Microturbine Decarbonization with Machine-learning Regression Modeling

Session Number

CMPS 01

Advisor(s)

Dr. Chandrachur Bhattacharya

Dr. Debolina Dasgupta , Argonne National Laboratory

Discipline

Computer Science

Start Date

17-4-2024 10:25 AM

End Date

17-4-2024 10:40 AM

Abstract

Authorities all across the world are trying to minimize carbon dioxide (CO2) emissions, and this decarbonization step is necessary if the global climate issue is to be resolved. Researchers at Argonne have modified a natural gas-burning microturbine to burn natural gas-hydrogen fuel blends with the aim to reduce CO2 emissions. Emissions and efficiency data are obtained viaexperiments performed at Argonne. This research project demonstrates the use of machine learning Gaussian Process Regression (GPR) approaches to build data-driven models of the microturbine system. The collected experimental data is used to train the GPR models. The models are then used to determine the optimal operational parameters of the microturbine that minimize harmful emissions while maximizing efficiency. This study uses multi-output regression and the model confidence intervals to evaluate the performance of the microturbine under the full range of operating parameters. This project also determines the extent to which CO2, CO, and NOx emissions are altered when hydrogen is used as a substitute for natural gas and will lead to future decarbonization prospects.

Share

COinS
 
Apr 17th, 10:25 AM Apr 17th, 10:40 AM

Microturbine Decarbonization with Machine-learning Regression Modeling

Authorities all across the world are trying to minimize carbon dioxide (CO2) emissions, and this decarbonization step is necessary if the global climate issue is to be resolved. Researchers at Argonne have modified a natural gas-burning microturbine to burn natural gas-hydrogen fuel blends with the aim to reduce CO2 emissions. Emissions and efficiency data are obtained viaexperiments performed at Argonne. This research project demonstrates the use of machine learning Gaussian Process Regression (GPR) approaches to build data-driven models of the microturbine system. The collected experimental data is used to train the GPR models. The models are then used to determine the optimal operational parameters of the microturbine that minimize harmful emissions while maximizing efficiency. This study uses multi-output regression and the model confidence intervals to evaluate the performance of the microturbine under the full range of operating parameters. This project also determines the extent to which CO2, CO, and NOx emissions are altered when hydrogen is used as a substitute for natural gas and will lead to future decarbonization prospects.