A multi-service approach for planning the optimal mix of energy storage technologies in a fully-renewable power supply

Energy storage systems (ESS) are a structural solution for the integration of renewable energy systems. To plan the optimal combination of ESS, storage expansion planning approaches are commonly used. They tend to focus on balancing the energy fluctuations from renewable technologies but are usually blind to the need for specific additional services required for dealing with forecast errors. Hence, they underestimate the real operating costs of the future power system and lead to suboptimal investment recommendations. In response, we propose a multi-service storage expansion approach. A linear programming optimization is developed, LEELO, to find the optimal investments in a 100% renewable system (based on solar photovoltaic and wind power) deciding on renewable generators and storage systems. In our formulation, we explicitly model the provisioning of power reserves and energy autonomy as additional services. A case study applies our model to Chile considering four regions and the (existing) hydropower park, for a complete year with an hourly resolution. We systematically assess how our novel multi-service planning differs from conventional energy-based planning in terms of total costs, operation, and investment decisions (with a focus on ESS). Considering power reserves and energy autonomy reveals on average 20% higher costs that otherwise would not be captured in the expansion planning process. Regarding operation, ESS show only slight differences in the two planning models. All ESS participate in the provision of energy. As might be expected, batteries are the main provider of (short-term) power reserves, assisted by pumped-hydro, whereas hydrogen storage is responsible for providing (long-term) energy autonomy. However, the storage investment decisions differ significantly between both models. In our multi-service model, the attained power capacities and energy capacities are up to 1.6 and 3.2 times larger, respectively than in conventional planning. The resulting storage mix changes even more strongly: a general shift towards hydrogen systems is observed. Mainly batteries are substituted, while pumped-hydro capacities stay relatively constant. The trend of the above results is consistent for various scenarios of wind and photovoltaic generation and for sensitivities of service parameters. Our findings underline the importance of modeling multi-services in the planning of renewable-based power systems