(675e) Carbon Capture Plant Model Identification through Simultaneous State and Parameter Estimation with Sensitivity Analysis

Net-zero emissions, which involve the elimination or offsetting of greenhouse gas (GHG) emissions [1], have recently garnered significant attention due to their environmental and sustainable benefits. According to recent studies [2], in 2020, electricity and heat generation ranked first worldwide in GHG emissions, and manufacturing and construction claimed the third spot. Hence, carbon dioxide (CO2), as the most important GHG [3], has gained considerable attention in these sectors, leading to the installation or plans to install carbon capture plants (CCP) on outlet gas streams of exiting processes to capture CO2 that would otherwise be emitted into the atmosphere (e.g., oil and gas [4, 5], cement production [6], etc.). For achieving various capture quotas or requirements from different stakeholders, advanced scheduling and control are required,
necessitating a high-fidelity model.

In this study, we focus on an industrial CCP illustrated in the attached figure. The plant comprises an absorber and a desorber, forming a vital component pair. Within the absorber, a liquid amine-based solution facilitates the absorption of CO2 from the flue gas. Adjacent to the desorber, a reboiler supplies energy to the process, aiding in the separation of CO2 from the liquid, with the captured CO2 collected from the top gas outlet of the desorber. An integral lean-rich heat exchanger (LRHE) positioned between the absorber and desorber effectively utilizes internal energy, thereby reducing the need for additional cooling and heating energy inputs.

The aim of this study is to identify a first-principles model for a CCP utilizing simultaneous state and parameter estimation with sensitivity analysis, leveraging temperature sensor measurements to predict captured CO2 . First, a first-principles model is derived. Then, simultaneous state and parameter estimation is employed to estimate the model parameters based on measured output data. In the process, the model parameters are treated as augmented system states. The sensitivity matrix of the measured outputs to the states and parameters is evaluated based on process data. Based on the sensitivity matrix, orthogonalization is used to determine estimable states and parameters. These estimable states and parameters are then estimated within a moving horizon estimation (MHE) framework.

The resulting model is utilized to predict captured CO2 , with its performance compared against benchmark approaches to demonstrate effectiveness. Preliminary results reveal approximately a 12.0% discrepancy between estimated and actual captured CO2 measurements, showcasing the potential for predictions in advanced control and planning. The subsequent part of the study investigates the benefits of simultaneous state and parameter estimation combined with variable selection using sensitivity analysis. Results indicate a notable improvement in estimation accuracy, with enhancements of 27.5% and 32.3% observed when compared to state-only estimation and state and parameter estimation without selection, respectively.

References
[1] Net-zero emissions by 2050 - canada.ca. https://www.canada.ca/en/services/environment/weather/climatechange/clim.... (Accessed on 04/06/2024).

[2] Hannah Ritchie, Pablo Rosado, and Max Roser. Emissions by sector: where do greenhouse gases come from? Our World in Data, 2020. https://ourworldindata.org/emissions-by-sector.

[3] Overview of greenhouse gases — us epa. https://www.epa.gov/ghgemissions/overview-greenhouse-gases#:~:text=Carbo...(CO2)%20is,gas%20emissions%20from%20human%20activities. (Accessed on 04/06/2024).

[4] Ccus in the decarbonization of upstream production in brazil — s&p global. https://www.spglobal.com/commodityinsights/en/ci/research-analysis/ccus-.... (Accessed on 04/06/2024).

[5] Labarge: Gateway to the future — exxonmobil. https://corporate.exxonmobil.com/what-we-do/delivering-industrial-soluti.... (Accessed on 04/06/2024).

[6] Why this company is adopting carbon capture and storage technology in its cement production — cbc news. https://www.cbc.ca/news/canada/edmonton/why-this-company-is-adopting-car... ction-1.7080725. (Accessed on 04/06/2024).