Prediction of flammability limits for ethanol-air blends by the Kriging regression model and response surfaces

Elkin I. Gutiérrez Velásquez, Christian J.R. Coronado, Juan C. Quintero Cartagena, João A. Carvalho Jr., Andrés Z. Mendiburu, José C. Andrade, Ely V. Cortez, José C. Santos

CHALLENGE - Flammability limits must be identified in order to assess and control handling risks of particular processes according to combustion product composition and environmental conditions.​ The present paper aims to present a model that can predict flammability limits for ethanol-air blends with at different moisture concentrations by Kriging interpolation techniques.

SOLUTION - Experimental data were processed through the Kriging data interpolation statistical technique by using a Gaussian variogram.​ The modeFRONTIER software was used for data processing, which allows generating a Java code algorithm that allowed predicting flammability indexes for ethanol-air blends under specific conditions.​ Once the prediction algorithm was obtained, its validation procedure was performed. For such a purpose, 342 experimental data records were used through which a good performance of the generated algorithm was evidenced.​

BENEFITS - The designed algorithm predicted the flammability condition of​ ethanol-air blends at temperatures ranging between 20 °C and 210 °C, pressure between 40 kPa and 101.3 kPa, ethanol moisture concentration at 0.5% and 8%, and ethanol volume percentages between 1% and 35%.​