Enhancing power distribution system resilience against urban flash floods using hierarchical combination of Monte Carlo technique and reinforcement learning / Suhail Afzal

• Main Author: Suhail , Afzal
• Format: Thesis
• Published: 2024
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://studentsrepo.um.edu.my/15412/1/Suhail_Afzal.pdf; http://studentsrepo.um.edu.my/15412/2/Suhail_Afzal.pdf; http://studentsrepo.um.edu.my/15412/

Power system is recognized as one of the lifeline systems of a community that is appreciably reliable, but vulnerable to natural hazards such as hurricanes, winter storms, and floods. In recent decades, urban flash floods have become more common because of climate change and are worsening with the intensification of short-duration rainfall extremes. This catastrophic natural hazard presents a significant threat to a power distribution system, thus enhancing system resilience against flash floods is essential and imperative. To achieve this goal, existing literature offers a wealth of knowledge, however in these works, two-dimensional (2D) surface flow models are used to solve the hydraulics. These 2D hydrodynamic models consist of a set of nonlinear differential equations and require high-resolution topographic data of the floodplain. Though these models can provide descriptions of overland flow propagation, they fail to provide overflow locations which are crucial in flash flood modelling. Additionally, these 2D flood models are computationally expensive, hence cannot be run in real-time. Furthermore, the flooding fragility model offered by the Federal Emergency Management Agency (FEMA) of the United States is adapted to formulate failure scenarios, that is inappropriate for the Malaysian distribution network. Moreover, researchers have proposed various service restoration models and techniques prioritizing critical load advocating the resilient operational prowess of diverse sets of distributed generators (DGs). However, mostly dispatchable DGs are modelled, and the time-based model has not been extensively taken into consideration. In addition to this, varying load profile, temporal fault incursion and DG profiles are also not investigated. Therefore, this study presents a probabilistic flood model that is easy to develop and can handle heavy uncertainties related to urban flash flooding. In this respect, a linear regression model is proposed to estimate flood elevation in an urban floodplain and the Monte Carlo technique is employed to predict overflow locations in a grid-based environment. Considering rainfall intensity, soil moisture, and curvature of the surface, reinforcement learning is then leveraged to trace the flow path of floodwater from these overflow locations, to identify distribution substations at the risk of inundation. The proposed flood model is applied to IEEE 33-bus and a real 23-bus distribution systems considering a hypothetical terrain and validated on a real urban area. Further to this, site surveys and historical data are used to develop flooding fragility curves for indoor electrical substations to determine their probability of failure. Finally, the spatiotemporal impact of flash flooding on a modified IEEE 33-bus test system is captured using the proposed flood model. The evolving substation faults are then included in the proposed resilience-oriented time horizon-based service restoration model that also considers dynamic load demand, heavy uncertainties related to renewable generation, and dependencies of a distribution network in and outside the power system. This work will assist decision-makers and utility operators in enhancing power system resiliency against urban flash floods while overcoming the barriers of limited data and time.