Biodiesel production from oil-rich feedstock: A neural network modeling. Issue 6 (19th March 2018)

• Record Type: Journal Article
• Authors: Deng, Chao; Gong, Shu; Gao, Wei
• Is Part Of: Energy sources. Volume 40:Issue 6(2018)
• Page Start: 638
• Page End: 644
• Publication Date: 2018-03-19
• Subjects: Accurate modeling -- biodiesel -- castor oil -- fatty acid methyl ester -- LSSVM-GA -- oil-rich feedstock; Natural resources -- Periodicals; Energy consumption -- Periodicals; Energy consumption -- Climatic factors -- Periodicals; Energy conversion -- Periodicals; Energy conversion -- Environment aspects -- Periodicals; Power (Mechanics) -- Periodicals; 333.7905
• DOI: 10.1080/15567036.2018.1454544
• Languages: English
• ISSNs: 1556-7036

Abstract

ABSTRACT: Biodiesel produced from oil-rich feedstocks is known as a green replacement for conventional petroleum diesel. Transesterification is the common method used for biodiesel production. Hence, in this contribution, neural network modeling and least square support vector machine (LSSVM) modeling were used to predict the transesterification of castor oil with methanol to form biodiesel. Also, genetic algorithm was used for the optimization of predictive model. Input and output parameter of predictive models for the prediction of biodiesel production yield and estimation of the efficiency of biodiesel production are catalyst weight (C), methanol-to-oil molar ratio (MOR), time (S), temperature (T), and fatty acid methyl ester (FAME) yield, respectively. Proposed LSSVM modeling predicts biodiesel production yield or FAME yield within ±2% relative deviation with a high value of coefficient of determination (0.99583) and a low value of absolute deviation (1.27) in which the mentioned statistical parameters represent the accuracy and robustness of the model.