Estimation of cold flow properties of biodiesel using ANFIS-based models

Abstract

ABSTRACT Biodiesel fuel tends to freeze at higher temperatures when compared with the traditional diesel fuel based on petroleum. Characterization and improvement of its poor cold flow properties (CFP) including cloud point (CP), pour point (PP), and cold filter plugging point (CFPP) have become one of the major challenges among researchers. The aim of the study was to develop a methodology to predict the three CFP of a biodiesel sample based on the fatty acid composition of its feedstock. CP, PP and CFPP prediction models based on Adaptive Neuro-Fuzzy Interference System (ANFIS) were developed. Saturated, monosaturated and polyunsaturated fatty acid contents were used as input variables. Seventy-five data sets including fatty acid contents of feedstocks and CFP of their biodiesels were used during training and verification stages. Biodiesel samples were produced from six different feedstocks either by base-catalyzed transesterification or by supercritical methanol transesterification methods in order to inspect the accuracy of the models. CFP of biodiesel samples were determined following the EN and also ASTM standards. It was noted that the CFP estimated by the ANFIS models were in close agreement with the measured values. The R2 values were found as 0.981, 0.985, and 0.981 for CP, PP, and CFPP, whereas the prediction performances of the models in terms of RSME were 1.066, 1.160, and 1.061, respectively. It was noteworthy that the present ANFIS models provided closer estimations than those models reported previously.