(11a) Design Under Uncertainty for Microgrids with Ammonia-Based Energy Storage

A microgrid provides a decentralized and flexible platform for integrating renewable energy sources (RESs) such as wind and solar power [1]. In microgrids, energy storage is crucial for handling the intermittency of RESs and the often limited access to the power grid. Batteries are the most common energy storage option, but they are not yet cost-competitive for long-duration and large-scale storage. Similarly, hydrogen produced via water electrolysis can be used to store renewable electricity but is very expensive due to its high storage cost. This has motivated the use of liquid e-fuels, which have much higher energy densities than hydrogen. In this context, ammonia has received significant attention in recent years as it has the key advantage of being completely carbon-free if produced using renewable electricity [2], and previous studies have demonstrated the potential economic benefits from using green ammonia as an energy storage medium in microgrids [3,4].

One major challenge in the design of renewables-based microgrids is the highly time-sensitive availability of RESs, which has a mixed pattern of short-term and long-term variations and is notoriously difficult to predict. Specifically, design decisions need to account for system lifetimes of multiple years, across which seasonal trends can change, while also considering how day-to-day operational decisions are made in light of short-term uncertainty. This has motivated many combined design and scheduling approaches [5,6], which consider seasonal changes, often by using representative days for the different seasons. On the other hand, short-term uncertainties are commonly addressed using feedback and frequent re-optimization of the current schedule [7]; however, such an online scheduling framework cannot be directly incorporated into a design optimization model.

In this work, we apply a simulation-optimization (SO) approach to optimize the design of ammonia-based microgrids under both short- and long-term uncertainty in RES availability, where the closed-loop performance of online scheduling is evaluated via a rolling-horizon simulation. In particular, we use Bayesian optimization (BO) [8] to guide the search in the design space, and rolling-horizon scheduling simulations over a one-year time horizon with an hourly resolution are performed on multiple scenarios that capture the uncertainty, which allows the evaluation of the expected operating cost. The use of BO is motivated by the fact that the number of design variables is relatively small (on the order of ten) while the rolling-horizon simulation is computationally very expensive (each run requiring on average 10 minutes). Indeed, BO has proven to be effective in finding very good solutions with relatively few iterations. In our case studies, we investigate the impact of the connectivity to the main grid on the design of the microgrid, showing that the islanded operating mode necessitates larger storage and production units for hydrogen and ammonia. The BO approach results in designs with total costs that are 8–14% lower than what can be obtained by random search methods. Further, a sensitivity analysis on the operating range of the Haber-Bosch process demonstrates that extending it from 90–100% to 10–100% yields an 18% cost reduction, underscoring the importance of flexible ammonia production.

[1] Faisal, Mohammad, et al. "Review of energy storage system technologies in microgrid applications: Issues and challenges." IEEE Access 6 (2018): 35143-35164.

[2] Salmon, Nicholas, and René Bañares-Alcántara. "Green ammonia as a spatial energy vector: a review." Sustainable Energy & Fuels 5.11 (2021): 2814-2839.

[3] Edmonds, Lawryn, et al. "Green ammonia production-enabled demand flexibility in agricultural community microgrids with distributed renewables." Sustainable Energy, Grids and Networks 31 (2022): 100736.

[4] Palys, Matthew J., et al. "A novel system for ammonia-based sustainable energy and agriculture: Concept and design optimization." Chemical Engineering and Processing-Process Intensification 140 (2019): 11-21.

[5] Palys, Matthew J., and Prodromos Daoutidis. "Using hydrogen and ammonia for renewable energy storage: A geographically comprehensive techno-economic study." Computers & Chemical Engineering 136 (2020): 106785.

[6] Guo, Li, et al. “A two-stage optimal planning and design method for combined cooling, heat and power microgrid system.” Energy Conversion and Management 74 (2013): 433-445.

[7] Baker, Kyri, Gabriela Hug, and Xin Li. "Energy storage sizing taking into account forecast uncertainties and receding horizon operation." IEEE Transactions on Sustainable Energy 8.1 (2016): 331-340.

[8] Paulson, Joel A., and Calvin Tsay. "Bayesian optimization as a flexible and efficient design framework for sustainable process systems." Current Opinion in Green and Sustainable Chemistry (2024): 100983.