Minimal Thermal Generation in Power Systems - Inferring Private Cost Parameters from Observed Firm Behavior

Most production processes can be turned off easily during times of low demand; this is not true for the generation of electricity. For a number of technical reasons, a certain number of conventional (thermal) power stations needs to keep producing electricity at all times, despite negative margins. Empirically, this is evidenced by the fact that even during times of low or negative prices, some power stations keep generating electricity. This paper estimates empirically the quantity of such minimal or “must run” generation levels in a large integrated power system, using observed price and production data. The level of must-run generation is not only an interesting aspect of power systems in itself, it is also an important parameter for power market and energy system modelling. It is particularly important in times of low demand for electricity from conventional plants, such as moments of high wind or solar in-feed, hence this parameter is an important input for high-renewables scenarios. Using German data, we find aggregated must-run generation to be on average 34 GW, about 60% of average thermal generation. However, the variation in this parameter is significant, with observed must-run levels between 20 GW and 49 GW. Regression analysis suggests that the must-run constraint is 2-3 GW higher in winter than in summer and 5-7 GW higher during the day than at night. While we expected technological learning and increased awareness among market participants to reduce minimal generation over time, the data shows an increase since 2012