Paper detail

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

Energy model distortions may unknowingly appear and represent unrealistic and non-physical effects that can mislead optimal model decisions. A prominent misleading effect is unintended storage cycling that was observed in previous literature by simultaneous charging and discharging of the same storage in 12 of 18 energy models. Especially for popular net-zero energy model scenarios, unintended storage cycling can cause high distortions and, hence, urges for its removal. Methods to remove such misleading effects exist, but are not computational efficient and sometimes not effective such as MILP formulations. Other techniques are successful, but only if renewable energy target constraints exist. This paper explores how to eradicate unintended storage cycling for models without renewable target constraints by correctly setting variable costs of relevant system components. We find through 124 simulations that determining appropriate levels of variable costs depends on the solver accuracy used for the optimization. If set too loose, the solver prevents the removal of unintended storage cycling. We further find that reliable data for variable costs in energy modelling needs to be improved and provide a list of recommended model inputs as well as a minimum variable cost threshold that can significantly reduce the magnitude and likeliness of unintended storage cycling. Finally, our results suggest that variable cost additives may remove other known unintended energy cycling effects, such as unintended line cycling or sector cycling.