Superstructure design, data-driven mixed-integer optimization, and guidelines for techno-economic-environmental enhancement of blended amine-based CO2 capture in natural gas combined cycle power plants

• Main Authors: Vo, Dat-Nguyen, Zhang, Zhiwei, Yin, Xunyuan
• Other Authors: School of Chemistry, Chemical Engineering and Biotechnology
• Format: Article
• Language: English
• Published: 2025
• Subjects: Engineering; Data-driven mixed-integer optimization; NGCC power plant
• Online Access: https://hdl.handle.net/10356/184153

Blended amine-based post-combustion CO2 capture (PCC) for natural gas combined cycle (NGCC) power plants lack efficient superstructure design, viable mixed-integer optimization approach, and practical guidelines to reduce capture cost, heating energy consumption, and amine emissions. This study introduces an efficient superstructure design comprising 12 feasible configurations, a viable data-driven mixed-integer optimization approach, and design and operation guidelines to enhance the techno-economic-environmental performance of decarbonizing a 400 MW NGCC power plant using Monothanolamine/Piperazine (MEA/PZ). First, we develop a high-fidelity process simulator to decarbonize the NGCC power plant using MEA/PZ. The validated simulator is then employed to create the superstructure design consisting of the 12 feasible configurations to reduce the heating energy consumption, capture cost, and amine emissions. Due to the cyclic nature of the closed-loop PCC process, each steady-state simulation takes 20 min to complete. To expedite the process, we develop a one-hot vector deep neural networks (OHV-DNN) model based on 4220 synthetic data cases, enabling accurate prediction of key performance indicators for the superstructure design in just one second. This DNN model is subsequently used to formulate economic mixed-integer optimization problems, which can be solved within one minute. Under the optimal conditions for capturing 90% of CO2 and emitting less than 1 ppm of MEA and PZ concentrations, the MEA/PZ-based optimal design significantly reduces the capture cost by 14.34% and 24.39% and cuts the heating energy consumption by 17.21% and 30.82% compared to the conventional design using MEA/PZ and MEA. Additionally, the analysis reveals the importance of design selection and optimal operating zone, as well as the impact of decision variables on the techno-economic-environmental performance of the MEA/PZ-based optimal design. The findings and the proposed approach are highly beneficial for decarbonizing NGCC power plants and can be extended to other concentrated CO2 sources.