Appraising raw exhaust pollutant gases emissions from industrial generators using statistics and machine learning approaches

Résumé

Industrial generators, widely used for backup power generation, emit significant levels of pollutant gases such as carbon monoxide (CO), carbon dioxide (CO2), hydrocarbons (HC), and nitrogen oxides (NOx). These emissions exacerbate air pollution and climate change, while their inhalation adversely impacts human health, leading to respiratory/cardiovascular diseases and increased mortality rates. Raw exhausts of CO, CO2, HC, NOx, and O2 from industrial generators were assessed using a portable analyser. Thereafter, the obtained dataset was analysed using multiple linear regression and Pearson’s correlation to quantify the synergistic impact of generator characteristics, while the study equally trained 70% of the dataset using machine learning (ML) classification models. The result showed that generators’ age and capacity impacted considerably on exhaust concentrations as the diesel-powered generators exhibited higher CO2 and NOx emissions at 76.1% and 7393ppm, respectively, compared to gas-powered generators. For diesel-powered generators, there was a moderate negative correlation at -0.49142 and p-value of 0.03281 for CO and NOx. For the gas-powered generators, the correlation is statistically significant for CO and HC, while there was an inverse association between NOx and O2. The employed ML models achieved high prediction accuracy range of 80.6?93.5 % for exhaust pollutant gases for OGEPA classification status. Based on this study, policy frameworks should be implemented up to impose stringent generator emissions standards to reduce air pollution, invest in expanding/upgrading the national electricity grid to reduce reliance, provide low-interest green loans to finance renewable energy systems, as well as access climate finance mechanisms to subsidise clean energy projects.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0