Book Reeviews - IEEE Xplore

1 downloads 0 Views 638KB Size Report
ysis-Samuel Eilon, C. D. T. Watson-Gandy, and Nicos Christofides. (London: Griffin, 1971 .... Ralph F. Miles, Jr., Ed. (New York: Wiley, 1973, 221 pp.). Reviewed by James S. Dyer, Graduate School of Management, University of. California, LosĀ ...
589

IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS, NOVEMBER 1974

Book Reeviews Distribution Management-Mathematical Modelling and Practical Analysis-Samuel Eilon, C. D. T. Watson-Gandy, and Nicos Christofides (London: Griffin, 1971, 240 pp.). Reviewed by Richard de Neufrille, Institute of Transportation and Traffic Engineering, University of Calijornia, Berkeley, Calif. 94720.

Despite the extensive advances that have been made over the last our ability to design optimal networks in practical situations is often extremely limited. For some problems, in particular for transportation and logistics, this capability has, in fact, been almost nonexistent. To be sure, our accumulated experience in transport planning does enable us to avoid the most ineffective kinds of networks. However, this is quite different from knowing what kinds of networks are truly optimal. For many modes of transport, we do not even know which of significantly different networks are better. In air transport, for example, a major argument is now taking place in Europe over whether a more

twenty years in network analysis,

connected or (as in the United States) a less connected network is to be preferred, and a similar debate is also taking place in the United States over whether the connectivity of the operational network on the railroads ought to be changed by the introduction of more through trains. Finally, there is continuous disagreement among those interested in urban goods movement about whether consolidation or similar warehouses are desirable. Curiously enough, these discussions are generally carried on without the benefit of any analysis, let alone of the kind of mathematical network analyses that have been so

successfully perfected. The book by Eilon, WatsonGandy, and Christofides is a significant contribution in this context. At last we have a reference that clearly describes and contrasts the different kinds of approaches that are available, that assesses their relative usefulness for practical problems, and that provides some clues as to what kinds of solutions are desirable. As a further advantage, the exposition is clear, and the book is easy to read.

The authors write with authority, justly due to their extensive academic and practical experience. Ihey are all on the faculty at Imperial College in London, the premier technical center in the United Kingdom, and have been teaching distribution and logistics for many years. In addition, their work there has been combined with extensive consulting work so that they have been able to develop a direct sense of what kinds of problems arise in practice and what analytic approaches are most useful. The first half of the book focuses on the depot location problem. It features an extensive review of the literature (much along the same lines as that published as a short article by ReVelle, Liebman, and Marks), coupled with a critical review of the relative modeling and computational advantages of the competitive approaches. This is illustrated through a case study of an actual large-scale problem. The second half covers the associated problems of vehicle scheduling, loading, and the determination of fleet size. The nature of the presentation is similar to that of the first half. The major difference is that the state of the art in these areas does not allow such definitive statements as for depot location. Thus, while the treatment is inherently somewhat inconclusive, it appears to be the best that I have yet encountered for

practical problems. In general, the text is an excellent introduction to practical analysis of networks for logistics and distribution management. Supplemented by descriptions of recent interesting advances, such as those of David Marks' group on garbage collection or of Nigel Wilson et al. on consolidation terminals for urban goods movements (all available through the M.I.T. Civil Engineering Systems Laboratory, Cambridge, Mass. 02139), this text is a good basis for a course in distribution. It also provides a first-rate reference for operation researchers practicing in this area. The book is highly recommended for both purposes.

It is unfortunate that this worthwhile work is unfamiliar to the U.S. public. This is possibly due to the fact that the U.K. publisher is

relatively unknown in North America. If this is the case,

one

would

hope that he, along with other publishers of good texts, would

co-

venture their work with transatlantic distributors so as to facilitate the

dissemination of such works.

Reguleringsteknikk (Control Engineering), vols. 1 and 2--Jens G. Balchen, vol. 3 Jens G. Balchen, Magne Fjeld, and Ole A. Solheim (Trondheim, Norway: Tapir; 1973, 1971, 1970; 274 pp., 331 pp., 392 pp.). Reviewed by Odd Pettersen, Department of Electrical Engineering, Technical University, Trondheim, Norway; Visiting Professor, Department of Electrical and Compater Engineering, University of Wisconsin, Madison, Wis. 53706. These three books cover different topics within the area of control engineering and may well be read separately. Together they constitute a complete set of courses on the undergraduate and partly graduate level. As such, they are used as main textbooks in courses at the Technical University of Norway, where the authors are professors and pioneers of automatic control, not only in theoretical aspects but also a wide range of areas of industrial application. The books are widely read and have gained a high reputation throughout Scandinavia, where the Norwegian language is easily understood. As a former student of Professor Balchen and a colleague of all three authors for several years, the reviewer is, perhaps, somewhat biased in favor of the books. On the other hand, he knows the books fairly well and has witnessed their effective use in teaching for many years. Volume One was originally published in 1965, based upon earlier lecture notes. It was later revised and reprinted several times, its last edition being the sixth repiint. The two other volumes are newer, which is easily verified by noticing the notation and the use of "modern control theory" with emphasis on vector-matrix formulation. Particularly, this approach is adopted in Volume Three. Volume One is organized into 12 chapters and covers the main parts of an elementary two-credit course. Because of its historical evolution through more than 15 years, Volume One is mainly concerned with the tranditional approach to control theory, based upon the Laplace transform. The book is updated, however, and state space analysis has been included as a separate chapter at the end of the book. Chapter 1 explains, in a very elementary and almost nonmathematical way, the nature and some important aspects and applications of automatic control and cybernetics, with examples from technical applications, biology, and other areas of man's activity and environment. Chapter 2 gives a mathematical basis for linear control systems, with basis in the Laplace transform. Chapters 3 and 4 start with an Introduction to the concept of block diagrams, and show how to derive time responses with the help of the inverse Laplace transform.

Furthermore, poles and zeros and their effects are explained, with a brief view of the root locus miethod. Chapters 5-7 deal with frequency analysis and stability; whereas Chapters 8 and 9 apply that theory in the synthesis of feedback control systems, with the mention of feedforward for the elimination of certain kinds of noise. Chapter 10 explains some aspects of rnutually-coupled systems, with the Laplace-transform and transfer-function approach. Several examples of two-variable systems are explained in some detail. Chapter 11 deals with the limitations inherent in linear systems, and chapter 12 gives the basic theory of state space analysis. In an appendix, the inverse Laplace transform is derived, and another appendix gives some "cook book" methods for the practical application and construction of frequency diagrams. At the end of the book is included a bibliography, with some reference to chapters in the book. Volume Two covers four distinct areas of control theory above the fundamental level: discrete control systems, nonlinear control systems, methods for statistical analysis, and experimenital determination of system parameters. Chapter 1, covering discrete systems, gives an easily understood engineer-oriented mathematical description of the

590 z transform. It includes a stability analysis and conditions and a short section introducing digital filtering. Chapter 2, about nonlinear control systems, begins with a short survey that shows the importance of nonlinearities in nature and in technical systems. The most important classes of nonlinearity are introduced. Limit cycles are treated in a separate section. Among the analysis methods explained are the Isocline method, Liapunov's second method for stability investigation of autonomous systems, describing functions, and Popov's method for investigation of global stability. One section gives a general state-space approach to the description of dynamical systems, without particularly emphasizing nonlinear systems. Chapter 3 thoroughly describes the basic aspects, terms, and methods of statistical analysis, including correlation functions, distribution, energy- and power-spectrums, and optimum filtering. Continuous and discrete systems are treated. In Chapter 4, a wide range of methods is introduced for the experimental determination of systems parameters. These include parameter estimation, the transient response method, cross-correlation, and adjustable models. Bibliographies are included following each chapter, but without much reference in text. Volume Three has the subtitle "Multivariable Systems" and follows entirely the modern approach of vector-matrix formulation. Chapters 1-3 mainly describe terms and present definitions. Chapter 4 deals with multivariable control and includes topics such as decoupling, modal control, state estimation with Kalman filtering, hierarchical systems, and adaptive control. Chapter 5 concerns optimal control of continuous systems and specifically describes static control, the calculus of variations, the maximum principle, optimization of linear systems with quadratic criterions, two-point boundary value problems, and optimum "tuning" of Kalman filters. Chapter 6 is called "Discrete Systems" and essentially covers the same topics as Chapter 5, but for discrete systems. Chapter 7 deals with distributed-parameter systems and begins with a collection of examples of different classes of problems. It continues with sections on stability, discretization, state-space formulation, different methods for approximation, optimal control, and ends with a section on estimation of state variables of distributed systems. Finally, in this volume is included two appendices: one that gives a concentrated, but quite comprehensive, view of matrix algebra and one that deals with the proving of the maximum principle. Like Volume Two, literature references are included after each chapter. Some reference to these are given in the text. None of the books includes exercises or problems for solution by the student. However, almost all sections and theoretical aspects are clarified by numerous examples, most of them from practical applications.

Systems Concepts: Lectures on Contemporary Approaches to SystemsRalph F. Miles, Jr., Ed. (New York: Wiley, 1973, 221 pp.). Reviewed by James S. Dyer, Graduate School of Management, University of California, Los Angeles, Calif. 90024. Systems engineering is a term that has been used to describe efforts directed toward the solution of large-scale complex technical problems. Perhaps the most outstanding example of the application of systems engineering is provided by the Apollo Program. More recently, systems engineering, or the systems approach, has been touted as a means of solving complex social problems in areas such as health, education, and transportation. If systems engineering can put a man on

the moon, so the argument goes, it must be capable of devising a plan to ensure adequate health care for all. Certainly there is a high degree of interest in systems engineering among professional engineers, managers, and laymen. However, many books dealing with this important topic primarily address philosophical issues, while others concentrate on a mathematical treatment of the elegant tools and techniques used by systems engineers. Systems Concepts: Lectures on Contemporary Approaches to Systems provides the complete overview of systems engineering that has been missing in other books. The presentation of the materials is sufficiently straightforward to allow nontechnically oriented readers to grasp the basic concepts. In addition, several chapters contain such insights that professionals in the field of systems engineering would be well advised

fEEE

TRANSACI'IONS ON SYSTEMS, MAN, AND CYBERNETICS, NOVEMBER 1974

to read them carefully. As an indication of the diverse appeal of the book, it is interesting to inote that favorable reviews have appeared previously in Fortune and Operations Research, which caier to nontechnical and technical audiences, respectively. The book contains eleven chapters based on a series of lectures on systems engineering presented at the California Institute of Technology, Pasadena, Calif., in the Spring of 1971. Three of the chapters are overviews of systems engineering (or systems analysis), four discuss the techniques used by systems engineers, two describe applications of systems engineering at the Jet Propulsion Ltaboratory, Pasadena, Calif, and at NASA, and two excellent chapters provide a critique arid a tew caveats concerning the approach. Perhaps the secret that allows this book to cover the technical and philosophical aspects of systems engineering so well is the outstanding list of contributors. For example, R. E. Machol (assisted by R. F. Miles, Jr.) provides a compact overview of the approach to engineering large-scale systems. R, A. Howard presents one of the most readable discussions yet printed of decision analysis and its use in the study of complex problemns. In addition,

C. West Chuichman contributes a numnber ot caveats coiicerning the use of the systems approach within social organizations. Readers who are famniliar with much of the literature on systems engineering niay feel that the views of these authors are alheady well known. These reader s should concentrate oiv the chapters of W. Edwards and R. Boguslaw. Edwards is a Professoi of Psychology whose interesting research on decision processes has appeared in journals perhaps not so famiiiliar to systems analysts and engineers. He reports the results of experiments that indicate that while humans caii imake excellent judgments based on a single itemn of data, they cannot synthesize a number of data properly. Specifically, he prov ides a convincing argument that humans use a conservative rule for the aggregation of probabilistic information relative to the nomrmative results of Bayes rule. To adjust for this teimdency, Emdwar(ds describes a nianmachine interactive system (The Probabilistic Iinformmiation Processor or PIP), which requires humans to make probabilistic judgments on single itemns, but uses a computer to aggregate these judgments and arrive at a final tesult. He reports that PIP has been used as an aid in medical diagnosis and to predict recidivism among juvenile delinquents. The views of Edwards complement the niore faiiiiliar (to systems analysts and engineers) views of Howvard, just as those of Boguslaw complement the vielws of Churchrnan. Boguslaw is conce rtied witlh the role of the engineer in society. He accepts the definitiomi that l cngineering is a discipline that concerns itself with the properties of inatter and the sources of energy irn nature and tries to make them useful to man." However, he points out that man plays a dichotomous i ole in this definition as both the client of the engineer and an irrmportant source of energy. Nonsocial systems, lhe argues, characteristically have a system objective forinulated by and relevant to someone outside the system itself. However, in a social system, the eleii-ents oi components of the systenih define system objectives in tern-is of their own special values arid needs. Thus the engineer who deals with a social system must be willing to accept a role similar to that of a labor arbitrator who is concerined with resolving the conflicts amnong parties in a dispute. Unfortunately, the engineer's role is even more difficult sinice it ilnay be impossible to identify all ot the parties who should be involved in the arbitration. Finally, Boguslaw argues that the engitieer of the future will not have the choice of dealing with either nonsocial or social systenms, since the definition of a social systemn is iniplicit in any work in which they are engaged. "As they understand their iole as designers, as well as occupants of social systetms, they will begin to participate more fully in the decisions which shape the conditions of life for thernselves and everyone else oni this plafmet." Naturally, this book is not without its faultsIThe coverage of such a wide range of topics in so few pages mneans that maniy important issues are discussed only superficially, while others are omitted entirely. For example, the important concepts of cost/benefit analysis and social welfare economics are mentioned only briefly. Happily, numerous references are provided for readers who are interested in learning more about the materials. In addition, the book suffers from a shortcoming that plagues almost all "readings." That is, the lectures are not as integrated as one would like, and thus both redLundancies and omissions of topics result. It is difficult to imagine that this book would be