|Oil analysis: an ensemble approach
|Big Data Chemometrics and Method Development(In-Silico)(KVCV)
|Mr Tom Hancock
|Dr Paul, J Gates
Dr Christopher, J Arthur
Dr Samuel Whitmarsh
Dr Christianne Wicking
Mr Samuel Ellick
Abstract Information :
Crude oil presents an analytical challenge as it consists of 100,000's of chemically similar compounds. No single analytical method is able to completely characterise the composition of such a complex matrix. Emerging trends in the life sciences attempt to address a similar challenge by linking multiple analytical methods to determine pharmacological properties of complex biological samples (for instance for disease diagnosis). This study demonstrates the application of a similar approach to link the spectroscopic and spectrometric analyses of oil to its physicochemical properties.
Nine crude oils selected from a broad range of global oil fields were analysed by infrared spectroscopy (IR), nuclear magnetic resonance (NMR), and mass spectrometry (MS). Data was combined and manipulated using open access tools such as KNIME and various python modules. The resultant dataset allows the classification of the oils by simple, easy-to-measure properties such as location of origin, specific gravity and sulfur content. These findings reinforce the notion that the analysis of complex samples is more appropriately addressed using an ensemble analytical approach and provides a framework to bring different datastreams together in a coherent manner.
In future work the classification of other physical properties will be explored. The development of such techniques could allow for prediction of refined product properties from different crude oils as well as other industrially relevant properties. In addition the approach described here will be used to develop methodology to characterise refined oils, such as base oils and formulated products.