Empirical evidence for slack in Modeling to Generate Alternatives

Modeling to Generate Alternatives (MGA) has been increasingly used with energy system optimization models to tackle uncertainty, stability of results, non-modeled objectives, and other complexities of the real-world energy transition. However, MGA results are highly contingent on the subjectively assumed value of slack, which defines how much higher the total system costs can be with respect to the costs of the cost-optimal scenario. Here, electricity system hindcasting in 31 European countries is used to quantify deviations between costs of the cost-optimal scenario and the real-world transition represented in an optimization model to provide first-of-the-kind empirical evidence for slack. For a multi-year model, the results indicate that the median value of cost deviations across the countries and across various parametric cost assumptions is 44% if modeling 30 years ahead and if accounting for costs during the modeled period only. A single-year model for 25 years ahead had a median cost deviation of 59%. As most current literature uses slack values that are lower or equal to 10%, these results underscore the importance of using larger and country-specific slack, and at the same time open more fundamental questions about the use of cost-optimization models.