Exploring uncertainty in energy systems: Using decision trees to assess the future role of onshore and offshore electricity storage and hydrogen infrastructure

Abstract

In coastal regions, the expansion of large-scale offshore wind is central for decarbonization. However, with increasing shares of on- and offshore variable renewable energy sources, their successful integration becomes increasingly challenging. Electricity storage and hydrogen technologies are mature options supporting this integration. In this work, we aim to understand the role of these integration options for a wide range of techno-economic assumptions. Specifically, we optimize the selection and size of integration options for a stylized energy system topology with an onshore and offshore node for a set of 2000 parameter combinations. These parameters include (i) topology parameters, such as renewable energy capacity distribution between onshore and offshore, distance between onshore and offshore subsystems, capacity of a conventional power plant, and (ii) economic parameters, such as investment costs, hydrogen price and fossil fuel costs. We then apply interpretable machine learning methods (random forests, decision trees, and multivariate linear regression) to identify links between parameter values and cost-optimal integration choices. Results indicate that hydrogen production to sell to an external market offers an economic pathway to utilize otherwise curtailed renewable electricity. For reaching stringent emission constraints, electricity storage becomes crucial. In more than 80% of cases, the economically optimal placement of both electricity storage and hydrogen production is onshore. Using hydrogen as a storage to balance electricity supply and demand is only economical under a combination of extreme assumptions (very low cost, small available backup capacity, and limited renewable capacities). Furthermore, our findings indicate that achieving high emissions reduction targets can be less expensive with some fossil-based power plant.