Risk Assessment of Electric Power Generation Systems by Stochastic Simulation UsingCLOUD Computing

This chapter reviews fundamental ideas in electric power systems reliability evaluation with emphasis on stochastic and data-scientific inferential estimation methods more comprehensive than conventionally deterministic single-value estimates. This review focuses on a discrete event simulator (DES), CLOURAM: CLOUD Risk Assessor and Manager that quantitatively estimates the risk in Electric Power CLOUD computing framework. The 2-state (UP, DOWN) and 3-State (UP, DOWN, DER) units are assumed to fail and recover with the Negative Exponential and/or Weibull densities. Interdependence between units’ failure and repair rates can also be modeled with respect to MVE (Multivariate Exponential Density). Unit start-up failure probability and start-up delay hour are importantly incorporated. The algorithms are implemented onto by-the-author-authentically-collected national power systems data-banks to assess risk content by computing the unreliability indices of LOLP (Loss of Load Probability), MW-h; LOLE (Loss of Load Expected in Hours), and USE (Unserved Energy in MW-h). Power Systems Modeling simulations generally assume constant failure and repair rates of generators; however, these rates can be randomized to reflect the variability of the units’ forced outage rates. Hence, the LOLE pdf as an end-result compared to a single-value becomes a value-added target. Various tabulations will compare the two-state generator approach to that of the three-state generator with selected 50% derated scenarios, and additionally, compare the Negative Exponential assumption to that of the Weibull density with varying shape parameters to reflect energy market’s operational realities. The input data management and interpretation of software outcomes for different operational variations in the electric power generation systems will be interpreted.