(734a) Multi-Objective and Dynamic Real-Time Optimization of Postcombustion Carbon Capture Processes for Cycling Applications

As the penetration of intermittent renewable energy sources, such as solar and wind, into the power grid increases^[1], there is a growing need for baseload fossil-fueled power plants to cycle their load accordingly^[2]. Cycling baseload plants may introduce unexpected inefficiencies into the system and increase costs, as the power plants are no longer operating at their optimal design point. Previously^[3], a dynamic real-time optimization (DRTO) framework was implemented on a reduced-order model of a methylethylamine (MEA)-based carbon capture subsystem (CCS) from a supercritical coal-fired power plant^[4]. This framework considered the cycling profile, electricity price market and cap&trade as forcing functions to determine the optimal output trajectories for different carbon capture landscapes. The obtained optimal dynamic trajectories can then ultimately be sent to lower-level advanced model-based controllers^[5].

Although the implementation of the DRTO framework for the CCS was performed successfully, in general, real-life situations are guided by more than a stand-alone economic objective to determine the optimum point at a certain time. Particularly, for the cycling scenario, objectives just as important such as emission reduction or minimization of wear and tear may be conflicting with the economic cost. To fill this gap, it is proposed in this work the incorporation of a multi-objective optimization (MOO) component into the DRTO framework for the calculation of optimal output trajectories considering all the objectives in tandem. In this case, more reliable optimal trajectories can be obtained, as the framework considers a more comprehensive context. As a trade-off, there is an increase in algorithm complexity, as new definitions (e.g., dominance) need to be introduced, and different strategies (e.g. aggregation, Pareto-front) need to be assessed to achieve the solution.

The reduced-order model of the subsystem considered in this work is a nonlinear autoregressive with exogenous inputs (NARX) with wavelets networks. The DRTO calculates the output optimal trajectories considering a one day-ahead prediction with a 1 hour simulated resolution. For the MOO component, it is desirable to minimize the decision maker interaction with the optimizer, so that the optimal trajectories can be determined and implemented at every time interval in an online and automated fashion. For instance, the optimizer should not wait for the decision maker to choose a point among a set of solutions to send the trajectories to the controllers. Thus, a multi-objective framework that guarantees a Pareto-optimal operating point is implemented in this work to keep the optimization independent of the hourly input of the decision maker. Furthermore, an algorithm without the necessity for system/search space convexity is also explored. The augmented framework is applied to a case study of the cycling carbon capture subsystem and the results are generated considering different carbon capture landscapes.

References

[1] REN21. (2018). Renewable Energy Policy Network for 21^st Century. Available at: http://www.ren21.net/status-of-renewables/global-status-report/. Accessed on April 16, 2018.

[2] BENTEK ENERGY LLC. (2010). How Less Became More: Wind, Power and Unintended Consequences in the Colorado Energy Market. Prepared for Independent Petroleum Association of Mountain States.

[3] KIM, R., LIMA F. V., (2017). Nonlinear System Identification and Dynamic Real Time Optimization of Postcombustion CO₂ Capture Processes for Cycling Applications. AIChE Annual Meeting, Minneapolis, MN.

[4] ZHANG Q., TURTON R., & BHATTACHARYYA D., (2016). Development of Model and Model-Predictive Control of an MEA-Based Postcombustion CO₂ Capture Process. Industrial and Engineering Chemistry Research. 55, 1292-1308.

[5] HE X, WANG Y, BHATTACHARYYA D, LIMA F. V, & TURTON R., (2018). Dynamic Modeling and Advanced Control of Post-Combustion CO₂ Capture Plants. Chemical Engineering Research and Design. 131, 430-439.