Reducing energy system model distortions from unintended storage cycling through variable costs

Abstract

Energy system models are used for policy decisions and technology designs. If not carefully used, models give implausible outputs and mislead decision-making. One implausible effect is "unintended storage cycling", which is observable as simultaneous storage charging and discharging. Methods to remove such misleading effects exist, but are computationally inefficient and sometimes ineffective. Through 124 simulations, we find that determining appropriate levels of variable costs depends on the variable cost allocation to certain components and the solver accuracy used for the optimization. For the latter, if the accuracy is set too loosely, the solver prevents the removal of unintended storage cycling. We further provide a list of recommended variable cost model inputs as well as a minimum threshold that can significantly reduce the magnitude and likeliness of unintended storage cycling. Finally, our results suggest that our approach can remove other similar misleading effects such as unintended line cycling or sector cycling.