Correlation of field, laboratory and standard lubricant tests using multivariate analysis

Abstract

Chemometric techniques were used to analyse engine lubricant test results in order to correlate activities within engine test procedures. Principal component analysis (PCA), partial least squares projection to latent structures (PLS) and cluster analysis were used to extract information from the data set. Analyses of the results by PCA revealed trends and outliers within the observations and showed relationships between observations and variables and among the variables themselves. The PLS analysis showed how each oil sample varied with time and the acquired prediction plot showed the extent of degradation of the oil samples, which was then used to rank them. A dendogram obtained from the cluster analysis was used to show clearly which oil samples were ranked together. The multivariate analysis approach was useful in extracting significant information from engine laboratory test data and, in parallel, engine lubricant field tests. Chemometric analyses have demonstrated a scaled relationship between a standard engine lubricant test, extreme sampling direct from the piston ring zone and also field tests.