Hydrogen generation from agriculture residues using the supercritical water gasification process
Date
2024-09-27
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
Type
Thesis
Degree Level
Doctoral
Abstract
Increased dependency on fossil fuel has led to surge in greenhouse gas emission from the combustion of fossil fuels. To replace fossil fuels, a better alternate clean source of energy such as biomass is required. Hydrogen is typically produced from natural gas via steam reforming of methane. A sustainable production of energy and hydrogen from biomass can reduce our dependency on natural gas for our energy requirements. Supercritical Water Gasification (SCWG) is a promising thermochemical conversion technology for the production of hydrogen from biomass resources. SCWG uses water at supercritical conditions (Temperature > 374.1 oC and Pressure > 22.1 MPa) to convert biomass materials into hydrogen-rich syngas. The ability to process high moisture content feedstocks without drying is one of the advantages of SCWG. The process is suitable for several feedstocks including lignocellulosic biomasses, algae, crude glycerol, and waste cooking oil. Most of the studies have investigated the SCWG of model compounds. However, literature is scarce on the SCWG of a mixture of different model compounds and the comparative studies between SCWG of different categories of real feedstocks.
In phase one, model compounds of crude glycerol, such as pure glycerol, methanol, and oleic acid were gasified in supercritical water (SCW) at varied feedstock mixture ratio. Objective of this phase was to understand the interactive effect of model compounds of a crude glycerol during gasification and develop detailed reaction mechnism of degradation of complex feedstock such as crude glycerol. Response surface methodology (RSM) using 2,3 simple lattice design of experiment was used to develop 3-D response curves and predictive models to predict yields of H2, CH4, CO2 and total gases from gasification of crude glycerol at different compositions. Results showed that the methanol demonstrated highest H2 yield and total gas yield of 7.7 and 8.6 mmol/g followed by pure glycerol and oleic acid. RSM plot and predictive equation entailed the interactive effects of composition of model compounds on gas yields. Furthermore, comparison of predictive gas yields with experimental gas yields of simulated crude glycerol confirmed the accuracy of developed equations.
Results from phase one provided valuable insights in degradation mechanism of a heterogeneous feedstock such as crude glycerol and interactive effects of its various components on gas yields. This laid the foundation for understanding the complex reaction mechanism of a real feedstock such as agriculture biomass. In phase two, different residual parts of an agriculture crop such as canola straw, canola meal, and canola hull, having different chemical composition were gasified and their gas yields were compared. Results showed that the canola straw had highest H2 yield of 7.07 mmol/g due to its highest cellulose and hemicellulose content among all canola residues. Canola straw was selected as the most suitable feed and its gasification was further studied at varying SCWG reaction conditions (350-500 ℃, 20-60 min, 10-25 wt.%). Increase in reaction temperature and time improved the hydrogen yield, while decrement in feedstock concentration increased the hydrogen yield due to enhancement in reforming, hydrolysis, and water-gas shift reactions. Highest hydrogen yield of 8.1 mmol/g was achieved with canola straw at optimized reaction conditions of 500 ℃, 23-25 MPa, reaction time of 60 min, and feed concentration of 10 wt.%.
Despite higher gas yields of canola straw at optimized reaction conditions, complete gasification of canola straw was not achieved. This results in valuable biomass resource not being fully utilized. Thus, in phase three various supported and promoted novel nickel based catalysts were systematically screened and compared for SCWG of canola straw at previously optimized reaction condition in phase two. Different supports such as hydrochar obtained from hydrothermal liquefaction (HTL-HC) and hydrothermal carbonization (HTC-HC), alumina (Al2O3), zirconia (ZrO2), activated carbon (AC), and carbon nano tube (CNT) were compared at constant nickel metal loading of 10%. Results showed that the Ni/ZrO2 showed highest hydrogen yield of 10.5 mmol/g due to its amphoteric nature which favoured the reforming and water gas shift reactions, resulting in higher hydrogen yield. Ni/ZrO2 catalyst was further modified with potassium (K), zinc (Zn), and cerium (Ce) promotors. Comparison of promotors showed that the Ce was most effective promotor with its highest hydrogen yield of 12.9 mmol/g due to its ability to improve the metal dispersion, and minimization of coke deposition and metal sintering.
Objective of phase four was to develop different supervised machine learning models to predict the individual gas yields (H2, CH4 CO, CO2 gas yields) of SCWG of lignocellulosic biomass. SCWG reaction conditions and biomass characteristics such as proximate analysis and ultimate analysis were used as input features to develop machine learning models. Eight different optimized machine learning models such as ((linear regression (LR), Gaussian process regression (GPR), artificial neural network (ANN), support vector machine (SVM), decision tree (DT), random forest (RF), extreme gradient boosting (XGB), and categorical boosting regressor (CatBoost)), with particle swarm optimization (PSO) and genetic algorithm (GA) optimizer for each gas yield were developed and compared for prediction of gas yields. PSO optimized XGB model demonstrated highest test R2 of 0.84 for prediction of H2 yield, while for prediction of CH4, CO, and CO2 gas yields, PSO optimized CatBoost showed highest test R2 of 0.83, 0.86, and 0.92, respectively. Feature analysis revealed that for hydrogen yield, reaction temperature was most dominating feature. Shapley additive explanation (SHAP) analysis also showed interactive behaviour of input features on prediction of gas yields highlighting complex nature of SCWG process.
In final phase five, a detailed technoeconomic analysis (TEA) and life cycle assessment (LCA) of a conceptual solar driven SCWG pilot plant was conducted for processing of 200 metric tons/day of canola straw to produce hydrogen. Economic analysis based on discounted cash flow estimated the minimum selling price (MSP) of hydrogen to be USD 3.38 /kg of hydrogen, which was comparable to non-renewable processes and lower than renewable process. Furthermore, SCWG plant had undiscounted net present value (NPV) of USD 548 M and internal rate of return (IRR) was 38.87%, showing the profitability of the process. Life cycle assessment showed that the SCWG of canola straw had global warming potential (GWP) of 1.91 kg CO2 eq./kg H2. This is significantly lower compared to non-renewable processes and comparable to renewable processes. However, if accounted for biogenic CO2 emission allocation by canola straw, GWP further decreased to 1.15 kg CO2 eq./kg H2.
Description
Keywords
Hydrogen, supercritical water, machine learning, biofuel, biomass
Citation
Degree
Doctor of Philosophy (Ph.D.)
Department
Chemical and Biological Engineering
Program
Chemical Engineering