Many countries worldwide are developing energy strategies to reduce reliance on fossil fuels and cut CO2 emissions to avoid the worst effects of climate change. Most models we use to support this decision-making process rely on very uncertain long-term forecasts for important parameters, such as fuel prices and energy demand. The dramatic unpredictability of energy markets urges us to factor this uncertainty into our energy models, which is seldom done in current modelling practice.

This talk discusses research that lies between energy systems modelling and mathematical optimization. After introducing what Energy System Optimization Models (ESOMs) are, we will quantify the uncertainty underpinning such models thanks to a backcasting analysis highlighting large errors in natural gas price forecasts over the last few decades. Then, we will show how we can consider uncertainty in energy models using robust optimization and apply it to the strategic planning of a national energy system modeled using EnergyScope, an open-source ESOM suitable for uncertainty applications. The results show that a robust investment strategy is up to 58% less likely to lead to overcapacity compared to the standard, deterministic decision-making approach which does not account for uncertainties. Overall, this proves that considering uncertainty in long-term energy planning models can reduce the risk of not meeting important climate targets.