Applications of artificial neural networks in prediction of performance, emission and combustion characteristics of variable compression ratio engine fuelled with waste cooking oil biodiesel

Abstract

The intention of this study is to predict the performance, emission and combustion characteristics of a single-cylinder, four-stroke variable compression ratio engine fuelled with waste cooking oil methyl ester and its blends-standard diesel with the aid of artificial neural network (ANN). The tests were performed with fuel blends of 20, 40, 60 and 80 % biodiesel with standard diesel, with an engine speed of 1,500 rpm and at compression ratios of 18:1, 19:1, 20:1, 21:1 and 22:1 under different loading conditions. Three different ANN models based on standard feed-forward back-propagation algorithm have been developed to predict the performance, emission and combustion characteristics of VCR engine. To train the network, compression ratio, blend percentage and percentage load were used as input parameters whereas engine parameters such as brake thermal efficiency, specific fuel consumption, brake power, indicated mean effective pressure, mechanical efficiency and exhaust gas temperature were used as output parameters for the performance model and exhaust emissions such as carbon dioxide, carbon monoxide, hydrocarbon and NO x were used as output parameters for emission model. Separate model is developed for combustion characteristics in which compression ratio, blend percentage, load percentage and crank angle were used as the input parameters whereas combustion pressure, heat release rate, ignition delay, combustion duration and mass fraction burnt were used as the output parameters. This study shows that there is a good correlation between the ANN-predicted values and the experimental data for different engine performance, emission parameters and combustion characteristics. Keywords Biodiesel Á Variable compression ratio engine Á Performance Á Combustion Á Emission Á Artificial neural networks List of symbols ANN Artificial neural networks B20 20 % Biodiesel ? 80 % Diesel B40 40 % Biodiesel ? 60 % Diesel B60 60 % Biodiesel ? 40 % Diesel B80 80 % Biodiesel ? 20 % Diesel BP Brake power BTE Brake thermal efficiency CA Crank angle CO Carbon monoxide CO 2 Carbon dioxide CNG Compressed natural gas CR Compression ratio HC Hydro carbon IMEP Indicated mean effective Pressure LM Levenberg-Marquardt (algorithm) MLP Multi-layer perception MSE Mean square error NO x Nitrogen oxides O 2

Experimental investigation of the cetane improver and bioethanol addition for the use of waste cooking oil biodiesel as an alternative fuel in diesel engines

“…Besides, waste cooking oil (WCO) significantly reduces the cost of production of biodiesel since it is cheaper than the raw materials mentioned above. In addition, the fact the use of WCO has no negative impact on food safety increases the importance of using it as a raw material in biodiesel production [5][6][7].…”

Application of Artificial Neural Network for Internal Combustion Engines: A State of the Art Review

“…Some authors initially present here, used the MISO models. Muralidharan et al [ 82 ] analyzed and developed a model to forecast the emission, performance, and combustion parameters of the engine using waste cooking oil biodiesel. Three different ANN models with back-propagation algorithms utilized to predict parameters separately.…”

Impact of tamanu oil-diesel blend on combustion, performance and emissions of diesel engine and its prediction methodology

“…This paper exploits the concept of neural network to predict the engine characteristics because neural network has been already proved for its successful prediction performance of different fuel resources [45] and engine models [46]. The neural network has been trained using Levenberg-Marquardt algorithm.…”