Classification of Oils by the Application of Pattern Recognition Techniques to Infrared Spectra.
Abstract
The classification of multicomponent petroleum oils (crude oils, lubricants, distillate and residual fuels) solely by their infrared absorption spectra is a difficult task. Crude oils alone include a phenomenal variety of systems, from heavy asphaltic crudes to light crudes that are similar to a No. 2 fuel oil. Furthermore, the distinctions between classes of fuel oils (i.e., Nos. 1, 2, 4, 5 and 6 fuels) are based upon ASTM specifications for continuous properties such as flash point and viscosity. In South Florida, for example, local fuel oil suppliers meet requests for Nos. 4 or 5 fuel oils by blending together appropriate proportions of Nos. 2 and 6 fuels. In order to reduce the amount of sampling required in the event of an oil pollution incident, it would be useful to be able to initially classify the pollution sample into one of the above groups. Infrared spectroscopy has been promoted as a useful analytical technique for oil classification and identifications, since it does provide some information on the aliphatic, aromatic, polynuclear aromatic, carbonyl, and organosulfur composition of an oil. Infrared spectra have been used in previous efforts to distinguish asphalts from residual fuels, and to provide a tool for 'fingerprinting' oils. Kawahara et al. applied linear discriminant function analysis (LDFA) to their infrared data to make the binary distinction between asphalts and residual fuels. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1976
- Accession Number
- ADA039387
Entities
People
- James S. Mattson
Organizations
- Rosenstiel School of Marine, Atmospheric, and Earth Science