(15e) A Two-Stage Stochastic Optimization Approach for the Design of Sustainable Ammonia Supply Chain Networks

Recently, renewable (“green”) ammonia has been proposed as an alternative to conventional fossil-based (“gray”) ammonia [2,3]. In this approach, renewable resources such as solar and wind are used to produce feedstock hydrogen via electrolysis and nitrogen via air separation as well as powering the ammonia synthesis loop itself. This ammonia production can be deployed locally to demand in areas of high renewable potential to reduce the carbon intensity of both production and distribution. Using renewable energy to make ammonia can also offer ammonia price stability because the cost of the renewable energy feedstock will be stable year-over-year. The renewable ammonia concept has considerable merit but nonetheless, transitioning existing ammonia supply chain networks to this more sustainable alternative will require multiple years. Recently, we have proposed a multiperiod deterministic (i.e., capacity expansion) model that optimizes such a transition by minimizing installation, operation, and distribution costs for new renewable ammonia production to augment conventional ammonia supply [4].

In this work, we consider the effect of uncertainty on the optimal design and transition of existing ammonia supply chain networks. The economics and design of such supply chain networks is affected by multiple sources of uncertainty, such as ammonia demand and conventional ammonia price. Ammonia demand can be estimated from crop acreage, whereas the price is more volatile as it evolves on the global market. We propose a two-stage stochastic programming approach where the first-stage decisions are renewable ammonia production investment decisions (e.g., installation year, location, and capacity) and the second-stage decisions are the flows of ammonia in the network from both conventional and renewable sources for different scenarios of conventional ammonia price. We use historical prices of ammonia and generate ten scenarios using the distribution matching approach [5]. The resulting problem is a Mixed Integer Linear Programming (MILP) problem.

We consider a case study on Minnesota’s ammonia supply chain network using a planning horizon from 2024 to 2032 with annual decision resolution. In a purely economically optimal supply chain transition, we find that accounting for uncertainty results in the installation of more renewable ammonia capacity and thus a reduction in the amount of ammonia purchased from conventional producers. We also consider the case where demand is met only from green ammonia by the end of the planning horizon. In this case, renewable ammonia production investments are made earlier in the planning horizon, compared to the deterministic case. Simulation of the supply network obtained from the stochastic and deterministic models shows that the stochastic design leads to lower net present cost. Overall, these results show that renewable in-state production of ammonia can act as a hedge against the volatility of conventional ammonia markets, leading to lower ammonia supply costs to the benefit of local consumers.

References:

[1]. Rouwenhorst, K.H., Travis, A.S. and Lefferts, L., 2022. 1921–2021: A Century of renewable ammonia synthesis. Sustainable Chemistry, 3(2), pp.149-171.

[2]. Wiskich, A. and Rapson, T., 2023. Economics of Emerging Ammonia Fertilizer Production Methods–a Role for On‐Farm Synthesis?. ChemSusChem, 16(22), p.e202300565.

[3]. Palys, M.J. and Daoutidis, P., 2022. Power-to-X: A review and perspective. Computers & Chemical Engineering, 165, p.107948.

[4]. Palys, M.J. and Daoutidis, P., 2023. Optimizing renewable ammonia production for a sustainable fertilizer supply chain transition. ChemSusChem, 16(22), p.e202300563.

[5]. Calfa, B.A., Agarwal, A., Grossmann, I.E. and Wassick, J.M., 2014. Data-driven multi-stage scenario tree generation via statistical property and distribution matching. Computers & Chemical Engineering, 68, pp.7-23.