The creation of the National Electricity Market in December 1998 represented a milestone in the Australian power energy industry restructure. The present work deals with a series of challenging issues which have appeared in recent years as a result of the deregulation process in this industry. Short-term demand forecast, multi-area unit commitment and unit dispatch are some of the most important issues which have changed dramatically from a monopolist to a competitive structure. In this thesis, a series of models are developed to address these issues.
The first model was developed to forecast the load demand, seven-day ahead, with high accuracy and low implementation cost. It is based on a non-linear regression method and has been applied to Victoria. Then, two different models, namely a sequential algorithm and a genetic algorithm have been developed to solve the multi-area unit commitment problem based on pool price and respectively bid
price. The most important features of the sequential multi-area unit commitment model are the inclusion of constraints and factors largely neglected in numerous studies, for example, inter-regional transmission limits, transmission costs, inter-regional transmission losses, auxiliary factor and spinning reserve. The model is developed solely using the sequential method and employs a bidding procedure to sequentially identify the next most economic unit to be committed. The sequential model is applied to solve the optimal operation of a large scale interconnected power systems with different number of regions in the Australian National Electricity Market and the optimal or near optimal solutions are obtained within a very low computation time, practically in 'real time'.
A range of sensitivity analyses is developed for the sequential multi-area unit commitment model applied to the National Electricity Market (four interconnected regions). Several different scenarios are
evaluated to find the impact on pool price and implicitly on the scheduling of units using a variety of levels of inter-regional transmission limits, a different bidding strategy and a forced unit outage.
The second model is developed solely using a genetic algorithm technique and is applied to solve an interconnected power system with two regions. The results between the two models are compared.