(434c) Integrating Price Predictions with Optimization Framework for Cost-Effective Industrial Electrification

The decarbonization of the chemical industry could be greatly aided by industry electrification. In the industrial sector, there are numerous benefits to switching from fossil fuel-based process heating to electrical heating. Nevertheless, operating costs are frequently significantly higher than fossil fuel-based heating [1]. A decrease in the price of electricity is crucial for encouraging the electrification of more industries. In chemical facilities of the future, electrified process heating could be coupled with energy storage and a backup green hydrogen burner to increase operational flexibility [1]. For instance, thermal energy storage (TES) utilizes low-cost storage mediums such as molten salt, rock, and sand that are non-toxic and more accessible, making it easier to scale up for industrial integration.[2] By incorporating the electricity procurement strategy with TES and the scheduling of backup green hydrogen burner, facilities could benefit from altering energy loads in response to fluctuating electricity prices. To effectively manage electricity costs, some researchers have proposed integrating electricity procurement and production scheduling using stochastic optimization. For example, Zhang et al. (2016) considered the case of purchasing electricity from power contracts and real-time spot markets [3], while other studies have explored long-term reserve market bidding [4-5], optimal day ahead plant scheduling [6-7], and real-time demand response programs [8-10]. Germscheid et al. (2022) have analyzed the potential for demand-response programs in chemical processes while accounting for both day-ahead market (DAM) and real-time market (RTM) pricing [11]. However, when the DAM decision is made, the price prediction of the intraday market is not considered. According to the 2020 hourly electricity price profile in Texas Houston area from ERCOT, one-third of the hours’ day-ahead market price is higher than the real-time spot market price. Therefore, to fully capitalize on price fluctuations, it may be necessary to predict both day-ahead market and intraday market prices and adjust energy usage accordingly. For large industrial customers, the ability to predict market prices and optimize their electricity procurement could lead to significant cost savings and increased profits.

Predicting electricity prices is challenging due to the volatile and unpredictable nature of the energy market. Electricity prices can be affected by a variety of factors, including weather conditions, supply and demand dynamics, government policies, geopolitical events, and technological advancements. Additionally, the integration of renewable energy sources such as wind and solar power has increased the level of uncertainty in the electricity market, as these sources are subject to intermittent and variable generation. As a result, accurately predicting electricity prices requires sophisticated models that can incorporate a wide range of variables and account for the inherent uncertainty and volatility of the market.

To address this, we propose a combination of a recurrent neural network (RNN) time series price prediction models of both DAM and RTM with a stochastic mixed-integer programming (MILP) optimization framework to address uncertainties in prediction. The objective of this framework is to minimize the operation cost and electricity costs while addressing uncertainty in prediction. The authors apply this framework to a multi-period heat integration (MPHI) problem in a chemical process, considering a set of hot and cold stream inlet and outlet temperatures, flow rates, heat capacity, operation parameters, and cost parameters of thermal energy storage (TES) and green hydrogen burner. To effectively manage electricity costs for the chemical process in each operational day, the problem is divided into four steps each day. First, based on the predicted prices of two electricity markets, the market with the lower price is selected for each hour of operation. If the day-ahead market (DAM) is chosen, the second step is attending the DAM auction (24 hours prioritized to the operation day), where the demand bidding load and price are submitted. Once the DAM market is cleared, the actual clearing price is received. If the clearing price is lower than the bidding price, the demand load is secured at the clearing price. However, if the clearing price is higher than the bidding price, the bidding is rejected, and the third step is entered. The third step involves purchasing electricity at the RTM price, which is updated by the known DAM price of the operation day. The RTM price remains unknown until electricity is purchased at each hour. Finally, the demand load profile is sent to the multi-period heat integration (MPHI) framework to determine the optimal scheduling for the process. Through a case study, we illustrate how the proposed prediction and optimization framework provides the optimal electricity procurement strategy and the corresponding scheduling of heat integration, TES charging/discharging and green hydrogen burner for a chemical process.

References

[1] Wei, M., McMillan, C. A., & de la Rue du Can, S. (2019). Electrification of industry: Potential, challenges and outlook. Current Sustainable/Renewable Energy Reports, 6(4), 140–148.

[2] Zantye, M. S., Gandhi, A., Li, M., Arora, A., & Hasan, M. M. F. (2022). A systematic framework for the integration of carbon capture, renewables and energy storage systems for Sustainable Energy. Computer Aided Chemical Engineering, 2089–2094.

[3] Zhang, Q., Cremer, J. L., Grossmann, I. E., Sundaramoorthy, A., & Pinto, J. M. (2016). Risk-based integrated production scheduling and electricity procurement for continuous power-intensive processes. Computers & Chemical Engineering, 86, 90–105.

[4] Schäfer, P., Westerholt, H. G., Schweidtmann, A. M., Ilieva, S., & Mitsos, A. (2019). Model-based bidding strategies on the primary balancing market for energy-intense processes. Computers & Chemical Engineering, 120, 4–14.

[5] Leo, E., & Engell, S. (2018). Integrated Day-Ahead Energy Procurement and production scheduling. At - Automatisierungstechnik, 66(11), 950–963.

[6] Zhang, X., & Hug, G. (2015). Bidding strategy in Energy and spinning reserve markets for aluminum smelters' demand response. 2015 IEEE Power & Energy Society Innovative Smart Grid Technologies Conference (ISGT).

[7] Leo, E., Dalle Ave, G., Harjunkoski, I., & Engell, S. (2021). Stochastic short-term integrated electricity procurement and production scheduling for a large consumer. Computers & Chemical Engineering, 145, 107191.

[8] Otashu, J. I., & Baldea, M. (2018). Grid-level “battery” operation of chemical processes and demand-side participation in short-term electricity markets. Applied Energy, 220, 562–575.

[9] Simkoff, J. M., & Baldea, M. (2020). Stochastic scheduling and control using data-driven nonlinear dynamic models: Application to demand response operation of a chlor-alkali plant. Industrial & Engineering Chemistry Research, 59(21), 10031–10042.

[10] Schäfer, P., Daun, T. M., & Mitsos, A. (2020). DO investments in flexibility enhance sustainability? A simulative study considering the German electricity sector. AIChE Journal, 66(11).

[11] Germscheid, S. H., Mitsos, A., & Dahmen, M. (2022). Demand response potential of industrial processes considering uncertain short-term electricity prices. AIChE Journal, 68(11).