Study On Bio-oil Production Characteristics And Pyrolysis Process Optimization Of Mobile Biomass Pyrolysis Systems

Biomass energy is a kind of renewable carbonous energy that can realize all-round substitution of fossil resources.The collection,storage and transportation cost of biomass is too high due to its dispersed distribution.As a result,the commercial development of biomass energy utilization technology is limited.Mobile pyrolysis systems can locally convert biomass into transportable liquid fuel bio-oil.Thus the collection and transportation cost of the feedstock can be much lower.Nowadays,the mobile pyrolysis technology is immature.A number of issues of the mobile pyrolysis technology need to be settled,such as the slow processing speed,the lack of knowledge on pyrolysis characteristcs and controlling mechanism.This dissertation,using experiment and modeling methods,devotes to solve the key scientific and technology problems in mobile pyrolysis systems.These problems mainly included the pyrolysis reaction intensity,the bio-oil production characterristics of various biomass feedstocks and the process optimization of biomass pyrolysis.The effects of pyrolysis conditions on reaction intensity were studied on an intermittent feeding pyrolysis experiment device.The pyrolysis conditions include pyrolysis temperature,heating rate,biomass particle size,fluidized state and biomass type.Based on partial least-square method and Levenberg-Maquardt method,correlation models related to the pyrolysis conditions and volatile devolatilization time was obtained.The influence of the pyrolysis conditions were analyzed quantitatively.Experiments showed that the devolatilization time of fast pyrolysis is as much as 4.4%of that of slow pyrolysis with heating rate as 10 ^oC/min.Model results show that the pyrolysis temperature plays the most important role in biomass fast pyrolysis,followed by particle size and gas flowrate.The devolatilization speed of various biomass feedstock is much different.The herbaceous plant feedstock,which is abundant in hemicellulose,usually has faster devolatilization speed and releases more CO₂.While the devolatilization speed of woody plant feedstock is slower because they have more stable lignin.The woody biomass releases more CO.The bio-oil production characteristics of various biomass feedctock were investigated on a continuous feeding pyrolysis fluidized bed.18 kinds of biomass feedctock,including woody plant,herbaceous plant,straw and other common biomass,were pyrolyzed on the experiment device.The effects of pyrolysis temperature and particle size on product distribution were studied.A database of bio-oil production characteristics was established based on experimental data.Multivariate non-linear regression models were established to predict pyrolytic product distribution.The bio-oil yield from woody feedstock pyrolysis is higher than that from herbaceous plant.Pine wood has the highest bio-oil yield,which is 58.4%at500 ^oC.The bio-oil yield of tobacco stalk,wheat straw and rice straw at 500 ^oC is 42.3%,51.7%and 41.3%,respectively.The bio-oil converted from woody biomass has much levoglucosan and little acetic acid,while that from herbaceous plant has much acetic acid and little levoglucosan.The correlation coefficient of bio-oil yield model,char yield model and gas yield model is 0.74,0.88 and 0.91,respectively.The accuracy of the regression models are advanced in comparison with other regression models.Based on the bio-oil production database,the correlation model of biomass pyrolysis with higher accuracy can be obtained using some machine learning algorithms.Support vector machine?SVM?algorithm and artificial neural network?ANN?algorithm were introduced to investigate the deep correlation between the feedstock composition,pyrolysis condition and product distribution.The correlation model was established,and the pyrolytic products were predicted based on this model.Optimized operating conditions and controlling mechanism were also obtained.It can be concluded that the predtcion performance of SVM algorithm can be better than that of ANN algorithm.The coefficient of determination?R²?of the SVM model for bio-oil yield prediction based on literature data can achieve 0.898,while that based on experimental data can be up to 0.912.A prediction model with higher accuracy can be obtained by adujusting weight coefficients.R² of the modified prediction model achieved0.963 finally.Then the support vector set matrix and its coefficient matrix were given.The operating conditions for each specific biomass feedstock were optimized,and the best bio-oil yield was predicted.The results show that the best pyrolysis temperature of woody biomass is higher than that of herbaceous plant.The process optimization of mobile biomass pyrolysis plant was investigated.The system modeling and techno-economic analysis of the mobile pyrolysis systems was conducted.The system model was built using Aspen Plus software.Different biomass-based liquid fuel production systems,including mobile pyrolysis system,were evaluated.The investment cost,operating cost and income were analyzed under the rural conditions in China.Simulated bio-oil yield from corn-cob pyrolysis is 55.6%,while the bio-char yield,gas yield and pyrolysis heat is 25.3%,19.1%and 527.5 kJ/kg,respectively.The simulation results are similar to the experimental results in respect of product distribution and energy consume.It can be concluded that the Fischer-Tropsch liquids production via biomass gasification?BG-FT?plant will be defective under low oil price situation.The bio-oil production cost of mobile systems can be only as much as 54.5%of that of fixed systems.Whether mobile systems or fixed systems,the rise of collection cost in growing season has influenced the total operating cost significantly.