Optimum parameter design for performance of methanol steam reformer combining Taguchi method with artificial neural network and genetic algorithm. (1st November 2017)

• Record Type: Journal Article
• Authors: Ouyang, Kwan; Wu, Horng-Wen; Huang, Shun-Chieh; Wu, Sheng-Ju
• Is Part Of: Energy. Volume 138(2017)
• Page Start: 446
• Page End: 458
• Publication Date: 2017-11-01
• Subjects: Methanol steam reformer -- Energy carrier -- Optimum parameter design -- Taguchi method -- Radial basis function neural network -- Genetic algorithm; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2017.07.067
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: The fuel cell is powered by H2 widely provided by reforming processes. A promising reforming process is methanol steam reforming which has received much attention. This study then attempts to acquire high hydrogen concentration, high methanol conversion efficiency and low CO concentration of methanol steam reforming. Three operating parameters were investigated: reacting temperature ( T = 220–280 ° C), steam-to-carbonate ratio ( S / C = 0.9 to 1.1), and the volume flow rate for nitrogen (N2 ) carrier gas ( Q = 40 to 100 cm 3 / min ) as the flow rate of aqueous methanol solution was set as 3.1 cm 3 /min. The integrated approach of combining the Taguchi method with radial basis function neural network (RBFNN) was proposed in this study to demand an optimum parameter design. The results showed that the optimum parameter design was: T = 267 ° C, S / C = 1.1, and Q = 40 cm 3 / min . The averaged percentage reduction of quality loss (PRQL) of 3.31% was obtained as optimum condition was implemented, in comparison with the starting condition (the largest reacting temperature, steam-to-carbonate ratio, and N2 volume flow rate). In addition, principal component analysis (PCA) is also investigated. The results obtained by PCA were compared with the ones by the integrated approach. Highlights: This work presents an optimum parameter design on the multi-objectives of methanol steam reforming. The optimum parameters acquire high H2 concentration, CH3 OH conversion efficiency and low CO concentration. We combine the Taguchi method with radial basis function neural network to determine optimum parameters. It can save 67% time and cost taken to perform experiments employing the Taguchi's L 9 orthogonal array. The optimum parameter design can reduce 3.31% of quality loss compared with the starting condition.