Biofuels production from pine needles via pyrolysis: Process parameters modeling and optimization through combined RSM and ANN based approach. (15th February 2022)

• Record Type: Journal Article
• Authors: Gupta, Shubhi; Patel, Pushpraj; Mondal, Prasenjit
• Is Part Of: Fuel. Volume 310:Part A(2022)
• Publication Date: 2022-02-15
• Subjects: RSM Response surface methodology -- ANN Artificial neural network -- FCCD Face central composite design -- ANOVA Analysis of variance -- MLP Multilayer perceptron feed-forward network -- LP Levenberg-Marquardt back propagation algorithm -- FC Fixed carbon content (wt. %) -- VM Volatile matter content (wt. %) -- GC–MS Gas chromatography and mass spectroscopy analysis -- FTIR Fourier transform infrared analysis -- XRD X-ray diffraction analysis -- SEM Scanning electron microscopy analysis -- BET Brunauer-Emmett-Teller analysis -- HHV Higher heating value (MJ/kg) -- MSE Mean squared error -- R2 Regression factor -- CV Coefficient of variation -- X Symbolizes independent variables -- Y Symbolizes response variables -- Yiexperimental Experimentally predicted response values -- Yipredicted Model predicted response values -- Yiaverageexperimental Average of the experimental values; Pine needles -- Pyrolysis -- Optimization -- Artificial neural network -- Response surface methodology; Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2021.122230
• Languages: English
• ISSNs: 0016-2361

Abstract

Graphical abstract: Highlights: Combined RSM and ANN based approach successfully modeled the pyrolysis process. Results showed temperature as the most significant factor affecting the process. ANN demonstrated better process prediction and generalization capability. RSM accurately predicted interaction and process parameter's significance. RSM and ANN predicted maximum bio-oil yield as 51.11% and 51.70%, respectively. Abstract: The current study determined the pyrolysis potential of pine needles with the aim to assess the influence of process parameters (namely temperature, heating rate and inert gas flow rate) along with their modeling and optimization through combination of response surface methodology (RSM) and artificial neural network (ANN) technique. Pyrolysis process output was predicted by ANN, while interaction and significance/insignificance along with optimization of process parameters was determined by RSM. A combined approach was employed to accomplish the individual limitations of both the modeling methods. R 2 close to 1 and low error demonstrates the viability of the developed models. Results showed comparatively higher R 2 and lower MSE value for ANN model, thereby elucidating superior capability of ANN for predicting process yield over RSM modeling; while RSM accurately predicted the process parameters interaction and significance. The study revealed that such a combinational approach has the better ability to model the pine needle pyrolysis process compared to the individual ones. Temperature had been determined as the most predominant variable influencing the yield of the products. Optimized condition had been predicted at 552.06 °C temperature, 50 °C/min heating rate and 164.40 mL/min inert flow rate by yielding maximum bio-oil as 51.11 and 51.70% from RSM and ANN modeling, respectively. Detailed characterization advocates high end-use of all the three products. GC–MS and FTIR techniques predicted that bio-oil was composed of different organic compounds and can be utilized in place of hydrocarbon fuels after certain upgradation processes and can also be extracted into various chemicals. Characterization of biochar and non-condensable gases also demonstrate their potential application as fuel and in several other fields. The study revealed that pine needles can be effectively employed for the production of bioenergy precursors.