(357a) Leveraging Hourly Variations in Electricity Pricing for Optimal Retrofit Electrification of an Oil Refinery

The chemical industry relies heavily on fossil fuels, presenting a significant challenge in meeting decarbonization goals. Despite the availability of technological solutions, the industry often hesitates to transition due to the value tied to existing infrastructure. To address this, we developed a comprehensive decision-making framework tailored for oil refineries. Our framework prioritizes the decarbonization of process heating and hydrogen production, leveraging advanced computational methods to identify the most cost-effective retrofitting plans.

Our preliminary research revealed that in the short to medium term, carbon capture technologies are preferred over electrification-based alternatives, unless there is a substantial decrease in electricity costs. Additionally, while the introduction of CO₂ taxes does incentivize the adoption of carbon capture technologies, it does not necessarily accelerate the adoption of electrification.

In the current work, in order to further enhance the viability of electrification solutions, and motivated by the general area of demand side management [1], we have investigated the effect of variability in electricity prices on an hourly basis, alongside exploring different storage options. Our findings indicate the potential for earlier adoption of electrified boilers and electrolytic hydrogen production when leveraging variable electricity prices.

In our results, we found hybrid technologies to be promising, such as combining electrified boilers with natural gas boilers, to optimize the cost of operation amidst hourly varying pricing scenarios. Additionally, leveraging storage options for hydrogen allows for increased flexibility, enabling higher production during periods of lower electricity prices.

To address the complexity of hourly electricity price variations and storage decisions, we adopted a representative days/weeks approach. By carefully selecting the number of representative time frames, we ensure that our model captures the full range of potential scenarios while maintaining computational tractability.

Moreover, tailored Benders decomposition is used to solve the optimization problem. By decomposing the problem into smaller, yearly subproblems, we enhance computational efficiency by parallelization without sacrificing accuracy. We add cuts to ensure that the subproblems are consistent with each other until convergence is achieved.

Finally, we briefly discuss how the proposed MILP models could be extended with stochastic programming techniques to explicitly account for uncertainties in the parameters of the MILP models.

Reference

Zhang, Q. and I.E. Grossmann, “Enterprise-wide Optimization for industrial demand side management: Fundamentals, advances and perspectives,” Chemical Engineering Research and Design 116, 114-131 (2016).