Machine Learning Enhanced LIBS to Measure and Process Biofuels and Waste Coal for Gasifier Improved Operation

The use of coal waste and biomass has significant environmental benefits, and it can contribute to promoting a low-carbon economy via hydrogen production. These feedstocks can be cost effective, are readily available, and in the case of biomass are renewable with the near elimination of greenhouse gases. But there are issues with gasifying waste coal and biomass. The first one is the widely varying organic makeup and moisture content of the feedstock which makes optimization and proper control of the gasifier challenging. The second is the impact on the slag properties of the inert part of the fuel, affecting reactor operation and reliability. To solve these issues the project team developed machine learning (ML) enhanced diagnostics to enable gasifiers to process coal waste and biomass optimally and economically; thus, avoiding them from being landfilled and taking advantage of their value and favorable environmental properties. Laser induced breakdown spectroscopy (LIBS) along with the use of advanced ML signal processing can accurately measure important properties of the feedstock such as proximate and ultimate analyses, and ash content, elemental concentrations, heating value, and ash slagging temperatures. This is done in-situ and in real time allowing the operators to accurately control their gasification process to tight tolerances never before possible. Phase I the project team successfully completed the feasibility testing of the ML-Enhanced LIBS instrument. We conducted extensive laboratory testing on several biofuels and waste coals and mixtures. We collected 64,000 spectral shots and converted them into to 12 parameters including heating value, fusion temperature, ash content and nine elements' concentration. We achieved outstanding accuracies and repeatability thus clearly demonstrating the instruments promise. In Phase II we will acquire a commercial-ready instrument (TRL of 7) and test it on a large number of coal waste and biomass while they are being moved on a conveyor belt. We will use the resulting ML algorithms and program them into the instrument. Data will be taken in realtime while the feedstocks are moving on the belt. Finally, we will use well-developed gasifier simulators and apply our instrument's data to determine the performance gains to be made. The instrument will be used on gasifiers which will provide real time data on their feedstock and allow them to better control their operation and optimize their performance. Other markets include the coal-fired power plant, Municipal Solid Waste (MSW) plant, and biofuel processing facilities.