การผลิตน้ำมันดิบชีวภาพจากไฮโดรเทอร์มอลลิควิแฟคชั่นของชีวมวลด้วยการจำลองแบบการเรียนรู้ของเครื่องจักรและการทดลอง

Abstract

Despite the COVID-19 outbreak, global energy demand continues to rise. meanwhile The production of energy from fossil fuels is also clearly detrimental to the environment and ecosystem balance. This forces humans to look beyond fossil fuels and look for alternative energy sources. Additionally, according to COP26 Glasgow, many countries have been asked to meet their global net zero emissions reduction targets by 2050 and maintain temperature changes of the Earth not to exceed 1.5°C. Among renewable and sustainable energy resources, biomass will play a vital role. Agricultural residues/wastes, energy crops, and lignocellulosic biomass, could potentially be major feedstocks for biorefineries. In Thailand, one of the most interesting energy crop is Grass species 4190, which is a high energy grass and less pollution than other grasses. It does not take long to grow, giving high yields. This has led to interest in using this grass to produce biocrude from hydrothermal liquefaction through machine learning simulations and experiments. From this thesis, it was found that machine learning simulation reduces time, budget for experimentation and help understand the direct and complex effects of features used in highly complex hydrothermal liquefaction processes. From the experimental results, it was found that machine learning gave a precise prediction of the yield and higher heat values of 87.11% with an error of 5.414 and matched the results obtained. In addition, the chemical composition of the products obtained from the hydrothermal liquefaction process was investigated by Gas Chromatography – Mass Spectrometry, Nuclear Magnetic Resonance Spectroscopy and Fourier Transform Infrared Spectroscopy.