Progress in pyrolysis conversion of waste into value-added liquid pyro-oil, with focus on heating source and machine learning analysis

Abstract This work emphases the influence of using different heating sources (direct thermal, solar, infrared, microwave heating) on the pyro-oil yield. The effect of the dominating process parameters, namely the heating rate and final temperature, are thoroughly discussed with respect to the heating and reaction mechanism involved. Emphasis is then placed on reviewing the application of microwave (MW) heating in pyrolysis as a relatively new technology with many promising features, particularly the little-known mechanisms of MW heating, new MW heating pattern and pathway using MW absorbents for pyrolysis of waste materials. Machine learning (ML) techniques were then used to statistically analyze the 182 observations in 59 pyrolysis cases obtained from previous pyrolysis practices. The ML linear regression model was developed to predict oil yield by five input variables (feedstock type, feedstock size, heating rate, final temperature, and heating source), which can be used as a guideline for pyrolysis production management. By comparing three heating sources (direct, solar and MW), MW heating is found to be the most efficient method to achieve the highest oil yield. The Decision Tree Analysis demonstrates that the importance order for key variables is as: Log feedstock size > Log heating rate > Heating rate > Temperature > Feedstock size > Heating sources > Feedstock type. Future work should focus on optimizing the heating method and heating rate to achieve optimal yield and quality of pyro-oil. The findings are envisaged to be useful for scaling up the pyrolysis of waste materials for industrial energy applications.