Transfer Function-Noise Modelling For Powerhouse Inflow Forecasting

Abstract

The problem of forecasting three-hour inflows and controlling a run-of-the-river hydroelectric plant is exanfiined with respect to remotely controlled powerhouses of a hydroelectric system operated by Alcan Smelters and Chemicals Ltd in the Province of Québec. Autoregressive-moving average (ARMA) models and transfer function-noise models are described, and recent developments with respect to the identification, parameter estimation, and diagnostic checking stages of model construction are reviewed for both types of models. The three stages of model construction are applied to sample data, and the most appropriate ARMA and transfer function-noise models are selected and discussed. The chosen transfer function-noise model relates flows at an upstream powerhouse and two tributaries to inflows at the downstream powerhouse. Flows from the tributaries are acquired via a GOES satellite. The selected model represents a very important component of the control strategy for the downstream powerhouse.

Résumé

Le problème de prévoir les apports à récurrence de trois heures et de contrôler une centrale hydro-électrique au fil de l’eau est examiné relativement à des centrales télécommandées d’un système hydro-électrique opéré par la Société d’électrolyse et de chimie Alcan Ltée dans la province de Québec. Les modèles de moyenne mobile autorégressive et de fonction de transfert avec bruit (“transfer function-noise models”) sont décrits et les développements récents relatifs aux étapes d’identification, estimation des paramètres et vérification dans la construction d’un modèle sont examinés pour les deux types de modèle. Les trois étapes de construction d’un modèle sont appliquées à un échantillon de données et les modèles les plus appropriés sont sélectionnés et discutés. Le modèle de fonction de transfert avec bruit choisi relie les débits d’une centrale en amont et ceux de deux tributaires aux apports à la centrale en aval. Les débits des tributaires sont acquis au moyen d’un satellite “GOES”. Le modèle sélectionné représente une composante très importante de la stratégie de contrôle de la centrale d’aval.

Additional information

Notes on contributors

Robert Thompstone

ROBERT M. THOMPSTONE was Senior Co-ordinator of the Hydraulic Resources Group, Power Operations – Quebec, Alcan Smelters and Chemicals Ltd, Jont quiere, Quebec, where he worked for the last nine years. Prior to that, he spent two years with a consulting firm specializing in water resources management. Recently he has been named Assistant Superintendent of Supply Services for Alcan Smelters and Chemicals Ltd installations in the Saguenay-Lac-Saint-Jean region of Quebec. Mr Thompstone holds a BASC in civil engineering and an MASC in management sciences from the University of Waterloo, and a BSPADM in business administration from the Universite du Quebec a Chicoutimi. He is currendy completing his PHD thesis in the Department of Systems Design Engineering, University of Waterloo.

Keith Hipel

KEITH W. HIPEL is an Associate Professor and Associate Chairman for Undergraduate Studies within the Department of Systems Design Engineering at the University of Waterloo. He obtained his Bachelors and PHD degrees in Civil Engineering; his Master's degree was in Systems Design Engineering at the University of Waterloo. In addition to being a member of the American Water Resources Association, the American Geophysical Union and the Association of Professional Engineers of Ontario, Dr Hipel is an Associate Editor of Water Resources Bulletin and is Chairman of the Surface Runoff Committee of the Hydrology Section in the American Ceophysical Union. Besides doing engineering consulting in operational research, Dr. Hipel has taught special courses and seminars in other countries, such as Brazil, China, Japan, Singapore, and the United States. His present research interests include the use and development of stochastic modelling techniques in the geophysical sciences and the incorporation of political and non-quantitative considerations into the systems design of large scale engineering projects. In addition to co-authoring two textbooks and also co-editing two books in stochastic modelling, Dr Hipel has produced research that has been published in various international engineering and operational research journals.

Ian Mcleod

A.I. MCLEOD is an Assistant Professor in the Department of Statistical and Actuarial Sciences at the University of Western Ontario. He holds a PHD in Statistics from the University of Waterloo for which he won the doctoral gold medal prize in 1978. His main research interests are in the theory and application of time series models, stochastic hydrology, and statistical algorithms.