Development of a methodology for voltage sensitive generation scheduling

Abstract

This thesis addresses scheduling conventional generation units keeping in view the low voltage problem. In a power system low voltage occurs when requisite reactive power support is not available. The problem will aggravate further in emerging power systems which will have electricity market oriented operation and large scale penetration of intermittent renewable based generation resources. Market operation makes the reactive power supports often scarce. Renewable sources involve DC-AC or AC-DC-AC conversion requiring adequate reactive power support. Besides this, renewable sources themselves may be unavailable totally in non-congenial weather. Voltage management through transformer tap changing, line switching, capacitor switching and lastly load shedding has long been practiced in power industry. But these may end up in voltage collapse or voltage instability due to hitting the ceiling (limit) of these devices/actions. So bus voltage in problematic areas may be improved if the generators in their neighborhoods can be scheduled. This is because reactive power is less mobile (i.e. cannot flow over a long distance unlike real power), and generators are active devices whose excitation can be controlled to supply or absorb reactive power. It appears that all the existing scheduling methods are sensitive to fuel-cost characteristics and select cheaper units considering only real power demand but do not select the units considering the impact on bus voltages. After scheduling the units the load is allocated among them by an optimal power flow (OPF) which takes care of operating and network constraints. But low voltage problem cannot be solved this way if the units near the problematic areas (even though their fuel cost may be relatively higher) are not selected at the scheduling stage before the OPF. In this research a new methodology for scheduling has been developed for considering bus voltages. This ensures that the generation units which are able to improve the system voltage profile will be selected. For this a new optimization function is formed augmenting the classical Lagrange function for the total time period (i.e. 24 hours) by a bus voltage sensitive function. Notably the classical Lagrange function for selection of units comprises only the fuel cost and load balancing constraint. In the voltage sensitive function, each bus voltage magnitude excepting that for the slack bus in each hour has been modeled in terms of (i) the elements of a network matrix, (ii) a negative injected current contributed by the shunt admittance including line charging susceptance and load (real and reactive) at each bus, and (iii) both real and reactive power outputs of the generation units to be evaluated for scheduling. The network matrix is similar to bus impedance matrix but it is formed excluding the slack bus and all shunt admittances and line charging susceptances. The real power (I2R) loss in transmission lines has also been embedded in the new Lagrange function as a fraction of the real power outputs of the units being evaluated for selection. The proposed scheduling method has been extensively tested on the IEEE 30 bus standard system under diverse forecasted loading scenarios for a period of 24 hours and various sets of cost parameters for the generation units. The results are compared in each case by running OPF separately using the units scheduled by respectively (i) the classical Lagrange method and (ii) the proposed method. In general compared to the classical method the proposed scheduling method has been able to so select the generation units that led to better OPF results i.e. better overall performance in terms of convergence, transmission loss and number of buses with voltage magnitudes significantly above the minimum value, at a marginal cost addition. The potentials of the proposed method (i.e. the voltage sensitive generation scheduling) in a real life system have also been demonstrated by applying and comparing it against the classical Lagrange method for 93-generator and 166-bus Bangladesh Power System (BPS). Suggestions have also been made on how the proposed method can be used for accommodating spinning reserve at the scheduling stage. The developed method is expected to be useful also for the ISOs (Independent System Operators) in deciding transmission network access to the generation units which have submitted on-line bids for a time period comprising a number of hours together.