(498f) Heat Transfer Measurement of Oil-Based Copper Nanofluids; A Different Approach to Data Analysis Using Self-Organizing Feature Maps (SOFM)

Large amounts of experimental data can be correlated via regression analysis yet even when plotted the results can be confounding. Self-organizing feature maps (SOFM) provide a powerful method of classifying multidimensional data into 2d or 3d plots which might visually clarify meaning.  This is one in a series of papers that describes the enhancement of  heat transfer characteristics of various base oils when infused with nanoparticles. In this study, nanofluids were prepared by dispersing three different sizes and amounts of spherical Cu nanoparticles into poly-alpha-olefin, (PAO), formulated motor oil. Heat transfer coefficients of base/nanofluids were measured using a heat transfer test-rig built at the University of Louisville. Analysis of the results using dimensionless numbers was illuminating yet did not provide all of the answers. A seven-dimensional set of 1519 data points (217 x 7), was submitted as input to a mapping program. The results give a surprising visual prescription for optimization of heat transfer enhancement: small particle size, low operating temperatures and low to medium Reynolds, Prandtl, and Peclet numbers. SOFM provide yet another analytical tool for the discriminating scientist/engineer.