Research on biomass pyrolysis characteristics and yield prediction model
Received Date:2024-09-12
Revised Date:2024-11-04
Accepted Date:2024-11-06
DOI:10.20078/j.eep.20241103
Abstract:The thermal gravimetric and other physicochemical properties of bulrush and red cedar were analyzed. Fast pyrolysis e... Open+
Abstract:The thermal gravimetric and other physicochemical properties of bulrush and red cedar were analyzed. Fast pyrolysis experiments were conducted using a horizontal fixed-bed reactor. The study explored changes in the yields of solid, liquid, and gas products from biomass feedstocks under different pyrolysis temperatures, pyrolysis times, and biomass particle sizes. A predictive model based on neural networks was developed to rapidly and accurately predict the yield distribution of the three phases under various reaction conditions. This model incorporates data on the yield of solid, liquid, and gas products from fast pyrolysis of different types of biomass reported in previous studies, facilitating predictions of pyrolysis product yields for various biomass types under different reaction conditions. The input features of the model are biomass characteristics (volatile matter V, fixed carbon FC) and pyrolysis conditions (pyrolysis temperature T, pyrolysis time t, and particle size D), while the output feature is the yield of three-phase products. The test set and predicted data were in good agreement. The average regression coefficient R 2 of the three-phase products in the test set was 0.97, and the predicted results had an error of only 5.31% compared to the experimental results, indicating the feasibility and reliability of the model for predicting the yield of pyrolysis products. Close-
Authors:
- ZHANG Tao1
- GAO Huanting2
- GONG Xun2,*
- PAN Haoxiang1
Units
- 1. Shanghai Power Equipment Research Institute Co., Ltd.
- 2. National Key Laboratory of Coal Combustion and Low Carbon Utilization, Huazhong University of Science and Technology
Keywords
- Biomass
- pyrolysis
- Threephase products
- Neural network
- Prediction model
Citation
Relate Article