(92b) Modular Process Intensification for Carbon Capture Using Data Driven Optimization

Modular chemical process intensification (PI) has attracted significant interest in the recent literature. Unlike conventional large-scale facilities, modular PI involves combining unit operations that usually appear together or dividing a unit into smaller subunits to enhance a specific function [1]. A modular facility consists of modules that can be easily added or removed, allowing more flexibility in capacity, product type, and geographic location [2]. Carbon capture is one area that can benefit from modularization. Depending on the source of emission, there exists a variability in the flue gas composition. For instance, post-combustion flue gas from coal-fired power plants has CO₂ concentrations ranging between 12.6-14%, while that of steel plants has a concentration ranging from 20-24% [3]. Furthermore, depending on the size and location of chemical plants, the modular design can be easily adjusted to handle different capacity and change in climate. Since most separation technologies only target a certain range of carbon content in flue gas under specific operating conditions, the performance and cost of carbon capture are largely affected by these variabilities. Thus, the identification of optimal modular design is crucial.

Optimization-based process synthesis can be used to determine the optimal modular design based on a given operating condition. Process synthesis involves developing a process superstructure and determining the optimal process configurations and parameters to achieve a certain process specification, such as minimizing energy requirement or maximizing profit. Process synthesis can be formulated into a mixed-integer nonlinear problem (MINLP) to allow selection of modules as well as optimal operating conditions. Accounting for all complex interactions between design decisions, optimization-based process synthesis is superior to heuristics-based process synthesis. However, one challenge of optimization-based process synthesis is that the resulting optimization problem is usually difficult to solve, especially when highly-complicated and accurate simulations are used to represent a module. The use of surrogate-based optimization has been proposed to reduce the complexity of the optimization problem [4, 5]. Surrogate-based optimization involves replacing complex computer simulations with surrogate models; the surrogate model is then optimized to determine the best modular design.

In this work, we propose an optimization-based modular PI strategy for carbon capture. We will specifically study modular vacuum pressure-swing adsorption (VPSA) with metal organic framework (MOFs). Several different types of MOFs will be integrated into modules, and the optimal modular design will be determined to address different variabilities intrinsic to carbon capture process, such as flue gas composition, geographic location of the plant, and desired capacity of the carbon capture facility. These variabilities will be analyzed using an actual CO₂ production data. Several factors of carbon capture process, such as purity, recovery, energy requirement, and productivity, will be used to analyze the performance of the modular design. The optimal modular design will be determined using surrogate-based optimization to reduce the complexity of the optimization problem.

References

1. Schug, B.W. and M.J. Realff, Design of standardized, modular, chemical processes. Computers & Chemical Engineering, 1996. 20: p. S435-S441.
2. Baldea, M., et al., Modular manufacturing processes: Status, challenges, and opportunities. AIChE Journal, 2017. 63(10): p. 4262-4272.
3. Hasan, M.M.F., et al., Modeling, Simulation, and Optimization of Postcombustion CO2 Capture for Variable Feed Concentration and Flow Rate. 1. Chemical Absorption and Membrane Processes. Industrial & Engineering Chemistry Research, 2012. 51(48): p. 15642-15664.
4. Caballero, J.A. and I.E. Grossmann, An algorithm for the use of surrogate models in modular flowsheet optimization. AIChE Journal, 2008. 54(10): p. 2633-2650.
5. Henao, C.A. and C.T. Maravelias, Surrogate-based superstructure optimization framework. AIChE Journal, 2011. 57(5): p. 1216-1232.