A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Environment, Resources & Development

Thesis (M.Eng.) - Asian Institute of Technology, 1984

The Testing of a power system tor each decision alternative is a computationally burdensome step. Change over from the currently established deterministic criteria to stochastic means would be advantageous . In this study, a short term load forecast for the Sri Lanka power system is reasonable estimated, considering its past trends of growth, using the method of least squares. The degree of uncertainty is quantified by the forecast variance. The propagation of the uncertainty of input variables to the output variables of the power system is determined, by the solution of the conventional load flow, followed by a series of noniterative steps. This was done by least squares estimation proposed by Do pazo et al. The final outcome of the study, is the predetermination of the expected values of the system responses and its possible deviations, due to the forecast inaccuracies of loads and generator outages, at their respective nodes. An algorithm and a programme have been developed and the applicability of the method in short term transmission planning and design is demonstrated, by using the Sri Lanka power network around the year 1990, as an example. Results are obtained and presented in a compact form in terms of the expected values and confidence intervals. The system security is also considered and certain suggestions are made with regard to the protection of the system.