Recent Papers

Bioethanol is a renewable, eco-friendly, and cost-effective alternative to fossil fuels derived from various biomass sources such as corn, sugarcane molasses, and other cellulosic materials. The usage and production of bioethanol has gained traction in recent years with good promise for future generations to come. In the Indian market, gasoline outlets are beginning to transition to a 20% blend of bioethanol called E-20 with gasoline from a 10% blend (E-10). This is done to maintain the calorific value of the blend along with reducing the emission of CO (carbon monoxide) into the environment. The study focuses on conducting a comparative analysis of analytical methods in the production of bioethanol, specifically aiming to compare the production of bioethanol from cellulosic material (Psidium Gujava) using Saccharomyces Cerevisiae (S. Cerevisiae). This is achieved by first collecting the leaves of Psidium Gujava and treating them initially to enhance the content of cellulose in the stock solution. After pre-treating the solution, the solution is sterilized and inoculated with S.Cerevisiae in the form of over-the-counter Bakers Yeast. Post-fermentation, the yield is purified, and yield is measured using a UV-Vis Spectrophotometer. Post the production of bioethanol, the yield concentrations are collected, interpolated to the required amount and analyzed using a Design of Experiments approach. The dataset collected in the form of CSV is put through various algorithms to predict the yield. The algorithms are developed, trained, and tested in Python using the sci-kit learn module in the case of Supervised learning models, and in the case of neural network regression, the algorithm will be developed in Pythons TensorFlow module using Google Colab. Regression algorithms like K-Nearest neighborsregression, Support Vector Regression, Decision Tree Regression, and Random Forest Regression were some of the models that were developed, trained, and tested and yielded a promising result with a good prediction of the bioethanol yield. Amongst the models above, the random forest regressor showed more accurate results with a cleaner prediction. Analysis of different built-in kernel types in Support Vector Machine Regression was also performed where the radial-based function (RBF) kernel showed more promise in terms of accuracy but the prediction value, the maximum predicted value was at around 10mgml whereas the actual maximum concentration value was at around 16 mgml. This situation was not the case for the other kernels, but the accuracy dropped drastically. Apart from the accuracy values, mean squared error, average of errors, and standard deviation in errors were also taken into consideration for analysis purposes.

Copyright © 2024 Vishal Murali. This is an open access article distributed under the Creative Commons Attribution License.