MOSCOW, 8 Aug – . A method for improving the production of biodiesel from the seeds of a poisonous tropical shrub was developed by Russian scientists from DSTU and FNAC VIM as part of a scientific group. According to them, the Barbados nut (Jatropha curcas) processing method developed by training a deep neural network will reduce labor costs and improve the quality of the resulting biofuel. The results are published in Sustainability.
Biodiesel is a long-chain fatty acid methyl ester derived from vegetable oil and animal fats. Experts noted that compared to petroleum-based diesel fuel, it has several advantages. The use of such fuel leads to a reduction in emissions of exhaust gases, particulate matter, unburned hydrocarbons. Biodiesel has improved biodegradability and is non-toxic. It can be used in any diesel engine in pure form or mixtures.
According to experts, the main problems in the production of biodiesel are high viscosity, poor volatility and polyunsaturation (the presence of a hydrocarbon chain with two or more carbon-carbon double bonds). To solve them, scientists use various chemical processes, one of which is interesterification.
This is a safe method of changing the physical and chemical properties of esters (fats, oils) by exchanging acidic residues between fats . Widely used by manufacturers of specialized fats in Europe and the former USSR.
< /span>
Scientists from the Don State Technical University (DSTU) and the Federal Scientific Agroengineering Center VIM (FNAC VIM) as part of an international scientific group used deep machine learning to optimize the parameters of the transesterification process of vegetable oils obtained from the seeds of the Barbados nut and predict the characteristics of the produced biodiesel fuel. According to them, this will improve the quality of the resulting fuel, the accuracy of forecasting its volume and reduce the labor costs of the production process.
«
«As a result of the work done, the most important parameters of the transesterification process were classified in terms of their maximum impact on the properties of biodiesel fuel», — said the head of the research laboratory «Intelligent electrical complexes of agricultural machines» DSTU, Vadim Bolshev, Senior Researcher of the Laboratory of Electricity and Heat Supply of the Federal National Research Center of VIM Vadim Bolshev. -crop-ratio=»0.667535853976532″ data-crop-width=»600″ data-crop-height=»401″ data-source-sid=»not_rian_photo» class=»lazyload» width=»1920″ height=»1282″ decoding=»async» />
He added that a positive correlation was found between temperature and process time and a negative correlation between methanol content and catalyst concentration.
«The highest possible ester yield (97%) was achieved using a reaction temperature of 60° C and concentration of potassium hydroxide catalyst in 1% by weight. In addition, it was determined that the concentration of methanol in the ratio of 6:1 allows you to get the maximum volume of biodiesel,» said Bolshev.
The researchers noted that comparing the quality metrics of the resulting model with other machine learning models showed the superiority of their proposed design. According to scientists, the accuracy of predicting the volume and viscosity of the resulting biofuel has been improved by 10% and 8%, respectively.
According to scientists, the study is particularly relevant for developing countries in Africa and South America, where the Barbados nut grows as a weed. It began to be grown commercially in India and China after experts pointed to this plant as one of the best candidates for future biodiesel production.
The study was conducted by scientists from the Don State Technical University, Federal scientific agroengineering center VIM, the Russian University of Transport and the University of Oil and Energy in Dehradun (India).
The international scientific team plans to continue working to improve the quality of forecasting of the developed deep learning model due to more training data, and also training the model on another type of raw material. In addition, there is a need to conduct a feasibility study, life cycle assessment, energy and exergy assessment (thermal efficiency assessment method. — Ed.) of the application of the model in the production of biodiesel fuel.
DSTU is a participant in the state support program for universities in the Russian Federation «Priority 2030».