Specific Fuel Consumption Prediction Model for Diesel Engines: A Preliminary Study

Abstract

Specific Fuel Consumption (SFC) is an indicator to measure the performance of power plants. Its value must be monitored and also predicted so preventive actions and maintenance can be formulated accordingly. To develop SFC prediction model, it is important to evaluate which factors are influential toward SFC values, how data collections are made, and which models are the best to use to predict SFC values with high accuracy. This paper provides a preliminary study to support the development of SFC prediction model. The results show that engine loading and type of fuel are the two major factors affecting the SFC values. Data collection to calculate SFC can be obtained through either controlled experiments or direct observations in power plants, each has its own pros and cons. SFC modeling can be done using regressions (linear, polynomial or SVR) and artificial neural network (ANN). Each method can be applied to get the modeling that produces the highest accuracy. However, the accuracy is also highly dependent on the validity of the input data.