Machine learning assisted techno-economic and life cycle assessment of organic solid waste upgrading under natural gas

Abstract

Conducting a techno-economic assessment (TEA) and life cycle assessment (LCA) is essential prior to scale up or industrializing new organic solid waste (OSW) upgrading processes, as well as before conducting a feasibility study on the utilization of new biomass in a previously industrialized process. These assessments involve time-consuming and expensive experiments aimed at determining the yield of the process. In this study, a machine learning method is utilized as a toolbox and is coupled with simulation to save time and money by omitting the above-mentioned experiments for techno-economic and life cycle assessment. The bio-oil yield in the pyrolysis process is predicted using the artificial neural network (ANN) model based on both ultimate and proximate analyses of the biomass and temperature of the pyrolysis process from the literature review. For tuning the ANN model, a semi-auto-tuning method was developed. The results show the high predictability of the model with an R2 of 0.81 for unseen data. Following this, a new process of upgrading organic solid waste using natural gas is simulated using the Aspen Plus software to determine the material and energy balances. According to the economic evaluation, using natural gas significantly reduces the minimum selling price (MSP) of renewable diesel. In this study, $3.5/gal is the minimum selling price, which is 22% lower compared to similar plants in other literature reviews. A Monte Carlo simulation was also performed to investigate the uncertainty, and the results indicated that, with a probability of 50%, the net present value (NPV) is greater than the NPV calculated deterministically. Based on the results of the life cycle assessment, the newly proposed process emits 66% less amount of greenhouse gases than other commercial processes.