Optimal design of multi-energy systems with seasonal storage

TLDR

Optimal design of multi‑energy systems with seasonal storage is hampered by the need for a year‑long, hour‑resolution horizon that creates a large, binary‑variable optimization problem. This study introduces mixed‑integer linear programming methods that enable year‑long, hour‑resolution modeling while substantially reducing problem complexity. The authors validate the approach on a simple system solvable without design days, then apply it to a Zurich neighborhood to optimize annual costs and CO₂ emissions, and finally conduct sensitivity analyses and Pareto‑set topology studies. The proposed methods produce results that closely match full‑scale optimization, correctly sizing storage and enabling long‑term operation while greatly simplifying the problem.

Abstract

Optimal design and operation of multi-energy systems involving seasonal energy storage are often hindered by the complexity of the optimization problem. Indeed, the description of seasonal cycles requires a year-long time horizon, while the system operation calls for hourly resolution; this turns into a large number of decision variables, including binary variables, when large systems are analyzed. This work presents novel mixed integer linear program methodologies that allow considering a year time horizon with hour resolution while significantly reducing the complexity of the optimization problem. First, the validity of the proposed techniques is tested by considering a simple system that can be solved in a reasonable computational time without resorting to design days. Findings show that the results of the proposed approaches are in good agreement with the full-scale optimization, thus allowing to correctly size the energy storage and to operate the system with a long-term policy, while significantly simplifying the optimization problem. Furthermore, the developed methodology is adopted to design a multi-energy system based on a neighborhood in Zurich, Switzerland, which is optimized in terms of total annual costs and carbon dioxide emissions. Finally the system behavior is revealed by performing a sensitivity analysis on different features of the energy system and by looking at the topology of the energy hub along the Pareto sets.