ISSN 2630-0583 (Print)

ISSN 2630-0656 (Online)

JCST

Journal of Current Science and Technology

http://jcst.rsu.ac.th

Journal of Current Science and Technology. Vol.8 No.2 , July - December 2018.

Downdraft gasifier identification via neural networks

Arpakorn Wattana, Sarawut Janpong, and Yannavut Supichayanggoon

Abstract

          This research presents the identification of producer gas resulting from the conversion of a given type of biomass in a downdraft gasifier, the use of a neural network (NN) to predict the identity for a given biomass type and  the comparison of the NN prediction to measured results of biomass fuel conversions.  Each type of biomass has different characteristics which affect the composition of the producer gas and thus its effective energy content.  This research predicts the composition of the producer gas from the characteristics of the biomass by creating a mathematical model using a neural network.  The model is then used to run simulations which are compared to actual measured values from experiments and then the accuracy of the simulations are verified with Simulink/MATLAB.  The results show that the simulation predicts the CO content of producer gas with an average error of 1.73%, 7.01% for H2, and 1.58% for CH4.  The simulation predicts the higher heating value with an average error of 0.73% and a lower heating value with an average error of 0.81%.

Keywords: downdraft gasifier, identification, neural network, water scrubber

Download Full Paper.

DOI: 10.14456/jcst.2018.10