Carbon Capture and Storage Optimization with Machine Learning using an ANN model

Abstract

The purpose of this study is to evaluate the accuracy of predictions regarding the work capacity of CO2 and the selectivity of MOF, using machine learning methodologies in relation to CO2/N2. A dataset was used that includes numerous characteristics of MOFs for the development of a neural network model. The factors that determined the operational capacity of CO2 and the CO2/N2 selectivity included pore size, surface area, chemical composition, among others. The model demonstrated its work capacity by evaluating the selectivity of CO2; the mean absolute errors for the CO2/N2 selectivity were 25 and 0.8 mmol/g, respectively. The correlation Analysis showed a fairly negative correlation (-0.014) between the operational capacity of CO2 and its chemical makeup and a very positive correlation ( 0.029) between the surface area and amount of pore size. Thus, the gas absorbability is not top-dependent exclusively; pore size and surface area of a material contribute to the capacity as well. More research should be carried out to evaluate a machine learning capability on predicting the nature of different Flow Object Models (MOFs) with an aim of increasing efficiency, precision and dependability of the models.

Keywords

• metal-organic framework
• machine learning
• co2 working capacity
• co2/n2 selectivity
• prediction