Safety assessment: predicting fatality rates in methanol plant incidents

In this article, the prediction of fatality accident rate at methanol (MeOH) plant was studied using different assessment methods. The prediction method was performed and simulated using HYSYS, ALOHA, MARPLOT, and MATLAB software. Recent studies for pressure variation up to 442 bar in MeOH synthesis by carbon dioxide (CO₂) hydrogenation showed that three times more MeOH was produced than in conventional plants, with 90% CO₂ conversion and 95% MeOH selectivity. However, safety concerns were noted when MeOH production was operated at pressures above 76-500 bar. Therefore, a safety assessment of the pressures between 76 and 500 bar was performed to predict the fatality rate at the MeOH plant. Adaptive Neuro-Fuzzy Inference System (ANFIS) was compared with a conventional analysis by using the consequence method to predict the fatality rate. First, 26 input parameters were simulated in HYSYS, ALOHA, and MARPLOT software by using the consequence method. Then, the input parameters were reduced to six, namely, pressure, mass, volume, leakage size, wind speed, and wind direction, for prediction using ANFIS tool in MATLAB. This study aimed to highlight the accuracy of the fatality rate prediction by using the ANFIS method. In this manner, accurate prediction of fatality rate for MeOH plant incidents was achieved. The prediction values for the ANFIS method was validated using the standard error of the regression. The percent error measurement obtained the lowest regression of 0.0088 and the lowest percent error of 0.02% for Hydrogen (H₂) Vapor Cloud Explosion (VCE) ident. Therefore, the ANFIS method was found to be a simpler and alternative prediction method for the fatality rate than the conventional consequence method.