Advanced process integration and machine learning-based optimization to enhance techno-economic-environmental performance of CO2 capture and conversion to methanol

Abstract

Developing economically sustainable CO2 capture and conversion processes is essential to realize carbon neutrality. This study proposed an integrated process for CO2 capture and conversion-to-methanol (CCTM) and applied machine learning-based optimization to enhance techno-economic-environmental performance. After validating CO2 capture and CO2-to-methanol sections, an advanced CCTM design was developed and compared with conventional one regarding techno-economic-environmental performance across various operating scenarios. The advanced CCTM exhibited significant improvements in energy consumption (14.73–16.30%), production cost (0.81–1.28%), and net CO2 reduction (3.13–3.38%) owing to efficiently reusing waste heat, off-gas, and water resources. The one-at-a-time sensitivity analysis revealed roles of each variable and nonlinear variable-performance tendencies among operating variables in the advanced CCTM process. Subsequently, a well-developed deep neural network (DNN) model precisely formulated the relationship between key variables and performances. The DNN-based optimization provided optimum operating conditions within a minute, resulting in an 8.21 $/tMeOH (∼0.81%) reduction in production cost compared to base case of CCTM. Notably, the total CO2 capture rate of 92.53% at an optimal condition highlighted the significant contribution of advanced CCTM to carbon neutrality. The findings provide a viable reference for the effective and sustainable design and operation of an integrated CCTM process.