18th European Symposium on Computer Aided Process Engineering – ESCAPE 18 Bertrand Braunschweig and Xavier Joulia (Editors) © 2008 Elsevier B.V./Ltd. All rights reserved. Numerical determination of heat transfer coefficient for nanofluids in microchannels. Kurowski Łukasz1, Thullie Jan1, Chmiel-Kurowska Klaudia1, Dzido Grzegorz1 1Silesian University of Technology, Department of Chemical and Process Engineering, Gliwice, POLAND, e-mail: lukasz.kurowski@polsl.pl Abstract CFD tools were successfully used to model laminar heat transfer of nanofluids. Quite good agreement with experimental results was obtained. Keywords: CFD, nanofluid, laminar flow, heat transfer, microchannels 1. Introduction Water and ethylene glycol are commonly applied heat transfer fluids. Nethertheless the heat conductivity for these fluids is not high. Intensive investigations were carried out for many years, seeking to develop some methods for enhancing their thermal performance. Most recently nanotechnology gives such possibilities of creating effective heat transfer media. The idea is to use a suspension of very small particles (nanoparticles) with diameters less than 100 nm in a base liquid (usually water, ethylene glycol or minerals oil). Such suspension called nanofluid, has effective heat transfer parameters greater than a base liquid. The effect is observed when nanoparticles have heat conductivity many times greater than the liquid. Usually Cu, Ag, CuO, Al 2 O3 or CNT are used. Nanofluids are characterized by greater heat conductivity than a base fluid even for a small concentration of nanoparticles [1, 2, 3]. An enhancement of 17% was observed in the heat conductivity of water based nanofluids with particle loading 0.4% in volume CuO and for ethylene glycol nanofluid an enhacement of 40% was observed with particle loading 0.3% in volume Cu. The mechanism itself of this phenomenon is not fully recognized. It results in augmentation of heat transfer coefficient. Convective heat transfer characteristics for nanofluids was experimentally investigated by Xuan and Li [4], Wen and Ding [5], Heris et al. [6] and Ding et al [7]. They found that inclusion of nanoparticles can significantly enhance the convective heat transfer coefficient. Unlike the results of forced convection Putra et al [8] and Wen and Ding [9] found deterioration of the natural convective heat transfer of water based nanofluids. Regardless extensive research there is no theory which could explain this results. Moreover, miniaturization of heat – exchange equipment applied in electronic, optoelectronic, motorization and medicine is connected with the increase of the amount of heat generated, which should be exchanged by a high-efficient cooling equipment. For this purpose microchannel cooling devices are applied for liquid coolant with hydraulic diameter < 2 mm. Further heat transfer intensification in such devices may be obtained by using nanofluids and taking advantage of their unique potential. Available data on such solutions are nowadays limited and some further investigations should be carried out [2]. Ł. Kurowski et al. 2 2. Mathematical models and numerical method The purpose of the work is modeling of an experimental plant used for investigation of nanofluids laminar flow convective heat transfer. The plant consists of a cooper tube 1.4 mm inner diameter and 620 mm length (Fig 1). The tube is heated up by an electric heater generating heat flux q = 2000 – 30 000 W/m2. Ethylene glycol is used as base fluid and Cu as nanoparticles (100 nm). Because of small concentration of nanoparticles (0.15% particle volume fraction) the pseudohomogeneous model was applied, so nanofluid is treated more like fluid rather than liquid-solid mixture. Similar approach was presented by Maiga and Palm [10]. Ltot Lh dQ=0 r Q dQ=0 z fluid flow inlet outlet z=0 z = Ltot r = Ri r = R0 Fig. 1 Experimental system The continuity, momentum and energy equations written in vector form are . v = 0 ( v. (1) ) ρv = - P + . (η v) c p v. T = k T 2 (2) (3a) For the tube wall the energy equation is k s 2Ts = 0 (3b) The boundary conditions are z 0 , 0 r Ri , u u 0 , T T0 0 z Ltot , r 0 , T u 0, s 0, v 0 r r r Ri , u v 0 The conditions for heated wall are (4a) (4b) (4c) Numerical determination of heat transfer coefficient for nanofluids in microchannels r R0 , q 0 k s Ts r 3 (5a) for heated part of the tube ks Ts 0 r (5b) for nonheated part of the tube z Ltot , 0 r R0 , T s 0 z (5c) The set of equations (1)-(3) together with boundary conditions (4-5) was solved by numerical method based on finite volume approach. The equations were solved in 2D with commercial CFD code FLUENT 6.3. In order to ensure that the solution does not depend on the number of grid points several different types of grids were tested. After preliminary calculations a non uniform grid 55 x 1296 was chosen. Finer grid point spacing was used at the entrance and exit region. Two wall heat fluxes were simulated, namely 14 331 and 19 108 W/m2. Supplied heat flux was not uniform along the tube. Appropriate measurements gave functional relationship used in boundary conditions. The results were compared with experimental data. 3. Thermophysical properties of nanofluids and heat transfer coefficient We assume that the nanoparticles are well dispersed within the base-fluid, so the effective physical properties are described by classical formulas: ρnf 1 Φ ρl Φρ p (6) cnf 1 Φ Cl Φ c p (7) however temperature dependence of l, p, cl and cp was taken into account. The dynamic viscosity of nanofluids was determined from functional relationship obtained by performing a least-square curve fitting of experimental data. The measurements were done by Brookfield DV-II+ Pro apparatus and plotted in Fig. 2 as temperature function. It shows that the temperature dependence of nanofluid viscosity as well as pure ethylene glycol viscosity is very strong. It is worth nothing that nanofluid viscosity at 318K is three times greater than ethylene glycol viscosity at the same temperature. The Maxwell correlation was used for determination of nanofluid effective thermal conductivity as follows k nf k p 2k l 2(k p k l )Φ k p 2k l (k p k l )Φ kl (8) Ł. Kurowski et al. 4 0.08 nanofluid nanofluid - experimental data EG EG - experimental data 0.06 [Pas] 0.04 0.02 0 280 300 320 340 360 T [K] Fig 2. Nanofluid effective viscosity compared to viscosity of pure base fluid Local Nu numbers and heat transfer coefficients were calculated according to equations hd i k (9) q Twx Tbx (10) Nu h Where q is the heat flux, Twx and Tbx are respectively, the wall and fluid bulk temperatures, d is tube diameter and k is the fluid thermal conductivity. The fluid bulk temperatures at x position were calculated according to the following equation: Tbx Tin PqLx mcp (11) where P is a perimeter of the test tube. 4. Results and conclusions CFD tools made important contributions to process modeling [11, 12] This work presents some capabilities of CFD technology to improve modeling of an experimental plant. Discrepancy between measured and calculated temperature profiles gives rise to careful reassessment of the wall boundary conditions and formulation of functional relationship for the heat flux supplied by the heater which gives dependence on the axial coordinate. Numerical determination of heat transfer coefficient for nanofluids in microchannels 360 360 2.7 l/h - CFD 2.7 l/h - exp. 5.2 l/h - CFD 5.2 l/h - exp. 6.7 l/h - CFD 6.7 l/h - exp. 350 2.7 l/h - cfd 2.7 l/h - exp. 4.0 l/h - cfd 4.0 l/h - exp. 5.3 l/h - cfd 5.3 l/h - exp. 6.7 l/h - cfd 6.7 l/h - exp. 350 340 5 340 T [K] T [K] 330 330 320 320 310 310 0 a) 0.2 0.4 0.6 0.8 1 x/L 0 0.2 b) 0.4 0.6 0.8 1 x/L Fig. 3 Axial temperature profiles in the tube wall cooled by: a) pure EG, b) nanofluid: EG + Cu Calculations were performed for ethylene glycol (100%) and nanofluid Cu/ethylene glycol for different Re numbers in laminar flow. The inlet fluid temperature was kept constant at 318 K. Wall temperature profiles were presented in Figs 3a and 3b. The curve parameter was mass flow rate of nanofluid and a fit with experimental points is good. Fig. 4 compares the wall temperature with uniform heat supply to experimental conditions. At the entrance to the tube the difference between profiles is quite noticeable. It is caused by the heater construction. Fig. 5 shows the mean heat transfer coefficient as a function of Re number. 350 q: f(x), prop.: f(T) q: const, prop.: f(T) experimental data 340 T [K] 330 320 310 0 0.2 0.4 0.6 0.8 1 x/L Fig. 4. Axial temperature profiles in the tube wall for uniform heated tube – simulated and for experimental conditions. Quite good agreement with experiments were obtained, however the scatter of experimental results suggests that the problem is still open. Ł. Kurowski et al. 6 4000 3500 3000 have [W/m2K] 2500 CFD - EG CFD - nanofluid: EG + Cu exp. data - EG exp. data - nanofluid: EG + Cu 2000 1500 0 40 80 120 160 200 240 Re Fig.5 Dependence of average heat transfer coefficient vs. Re number References [1] V.Trsaksri, S. Wongwises; Renewable and Sustainable Energy Reviews, 11, (2007), 512 [2] W. Daungthongsuk, S. Wongwises; Renewable and Sustainable Energy Reviews, 11, (2007), 797 [3] X-Q Wang, A.S. Mujumdar; International Journal of Thermal Sciences, 46, (2007), 1 [4] Y. Xuan and Q. Li; Journal of Heat Transfer 125, (2003), 151 [5] D. Wen, Y. Ding; International Journal of Heat and Mass Transfer, 47, (2004), 5181 [6] S. Heris, S.G. Etemad, M. Esfahany; International Communications in Heat and Mass Transfer, 33, (2006), 529 [7] Y. Ding, H. Alias, D. Wen, R.A. Williams; International Journal of Heat and Mass Transfer, 49, (2005), 240 [8] N. Putra, W. Roetzel, S.K. Das; Heat and Mass Transfer 39, (2003) 775 [9] D. Wen, Y. Ding; International Journal of Heat and Fluid Flow, 26, (2005), 855 [10] S.E.B. Maiga, S.J. Palm, Internationa Journal of Heat and Fluid Flow 26, (2005), 530 [11] C,C, Pantelides, Computer Aided Process Engineering, 9, 2001, 15 (ESCAPE 11) [12] F. Bezzo, S. Macchietto, C.C. Pantelides, Computers and Chemical Engineering, 24, (2000), 653. Nomenclature cp – specific heat capacity, J/kgK Ri – inner radius, m h – heat transfer coefficient, Ro – outer radius, m W/m2K k – thermal conductivity of fluid W/mK T – fluid temperature, K ks – thermal conductivity of solid W/mK Ts – wall temperature, K Ltot –total length of the channel, m u, v – velocity components, m/s Lh –heating length of the channel, m z – axial coordinate, m q –heat flux, viscosity, Pa s W/m2 r –radial coordinate, m -density, kg/m3