Prosopis juliflora valorization via microwave-assisted pyrolysis: Optimization of reaction parameters using machine learning analysis

Microwave power and pyrolysis temperature are essential parameters in optimizing the bio-oil yield and quality in microwave pyrolysis. This study focused on understanding the interactions between the microwave power/ heating rate and pyrolysis temperature in microwave-assisted pyrolysis of Prosopis juliflora. For optimum bio-oil yield, a discrete set of microwave powers (280 W, 420 W, and 560 W) and pyrolysis temperatures (200 degrees C, 350 degrees C, and 500 degrees C) were selected. A central composite design (CCD) was adopted to analyze the effect of mi-crowave power and the pyrolysis temperature on product yields, heating rate, microwave conversion efficiency, and heat losses in pyrolysis. Moreover, the effect of heating rate, reaction time, specific microwave power, specific microwave energy, and conductive heat loss on gas, char, and liquid yields was evaluated using sta-tistical machine learning techniques. Moreover, a new parameter, pyrolysis index, is calculated under different conditions to understand the extent of pyrolysis intensity using pyrolysis time, heating value, feedstock mass and conversion, and microwave energy conversion. The yields of bio-oil, biochar, and gas were 25-40 wt%, 25-35 wt %, and 35-40 wt% at different experimental conditions. Bio-oil consists of a mix of organic compounds with methoxy phenols at high selectivity, and the calorific value of bio-oil was in the range of 26-28 MJ/kg. Carbon number analysis revealed higher presence of C5-C9 compounds. This study shows the role of machine learning in understanding the effect of various parameters effectively and optimizing the experimental conditions accordingly.

Automated summary

This study focuses on the interactions between microwave power/heating rate and pyrolysis temperature in microwave-assisted pyrolysis of Prosopis juliflora. A central composite design is used to analyze the effects of microwave power and pyrolysis temperature on product yields and heat losses in pyrolysis. Statistical machine learning techniques are also used to evaluate the effects of other parameters on gas, char, and liquid yields.