Reliability modelling of power system components using the CEA-ERIS data base
Debnath, Kamal Krishna
The Canadian Electrical Association (CEA) in its Equipment Reliability Information System (ERIS) collects power system component data through the member utilities. The objective of the research project reported in this thesis is to explore the CEA-ERIS data base and determine the extent to which these data can be utilised to support the existing reliability techniques and to develop new models for quantitative power system reliability evaluation. Parameters for conventional generating unit models were computed using the CEA-ERIS generation data. A basic eleven-state model was first developed. The continuous state monitoring approach employed by CEA in the generating unit data collection procedure allows the computation of both state residence times and transition rates. Parameters for a basic two-state model for base load units or a four-state model for peak load units can be obtained from the eleven-state model. Important parameters for these two models are forced outage rate (FOR) and utilisation forced outage probability (UFOP). Residence times associated with derated states can be adjusted to yield "equivalent" forced outage times. If such an adjustment is made, the basic unit parameters become the derating-adjusted FOR (DAFOR) and the derating-adjusted UFOP (DAUFOP). Analysis of the ERIS data has indicated that some fundamental changes are required in the basic models. Modification to the four-state peak load unit model resulted in the modified UFOP (MUFOP) and modified DAUFOP (MDAUFOP) statistics. These parameters have been calculated and are reported in this thesis. A multi-state generating unit model is a more valid representation than a derating-adjusted two-state model. A nine derated-state model is, therefore, presented and the associated parameters calculated using the CEA-ERIS data base. This model can be reduced to a three-state model containing a single "equivalent" derated state. A generalised model containing fifty one states has been developed to store the generating unit data in a condensed form. This model can be used to obtain parameters for other models. Analysis of state residence times revealed that the operating times follow hyper-exponential distributions whereas the repair times follow lognormal distributions. The basic model applicable to all transmission equipment is the two-state model. Failure frequency, repair and switching times and unavailability have been computed for all the major transmission equipment. The total failure rates are divided into different failure mode categories. Active and passive failure rates can be separately computed from this information. Additional failure rates associated with a circuit breaker have also been calculated to permit detailed modelling. Common-mode failure rates for transmission lines were calculated and those involving lines on common towers separated. Possible application of the generation and transmission equipment models in the quantitative power system reliability evaluation is briefly discussed. Modifications in the data collection procedures are suggested in situations where certain required information can not be obtained from the CEA-ERIS data base.