Umeå universitets logga

An engine oil has several important properties it must fulfil in order to provide efficient engine operation. In addition to damping friction by lubrication, an engine oil must be able to cool, protect against corrosion, dampen vibrations and clean surfaces from contamination, wear and residual products. During the lifetime of engine oil, it is subject to degradation processes which impair its properties. In order to predict oil degradation and oil changes more accurately, there is a need to improve the tools and methods used to study the characteristics and chemistry of oil. The thesis investigates data and statistical models relevant to describing the nature of engine oil and its degradation, and also discusses the use cases of statistical models in process control. The results are statistically derived and confirm statements found in the literature review. Factor analysis was used to find underlying concepts such as wear materials, additive groups and the relation between different lubricant analysis methods. Principal component analysis was used to cluster oil types and conclude that there is a difference in oxidation rate between the engine oil in buses and trucks. Multilayer perceptron was used to create an oxidation model with an R-squared value of 0.7.