Simulation And Experimental Study Of Biomass Anaerobic Digestion Process For Integrated Pyrolytic Gasification

Biomass has gradually become a potential substitute for fossil fuels due to its characteristics of renewable,low pollution and wide distribution.According to the different utilization technology,the field of biomass energy can be divided into thermal energy and biotechnology energy.Among them,biotechnology takes anaerobic digestion technology as the main body.Thermochemical energy production includes gasification,pyrolysis,hydrothermal liquefaction,and hydrothermal carbonization.In recent years,with the increase of investment and research focus,the above two biomass technologies have respectively achieved quite remarkable results.But at the same time,the two technologies also exposed their own bottlenecks to varying degrees.The integration of thermochemical and biochemical technologies can compensate for the shortcomings of individual technologies,but coupled systems require more complex regulation than biotechnology or thermochemical technologies alone.Therefore,a reasonable design is needed to achieve the best benefits of the coupled system.However,many integration experiments with different degrees require too long a period and face many uncertainties.In contrast,accurate coupled system modeling with visual regulation mechanism is a more effective way.Based on the above research,the traditional dynamic model ADM1 is coupled with the numerical model machine learning,and the sensitivity parameters in ADM1 are predicted by using the machine learning method with ADM1 as the main body,to form the M-ADM1model that can predict a variety of biomass raw materials.In order to realize the regulation of integration technology and explore the relevant mechanism,M-ADM1 was coupled with the T-ANN model previously formed in this study to form an anaerobic gasification integration system.Carbon emission was taken as the research index to evaluate and optimize the integration model.The specific results are as follows:In this study,75 biomass samples were used to build a machine learning model.The machine learning model input used to predict dynamic parameters includes C,H,O,N,S content and anaerobic digestion temperature.The sensitivity of 17 dynamic parameters was evaluated in this model,and the 7 dynamic parameters with the highest sensitivity were selected as the model output.ADM1 optimized by machine learning model is called M-ADM1.According to the model accuracy test,the average R² of the seven dynamic parameters predicted by the research reaches 0.92,and the root-mean-square error reaches0.167.To some extent,this indicates that the machine learning model is accurate.The simulation prediction of anaerobic digestion was carried out for municipal household waste,restaurant waste and sludge,and the relevant prediction results were compared with the experimental values.The results show that the overall M-ADM1 accuracy expressed by the Taylor inequality coefficients of MSW,kitchen waste and sludge are 0.0163,0.0327 and0.0361,respectively.The results verify the hypothesis that the machine learning model can accurately predict some key intermediate parameters to improve the performance of traditional ADM1.Then,the anaerobic digestion model M-ADM1 and gasification model T-ANN were integrated to form a set of integration models that can effectively simulate the biomass AD-GS integration technology.In this process,this study explores the changes of various intermediates and biogas residue elements in the integration process.The results showed that the accumulation rate of CH₄ increased significantly in the first 15 days,then began to decline in the 15th to 30th days,and basically maintained a balance after the 30th day.In the study of three small molecules（small sugar,LCFA,and small amino acid）in the digestion liquid during the simulation process,it was found that the contents of the three substances peaked5 days before digestion,and the content and rate of LCFA were the most obvious changes,followed by sugar and amino acid.This phenomenon was consistent with the characteristics of changes in the composition of anaerobic digestion liquid of kitchen waste.The content of C,H,and O elements in biogas residue was analyzed,and the results showed that the percentage of three elements showed a decreasing trend.In the simulation of M-ADM1,elements C,H,and O decreased with the reduction of organic compounds,as expected.But when the reaction is hydrolyzed,the H and O elements increase appropriately due to H₂O.It is worth mentioning that the overall percentage of H element in the solid residue showed a downward trend,but the range of change was not obvious.This result may be due to the hydrogen inhibition provided in the anaerobic digestion model.The analysis results of gasification products at different digestion times showed that the percentage contents of CH₄,H₂,CO,and CO₂ showed a decreasing trend.The main reason may be that the content of organic matter in kitchen waste decreased with the extension of anaerobic digestion time.In the operation of the integration model,carbon emission reduction is used as the evaluation index of the integration model in order to better explore the optimal integration point position.Specifically,in this study,the composition of biogas slurry predicted by anaerobic digestion simulation was used to prepare biogas slurry fertilizer,and the biogas residue was put into gasification simulation to produce syngas.In this process,biogas slurry fertilizer can replace industrial fertilizers,and biogas and syngas obtained from digestion can replace natural gas,thus obtaining carbon emission reduction index.This study found that when the duration of anaerobic digestion was 0-15 days,the reduction of carbon emissions increased rapidly.The reduction in carbon emissions peaked on day 15.Carbon emission reduction value is 0.1828g CO₂eq.