(67c) Improving Full-Scale Models of New Carbon Capture Technologies with Uncertainty Quantification

Carbon Capture technologies for combustion power plants aim to remove CO₂ from flue gas. One such technology is the novel CO₂-Binding Organic Liquid (CO₂BOL) process developed at the Pacific Northwest National Laboratory. The process utilizes their anhydrous CO₂BOL solvent in place of amine mixtures to reduce the energy penalty. A full-scale model of this system based on NETL’s Case 10 power plant is projected to produce 7-16% more net electric power over a traditional MEA system for the same plant. These full-scale model predictions are promising, however full-scale simulations are difficult to validate using only bench-scale data. There are errors in both the measurements and models that need to be considered in addition to uncertainties introduced by up-scaling. Uncertainty quantification (UQ) is a statistical framework used to better understand these uncertainties as well as data gaps in models. By constraining models to data, distributions of model parameters are estimated and then propagated through the model to obtain distributions of key outputs such as carbon capture percent, energy penalty, stripper temperature, etc. The results from this UQ analysis aid the design of experiments (DoE) which identifies data gaps to fill to improve model accuracy. Using UQ and DoE effectively can reduce the development time of new carbon capture by years and save time and money on the path to a pilot, and ultimately full-scale, plant.