A novel method to evaluate nanofluid stability using multivariate image analysis

Abstract

A multivariate image analysis (MIA)-based approach in conjunction with chemometrics is proposed to evaluate the stability of nanofluids prepared by dispersing cobalt ferrite nanoparticles in mineral insulating oil with vigorous mechanical stirring (20.000 rpm, Ultra-Turrax). Three different magnetic fluids were evaluated: (a) oleic acid-coated magnetic fluid (OAMF) at 0.00001% (m/v), (b) stearic acid-coated magnetic fluid (SAMF) at 0.01% (m/v), and (c) non-coated magnetic fluid (NCMF) at 1% (m/v). Magnetic nanoparticles as powders or dispersed in oil were characterized using XRD, FTIR spectroscopy, Mössbauer spectroscopy, and DLS. Glass test-tubes were filled with magnetic fluid and digital images were recorded during 67 days for OAMF, 20 days for SAMF, and 90 min for NCMF. According to the principal component analyses of the acquired digital images, OAMF remained stable during 39 days. On the other hand, the less stable fluids, SAMF and NCMF, showed a drastic reduction in their sedimentation rates after 10 days and 26 min, respectively. Multivariate regression methods (MLR, PCR, and PLS) combined with a genetic algorithm (GA-MLR, GA-PCR, and GA-PLS) were also employed in order to estimate the NCMF sedimentation times and cobalt ferrite concentrations in OAMF. GA-PLS provided the best sedimentation time estimates and PCR showed better performance when estimating the cobalt ferrite nanoparticle concentrations. As a result, the proposed method is efficient, fast, non-destructive, low-cost, accurate, and can be employed from low to high concentrated nanofluids, even when they are dark in colour.