Chowdhury, N. A.2018-12-142018-12-141994-09Septemberhttp://hdl.handle.net/10388/11661The research work reported in this thesis deals with the application of probabilistic methods to adequacy assessment at hierarchical level I (HLI) with due considerations given to operating criteria. In a conventional adequacy assessment at HLI, generating unit models are lumped regardless of their need or order to form an equivalent capacity model. The capacity model is examined to determine its adequacy to meet the total system load requirement. In this technique, generating units are assumed to be in their operating states unless on forced outage states. In practice, a generating unit is taken in service, if it is needed. An attempt has been made in this thesis to illustrate the impact of operating considerations in adequacy assessment at HLI. Existing analytical methods and mathematical models are employed in this research work and they have been enhanced and improved wherever required to do so. A four-state model of a generating unit can include the operating considerations such as loading order, duty cycle and start-up failures. A technique to modify four-state models to reflect the operating factors has been developed and illustrated in this thesis. A technique has been developed to estimate adequacy indices with the help of multi-period indices. In practice, all generating units are not utilized during an entire period and, therefore, a capacity model cannot be valid for an entire period. A study period is divided into multiple short periods. Each short period is then utilized to include the operating factors of a unit. Several load models have been utilized to determine the impacts of load modeling in adequacy assessments. A new probabilistic margin approach has been developed to assess risk and required margin in a system. The probabilistic margin approach evaluates static reserve capacity and margin for a given risk criterion. Four-state models can be modified in two ways to represent units with energy limitations. One method is developed based upon a peak shaving approach. The second method proposes a four-state model where the generating unit data are modified by the expected allotted energy of the unit. Two representative test systems are used throughout the thesis to provide numerical examples.Thesis