activity and a waste management activity. Through further stepwise refinement, the real3 industrial activities may be seen, with a successively increased level of detail. The possibility of aggregating and disaggregating activities was modelled by extending the meaning of the activity concept: An activity is either aggregated, which means that it has internal activities, or is it a simple activity. The aggregated activity is the key to allow for transparent LCI reporting, reuse of plain data on technical systems and reuse of entire models of technical systems: • By internalising a plain activity within an aggregated activity, the entire set of references and explanations is also internalised, hereby guaranteeing absolute transparency. • A plain activity can be reused within an unlimited number of aggregated activities. • Since both plain and aggregated activities are treated identically, an aggregated activity may include any number of both plain activities as well as aggregated activities.

The Concept of Flow It is not possible to anticipate all future needs regarding types of inputs and outputs. Some typical types are: ‘raw material’, ‘electric energy’, ‘fossil fuel’, ‘product’, ‘waste’ and ‘emission’. Depending on how and why an LCI is made, there may be a need to include different numbers of types. For example, it may be necessary to distinguish between ‘virgin’ and ‘recycled’ raw material, or ‘main products’, ‘by-products’ and ‘waste’. For this reason it 3

Here ‘real’ means a level of process aggregation for which data is accessible, sometimes referred to as ‘unit processes’ [9]. 6(14)


would be unnecessarily inflexible to design a conceptual model for some fixed set of types. Instead, we modelled only two general types, named ‘inflow’ and ‘outflow’. The two are conceptually referred to as flow, and they can be of any subtype, in concert with a purpose. Any activity has an unlimited number of inflows and outflows. Identification of activities to include in the technical system of the inventory is done by starting from a specific product or service. From this starting point, the matter and energy paths are followed, upstream to the cradle of the raw material and downstream to waste treatment. The result is a technical system which is not merely a collection of activities and subsystems, as shown in figure 3 above, but rather a network of activities. Each activity is incorporated into the technical system on the basis of its exchange of matter or energy with the previously incorporated activities4. A network is built by connecting flows of activities within an aggregated activity (figure 4). Since resource use, emissions, waste and by-products are also flows, there will be a number of flows not connected to any other flows of other internal activities. These free flows represent resource use, use of raw material, emissions, waste and by-products out of or into the total technical system of the study: The flows of an aggregated activity are the sum of the free flows of its internal activities (figure 4). free flows of the internal activities become flows of the aggregated activity

an inflow and an outflow connecting two internal activities

with all free flows summed up, an aggregated activity can be regarded as a plain activity

Figure 4. Flows of internal activities of an aggregated activity.

The fact that an aggregated activity can have flows, as can any other activity, makes the conceptual model of activity and flow extremely general: any technical system can be transformed into an activity, and can therefore be incorporated into any other technical system. The fact that an aggregated activity can have flows qualifies it to participate in flow connections, which means that it can exchange both matter and services with other internal activities inside an aggregated activity. The flow concept allows an activity to have any number of products. The data model also supports any allocation method applied to such a production system. This support has been differently implemented for different allocation methods.

4

It is not necessarily true that all activities are connected within a technical system. For example, when allocation is avoided by applying system expansion [20], it may lead to technical systems consisting of two or more separated technical subsystems. 7(14)


Avoidance of allocation, by system expansion or by an increased level of detail, as suggested in 1996 in the ISO draft CD 14041.2 [13] is inherently supported by the conceptual activity model: Any extra technical systems can be added to an aggregated activity, and any single activity can be exchanged with an aggregated activity to increase its level of detail. A general method for both non linear and linear allocations is also supported by the model: Allocation based on non linear process behaviour is not supported in the present implementation of the data model, but it might very well be expanded to support also this. Linear allocation, based on, for example, relative economical value, is fully supported in two ways: Any substance and any flow may be subscribed an arbitrary number of quantitative properties, for example economical value. Thereby giving possibility to register the data needed for linear allocation decisions. In addition to this, it is also possible to register linear, quantitative relationships between any pair of flows of an activity internal of an aggregated activity. By having defined the activities as systems, any principle of allocation may be applied to them, either automatically by rules implemented in software, or by manually defining a new allocated activity which references the original data set. Also, it is possible to describe any allocations that may have been applied. This, of course, is especially useful for data without references within the database, since many decisions must be made when isolating any technical system from its (technical and economical) environment. This is further discussed below, when describing the meta data abilities of SPINE.

LCI Data As a basis for the description of LCI data, data is separated into two different types: computable data5 , which is data that can be calculated with, and explanatory data, here referred to as meta data6.

Computable Data Amounts of flows are computable data, as is transport distances and relative economical value. There are many possible ways to represent amounts. The choice depends both on how the retrieved data is formatted and on whether or not the normalisation of a technical system includes statistical calculations. The possible formats prepared for in the present implementation of SPINE is: Single numerals (quantity = 35.2), assumed Normal-distributed mean values ( x ) supplied with a variance σ (quantity = 35.2, σ = 4) or any other type of statistically interpretable value, supplied with a minimum and a maximum value in accordance with some statistics for the values (quantity = 35.2, min = 28.3, max = 41.0)). However, we are fully aware that this approach will need to be further developed in pace with an emerging use of statistical methods in LCA methodology, and an emerging growth of the statistical understanding of retrieved LCI data.

5

We use the term ‘computable’ in a contextual sense, meaning data computable within the LCA methodology. 6 ‘Meta’ - from the Greek meaning ‘transcending or going beyond’. Used here to mean ‘Data about data’. 8(14)


Meta Data The concept of activity and flow is an image, or a model, which may be communicated by itself, in order to synchronise an implicit and collective understanding of computable LCI data within a group of data users and suppliers. However, due to inventory subjectivity, data gaps, system boundary difficulties and other pitfalls of LCI, there is also a need to supply data with explanatory data, meta data. In general there are two types of meta data needed for LCI data: • The physical object: A relevant description of the physical object which data is meant to represent. • The principles, procedures, conditions and decisions applied and obeyed: Relevant descriptions of the principles, procedures, conditions and decisions applied and obeyed when retrieving and translating information into the present data representation of the physical object. These two meta data types will be further described in the two following subsections. Describing Physical Objects Activities may have similar names and produce similar products but may still be environmentally different, due to differences at the system level. These differences may be found in the internal functional subsystems or in the choice of system boundaries (technical, geographical, time, …). Therefore such information needs to be supplied with the data. Examples of relevant information on the internal of systems are shown in table 1 below. Table 1. Internal aspects of technical systems which may relevantly affect their environmetal behviour in relation to a similar system producing the same product.

Internal aspect of system

Example

Example in general terms

Factual technology

The name of the technology There may exists more than applied to produce a product. one alternative production technology.

Emission reducing devices

Filters or waste water treatments are present

Type and location for the environmetal impact has been changed.

Technological improvements Decreased temperature for chemical reaction to occur.

Changes made to a standard procedure or process, thereby reducing use of resources.

Energy saving devices

Waste has been turned into goods by technological innovation.

Heat exchangers are installed.

Examples of system boundary conditions that needs to be described are exemplified in table 2 below.: Table 2. System boundary aspects that may affect the environmental impact of the technical system.

System boundary condition

Example

Example in general terms

9(14)


Technical system boundaries

Administrational supply and internal transports.

Included or excluded system components.

Geographical system boundaries

Geographical site of electric energy generation.

The region within which a technical system is extended, and the surrounding, not included regions.

Time boundaries

The time for which the data set is relevant.

Any time related aspects of the system that may affect its current state.

The implementation of SPINE allows for structured storage, retrieval and administration of this descriptive information, together with the computable data. Material and substance flows must also be described, in order to unambiguously identify them. This is mainly due to naming ambiguity. For example, ‘Iron’ could mean ‘Iron ore as a natural resource’, ‘Mined iron ore’, ‘Cast iron’, or even ‘Steel’ or ‘Steel product’. Therefore, vaguely defined names, together with vaguely described technical systems, may lead to erroneous LCA results, due to double accounting or unnoticed lack of subsystems. The SPINE model gives structured support for implementing and applying any hierarchical nomenclature for material and energy: In order to give name to a flow, the name must first be inserted into the hierarchical substance nomenclature. Thereafter the name can be chosen for the flow. This arrangement ensures a second thought when naming data, and it provides a practical means for standardising naming within groups of LCA users: Data suppliers can be supported with ready made and simply used nomenclatures within their database, thereby applying the same foundation for naming. The same nomenclature technique is implemented for many other names, for example process types, industrial sector, geography and environment. Describing Principles, Procedures, Conditions and Decisions A set of data representing a technical system is the result of a applied principles and procedures in compromise with subjective decisions due to factual conditions: The technical system must be selected: a specific plant, for example, or an average formed from a population of similar plants. It must also be decided which parameters to include, how to draw the system boundaries, how to deal with time, etceteras. It is not likely that any two representations will be the same if created by different people under different conditions. Therefore, in order to correctly understand and interpret the data, it is necessary to supply the data with a description of the decisions that led to the results, as well as of the people responsible for the result, and of the conditions under which the data was put together. SPINE has been designed to structurally allow for such descriptions on any data set. Examples of how the principles and conditions can be applied when defining a system boundary is given in table 3 below. Table 3. System boundary aspects and the subjective aspects resulting in these boundary choices.

System boundary aspect

Example

Subjectivity aspect

10(14)


Technical system boundaries

Applied allocation or cut off rules and assumption regarding their relevance.

Applied principle for allocation and compromised application of this principle under actual condition.

Geographical system boundaries

Description of assumptions regarding geographical extensions of technical support systems.

Applied principle for cutting support systems at national borders and compromised application of this principle under actual condition.

Time boundaries

Description of assumptions regarding the time for data relevance.

Applied future scenario method and compromised application under actual conditions.

Environmental boundaries

Description of applied criteria Applied impact assessment for selecting the relevant and model for identifying the relevant parameters and significant flows. compromised application of this model under actual condition.

Computable data of course also is the result of compromises between principles and facts. Examples of descriptions that may supplied with each separate computable data entity are shown in table 4. Table 4. Aspects of computable data that are subject to subjectivity and therefore should be supplied with a description.

Aspect of computable data

Example

Subjectivity Aspect

The origin of the computable data entity

How and when they were measured, or otherwise retrieved.

Identification of empirical value of the computable data entity.

The statistics of the computable data entity.

How they were mathematically and/or statistically treated.

Identification of the scientific value of the methods applied to the computable data entity as well as the value of the statistics provided with the entity.

The compilers interpretation of the computable data entity

A recommendation on how An applicability guide for the they should be interpreted and data user supplied by the used. provider of the data.

Each computable data, within a data set on an activity, may have been retrieved differently. Therefore, this information can be supplied for each numerical value.

Meta Data on Receiving Body

11(14)


For the impact assessment stage of LCA it is necessary to have information on the environmental interference caused by the free flows (described in section The Concept of Flow) of the technical system.

Environmental system

Technical system being studied

Other technical systems

Flows: Resources from nature

Waste

Emissions

Products

Figure 6. Relations between the technical and the environmental systems.

Figure 6 shows a finite technical system with free flows (compare with figure 5). Some of these free flows reach another technical system, either as raw material or energy supply, or as products and waste left for waste treatment. The rest of the free flows reach the environmental system as resource extraction, or as emissions or waste. In general impact assessment methods regards the environmental system as one large ‘average environment’. To perform an impact assessment with such a method, the occurrence of free flows is sufficient meta data. Discussions have been held regarding a more spatially differentiated view of the environment, also taking regional or local differences in the environment into account [14] [15]. To perform an impact assessment applying such methods, free flows, summed parameter for parameter, is not sufficient. Additional information must be given in order to identify the local or regional environment. When designing SPINE it was decided to make it possible either to describe the environment in terms of environmental types and/or in terms of geographical locations. When the receiving body is described in terms of locations, the environmental type must be found from external meta data, for example environmental geographical information systems (GIS) or habitat models.

Discussions & Conclusions The data model has been proven successful: It allows full transparency and reuse capabilities of LCI data. SPINE can be modified to store any meaningful meta data on any computable

12(14)


data and on any system property. SPINE also allows inventories of technical systems to be reused like any originally collected data. The model is also methodologically flexible, supporting data handling for several types of LCI methodology. There are SPINE implementations today, both large databases holding collections of LCA data and commercially available software applications using SPINE as a storage format [16] [17]. There are also commercially available electronic LCI data communication software based on the same model. It should be stressed that data can be displayed in any way a user desires, for example as graphical flowcharts, as SPOLD [9] forms, or via software directly accessible via the internet. This generality has been achieved both by focusing on data modelling, but also from having chosen the relational database technology for the implementation.

Future development Having a technically and methodologically functional data model and well functioning database implementations for LCI data is a good step forward for LCA data handling. But still there are some important steps to be taken. In order to efficiently exchange and reuse LCI data, LCI databases and other useful data sources needs to be compatible with each other. The work of SPOLD [9] has meant much in terms of a common semantics for LCI data, but there still is no common view of the concepts and the logic of such a format. We recommend that a common view of LCI data, in all aspects, should be done by applying a data modelling approach. A consequence of what was described in the subsection Meta Data on Receiving Body, SPINE is designed as a module for data on technical systems. Any compatible module modelling the environmental system can be joined with this module. It would be of great interest and importance to the LCA methodology to have good, compatible models of the environmental system. To fully exploit the capabilities of LCI data models, such modelling still needs to be done.

References 1. Steen, B. et al. (1995), SPINE, A Relation Database Structure for Life Cycle Assessments, Swedish Environmental Research Institute (IVL), IVL-Report B 1227, Göteborg, Sweden. 2. Ryding S-O et al. (1993) The EPS system - A life cycle assessment concept for cleaner technology and product development strategies, and design for the environment, EPA Workshop on Identifying a Framework for Human Health and Environmental Risk Ranking, Washington DC, USA. 3. LCA Inventory Tool (1994), Chalmers Industriteknik (CIT), Göteborg, Sweden. 4. Carlson, R. (1994), Design and Implementation of a Database for use in the Life Cycle Inventory Stage of Environmental Life Cycle Assessments, Department of Computing Science, Chalmers University of Technology, Göteborg, Sweden. 5. Frischknecht, R et al. (1994), Ökoinventare für Energiesystemen, ENET, Teil 1-3, Zürich, Switzerland. 6. European Database for Corrugated Board Life Cycle Studies (1996) 1996/1, FEFCO Paris, Groupement Ondulé Darmstadt, KRAFT Institute Stockholm, Sweden. 7. I. Bousted, General principles for life cycle assessment databases, Journal of Cleaner Production, 1993, Vol. 1, Number 3-4. 13(14)


8. Ekvall, T. (1994), Criterias for a national database for life cycle assessment, CIT EkoLogik, Göteborg, Sweden. 9. Singhofen, A. (1996), Introduction into a Common format for Life-Cycle Inventory Data, Society for the Promotion of LCA Development (SPOLD), Brussels, Belgium. 10.An International Database for Ecoprofile Analysis, A Tool for Decision Makers (IDEA) (1991), WP-91-30, International Institute for Applied Systems Analysis (IIASA), Laxenburg, 11.Haglind, I. et al. (1994), STFI bygger databas för skogsindustriella livscykelanalyser, p 34-35, Svensk Papperstidning/Nordisk Cellulosa nr 2, Sweden. (in Swedish) 12.SimaPro Demo v 3.1 (1995), PRé Consultants, Amersfoort, Netherlands. 13.ISO/CD 1041.2; International Organisation for Standardization (ISO), 1996. 14. Udo de Haes, H.A. et al. (1996), Towards a Methodology for Life Cycle Impact Assessment, Part 1: Discussions of general principles and guidelines for practical use, SETAC. 15.Schaltegger, S, (1996), Eco-Efficiency of LCA, The Necessity of a Site-Specific Approach, Life Cycle Assessment (LCA)-Quo vadis?, Birkhäuser Verlag. 16.EcoLab (1997), Nordic Port AB, Göteborg, Sweden. 17.EPS 3.0 Design System (1997), Assess AB, Göteborg, Sweden.

14(14)

Journal of Hazardous Materials 61 Ž1998. 67–75

An approach for handling geographical information in life cycle assessment using a relational database Magnus Bengtsson) , Raul Carlson, Sverker Molander, Bengt Steen Chalmers UniÕersity of Technology, Department for Technical EnÕironmental Planning, S-412 96, Goteborg, ¨ Sweden

Abstract A new data model has been developed to handle information relevant to site-specific life cycle assessments ŽLCA.. The model is orientated towards GIS-representations of three generalised subsystems; the technical, the environmental and the social subsystems. The technical and environmental systems are mainly linked through flows of energy and matter, which are the causes of environmental impacts, which subsequently is perceived, evaluated and acted upon by the social subsystem. For all three systems important differences, attributable to geographical locations can be determined. With the new data model a possibility to enhance LCA and reach more relevant results emerge due to a higher site specificity. The high level data model is expressed as relations between different entities using the entity relationship ŽER. modelling language. An existing LCA-database, SPINE, which is already used by several companies for decision support in product development, can be utilised since the structure of the database supports geographical information. So far, applications with GIS-data are limited, but examples of area specific LCA impact characterisation factors exist. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Life cycle assessments ŽLCA.; Entity relationship ŽER.; Relational database

1. Introduction Life cycle assessment ŽLCA. is a tool for assessing the environmental impact caused by a product or service. The basic principle for LCA is to follow the raw material and the energy used by the manufacturing process for the product or the service-performing process upstream, to the virgin resource extractions and downstream, to the final disposal of the waste produced at each stage during the entire life cycle. )

Corresponding author.

0304-3894r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 3 0 4 - 3 8 9 4 Ž 9 8 . 0 0 1 0 9 - 5

68

M. Bengtsson et al.r Journal of Hazardous Materials 61 (1998) 67–75

The environmental impact of a product or process is caused by exchange of energy and matter between the technical Žsub.systems, and their surrounding environment. In the case of agriculture, for example, this interaction may be complex, involving several kinds of impacts such as, e.g. land-use Žincluding the occupation of space., leakage of nutrients, emissions from machinery and manure, and compaction of soil. During the LCA modelling each subsystem of the life cycle is linked together into a chain of processes, in one end extracting resources and in the other giving various types of emissions or waste. This chain of linked processes is referred to as the technical system. In reality a technical system is under some sort of human control and designed for a certain purpose, to deliver a certain benefit or good, which in the LCA is expressed through the functional unit of the system. The processes are also located somewhere, which implies that they can be geographically referenced. Environmental impact caused by a technical system, or its LCA equivalent, the functional unit, is estimated in terms of the negative change implied by the technical system upon the enÕironmental system, as evaluated by the social system. These systems may also be geographically referenced, which is an important starting point for a consideration of the relations between LCA and localised environmental impacts. The traditional, generic, LCA methodology has been developed out of mass balances studies of production systems covering ‘cradle to grave’, where both cradle and grave belong to the environmental system. Materials and energy flows in the technical system are thus followed from extraction to final deposition. However, as the models of the systems become large and complex, information on locations of different activities are mostly lost when aggregating data on similar material and energy flows. This in turn becomes a problem when trying to assess the impact of different flows to and from the environmental system since the sensitivity of ecosystems are varying. The aim of the work reported here is to present a data model of the LCA procedure and to include geographical information in such a way that site-specific impacts may be assessed. A geographical expansion of LCA methodology rises demands on data. A such LCA should thus not be considered as part of the standard method of LCA but as a possibility to be used when the study can benefit from this further work, e.g., when the significant environmental impact is a contribution to acidification where differences of ecosystem sensitivity are well established. It seems not likely that geographically referenced information can be used commonly in an LCA today or in the nearest future. However, a data model in itself is not dependent on the amount of data that are arranged according to the model.

2. A data model of LCA In order to use a relational database as a tool for LCA the method first needs to be expressed, or modelled, into a language that can be used for database design. Fig. 1 depicts a high level model of the relationship between the technical, environmental and social systems, described in the introduction, by using the entity relationship ŽER. modelling language w1x. In the ER-language, a box represents an entity, a diamond


69

Fig. 1. An ER data model of LCA, that describes the relationship between the three different systems.

represents a relationship, and an oval represents an attribute of an entity like, for example, the flows out from a technical system. The key to understanding Fig. 1 is given in the following subsections Žthe relevant attributes of the entities have been lifted out for explication.. 2.1. About the systems 2.1.1. The technical system (TS) A model of a technical system may have zero or many flows Žthe ‘0,m’ between the ‘has’-relation and the ‘Flow’-attribute.. 2.1.2. The enÕironmental system (ES) A model of the environmental system may have zero or many impact categories Žthe ‘0,m’ between the ‘has’-relation and the ‘ImpactCategory’-attribute.. 2.1.3. The social system (SS) A model of a social system’s valuation of the impact on an environmental system may concern zero or many impact categories Žthe ‘0,m’ between the ‘has’-relation and the ‘ImpactCategoriesRelativeWeight’-attribute..

70


2.2. About the relations between the systems 2.2.1. Relations between the technical system and the enÕironmental system Ø Constituted by material and energy flows of TS causing identified changes in ES: The flows of TS may be described in accordance with the model of the impact categories of ES. One specific flow may affect many or none impact categories Žthe ‘0,m’ between the ‘Contributor’-relation and the ‘ImpactCategory’-attribute., while one specific impact category may be the consequence of many different flows Žthe ‘m’ between the ‘Flow’-attribute and the ‘Contributor’-relation.. Please note that this means that the model does not consider environmental impacts caused by other sources than technical systems. Ø Constituted by the change of material and energy flows between the two systems, as implied by changes in ES: Changes in ES force changes on the flows of TS. ŽResource depletion enforces substitutions and decrease of certain flows.. The model considers no other than material and energy flows between the technical and the environmental systems. 2.2.2. Relations between the technical system and the social system Ø Constituted by information flow from TS to SS: SS retrieves information from TS for evaluating the good supplied by TS in relation to the environmental changes it causes. This information generally is carried as the quality or the value Žphysical or economical. of the good produced by the technical system. In LCA this is an implicit part of the evaluation model, in which the good supplied is the purpose of the technical system and the basis for, for example, economical valuation of environmental changes vs. produced products. ŽThe ‘0,m’ between the ‘Control and evaluation of purpose’-relation and the entity ‘Model of a Social System’s Valuation of the Impact on an Environmental System’.. Ø Constituted by information flow from SS to TS: SS controls TS by information input, as a result of an evaluation process. This is the consequence of an LCA, when the result is used to control, manipulate or take any other decision regarding the technical system studied. ŽThe ‘0,m’ between the ‘Control and evaluation of purpose’-relation and the entity ‘Model of Technical System’.. It should be emphasized that the analyst’s information retrieval, for the analysis, is not included in the model, i.e. the analyst is not considered part of the social system, which, of course, is a simplification. 2.2.3. Relations between the enÕironmental system and the social system Ø Constituted by information flow from ES to SS: Environmental change needs to be identified in order to assess its weight in relation to other identified environmental changes. ŽThe ‘0,m’ between the ‘Assumed relationship’-relation and the entity ‘Model of a Social System’s Valuation of the Impact on an Environmental System’.. Ø Constituted by information flow from SS to ES: SS has no active relation towards ES. ŽThe ‘0,m’ between the ‘Assumed relationship’-relation and the entity ‘Model of Environmental System’.. The meaning of this relation may be discussed, since the idea of ‘technical system’, implies that any action performed by the social system is realised through the technical system.


71

3. On ER-modelling When representing an entity as a box, as in the ER-language, the entities’ inner structures are neglected. This allows for a top-down modelling approach, which has two important implications for the modelling of GIS as a tool for LCA. First it allows for focusing on the relations between the entities without having to be concerned about their inner structures. When applying a systems analytical approach it is beneficial to regard these relations as interfaces, each relation defines the interface between adjacent systems. This is the interdisciplinary approach of LCA: the relations between the systems can be defined, without having to deal with the complexities within each one of them. Hereby enabling the design of a robust high-level data model of LCA as a tool. Having designed the high-level model, the competence required while designing the inner structure of any entity need not be interdisciplinary. As long as the high-level model and the relations between the different systems are understood and obeyed by the modelling teams, i.e. as long as the interfaces of adjacent systems are consistent, each team is free to focus on how to model the inner structures of any specific entity.

4. SPINE SPINE w2x is a relational database structure, designed for storing LCA data. The structure has been used to implement LCA databases at many different locations, both for large databases and as storage device for LCA analysis software. The structure of SPINE follows the model described in Fig. 1. The inner structure of the entity describing the technical system has been thoroughly modelled in SPINE. This was done in 1994 as a close co-operational project, including chemical and technical engineers and data modelling experts. The result was a structure for the analytical model and data on the technical system and an integrated data dictionary for data describing the data Žmetadata.. SPINE also includes models of the environmental and the social systems, but these are not yet as well modelled as the technical system. In addition to this, SPINE was prepared also for handling geographical information. In order to realise this enhancement of SPINE, LCA needs to be further analysed from a geographical viewpoint.

5. LCA and geography Modelling LCA as consisting of three related systems implies that the geographical aspects of LCA is the aggregate of the geographical aspects of each of these systems. The nature of each such geographical aspect is presented in the following three subsections.

72


5.1. Geography and technical systems A technical system generally includes processes which might be geographically referenced, and which are connected via different types of transport systems, such as goods Žroad, sea, air. or energyware Žpipelines, electricity grid. distribution systems. The geographic location and extension of such a technical system gives relevant information for the modelling and assessment of its environmental impact w3x. 5.2. Geography and enÕironmental systems Given the geographical location of the different parts of the technical system, it is possible to model the dispersion of various agents, so that the varying sensitivity of ecosystems, regions, etc., can be taken into account where this is relevant w4x. 5.3. Geography and social systems A geographically large technical system Žand the environmental impacts of such a system. may cross national, regional and even continental boundaries, and therefore also affects different cultures or groups of people, holding different attitudes towards changes in the environment. Since the environmental impact is a composite measure, based on

Fig. 2. A data model of LCA complemented with an entity for GIS.


73

both the changes that take place in the environment and on the social system’s attitudes towards such changes, a relevant assessment should have the possibility of taking into regard the attitudes of different social groups w5x. In the LCA study the relevant social system is defined, i.e. what groups of people are relevant in the case in point. Dominant attitudes towards certain environmental changes may be modelled geographically for different social groups.

6. A data model of LCA and geography We may now add to Fig. 1 a new entity, the model of the geographical system, as depicted in Fig. 2. In accordance to what has been said above, the three other entities, or system models, can be described in relation to this new entity. This new model gives that the relations between the model of the geographical system and any of the other systems are strictly related to geographical information, i.e., a geographical reference. The relational complexity of the three previously described systems is preserved, which indicates that the system model of LCA with the technical, environmental and social systems harmonises well with an expansion covering geographically referenced data. Fig. 2 may now be simplified, as depicted in Fig. 3. It is assumed that the ‘Model of the geographical system’ is expressed in terms of a geographical information system

Fig. 3. The relations between a GIS model and a model of LCA. Please note that B is a simplification and a redrawing of A.

74


Fig. 4. The three systems in LCA represented as GIS layers.

ŽGIS., and all LCA specific information has been modelled into an entity called ‘Model of LCA’. Due to the geographical independence of the internal relations within the LCA model, there are three parallel and independent relations between the two entities.

7. LCA and GIS Having organised the information as described in Figs. 2 and 3, an LCA can be performed as follows: the technical system of LCA is aggregated from geographically described subsystems. The environmental changes caused by this system are calculated by use of geographically related data on both the technical and the environmental system. The impact assessment is calculated, or otherwise analysed, by relating each environmental change at each geographical location to the information on each appurtenant local, regional or global attitude to this change ŽFig. 4..

8. Discussion and conclusions LCA methodology is highly dependent on available information from risk assessments and impact assessments. It is not realistic to expect any new such assessments being made within an LCA study. If properly designed with respect to geographical information, LCA methodology may however be a powerful tool to incorporate and communicate knowledge on environmental issues into various types of decision-making such as product development and purchasing. A stable model of LCA, expressed in terms of related systems, an ER-model and a database structure provides a good foundation for developing GIS supported tools for LCA analyses. Implementation of any such tool is a matter of software development. However, the efforts required to fill these models with workable data should not be underestimated.


75

Within the nearest future it seems reasonable to assume that information regarding regions and countries will only be used for certain impacts such as acidification and eutrofication and for weighting using comparisons with political goals. Site-specific data may perhaps be used where a sensitivity analysis has shown that local impacts from a certain site is of particular importance for the results. It should however be noted that the data model itself does not require that geographical information is given, it only provides means for storing such data in a structured way.

References w1x N. Elmasri, S.B. Navathe, Fundamentals of Database Systems, The BenjaminrCummings Publishing Company, Redwood City, 1989, pp. 409–451. w2x B. Steen, et al., SPINE: A Relational Database Structure for Life Cycle Assessments, B1227, IVL-Swedish Environmental Research Institute, Stockholm, 1995. w3x S. Schaltegger, in: S. Schaltegger ŽEd.., Life Cycle Assessment ŽLCA. —Quo vadis?, Birkhauser, Basel, ¨ 1996. w4x J. Potting, K. Blok, in: H. Udo de Haes ŽEd.., Integrating Impact Assessment into LCA, SETAC-Europe, Brussels, 1994. ¨ w5x P. Hofstetter, in: A. Braunschweig et al., ŽEds.., Developments in LCA Valuation, IWO-Diskussionsbeitrag No. 32, Universitat ¨ St. Gallen, St. Gallen, 1996.

Carlson, Pålsson; Industrial environmental information management for technical systems, Journal of Cleaner Production, 9 (2001) 429-435

Industrial environmental information management for technical systems Raul Carlson, tel + 46-31-772 21 73, e-mail: [email protected] Ann-Christin Pålsson, tel + 46-31-772 56 46, e-mail: [email protected] Centre for Environme ntal Assessment of Product and Material Systems, Chalmers University of Technology, 412 96 Göteborg, Sweden

Abstract This paper describes a procedural approach for modelling a technical system for industrial environmental assessments. The approach is fo rmulated as a procedural and conceptual model, denoted PHASETS (PHASEs in the design of a model of a Technical System). The objective of PHASETS is to serve both as a linguistic model for documents describing a technical system, as well as to serve as a guideline when accomplishing or assessing a model of a technical system. The PHASETS approach for information management and data quality control is independent of data format but relies on organisational commitment, for authorisation and for supplying the necessary competence needed to maintain the quality routines. Some applications for PHASETS are described. Keywords: integration; communication; information system; environmental management system; life cycle assessment.

1 Introduction Industrial sustainable development requires analyses and assessments of vast amounts of information about technical systems. An increased understanding of what sustainable development means has led to the formulation of new requirements and duties, which the industry needs to meet and fulfil. Today the information amounts handled within industrial environmental information systems are huge and difficult to overview. Harmonisation and standardisation efforts, regarding management [1][2], tools [3][4][5] and formats [6][7][8][9][10] are attempts to facilitate the handling and the interpretation of this information. However, to fully make use of harmonisation of industrial environmental management activities and information formats there is a need also to have a common understanding of how descriptions of industrial activities are created, what they mean, and how they should be communicated. This paper describes PHASETS (PHASEs in the design of a model of a Technical System). PHASETS is a procedural description for how a model of a technical system is compiled. PHASETS is one specification of the general model PHASES (PHASEs in the design of a model of a System), which describes from a top down or bottom up perspective how information about a system model is acquired, communicated, aggregated, and reported [11]. In one end the definition of the measured entity, and in the other end the reporting-result. The description of PHASES, and implicitly PHASETS has a number of practical implications. The procedural description is divided into six separate phases, distinguishing separable work tasks. This view of PHASETS can be used as a template for an organisational schema for environmental information management. PHASETS is also a communication model, which describes the paths traversed by information within the information system. The PHASETS procedural description includes all tasks performed when modelling a technical system from a

1


general perspective, hereby supplying a common conceptual frame aiding the apprehension of an otherwise complex idea. A common conceptual frame is also a necessary foundation for a shared reference model for design of industrial environmental information systems and information management routines. PHASETS is intended for different applications, such as: • Design of routines for generation of reliable and credible industrial environmental information and data • Design of routines for reliably communicating credible industrial environmental information and data • Estimation and control of costs for industrial environmental data acquisition • Development of a consistent security system for corporate environmental information management • Define and enable data transparency • Achieve data quality The use of PHASETS in these applications is further discussed in section 4.

2 Environmental data on technical systems Environmental management concerns management of the environmental performance of technical systems, and therefore needs information that describes the managed technical systems. For example, Environmental Management Systems, EMS, and the inventory part of Life Cycle Assessment, LCA, generate vast amounts of information describing environmental aspects of technical systems, e.g. data on physical flows or environmental performance indicators (EPI:s) [5]. Depending on the management’s responsibility, the managed systems are different, and they are differently modelled and described. However, the managed systems has similar information needs, in that they require information about the environmental impact from industrial activities (figure 1): A. They consider the scope of the technical system, i.e. the technical system components and system boundaries. B. They consider the flows of environmentally significant matter and energy into and out of the technical system, i.e. the in- and outflows. A. Technical system components and system boundaries

C. Functional flow, or other functional unit B. In- and outflows

Figure 1. A model of an environmentally managed technical system.

For the example of EMS and LCA the differences in the information required is that LCA considers also the function of the technical system. This is in figure 1 expressed by C, the functional flow, or other functional unit. The system boundaries and the system content for models applicable in EMS are generally different from models applicable for LCI. When modelling a technical system applicable for EMS, the models are drawn to include the

2


management’s organisational units, while models useful for LCA are drawn to include technical subsystems that are associated with a specific function, such as the production of one or more products or services.

3 PHASETS - PHASEs in the design of a model of a Technical System 3.1 Overview When designing a model of a technical system, modelling actions are taken, which may be more or less conscious or deliberate. Throughout the modelling different information about the observed system is used, such as results from measurements and estimates, descriptions of processes etc. The quality of the resulting model, and implicitly its usefulness for decisionmakers depends on both the modelling actions taken and on the quality of the information used in the modelling. Thus, to assess the quality of such a model one should assess the quality of both the information used for the modelling, and the quality of the modelling actions. The PHASETS model structures both the generatio n of the information used in the modelling, as well as the modelling actions. The PHASETS model describe the path that data management should follow while generating a model of a technical system, from defining measured entities, through the modelling, to the final communication of the result. The path is divided into six phases, each of which is self contained and distinctly separated from previous and subsequent phases. Within the phases, well-defined distinguishable tasks are performed, which requires a homogeneous input and which supplies a meaningful, homogeneous and interpretable output. Communication between each phase is stressed in the PHASETS model, and is defined and described generally, as one separate and generic phase. Each phase may in itself include many tasks, complex or simple. To keep the model at a general level, it does not allow for further sectioning of the existing phases. The tasks within each phase may, however, be described in detail in specific implementations. It is also likely that in different information systems and organisational routines some of the phases are merged or grouped differently. The structure, i.e. the formatting of the actual information handled within and communicatded between the phases is not covered in this paper. It should, however, be recognised that in order to enable data quality management, one needs both a format for communication and storage of the information and a structured way to handle information. The PHASETS model provides a structured way to handle the information, an in any given application of PHASETS a format has to be chosen or specifically developed.

3


3.2

The PHASETS model

PHASETS is described as a procedural and conceptual model based on communication and information aggregation. Data and information pass through phases, where sets of simple or basic concepts are communicated to formulate more aggregated and complex concepts within each phase.

Information aggregation

55 Communicating Communicating information information between between different contexts 44 Aggregating Aggregating models models of of technical technical systems systems 33 Synthesizing Synthesizing aa model model of of aa technical technical system system 22 Forming Forming a frequency frequency function function from a set of from a set of sample sample values values

3,7 g NO 2

11 Sampling Sampling an an individual individual value value

3,7 g NO2

00 Defining Defining an an entity entity for for aa selected selected parameter

Figure 2. The phases in the PHASETS model

Following the direction information takes for reporting, PHASETS may be described sequentially as (figure 2), from bottom up: 0. Defining an entity for a selected parameter; The choice of entity to measure and the setting up of the measurement system defines the simplest concept; i.e. the meaning of a measured value. 1. Sampling an individual value; The sampling results in a value for the simplest concept, i.e. a measured value 2. Forming a frequency function from a set of sample values; The frequency function aggregates sets of measured values into statistically expressed concepts. 3. Synthesizing a model of a technical system; The systems synthesis further aggregates the frequency functions from phase 2 into structured models of technical systems. 4. Aggregating models of technical systems; The models of technical systems synthesised in phase 3 may be aggregated into complex concepts describing e.g. averages or cradle to gate systems 5. Communicating information between different contexts; between any two phases 0-4 the resulting data and information, is communicated from the generator to the consecutive phase. The six phases in the PHASETS model will be described in detail in the following sections. The model is approached top down, the way it will be seen from e.g. a decision-maker.

4


3.2.1 Communicating information between different contexts, Phase 5 This section refers to phase 5 in the PHASETS model, see figure 2. Information is correctly communicated between different users within different contexts if the receiver understands the meaning of the information as it was intended by the sender [12]. Meaning is preserved during communication if the message contains enough information for the meaning to be reconstructed by the receiver. Considering that sender and reciever has differetn contexts the message must bridge the differences in terminology, language and implied meanings between the context of the sender and of the receiver. For example, literally equal terms may address entirely different concepts within the receiving contextual environment, implied meanings may not be identified, and not intended meanings might be imposed. The sender should be aware of this and design the message accordingly. This is general for communication between contexts, and is addressed in organisational communication theory [13] and cognitive theory [12]. It is also the key issue for standardisation work [14]. Practically, there are three ways of communicating meaning during contextual transfer of information: 1. Documentation Supplying a rich and sufficient documentation using terms and language shared by the communicating parties. Such may be the result from standardisation, common education and commonly addressed dictionaries. 2. Dialogue The information receiver may ask the information supplier for clarifications of meaning. To ensure correct use, the supplier may also review the receiver’s handling of the information. 3. Implicit When observing or participating in the actual data generation communication is implicit, i.e. the receiver implicitly retrieves information as experience. The current lack of documentation of environmental information indicates that present communication is made as dialogue or implicit. These forms generate no documentation that can be addressed or referred to after the communication is closed. This results in difficulties in terms of quality control, review, and assessment. As most informal communication dialogue and implicit communication are expensive since they require personal contact between sender and receiver, as well as reconstruction of facts each time they are needed. Formal communication should primarily be based on documentation, which enables effective and independent information review and retrieval. This paper only addresses this type of communication, i.e. it will from here on be assumed that communication between each phase in the PHASETS model is based on written documentation, e.g. electronic or paper based. 3.2.2 Aggregating models of technical systems, Phase 4 This section refers to phase 4 in the PHASETS model, see figure 2.

5


Models of technical systems (see section 3.2.3) can for different purposes be aggregated into new models of technical systems, e.g. when compiling the product system for a life cycle inventory (LCI) or when compiling different types of averages of technical systems. These compiled aggregates are new models of technical systems, with specific purpose, properties, functions and scopes. As examples, two important types of aggregates are discussed; flow charts and averages. 3.2.2.1 Flow chart modelling

Flow chart aggregation

Figure 4. Aggregation of technical system models by a flow chart modelling.

Flow chart modelling means aggregation of different system models into a flow chart (see figure 4). Examples are process flow charts for plant analyses and LCI product systems. Depending on the use of the model different modelling methodology and mathematics are applied [15] . 3.2.2.2 Averaging

+

+

Averaging 3

Figure 3. Aggregation of technical system models forming an average technical system model.

Averaging means mathematical aggregation of similar models of technical systems into a new virtual model, representing an average system model. The modelling includes specification of the scope and requirements on the included system models and the mathematical summation of frequency functions of flows (see figure 3). Mathematical treatment for formation of averages requires each of the summed models to be equal with regard to a set of specified physical and mathematical aspects [16]. Examples of averages are: • Industrial averages - different technical systems Industrial averages are compiled by averaging different but physically similar models of technical systems into a new system model. Industrial averages may describe a specific sector, market or product.

6


•

Interval averages - the same technical system Interval averages are compiled from system models describing the same physical technical system during different intervals e.g. time intervals or product volume intervals. The resulting system model is a model of a technical system describing a specific time or product volume interval. An example of such an average is a corporate annual environmental report describing the environmental performance of a plant.

Interval averages differ from industrial averages in that they include averaging over the same physical technical system, while industrial averages are compiled from models of different physical technical systems. 3.2.3 Synthesising a model of a technical system, Phase 3 This section refers to phase 3 in the PHASETS model, see figure 2. A model of a technical system is originally defined from a synthesising procedure e.g. by synthesising results from detailed analyses of a real technical system, or by synthesising estimates of parameters into a coherent model [17]. A technical system model is synthesised for a specific purpose from which it derives all properties [18]. A model of a technical system has well-defined properties, i.e. a system model scope limited by boundaries, and an interface describing the interaction between the system and its environment, cf. figure 1. In practice, such systems are not independent of the observer, i.e. the modeller, which means that a model of a technical system does not necessarily describe a physically existing and separable technical system [19]. When communicating information describing a synthesised technical system model special care should be taken to communicate information about the system boundaries and the system content and of any modelling made to e.g. transform input data on flows supplied from phase 2 (see section 3.2.4). 3.2.4 Forming a frequency function from a set of sample values, Phase 2 This section refers to phase 2 in the PHASETS model, see figure 2. As basis for synthesising a model of a technical system, as described in section 3.2.3, the modeller needs to have numerical data on relevant parameters, prepared in the form of statistically expressed numerical values. Results from measurements, e.g. sets of sample values, are generally not useful in the form in which they are originally recorded. Instead they are aggregated into statistical parameters, which describes frequency functions. Such functions statistically describe trends and levels of the measured entities. To acquire these statistical parameters, statistical data analysis is applied to the sets of sample values. The choice of statistical methods is not arbitrary but is inherent from the measurement system, which defines the meaning of the measured values (dealt with in phase 0, below) and the sampling process that generates the sample set (dealt with in phase 1, below) [20].

7


When communicating information that describes a frequency function, special care should be taken to clearly communicate the interval for which it is valid, as well as any relevant discrepancies during that interval. 3.2.5 Sampling an individual value, Phase 1 This section refers to phase 1 in the PHASETS model, see figure 2. The origin of data about a physical entity is an observation, e.g. a sampling of an individual value from a measurement system. A measurement system is the systematic monitoring of one physical entity [18]. For example, it may monitor a total emission outlet by recording each individual sampling result. Data may also be the result from e.g. estimations based on experience, retrieval of data from literature or the execution of a mathematical model. Also these recordings are here referred to as sampling. When communicating a sample value it should be supplied with documentation of whether the circumstances under which the sampling were made were in accordance with the specifications of the systematic of the measurement system or if any deviations from the specified sampling routines were made. 3.2.6 Defining an entity for a selected parameter, Phase 0 This section refers to phase 0 in the PHASETS model, see figure 2. This phase defines the origin of data on any measured entity in PHASETS. Unless an unambiguous definition of this entity is established there is little meaning of communicating information about it. A measured entity originates from a measurement system, which is designed with regard to what aspect to control by the resulting information [18], e.g. control of the production with regard to legal aspects or environmental market requirements. The design of the measurement system should also consider whether the entity should be correlated with other measured entities. For example, if the amount of emissions should be correlated with the production rate, the measured entities and the measurement systems must be synchronised. A measurement system may be simple, e.g. a computer recording quantitative properties of occurrences of the entity being measured, or it may include several steps of sample analysis and data treatment before one sample value can be recorded. That value may in itself be described in terms of probability etc. The meaning and the interpretation of a measured entity are equally important for estimates, data taken from literature, or results from mathematical models. When documenting a measurement system, special care should be taken to ensure that known relevant aspects under which the measurement system is valid are clarified, e.g. measurement

8


ranges and uncertainty intervals, as well as correlation with other measurement systems and entities should be clearly described.

4 Applications In this section some applications of the PHASETS model for industrial handling and communication of environmental information and data will be described. The PHASETS model may be applied as a structuring aid for: •

Design of routines for generation of reliable and credible industrial environmental information and data Business requires effective environmental information management, i.e. the environmental management needs a reliable decision base and the market requires credible environmental market statements. Business competition requires efficient management i.e. the routines for the environmental information and data needs to be efficiently integrated with the established corporate routines. PHASETS is an organisationa l reference model when identifying the data management routines that need to be made reliable.

•

Designs of routines for reliably communicating credible industrial environmental information and data As environmental information and data are communicated from original source to final reporting, the data and information passes through many organisational units. The reliability of the data and information is affected by different expertise within the involved organisational units. In the communication, there is a risk that misunderstandings may introduce unpredictable and untraceable errors in the final results. Consequently, there is a need for a reliable quality assurance of the environmental information management. The PHASETS model not only the activities involved in the handling of environmental information, but also the communication between the different activities. If PHASETS is used for the design of information management routines it will also facilitate identification of areas where communication errors may be introduced.

•

Estimation and control of costs for industrial environmental information and data acquisition Today companies face a number of parallel environmental information acquisition duties, e.g. authority or corporate environmental reporting and life cycle inventory (LCI) data requests from customers. With an integrated strategy for these duties, costs may be shared with other duties and reduced by allocating costs to already implemented activities. By applying the PHASETS model, the different activities involved in the handling of environmental information can be clearly defined and distinguished, and a structured overview of the entire corporate environmental information system may be obtained. A result from this is that the costs for these activities may be estimated, budgeted and controlled.

•

Development of a consistent security system for corporate environmental information management Environmental business requirements put demands on new potentially sensitive information to be acquired, compiled, stored and communicated within the organisation. This leads to emergence of a new corporate information system, which needs to be designed as thoroughly as e.g. the economical information system, to meet security management standards. PHASETS structuring of internal and external environmental

9


information management supplies support to such a thorough design process. All users, communication channels and access roles may be well identified, which allow for a wellstructured secrecy control. •

Define and enable data transparency Credibility is based on openness, and to achieve openness concerning industrial environmental information, one must enable review of information and data. Review requires transparency. Transparent information can only be achieved with a structure supporting transparent information management. PHASETS is such a structure, which enables review at all levels of data and information handling and communication.

•

Achieve data quality Considering quality as a measure of degree of fulfilment to a specification, an environmental management system designed on the basis of the PHASETS model will enable achievement of any data quality specification. The paths through which the data quality specifications are broken down and distributed to the responsible persons are given by PHASETS.

It should be recognised that the phases 0-3 may be performed differently with higher usability in the consecutive phases by applying statistical experimental design methods [18].

5 Implementing PHASETS PHASETS is first of all a conceptual model describing a procedural path. It requires interpretation for each specific application, but then serves as a common basis for understanding for the different actors. Users and roles, organisational units and communication channels, methods and equipment can be structured in an efficient and effective way in accordance with the phases of the model. Practical work with implementing PHASETS in e.g. the Swedish forestry, pulp and paper industry shows that when implementing PHASETS the information requirements of the resulting reports should first be investigated and well defined. It is natural to then consequently continue down through the model, starting with defining the models to acquire information about. In phase 4 the required system model units are identified. In phase 3 these models are defined, including the system content, its boundaries and its environmentally relevant parameters. From phase 3 it is practical to continue directly to phase 0, to identify or implement measurement systems for relevant parameters, as well as to formulate sampling routines. Phases 0 and upward are then a practical execution of the PHASETS model, which follows the preparing steps (definitions of phases 5, 4, 3, 0) of the implementation. Each phase is connected using the rules of communication and preservation of meaning described in phase 5.

6 Conclusions The PHASETS model aids in the identification of mutual models for environmental information managment for different applications, e.g. EMS monitoring and LCI data acquisition. With PHASETS companies may considerably increase the capacity for their voluntary environmental control, at e.g. site, corporate, or life cycle levels. The PHASETS model also enables quality control of environmental information management, and if fully

10


implemented throughout a supply-chain, it enables comprehensive data transparency for entire LCI studies. Due to choice of organisational structure the acquisition of data for use within the EMS is often separated from the generation of data for LCA purposes. This is true for companies not being streamlined for market orientation, i.e. not being product flow oriented or for companies which have formulated environmental goals for their organisation but not for their products. Within such organisations LCI data is not generated within the EMS routines. LCI data acquisition and generation is instead run as separate projects on demand. Regardless of routines organisations may still apply the same general structure for information management for e.g. EMS and LCI as formulated by PHASETS. PHASETS reduces the complexity of information describing models of technical systems, by structuring the data handling, instead of removing information. By practically connecting final reports with measurement systems and intermediate data treatment, PHASETS serves as a model for ideal transparency of industrial environmental data and information.

References [1] International Standard ISO 14001:1996(E) Environmental Management Systems Specifications with Guidance for Use [2] The Council of European Communities, Eco-Management and Audit Scheme (EMAS), The Council Regulation (EEG) No 1836/93, 1993 [3] International Standard ISO 14040:1997(E) Environmental management - Life cycle assessment - Principles and framework [4] ISO/TR 14025:2000(E) Environmental labels and declarations - Type III environmental declarations - Guiding principles and procedures [5] International Standard ISO 14031:1999 Environmental management - Environmental performance evaluation - Guidelines [6] Carlson R, Steen B, Tillman A-M, Löfgren G. LCI data modelling and a database design. Int. J. LCA 1998: 3(2) 106-113 [7] Singhofen A, Hemmings CR, Weidema BP, Griesel L, Bretz R, De Smet B, Russel D. Life cycle inventory data: development of a common format. Int. J. LCA 1996; 1(3): 171-178 [8] International Standard ISO 7168-1:1999(E) Air quality - Exchange of data - Part 1: General data format [9] ISO/CD 14048 Environmental management - Life cycle assessment - Life cycle assessment data documentation format [10] Sustainability Reporting Guidelines on Economic, Environmental and Social Performance, Global Reporting Initiative, Boston, 2000

11


[11] Carlson R, Pålsson A-C. PHASES - Information models for industrial environmental control. CPM-report 2000:4, Chalmers University of Technology 2000 [12] Ellis H, Hunt R. Fundamentals of Human Memory and Cognition, Fourth edition. Dubuque, Iowa: Wm. C. Brown Publishers, 1989 [13] Kreps G. Organizational Communication Second edition. New York: Longman, 1986 [14] Hawkins R, Mansell R, Skea J, Elgar E (ed.). Standards, Innovation and Competitiveness: The Politics of Standards in Natural and Technical Environments. Aldershot, 1995 [15] Forsberg P. Modelling and Simulation in LCA, CPM-report 2000:1, Chalmers University of Technology, 2000 [16] Klir G. Facets of Systems Science. IFSR International Series on Systems Science and Engineering, Vol 7. New York: Plenum Press. 1991 [17] Singh M (editor in chief). Systems & control encyclopedia theory, technology, applications. Vol 2, first edition, Oxford: Pergamon, 1987 [18] Turner K, Hicks C. Fundamental concepts in the design of experiments. Fifth edition, New York: Oxford University Press Inc. 1999 [19] Checkland P. Systems thinking, systems practice. Chichester: John Wiley & Sons Ltd., 1995 [20] Rice J. Mathematical Statistics and Data Analysis. 2nd ed. California: Duxbury Press, 1995

12

Carlson R., Forsberg P., Dewulf W., Ander Å., Spykman G., "A Full Design for Environment (DfE) Data Model", PDT Europe 2001, April 24th-26th, 2001, pp. 129-135.

A full Design for Environment (DfE) data model Raul Carlson and Peter Forsberg, CPM, Chalmers University of Technology, SE-412 96 Göteborg, Sweden Wim Dewulf, KU Leuven, Mechanical Engineering Department, Celestijnenlaan 300B, B-3001 Heverlee, Belgium Åsa Ander, Adtranz Sweden, Dept. ICT/K, SE-721 73 Västerås, Sweden Gerold Spykman, GEP Gesellschaft für Entwicklungsberatung & Produktrecycling mbH, Olympiastrasse 1, D-26419 Schortens, Germany

1. Introduction A data model covering all product data and information needs for Design for Environment (DfE) of rail vehicles (trains, trams) has been designed by the project RAVEL (RAil VEhicLe eco-efficient design). The data model is referred to as the RAVEL information platform [1]. RAVEL is a Brite EuRam project partly (50%) financed by the European Commission. The participants in the RAVEL project are ABB, Adtranz, Chalmers, DSB, GEP, KU Leuven, , SJ and Woodville Polymers. For more information please visit the web-site of the project [2]. The RAVEL project started in November 1998 and will end in October 2001. The RAVEL project develops a methodology, together with web-based software tools with a knowledge base to enable improved eco-efficiency of train vehicles for the entire rail industry, including suppliers and sub-suppliers, manufacturers, operators, and scrappers. The environmental performance of the rail vehicle is considered from a life-cycle perspective, i.e. the environmental performance indicators cover resource use, emissions, waste generation, toxicity, energy use, recycling, etc. expressed in terms of design aspects like choice of material and e.g. properties of a supplier of components. The RAVEL information platform is a standard data structure for data storage and data communication of DfE for rail vehicle design. It is intended for software and information systems development. The data and information needs for DfE for rail vehicles include • a knowledge base consisting of purchaser and designer guidelines • good examples of rail vehicle designs • help for the users of the web-based information system • definition of eco-efficiency indicators for the rail design • targets for design projects • material and component lists for rail vehicles • environmental and economical life cycle data • etc. The identification of the rail vehicle DfE data and information needs and the design of the data model for the RAVEL information platform were performed in the RAVEL project during 1999 and 2000. The work started with a thorough investigation of the user requirements at especially Adtranz, SJ, DSB and Woodville Polymers in 1998. In early 1999 a first draft RAVEL information platform was developed during a three days workshop consisting of experts from Adtranz, GEP, KU Leuven, ABB, and Chalmers (figure 1). The system architecture for the RAVEL information system was sketched during the same time, from identification of anticipated and required functionality, prototype designs, and a specification of the scope of the information system (figure 2). In the beginning of 2000 the specification of the RAVEL information was sufficiently well known to finalize the RAVEL data modeling in a few months of efficient work. The data model is expressed as a relational database schema and an EXPRESS model.

1(7)


2. The Scope of the RAVEL Information platform The RAVEL information platform includes all data and information necessary for a systematic DfE process for rail vehicles, considering e.g. material contents of the product, life cycle assessment (LCA) data [3],[4], life cycle cost (LCC) and environmental design advice. The scope can be described in two ways; in terms of which information to provide (the information platform) and in terms of the scope of the information system called the RAVEL workbench. The two will be described in the following.

2.1. The RAVEL information platform

4 3

Project target Knowledge system

1

2 Project

5 Target setting

Product

Design

Environment

Indicator

Figure 1. High-level model of the RAVEL information platform, including major concepts and information areas. Figure 1 shows a high level model of the information platform data model based on the analysis of user requirements from Adtranz, Woodville Polymers, SJ and DSB. The model includes the main concepts of the information platform, and it provides a key to describe the overall RAVEL methodology and the functionality of the RAVEL information system and the software, the RAVEL workbench. The numbers in figure 1 represent; 1. Project management data, 2. Basic database including environmental data about products, described as design indicators, 3. Knowledge base including guidelines and help system, 4. Project target including quantitative goals for a design, 5. Information and data to configure tools supporting the RAVEL methodology and the design activities, e.g. target setting and assessment of a design. From figure 1, one may derive that a rail vehicle design (5. Design) is made within a project (1. Project), that a design is related to a general product (2. Product), that a product has environmental properties called indicators (2. Environment, Indicator) and that quantitative results of the design indicators may be compared with the project target (4. Project target). Project targets may be set for each project (5. Target setting). The knowledge base provides knowledge, guidelines and help to the user during all tasks of the DfE process (3. Knowledge system).

2(7)


In the following all five information platform elements are presented individually. 1. Project management Projects are described with sufficient detail so that a more comprehensive project description may be found outside of the RAVEL information platform. Users are identified so that different user profiles can be set, and so that different roles can be assigned to different users, e.g. designers and project managers. 2. Basic database Products may be described from a number of different viewpoints that are relevant to DfE. Any component can be hierarchically constructed and decomposed into an arbitrary depth of lowest level materials. The deepest depth is defined by the RAVEL material list, a user configuration describing materials for which environmental properties may be described. Except for enabling the hierarchical modeling of components up to full rail vehicles, it is also possible to describe the linkage between different assembled or joined sub-components, e.g. whether they are glued, welded, adhered on surface, etc. to each other. An arbitrary number of different product life cycle scenarios can be ascribed to any material or component, e.g. a cradle to gate scenario based on an actual component provider, a generic cradle to gate scenario, a general full life cycle scenario, or different use phase or end of life scenarios. The life cycle scenarios may consider any life cycle modeling necessary to evaluate the eco-efficiency of a rail vehicle, as described by e.g. life cycle assessment (LCA) according to ISO 14040 or life cycle cost (LCC). It is a free choice to decide whether the RAVEL methodology should include a detailed LCA or LCC assessment for each design assessment, or whether aggregated LCA or LCC values should be assigned to materials and components. The basis of the material and component data modeling has been mapped and copied from the SPINE data model [5][6][7]. 3. Knowledge base The knowledge base of the RAVEL information platform allows for structuring of four different types of documents; texts in text-fields in the database of the information platform, arbitrarily formatted electronic documents in the same computer environment as the RAVEL information platform, arbitrarily formatted electronic documents in other computer environments, and references to paper copies such as library books, files or reports. All documents in the knowledge base, regardless of their location, can be contextually related to any part of the rest of the information platform. A user wanting to know more about, e.g. a composite found in the basic database may find all documents not only within the database of the RAVEL information platform, but also in e.g. physical libraries within the actual organization or anywhere else that the database maintainer has provided a reference to. 4. Project target The project target is a small part of the RAVEL information platform, but it is crucial. It makes it possible to store the design target that must be met for each component level of the rail vehicle. The overall target may be set by the customer already during the call for tender, and it may be adjusted during the design project. The overall design target may be broken down into sub-targets so that designers of e.g. the break system need not to take into consideration dimensions of the target that is not relevant for their task, or so that the different sub-component designers can be made responsible for different shares of the overall target. 5. Tools supporting the RAVEL methodology The scope of the RAVEL DfE methodology includes all processes from the procurement stage at the railway operator, throughout the design at the vehicle manufacturer, and upstream to sub-suppliers of components for rail vehicles, etc. The tools included in the RAVEL methodology for these different users are different software for managing the data and information in the database(s), as well as assessing the eco-efficiency performance based on indicator-values of a design. The software also enables comparison to target and presentation of contextually correct documents from the knowledge base.

3(7)


The RAVEL information platform is designed so that it should be possible to design many different software tools for different users and tasks, as well as to store different types of software configuration data within the platform.

2.2. The Scope of the RAVEL Information System - the RAVEL workbench In the RAVEL project plans, the RAVEL information system was called a workbench, the RAVEL workbench. This name refers to a software system with a user interface for many different DfE purposes. During the project the RAVEL workbench has been designed as a web-based information system, with a well-defined database structure (the RAVEL information platform) and information environment and with the possibility to easily design different user interfaces for different users, tasks and organizations. This possibility to further develop the workbench stresses the need to define the scope of the information system, i.e. what is included with the workbench and what is not included, and how to exchange data between the RAVEL information platform and its data environment. RAVEL Workbench

Agent

RAVEL Knowledge system

Design project members

Project management

RAVEL Function dictionary Target setting

LCC

Agent

Design

Measure design performance

''Stripper' Functions for measurement of performance indicators

Picasso External information source containing LCC data (process chains/aggregated systems with e.g. costs)

''Stripper' LCI/A

''Stripper'

External information source containing LCI data (process chains/aggregated systems with e.g. resource use, emissions, waste generation)

RAVEL basic database

External information source containing product design data, e.g. CAD, PDM, …

RAVEL Data dictionary

Figure 2. The scope of the RAVEL information system - the RAVEL workbench. Figure 2 describes the scope of the RAVEL workbench [8]. The RAVEL knowledge system, the RAVEL function dictionary, the RAVEL basic database and the RAVEL data dictionary are all parts of the RAVEL information platform. Within the definition of the RAVEL workbench are also tools for accessing the correct documents from the knowledge system (agents), tools to support target setting, to measure design performance, and tools to sketch simple abstracts of DfE models of train vehicle designs (picasso). Project management and the real design of the vehicle is not made with tools provided by the RAVEL workbench, since those tools already exists within the organizations where the RAVEL workbench will be used, and are better developed by other competencies. Instead the workbench and its information

4(7)


platform is designed to match with its neighboring information systems by e.g. having space for addresses and identifiers to projects and project members, and design drawings. The workbench could however offer support for administrating DfE responsibilities. The RAVEL workbench is also designed to handle information from outside of its scope, by importing necessary information from other tools, e.g. material and component structures from CAD and PDM systems, and life cycle data from LCA or LCC tools. The general strategy and principle to enable such import is the data model and the structure of the RAVEL information platform. By having a clear and concise structure within the boundaries of the RAVEL information system, it is possible to unambiguously define the mapping from any other well-structured information system. In figure 2 the 'strippers' should be understood as different software modules importing parts of the information from other files or databases through such well-defined mapping schemes that are defined specifically for each different external data source. The principle for the scope of the RAVEL workbench can be summarized as that it consists of information management tools for the intersection of all information necessary to perform a comprehensive DfE of a rail vehicle. However, it does not include any of the tools that are necessary to produce intermediate results e.g. LCA, LCC, DfD (Design for Disassembly). If suitable tools of such kinds do not exist elsewhere, the RAVEL information platform allows for them to be added, and is designed to store the required information.

3. Modeling languages It should be possible to develop industry standards from the results of the RAVEL project, as well to make use of standards and standard technology for the implementation of the RAVEL workbench. It was therefore decided that the RAVEL information system should be based on well-established technologies with independence from specific vendors. The information platform was first designed as a relational database structure (ANSI SQL) [9], so that anyone implementing the workbench should be free to chose any of the different relational database systems available on the market. The actual implementation for the prototype system developed in the project was based on Microsoft Access, since that made it easy and economical when distributing files describing the database definition. An upscale to any other relational database system (ORACLE™, Sybase™, SQL Server™ etc.) is a matter of configuration for a specific industrial computer/software environment. In addition to describing the information platform as a relational database schema, special effort was made to also describe the model in the EXPRESS language (ISO 10303)[10]. This effort was assisted by the Norwegian EXPRESS consultant agency EPM. The choice to define both a relational database structure and an EXPRESS model was based on the understanding that it will be necessary both to have a vendor-independent solution for the database structure, and another vendor-independent definition for the data communication within the system and with its data environment. The EXPRESS modelling effort was made with the original intention to make use of standardised part libraries in the STEP standard, but it was soon found that such an ambition was well beyond the scope of the research and development project RAVEL. It was therefore decided that the effort should be focussed only on describing the relational schema using the EXPRESS syntax. This choice led to that the current EXPRESS model of the RAVEL project is not object-oriented but relational. This has an immediate benefit when mapping between communication files and the relational RAVEL information platform, since a mapping towards the EXPRESS model also is mapping towards the database definition. However, due to the differences between object oriented and relational paradigms, it may also introduce difficulties when mapping towards object oriented data structures. The differences between the relational and the object-oriented paradigms are most apparent in regards of identification of data sets and sub-sets. Whilst the object-oriented paradigm has an implicit referential identification of objects (is, has, etc.) the relational paradigm use primary keys and foreign keys for the same type of identification. This difficulty was noted, but a further development of a full object-oriented and STEPcompliant specification of the data communication formats is left for future efforts in ISO/TC 184.

5(7)


During the project, the data model has been implemented and used in the form of a relational database. It has been tested both in regards of how well the model fulfils the needs of DfE for rail vehicles (as the web-based RAVEL workbench) and in terms of how well it is suited for communicating data from tools external of the RAVEL workbench. The tests and the prototype are very promising and there is also a clear interest for the industrial partners of the project to start using the result in their operational DfE work.

4. Domain compatibility and SPINE The RAVEL workbench is designed to work for many different users and to be compatible with many different tools in many different working-environments. It supports the train operators' call for tender, the manufacturers' assessment of a conceptual design and relationships with sub-suppliers, as well as the detailed design and the final reporting of the environmental performance to the customer, the train operator. The different tools that the RAVEL workbench is designed to work with are e.g. LCA tools, materials selection tools, black and gray lists, LCC tools, CAD tools, and PDM systems. The RAVEL information platform is designed to be compatible with these different environments through its EXPRESS data model. By having the open and well-defined RAVEL data model including all necessary information items, it is possible to match any external file format with relevant data to the information platform and to design translation software for the data communication. Thereby enabling compatibility between tools. In this way the RAVEL data model does not require any specific data structure in the tools that support a systematic DfE, e.g. LCA, LCC, DfD, etc. It does however require open formats and models and rely on the mapping between open formats used in external tools onto the explicit and open structure of the RAVEL information platform. In this way, by supplying an open model of the information platform the project result enables that anyone with sufficient modeling competence can map the data from other necessary tools available on the market, with only a reasonable effort. For the LCA-part of the RAVEL Information platform, the SPINE (Sustainable Product Information Network for the Environment) has been used [5][7]. The SPINE model has been practically tested and industrially applied since 1995. SPINE is also the initial contributor to the standardization of ISO 14048, an LCA data documentation format [4], which is based on the LCA standards in the ISO 14040-series [3].

5. Conclusion and further development The tests that have been run within the RAVEL project show that the RAVEL information platform is a good solution for handling environmental product data as well as other supporting information for a DfE process. It is also likely that the general approach that has been considered during the development will make the data model useful also for DfE in other industrial sectors. Contacts with various organizations have shown that there is a wide interest for this type of solution outside the RAVEL project and the railway industry. The project will hand over the results towards a standardization effort, firstly for wide spread usage within the rail sector but also, in e.g. ISO/TC 184 in order to make the information platform data model a STEPcompliant solution. In this way the future maintenance of the model will be ensured at the same time as a smaller DfE-specific part could be needed to complete the STEP parts libraries. It is of great importance that commercial software vendors find a business interest in the development of tools for the RAVEL methodology, and that the software are also compatible with the RAVEL information platform. Such tools may be software for material selection, LCA, LCC, DfD, etc. At the moment the RAVEL workbench is tested in different situations within the RAVEL consortium. These tests will be followed by demonstration and verification sessions within the industrial member companies of RAVEL. The tests will lead to more requirements and demands on how to make full use of the RAVEL information platform. However, it already looks promising, this new way to manage industrial DfE projects.

6(7)


References [1] Carlson R. et. al., The RAVEL information platform data model, RAVEL internal document number CPM-000919, September 2000 [2] http://www.ravel-project.de [3] International Standard ISO 14040:1997(E) Environmental management - Life cycle assessment Principles and framework [4] ISO/CD 14048 Environmental management - Life cycle assessment - Life cycle assessment data documentation format [5] Carlson R., Löfgren G., Steen B., "SPINE, A Relation Database Structure for Life Cycle Assessment", Report B1227, Swedish Environmental Research Institute, Göteborg, September 1995 [6] Carlson R., Steen B.; "A Data Model for LCA Impact Assessment"; Presented at 8th Annual Meeting of SETAC-Europe 1998 14-18 April; Bordeaux [7] Carlson R, Steen B, Tillman A-M, Löfgren G. LCI data modelling and a database design. International Journal of Life Cycle Assessment 1998: 3(2) 106-113 [8] Karlson L. et. al., Final report subtask 2.1, RAVEL internal document number ABBCRC/D99/460, December 1999 [9] ISO/IEC 9075:1992, Information Technology Database Languages SQL, or ANSI X3.135-1992, Database Language SQL [10] ISO 10303 - Industrial automation systems and integration - Product data representation and exchange, Parts 21 and Parts 11

7(7)

OMNIITOX

Opening

OMNIITOX: Opening

OMNIITOX Concept Model Supports Characterisation Modelling for Life Cycle Impact Assessment Raul Carlson, Maria Erixon, Ann-Christin Pålsson and Johan Tivander* Industrial Environmental Informatics, Chalmers University of Technology, SE-412 96 Göteborg, Sweden * Corresponding author ([email protected])

DOI: http://dx.doi.org/10.1065/lca2004.08.168 Abstract

Goal Scope Background. The main focus in OMNIITOX is on characterisation models for toxicological impacts in a life cycle assessment (LCA) context. The OMNIITOX information system (OMNIITOX IS) is being developed primarily to facilitate characterisation modelling and calculation of characterisation factors to provide users with information necessary for environmental management and control of industrial systems. The modelling and implementation of operational characterisation models on eco and human toxic impacts requires the use of data and modelling approaches often originating from regulatory chemical risk assessment (RA) related disciplines. Hence, there is a need for a concept model for the data and modelling approaches that can be interchanged between these different contexts of natural system model approaches. Methods. The concept modelling methodology applied in the OMNIITOX project is built on database design principles and ontological principles in a consensus based and iterative process by participants from the LCA, RA and environmental informatics disciplines. Results. The developed OMNIITOX concept model focuses on the core concepts of substance, nature framework, load, indicator, and mechanism, with supplementary concepts to support these core concepts. They refer to the modelled cause, effect, and the relation between them, which are aspects inherent in all models used in the disciplines within the scope of OMNIITOX. This structure provides a possibility to compare the models on a fundamental level and a language to communicate information between the disciplines and to assess the possibility of transparently reusing data and modelling approaches of various levels of detail and complexity. Conclusions. The current experiences from applying the concept model show that the OMNIITOX concept model increases the structuring of all information needed to describe characterisation models transparently. From a user perspective the OMNIITOX concept model aids in understanding the applicability, use of a characterisation model and how to interpret model outputs. Recommendations and Outlook. The concept model provides a tool for structured characterisation modelling, model comparison, model implementation, model quality management, and model usage. Moreover, it could be used for the structuring of any natural environment cause-effect model concerning other impact categories than toxicity. Keywords: Characterisation model; concept model; informatics;

information system; OMNIITOX

1

The Need for a Concept Model and System Requirements in OMNIITOX

The OMNIITOX project is of a multidisciplinary nature. Life cycle assessment (LCA) practitioners, chemical regulatory risk assessment (RA) expertise, as well as expertise from industry, academia, and government bodies are all foreseen to make use of the project results (Molander et al., this edition). Different users have different needs and can use different terminology for the same or related concepts. The task at hand is to facilitate efficient and understandable communication of information in this versatile user environment. The main focus in OMNIITOX is on characterisation models in an LCA context, and the OMNIITOX information system (OMNIITOX IS) is being developed primarily to facilitate toxicological characterisation modelling and calculation to provide users with information necessary to manage and control industrial systems towards sustainability [1]. The modelling and implementation of operational characterisation models on eco and human toxic impacts developed in OMNIITOX require the usage of data and modelling approaches originating from the RA and related disciplines [2]. The ISO 14040 standard series provides a framework of concepts and terms suitable for LCA, but there is an additional need of a framework where data is consistently interchanged between contexts of natural system modelling approaches. The implementation of an information system puts specific requirements on the structuring of information as it is communicated between users via the technical and organisational components of the OMNIITOX IS. The terminology must not only be understandable by humans, but also translatable into a language interpretable by computers. A concept model is a collection of named and described concepts including their relations. The OMNIITOX concept model is intended to support the structuring of natural system models for industrial applications in general and for operational characterisation models in particular. Moreover, it should describe the meaning of all types of information that are used by all parts – humans and technical components – interacting in and with the OMNIITOX IS. In this way, the concept model can serve as an essential support in characterisation modelling, model comparison, model implementation, model quality management, and model usage as information is communicated between the components inside the OMNIITOX IS and between external data sources and receivers [3].

Int J LCA 9 (5) 289 – 294 (2004) © ecomed publishers, D-86899 Landsberg, Germany and Ft. Worth/TX • Tokyo • Mumbai • Seoul • Melbourne • Paris

289

Opening

2

Concept Modelling Methodology Applied in OMNIITOX

In OMNIITOX, a consensus based concept modelling methodology has been applied as the basis for developing the required multidisciplinary communication platform. The methodology also builds on the six database design principles as set up by Elmasri and Nanthe [4]. 1. 2. 3. 4. 5. 6.

Requirement collection and analysis Conceptual database design Choice of Database Management system (DBMS) Data model mapping Physical database design Database implementation

The first step – requirement collection analysis – concerned the definition of the scope and required functionality of the OMNIITOX IS. The participating organisations in the project are a cross-section of the intended users of the OMNIITOX IS with representatives of LCA practitioners, RA, as well as expertise from industry, academia, and government. In addition, the international character of the OMNIITOX consortium adds the aspect of cultural and lingual differences also within disciplines. It has therefore been a crucial aim to give all participants the opportunity to contribute with their needs, ideas, and terminologies. In the second step – Conceptual database design – the analysis and the synthesis of the result from the first step were taken further to find a suitable concept model. In practice, this involved finding the essence in the field of study by analysing intersections and differences of related concepts used in the different disciplines. Initially a quite complex picture arises with ambiguities between the meanings of abstractions from different disciplines. Ontological principles are applied in order to create an understandable structure that sorts out what really is the object of study, i.e. the concept modelling aims at structuring information based on the physical causality found in the real world. The modelling has also applied the general design principles as described in the PHASENS model [5,6] to provide a baseline for model transparency and consistency. PHASENS – PHASEs in the design of a model of a Nature System – describes how information about changes in the natural systems (not being controlled by humans) are communicated and analysed. Additional important inputs to the concept modelling have been existing model frameworks including the toxicity impact assessment framework as described by SETAC working groups on impact assessment [7], as it has provided understanding of the state of the art in toxicity impact assessment modelling. Step 1 and 2 was iterated until a first prototype version of the concept model was achieved. The information system developers in the OMNIITOX project performed the analysis and compilation, and the resulting concept model was presented. In order to make the concept model interpretable by computers, the process continued by implementing the prototype concept model in a database through step 3–6. The Database Management System (DBMS) chosen is relational databases R-DBMS. This choice was made prior to the concept modelling process and was based on the fact that rela-

290

OMNIITOX tional databases are a well established and well understood standard technology DBMS [8]. It is tested in a vast number of applications, and previous experiences have shown the high applicability of R-DBMS in environmental information management [9–12]. At the fourth step, the prototype concept model was transformed into a relational data model where entities and relations are translated into tables, primary keys, and foreign keys. The relational data model was written down as Structured Query Language (SQL) statements in step five. In the final step, the database objects were created by executing the SQL statements through the R-DBMS. This overall process – step 1 to 6 – has been iterated until the user requirements are incorporated and the concept model follows R-DBMS rules. In this way, the OMNIITOX data format and the OMNIIITOX database were created simultaneously as the concept model was developed. The understanding and acceptance of the concept model has also increased in the iterative consensus process. 3

The Core of the OMNIITOX Concept Model

The design of the OMNIITOX concept model is based on the fact that a cause, an effect, and the relation between them within a defined system frame can be identified for all types of models of the natural system used in industrial applications. The concepts load, indicator, mechanism, nature framework, and substance are the cornerstones in the OMNIITOX concept model (Fig. 1). The arrows indicate relational dependency: an arrow that points from A to B means that A is dependent on B and requires the existence of B. In addition B is independent on A. The concepts load, indicator, substance, and nature framework are independent of other concepts, whereas mechanism requires the existence of the other four concepts. This is a key to the information structure of a cause effect model as it is not possible to define the relation between a cause and an effect without first defining what the cause and effect is. In addition to the core concepts, a number of concepts have been defined to support the core. This includes the structuring of substance property and nature property data, quality meta-data, units, reference documentation, mechanism algorithm, etc. A detailed description of the OMNIITOX concept model can be found in the OMNIITOX concept model and data format report [13].

Load

Mechanism

Indicator

Definition of cause

Relation between cause and effect based on substance and nature framework data

Definition of effect

Substance

Nature framework

Property data

Property data System boundaries

Fig. 1: The core of the OMNIITOX concept model. The arrows indicate relational dependency

Int J LCA 9 (5) 2004

OMNIITOX

3.1

Opening

3.2

Substance and nature framework

In OMNIITOX, a substance consists of physical matter, which can be an element, a chemical, a compound, a material, etc. A substance can be viewed in various contexts, such as in an emission quantity in an LCA study, or in a concentration in the natural system. However, the context in which a substance is viewed is not an intrinsic property of a substance. Substances have substance properties e.g. mass, colour, dissociation constant, degradability, acute toxicity to fish, etc. Definitions of substance properties are called substance property specifications. Measured data are connected to a substance and a property specification as a substance property value. The value may be dependent on conditions such as the temperature at the measurement. Fig. 2 shows relations between substance property data.

Load and indicator

The load concept refers to the cause, i.e. what is considered to trigger the mechanism and in turn has an impact on an indicator. The load is independent of the mechanism as the same load may trigger many mechanisms. By just looking at the load, a specific mechanism cannot be distinguished. An example of a load definition is shown in Fig. 4. An example of data in an LCI result

55 kg kg of of CO CO2 emissionto toair airininEurope Europe 2 emission

The LCI data is specified by nomenclaturised parameters

Unit

The definition of a load matching the LCI result data example

7664 Substance 50 -00-0 Phosphoric acid Formaldehyde H3PO4 CH 2O ...

13.3 2.13

298 298 Condition value

Acidity dissociation constant constant [-] Substance property specification

Temperature Temperature [K] [K] Condition specification

Fig. 2: Example of substance property data. A substance property value is defined by referring a substance and a substance property specification. In this case the substance with the names 50-00-0, Formaldehyde, and CH2O is connected to the substance properties Melting point and Acidity dissociation constant

The nature framework concept refers to system frame models of the environment. Nature frameworks have system boundaries and properties that are independent of substances such as geographical location, wind speed, time of rain period, etc. These properties are called nature properties in the OMNIITOX concept model. Concepts for nature framework data are nature property specifications and nature property values in equivalence to substance, as shown in Fig. 3.

European European aquatic aquatic and and terrestrial terrestrial ecosystem ecosystem Nature ... framework

Nature property values

Density Density of of soil soil solid solid phase phase 3 [kg/m ] [kg/m3]

2.13 0.5 0.5

Precipitation Precipitation rate rate constant [m/yr] [m/yr] Nature property specification

Fig. 3: A nature property value is defined by referring a nature framework and a nature property specification. In this example the nature framework named European aquatic and terrestrial ecosystem is connected to the nature properties Density of solid phase soil and Precipitation rate

Int J LCA 9 (5) 2004

Unit: Unit: Substance: Substance : Type: Type : Medium: Medium: Geography: Geography : Description: Description :

Geography

Medium

emission emission of of CO CO2 toair air 2 to in in Europe Europe kg kg CO CO2 2 emission emission air air Europe Europe Representing Representing emissions emissions from from technical technical systems systems Data category Data category in in an an LCI LCI context context

Fig. 4: Example of a load definition based on nomenclaturised parameters that matches LCI result data

An indicator is a quantifiable entity of interest within the nature system. In a cause-effect model the indicator is the definition of the effect. In the same way as for a load, an indicator is independent of any mechanism. Many mechanisms may affect the same indicator, and it is not possible, by just looking at the indicator, to distinguish which specific mechanism may affect it. The indicator is nevertheless defined for a specific reason. The impact indication principle, i.e. the reason why and the policy applied when an indicator is chosen must therefore be stated with the definition of the indicator. An example of an indicator definition is shown in Fig. 5. It is important to distinguish the definitions of loads and indicators from values of loads and indicators. Loads and indicators only define the domains of valid load values and indicator values.

Example of an indicator definition

315.4 1800 1800

Name: Name :

Load Load selection selection Principle: principle

Melting Melting point point [K] [K]

315.4 181.2 Substance property values

Type

Substance

Name: Name:

Concentration Concentrationof of formaldehyde formaldehydein in fresh fresh water water compartment compartmentin in Europe Europe Unit: kg/m³ kg/m3 Substance: Formaldehyde Substance: Formaldehyde Medium: Fresh Fresh water Geography: Europe Geography: Europe Description: Representing Description: Representing increase increase of of concentrationover over background backgroundof of formaldehyde formaldehydein in fresh fresh water water compartment compartmentin in Europe Europe Impact Indication indication ERA ERA fate fate modelling modelling Principle: principles principle principles

Fig. 5: Example of an indicator definition

291

Opening

3.3

OMNIITOX

Mechanism

Once the load and indicator are defined in a cause effect chain model the relation between them can be defined. The mechanism concept refers to this relation. It determines the magnitude of the impact on an indicator due to the load. This relation is based on a defined set of substance and nature properties. The system boundaries of the nature framework also define system frame of the mechanism. In this way the mechanism model connects the other independent core concepts together. One or several loads and indicators may be referred to by the same mechanism. A descriptive explanation of the mechanism is required as well as a mathematical definition of any algorithms applied. In a linear characterisation model the mechanism is mathematically expressed as characterisation factors, see Chapter 4. 3.4

Comparison with related concepts

The core concepts represent key elements of a rich language and point out the common denominators of modelling approaches in relevant disciplines. In characterisation models in LCA, toxicological measurements, exposure and effect assessment in RA methodology, it is possible to distinguish the core concepts. Internally, the different modelling approaches use different concepts with slightly different meaning as exemplified in Table 1. By structuring the data of these models according to the OMNIITOX concept model, it is possible to consistently compare numeric outputs of models and it provides possibilities to assess the reusability of data and modelling methodology. 3.5

Substance Emission Model

S=M•F•E

The SETAC Europe's First Working Group on Life Cycle impact assessment (WIA1) proposed a framework for ecotoxicity and human toxicity characterisation within life cycle impact assessment (LCIA) [7], in the following text referred to as the SETAC framework. The SETAC framework builds on experiences primarily from RA and contains many aspects relevant to OMNIITOX. It has been an important starting point to structure toxicity characterisation models. Fig. 6 shows an interpretation of the SETAC framework. In the terminology of the SETAC framework the load, M, represents the amounts of substances emitted from the technical system. A substance emission model in LCA is equivalent to a model of the technical system in an LCI and M is

Fate and Exposure Model

F

Effect Model

E

Category indicator

Characterisation factor = F • E

Fig. 6: General model framework for ecotoxicity and human toxicity characterisation within life cycle impact assessment according to SETAC Europe's First Working Group on Life Cycle impact assessment (WIA1) [7]. Arrows indicate cause effect chain causality

the LCI result. F is the exposure factor that is calculated by a fate and exposure model. The effect factor, E, is calculated by the effect model. The characterisation factor is an aggregation of the whole model into a single value per substance and emission scenario and is calculated by multiplying the exposure factor by the effect factor. The calculation of the impact score on the category indicator, S, is performed by multiplying the characterisation factor by the load. When interpreting the SETAC framework with the terminology of the OMNIITOX concept model, three mechanism models can be distinguished: the fate and exposure model, the effect model and the overall characterisation model. The indicator of the effect model and the characterisation model are the same, i.e. the category indicator. The output from the exposure model must mathematically and contextually match the input to the effect model. The load of the fate and exposure model and the overall characterisation model is the same as the load concept, M, in the SETAC framework. 3.6

Comparison with SETAC framework

M

Support for consistent cause-effect chain modelling

The SETAC framework shows the possibility and the need to model cause-effect chains of the natural system. However, the structuring of a characterisation model does not necessary have to follow the SETAC structure where a fate and exposure model links to an effect model. Other approaches such as fate linking to exposure linking to toxic effect linking to economical cost etc. are equally valid. Therefore, the OMNIITOX concept model should provide the possibility to model an arbitrary number of cause-effect links. The key lies in the connectivity of the indicator of a mechanism with the load of the consequent mechanism. In an industrial perspective we are concerned with how the technical system affects indicators in the natural system. In

Table 1: Core concepts of the OMNIITOX concept model and related concepts used in the contexts LCA, RA, and (eco)toxicology

OMNIITOX Life cycle assessment

Risk assessment

(Eco)Toxicology

292

Load Environmental aspect, LCI result, inputs and outputs, emission, resource, load Input, substance, chemical, stressor, dose Stressor, input, dose

Indicator Category indicator, midpoint, endpoint, damage

Mechanism Environmental mechanism characterisation method, characterisation model

Substance Substance, material, inputs and outputs name

Nature framework System boundaries of characterisation method, receiving environment, landscape data

Risk, response, toxic effect

Dose-response model, fate exposure effect model

Chemical, substance, compound

Compartment, system boundaries

Response, toxic effect

Dose-response model, exposure effect model

Chemical, substance, compound

Measurement conditions, laboratory set-up, compartment, system boundaries

Int J LCA 9 (5) 2004

OMNIITOX

Opening

Aggregated mechanism

Load

Sub mechanism

Indicator

Sub mechanism

Load

Indicator

Fig. 7: A cause-effect chain can be modelled if the indicator of a mechanism is identical to the load of a subsequent mechanism. The aggregated mechanism encompassing sub-mechanisms can be regarded as a 'unit' mechanism. In this way, an arbitrary number of sub-levels of mechanisms may be modelled, documented, and communicated by using the OMNIITOX core concepts. Arrows indicate relational dependency

LCA, the emission and resource flows in an LCI result are an interface between the technical system and the natural system. For consistent use of LCI data in natural system models, a definition of a load that refers to technical systems flows should precisely match the information of this flow interface. It is not necessary that a load originates from a technical system. The load may originate from any adjacent system of a mechanism, e.g. the load may refer to changes in environmental concentration that may be caused by natural mechanisms. This generalisation allows for the modelling of cause effect chains within the natural system with well-defined model inputs and outputs as shown in Fig. 7. In Fig. 7, the arrows indicate relational dependency. This gives that first the loads and indicators must be defined. Mechanisms are then defined as the relations between the loads and indicators. In this sense a load is an indicator that can trigger mechanisms. In other words: a mechanism is the relation between indicators in the natural system. For example, an 'exposure and effect' mechanism model can describe the relation between the indicators 'concentration of nitrogen in water' and 'biomass of algae in water'1. Other mechanisms may describe the relation between 'biomass of algae in water' and 'biomass of fish in water', etc. The interconnection of mechanisms, as shown in Fig. 7, implies a widening of the overall system boundaries, i.e. the connected sub-mechanisms are all members of a superior aggregated mechanism. If the inner sub-mechanism structure of the aggregated mechanism is disregarded, it is itself a 'unit' mechanism. The superior mechanism can in turn be incorporated as a sub-mechanism of yet another mechanism. In this way, an arbitrary number of sub-levels of mechanisms with different levels of detail and resolution can be transparently interconnected in the same model by using the same core concepts. 1

To point out the generality of the cause-effect chain of load-mechanism-indicator the opposite relation is also feasible, i.e. the effect that 'biomass of algae in water' has on 'concentration of nitrogen in water' could be modelled.

To exemplify the indicator-load connectivity the OMNIITOX base model [2] is outlined in Fig. 8. The human impact mechanism is the relation between emissions from technical systems and two human toxic effect indicators. It encompasses a fate, an exposure, and an effect mechanism. The ecotoxicity impact mechanism is the relation between emissions from technical systems and one eco toxic effect indicator. Here the same fate mechanism as in the human impact mechanism is re-used and linked to an ecotoxicty effect mechanism. 4

Using the OMNIITOX Concept Model in LCA Characterisation Modelling

The concept model is compatible and interpretable with related established LCA concepts as defined in the ISO 14040 standards series [14–17]. In LCA a conceptual distinction between the concepts category indicator, and category endpoint [17] is made. In the OMNIITOX concept model these concepts are examples of indicators. Any indicator can be incorporated in an LCIA method by defining them as category indicators that can be weighted in accordance with LCA methodology. The characterisation step can be mathematically expressed as a general characterisation function: CR=f(L, CP), where CR is the characterisation result i.e. the indicator value, CP are characterisation parameters, and L the LCI result, i.e. the load value. Here, the function f is the mechanism algorithm in its most aggregated form. The mechanism is the modelled physical representation that transparently explains the shape of the function f including all parameter derivations, assumptions, simplifications, and data aggregations modelled. The calculation of the characterisation parameters generally follows CP=f(S,N) where S are the required substance property values and N are the required nature property values. Characterisation factors (CFs) are a specific case of characterisation parameters. If CFs are applied, the mechanism describes a direct proportional relation between load and indicator: CR=L · CF.

Human impact mechanism Human exposure mechanism

Emissions from technical systems

Fate mechanism

Intake fractions

Human effect mechanism

Human toxic effects

Masses in compartments (time integrated)

Ecotoxicity impact mechanism

Ecotoxicity effect mechanism

Eco-toxic effect

Fig. 8: The OMNIITOX base model outlined in terms of loads, mechanisms, and indicators. Arrows indicate relational dependency

Int J LCA 9 (5) 2004

293

Opening When numerically comparing models, it is necessary to be clear about differences in loads and indicators between models. If they are in any way different, numerical differences in calculated characterisation parameters can be expected. 4.1

Usage and implementation of operational models

In order to create operational models and industrial tools an important requirement is to make it possible to easily extract the information relevant for usage of these often very complex models, i.e. it should be easy to understand how to apply a model and to assess the applicability of a model for a specific study. It is a simple task to assess how a characterisation model documented according to the OMNIITOX concept model should be applied in an LCA context. The load is a precise definition of what LCI data categories are applicable to the model. Any indicator referred by the mechanism can be defined as a category indicator. The nature framework gives the user an explanation of system boundaries that can be compared with the scope of the specific LCA study. If the load, indicator, and mechanism algorithm of a model is unambiguously defined, it can be translated into well formatted data and software code by programmers and implemented on a computer. Upon implementation of the OMNIITOX characterisation models this structure has been used as a tool to assess model completeness and consistency. 5

OMNIITOX information in the OMNIITOX database. The database is at the heart of the OMNIITOX information system, and the fields and tables of the OMNIITOX database are directly derived from the OMNIITOX concept model, which makes it possible to have the same understanding of the data regardless of user perspective. It is a crucial support for data quality management within the OMNIITOX IS. Though the scope of OMNIITOX only covers eco and human toxicity, the OMNIITOX concept model is designed to be applicable to any characterisation model. In addition, as the OMNIITOX concept model is created in a consensus process with participants from the LCA, RA, and toxicology disciplines, it is possible to use the OMNIITOX concept model as a tool in RA and toxicity modelling. Acknowledgement. The authors gratefully acknowledge the financial support of the European commission through the Sustainable and Competitive Growth-research program given to the OMNIITOX-project (Operational Models aNd Information tools for Industrial applications of eco/TOXicological impact assessments, EC Project number: G1RD-CT-2001-00501).

References [1]

[2]

Conclusions

The OMNIITOX concept model has been developed in a consensus based process by participants from the LCA, RA, and informatics disciplines. The SETAC framework for ecotoxicity and human toxicity characterisation within life cycle impact assessment has been a starting point for the design of the OMNIITOX concept model. A characterisation model following the design of the SETAC framework can be easily interpreted into the OMIINTOX concept model. However, the OMNIITOX concept model has a higher level of flexibility and can be regarded as a generalisation of the SETAC framework to allow for scalability and arbitrary structure of cause-effect chain models. The OMNIITOX concept model is compatible and interpretable with related established LCA concepts as defined in the ISO 14040 standards' series. Load definitions referring to technical systems specify the applicable data categories of the characterisation model. A category indicator is an indicator applied in an LCIA method, characterisation is the relation between an LCI result and a category indicator, which is identical to a mechanism that describes the relation between a load and an indicator and has system boundaries defined as a nature framework. From a user perspective the OMNIITOX concept model facilitates the understanding of applicability, usage and interpretation of model output. Thus, the OMNIITOX concept model provides a common format for documenting, storing, reporting, and exchanging information in the multidisciplinary environment within the scope of OMNIITOX.

[3] [4] [5] [6] [7]

[8] [9]

[10] [11] [12] [13]

[14] [15] [16] [17]

6

Outlook

In the further work with the implementation of the OMNIITOX IS the OMNIITOX concept model will describe all pieces of

294

Molander S, Lidholm P, Schowanek D, Mar Recasens M del, Fullana i Palmer P, Christensen F, Guinée JB, Hauschild M, Jolliet O, Carlson R, Pennington DW, Bachmann TM (2004): OMNIITOX – Operational Life-Cycle Impact Assessment Models and Information Tools for Practitioners. Int J LCA 9 (5) 282–288 Guinée JB, de Koning A, Pennington DW, Rosenbaum R, Hauschild M, Olsen SI, Molander S, Bachmann TM, Pant R (2004): Bringing Science and Pragmatism together in a Tiered Approach for Modelling Toxicological Impacts in LCA. Int J LCA 9 (5) 320–326 Erixon M, Carlson R, Flemström K, Pålsson A-C (2004): The Data Quality Foundation in OMNIITOX Information System. Int J LCA 9 (5) 333 Elmarsi R, Navathe SB (1989): Fundamentals of Database Systems. Addison Wesley World Student Series, ISBN 0-201-530090-2 Carlson R, Pålsson A-C (2000): PHASES Information models for industrial environmental control, CPM-report 2000:4, Chalmers University of Technology Carlson R, Pålsson A-C (2001): Industrial environmental information management for technical systems, Journal of Cleaner Production 9 (5) 429–435 Jolliet O (ed) et al. (1996): Impact assessment of human and ecotoxicity in life cycle assessment, pp 49-61. In: Udo de Haes H A (ed): Towards a methodology for life cycle impact assessment. Society of Environmental and Toxicology and Chemistry Europe, Brussels Cronin B (ed) (1985): Information management: From strategies to action. London, Aslib, ISBN 991-557677-3 Carlson R (1994): Design and Implementation of a Database for use in the Life Cycle Inventory Stage of Environmental Life Cycle Assessment. Diploma work in Computing Science, Department of Computing Science, Chalmers University of Technology and Gothenburg University Carlson R, Steen B, Löfgren G (1995): SPINE, A Relation Database Structure for Life Cycle Assessment. Gothenburg, IVL-REPORT B1227 Carlson R, Pålsson A-C (1998): Establishment of CPM's LCA Database. Gothenburg, CPM Report 1998:3 Dewulf W, Duflou J, Ander Å (eds) (2001): Integrating Eco-efficiency in Rail Vehicle Design. Final report of the RAVEL project, Leuven University Press, ISBN 90-5867-176-3 Carlson R, Erixon M, Geiron K, Tivander J (to be published): OMNIITOX concept model and data format report. Deliverable 20 and 26 of the OMNIITOX project; Industrial Environmental Informatics; Chalmers University of Technology ISO 14040 (1997): Environmental management – Life Cycle Assessment – Principles and Framework ISO14041 (1998): Environmental management – Life Cycle Assessment – Goal and scope definition and inventory analysis ISO 14042 (2000): Environmental management – Life Cycle Assessment – Life cycle impact assessment ISO/TS 14048 (2002): Environmental management – Life Cycle Assessment – Data documentation format Received: October 17th, 2003 Accepted: August 18th, 2004 OnlineFirst: August 20th, 2003

Int J LCA 9 (5) 2004

Advances in Environmental Research 5 Ž2001. 369᎐375

System for integrated business environmental information management R. CarlsonU , M. Erixon, P. Forsberg, A.-C. Palsson ˚ Centre for En¨ ironmental Assessment of Product and Material Systems, Chalmers Uni¨ ersity of Technology, S-412 96 Goteborg, Sweden ¨ Accepted 2 January 2001

Abstract By making use of current business information technology, such as Internet-accessible tools, and industrial environmental management tools, standards, policies and legislation an information system for environmental information management has been designed. The system is named Integrated Business Environmental Information Management ŽIBEIM., and it includes operational, procedural and organisational support for a business’ entire environmental information management. IBEIM consists of a system architecture and an information and data content. The system architecture is designed from three thoroughly developed information models, i.e. a reference model for information aggregation and communication, an information model for data structuring, and a modularization including module interface specifications. The content is stored in a common information platform, and includes, for example, a structured knowledge system. IBEIM efficiently supports and integrates environmental information management for Environmental Management Systems ŽEMS. tools, LCA and other environmental process modelling tools, and Design for Environment ŽDfE. tools. With this system, integration communication and reports are handled in a consistent and compatible way all through an organisation of any size. IBEIM is also designed for supply chain communication. 䊚 2001 Elsevier Science Ltd. All rights reserved. Keywords: Information system; Modularised system architecture; Industrial environmental management; PHASETS; SPINE; ISO 14048; STEP; Environmental database; Environmental supply-chain management; Environmental reporting; Information quality maintenance

1. Introduction and purpose of the IBEIM system Since the 1960s, the industries have faced ever increasing environmental demands externally, from customers, authorities and NGOs, and, during recent

U

Corresponding author. Tel.: q46-31-772-2173; fax: q4631-772-5649. E-mail address: [email protected] ŽR. Carlson..

decades, also from internal business control functions, i.e. environmental management systems. To meet these demands, the needs for environmental information management have grown simultaneously. Every new demand has been met by implementing a new information management activity, which has led to the current situation, where most industries have many parallel and unsynchronised information management activities. For instance, it is commonplace to have separate systems for governmental reporting, internal yearly re-

1093-0191r01r$ - see front matter 䊚 2001 Elsevier Science Ltd. All rights reserved. PII: S 1 0 9 3 - 0 1 9 1 Ž 0 1 . 0 0 0 8 8 - 0

370

R. Carlson et al. r Ad¨ ances in En¨ ironmental Research 5 (2001) 369᎐375

porting, internal environmental management, life cycle assessment, etc. However, much of the information used in these separate systems is the same, and it is an economical sub-optimisation to independently acquire, document, analyse, store, communicate, and quality control the same information in all different systems. The system for Integrated Business Environmental Information Management ŽIBEIM. is designed to encompass and integrate all environmental information management activities within business organisations, and in their external communication. The design is based on an overall view including all current needs and has an open design towards future demands, as long as these future demands are formulated in a way similar to current demands. This means that environmental management concerns industrial and other human activities and is expressed in terms of environmental indicators and physical flows that are assessed with for example impact and risk assessment methodologies. IBEIM structures information acquisition and storage. It supplies a structured approach to information communication and aggregation, all the way from measured entity and up to highly aggregated reports. In addition, it supplies standard communication formats as well as translations between different formats. All these functions and structures are needed for a truly integrated information system. IBEIM makes use of current business information technology, i.e. Internet technology and client server modularization. Currently, the ideas of IBEIM are being applied in the development of different corporate environmental information systems within, e.g. the Swedish forestry, pulp and paper industry, the European RAVEL ŽRail Vehicle eco-efficient design. project, and ABB. The Swedish forestry, pulp and paper industry is introducing the system as work routines that may allow for implementation into computer systems, the RAVEL project aims for a methodology and a computer system supporting the total supply-chain during train design projects.

for example according to ISO 14020 series Ž1998. ŽISO 14020:1998., presented to customers. IBEIM is prepared for full supply-chain information management, meaning it is open for communication with each customer᎐supplier interface. Many different reports can be generated as information for customers, and the customer may be free to either import the report into their own IBEIM, or to acquire them as separate pieces of information. In addition, IBEIM generates supplier-questionnaires on demand and may even allow suppliers to enter information directly into the IBEIM information platform. In a fully integrated IBEIM, the customer and supplier may share access to parts of each other’s systems Žsee also Fig. 2.. Any industry already has much of the information needed in IBEIM structured in other information management systems, such as economic, logistic, product data management, etc. systems. IBEIM does not require parallel storage and structuring of this information, but instead connects to and reformats the information in other systems, on demand. IBEIM does not require additional tasks or activities, additional personnel or much additional education. Instead it simplifies and increases efficiency in already established routines and tasks.

3. Components of system IBEIM is a structured information system, i.e. a system architecture with an information and data content. The IBEIM architecture is built from three elements Žsee also Fig. 1.: 䢇

2. Scope of system 䢇

The scope for IBEIM is the same as the scope for the industrial environmental responsibilities and activities described in the ISO 14000 standards. One boundary for the scope is drawn at defining environmentally relevant physical entities and indicators, as described in ISO 14031:1999. The reporting to an enduser of the system draws another boundary, for example, internally for aggregated environmental performance reports or statements for the director, or externally for simplified environmental product declarations,

A basic information platform, the data structure, implemented as a relational database for storage, and a standardised format ŽSTEP. for data communication within the system and between separated sub-systems ŽISOrIEC 9075:1992; ANSI X3.1351992; ISO 10303-11:1995, 10303-21:1995.. Using STEP as the communication format will ensure external compatibility with other software and internal transparency between system modules. PHASETS ŽPHASEs in the design of a model of a Technical System., which is a reference model including six phases for communication and aggregation of industrial environmental data and information ŽCarlson and Palsson, 2000.. Each phase in the ˚ model describes functions and management tasks, without specifying how these tasks should be performed. The reference model serves as a template for the architecture of the IBEIM system, and as a task and a reporting sequence for the different types of information handled within the system.


䢇

Fig. 1. Schematic illustration of the IBEIM system modularisation.

䢇

A modularization based on the information platform, i.e. the data structure of the database, the distinguished phases of the reference model PHASETS, and on organisational tasks that the users perform, i.e. the organisational structure. The modularization is documented as an application programmers interface ŽAPI. specifying the interfaces and the communication and collaboration between software modules of the system. The formulation of an API makes it possible to addrchange modules according to present needs in the form of for example plug-ins. The API documentation will be maintained in a public forum and will be publicly available for different software developers to apply and conform to. The current IBEIM systems being built do not make use of the IBEIM API. The IBEIM information and data content:

䢇

䢇

Descriptions of all activities related to each of the six phases in PHASETS, supporting communication and aggregation of industrial environmental information ŽCarlson and Palsson, 2000.. Except for the ˚ description and documentation of the main phase activity, each phase also involves related information, such as procedures for the control of documents, education, auditing, and following-up, etc. Roles, responsibility, and authorities can be defined, and be sufficiently described as well. A knowledge system, referencing external or inter-

371

nal information sources, such as environmental and quality standards wISO 14000 family of standards, EMAS ŽThe Council of European Communities, 1993. and ISO 9000 family, etc.x, national and international environmental legislation and regulation, environmental customer demands to support the company purchase department, environmental benchmarking for the line of business, etc. The knowledge system is an integral part of IBEIM, in that its structure is described in the IBEIM database and data communication structures, and in that data in the database reference documents that can supply the user with definitions or general information through context sensitive pointers within the system. Environmental records and communication forms, such as policy and goal statements, results of environmental assessments Žlife cycle assessments, risk assessments, SWOT-analyses, etc.. audits and reviews. These documents can consist of internal or external reports, working material, official environmental product declarations, information brochures for customers, environmental accounts, diagrams and tables for presentations, etc.

4. Technology applied IBEIM is based on standard technology. Special efforts have been laid on the information platform in order for it to be stable, regardless of implementation of specific modules. Information structure should be left unchanged regardless of choice of methodology, procedures or software. Therefore, the information platform is documented as a conceptual entity relationship model, from which both the relational database design and the STEP exchange file specification has been derived. By choosing relational databases for data storage, there is a wide range of standard operating systems and database management systems that may be used as database servers for the system. STEP is a standard for implementing a data exchange format, as well as the standard describing an implemented format. Currently, IBEIM has only followed the STEP methodology for implementing and describing the exchange file format. It has not been suggested as a standardised format for integrated industrial environmental information management. The PHASETS reference model is not standard technology, but is based on the principle of the ISOrOSI reference model Ž responsible body ISOrTC97. which describes data aggregation and communication from a lowest physical layer up to the application layer. PHASETS describes aggregation and

372


communication from definition of a measured entity up to compilation and submission of different environmental reports. The modularization of the system is made in a standard way through the documentation of the API. This makes any module replaceable and it makes it possible to build and maintain the system into its future needs. By having the API specified, it is also possible to build a copy of any existing system on another operating system platform or to build it as a platform independent system using for example JavaTM technology.1 The system modules are independent of each other and can thus be both developed and used separately. The system is intended for Internet technology, meaning that users connect to the IBEIM system by their Internet browsers. All user interaction is built to be platform independent and runs at the client, while databases and demanding calculations runs at the servers, establishing a well-optimised client-server environment.

5. Organisation for system The IBEIM is designed for integration with any business or industrial organisation. Before starting to use the system, it may be configured with descriptions of the organisational units that will make use of the system, the users within these units, and the different relationships between the organisational units and users. Since the system is based on Internet technology, it is distributed to each different user as web pages. Access rights, user restrictions, and appropriate functionality is prepared for the user, based on login identities. The system is designed for three levels of access rights for each user login: core rights are given for each form or functionality that the user needs for his direct work in the system. A user generally has insertion and updating rights on all core information. Contextual rights are given to a wider area of information and functions, to enable a user to view and analyse a larger information base than is directly needed for the work. The user does not have updating rights on the information with contextual rights. All information outside of core and contextual rights is hidden to the user. IBEIM supports a number of common and crucial organisational roles. The basis for the system is the information quality management. The responsibility for the total information quality is distributed through the entire organisation, and is strongly supported by the

1

Java is a trademark of Sun Microsystems Inc.

system. ŽCarlson and Palsson, 1998a,b.. The quality ˚ control of each role is considered, from the environmental management and co-ordinators identifying, defining and selecting environmental indicators, the measurement engineers being responsible for defining measured entities and the setting up and maintenance of measurement equipment, through the operational registering of quantitative information, mathematical and systems analysis and compilation of reports. The quality is maintained by standard operations, documentation, and transparent communication. End users of information can always trace operations and manipulations made on data and information. The management can use IBEIM to delegate new routines for data acquisition, analysis and compilation, and can through the system supervise a correct data handling. The business’ marketing departments, the management and the board, and the product and service development departments can access the system for their decisions or to design special reports displaying specific environmental issues. Such reports may then be edited in for example standard graphical design software, for powerful and credible presentations.

6. System architecture and functionality IBEIM is a modularised, distributed system designed to support the six phases of environmental data management described by the PHASETS model ŽCarlson and Palsson, 2000.. Each phase is designed as a soft˚ ware layer in the system, including software that performs well-distinguished tasks Žsee Fig. 1.. The bottom layer Ž0. consists of tools for entering documentation describing quantitative entities, such as environmental indicators, and equipment and methods for measuring these entities. The first active layer Ž1. consists of tools for registering quantitative data and sampling. This can either be user interfaces for manual data insertion, or automatic routines sampling directly from, for example, a measurement system. The second active layer Ž2. consists of tools aiding the user or the system to perform statistical analysis on sets of individual quantitative data. The third and fourth active layers Ž3. and Ž4., consist of systems analytical tools, such as process modelling tools, LCA tools, environmental management systems ŽEMS. tools, design for environment ŽDFE. tools etc. With these tools, the user may generate descriptions of the environmental performance of technical systems, subsystems and products. The fifth active layer Ž5. consists of configurable report generation tools, communicating the contents of the databaseŽs. to different IBEIM information users. Each layer communicates using the same data model


that defines the database structure. This may be done either by communicating through data exchange at the database, or through direct communication between the layers, using the data model implemented as a STEP ŽISO 10303. parts 21 file ŽISO 10303-21:1995.. Communication within IBEIM and to its environment is regulated by rules described in the fifth layer, basically prescribed as routines for quality assurance regarding preservation of meaning between information sender and receiver. A formal description of the underlying format makes it possible to use existing techniques for verification of consistency of the format. Tools and modules for the generation and interpretation of documents that comply with the format are already available as elements similar to these commonly used in software compilers for computer languages ŽC, Pascal, Ada etc... Such technology has also become commonly used for constructing recognisers and translators for different kind of text processing, such as HTML, SGML and STEP. An IBEIM system is a network capable unit, which may communicate with other IBEIM compliant systems. For instance, IBEIM may be fully or partially implemented in different organisational units within the same company, and will then work fully and separately for each organisational unit. Any two IBEIM implementations may then communicate, i.e. exchange information and data with each other, to generate for example unifying reports for the management. IBEIM is also designed for supply chain management, which means that companies with business relations may exchange relevant environmental business and product data. In fact, this communication needs only the agreement of communicating the IBEIM prescribed STEP parts 21 files between the companies ŽFig. 2..

7. Applications for system IBEIM has been developed to support the industry and other business organisations with an intelligent environmental information management system that will assist these organisations to achieve environmental and economic goals. IBEIM co-ordinates and integrates the various information needs supplying the vast number of environmental management tools used today, such as: 䢇

environmental management system, for example, information regarding the operation of the system, the data aggregation phases, legislative, regulatory, and other applicable policy requirements, such as ISO 14001:1996, EMAS ŽThe Council of European Communities, 1993., environmental assessments,

373

Fig. 2. IBEIM as a network capable unit, enabling environmental supply chain management.

䢇

such as risk assessment, life cycle assessment and environmental evaluation of suppliers; and environmental records and communication forms, for example, internal and external environmental reporting and environmental labelling.

The system is designed to increase the efficiency and the quality of the information management for any of the above listed tools. The best efficiency, however, is reached by integrating more than one tool with the system. In this way, one may make use of the same information for different purposes, and use the same software modules for different tasks and routines. A core design feature of the system is the format transformation functions, which translates information and data from the IBEIM internal data format into or from data formats for commonly used environmental information management software. This feature implicates that the IBEIM recommends system users to make use of the most suitable software tools for each environmental information management task, without enforcing any specific tool.

8. Further work Currently, there are many different environmental information management systems being designed and developed within different industries and other businesses. These systems are integrated with, for example, product data management and computer aided design systems. Within each of these separately designed systems, the environmental information is differently

374


modelled and formatted. Since IBEIM is designed for format translation between different information sources, each new format to translate to and from will imply the need for a new formal translation-routine in the system. Frequent launching of new formats for every new system will, therefore, lead to very high costs for a general IBEIM which can handle all format translations needed. Unless the continuous launching of new formats for environmental information is halted, costs for environmental management will increase uncontrollably. This may eventually lead to a situation where environmental management cannot be economically sustainable. The only way to avoid this situation is through harmonisation and standardisation of reference models such as PHASETS ŽCarlson and Palsson, 2000., and data for˚ mats such as ISO 14048 ŽISOrCD 14048.. To reach full functionality and efficiency, IBEIM should be integrated with other core business information systems, such as automated manufacturing systems, economical systems, logistic systems, and support systems for purchasing and marketing. Integration with business information systems may be approached from two different directions. One is by expanding IBEIM’s format translation functionality to include all relevant information exchange formats needed to communicate with other business systems. As soon as a format is expressed in any standard formal way, a new interpretation unit for the language can be constructed. A requirement is that the format can be given a consistent construction, otherwise it cannot be expressed in a logical manner. The other approach is to integrate IBEIM models, standards, functionality, and modularity with business systems products available on the market. Basically this would mean that business system vendors develop their own competitive IBEIM solutions. This may be profitable for the vendors when industry standards for environmental information and data has been introduced and successfully used within different industries. Such increased industrial assimilation may initiate a demand for IBEIM compliant and integrated products.

9. Conclusions

PHASETS reference model and to the format for the environmental information and data. The successful design of IBEIM is due largely to the PHASETS reference model, in that it supplied a thorough model for information management at different organisational levels and with different information aggregation. PHASETS serves as a design guideline through the development of an information system reaching from the level of physical flows at manufacturing plants, up to strategic organisational levels, such as the executive management board and the marketing departments. It also serves as a high-level software modularization schema, including high-level communication interfaces and task descriptions. Another valuable component for the design of IBEIM is the underlying model describing the format for the environmental information. This model originates from the work made in the Swedish industry, with the SPINE model and format ŽCarlson et al., 1995; Carlson and Palsson, 1998a,b., and on the work made ˚ within ISO, with standardising a format for LCA data documentation, the upcoming ISO 14048 ŽISOrCD 14048.. This work has revealed the important aspects of information describing the environmental performance of technical systems, i.e. production processes, service activities, transports, etc. It has also resulted in a substantially improved understanding of the different and special data quality aspects of such information, and has identified methods for administrating and maintaining the quality of such information systems ŽCarlson and Palsson, 1998b.. These accomplishments ˚ have guided both the development of the PHASETS model, as well as the general design of the IBEIM system. The importance of well described data formats and well-designed transformations between systems cannot be underestimated, and it sets the foundation for the quality of the entire information system. IBEIM aids users with very fast communication, by making use of computer network technology, i.e. client server components, database servers and centrally published web user interfaces. By making parts of the IBEIM accessible for both suppliers and customers, environmentally relevant business communication may be as fast and feasible as internal communication. References

In spite of its apparent ambitious scope, IBEIM is relatively inexpensive to technically implement. This is due to the focus directed towards formats, standards and standard technologies. By the application of standards, no additional work is needed for systems development of the core elements, i.e. information requirement analysis and design. IBEIM is also inexpensive to technically implement because of the thorough modelling of the system, especially with regard to the

ANSI X3.135-1992. Database Language SQL. Carlson, R., Lofgren, G., Steen, B., 1995. SPINE, A Relation ¨ Database Structure for Life Cycle Assessment. Report B1227. Swedish Environmental Research Institute, Goteborg. ¨ Carlson, R., Palsson, A-C., 1998a. Establishment of CPM’s ˚ LCA Database. CPM-report 1998:3. Centre for Environmental Assessment of Product and Material Systems, Chalmers University of Technology, Goteborg. ¨

R. Carlson et al. r Ad¨ ances in En¨ ironmental Research 5 (2001) 369᎐375 Carlson, R., Palsson, A-C, 1998b. Maintaining Data Quality ˚ within Industrial Environmental Information Systems. Computer Science for Environmental Protection ’98, 12th International Symposium, Bremen 1998, Band IrVolume I. Metropolis-Verlag, Marburg, pp. 252᎐265. Carlson, R., Palsson, A.-C., 2000. Industrial Environmental ˚ Management for Technical Systems. Accepted for publication in J. Cleaner Prod. ISO 10303-11:1995. Industrial automation systems and integration. Product data representation and exchange. Part 11: description methods: the express language reference manual. ISO 10303-21:1995. Industrial automation systems and integration. Product data representation and exchange. Part 21: implementation methods: clear text encoding of the exchange structure.

375

ISO 14001:1996. Environmental Management Systems ᎏ Specifications with Guidance for Use. ISO 14020:1998. Environmental labels and declarations ᎏ General principles. ISO 14031:1999. Environmental management ᎏ Environmental performance evaluation ᎏ Guidelines. ISOrIEC 9075:1992. Information Technology Database Languages SQL. ISOrCD 14048. Environmental management ᎏ Life cycle assessment ᎏ Life cycle assessment data documentation format. The Council of European Communities, 1993. Eco-Management and Audit Scheme ŽEMAS., The Council Regulation ŽEEG. No 1836r93.

Carlson R.: "Learning from management of LCA data" Journal of Life Cycle Assessment, Japan, Vol. 1 No. 2 Jul. 2005, pp 102-111

Synopsis 5

10

15

The Swedish national LCA database was built to support industrial LCA needs. All experiences during the build-up and all data were documented so that the acquired knowledge should be useful in the future. This article describes many learnings and positive side-effects from ten years of establishing that database. The national co-operative form is presented, as well as the layout of the project. The data format and data quality approaches are described in terms of relevant choices and learnings. Learnings acquired in the project were put into use early, and integration of LCI data acquisition with industrial environmental information systems were tested in industries as early as 1997, and full-scale tests were proven successful in 2001. Integration with DfE methodologies has also been tested at industrial scales, as well as semiautomated environmental reporting. In 2005 the database is still running and is actively maintained and is further developed. Companies start to want to update the database, and all key learnings from the ten years are available and can be utilised as competence for new projects, strategies and visions for more and better LCA data. Keywords: environmental information management, environmental management system, design for environment, LCA database, data format

Introduction Starting the Swedish national LCA database 20

Plans for the Swedish national LCA database were sketched in the early 1990s [1], from lessons learned from data acquisition in different LCA studies in industry [2] and for the Swedish government, such as the Swedish packaging LCA performed in 1991 [3].These were similar to those that later led to e.g. the SPOLD data format [4] and formulations from LCANET in 1997 [5].

25

Included with the Swedish plans were lists of criteria, such as that: • The data should be documented in English. • The database should suit different types of LCA. • The data should contain meta-data. • It should be possible to connect the database to different (software) applications. • Data quality should be assessed before data were entered into the database. • The database should be continuously updated.

30

In 1994 there were few experiences from building general, sector independent, transparent and quality assessed LCA databases. Learnings had to be acquired from own experiences. 35

40

45

The co-operative national effort The starting point was the forming of the Swedish national competence centre CPM [6]. CPM started in early 1996, as a joint research forum between Chalmers University of Technology [7], 10-15 Swedish industries and government (through the agency NUTEK [8], later VINNOVA [9]). A main purpose of CPM was to establish and maintain a Swedish national LCA database, and to conduct research in LCA and product related environmental assessments. CPM started as a ten year governmental investment, financed with equal parts from government, industry and Chalmers. The CPM project for establishing the database started in April 1996, and according to the plan the objective was to increase availability, usefulness and quality of LCA data by installing and managing a physical database for LCA data, by developing and managing the database’s conceptual data model (see section ‘The LCI data format’ below), and to work towards a standardised LCA data communication format. The project should also acquire and enter data into the physical database. The database should also have a data quality management system. During the project runtime two small reports were produced, to answers 12 common questions or misunderstandings during the project work [10], and to clarify the strategy of the project [11]. The final

1(10)


50

55

result of the project was reported two years later in the project report [12]. It was written as guidance for others who need to establish a similar database. Learnings from the project are that the strength from having ten years of combined governmental back-up, industrial commitment and funding for academic research provided the necessary courage for CPM to formulate serious questions and set relevant goals for the work; the LCA database should comply with industry standards so that LCA would comply as an industrial methodology and tool. In the following a number of detailed spin-offs and learnings will be presented.

Data The LCI data format 60

65

The data format for the database was developed in a joint Nordic project, with companies and research institutes from Norway, Sweden and Finland into a commonly agreed upon data format. The format was described in a public report [13], and the format was given the name Sustainable Product Information Network for the Environment (SPINE) [14]. During the development it was anticipated that SPINE should apply to any type of LCA, and also for environmental accounting, environmental management systems (EMS), design for environment (DfE), and for simultaneous environmental reporting. It was also designed flexibly to be compatible with other LCA data formats.

The impact assessment data format

70

75

In 1998 the ISO standard for life cycle impact assessment, ISO 14042 [15], was in its finalisation phase, and the LCI part of the SPINE data format was well tested in industry and by software developers. A compatible data format for impact assessment data, according to ISO14042 was therefore developed [16]. It was tested, and was made public in the year 2000 as a transparent and reviewable impact assessment database combined with the Swedish LCI database, into web-based LCA tool [17]. The database contains the most well known and commonly applied LCA impact assessment methods at that time, complete and well-documented; Eco Indicators 99 [18], EDIP [19][20], and EPS 2000 [21][22]. The learnings from documenting these three impact assessment methods in the format were published as documentation guidelines for impact assessment data [23].

The data quality criteria

80

85

90

95

When establishing the national LCA database a separate sub-project was dedicated to define the quality criteria for the database. The sub-project was industrially driven and based on relevance and consensus. The result was the CPM LCI data quality criteria [24] which were in fact documentation criteria (English version of criteria in terms of ISO 14041 was published in 1999 [25]). These criteria were expressed as a list of mandatory data fields of the SPINE format; each mandatory data field should be filled in to allow for publication in the CPM database. As a result, since 1997 each data is manually reviewed for clarification and consistency, and to facilitate compliance with these criteria, manuals for data documentation [26] and for data review [27] were produced, and courses were held for industry, institutes and academia. In parallel to developing the documentation criteria, CPM also performed theoretical data quality studies, with the aim to find a quantitative approach to complement the documentation criteria. These studies resulted in a licentiate thesis in 2001 [28], but the findings have not been possible to operatively integrate with the database. The CPM data documentation quality criteria are valuable since they are easy to understand and flexible to any industry or any application. They also provide the natural interpretation of the data format, and database users highly appreciate to get documented data. The weaknesses are that users find it tedious and time consuming to document the data, and that that the quantitative data are not reviewed. To review that quantitative LCA figures are correct would require competence from many or all industry sectors. This has instead been approached by developing methodologies for environmental information quality management in production plants.

2(10)


Information and database management Data quality is the result of quality processes 100

105

110

115

The data quality criteria developed by CPM requires documentation of data, and in the project an obvious fact was learned; If data is not documented when presumptions, assumptions, errors, limitations and other important facts about data are known, then these facts will never be known again and will never be documented later on. The implication is that data need to be documented already when it is produced, e.g. as a sampling value at the measurement site or when calculated by an emission-model. Such LCI data documentation was not generated anywhere in 1996 and the methodology had to be developed. A general methodology for how documentation could be merged with the handling of environmental information within industries were described already in 1998 [29], and refinements were made due to practical tests in industry [30]. The aim was to reduce costs by integrating environmental information management systems with other business functions, while improving precision and documentation quality of the data. The theoretical framework named PHASES (PHASEs in the design of a System) supports this [31][32]. PHASES conceptualises the different steps needed to produce environmental data, from defining and setting up a measurement system to the final reporting of an LCI data set or a complete LCA. Among other applications, PHASES was adapted in the European thematic network CASCADE (Co-operation and Standard for Life Cycle Assessment Data in Europe) [33] as a data collection process that results in high quality LCA data [34][35]. One extra benefit from having a quality management view on LCA data handling is that it provides better understanding about the quality of industrial environmental data management in general; there is a dependency between LCA data quality and the quality of EMS.

Environmental data management in industry

120

125

130

From learning that LCA data quality results from information management within the EMS, a number of master thesis studies were performed to investigate this relationship [36][37]. At the SETAC meeting in Leipzig in 1999 a general principle of industrial environmental information quality management system was presented [38]. Between 2001 and 2004 a number of CPM companies ran a project aimed at practically improving environmental data handling within industrial EMS [39][40][41]. It was based on a direct coupling between a company’s environmental policy and its environmental information system, and quality implications from this. One finding was that there are data gaps in current EMS. This conclusion was further underlined by a data requirements gap analysis of the ISO 140ff standards that was performed in 2004 [42]. The gap analysis shows that the standards include expectations on data quality but supply insufficient support for how to meet these expectations. From working with LCA data quality management as part of the industrial EMS it was learned that there are gaps between company’s policy levels and the operational levels. Hence, the building up of the Swedish national LCA database developed far from scientific data and statistics, into the core of industrial EMS.

Applications of methodology in different industrial sectors Tests and feasibility studies

135

140

The SPINE data format was designed to be independent of industrial sectors, and this claim is tested and proved within different industries that run SPINE databases as corporate LCA network databases or as standalone software applications. Except for practical and commercially based SPINE installations, feasibility assessments of the format have been made in the building sector [43], the waste management sector [44], the electronics sector [45] and the aluminium sector [46]. All tests have given positive results, but it may be that the broad practical success is one reason why it was not fully appreciated in different LCA communities outside of Sweden; the practical quality tests used up the resources that could have been available for scientific dissemination and practical marketing.

3(10)


Information management system in paper and pulp industry

145

150

Approaches for industrial environmental information management were tested thoroughly in the Swedish paper and pulp industry between 1999 and 2002. The PHASES method [32] was refined and tested for environmental information management and LCA data acquisition at industrial production plants, with the intention to decrease costs, increase quality and coordinate environmental information management. The project developed a practical methodology that fulfilled these intentions. The results were published as a methodology report [47], a manual [48], a technical licentiate degree [49] and articles presented to the international pulp and paper sector [50]. Learnings were that it is practically possible and economically beneficial to produce well documented LCI data already at the production site. The information becomes quality assured and the routines become independent of the persons who work. It was also learned that such benefits are not sufficient for introducing systematic environmental information management; most companies chose not to implement the system after the project was over. The reasons need to be further assessed and understood.

DfE information management in rail industry 155

160

165

The first system based on the data format for impact assessment (see section ‘Data’ above) was tested within CPM, for electronics manufacturing [51] and later also for DfE applications in the European project RAVEL (Rail VEhicLe eco-efficient design)[52]. An extension of the SPINE format and an application of the CPM data quality management were integrated with a DfE system and an eco-procurement system for the European railway sector. One result is that the host of the Swedish national LCA database today also holds a database of environmental properties of materials for measuring environmental performance of rail vehicles. A valuable learning from the RAVEL project is the understanding of a direct relationship between environmental indicators and choice of contents of an environmental database; data should not be collected before it is known how it measures environmental performance, and environmental indicators should not be defined unless one is clear about data availability. Environmental performance indicators and database contents are two sides of the same thing, the environmental information system.

Sector independency Semi-automated reviewable reporting 170

175

The Swedish Environmental Product Declarations (EPD) [53] system was developed in parallel with the Swedish LCA database. It was intended that the database should support the EPD system. The first contribution from the LCA database was to share its review principles; the EPD system was to apply the same documentation criteria as the database [54]. Many Swedish LCAs for EPD purposes have utilised data from the Swedish LCA database, and the database has developed to more efficiently support the EPD system. A methodology to generate fully transparent and reviewable complete LCA studies [55] is described, and parts of it will be implemented as prototype tool in a CPM project named IMPRESS [56]. It is expected that integration of web-based LCA tools for reviewable LCA studies with the Swedish national LCA database will reduce costs for performing reviewable LCAs.

Approaches to simplify data acquisition 180

185

In parallel with integrating the LCA database with different software applications and LCA data acquisition routines with operative EMS, approaches to simplify continuous LCA data acquisition have also been made. Three examples are: • It was assessed whether the environmental reporting that Swedish companies are obliged to submit to the government includes the information that is needed to perform an LCA [57]. It was found that since they lack data on production volumes they were not fit. (Assessments of usability of a combination of publicly available economic and environmental reporting have not yet been made.) • Principles and tools were developed to support and encourage LCA data providers to establish mutual LCA data trading [58]. Courses were provided by CPM, and software, common nomenclatures and a common data exchange format were made available at no charge. In spite of this, the idea was not implemented. This was largely due to competition between initiators, and

4(10)


190

195

•

today a new technical platform has been developed to facilitate for different ideas and business principles [59]. A guideline for LCI data acquisition strategies for companies or LCA projects were produced [60]. The guidelines are based on limited practical tests in the electronics sector.

Learnings from simplifying LCA data acquisition into general strategies and business networks may form basis for next generation of LCA data support systems. Currently there is no approach that both meet LCA data documentation requirements and that are propelled by an economically sustainable business model. The LCA data acquisition problem is not solved by a database.

Harmonisation and standardisation 200

205

210

215

When the Swedish national LCA database-work started in the mid 1990s it was understood that life cycle assessment was in its infancy. Much was to come, such as the integration with EMS and international exchange of environmental product data. Therefore, already from the beginning it was understood as necessary to work towards international harmonisation of life cycle data formats, and that these needed to be flexible beyond LCA. The development of the common Nordic format SPINE [13] was a first step in this direction. The next step was taken in 1997 to propose initiation of an ISO standardisation of a LCA data documentation format. In parallel to this CPM also participated to develop and improve the SPOLD format [4] [61], and in Sweden harmonisation efforts aimed to facilitate data exchange between software and LCA users in consultant agencies, industrial companies and research institutes [62]. When ISO released the technical specification ISO/TS 14048 Data documentation format in 2002 [63] CPM published examples [64], a computer implementable version of the format [65], a mapping-table between SPINE and ISO/TS 14048 [66], a guide for documenting LCA data using the ISO specification [67], a report that describe how to apply the CPM documentation criteria on the ISO format [68], and a quick guide for LCA practitioners to switch from SPINE to ISO/TS 14048 [69]. Learnings are that harmonisation investments are worthwhile. The Swedish contributions to documentation requirements on LCA data are globally accepted, though not always applied. Another learning is that harmonisation takes time; over ten years of Swedish development is relatively quick when compared internationally, much due to the steady financing of CPM and the close relationship with industry work. Harmonisation efforts may sometimes be regarded as threats, unless care is taken to help others to catch up and prosper from the mutual benefits.

The Swedish LCA database today 220

225

230

235

240

In March 2005 the Swedish LCA database holds 536 LCI data sets, covering energy and transport sectors and many industrially consumed raw material and products. Data sets are sold for a price in the range of USD 150 per data set. The data can be viewed and purchased in the format of either SPINE or ISO/TS 14048. These and other facts about the database were presented at The Sixth International Conference on EcoBalance, Oct. 25th - 27th, 2004, Tsukuba, Japan [70]. Over time, the different learnings described in this article have been disseminated to industrial companies in CPM since the mid 1990s. Two distinct learnings about LCA databases in general have been noted during the way: • How to efficiently set up a practically useful LCA database [71]. • How to continually improve an LCA database according to practical experiences [72]. Further key learning is that well-structured documented data is valuable and can be reused; the industrial LCA practitioners that have worked with the SPINE format and who applied the CPM documentation criteria find it easy to understand the contents of their own LCA databases long after they entered them into the database. Involving new colleagues is easy, whether it concerns the in-house LCA database or the national database. Much LCA costs have been saved due to that. But many LCA practitioners still perform data collection without documenting sufficiently. Long-term economic benefits are difficult to communicate. Over the years, the Swedish national database administration matured into a research unit within the university, with the name of Industrial environmental informatics, IMI [73]. The research is based on questions and findings when establishing the national LCA database and addresses issues such as environmental indicators and performance measurements, setting up databases, defining data quality

5(10)


systems and data acquisition. Much of the research is performed through software prototypes and different industrial environmental databases.

Conclusions 245

250

255

260

265

270

275

All conclusions from establishing the Swedish national LCA database in joint CPM co-operation cannot be condensed in one article. But some essential conclusions will be noted here. Many of the original criteria for the national database proved true, such as that: • The data should be documented in English. It facilitated that the database can be appreciated also in 2005 when all Swedish companies that entered CPM in 1996 are now parts of global corporations. • The database should hold data for different types of LCA and the database should contain metadata and be transparent. It ensured that users in 2005 can review data entered in 1996. • It should be possible to connect the database to different (software) applications. A fact that has been proven by integration into DfE-tools, web-based LCA tools and impact assessment databases etc. • The data quality must be assessed before data are entered into the database. This has facilitated data management and has held maintenance costs at a minimum, since each data set need to be assessed only once. • The database was expected to become systematically updated. But this has not been done, though voices today claim that it should. There also are learnings that are results from choices of technologies and strategies, rather than from original criteria. Examples of such learning are • The LCA database should comply with industrial technology and quality standards, which has facilitated integration with industrial data systems and EMS. • Ten years of governmental back-up, industrial commitment and academic funding provided the courage to form serious questions that established new knowledge about data for product life cycles. • The quality management view of LCA data learned more than that had been questioned before, i.e. that it had direct implications to quality management of EMS in general. • There were great benefits from having compatible data formats for LCI and impact assessment data. Integration of LCA tools for transparent LCA studies is an example of that, and costs for developing different LCA tools can be decreased. • The CPM data documentation criteria are easy to understand and flexible to any industry or any application. The weakness is that users find it tedious and time consuming to document the data. • Harmonisation efforts are well worth the investments and need to take time, but the broad practical success may be the reason why the Swedish approach have not been fully appreciated in different LCA communities; practical tests used resources that could have been used for dissemination and marketing of results.

280

285

290

From the close relationship with industry, the development of LCA applications could be foreseen early, in terms of reuse of LCA data, integration into industrial environmental management and design for environment and green procurement. CPM development has made some early findings: • It is practically possible and economically beneficial to produce well documented LCI data already at the production site. However, these benefits are not sufficient for introducing such routines (see section ‘Information management system in paper and pulp industry’). • Environmental performance indicators and database strategy are two sides of the same thing; there is a direct relationship between environmental indicators and choice of contents in an environmental database for environmental performance measurements (see section ‘DfE information management in rail industry’). • Documented and well-structured data can be reused, and LCA costs can therefore be saved. However, many LCA practitioners still do bad documentation since such long-term benefits seem meaningless to inexperienced practitioners (see section ‘The Swedish LCA database today’).

6(10)


295

300

305

310

315

320

CPM learnings has been disseminated in e.g. Swedish EPA reports [74] and other articles and reports, and as results out from e.g. the project DANTES [75]. CPM’s international lead concerning LCA data handling has also been confirmed by three international assessments of CPM [76]. A general learning is that when producing a new type of database one should expect to have to solve interesting problems, and this work might lead farther than was intended by the actual task. From the viewpoint of the new research field of industrial environmental informatics, the intensive work at the cross section between industry and academy during the establishment of the Swedish national LCA database was an ideal spot for evolution of new ideas. Development, rejection and redevelopment of tools, methodologies and a new scientific approach have been quick.

Future development The Swedish national LCA database is a fundament on which new work can be built. Practical issues to deal with already are the updating, or the second version of the database. Since one of the key learnings is that there is a direct relationship between environmental performance indicators (EPIs) [77] and the database content, future databases should be established on well defined EPIs, and much effort should be spent on identifying the intended user groups and their actual needs. This will save cost and ensure relevance. Data acquisition of different kinds, such as laboratory tests and measurement in natural environments should be performed to improve characterisation modelling, and demographic studies should be performed to improve weighting and prioritisation modelling. LCI data is not valuable on its own. Harmonisation efforts are regarded as threats and care must be taken to assist and disseminate results. This will be achieved by stronger participation in different international networks, e.g. UNEP/SETAC LC Initiative [78] etc. But the LCA database management shall also still function as a living learning forum, for integration of industrial environmental information systems [79], and for further studies that relate environmental information and data quality issues with aspects of the inter disciplinary fields of cognitive science [80].

References 325

330

335

340

345

[1] Svensson G., Ekvall T.: “Criterias for a National Database for Life-Cycle Assessment”, Chalmers Industriteknik, Gothenburg, AFR-REPORT 37 (1994) [2] Ryding S-O.: “Miljöanpassad produktutveckling”, Sveriges industriförbund, Stockholm, Sweden, Industrilitteratur (1995) [3] Tillman, A.-M., Baumann, H., Eriksson, E. och Rydberg, T: “Packaging and the Environment”, Offprint from SOU 1991:77, CIT Chalmers Industriteknik, Göteborg (1992) [4] Singhofen A., Hemming C., Weidema B., Grisel L., Bretz R., de Smet B., Russell D.: Int. J. LCA, 1(3) (1996), pp. 171-178 [5] Udo de Haes, H.A., Wriesberg N.: “Life-Cycle Assessment: State-of-the-Art and Research Priorities - Results of LCANET; a Concerted Action in the Environment and Climate Programme (DGXII)”, LCA Documents Vol 1. (1997), pp. 120-147 [6] Competence center CPM Center for environmental assessment of Product and Material system, Chalmers university of technology (Date: 6/2/2005) [7] Chalmers university of technology, Chalmers, (Date: 6/2/2005) [8] NUTEK, the Swedish Business Development Agency, (Date: 6/2/2005) [9] VINNOVA, the Swedish Agency for Innovation Systems, (Date: 6/2/2005) [10] Carlson R.: “12 Vanliga Frågor med Svar om CPM:s LCA-Databas”/”12 Frequently Asked Questions on CPM:s LCA Database”, CPM-rapport 1997:3, CPM, Chalmers University of Technology (1997) [11] Carlson R.: “Strategin kring arbetet med CPM:s LCA-Databas”/”Strategy for the Work on CPM:s LCA Database”, CPM-rapport 1997:4, CPM, Chalmers University of Technology (1997) [12] Carlson R., Pålsson A-C: “Establishment of CPM's LCA Database”, CPM Report 1998:3, CPM, Chalmers University of Technology (1998)

7(10)


350

355

360

365

370

375

380

385

390

395

400

[13] Carlson R., Löfgren G., Steen B.: “SPINE, A Relation Database Structure for Life Cycle Assessment”, IVL Swedish Environmental Research Institute, Göteborg, Sweden, IVL-Report B 1227, (1995) [14] Carlson R., Löfgren G., Steen B., Tillman A-M.: Int. J. LCA, 3(2) (1998), pp. 106-113 [15] ISO 14042: Environmental management - Life cycle assessment - Life cycle impact assessment, ISO International Organisation for Standardisation (2000) [16] Carlson R., Steen B.: Proc. 8th Annual Meeting of SETAC-Europe, Bordeaux, France (1998) [17] World Wide LCA Workshop (Date: 6/2/2005) [18] Goedkoop M., Spriensma R.: “The Eco-indicator 99, A damage oriented method for Life Cycle Impact Assessment Methodology” Report 2:nd ed. (2000) [19] Wenzel H., Hauschild M., Alting L: “Environmental assessment of products Vol. 1 Methodology, tools and case studies in product development”, London Chapman & Hall (1997) [20] Hauschild M.,Wenzel H., “Environmental assessment of products Vol. 2 Scientific background”, London Chapman & Hall (1998) [21] Steen B.: “A Systematic Approach to Environmental Priority Strategies in Product Development (EPS) Version 2000 - General System Characteristics”, CPM-report 1999:4, CPM, Chalmers University of Technology (1999) [22] Steen B.: “A Systematic Approach to Environmental Priority Strategies in Product Development (EPS) Version 2000 - Models and Data of the Default Method”, CPM-report 1999:5, CPM, Chalmers University of Technology (1999) [23] Carlson R., Pålsson A-C.: “Documentation of environmental impact assessment, compatible with SPINE and ISO/TS 14048” IMI-report 2002:1, IMI, Chalmers University of Technology (2002) [24] Arvidsson P. (Editor) et al: “Krav på datakvalitet CPM:s databas 1997”, CPM-rapport 1997:1, CPM, Chalmers University of Technology (1997) [25] Arvidsson P. (Editor) et al: “An interpretation of the CPM use of SPINE in terms of the ISO 14041 standard”, CPM-rapport 1999:9, CPM, Chalmers University of Technology (1999) [26] Pålsson A-C.: “Introduction and guide to LCA data documentation using the CPM data documentation criteria and the SPINE format”, CPM Report 1999:1, CPM, Chalmers University of Technology (1999) [27] Pålsson A-C.: “Review of LCI-data at SPINE@CPM”, CPM Internal Report 1999, CPM, Chalmers University of Technology (1999) [28] von Bahr B (2001): “The Relevance Aspect of Life Cycle Inventory Data Quality”, Licentiate thesis, Environmental Systems Analysis, Chalmers University of Technology (2001) [29] Carlson R., Pålsson A-C.: Proc. 12th International Symposium Computer Science for Environmental Protection, Bremen, Band 1/Volume 1 (1998), pp. 252-265 [30] Carlson R., Pålsson A-C.: Proc. Int. Workshop on Quality of LCI data, UNEP/SETAC Life Cycle Initiative, Oct. 20-21, 2003, Forschungszentrum Karlsruhe, Germany (2003), pp. 35-38 [31] Carlson R., Pålsson A-C.: “PHASES Information models for industrial environmental control”, CPM-report 2000:4, CPM, Chalmers University of Technology (2000) [32] Carlson R., Pålsson A-C.: Journal of Cleaner Production, 9, (2001) pp 429-435 [33] Website presenting the thematic network CASCADE (Co-operation and Standard for Life Cycle Assessment Data in Europe) financed by the European Commission: (Date: 6/2/2005) [34] Weidema B. P, Cappellaro F., Carlson R., Notten P., Pålsson A-C, Patyk A, Regalini E., Sacchetto F., Scalbi S.: “Procedural guideline for collection, treatment, and quality documentation of LCA data”. Document LC-TG-23-001 of the CASCADE project, (Date: 6/2/2005) [35] Carlson R., Pålsson A-C, Notten P, Cappellaro F, Scalbi S, Patyk A.: Proc. Int. Workshop on Quality of LCI data, UNEP/SETAC Life Cycle Initiative, Oct. 20-21, 2003, Forschungszentrum Karlsruhe, Germany (2003), pp. 51-54 [36] Taprantzi, A.: “A systematic approach for acquiring industrial data and information for environmental applications”, Master’s thesis, Uppsala University, Uppsala, Sweden (2002) [37] Nilsson I.: “Integrating Environmental Management to Improve Strategic Decision-Making”, CPM Report 2001:2, CPM, Chalmers University of Technology (2001) [38] Carlson R., Pålsson A-C.: Proc. 9th Annual Meeting of SETAC-Europe, Leipzig, Germany (1999)

8(10)


405

410

415

420

425

430

435

440

445

450

455

[39] Häggström S.: “Policy controlled environmental management work - Problem Inventory Report” CPM-report 2004:9, CPM, Chalmers University of Technology (2004) [40] Carlson R., Häggström S., Pålsson A-C.: “Policy controlled environmental management work Final report” CPM-report 2004:10, CPM, Chalmers University of Technology (2004) [41] Carlson R., Häggström S., Pålsson A-C.: “Policy controlled environmental management work Manual”, CPM-report 2004:11, CPM, Chalmers University of Technology (2004) [42] Pålsson A-C., Flemström K.: “Gap analysis of the documents in the ISO 14000-series with regard to quality management of environmental data and information”, CPM-report 2004:3, CPM, Chalmers University of Technology (2004) [43] Björklund T., Carlson R., Pålsson A-C.: Proc. CIB World Congress 1998, Gävle, Sweden (1998) [44] Carlson R., Tillman A-M.: Proc. Systems engineering models for waste management –Int. workshop in Göteborg, Sweden Feb. 25 -26 1998 (1998) (Date: 6/2/2005) [45] Erixon M.: “Practical Strategies for Acquiring Life Cycle Inventory Data in the Electronics Industry”, CPM Report 1999:3, CPM, Chalmers University of Technology (1999) [46] Erixon M.: “Formatting Data for EAA According to the CPM Data Documentation Criteria” CPM report 2002:2, CPM, Chalmers University of Technology (2002) [47] Pålsson A-C, Enqvist A., Karlsson G., Loviken G., Möller Å, Nilseng AB, Nilsson C., Olsson L., Svending O.: “Methodology for handling forest industry environmental data – Method report”, CPM-report 2005:1, CPM, Chalmers University of Technology (2005) [48] Pålsson A-C, Enqvist A., Karlsson G., Loviken G., Möller Å, Nilseng AB, Nilsson C., Olsson L., Svending O.: “Methodology for handling forest industry environmental data – Manual”, CPMreport 2005:2, CPM, Chalmers University of Technology (2005) [49] Svending O.: “Internal allocations in the Swedish pulp and paper industry”, CPM report 2002:3, CPM, Chalmers University of Technology (2002) [50] Pålsson A-C., Svending O., Möller Å., Nilsson C., Olsson L., Loviken G., Enqvist A. Karlsson G., Nilseng A.: Proc. 7th International Conference on New Available Technologies, Stockholm, June 4-6 (2002), pp.177-181 [51] Erixon M.: “Information system supporting a web based screening life cycle assessment tool”, CPM Report 2001:14, CPM, Chalmers University of Technology (2001) [52] Dewulf W. (Ed.) et al: “Integrating Eco-Efficiency in Rail Vehicle Design”, Leuven University Press, Leuven, Belgium (2001) [53] The international site for Environmental Product Declaration, EPD (Date: 6/2/2005) [54] Eriksson E., Lindfors L-G., Pålsson A-C., Ribbenhed M.: “Manual för granskning av livscykelanalyser - med applikation på EPD”, AFR-report 248, Swedish Environmental Protection Agency (1999) [55] Erixon M., Carlson R., Pålsson A-C.: Proc. Int. Workshop on Quality of LCI data, UNEP/SETAC Life Cycle Initiative, Oct. 20-21, 2003, Forschungszentrum Karlsruhe, Germany (2003), pp. 24-27 [56] Website describing CPM project IMPRESS (Implementation Of Integrated Environmental Information Systems) (Date: 6/2/2005) [57] Erixon M. Ågren S.: “Miljörapporter som underlag till livscykelanalys”, CPM-rapport 1997:5, CPM, Chalmers University of Technology (1997) [58] Carlson R. Erixon M., Pålsson A-C.: “Etablering av handelsstrukturer för LCI-data, En rapport som beskriver CPM:s strategi för utveckling av LCI-datahandel”/”Establishment of Trade Structures for LCI Data”, CPM-rapport 2000:3 CPM, Chalmers University of Technology (2000) [59] Mustafa M., Wigren P.: “Secure XML file sharing in a JXTA P2P network for inter-organizational industrial collaboration” Master's thesis, Chalmers University of Technology, Göteborg, Sweden (2004) [60] Carlson R., Erixon M.: Proc. 4:th International Conference on EcoBalance, Tsukuba, Japan (2000), pp. 719-722 [61] Erixon M. Ågren S.: “An Assessment of the SPOLD format”, CPM Report 1998:5, CPM, Chalmers University of Technology (1998) [62] Erixon M. (Editor) et al: “Facilitating Data Exchange between LCA Software involving the Data Documentation System SPINE”, CPM Report 2000:2, CPM, Chalmers University of Technology (2000)

9(10)


460

465

470

475

480

485

490

495

[63] ISO/TS 14048: (E) Environmental management – Life Cycle assessment – Data documentation format, ISO International Organisation for Standardisation (2002) [64] Carlson R., Pålsson A-C.: “First examples of practical application of ISO/TS 14048 Data Documentation Format”, CPM-report 2001:8, CPM, Chalmers University of Technology (2001) [65] Carlson R., Tivander J.: “Data definition and file syntax for ISO/TS 14048 data exchange”, CPM Report 2001:9, CPM, Chalmers University of Technology (2001) [66] Carlson R., Erlandsson M., Flemström K., Pålsson A-C., Tidstrand U., Tivander J.: “Data format mapping between SPINE and ISO/TS 14048”, CPM report 2003:8, CPM, Chalmers University of Technology (2003) [67] Flemström K., Pålsson A-C.: “Introduction and guide to LCA data documentation using the CPM documentation criteria and the ISO/TS 14048 data documentation format”, CPM report 2003:3, CPM, Chalmers University of Technology (2003) [68] Flemström K., Pålsson A-C.: “An interpretation of the CPM data quality requirements in terms of ISO/TS 14048 data documentation format” CPM report 2003:4, CPM, Chalmers University of Technology (2003) [69] Flemström K., Pålsson A-C.: “Quick guide when switching data documentation format from SPINE to ISO/TS 14048” CPM report 2003:5, CPM, Chalmers University of Technology (2003) [70] Pålsson A-C., Carlson R.: Proc. 6th International Conference on EcoBalance, Tsukuba, Japan (2004), pp. 147-150 [71] Carlson R., Pålsson A-C., Karlson L., Wennsten J., Brohammer G., Blanco L., Swahn G., Arvidsson P.: Proc. LCM 2001, 1st International Conference on Life Cycle Management, Copenhagen (2001), pp. 269-272 [72] Carlson R., Erixon M., Pålsson A-C., Mattson G., Ny H., Hallberg K., Person L., Blanco L., Broberg R.: Proc. LCM 2001, 1st International Conference on Life Cycle Management, Copenhagen (2001), pp. 273-276 [73] Website of Chalmers research group IMI Industrial Environmental Informatics, (Date: 6/2/2005) [74] Erixon M., Carlson R., Pålsson A-C.: “Measuring the environmental impact of products”, CPM report 2003:1, Swedish Environmental Protection Agency Report 5349 (2003) [75] Website of EU financed demonstration project DANTES Demonstrate and Assess New Tools for Environmental Sustainability (Date: 6/2/2005) [76] Baras J. S., Frysinger S., Inaba A., Stenius P.: “CPM International Evaluation Report”, CPMreport 2004:2, CPM, Chalmers University of Technology (2005) [77] Carlson R.: “Environmental Performance Indicators”, INSIGHT, The International Council on Systems Engineering (INCOSE), Vol 5 Issue 2 (2002), pp. 22-23 [78] The United Nations Environment Program (UNEP) and the Society for Environmental Toxicology and Chemistry (SETAC) global Life Cycle Initiative (Date: 6/2/2005) [79] Carlson R., Erixon M., Forsberg P., Pålsson A-C.: “System for Integrated Business Environmental Information Management”, Advances in Environmental Research, 5/4 (2001), pp. 369-375 [80] Carlson R., Pålsson A-C.: Proc. 6th International Conference on EcoBalance, Tsukuba, Japan (2004), pp. 339-340

10(10)