Prediction of biodegradability of mineral base oils from chemical composition using artificial neural networks

Abstract Biodegradability studies of base oils are important for designing and development of environment-friendly lubricants. Biodegradability of base oils and lubricants have been determined by a large number of test methods. Among these, the 21 day test developed by Coordinating European Council and designated as CEC-L-33-A-93 has been accepted worldwide. In this work, artificial neural network (ANN) technique has been used to construct mathematical models for predicting biodegradability of base oils based upon their chemical composition, viscosity and viscosity index. The chemical composition has been determined either by NMR or mass spectrometry and two models have been developed. Thirty-one base oils of different origin and processing schemes, and their blends with polyalphaolefin (PAO) were analyzed for chemical composition and biodegradability. Part of the data was used for developing the models using backpropagation ANN, while the remaining data were used for evaluating the predictive ability of the models (correlation coefficient, R 2 ∼0.97). The models can serve as useful tools for screening base oils before subjecting them to 21 day biodegradability test.