On stochastic unit commitment for thermal power plants

Abstract: Climate change is a fact, a crisis threatening every country, economy and human. Toprevent this crisis, the emission of greenhouse gases needs to decrease dramatically. 72%of global greenhouse gas emissions in 2016 came from energy production where electricityand heat account for 42% of the 72%. Nevertheless, coal power grew with 28% in2018 to meet the increased demand of electricity. It is therefore of utmost importancethat the resources used in power plants are distributed as efficiently as possible. Unitcommitment is a short-term planning formulation which is part of the planning chain forproduction of electrical energy. An accurate unit commitment can decrease emissionsand costs.The aim of this study is to implement a model for the stochastic behavior of the electricalload into unit commitment. With this, it shall be evaluated, whether this solutionis robust enough for usage in network control. The evaluation needs to assess the reliability,economic impact and the computational e↵ort for solving the stochastic unitcommitment problem.A test system has been created in MATLAB to evaluate the stochastic versus deterministicunit commitment formulation. Scenarios for the stochastic unit commitmenthave been generated by using a stationary, discrete-time Markov Chain to generate loadforecast errors. The Fast Forward Selection method has been used to reduce number ofscenarios to minimize computational e↵ort. The quality of the solution has then beenevaluated with value of the stochastic solution for economic analysis. Loss of load probabilityand energy not served have been used to evaluate the reliability.A stochastic approach gives a more robust solution but can be more expensive in termsof costs. Five scenarios were the optimal choice for the stochastic unit commitmentformulation. Increasing number of scenarios did not improve the reliability and resultedin a more expensive solution. The conclusion of this work can be contradictory but highlightsone of the challenges in electric power systems. A more robust system is usuallymore costly and therefore the players in the system must decide what is most desirablein this particular system. A more reliable but expensive system or a less reliable andless costly system.