Application of Monte Carlo Simulation for risk-based decision making in MV distribution networks

Due to the liberalization of the electricity sector and the introduction of regulatory agencies to ensure service reliability, while at the same time controlling costs, distribution system operators (DSOs) are increasingly focusing on the concept of asset management. Asset management involves decision-making for investments where a balance between cost, performance and risk of the system needs to be found. Decisions within asset management are often risk-driven. Risk assessment methods within DSOs are relatively new and the available methods have, up to date, not yet been extensively applied for electric distribution systems. Nonetheless, there is a worldwide increase in interest to do so, which is supported by the specification PAS-55 for a number of years and by the in 2014 released ISO55000 series for asset management. The challenges for DSOs with risk-based methods lies with the size of the distribution system consisting of a wide variety of large numbers of assets, the availability of data, the probabilistic behavior of the power system, dynamic system configuration, ageing of components and environmental influences. The main purpose of this thesis is to study the applicability of the Monte Carlo Simulation (MCS) method, utilizing the currently available failure data recorded at DSOs, and to investigate whether the simulation results can provide guidance in typical risk-based asset management decision-making processes. The MCS method is modelled for a practical application in the distribution network of Stedin. In this thesis the focus is on forecasting costumer interruptions through the MCS method. The study cases consist of 12 randomly selected 10 kV radial feeder systems from a distribution network of Stedin. The significant components contributing to the occurrence of customer outages are investigated by statistical analysis. According to the historic failure records cables and cable joints are found to contribute to majority of MV grid related outages and therefore their behavior is investigated. The strategy applied to the study cases is to initiate a MCS at an earlier point in time (at an initial starting year) and simulate for a mission time up to present time. This strategy is chosen in order to make it possible and realistic to compare the simulated customer outages with the actual occurred amount of customer outages for validation of the model. Moreover, for application of the MCS, the necessary input data of the components in the system are their times-to-failure (TTF) distributions. According to the available information, the TTF distribution of paper-insulated lead-covered cables (PILC) and cross-linked polyethylene cables (XLPE) are assumed to follow an exponential distribution. The failure distributions of cable joints are assumed based on results obtained from a previous research. The failure probability distributions as function of age are found to be, a normal distribution for oil-filled joints and Weibull distributions for resin-filled and synthetic joints. The TTF distributions modeled through the application of truncated distributions at their ages. The MCS algorithm for the study cases in this research is modelled in Matlab. From the validation of the results of the 12 historic radial feeder systems, by comparison of the actual occurred number of outages, it is concluded that 100% of the actual occurred outages were predicted through the simulation, i.e. the actual occurred outages of all the systems were possible outcomes of the simulation. Further, 33% occurred with a probability of less or equal than 0.1 and thus represent extreme events. Through brief investigation of possible causes of extreme events it can be argued that the cable length, number of joints in the system and the soil type contribute to the occurrence of these extreme events predicted though the MCS model. The predictive risk assessment method presented in this thesis can provide the asset manager with inputs to assist in the decision-making process. The conceptual framework for the utilization of the results is presented and a decision tool for the execution of risk-reducing investment plans is provided. This concept is indicated in the flow chart below.Electrical Engineering, Mathematics and Computer Scienc