(373ao) A Two-Layer Integrated Scheduling and Economic Control Scheme for Demand Side Management of Chemical Process

Scheduling and control of industrial processes have historically been treated as separate decision layers operating on different time scales. When, however the scheduling horizon involves significant transient operation - e.g., decisions based on process and control dynamics - disregard of the latter leads to poor economic efficiency [1]. Hence, strategies to integrate scheduling and control are required [2, 3, 4]. Process integration in a demand-response (DR) paradigm includes reaction to frequent price fluctuations in different response schemes and, thus falls in the category [5, 6].

The two most investigated approaches for integrating scheduling and control are integrated dynamic scheduling (IDS) and economic nonlinear model predictive control (eNMPC) [3]. The former accounts for the process dynamics in the scheduling layer and retains the hierarchical automation architecture [7]. The latter integrates the scheduling objective into a model predictive controller [8]. A recent comparison of the two approaches showcases higher economic savings but also higher computational online costs of eNMPC compared to IDS [4]. In particular, the combination of fine control parameterization and long prediction horizon renders eNMPC computationally intractable.

Two characteristic examples of electricity markets considered in DR programs are the auction day ahead (DA) and the real-time intraday (ID) market. The former performs price clearing once per day and results in hourly constant prices. The latter, operates in a pay-as-bid fashion and provides 15-minute price granularity, enabling trading even a few minutes before delivery [9]. Numerous studies have considered process operation in multiple electricity markets, usually in the form of a two-stage bidding and optimal scheduling problem [10, 11]. However, there remains the open research question as to how a process can act as an electricity redispatch editor in multiple markets while simultaneously accounting for the closed-loop control dynamics.

Herein, we propose a “two-layer” (TL) multi-market scheduling and control strategy attaining a conjunction between the IDS and eNMPC paradigm. More precisely, we suggest the solution of an upper-level IDS problem which incorporates the concept of lower-order data-driven scale-bridging models (SBMs) [12] to decrease computational demands for long-term trading in the DA spot market. To this end, we use SBMs [12] to formulate the IDS problem in the first upper-level stage. In a second continuous online stage, we solve eNMPC problems with a short prediction horizon to account for redispatching in the continuous ID electricity market at a high frequency. We apply our proposed TL concept to an air separation unit (ASU) [13] and consider DA and continuous ID EPEX Spot markets [9]. We compare our results to the traditional IDS and eNMPC paradigms.

The results reveal superior economic savings of the eNMPC compared to the IDS approach for comparable, i.e., same average price, DA and ID prices. This is due to the restricted operation close to the tracking setpoints and use of simplified models in IDS [4]. TL showcases improved economic gains compared to IDS in the DA market. If juxtaposed with eNMPC with long prediction horizon, similar results are noted due to the maximum utilization of the ID price variability respecting the process constraints. However, TL provides greater savings even for short prediction horizons, showing that IDS can be leveraged to considerably reduce the required prediction horizon of eNMPC at no economic loss. Unlike eNMPC, the profit of TL remains significantly higher when the average ID prices is higher than the average DA prices. This indicates that TL is less sensitive to the price level, as it preserves the economic superiority of eNMPC while enabling further benefit from arbitrage. Overall, the results underpin the real-time applicability of the proposed method, while ensuring compliance to the process constraints.

Acknowledgements

The authors gratefully acknowledge the financial support of the Kopernikus project SynErgie by the Federal Ministry of Education and Research (BMBF), and the project supervision by the project management organization Projektträger Jülich (PtJ). They additionally acknowledge the financial support of and the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—333849990/GRK2379 (IRTG Hierarchical and Hybrid Approaches in Modern Inverse Problems).

References

[1] T. Tosukhowong, J. M. Lee, J. H. Lee, and J. Lu. “An introduction to a dynamic plantwide optimization strategy for an integrated plant”. Computers & chemical engineering 29.1 (2004), pp. 199–208.

[2] J. Kadam, W. Marquardt, M. Schlegel, T. Backx, O. Bosgra, P. Brouwer, G. Dünnebier, D .Van Hessem, A. Tiagounov, and S. De Wolf. “Towards integrated dynamic real-time optimization and control of industrial processes”. Proceedings foundations of computer-aided process operations (FOCAPO2003) (2003), pp. 593–596.

[3] M. Baldea and I. Harjunkoski. “Integrated production scheduling and process control: A systematic review”. Computers & Chemical Engineering 71 (2014), pp. 377–390.

[4] A. Caspari, C. Tsay, A. Mhamdi, M. Baldea, and A. Mitsos. “The integration of scheduling and control: Top-down vs. bottom-up”. Journal of Process Control 91 (2020).

[5] A. Flores-Tlacuahuac and I. E. Grossmann. “Simultaneous cyclic scheduling and control of a multiproduct CSTR”. Industrial & engineering chemistry research 45.20 (2006), pp. 6698–6712.

[6] A. Mitsos, N. Asprion, C. A. Floudas, M. Bortz, M. Baldea, D. Bonvin, A. Caspari, and P. Schäfer. “Challenges in process optimization for new feedstocks and energy sources”. Computers & Chemical Engineering 113 (2018), pp. 209–221.

[7] A. Helbig, O. Abel, and W. Marquardt. “Structural concepts for optimization based control of transient processes”. Nonlinear Model Predictive Control. Springer, 2000, pp. 295–311.

[8] M. Ellis, H. Durand, and P. D. Christofides. “A tutorial review of economic model predictive control methods”. Journal of Process Control 24.8 (2014), pp. 1156–1178.

[9] Fraunhofer Institute for Solar Energy Systems ISE. Energy Charts. Accessed: 2024-03-13. 2023.

[10] J. M. Simkoff and M. Baldea. “Stochastic Scheduling and Control Using Data-Driven Nonlinear Dynamic Models: Application to Demand Response Operation of a Chlor-Alkali Plant”. Industrial and Engineering Chemistry Research 59 (2020).

[11] T. Varelmann, N. Erwes, P. Schäfer, and A. Mitsos. “Simultaneously optimizing bidding strategy in pay-as-bid-markets and production scheduling”. Computers & Chemical Engineering 157 (2022), p. 107610.

[12] R. Mahadevan, F. J. Doyle III, and A. C. Allcock. “Control-relevant scheduling of polymer grade transitions”. AIChE Journal 48.8 (2002), pp. 1754–1764.

[13] R. C. Pattison, C. R. Touretzky, T. Johansson, I. Harjunkoski, and M. Baldea. “Optimal process operations in fast-changing electricity markets: framework for scheduling with low-order dynamic models and an air separation application”. Industrial & Engineering Chemistry Research 55.16 (2016), pp. 4562–4584.