Abstract
One of the major factors affecting the performance, emissions and life of a CI engine is the quality of fuel used. The cetane number is a direct measure of the quality of any fuel. Optimum value of cetane numbers lead to decreased knocking and improved combustion characteristics resulting in better performance, lower emissions, lesser noise and lower vibration levels thereby improving the life of engine components. Hence the cetane number is of immense importance which controls the suitability of any fuel. The usual method for finding cetane number involves a tedious process.
In order to make the work simpler, this paper computes cetane number based on in-cylinder parameters using Artificial Neural Networks (ANNs). ANNs can be used for prediction purposes by training them with a series of known data. A multilayer perceptron backpropagation neural network model is employed. The in-cylinder parameters like peak pressure, ignition delay, crank angle at which peak pressure occurs etc. are given as inputs to the first ANN and the corresponding cetane number constitutes its output. This network, with known in-cylinder parameters is trained to compute the cetane number which is fed as an input along with in-cylinder parameters to the second network based on which it provides the engine performance characteristics like Brake Thermal Efficiency (BTE) and Specific Fuel Consumption (SFC), emissions of Hydrocarbons (HC) and NOX as output. Performance tests have been carried out on a series of fuels (bio-diesels derived from different sources blended in various proportions with diesel) with known cetane numbers and in-cylinder parameters, performance and emissions are recorded. The above data is used to train the above networks and the networks are validated with another series of test data. Thereby, this paper attempts to first find the optimum cetane number by correlating it with in-cylinder parameters and then predicts the performance and emission characteristics based upon the in-cylinder parameters and the obtained cetane number. Thus, by knowing the in-cylinder parameters alone, this paper not only computes the cetane number of the fuel used but also estimates the performance and emission of the CI engine. In addition, this paper makes a study by comparing the performance and emission characteristics obtained from the previous setup (of the above two networks) with a new network which directly correlates in-cylinder parameters with performance and emission characteristics.
Thus this paper provides a comprehensive study of the correlations that exist between incylinder parameters, cetane number, engine performance and emission characteristics by establishing artificial neural networks and can be used to aid in the selection of fuels with optimum cetane numbers for best performance and emission characteristics of a CI engine.
Keywords: CI engine, Cetane number, In-cylinder parameters, Performance and Emission characteristics, Artificial Neural Networks