Dynamic Load Modelling for Close to Real Time Demand Side Management

Voltage unbalance negatively affects both distribution network operators (DNOs) and customers by introducing overheating, accelerated thermal ageing of equipments, reduced efficiency and therefore additional financial cost. Once a new source of unbalance appears in the network, the overall degree of unbalance of the network will be increased, contributing to the system losses additionally. The research develops a methodology for probabilistically estimation of unbalance in networks and exploration of the pattern how the effects of multiple sources of unbalance superpose. Simulated with a real 24-bus network using Matlab, by involving normally distributed load and daily loading curve, which stand for the time-varying asymmetrical loading during a day, unbalance-related parameters are computed and analyzed to discover the pattern for superposition. The validated pattern enables DNOs to closely derive the level of unbalance in the network from either historical data or the probabilistic estimation even when full monitoring of the network may not be available. 40 Dynamic Load Modelling for Close to Real Time Demand Side Management Yizheng Xu, J. V. Milanović School of Electrical & Electronic Engineering, The University of Manchester Manchester, M13 9PL, United Kingdom [email protected], [email protected] Abstract – In order to illustrate daily variation of dynamic characteristics of the load for different types of load, time-varying aggregate load responses are derived and presented in this poster. Different from the modelling approach in previous literatures, where either measurement-based approach or componentbased approach is adopted, this project combines both modelling approaches to compensate the weakness brought by one from the other. Hourly participation of different load categories and dynamic responses of individual load categories are compulsory information for the derivation of time-varying aggregate load responses. The load participations of individual load types in aggregate load are represented by decomposed daily load curves, and they are derived from surveys or directly measured daily load curves of different load types conducted in the past work. Dynamic responses of individual load categories are captured from the simulation or field measurements and emulated with suitable mathematical formula via measurement-based approach. The aggregate dynamic responses are derived by scaling the individual load dynamic responses with respective load contributions and finally summing all the scaled individual load dynamic responses up, resulting in close to real time power and reactive power demand dynamic response curves. All hourly curves are finally presented in daily demand response surfaces which simultaneously provide the information of dynamic responses and hourly demand, shown in Fig. 1. In order to illustrate daily variation of dynamic characteristics of the load for different types of load, time-varying aggregate load responses are derived and presented in this poster. Different from the modelling approach in previous literatures, where either measurement-based approach or componentbased approach is adopted, this project combines both modelling approaches to compensate the weakness brought by one from the other. Hourly participation of different load categories and dynamic responses of individual load categories are compulsory information for the derivation of time-varying aggregate load responses. The load participations of individual load types in aggregate load are represented by decomposed daily load curves, and they are derived from surveys or directly measured daily load curves of different load types conducted in the past work. Dynamic responses of individual load categories are captured from the simulation or field measurements and emulated with suitable mathematical formula via measurement-based approach. The aggregate dynamic responses are derived by scaling the individual load dynamic responses with respective load contributions and finally summing all the scaled individual load dynamic responses up, resulting in close to real time power and reactive power demand dynamic response curves. All hourly curves are finally presented in daily demand response surfaces which simultaneously provide the information of dynamic responses and hourly demand, shown in Fig. 1. Figure 1 Daily Demand Response Surface for Power 0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24 40 60 80 100 Hour time (sec) P ow er , P (% )