(398h) Multi-Objective Optimization of Solid Sorbent-Based CO2 Capture Systems

Authors: 
Zamarripa, M., National Energy Technology Laboratory
Eslick, J., National Energy Technology Laboratory
Miller, D. C., National Energy Technology Laboratory
This work presents a multi-objective optimization framework that uses a mixed integer nonlinear programming model (MINLP) for the optimal design and operation of carbon capture plants. A mix of first principle and surrogate-based models have been developed to characterize the process design and operation.

Coal-based power generation is one of the most important sources of energy around the globe. As such, it is important to develop cost-effective carbon capture and storage (CCS) technologies to reduce the environmental impact of power plants. Reducing emissions without affecting the overall cost of the system is difficult. Therefore, to examine the trade-off between cost and emissions, advanced energy analysis tools that consider multiple criteria are needed. The state of the art in technologies for carbon capture include solvents, membrane and solid sorbent systems. This work analyzes the solid sorbent-based adsorption system under different objectives (such as minimizing the cost of electricity and maximizing capture rate, operating vs capital cost, etc.).

In solid sorbent-based capture systems, the process is mainly driven by the interaction of the solid sorbent particles with the flue gas (mainly consisting of N2, CO2, and H2O). The main process units of the solid sorbent systems are multiphase reactors, such as moving bed and fluidized beds. Given the modeling complexity and the difficulty of formulating algebraic models for such processes, previous studies either are generally limited to process-simulation1 based frameworks or simplified models for mathematical optimization. Process simulators limit the exploration of different process configurations, while in the latest example the validation of the process behavior is limited. Our previous work proposed a mathematical optimization framework, in which both first principles-based and surrogate models2 were developed to obtain the optimal plant design and operating conditions3. Carefully-tuned detailed models implemented in commercial process simulators have been exploited in the preparation and characterization of surrogate models with ALAMO2. Model validation and sample generation play a very important role in this mathematical framework.

The proposed mathematical model considers a set of components (CO2, N2, H2O, and Carbamate), streams (utilities, gas, and solids), a set of units (adsorbers and regenerators mixers, splitters, compressors, pumps and heat exchangers), and a set of technologies (two reactor configurations for adsorption and regeneration). Design decisions determine the plant layout, by selecting the optimal number of parallel trains, process configuration (number of beds for adsorption and regeneration processes), and equipment design for each unit (diameter, length, area, etc.). At the operating level, the decisions involve the optimal flowrates, temperatures, molar fractions of gas and loading of the solids.

Making use of the reduced-order models generated from our previous work, this work uses the normal-constraint method for decision making under multiple objectives. The general idea is to obtain the Pareto frontier generating a set of solutions by solving multiple optimization problems4. Finally, one of the Pareto optimal solutions will be selected based on the Utopia-tracking approach5.

[1] Eslick J. C., Tong C., Lee A., Dowling A. W., Mebane D. S. (2015). Optimization under uncertainty with rigorous process models. 2015 AIChE Annual Meeting.

[2] Cozad A., Sahinidis N. V., Miller D. C. (2014). Automatic Learning of Algebraic Models for Optimization. AIChE Journal, 60, 2211-2227.

[3] Zamarripa M., Eslick J., Lee A., Ajayi O., Wilson Z., Sahinidis N., Miller D. (2016). Optimal Design and Operation of Hybrid CO2 Capture Systems. 2016 AIChE Annual Meeting.

[4] Messac A., Ismail-Yahaya A., Mattson C. A. (2003). The normalized normal constraint method for generating the Pareto frontier. Structural and Multidisciplinary Optimization, 25, 86-98.

[5] Zavala V. M. (2015). Managing Conflicts among Decision-Makers in Multiobjective Design and Operations. 2015 AIChE Annual Meeting.