Modeling and optimization of the yield of pyrolytic oil from waste face masks using RSM‐ANN‐LM hybrid approach

Abstract

Pyrolysis is one of the most widely practiced thermochemical conversion technique to convert biomass into bio fuel. In this investigation, waste surgical masks were taken for pyrolysis process and the pyrolysis oil yield was determined experimentally. The experiments were designed and optimized based on Box Behnken Design (BBD) with operating parameters (feed size, temperature, and time) to study the effect on the yield of the pyrolysis oil and char. Experiments were conducted by varying the feed size (10–50 mm), pyrolysis temperature (400–600°C), and time (30–90 min). A hybrid Response Surface Methodology –Artificial Neural Network‐ Levenberg–Marquardt algorithm (RSM‐ANN‐LM) modelling approach has been to optimize the process parameters for the prediction of pyrolysis oil yield. The optimized network architecture was found to be 3‐10‐1 and the authenticity of the developed model has been evaluated using Regression Co‐efficient (R2) and Mean Squared Error (MSE). The developed RSM‐ANN‐LM model has outperformed the RSM model on predicting the pyrolytic oil yield. For the optimized conditions of 10 mm size, 600°C of pyrolysis temperature and 90 min pyrolysis time, 85.30% oil yield was obtained.