02/04/2025 - 15:00 - CORE C.035
 

Diederik Coppiters

(UCLouvain - Institute of Mechanics, Materials and Civil Engineering) 

Will give a presentation on:

When the optimum is not enough: designing energy systems for an unpredictable world

Abstract: 

Computer models have transformed how we analyze and optimize energy systems. However, the designs they produce are only as reliable as the assumptions they make about future operating conditions—conditions that are dynamic, complex, and deeply uncertain.
This talk examines how to place uncertainty at the center of energy system design. I first present methods to characterize uncertainty under data scarcity and to propagate it through computationally demanding models using surrogate-based approaches. I then show how these methods can be used to study the sensitivity of optimal designs in linear programming models, and how they can be embedded in metaheuristic optimization frameworks for nonlinear system models to target expected performance, robustness, and, for the first time, antifragility—systems that benefit from variability rather than merely withstand it.
Finally, I address structural uncertainty: factors such as stakeholder preferences, implementation constraints, and social acceptance that are difficult to formalize, but often decisive in practice. Rather than focusing on a single optimum, I show how Modeling to Generate Alternatives can identify structurally different near-optimal designs, and how interpretable machine learning can turn the large solution sets into actionable design rules and policy-relevant archetypes.
The methods are illustrated with applications ranging from detailed process engineering models to national-scale energy system analyses. Taken together, they shift the focus from optimizing for a single predicted scenario to designing systems that remain effective in an unpredictable future.

More info about the speaker