CFD analysis and experimental investigations towards optimizing the parameters of RanqueHilsch vortex tube

Upendra Behera a, P.J. Paul a, S. Kasthurirengan b, R. Karunanithi b, S.N. Ram b, K. Dinesh b, S. Jacob b,*
a

Department of Aerospace Engineering, Indian Institute of Science, Bangalore 560 012, India b Centre for Cryogenic Technology, Indian Institute of Science, Bangalore 560 012, India

Abstract Computational uid dynamics (CFD) and experimental studies are conducted towards the optimization of the RanqueHilsch vortex tubes. Dierent types of nozzle proles and number of nozzles are evaluated by CFD analysis. The swirl velocity, axial velocity and radial velocity components as well as the ow patterns including secondary circulation ow have been evaluated. The optimum cold end diameter (dc) and the length to diameter (L/D) ratios and optimum parameters for obtaining the maximum hot gas temperature and minimum cold gas temperature are obtained through CFD analysis and validated through experiments. The coecient of performance (COP) of the vortex tube as a heat engine and as a refrigerator has been calculated.

Keywords: RanqueHilsch vortex tube; CFD simulation; Experimental validation; Design parameters; Thermal performance

1. Introduction RanqueHilsch vortex tube having, relatively simple geometry, no moving mechanical parts and no need for absolute sealing element, was invented by Ranque [1] demonstrating the eect of temperature separation of gases. Later, Hilsch [2] described this effect in detail. Intensive experimental and analytical studies of RanqueHilsch eect began since then and continue even today. According to these studies when a vortex tube is injected with compressed gas through tangential nozzles into its scroll chamber, a strong cir-

Corresponding author. Tel./fax: +91 80 23601612. E-mail address: jacob@ccf.iisc.ernet.in (S. Jacob).

cular ow eld is established. This vortex in the inlet area causes pressure distribution of the ow in radial direction. As a result a free vortex is produced as the peripheral warm stream and a forced vortex as the inner cold stream. The ow pattern is schematically shown in Fig. 1. The tangential velocity of the uid element is related to the distance r from the axis of rotation as, v / r, for forced vortex or quasi-solid rotation and v / 1/r, for free vortex or potential vortex. The variation of tangential velocity along the radius for such a case in a vortex tube representing forced region and free region is shown in Fig. 2. Many investigators, Hilsch [2], Deissler and Perlmutter [3], Kurosaka et al. [4], Ahlborn et al. [5], Cockerill [6], Gutsol [7], have suggested various theories to explain the Ranque eect. Ahlborn et al. [14] have postulated a

Nomenclature A CFD COP D dc h he k L m p R re r Sm Si T t u ui V0 v w xi z area (m2) computational uid dynamics coecient of performance diameter of vortex tube (mm) cold end diameter (mm) enthalpy (kJ kg1 K1) heat engine turbulence kinetic energy (m2 s2) length of vortex tube (m) mass ow rate (kg s1) piezometric pressure (Pa) radius of vortex tube (mm) refrigerator radial distance from axis mass source momentum source components temperature (K) time (s) swirl velocity (m s1) absolute uid velocity component in direction xi maximum swirl velocity (m s1) radial velocity (m s1) axial velocity (m s1) Cartesian coordinate (i = 1, 2, 3) axial length from nozzle cross section Cei DTch DTih DThi coecients (i = 1, 2, 3, 4) used in Eq. (4) temperature dierence between cold and hot end temperature dierence between inlet and hot end temperature dierence between hot end and inlet

Greek symbols a cold gas fraction b coecient used in Eq. (4) e turbulence dissipation rate (m2 s3) l dynamic viscosity (kg m1 s1) lt turbulent viscosity (kg m1 s1) g0 coecient used in Eq. (4) q density (kg m3) r stress (N m2) s shear stress (N m2) sij stress tensor components Subscripts c cold gas h hot gas i inlet in vortex tube r constant reference value

theory of temperature separation based on heat pump mechanism enabled by secondary circulation ow in vortex tubes. However till today no exact theory has come up to explain the phenomenon satisfactorily. Thus much of the design and development of vortex tubes have been based on empirical correlations leaving much scope for optimization of critical parameters. Computational uid dynamics techniques have revolutionized engineering design in several important areas, notably in analysis of uid ow technology. CFD can also be used as a minimal adequate tool for design of engineering components. A careful scanning of various numerical investigations on the mech-

anism of thermal separation in vortex tubes indicate that barring a few [10], no serious attempts have been made to use CFD techniques to simulate the ow patterns of vortex tubes. In the present study, a detailed analysis of various parameters of the vortex tube has been carried out

Fig. 1. Schematic ow pattern of RanqueHilsch tube.

Fig. 2. Schematic velocity distribution in RanqueHilsch tube.

through CFD techniques to simulate the phenomenon of ow pattern, thermal separation, pressure gradient etc. so that they are comparable with the experimental results. The paper reports also its utility as a tool for optimal design of vortex tube towards the optimization of number of nozzles, nozzle proles, cold end diameter, length to diameter ratio, cold and hot gas fractions and comparison of the experimental results with corresponding CFD analysis.

is very high, Renormalization Group (RNG) version of ke model has been used as this can be employed for law of the wall functions. The equations involved in RNG ke model for turbulence energy and turbulence dissipation rate are given as   o o l ok qk q~j k eff u ot oxj rk oxj   2 oui oui lt qk 3 lt P P B qe 3 oxi oxi   o o l oe qe q~j e eff u ot oxj re oxj     e 2 oui oui e lt C e3 lt P B qk C e1 lt P k 3 k oxi oxi C e2 q e2 oui C l g3 1 g=g0 qe2 C e4 qe 1 bg3 k oxi k 4

2. Numerical modeling of vortex tube The numerical modeling of the vortex tube has been carried with the code system of Star-CD. This programme allows calculations for incompressible, compressible and turbulent ow. The mass and momentum conservation equations per unit volume with respect to time (Star-CD Methodology and User Guide [11]) solved by Star-CD for general incompressible and compressible uid ow are given by o o q~j S m u q ot oxj o o op qui q~j ui sij u Si ot oxj oxi 1

where Ce1, Ce2, Ce3 and Ce4 are given by 1.42, 1.68, 0.0 or 1.42 (if PB > 0 and 0 otherwise) and 0.387 respectively. leff l lt P  2S ij oui ; oxj PB gi 1 oq ; rh q oxi g  Sk=e;

S  2sij sij 1=2 ;

g0 4:38;

b 0:01

where ~j uj ucj , relative velocity between uid and u local (moving) coordinate frame that moves with velocity ucj. The specialization of the above equations to a particular class of ow involves: application of ensemble or time averaging if the ow is turbulent; specication of a constitutive relation connecting the components of the stress tensor sij to the velocity gradients; specications of the source, Si, which represents the sum of the body and other external forces, if present. The scope of numerical model was restricted through few assumptions like steady state, symmetrical and compressible ows where the density itself could be estimated using the equation of state with the calculated temperature and pressure values. 2.1. Turbulence model

3. Physical modeling of vortex tube The vortex tube ow was investigated by modeling the 12 mm diameter vortex tubes used for the experimental programme with the Star-CD code. The meshing and analysis of vortex tubes have been carried out in a P4, 2.4 GHz and 1 GB RAM machine with renement of meshes near the critical regions such as nozzle inlet areas, hot end exit area and at wall boundary. The analysis is carried out to arrive at the optimum number of nozzles, nozzle prole, cold end diameter (dc), length to diameter (L/D) ratio, cold and hot gas fractions. The diameter of vortex tube, D = 12 mm and ratio of nozzle inlet area to vortex tube area Ai/A = 0.07 are kept nearly constant in the analysis. The L/D ratio assumed for the studies ranged from 10 to 35. 3.1. Boundary conditions for analysis

Flow in the vortex tube is highly turbulent. Star-CD employs mathematical models of turbulence to determine the Reynolds stresses and turbulent scalar uxes. The steady state assumption and practical considerations indicate that a turbulence model must be employed to represent its eects. As the Reynolds number

The pressure and temperature data obtained from the experiments are supplied as input for the analysis. The boundary conditions given to simulate the vortex tube phenomenon at dierent regions are as follows.

 Stagnation boundary condition to the inlet region of the vortex tube (at inlet of nozzle) with total pressure of 0.5422 MPa (absolute) and total temperature of 300 K. Pressure boundary condition to cold end region of vortex tube with pressure of 0.136 MPa (absolute). Pressure boundary condition to the hot end region of vortex tube with pressure varying so as to vary the mass fraction at the hot end to obtain the optimum value. The temperatures at hot and cold ends are given zero gradients in calculation and the mass ow rate at the outlet of nozzle is constant and the ow is compressible. As the ow inside vortex tube is axisymmetric, a part of sector is taken for analysis giving cyclic boundary condition. The three-dimensional model showing the boundary regions used in calculation is represented in Fig. 3(a) and (b). The reference pressure pr = 0.136 MPa and reference temperature Tr = 293 K are specied. The tube walls are considered to be adiabatic and no slip conditions are used. For the numerical solution of the equation for compressible ow, a variant of well-known SIMPLE algorithm method is used.

3.2. Grid dependence study To eliminate the errors due to coarseness of grid, analysis has been carried out for dierent average unit cell volumes in a vortex tube of L/D = 10. The variation of key parameters such as total temperature dierence and maximum swirl velocity are shown in Fig. 4(a) and (b) for dierent unit cell volumes. It can be seen that not much advantage in reducing the unit cell volume size below 0.003 mm3 which corresponds to 0.75 million cells for the conguration studied.

4. Velocity and temperature proles in vortex tube from CFD analysis For a specic vortex tube of diameter D = 12 mm, L/D = 20, cold end diameter dc = 7 mm with six numbers of convergent nozzles the following analysis is made. 4.1. Velocity components and ow pattern It has been known [6] that experimental determination of velocity components is rather dicult in vortex tube, because the measuring probes can considerably

Fig. 3. (a) Three-dimensional model of vortex tube in sector; (b) the hot end and cold end of vortex tube in sector provided with renement in mesh.

Total temperature difference (K)

60 55 50 45 40 35 30 0 0.003 0.006 0.009 0.012 0.015 Average unit cell volume (mm3)
280

(a)

Swirl Velocity (m/s)

276 272 268 264 260 0 0.003 0.006 0.009 0.012 0.015

diameter and the core ow lls up the cold ow outlet area (7 mm diameter) completely. Stagnation point can be established from the velocity prole along the tube length at the point where axial velocity cease to have a negative value. The radial velocity (v-component) is signicantly low in magnitude compared to the u and w components. It has a negative component at the inlet zone at radius %0.3R towards the tube axis and a positive component upto the tube wall. This indicates possibilities of energy transfer in the radial plane. The above velocity proles obtained at dierent z/L sections of vortex tube are in good conformity with those shown by Gutsol [7]. The experimental observation of ow pattern in metallic vortex tube is rather dicult, while CFD analysis can provide better understanding of this. The ow pattern obtained by CFD analysis for sectional lengths of 0.05 m at specic z/L location is shown in Fig. 5(d). The core and peripheral ow can be clearly seen at z/L sections at 0.0375 and 0.46 but at z/L = 0.95 shows only the presence of peripheral ow towards the hot end. Thus CFD analysis can locate the stagnation point at appropriate z/L location of the vortex tube. 4.2. Temperature distribution

(b)

Average unit cell volume (mm3)

Fig. 4. Grid size dependence study on: (a) total temperature dierence at dierent average unit cell volume; (b) maximum swirl velocity at dierent average unit cell volume.

alter the ow pattern. Even using a 1.6 mm diameter Pitot tube for a 25.4 mm diameter vortex tube, 8% of vortex tube cross section is blocked [13]. Therefore determination of the velocity components from CFD assumes great signicance. The radial proles of velocity components swirl, radial and axial are shown in Fig. 5(a)(c) and the ow patterns at specic sectional lengths in Fig. 5(d). Comparing the velocity components, it is observed that swirl velocity (u-component) has the highest value. It has a value almost equal to inlet tangential ow V0 in the nozzle inlet zone, which rapidly decreases in amplitude towards the hot end discharge. The radial prole of the swirl ow indicates a free vortex near the wall but predominantly forced vortex especially at inlet section, which is in conformity with the observations of Kurosaka [4], Gutsol [7]. The small diameter of the vortex tube (12 mm) also leads to a predominantly forced vortex ow pattern. The axial ow velocity (w-component) shows a maximum value w/V0 % 0.365 at the tube axis near the inlet zone. The axial ow prole shows that the ow reversal of the forced vortex core ow takes place at about 8 mm

The total temperature from CFD analysis plotted as contour plot in Fig. 6. It shows that peripheral ow is warm and core ow is cold relative to inlet temperature for a hot gas ow fraction of about 28%, giving maximum hot gas temperature of 353.7 K and minimum cold gas temperature of 266.2 K. For experiments conducted with nearly same conditions of hot gas ow fraction, the maximum hot gas temperature obtained is %360 K and minimum cold gas temperature %281 K. Thus there is fair agreement in the total temperatures obtained through experiment and by CFD simulations.

5. CFD analysis towards optimization of parameters 5.1. Nozzle prole and number of nozzles Earlier, investigators [9] had to carry out laborious fabrications and experimental programme to deduce the optimum nozzle proles and evaluate their performances. CFD analysis can minimize these diculties. CFD analysis has been carried out on dierent types and number of nozzles to arrive at the optimum prole of the inlet nozzle and number of nozzle(s) for enhanced swirl generation. In order to provide uniform basis for comparison of results and to ensure symmetric conditions, the analysis have been done for the entire cross section and not for 60 sector that is possible for six

Fig. 5. (a) Radial proles of swirl velocity (u-component); (b) axial velocity (w-component); (c) radial velocity (v-component) at z/L = 0.0375, 0.46, 0.95; (d) ow pattern for 0.05 m length sections at z/L = 0.0375, 0.46, 0.95.

numbers of nozzles or 180 sector that is possible for two numbers of nozzles. This may yield a lesser value of total temperature dierence compared to sector analysis. However for comparison of nozzle proles and number

of nozzles, this approach is adequate. The geometric dimensions, boundary conditions and mass ow are kept nearly constant for all nozzle proles in the analysis.

Fig. 6. Temperature distribution in axial direction of vortex tube in sections.

The swirl pattern in the vortex generator for dierent nozzle proles are given in Fig. 7(a)(e) and the result of analysis is given in Table 1. The performance of swirl generators can be characterized by two important factors, namely the magnitude of swirl velocity and radial symmetry of ow. The ow pattern obtained for two numbers of convergent nozzles is oval in shape though the swirl velocity is highest. In this case the non-radial symmetry of the ow results in less total temperature dierence. In the case of six convergent nozzles, radial symmetry of ow is obtained along with good swirl velocity, which has resulted in the highest total temperature dierence compared to all the nozzle proles. This is in conformity with the studies as reported by Gutsol [7], which mention that not less than four tangential nozzles are necessary to attain proper radial symmetry of ow. The swirl velocity attained for circular and rectangular helical single nozzles are signicantly less than for conical nozzles and hence they give lower total temperature dierence. Therefore a swirl generator with six numbers of convergent nozzles is selected for further design and analysis of vortex tubes. 5.2. Cold end orice diameter The temperature dierence DTch obtained experimentally for L/D = 10 at specic hot gas fractions and at different cold end diameters of 5, 6, 7, and 7.5 mm and CFD simulation for 6 and 7 mm are shown in Fig. 8(a). It can be seen that maximum temperature dierence of 53.4 K (at 15.75% of hot gas ow) is obtained from CFD analysis and about 62.9 K (at 13.51% of hot gas ow) is obtained from experiments at dc = 7 mm. Thus CFD analysis is in reasonable agreement with experimental results. From the analysis and experiments it has been observed that vortex tube can be operated in such a way

that it can be used to produce maximum hot gas temperature and/or to produce minimum cold gas temperature. The CFD analysis and experimental results are plotted in Fig. 8(b) and (c) showing hot gas temperature and cold gas temperature at dierent hot gas fractions for specic cold end diameters at L/D ratio of 10. Fig. 8(d) and (e) shows the experimental data of maximum hot and minimum cold gas temperatures obtained for dierent cold end diameters and L/D ratios. The optimum value of cold end diameter dc to get maximum hot gas temperature is 7 mm where as to get minimum cold gas temperature dc is 6 mm, for all the values of L/D greater than 10. These values of cold end diameter fairly match with values reported by Takahama [12] and the correlation dc/D = 0.350 + 0.313a as given by Merkulov [8]. This study conrms that CFD modeling carried out in this work is capable of selecting the correct dc value for a vortex tube, without resorting to the empirical correlations as a design guide or a laborious experimental programme. 5.3. Length to diameter ratio Initial CFD analysis for specic L/D ratio of 10 showed that the swirl intensity extends beyond the length of hot end exit of the tube and maximum DTch to be 53.4 K. Further investigations were carried out at dierent L/D ratios in the range of 1030 to nd the length of tube such that the stagnation point is within the tube. They led to the conclusion that increase in the length of tube enhances the temperature separation upto the condition that stagnation point is within the length of tube. The corresponding L/D ratio is in the range of 2030. Experiments were carried out for dierent L/D ratios of 10 to 35. Fig. 9 shows the experimental and CFD analysis data on temperature dierence DTch for dierent L/D ratios. It can be noted that the peak value in DTch is obtained experimentally for L/D ratio of 30. The studies highlight that CFD has reasonable accuracy in predicting an optimum L/D ratio and is an adequate tool for the design of the vortex tube. 5.4. Hot gas fraction The inuence of hot gas fraction on the hot gas temperature and cold gas temperature from CFD analysis is shown in Fig. 8(b) and (c) for vortex tube of L/D = 10. The results indicates that maximum hot gas temperature is obtained for lower mass fraction at the hot end and similarly lowest cold gas temperature is obtained at lower mass fraction at cold end (higher mass fraction at hot end). Fig. 10(a) and (b) represents the experimental data of hot gas and cold gas temperatures at dierent hot gas

Fig. 7. Swirl patterns at the inlet zone obtained from CFD for: (a) two numbers of convergent nozzles; (b) single helical circular nozzle; (c) single helical rectangular nozzle; (d) straight six numbers of nozzles; (e) convergent six numbers of nozzles.

fractions and L/D ratios for dc = 7 mm and dc = 6 mm respectively. It is seen that a hot gas fraction of 1215% gives maximum hot gas temperature of 391 K for L/ D = 30 and for cold gas fraction %60% (lowest cold gas fraction possible with the present experimental system) gives lowest cold gas temperature of 267 K for L/D = 35.

6. Secondary circulation ow CFD tence of [13] and dierent studies were carried out to verify the exissecondary circulation ow in vortex tube its inuence on temperature separation for dc/D values. Analysis were carried out for

Table 1 Comparison of nozzle exit velocity and total temperature dierence between cold and hot ends for ve dierent nozzle congurations Nozzle prole Convergent Circular helical Rectangular helical Straight Convergent No. of nozzles 2 1 1 6 6 Mass ow rate at inlet (kg/s) 0.00890 0.00905 0.00892 0.00921 0.00917 Nozzle exit velocity (m/s) 362.8 303.3 310.6 420.4 537.9 Total temperature dierence, DTch (K) 20.2 19.8 19.6 21.6 22.3

12 mm diameter vortex tube for dc/D values of 0.323, 0.417, 0.5 and 0.58 for L/D = 30. We nd that there is secondary circulation ow superimposed on the primary forced vortex for lower values of dc/D as reported by Ahlborn et al. [13]. However we nd that as dc/D value is increased, the magnitude of secondary ow decreases and is completely eliminated for dc/D = 0.58 (cold end diameter = 7 mm). The velocity proles of the ow near the cold end exit for dc/D = 0.323 and dc/D = 0.58 are shown in Fig. 11(a) and (b) respectively. They clearly show the swirling back of the secondary ow near the cold end exit for vortex tube with dc/D = 0.323 and absence of secondary ow for dc/D = 0.58. The stream function plotting for the vortex tube of dc/D = 0.417 near the cold end exit zone in Fig. 12 shows a few of the secondary ow loops as well as the free and forced vortex ow components. We have evaluated the ratios of measured exit temperatures as a function of calculated values for 12 mm diameter vortex tube at dierent values of dc/D based on the analytical model of Ahlborn et al. [14]. The results are shown in Fig. 13. It is observed that for lower values of dc/D, the ratios of exit temperatures are closer to the analytical model but for higher dc/D values, the measured values are much higher than the predicted values. Therefore the analytical model by Ahlborn et al. [14] appears to be valid for the specic vortex tube with low dc/D value of 0.323. The temperature separation in vortex tube with low dc/D values was lower compared to higher dc/D values (Figs. 8(a) and 13). It is to be noted that Ahlborn et al. [13] have observed secondary circulation ow for a vortex tube with a rather low dc/D value of 0.323 and have not evaluated the ow for any other dc/D value. Secondary circulation ow appears to be characteristic of vortex tubes with low dc/D values as conrmed by our studies and could be a performance degrading mechanism. The degradation could be due to transfer of colder uid elements near the cold end exit zone through the swirling secondary loop to the warmer ow region causing decrease in the hot end temperature and transfer of warmer ow elements back to the cold end exit zone causing increase in cold end exit temperature of the ow.

7. Thermal analysis of the optimized vortex tube The performance characteristics of the vortex tubes have been estimated by calculating the coecient of performance (COP) of this system as a heat engine and as a refrigerator. As a heat engine the COP of the vortex tube is given as COPhe mh hh =mi hi 1 ahh =hi COPre mc hc =mi hi ahc =hi 5

As a refrigerator the COP of the vortex tube is given as 6

The COP values of the vortex tube for L/D equal to 10 and 30 as a heat engine and as a refrigerator are shown in Fig. 14. It is observed that the COP of the vortex tube as a heat engine varies from 0.16 to 0.59 for L/D = 10 and varies from 0.16 to 0.46 for L/D = 30. The COP of this vortex tube as a refrigerator varies from 0.42 to 0.83 for L/D = 10 and varies from 0.55 to 0.82 for L/D = 30. This shows that L/D ratio has negligible eect on COP of the vortex tube. The COP values of Carnot cycle for the same temperature conditions as shown in Fig. 14 for L/D equal to 10 varies from 10.5 to 26.5 as a refrigerator and 6.8 to 10.9 as a heat engine. For L/D equal to 30 the COP values varies from 8.6 to 16.9 as a refrigerator and 4.3 to 6.1 as a heat engine. The COP of vortex tube is very low as compared to COP of Carnot cycle. However this is a unique device which produces both heating and cooling eects simultaneously without using any other form of energy than compressed air at moderate pressure. This device can be used eectively in process environments where heating and cooling outputs of vortex tubes can be concurrently used.

8. Conclusion A new approach in optimizing the design of vortex tube has been evolved through CFD analysis using ke turbulence model of the Star-CD code, replacing experimental correlation used hitherto by other researchers. The accuracy of the CFD simulations has been validated by experiments. The swirl velocity, axial velocity and

Total temperature difference (K)

80 70 60 50 40 30 5 15 25 35

Hot gas temperature (K)

dc=7mm(Experiment) dc=6mm(Experiment) dc=7mm(CFD) dc=6mm(CFD) dc=7.5mm(Experiment dc=5mm(Experiment)

360 350 340 330 320 310

dc=7mm(Experiment) dc=6mm(Experiment) dc=7mm(CFD) dc=6mm(CFD) dc=7.5mm(Experiment) dc=5mm(Experiment)

45

55

65

15

25

35

45

55

65

(a)
296 Cold gas temperature (K) 292 288 284 280 276 272 5 15

Hot gas fraction (%)

(b)
Maximum hot gas temperature (K)

Hot gas fraction (%)

dc=7mm(Experiment) dc=6mm(Experiment) dc=7mm(CFD) dc=6mm(CFD)

400 390 380 370 360 350 340 330 320 5 10 15 20 25 30

25

35

45

55

65

(c)

Hot gas fraction (%)

35

40

(d)

Length to diameter ratio (L/D)

282
Minimum cold gas temperature (K)

280 278 276 274 272 270 268 266 5 10 15 20 25 30

dc=7.5mm dc=7mm dc=6mm dc=5mm

35

40

(e)

Length to diameter ratio (L/D)

Fig. 8. (a) Total temperature dierence at dierent hot gas fractions and cold end diameters for L/D = 10; (b) hot gas temperature at dierent hot gas fractions and cold end diameters for L/D = 10; (c) cold gas temperature at dierent hot gas fractions and cold end diameters for L/D = 10; (d) experimental maximum hot gas temperature obtained at dierent L/D ratios and cold end diameters; (e) experimental minimum cold gas temperature obtained at dierent L/D ratios and cold end diameters.

radial velocity components of the ow and the ow pattern have been obtained through CFD, which are dicult to obtain experimentally due to disturbance of ow by measuring probes. The analysis shows that the ow has forced and free vortex components up to stagnation point and temperature dierence between hot and cold gas ow can be maximized by increasing the

length to diameter ratio of vortex tube such that stagnation point is farthest from the nozzle inlet and within the tube. Many investigators have so far determined the optimum nozzle prole and number of nozzles of vortex tube by experimental investigations. The work reported here shows that CFD analysis provides an elegant and

120 Temperature difference (K)

Experiment CFD

100

80

60

40 5 10 15 20 25 30 35 40
Lengh to diameter ratio (L/D)

Fig. 9. Temperature dierence between hot and cold gas for dierent L/D ratios.

400
Hot gas temperature (K)

390 380 370 360 350 340 330 320 0 10 20 30 40

L/D=10 L/D=15 L/D=20 L/D=25 L/D=35 L/D=30

50

60

(a)
295
Cold gas temperature (K)

Hot gas fraction (%)

Fig. 11. (a) Flow pattern near cold end exit for dc/D = 0.323 showing secondary circulation ow; (b) ow pattern near cold end exit for dc/D = 0.58 showing no secondary circulation ow.

290 285 280 275 270 265 0 10 20 30 40 50

60

(b)

Hot gas fraction (%)

Fig. 10. (a) Hot gas temperature at dierent hot gas fractions and L/D ratios at dc = 7 mm; (b) cold gas temperature at dierent hot gas fractions and L/D ratios at dc = 6 mm.

more accurate way to arrive at these critical parameters. For a 12 mm diameter vortex tube this study has shown that swirl generator with six numbers of convergent nozzles gives the best performance.

The critical design parameters of the vortex tubes, namely the cold end diameter (dc) and the length to diameter ratio (L/D) also could be most eectively decided by CFD techniques rather than depending on experimental correlations, which are conventionally used by vortex tube designers. The accuracy of these simulated results for 12 mm diameter tube have been validated by experimental investigations. The CFD and experimental studies have shown that for 12 mm diameter vortex tube, the cold end diameter of 7 mm is ideal for producing maximum hot gas temperature, while cold end diameter of 6 mm is optimum for reaching the minimum cold gas temperature. The investigations have shown that L/D ratio in the range of 2535 is optimum for achieving best thermal performance for 12 mm vortex tube. These optimized vortex tube could produce maximum hot gas temperature of 391 K at 1215% hot gas fraction and a minimum cold gas temperature of 267 K at about 60% cold gas fraction.

Fig. 12. Stream function for dc/D = 0.417 near cold end exit zone.
1.4 1.35 1.3 Th/Tc (Measured) 1.25 1.2 1.15 1.1 1.05 1 1 1.1 1.2 Th/Tc (Ahlborn et al. [14]) 1.3 1.4
dc=4mm: dc/D=0.323 dc=5mm; dc/D=0.41 dc=6mm; dc/D=0.5 dc=7mm; dc/D=0.58

of forced and free vortex ows and with optimal dc/D value, the secondary ow is eliminated which results in its higher temperature separation between cold end and hot end. For the 12 mm vortex tube the optimum dc value is 7 mm corresponding to dc/D = 0.58. The maximum COP of the vortex tube is found to be 0.59 as a heat engine and 0.83 as a refrigerator. Though the COP of vortex tube is low as compared to the COP of Carnot cycle for the same temperature conditions, the thermal separation produced by the vortex tubes can nd ready applications where there are modest heating and/or cooling requirements and ready availability of compressed air.

Fig. 13. Ratio of measured exit temperatures as function of calculated values based on analytical model of Ahlborn et al. [14] for dierent cold end diameters.

References
[1] G.J. Ranque, Experiences sur la detente giratoire avec productions simultanes dun echappement dair chaud et dun echappement dair froid, J. Phys. Radium 4 (7) (1933) 112114. [2] R. Hilsch, The use of the expansion of gases in a centrifugal eld as cooling process, Rev. Sci. Instrum. 18 (2) (1947) 108113. [3] R.G. Deissler, M. Perlmutter, Analysis of the ow and energy separation in a turbulent vortex, Int. J. Heat Mass Transfer 1 (1960) 173191. [4] M. Kurosaka, Acoustic streaming in swirl ow and the RanqueHilsch (vortex-tube) eect, J. Fluid Mech. 124 (1982) 139172. [5] B. Ahlborn, J.U. Keller, R. Staudt, G. Treitz, E. Rebhan, Limits of temperature separation in a vortex tube, Phys. D: Appl. Phys. 27 (1994) 480488. [6] T.T. Cockerill, Thermodynamics and Fluid Mechanics of a Ranque Hilsch Vortex Tube, Masters Thesis, University of Cambridge, England, 1995. [7] A.F. Gutsol, The Ranque eect, Phys.Uspekhi 40 (6) (1997) 639658. [8] A.P. Merkulov, Vikhrevoi Eekt I Ego Primenenie V Tekhnike (Vortex Eect and Its Application in Technique), Mashinostroenie, Moscow, 1969. [9] V.I. Metenin, Investigation of vortex temperature type compressed gas separators Translated from, Z. Tekh. Fiz. 30 (9) (1960) 10951103. [10] W. Frohlingsdorf, H. Unger, Numerical investigations of the compressible ow and the energy separation in

100

Temperature difference in (K)

80 70 60 50 40 30 20 10 0 0 0.1 0.2 0.3 0.4

0.7 0.6 0.5 0.4 0.3 0.2 0.1

0.5

0 0.6

Hot gas fraction (%)

Fig. 14. Coecient of performance of the optimized vortex tube as a heat engine and as a refrigerator for L/D = 10 and L/D = 30 at inlet pressure of 0.5422 MPa (absolute) and inlet temperature of 300 K.

The studies conducted by us have conrmed the presence of secondary ow for vortex tubes with low dc/D values. However CFD analysis and experimental results indicate that secondary circulation ow could be a performance degrading mechanism in vortex tubes. An optimally designed vortex tube has only two regimes

Coefficient of performance

90

Thi (L/D=10) Tic (L/D=10) Thi (L/D=30) Tic (L/D=30) (COP)he (L/D=10) (COP)re (L/D=10) (COP)he (L/D=30) (COP)re (L/D=30)

0.9 0.8

RanqueHilsch vortex tube, Int. J. Heat Mass Transfer 42 (1999) 415422. [11] Star-CD, Version 3.10A: Methodology and User Guide, Computational Dynamics Limited, UK, 1999. [12] H. Takahama, Studies on vortex tubes, Bull. JSME 8 (31) (1965) 433440.

[13] B. Ahlborn, S. Groves, Secondary ow in vortex tube, Fluid Dyn. Res. 21 (1997) 7386. [14] B.K. Ahlborn, J.U. Keller, E. Rebhan, The heat pump in a vortex tube, J. Non-Equilib. Thermodyn. 23 (2) (1998) 159165.