Evaluation of fuel cost savings for optimal loading of the generating units of Bangladesh power system

Abstract

Bangladesh Power System (BPS) does not use optimization technique to schedule its generating units. This results in higher production cost. Moreover, the hydro potential of Kaptai is not utilized properly to generate electricity. This thesis evaluates the production cost of BPS considering appropriate generation scheduling and exhaustive use of hydro potential. This thesis uses dynamic programmll1g (DP) method for the generation scheduling. It neglects the transmission loss. A multi-state model of Kaptai hydro plant is developed using the historical data of water head of Kaptai reservoir and forced outage rate (FOR) of the units of the plant. The input output characteristics of the units of BPS are not readily available. These characteristics are developed by collecting data from different power stations. The startup and shutdown cost of the generating units are also collected. This thesis evaluates production cost of BPS for four cases. First case considers only the thermal power generating units optimally scheduled using DP and hydro units are scheduled as peak shaving unit. All the thermal generating units of BPS are optimally scheduled and the hydro unit is scheduled as energy limited (El) unit in the second case. In both cases, first and second, the duration of study is considered to be 2006. The production.cost savings is calculated by comparing the actual energy production cost of2006. In the third case the production cost of2006 is evaluated using segmentation method, which considers merit order loading based on average incremental cost of the units and neglecting startup and shutdown cost. The results are compared with those obtained by optimal scheduling of the units by DP method without considering startup and shutdown costs. The fourth case evaluates the energy generation cost of BPS for a period from 2007 to 2025 using DP optimal scheduling approach. The savings in production cost is evaluated by comparing with that obtained using merit order loading.