This paper presents a numerical study of the natural convection of gases in a horizontal armulus of gas-umdated transmission lines (GIL) It consist of thermal analysis of high voltage power gas insulated transmission lines. The inner cylinder is heated of a constant heat or kept at a constant temperature, the outer cylinder is maintained at constant temperature. The two dimensional analyses of the heat transfer and fluid motion are performed for Rayleigh number ranging from 10S to 10'0, the radius ratio is at 2,
This paper presents a numerical study of the natural convection of gases in a horizontal armulus of gas-umdated transmission lines (GIL) It consist of thermal analysis of high voltage power gas insulated transmission lines. The inner cylinder is heated of a constant heat or kept at a constant temperature, the outer cylinder is maintained at constant temperature. The two dimensional analyses of the heat transfer and fluid motion are performed for Rayleigh number ranging from 10S to 10'0, the radius ratio is at 2,
This paper presents a numerical study of the natural convection of gases in a horizontal armulus of gas-umdated transmission lines (GIL) It consist of thermal analysis of high voltage power gas insulated transmission lines. The inner cylinder is heated of a constant heat or kept at a constant temperature, the outer cylinder is maintained at constant temperature. The two dimensional analyses of the heat transfer and fluid motion are performed for Rayleigh number ranging from 10S to 10'0, the radius ratio is at 2,
Transmission Lines (GIL) Abdellah Chakir and Hermann Koch ,41rstmct - This paper presents a numerical study of the natural convection of gases in a horizontal armulus of gas-umdated transmission lines (GIL). It consist of thermal analysis of high voltage power gas insulated transmission lines, In different case studies the inner cylinder is heated of a constant heat or kept at a constant temperature, the outer cylinder is mamtained at constant temperature. The cylinders are long ancl the flow is assumed to have axially independent properties. The two dimensional analyses of the heat transfer and fluid motion are performed for Rayleigh number ranging from 10S to 100, the radius ratio is at 2,5. Comparisons are made with experimental test set up and measurements. Index Terms - Gas-Insulated Transmission Line, natural convection, thermal behavimrr, numerical study, amnrli, experimental test set-up, measurement I. INTRODUCTION Flows due to the natural convection between two horizontal isothermal cylinders has been widely studied by many authors. The study of this type of problem is usually encountered in many fields specially in solar concentrators, thermal storage plants, pressurised water reactors and electrical Gas insulated transmission lines. Many experimental and numerical works are carried out for laminar case or turbulent flow with low Rayleigh number (till 10) and for isothermal cylinders as boundary conditions. Only a few papers studied the convection with non isothermal boundary conditions and for a Rayleigh over than 107. The results of these experimental and theoretical-numerical studies show the dependence of the flow patterns on the Rayleigh number, the Prandtl number and radius ratio. Investigations of the problem experimental] y and numerically determined the temperature distribution and the local heat transfer coefficients in air and in water. The use of two fluid showed the Prandtl number affect the transition characteristics. In another experimental study it was shown, that the heat flow is substantially altered on both cylinders due to the influence of the eccentricity of the inner cylinder. Good agreement for the temperatures profiles and the flow pattern was found and the Graschoff transition number was determined indirectly. It was shown, that the stream fimction remain negative, i.e., no secondary counter-rotating cells generation. The flow regions of convection a three dimensional spiral flow divided into 5 regions: an inner boundary layer near the inner cylinder, an outer boundary layer near the outer cylinder, a vertical plume region above the inner cylinder, a stagnant region below the inner cylinder, and a core region surrounded by the four regions. This study consists of thermal analysis of high voltage power gas insulated transmission lines (GIL) based on the above mentioned basic knowledge. Two coaxial cylinders of a metallic sheath, an inner conductor and outer enclosure are separated by a dielectric medium, mixture of gas SF6 and Nz. The GIL is designed to be laid in underground or laid in Tunnel with a cooling air flow. The cylinders are very larger than the gap between the inner and outer cylinder, then the two dimensional is applicable. The results of the calculations and the experimental measurements show that for the GIL a ve~ good heat transfer from the conductor, which is represented by the inner pipe, to the outer enclosure is given. The temperature drop between conductor and enclosure is below 10 K. For the practical application of the GIL this means that for a directly buried GIL, or a GIL in a tunnel the thermal heat produced by the current through the conductor is transported fast to the surrounding of the GIL the soil or the ambient air, and leaves low maximum temperatures at the conductor. Compared to a solid insulated cable the maximum temperatures are much lower at the transition point between the outer cable or GIL surface and the soil or air around. Specially in cases of high transmission power ratings of 1500 to 2000 MVA at 420 kV or 550 kV transmission voltage levels the better heat transmission of the GIL is an important advantage. A. Chakir and H. Koch are with Siemens AG, Power Transmission and Distribution, P. (). Box 3220, 91050 Erlangen, Germany (e-mad: abdelIah,chakw@ev. slemens.de, hermann.koch@ev. siemens.de) 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 162 II. PROBLEM FORMULATION In the present study numerical solutions are presented for the steady-state, two dimensional natural convection in the annulus between two horizontal concentric cylinders. The inner, r,, is heated at constant flux or is kept at a constant Temperature T, and the outer, rO, is cooled at constant temperature TO. The thermophysical properties of the fluid are assumed to be constant except the density in the buoyancy term in momentum equation. The Rayleigh number range from 10 to 10Oand the radius ratio are 2,5; 2,6. The gravity force on the system effects vertically downward. The set of equations governing heat flow has been derived in [1], [2], [5], details of development and assumptions are available in these papers. The equations of conservation of mass, momentum and energy are written, after introducing the decomposition between mean and fluctuating quantities. 111. BOUNDARY CONDITIONS Solution of the equations requires specified boundary conditions along the wall encloses the flow field. In different case studies the inner cylinder is heated at constant flux and kept at a constant temperature, the outer cylinder is considered to be at a uniform temperature To. The geometry considered a vertical symmetry plane, then the problem is solved only for the vertical half plane. The axial and vertical velocity is set to zero along each wall. At the symmetry lines the axial velocity and the heat flow are assumed to be zero. Additionally, in the case of constant heat flux, radiation is applied between the surfaces of the two cylinders. [V. RESULTS AND DISCUSSION Case studies Table 1 shows a summary of the considered cases studied in this papers. The radius ratio and the Prandtl number are nearly constant at Pr=0,77 and R+=2,5. TABLE 1 PRESENTED CASE STUDIES Case Ra R+ Pr B.C. inner cylinder 1 2,5e6 2,6 0,7 temperature; T,=373 K 2 2,1e9 2,5 0,75 heat flux; q,=l 01 ,8W/m2 3 1,7e9 2,5 0,75 heat flux; q,=110,8W/m 4 1,4e8 2,5 0,7 temperature; Ti=315 K 5 4,2e9 2,5 0,77 temperature; Ti=315 K 6 5,7e9 2,5 077 heat flux; q,=72,6Wlm 7 2,5e8 2,5 0,7 heat flux; q,=72,6Wlm2 At first a grid size study was made to choose the right mesh. This is important for the calculation time and the quality of the results. Therefore the first case was examined by choosing different grid sizes. The annulus was divided in the radial and the angular direction. The mesh resolution became higher near the lines. Four calculations with different meshes from 41*41 to 100*100 were compared by showing the heat transfer rates for the inner and outer cylinder (local Nusselt number as function of the angular position). A mesh size of 41*41 shows reliable results at the inner cylinder. The values are nearly the same for all resolutions at the top of the cylinder. The inner cylinder shows different Nusselt-profiles for the 4 meshes, but the difference is small enough to choose a grid size of81*81 for all calculations. In order to make a further check of the model, comparisons in case of an isothermal inner wall heating have been made with experimental results of Kuehn and Goldstein [1], [2] for radius ratio R+=2.6, Ra=2,5 10c (Rayleigh number based on the radius difference) and Prandtl number Pt=O.7. In their study, the end losses and radiation correction are made to determine only the convective heat transfer coefficients. Our calculation is therefore made without radiation. The numeri- cal results are carried out with three different wall functions, Spalding, Van Drist and the Equilibrium model. Good agreements were achieved by comparing Nusselt numbers at the inner cylinder fig. 1 and the outer cylinders fig. 2. The different wall functions have nearly no effect on the heat transfer. The temperature distribution across the annulus is presented in fig. 3. The temperature continues to decrease slowly in the core along the radial direction until the outer boundary layer, in which the temperature falls sharply. For angles smaller than 120, a temperature falls sharply in the rRi boundary layer of the inner cylinder ( <0.1 ) and there L appears an inversion in the centre of the annulus. The numerical mean Nusselt numbers at the inner and outer cylinder show good agreement at the outer cylinder, while the inner one shows large disagreement for an angle smaller than 40. The calculation underestimate the heat transfer on the bottom side of the inner cylinder. Figure 4 depicts the isothetms and streamline contours. In the middle of the core the fluid in the top of the inner cylinder moves upward rind recirculates near the outer wall, gets warming up slightly, while in the bottom a cold fluid stays stagnant. OUTER CYLINOER 35 ., 30- D - Spahmg - van Dr,e,[ - EqIib, 25- . Goldstein ~20- g Z15 . 10- I 5 0 0 20 40 60 80 lIXI 120 140 180 180 Angle (degrees) Fig, 1. Nusselt number distribution of the outer cylinder calculated with different wall function in comparison to experimental data presented by Goldstein (Case 1) 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 163 o ~~ o 20 40 en 80 IIXJ 120 140 160 180 Alw@ (degreas) Fig. 2. Nusselt number distribution of the inner cylinder calculated with different wall function in comparison to e~perimental data presented by Temperatures for the Angles 60, 70, 60 1 ~ 0<7 ~ ~,Q --- ~ 0,5- % ; 0,4 * . g 0,3- 0,1 i 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 Radial Iokstion (r-Ri)/L Fig. 3. Temperature distribution in the annulus in comparison to ex~erimental data presented by Goldstein (Case 1) Goldstein (Case 1) A =-. 003194 B =-.002815 c =-.002436 D = ,0020,6 e .-.0016-/7 F =-.001298 G =-.9198-03 H =-. 53,.-03 I .-.1.0, -03 .4 ~ -_ . . . A .--. Fig. 4. Case 1: (ai isotherms, temperature in [K] Other calculations are carried out in order to compare with the results of 2 significant experiments made in the test field for R+=2.5, Pr=O.75 and Rayleiglh number 2,110 and 1,7 1o. (b) streamlines Fig. 5a. Experimental set-up of a GIL thermally isolated by compressed glass wool. The inner and outer cylinder of the GIL of the experimental set-up (Fig. 5a) was made from an aluminium alloy. The outer diameter of the inner cylinder was 250 mm with a wall thickness of 16 mm, the inner diameter of the outer cylinder was 630 mm with a wall thickness of 10 mm. Between them, there was a gas mixture of SFG and N1 as insulated medium. Both cylinders were heated carrying electrical current. The cylinders were stored horizontally, the length of the cylinders were 12 m. The outer cylinder was isolated from the air by compressed glass wool. The temperatures were measured at the inner and outer cylinder at angles 0 (position 1, 2), 90 (position 3,4) and 180 (position 5,6) by thermocouple indicators at the wall. Due to the long distance to the end walls no side effect appeared. The experimental data presented her were measured 12 hours after the experiment started, so that a steady state operation occurred. Figure 5b shows the arrangement. The results of temperature measurements are shown in Table 2. 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 164 Fig. 5b. Measuring points TABLE 2 EXPERIMENTAL RESULTS FOR CASE 2 AND 3 E : ~ Ra=2,1 10 Ra=l,7 10 case 2 case 3 Positio inner cylinder outer cylinder inner cylinder outer cylinder n of the T(K) T(K) T(K) T(K) sensor o 325.56 (point 309,16 (point 3!;3.46 (point 335.96 (point 90 1) 2) 1) 2) 180 325.06 (point 311.96 (point 3:$2.86 (point 339.96 (point 3) 4) 3) 4) 325.46 (point 313.36 (point 353.36 (point 341.86 (point 5 6 5) 5 Heat flux 101.8 110.5 (Wlm2 TABLE 3 COMPARISON OF NUMERICAL AND EXPERIMENTAL ( &sa Angle r) experiment Equilibrium The calculations are made for a constant heat flux 101.8 W/m, 110.5 W/m applied at the inner cylinder obtained from the experiments and with a temperature constant at outer cylinder, which is the average value of the three values above (311 .5K, 339.3 K). The numerical results are presented in the Tab. 2. The calculation was carried out with radiation be- tween the two cylinders. The same numerical results are obtained with the Van Drist and Spalding model [3]. They arc slightly over the value obtained with the Equilibrium model. The calculated temperature clifference between top and the middle is over than 5, however the experimental difference is not significant. One can explain the constant value of temperature at the inner cylinder in the experiment by the more important conduction heat phenomena in the wall of inner cylinder than the convection phenomena at the boundary layer. This is shown by the Biot number resistance in the wall (Bi= a =0.02), i.e., the conduction heat is JCY,inde, I Lo much smaller than the convection heat resistance. The Nusselt number at the inner and outer cylinder for the three wall function are presented in fig. 5 for case 2. The inner cylinder is at a constant heat flux. In the outer cylinder the Nusselt number increase sharply from the bottom till the horizontal plan (90 )where remain constant until the region near the top. For the inner cylinder the temperature decrease slowly from angle 0 till angle 160, at which an inversion increase the Nusseh value to the one at OO. The equilibrium model presents an increased Nusselt-number, while the Spalding and Van Drist [4] models presents a decrease (fig. 5). This phenomena of Nusselt-number decrease for Spalding and Van Drist models appear only for high Ra number (see fig. 1: Ra = 2,5 10). Both wall function Spalding and Van Drist seem not good for high turbulence. The numerical results obtained with the equilibrium model presents the best agreements with the experiment values (Tab 3). So the calculation for the second experiment are made only with this model and results of temperatures for inner cylinder are presented in Tab. 4 TABLE 4 COMPARISON OF NUMERICAL AND EXPERIMENTAL RESULTS (CASE 3) (Case 3) Ra=l,7 10 inner cylinder k Angle () o 90 180 7 - experiment I Equilibrium model T(K) T(K) 353.3 361.4 352.7 354.6 353.2 I 360.8 OUTER CYUNDER . El * Eqlb 60 + S@ldlm -c- Van Dnst m- =40- ! zsc- 20- o ~~ o 20 40 60 S6 100 120 140 160 180 Angle (degrees) Fig. 5. Nusselt number distribution of the outer cylinder calculated with different wall function for a high Rayleigh number (case 2) Three different studies were made to examine mainly the effects of a Rayleigh number variation on the convection heat transfer at the inner and outer cylinder (fig. 6, 7). The temperature was constant at both cylinders. The radiation was not considered. The Rayleigh number reach from 2,5e6 to 4,2e9, the other parameter are nearly constant at R+=2,5 and PI=0,7 (case 1,4,5). 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 165 OUTER CYLINDER L The temperatures over the annulus center are nearly constant for gas 1, only the boundary layers have strong gradients (fig. 9). The layer becomes smaller with higher turbulence (fig. 3 i 3C4 and 9) and the temperatures more uniform in the middle part. 7A INNER CYLINDER 20 325- 10 ~-- o 0 20 40 60 80 100 4:0 140 480 180 : AIwle (dwmes) } Fig. 6. Nusselt number distribution of the outer cylinder for different Rayleigh numbers (case 1,4,5) 317 INNERCYLINDER An-- <0 l-- --x%.% 5 4 OL: I 0 20 40 60 80 fOO 120 140 ?60 180 A@e (dwreas) Fig. 7. Nusselt number distribution of the inner cylinder for different Rayleigh numbers (case 1,4,5) The Nusselt number at the outer cylinder has an increasing trend from the bottom to the top, where the highest values are reached. The highest heat transfer rate is at 180 with Rafi,2 10. Noticeable is lhe strong increasing heat transfer in the middle (80-1000) with higher Rayleigh numbers. The same trend is existing at the inner cylinder, thus indicating a becoming higher turbulence in the middle part. The Nusselt number (inner cylinder) is relative constant between 10 and 160 for the Rayleigh numbers Ra=2,5 10~ and 1,4108. It is decreasing slightly from the bottom to the top. For Ra= 4,2 10g is the heat transfer rate loosing the uniformity. It is increasing to the middle of the cylinder. The influence of different fluids under pressure is investigated on the Temperatures in the annulus in fig. 8. Gas 1 is a mixture of SF6 and N2 (Pr=0,77; R.a=5,7e9), gas 2 is simple air (Pr=0,7; Ra=2,5e8). Fig. 8 shows the temperature distribution at the inner cylinder and the mean temperatures. Gas 1 temperature is lower due to the higher turbulence for high Rayleigh numbers. Both temperatures are decreasing to the middle of the cylinder, where they reach a minimum. 315~4 o 20 40 60 80 100 120 140 160 180 Awl. (dwre.s) Fig. 8. Temperature distribution T, at the inner cylinder for two different gases (case 6,7) and the calculated average temperature Tl~ (case 6,7) 320 310 308 306 ~ / 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,S 1 Radial Lokation (r-Ri)/L Fig. 9. Temperature distribution T, in the annulus for a high Rayleigh number (case 6) The temperature in the center is nearly constant and the gradient over the annulus is high. Additionally the kinetic energy and the flow field is presented for case 2, Basically the flow is composed of an increase of thermal boundary layer along the inner cylinder, of a rising of a hot buoyant plume toward the top of the cylinder, an entrainment of the buoyant plume into the thermal boundary layer along the outer cylinder, and finally the separation of the boundary layer from the outer cylinder to complete the recirculation pattern inside the cavity. The turbulent kinetic energy contours show a high turbulence near the top of the outer cylinder and near the center of the recirculating flow. 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 166 V. CONCLUS1ON An investigation of turbulent natural convection in a horizontal concentric annulus wa:s carried out for high Rayleigh numbers up to Ra=l O1. The results were presented in form of temperatures and Nusselt numbers at the inner and outer cylinder. The numerical results were compared with experimental data. Three different wall functions were used to find an accarate model. There were no effects noticeable for a low tttrlbulent flow. The differences to the achieved Nusselt results of Goldstein were small. The maximum temperature deviation was 15Yo. The equilibrium model shows the best results compared to the carried out experiment. The temperatures of the middle inner cylinder were nearly equal. However, high differences appeared at the top/bottom of the cylinder. This were the results of the much smaller heat resistance in the wall than the convective resistance. The calculated temperatures at the inner cylinder had a uniform trend except the top and bottom, where they were rising to the highest values at 0 and 180. The Nusselt number as a function of Rayleigh showed the high dependency. The increasing heat convection for high Rayleigh numbers occurred especially in the middle part of the cylinders and in the top of the outer one. The temperatures in the center of the annulus became constant for higher turbulence. The temperature gradient at the wall boundary layers was rising. The trend for high turbulence is going to small thermal boundary layer at the inner and outer cylinder with uniform temperatures at the annulus center. [1] [2] [3] [4] [5] VI. REFERENCES T.H. Kuebn und R.J. Goldstein, An experimental and theoretical study of nataral convection in the annulus between horizontal concentric cylinders, J Fhrid Mech. 74, 695-719(1976). T. H. Kuehn, R. J. Goldstein, An experimental study of natural convection heat transfer in concentric and eccentric horizontal cylindrical annuli, ASME J. Heat Transfer 100,635-640 (1978). B. E. Launder, D. B. Spalding, The numerical computation of torbulent flows, Computer methods in applied mechanics and engineering Vol. 3,269-289 (1974). E. R. van Drist, On Turbulent Flow Near a Wall, Journal of the Aeronautical Sciences, vol. 23, 1956, p. 1007 Fluid flow, Ansys Theory Reference release 5.4, chap.7, pp 7.1 -pp 7-39 ( 1996) VII. BIOGRAPHIES Institute was finished with Dr.-hrg. Hermann Koch was born in November 1954 in Hauswarz, Germany. From 1976-1979 he studied Electrical Engineering at the Fachhochschtde Riissels- heim, where he graduated with the diploma of engineering (Dipl.-lng.). From 1980-1981 he studied on a Ftdbright Scholarship at New Jersey Institute of Technology, Newark, New Jersey, USA. From 1981-1986 he studied Electrical Engineering at the Technical Uni- versity of Darrnstadt, where he graduated with the diploma of engineering (Dipl.-Ing.). His continued research work at the High Voltage his doctoral work about Partial Discharges in Low Voltage Equipment and the degree of Dr.-hrg. Since 1990 he is working with Siemens High Voltage Division where he ia head of Gas Insulated Lines Department. He is committee member of the IEEE Power Engineering Society in the Gas Insulated Substation Committee. In the IEC he is Secretary of Technical Subcommittee SC I 7C and in CENELEC he is Secretary of the Technical Committee TC 17. in research and development Insulated Substations. Dr.-lng. Abdellah Chakir was born 1965 in Casablanca, Morocco. He received his Dipl. Ing. und M. SC. degrees in mechanical and aeronautical engineering in 1990 from ENSMA (Ecole Nationale Superieure de M4canique et dA&otechnique de Poitiers, France) and received the Ph. D. degree in Mechanics in 1993 from Poitiers University, France. From 1993-1996 he was Auxiliar Professor in mechanical engineering at the Polytechnic Institute and ISACE of Guarda, Portugal. Since 1996 he is working with Siemens High Voltage Division, Gas- 0-7803-7031-7/01/$10.00 (C) 2001 IEEE 0-7803-7173-9/01/$10.00 2001 IEEE 167