Designing a parallel nonlinear model for predicting nitrogen oxide emissions

• Published in: AIP conference proceedings Vol. 3094; no. 1
• Main Authors: Migallón, Violeta, Navarro-González, Francisco José, Penadés, Héctor, Penadés, José, Villacampa, Yolanda
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Melville American Institute of Physics 07.06.2024
• Subjects: Algorithms; Emissions; Gas turbines; Goodness of fit; Message passing; Natural gas; Nitrogen oxides; Radial basis function
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0210140

A current grand challenge in the world is to reduce gas turbine-related pollutant emissions while simultaneously optimising their efficiency. The emissions of nitrogen oxides (NOx) are the most significant emissions when combusting natural gas in combustion turbines. The stringent regulations on emissions have led to a surge in research focused on decreasing NOx levels in gas turbines. In this work, a nonlinear radial basis function-based method has been analysed for predicting NOx emissions from gas turbines. Due to the computational demands of this method, parallelisation becomes indispensable to address this problem effectively. For this purpose, we have developed and analysed an MPI (Message Passing Interface) parallel algorithm. On the other hand, this method has been compared with the well-known K-nearest-neighbours (KNN) technique, which has recently been suggested for predicting NOx emissions. The efficiency of this parallel approach is shown using various measures of goodness of fit.