Estimation of liquid viscosities of oils using associative neural networks

Dynamic viscosities of a number of vegetable oils (castor oil, palm oil, sunflower oil and coconut oil) and lubricant oils (2T and 4T) have been determined at temperature range 30 degree -90 degree C using Ubbelohde viscometer. An associative neural network is used to compute the viscosities of oils for unknown temperatures after training the neural network with type of oil, temperature as input and viscosity as output. Predicted results agree well with the experimental results. Simplified and modified form of Andrade equations that describe the temperature dependence of dynamic viscosities are fitted to the experimental data and correlations for the best fit are presented. The results obtained from associative neural network and best correlation equation show that both predict the viscosities very well with correlation coefficient R(2)= 0.99.