PI: Zheng Yao
Co-PI(s): Carlos Romero
University: Lehigh University
Industry partner: N/A
The proposed project will investigate and assess the capabilities of Machine Learning with Laser Induced Breakdown Spectroscopy (LIBS) to measure the higher-order parameters of the feedstock of interest to gasifier operators, such as thermal conductivity. The feedstock will include biomass, waste plastics, and waste coal, individually and in blends. It is expected that the proven approach will make it possible to improve feedstock characterization throughput over baseline manual methods and will lead to a leap in developing online measurement technology for application in gasifiers for hydrogen production. The conventional baseline method used in the characterization of gasifier feedstock consists of numerous manual samplings of the feedstock material, from either a feed conveyor or piles, mixing and compositing those samples to achieve a representative sample, followed by laboratory analysis of those representative samples by standardized techniques, typically involving ultimate and proximate analyses, as well as calorific value of the sample. The results of this baseline method take too long (days whereas minutes are required) for efficient characterization of feedstock, and more importantly, for efficient utilization of the feedstock information in the conversion-to-hydrogen processes, where real-time feedstock information is critical for process control.