Multi-area power system reliability modeling

Typescript (photocopy).The dissertation presents a transmission network model suitable for use in multi-area reliability models which utilize a Monte Carlo simulation approach. The transmission network model utilizes the REI equivalent which permits accurate modeling of transmission ties between areas while reducing computational burden and permitting variations in area generating capacity and loads as required in reliability studies. The study also presents a computationally efficient approach called heuristic redispatch to generation redispatch as required to model mutual assistance between areas in multi-area reliability studies. The variance reduction techniques are applied to reduce the size of samples in Monte Carlo simulation by reducing the variance of reliability indices which we want to obtain. A 3-area power system which is an arbitrary decom position of the IEEE 116 bus system and a large power pool which has 13 areas were modeled using the REI equivalent method. The REI equivalent transmission model is shown to give reliability index results very close to those obtained using a fully detailed transmission network modeled. This confirms the suitability of the REI equivalent model for multi-area reliability studies. Total computer time using the REI equivalent was only 47 per cent of that required using the detailed transmission model for the 3-area sample system. The computer time reduction with the REI equivalent can be expected to be much larger for large systems with more areas and transmission lines. It is shown that the heuristic redispatch approach greatly reduces the number of times linear programming solutions are required and consequently the computer time required. The more detailed heuristic algorithm II is seen to give the best results for both the small and large sample systems. Hence, the greater computational burden of algorithm II seems to be justified by its superior performance. Sample system studies show computer time reductions of 56 per cent for a small 3-area system and 97 per cent for a large 13-area system. Several variance reduction techniques are shown to be effective in reducing the number of replications and computer time required to obtain statistical convergence of reliability indices. In particular a joint use of variance reduction technique, importance & monthly stratified sampling, gives more than 90 per cent reduction of computing time and number of replications for a 3-area power system