1 Wind power prediction and power plant scheduling in ... - CiteSeerX

18 downloads 1168 Views 783KB Size Report
The paper describes part of a Wind Energy Forecasting, Hydrogen Storage and Fuel Cell Application ... Board (ESB) to a new energy source, which cannot.
Wind power prediction and power plant scheduling in Ireland Corinna S. Moehrlen1and Eamon J. McKeogh2 Sustainable Energy Research Group, Department of Civil and Environmental Engineering, University College Cork, Cork, Ireland 1 Phone: +353-21-903025, Fax: +353-21-276648, email: [email protected] 2 Phone: +353-21-902524, Fax: +353-21-276648, email: [email protected]

ABSTRACT The paper describes part of a Wind Energy Forecasting, Hydrogen Storage and Fuel Cell Application project currently under way, funded by the HEA Program Research in Third level Institutions 2000-2003. The need for an automatic on-line scheme for operational purposes in the daily planning and dispatch of electricity was first identified by the Irish Electricity Supply Board (ESB) to overcome technical barriers and ensure secure functioning of Unit Commitment and Economic Dispatch in an environment of increasing wind energy. The aim of the ‘Wind Energy Forecasting’ is to (1) gain experience in the wind forecasting for operational purposes using HIRLAM and WPPT, (2) investigate the potential of using a non-hydrostatic model with high spatial resolution for operational purposes and resource assessments. Keywords: 1. wind forecasting,,2. meteorological weather prediction, 3. wind power prediction

1. INTRODUCTION AND BACKGROUND Even though Ireland has one of the best wind energy climates in Europe the full potential is not realised due to perceived technical and nontechnical barriers associated with the characteristics of the wind resource, including difficulties associated with planning and availability of financial investment [1]. The technical barriers relate primarily to the electrical grid transmission and distribution networks' ability to operate in a stable way when subject to variable power input. The non-technical barriers related in the past to conservative attitudes within the Electricity Supply Board (ESB) to a new energy source, which cannot be controlled in the same way as conventional fossil fuel based power stations and therefore power cannot be dispatched in the traditional way [2]. This has changed with the opening of the electricity market in February 2000. ESB is now aware of the need to develop new strategies for the implementation of wind energy into the grid, which will give a higher value to the wind energy installed capacity and will reduce the need for backup capacity. ESB pointed out that Unit Commitment and Economic Dispatch functions need to be developed to consider the new operating conditions. Furthermore, the Renewable Energy Strategy Group recommended that appropriate research studies should be carried out in the short term to ensure that the wind energy generated electricity can be accommodated without becoming a constraint to reaching current and future targets for wind energy penetration. In their newest report [3] the Strategy group pointed out that “A study should be undertaken on all existing windfarms in Ireland to evaluate the short-term impact of wind power

output variation on frequency regulation. This type of study requires the collection of high frequency … “ The development of an automatic on-line operational scheme for the daily planning and dispatch of electricity is therefore needed. For operational purposes, this means that wind speed at particular wind farm sites and the potential power production for penetration into the electricity grid has to be modelled 48 h ahead. Additionally, windfarm owners indicated increased interest in wind forecasts in order to improve their power production forecasts. In the liberalised electricity market a wind farm owner has to predict his power input into the grid one week in advance and balance any differences. The project therefore aims to develop a wind forecasting system according to these new objectives. 2. OUTLINE OF THE PROJECT The project is divided into three phases, which account for national developments and technical constraints, specific to the forecasting environment. State-of-the-art operational meteorological weather prediction from the national meteorological service (Met Eireann) is based on the limited area model HIRLAM (High Resolution Limited Area Model) and produces forecasts on a horizontal grid with 218 x 144 grid points on a 0.297° x 0.297° horizontal Arakawa C-grid (approx. 33km x 33km) on 24 vertical levels. With regard to operational wind forecasting, this resolution is too coarse for existing operational models (i.e. WPPT, Predictor) to be applied. In this sense the Irish situation with

1

regard to operational wind forecasting is different from most European countries, where weather forecasts are produced on 10 km to 5 km horizontal grids . Phase I The first phase deals with the downscaling of the operational weather forecast model to a 5km x 5 km horizontal resolution and a reduced modeling area. The 5km x 5km model runs in a one-way nesting configuration, i.e. initial and time dependent boundary conditions are obtained from the operational model. The first phase is being carried out in collaboration with Met Eireann. The collaboration includes the provision of the modeling system HIRLAM as well as boundary fields from the operational model, which are updated four times per day. The main operational model produces 0–48 hour forecasts at 00, 06, 12 and 18 UTC. Figure 1a and 1b show the horizontal and vertical profiles of the operational 33km model and the 5km model from a test run. The new modeling area is placed from 15° W to 0° E and 50° S to 60° N, with 24 vertical levels. The horizontal grid has 299 x 199 grid points on a 0.05° x 0.05° horizontal Arakawa C-grid.

Figure 1a: Horizontal grid of the operational HIRLAM model (Met Eireann) and the new area. A second aspect of the first phase is the collection of real-time wind speed and power output data from existing windfarms throughout Ireland and the

extension of the current installed wind monitoring masts in Co. Cork. There are currently 13 windfarms in Ireland with an installed capacity of 70 MW, of which most are located on the west coast.

Figure 1b: Vertical profiles of the operational model (left) and the new model (right). The forecasts indicate that by 2005 there will be 600 MW installed electricity generating capacity. This shows clearly the importance of a monitoring system for the evaluation of the short-term impact of wind power output on frequency and the requirements for stability and frequency regulation of electricity supply. The data will serve to evaluate model output errors from the meteorological model and the power prediction models. The collected data will be extremely important for the second phase of the project, i.e. the accuracy of actual wind power output predictions from wind speed. Phase II The second part of the project deals with the modeling of wind power output. In collaboration with the Institute of Mathematical Modeling (IMM) at the Technical University of Denmark, the modeling system WPPT (Wind Power Prediction Tool) will be installed and tested. WPPT forms subareas within a larger area, each covered by a reference windfarm. Predictions are made on a time horizon from half an hour to 36 hours ahead using local climatic variables and meteorological forecasts of windspeed and direction from a NWP model (in our case HIRLAM). An ARX (AutoRegressive with eXoneous input) type model is used for predictions based on measured data. Meteorological forecast variables are then included through a power curve extension model. In the final step, the predicted wind speed for each sub-area is scaled up to cover all wind turbines in the sub-area before it is summarised to give a prediction for the defined reference area [4,5,6]. WPPT comprises a

numerical part and a presentation part with graphical user interface (GUI). It can be considered a “ready-to-go-model”, as it has been in operation in the control centers of Elsam and Eltra in Denmark since October 1997 [5]. WPPT therefore is ideally suited develop the first operational system for wind energy forecasting in in Ireland. A second approach will evaluate the potential of using a regional weather predicition model for wind energy forecasting. The advantage of a numerical weather prediction is that it solves all important dynamical and physical processes explicitly rather than through parameterization. When concentrating on a limited domain, it is possible to model with high spatial resolution. The PSU/NCAR mesoscale model MM5 will be used. MM5 is a limited-area, non-hydrostatic, terrain-following, sigmacoordinate model designed to simulate or predict mesoscale and regional-scale atmospheric circulation. Using MM5 with historical data will, in the early stage, primarily serve to upgrade the existing information database on wind resources in Ireland. It is planned to set up MM5 with an outer horizontal grid of 3km x 3 km and a nested 1km x 1km grid. Grid point predictions and simple model output adjustments will be used to convert wind speeds to wind power output. This method is described in Barbour and Walker [7]. In their approach, model output adjustment is performed with a simple linear regression. Preliminary studies indicate reasonable results, when using the correct calibration periods for the adjustments. The program WindFarm (by ReSoft Ltd.) has been selected to conduct comparisons with the WPPTapproach and wind energy resource assessments. WindFarm uses MS-Mihcro/MS3DJH from the Atmospheric Environment Service of Canada and Zephyr North. The original MS3DJH program was developed for accurate and efficient predictions of boundary layer wind perturbations caused by local topography. Briefly, the model is based on a division of an assumed neutrally-stratified flow field into inner and outer layers. The outer layer is characterised by inviscid, potential flow while, in the inner layer, a balance between advective, pressure-gradient and turbulent-viscous forces is assumed and turbulent transfers are modelled with a simple mixing length closure scheme. Fourier transforms are used to provide the solution. An alternate roughness change model is also available. For specific windfarm site forecasts Windfarm includes the effect of roughness changes, integrates topographic effects and includes added turbulence due to the wind turbines. Although Windfarm cannot be used for operational purposes, it’s features are ideal for a reasonable accurate presentation and analysis of surface winds and

energy yield at potential windfarm sites from historical data, when using meteorological forecasts of wind speed and direction downscaled to a horizontal grid of 1km with MM5. To summarise, the aim of phase II is to gain experience in the wind forecasting for operational purposes using HIRLAM and WPPT, and to investigate the potential of using a non-hydrostatic model with high spatial resolution for operational purposes and resource assessments. Figure 2 shows the expected system. The third part of the project mainly deals with the refinement and assessment of the model system under operating conditions. Analyses of the relationship of forecasted with observed wind speed, wind direction and wind power output will be carried out The evaluation has to address a variety of characteristics of the meteorological forecasts and wind power prediction. The meteorological forecasts have to be evaluated against general influences at the land-sea boundary, i.e. u, v component correlation to observations.

operational

historical

Figure 2: Principle of the Forecasting system In Ireland the Westerlies are predominant, and a higher correlation of the u-components to observations is expected. Other aspects to be evaluated are biases, systematic and non-systematic errors (RMSE, Correlations with observations etc.), terrain features and local pressure gradients. The evaluation of power prediction has to focus on time horizons, choice of vertical levels used for the downscaling onto the surface and the treatment of model output noise/errors. In a study carried out from IMM [5] it has been shown how important the

investigation and evaluation of these parameters is for the operational use of the forecasting model. In particular, the mean value and variance of forecasted wind speed and direction strongly depend on the forecast horizon [5]. Additionally, the relationship between forecasted and observed wind speeds has to be identified. This includes both, the forecasts from the meteorological model and the statistical model (WPPT). Whereas for WPPT this relationship will only include the specific windfarms/sites, the meteorological model will be evaluated against observed wind speed and direction of the monitoring system of Met Eireann and the project monitoring system as described in part I.

3. Conclusion As a result of the Kyoto protocol commitments, power production from renewable energy resources has become a much more important source of future energy supply. The liberalisation of electricity markets worldwide also calls for new strategies for electricity dispatch and planning (spinning reserves, fossil fuel power unit schedules etc.). Within this framework, wind forecasting has been identified worldwide as an important part of the development of wind energy generated power. The R&D Wind Annex of the International Energy Agency held an Expert Meeting on Wind Forecasting in April 2000. The outcome of the meeting was to propose a new Annex for wind forecasting techniques to the IEA. The approval of the formal annex proposal is scheduled for a meeting in October 2000. Although weather prediction is an established research area, there is still a lot of development needed to establish an effective environment for wind energy forecasts. Especially observations are purely based on the requirements for traditional weather prediction. Our focus therefore is to establish a link to the traditional weather forecast community and the electricity suppliers. Three major parts have been identified. (1) Furthering the research in the numerical weather prediction on high spatial resolution towards the needs of wind energy predictions. (2) Establishing and testing an operational wind energy prediction system. (3) Extending the existing observation system for initialisation of the forecast models and planning purposes.

Acknowledgements The authors would like to thank Met Eireann for collaboration, support and providing the HIRLAM system and their Graphics Program. The HIRLAM System was developed by the HIRLAM Project

group, a cooperative Project of the national weather services in Denmark, Finland, Iceland, Ireland, the Netherlands, Norway, Spain and Sweden. Special thanks to Ray McRath and James Hamilton at Met Eireann and Henrik Madsen at IMM, Technical University of Denmark for their support and inspiring discussions. References [1] A Renewable Energy Development Strategy, Project report, XVII /4.1030 /Z/95-002,(1997). [2] Green paper on Sustainable energy. Sustainable Energy Division, Department of Public Enterprise, ISBN 0707-66263, Sept. (1999). [3] Strategy for Intensifying Wind Energy Deployment. Report of the Renewable Energy Strategy Group, Government of Ireland Publication, ISBN 0-7076-9225-3, July (2000). [4] Landberg, L., Joensen, A., Giebel, G., Madsen, H., Nielsen, T.S., Short Term prediction towards the 21st Century. In Wind Forecasting Techniques, 33 Meeting of Experts, Technical Report from the International Energy Agency, R&D Wind, Ed. S.-E. Thor, FFA, Sweden, 7785, July (2000). [5] Nielsen, T.S., using Meteorological Forecast in On-line Prediction of wind Power. Technical Report IMM-ELTRA-ELSAM, Sept. (1999). [6] Nielsen, T.S., Madsen, H., Tofting, T., WPPT, A Tool for On-Line Wind Power Prediction. In Wind Forecasting Techniques, 33 Meeting of Experts, Technical Report from the International Energy Agency, R&D Wind, Ed. S.-E. Thor, FFA, Sweden, 93-116, July (2000). [7] Barbour, P.L., Walker, S.N., Wind forecasting activities of the Oregon State University Wind Research Cooperative. In Wind Forecasting Techniques, 33 Meeting of Experts, Technical Report from the International Energy Agency, R&D Wind, Ed. S.-E. Thor, FFA, Sweden, 6371, July (2000).