Systems design and uncertainty quantification of co2 capture from air
MetadataShow full item record
The rapid increase in concentration of atmospheric CO2 has stimulated the recent development of CO2 capture technologies. One of the strategy is to capture CO2 directly from ambient air which, if successfully implemented, could result in capture of CO2 from disperse emission sources. My work proposes comparison of adsorbents for Direct Air Capture (DAC) through temperature vacuum swing adsorption (TVSA). The adsorbents are grown as films inside monolith contactors which offer low pressure drop during the adsorption step. A cyclic steady state process has been simulated and detailed techno-economic study has been performed on DAC. I have identified sensitive parameters such as gas flow rate, cycle time, adsorbent purchase cost, which effects the overall energy requirements and net economics of the Direct Air Capture process. Another area of CO2 capture is enclosed environments. Removal of CO2 from enclosed environment, such as commercial buildings is also of critical importance, primarily due to health risks associated with high CO2 concentration (>0.5%). In this study, I have designed and modeled a modified air conditioning system which regulates temperature as well as air quality inside an enclosed environment. The model is simulated on a 24 hour timescale including time varying human occupancy of the room. The system is optimized keeping realistic constraint as per ASHRAE standards to maintain a complex state of CO2, O2 and humidity levels with operation of air flow through different beds and recirculation. The optimized model shows improvement in performance as compared with conventional ventilation systems. CO2 capture systems have been modeled by understanding the physical process and solving coupled heat and mass balances equations through numerical simulators. Model parameters such as mass transfer coefficient, inlet flow rates etc. often remains uncertain due to lack of precise measurement as well as incomplete information. The accuracy of model outputs and performance metrics such as the net energy and cost can be questioned because of these parametric uncertainties. Hence, quantification of these uncertainties is necessary for more useful model predictions. I have employed Polynomial Chaos Expansion (PCE) methods (intrusive and non-intrusive) to quantify model uncertainty and compared it with Monte Carlo Simulations (MCS). Galerkin method is used to solve intrusive case, Latin Hypercube is used as sampling method and Nataf transformations are used for handling correlations in random variables. The PCE methods have shown better performance in terms of computational time, as compared to MCS. The intrusive PCE method shows higher complexity in model formulation but reduced computational time as compared to non-intrusive PCE method. In order to model all these systems, partial differential algebraic equations have been implemented in gPROMS, which is a commercial dynamic process modeling and optimization software. Linear driving force (LDF) model is used to approximate the rate of CO2 adsorption. Further, MATLAB lsqnonlin solver is used to solve nonlinear curve fitting for parameter estimation of isotherm data points obtained from experiments.