Optimization through Taguchi and artificial neural networks on thermal performance of a radiator using graphene based coolant

Abstract

As a typical coolant in a radiator, a combination of water and ethylene glycol is often used. Since it has a lower thermal conductivity, the addition of nano particles can increase the performance of the coolant which aids in dwindling the weight and size of the radiator. This paper deals with effect of various input variables like flow rate, inlet temperature of the coolant and Vol% of nano particles (NP) on a radiator’s heat transfer parameters like heat transfer rate (Q), convective heat transfer co-efficient (h), Reynolds number (Re), Nusselt number (Nu) and friction factor (ff) were estimated using optimization methods through Taguchi, ANOVA and ANN. Initially Experimental trials were carried to evaluate heat transfer parameters of a radiator under forced convection by changing inlet temperature, flow rate and NP addition of a coolant. Base fluid water and EG were taken in the ratio of 70:30. To that, Nano particles of graphene were dispersed in the base fluid in the range of 0.1–0.3 Vol%. Coolant inlet temperature, flow rate and addition of nano particles were considered as input parameters (I/P). TheL27 orthogonal array was used as Design of Experiments (DOE). Q, h, Nu, Re and ffwere selected as performance variables. In addition to that, an ANOVA test through Minitab 15 was employed to evaluate the performance parameters in order to estimate each variable and its contribution in percentage. It was established that parameters like Q, h and Nu were largely influenced by the inlet temperature of a coolant, where Re and ff were impacted by flow rate. Estimation of heat transfer parameters like Q, h, Nu, Re and ff were done by using MATLAB through Artificial Neural Networks (ANN) and were found to be in good agreement with experimental data.

Circular Mining Wastes Management for Sustainable Production of Camellia sinensis (L.) O. Kuntze

“…When choosing the methods for processing arrays of three-dimensional data, we proceeded from the experience of solving such types of problems in the relevant fields of knowledge [62][63][64][65]. The closest analog is the spatial-temporal problems of geoecology: the assessment of the aerogas regime of the surface layer (earth breathing) [66][67][68][69], forecasting the long-term time series of data [70], geostatistical modeling using GIS technologies [71][72][73], applying fractal models [74], deep learning [75,76], or perceptrons [77,78]. In contrast to using splines [79,80] or the local polynomial regression (LOESS) method [81,82], the primary data were not smoothed in this work due to the small sample size.…”

Substantiation of Drilling Parameters for Undermined Drainage Boreholes for Increasing Methane Production from Unconventional Coal-Gas Collectors

“…For example, a two-dimensional curve characterized by the relationship between the methane concentration and distance S at different times, as well as the methane concentration change over time at different positions, is insufficiently representative (see Figure 2 in [26]) owing to the deformation of a three-dimensional surface when it is projected onto a two-dimensional plane. This conceptual imperfection in the methodological approach does not allow for adequately Spatial analysis of heterogeneous data remains one of the most difficult tasks of predicting the distribution of the response function over the factor space [52], solved using different approaches: fuzzy logic in MATLAB fuzzy logic [53]; stochastic modeling [54][55][56]; wavelet analysis with the Morlet algorithm (CWT) [57,58]; analytical methods based on using trigonometric relationships for quadratic surfaces [59]; nearest neighbor method [60]; inverse weighted distance (IDW) method [61]; multivariate nonlinear regression in SPSS software, www.ibm.com/spss [62]; and machine learning [63], fuzzy cognitive map (FCM) [64], or artificial neural networks (ANN) [65], using GIS technologies-crunching methods [66][67][68]. At the same time, deterministic methods of three-dimensional data interpolation have not lost their relevance [69].…”