' AlAA 98-2882 The Use of Heavy Gas for Increased Reynolds Numbers in Transonic Wind Tunnels J. B. Anders and W. K. Anderson NASA Langley Research Center Hampton, VA A. V. Murthy Aerotech, Inc. Hampton, VA 20th AlAA Advanced Measurement and Ground Testing Technology Conference June 15-18 , 1998 I Albuquerque, NM For pecmlssiacr to copy or npubJbh, contact tho Amdarn ~nrtftudo.f krorrrvtla ad Aabonrutlcs 1801 AkxmdnB ell Drlw, Sulte 50, Reston, VA 201 91 I American Institute of Aeronautics and Astronautics AIAA-9 8-2 8 82 THE USE OF HEAVY GAS FOR INCREASED REYNOLDS NUMBERS IN TRANSONIC WIND TUNNELS J. B. Anders,”’*W . K. Anderson,”’+a nd A. V. Murthyh5 ”NASA Langley Research Center, Mail Stop 170, Hampton, VA 23681-001, bAerotech, Inc. Hampton, VA 23666 Abstract The use of a high molecular weight test gas capable of producing the required flight-level to increase the Reynolds number range of transonic Reynolds numbers at transonic speeds are cryogenic wind tunnels is explored. Modifications to a small tunnels. Two such tunnels in the U. S., and one in transonic wind tunnel are described and the real gas Europe are available for commercial testing, but properties of the example heavy gas (sulfur wind tunnel testing at 100 degrees Kelvin is far from hexafluoride) are discussed. Sulfur hexafluoride is routine and requires special care with regard to shown to increase the test Reynolds number by a instrumentation and model construction. These factor of more than 2 over air at the same Mach cryogenic facilities rely on a combination of number. Experimental and computational pressure increased pressure and reduced temperature to distributions on an advanced supercritical airfoil increase the density and reduce the viscosity of the configuration at Mach 0.7 in both sulfur hexafluoride test gas (nitrogen). This technique can produce free- and nitrogen are presented. Transonic similarity stream unit Reynolds numbers in excess of 100 theory is shown to be partially successful in million per foot and, augmented by the additional transforming the heavy gas results to equivalent benefit that the lower speed of sound at cryogenic nitrogen (air) results, provided the correct definition temperatures results in reduced drive horsepower and of gamma is used. model loads, offers the only currently viable solution for high Reynolds number testing. However, the cost Introduction and difficulty associated with cryogenic testing, and the unlikely possibility of constructing additional The shortfall in Reynolds number in current cryogenic facilities to relieve the crowded test U. S. transonic ground test facilities is well known schedules has initiated research at NASA into (Mack, et al.’), and with the advent of the proposed, alternate techniques for increasing test Reynolds so-called “megaliners” (600 to 800 passenger numbers. transonic, transport aircraft) this shortfall is likely to One technique that may hold promise is that grow significantly in the coming years. Test data of increasing the Reynolds number through the use needed by the designers of these new large-scale of a high molecular weight test gas. The advantages aircraft will require new, higher Reynolds number of such a technique are that testing can be done at test facilities, and recent estimates indicate that ambient temperatures, and existing tunnels can be large-scale, high Reynolds number, transonic wind retrofitted to use “heavy gas”. This latter advantage tunnels will be extremely expensive to build and becomes even more significant in view of the high operate. All of this is occurring in an era of cost of constructing new facilities. increased commercial competition to reduce the The use of heavy gas to increase Reynolds design cycle time and lower the cost of developing number is an old idea dating back to work by Hube?, new transport aircraft, and in a time of declining Schwartzberg3, vonDoenhoff, et al.4 , and Chapman5. research budgets. The only facilities currently Chapman’s work was an exhaustive examination of candidate gases that could be combined to produce * Senior Member AIAA gas mixtures with a ratio of specific heats (y) near 9 Associate Fellow AIAA 1.4. Later work by Pozniak6, Treon, et a17, Yates and Copyright 0 1998 by the American Institute o f Aeronautics and Astronautics, Sandford’, and others focused on relating results in Inc. No copyright is asserted in the United States under Title 17, US Code. gases where gamma did not equal 1.4 to equivalent The US Government has a royalty-free license to exercise all rights under the copyright herein for Government purposes. All other rights are reserved by results in air. Results from these earlier studies were . the copyright owner. 2 American Institute of Aeronautics and Astronautics inconclusive, and the technique has lain essentially schematic of the modified tunnel with the additional dormant for the past 20 years. However, in 1991, systems required for heavy gas operation denoted. Anderson' conducted a computational study of transonic airfoils in sulfur hexafluoride and found that I I transonic similarity scaling could be used to relate I ASNFAr, LGYAZSE R ' 'cHcP7I7M I' L BINEGR TEST INJECTION SECTION PORTS results in SF, to equivalent air results for primarily inviscid flows, as long as the proper definition of gamma was used. Anderson also found that as viscous effects became more pronounced, and as the flow became more compressible, the scaling was SFc Return less effective. Later, Bonhaus, et a1.l' applied I' \ Anderson's code to two multi-element high-lift HEAT EXCHANGER configurations in SF, and concluded that transonic similarity scaling was adequate for such flows. As a result of Anderson's work, NASA began Fig. 1. Schematic of 0.3-Meter Transonic a program in 1991 to convert a small transonic wind Cryogenic Tunnel with modifications for heavy tunnel to SF, operation in order to provide some gas operation. experimental confirmation of Anderson's results, and to further define the range of application of transonic Figure 2 shows the operating envelope of the similarity theory to heavy gas flows. A progress tunnel. Although the Reynolds number range for SF, report on this program has been given by Anders". is approximately half that of cryogenic nitrogen, the The current paper will discuss the results of an initial upper limit approaches 60x106/ft., substantially study of a supercritical airfoil at transonic speeds in greater than for ambient temperature air operation. SF, compared with nitrogen results using transonic similarity scaling. Computational results for both Model SF, and air will be compared to the experimental data. For the purposes of this paper air and nitrogen are assumed to be the same gas. Methods ExDerimental Facilitv The test facility used for the current study is the 0.3-Meter Transonic Cryogenic Tunnel at the Langley Research Center. This facility, described in some detail in Mineck and Hill", was used to develop the cryogenic test concepts leading to NASA's National Transonic Facility (NTF). The facility operates at pressures up to 6 atmospheres, temperatures from 100" to 300" K, and Mach numbers from 0.15 to 1.0. The test section is approximately 13" x 13" and the upper and lower walls are adaptive. The modifications required for operation with SF6 included a gas reclamation unit for charging and reclaiming the test gas, a gas analysis unit for real-time monitoring of gas composition, a gas warning system for personnel safety, and a The model was tested over a fairly limited specially designed heat exchanger. These angle-of-attack range (from +1.0 degrees to -1.0 modifications are discussed fully in Anders", and degrees), and the upper and lower tunnel walls were with the exception of the custom-designed heat contoured to minimize tunnel wall interference exchanger, all of these units are off-the-shelf, effects". The majority of the data were taken in the commercially available systems. Figure 1 shows a Mach number range from 0.70 to 0.72. The Reynolds 3 American Institute of Aeronautics and Astronautics number based on airfoil chord was 30 x106, although a2 -49.9051433 N-m4/k2 = a limited amount of data was obtained at 15 ~10,. b2 5.485495~10N-~-m 4/k2- K = The accuracy of the measured static airfoil c2=- 2.375924505~10N~-m 4/kg2 pressures is estimated as f0.5%. The accuracy of the a3 4.124606~10N-~-m 7/kg3 = quoted Mach number is f0.002, and the accuracy of b3 = -3.340088~10"N -m7/kg3- K the angle-of-attack settings is f0.05 degrees. Figure 3 c3 2.819595 N-m7/kg3 = shows an example of the repeatability of the airfoil a, -1.612953~10"N -m"/kg4 = surface pressure measurements in SF,. b, = 0 c, 0 = , a, -4.899779~10-"N -ml3/k2 = -1 5, , , , , , , , , , , , , . , , , , , , b, = 1.094195~10-"N -m13/kg5-K c, -3.082731~10"N -m13/kg5 = -1 k 6.883022 = T, 318.8 K = -0 5 ... d = 3.27367367~10m~3 /kg Anders" discusses the real gas behavior of cP O SF, and concludes that the thermal imperfections of the gas, while not large at typical wind tunnel 05 stagnation pressures, are, nonetheless, significant. Also, the caloric imperfections are shown to be quite large, and the variable specific heats must be accounted for in calculating the isentropic flow properties. In addition, Anders" concludes that small 0 02 08 1 O xic O amounts (less than 10%) of air and water vapor contamination of the test gas have little or no effect Fig. 3 Repeatability of SF, pressure distributions, Re,=30x106, M=0.72, a = 1.0 deg. on the free-stream flow properties, but do effect the mixture viscosity slightly. In the present experiment the test gas composition was continuously monitored Characteristics of Sulfur Hexafluoride during tunnel runs and indicated an SF, purity of approximately 98%. Jenkins', has developed a computer code to calculate the isentropic flow Sulfur hexaflouride (SF,) is a colorless, properties of SF, and this code was used for the data odorless, non-toxic gas with a molecular weight of reduction of all the experimental results presented 146. It's principal commercial use is as a dielectric here. in high-voltage switchgear for arc suppression, and it Assuming for the moment that SF, is a is readily available from a number of manufacturers. perfect gas, it is easy to show that the Reynolds It is essentially inert at laboratory temperatures, and, number increase for SF, over air at the same Mach although it has no ozone depletion potential, it is a number and total pressure can be written a d greenhouse gas. SF, is both calorically and thermally imperfect, but its properties are well-known and documented. Equipment for handling SF, (condensers, vaporizers, storage tanks, pumps, etc.) are manufactured commercially for power companies and are readily available. The principal hazard from where MWt = the molecular weight. this gas is that of an asphyxiant since it displaces air. Similar expressions can be developed for the The equation of is well-represented dynamic pressure and the required drive horsepower: by: -kFc RT 5 ui+biT+cie P= -v-+d J i 1 4 (v-d) where R = specific gas constant and the various constants are given as: 4 American Institute of Aeronautics and Astronautics Although sulfur hexafluoride was one of the candidate gases originally proposed by Chapman, -1 5 , , , , , , , , , , , , , , , new refrigerant gases such as R134A (the replacement gas for Freon- 12) have become common since that time and may, in fact, be preferable due to environmental (greenhouse) and cost concerns. -0 However, SF, is representative of the class of high molecular weight gases that can, as shown above, cP increase the Reynolds number and decrease the dynamic pressure and drive horsepower. These 0 characteristics may be particularly attractive for selected, existing transonic wind tunnels to provide an enhanced Reynolds number test range. Table 1 compares some of the physical properties of SF, with air and with two of the new refrigerant gases. 0 02 04 06 08 1 xic Technically, the ratio of specific heats (y) is not a constant for the three heavy gases listed in Table 1, but for the purpose of estimating the Reynolds number increase, it is treated as such. For all other purposes in this paper, the ratio of specific heats is treated as a variable. 1 5 " " " " " " " " " " ' 02 04 06 08 xic (b) N * Fig. 4. Pressure distributions on test airfoil in sulfur hexafluoride and nitrogen, M= 0.70, Re, = 30 x lo6. * p in lb, sec/ft2x108 Figures 5a, b, and c show a direct comparison between the two gases at three angles-of- Table 1. Gas characteristics. attack. The agreement on the lower surface and on the rear of the upper surface, where the local Mach Results and Discussion numbers are low, is fairly good, as would be expected for a nearly incompressible flow. However, ExDerimental Results on the forward part of the upper surface, as angle-of- attack increases and compressibility effects become Typical pressure distributions over the test larger (the local Mach number upstream of the shock supercritical airfoil in both SF, and cryogenic N2 are for nitrogen reaches a maximum of 1.3), there are shown in figures 4a and 4b. There are clear significant differences, especially in the region of the differences in the pressure distributions between the shock. Obviously, thermodynamic differences two gases, especially for the cases at higher angles- between the two gases become important for of-attack where a shock forms on the forward part of compressible flow. The shock location on the airfoil the airfoil upper surface. The rear of the upper and in nitrogen always appears downstream of the lower surfaces show little effect from either gas or location in SF, at the same Mach number. angle-of-attack. 5 American Institute of Aeronautics and Astronautics Computational Method and Results 1 1 ................I j -1 -1 .................;. ............... ................. ................ The computations have been done using the unstructured grid Navier Stokes ~olverw'~it h suitable modifications to properly account for the non-ideal gas behavior of SF,. Figure 6 shows examples of the computational results at Re, = 30 x 10, and a = 0.75 degrees compared with the measured pressures for both SF, and N2. The agreement between computation and experiment is considered to be quite good except for a consistent overprediction of the upper surface pressures downstream of the shock. It should be pointed out that the computation results in Fig. 6b are actually for air rather than nitrogen. This was done as a matter of convenience since for all practical purposes the two gases are aerodynamically identical. 0 02 04 06 08 1 xic 0 02 04 06 08 1 (b) a=0.50 deg. xic -1 5 , , , , , , , , , , , , , , , , , , , 0 02 04 06 08 1 0 02 04 06 08 1 xic xic (c) a=1.0 deg. (b) Nz Fig. 5. Comparison of Nz and SF, pressure Fig. 6. CFD comparison with experimental data, distributions at M= 0.70, Re,= 30x10,. Re,=30x106, M = 0.70, a=0.75 deg. 6 American Institute of Aeronautics and Astronautics The agreement shown in figure 6a gives Scaling Amlied to ExDerimental Results some assurance that the real gas behavior of SF, is correctly captured in the computations Typical transonic similarity scaling variables for the current experiment are given by: Transonic Similaritv M,, = 0.70 M,,, = 0.72 According to inviscid, small disturbance ?/ = 1.04 A = 0.967 transonic similarity theory’,, the flows between air and SF, will be equivalent provided that the Using the similarity scaling laws just described, the similarity parameters are equal. nitrogen data at M=0.70 is compared in Figures 7a, b, and c with SF, data at the similarity Mach number 1-M: 1-M: of M = 0.72. The C, values have been scaled using the parameter A defined above. As the figure illustrates, the comparison is good over most of the heavy gas nitrogen airfoil, with some differences on the upper surface, where 2 = t/c = max. thicknesdairfoil chord especially in the vicinity of the shock at the highest angle-of-attack. Comparisons at Re,=l5 x 1O 6 (not For a given profile where the thickness, 2, is shown) indicate a similar difference in the region of fixed, an appropriate definition of y may be used the shock. In fact, for all cases in the current with the above equation to determine a freestream experiment where a shock was present, the shock Mach number in SF, so that the similarity parameter location in SF, (at the similarity Mach number) was will match that of air. After the fact, the lift and always slightly downstream of the shock location in pressure coefficients can be corrected by: N,. This difference was so consistent in the - A experiment that it can not be attributed to ‘1,nitrogen - ‘1,heavy gas - experimental inaccuracy. It is likely that there are A ‘p,nitrogen- ‘p,heavy gas differences in the viscous shock-boundary layer and the parameter A is given by interaction process between the two gases, and evidence of this can be found in the following comparisons with computations. Figures 8a and 8b are plots showing SF, and N, experimental results at the similarity Mach number, along with computational results for air and Anderson’ demonstrated computationally SF,, again at the similarity Mach number. The that inviscid flows in SF, can be scaled to give vertical scale has been expanded to illustrate that excellent agreement with air provided that the proper the computations show exactly the same trend in the definition of gamma is used. This is given by’ region of the shock as the experimental data. That is, the SF, calculations at the similarity Mach number predict a shock location slightly downstream of the shock location for air (nitrogen in the case of the where a = speed of sound and h = enthalpy. experimental data). Also, note that downstream of At a given freestream Mach number in air, the shock the pressure coefficients are slightly more the Mach number in SF, is determined by first using negative for SF, than for air (nitrogen) in both the the freestream pressure and temperature in SF, to computations and the experiment. The reasons for determine y’. The new Mach number can then be these differences are not entirely clear, but determined using the similarity parameter given Anderson’ noted this same effect in his earlier above. computational study and indicated that the SF, Using the above definition of y’ in boundary layer was appreciably thinner than the air conjunction with the transonic similarity laws, boundary layer, which could result in less upstream inviscid computational results obtained in SF, have movement of the shock due to displacement effects. been demonstrated to consistently scale to yield A few comments in the next section on viscous excellent agreement with computations in air over a similarity may shed additional light on some of the wide range of pressures and temperatures’. However, viscous mechanisms that may be responsible for application of these scaling laws for computed these differences. viscous flows has been only partially successful. 7 American Institute of Aeronautics and Astronautics -1 2 , , , , , , , , , , , , , , , , , , , -1 1 ................ :. ................ :. ................ :. ................ :. ............... -1 ................. j .................; . ............... ;.. ............... ;. ................ .I -0 5 -0 9 CP -0 8 cP O -0 7 05 -0 6 -0 5 1 5 ~ " " "0"2" " " "0"4 ' l 06 08 -0 4 xic 0 02 O xlc0 6 08 1 (a) a=O.O deg. a) alpha=0.50 deg -1 3 -1 2 -1 ................. ................. ................. ;. ............... -1 1 -0 5 -1 cP cP O -0 9 05 -0 8 -0 7 1 ................ :. ................: . ................ :. ................ :. ................ -0 6 15 " ' ~ " ' ~ " ' ~ " ' ~ " ' 02 04 06 08 -0 5 0 02 04 xic 06 08 1 (b) a=0.5 deg. b) alpha=l.O deg -15 , , , I , , , I , , , I , , , I , , , Fig. 8. Comparison of scaled experimental and computational results. Rec=30x106, MN2=0.70, MsF6=0.72 Comments on Viscous Similaritv The fact that transonic similarity theory fails in the shock-boundary layer interaction region is not unexpected since it is based on small disturbance potential flow theory. Clearly, if the boundary layers are different in the two gases, then an inviscid transformation process will not account for this difference. 1 5 ~ " " " " " " " " " ' 1 Calculations of the growth of the 02 04 06 08 displacement thickness on the upper surface of the xic (c) a=1.0 deg. airfoil for both air and SF, (again, at the similarity Mach numbers of 0.70 for air and 0.72 for SF,) Fig. 7. Transonic similarity scaling applied to indicate that the air boundary layer grows at a experimental results, ReC=30x1O6M, N2=0.70, significantly greater rate (also see Anderson'). Figure MSF=O~ - 72 9 shows that for the present airfoil at roughly the 8 American Institute of Aeronautics and Astronautics mid-chord location, the air displacement thickness is Concludin? Remarks roughly 10% greater for air than for SF, It is tempting to suggest that this difference Previous computational studies showed that in 6*, which results in an effective difference in the transonic similarity theory is successful in airfoil thickness between the two gases, may be transforming transonic airfoil pressure distributions in accounted for by simply using an effective 2 in the SF, to equivalent results in air when the flows are transonic similarity relations presented earlier. primarily inviscid. However, this earlier work also However, it is easy to show that if 2 for the airfoil in indicated that the scaling is less successful when air is increased by lo%, then the similarity Mach viscous effects are included. The current number for SF, increases from 0.72 to approximately experimental results confirm the viscous 0.736, which would result in a further downstream computations and show that in the region of a shock- movement of the SF, shock location relative to the boundary layer interaction the scaling always yields nitrogen case, therefore increasing the disagreement a shock location for heavy gas that is slightly between the two gases in the shock region. downstream of the location for nitrogen (air). This mismatch is small, however, and the scaling may be adequate for many applications where only small regions of compressible flow are present. The interesting questions that remain to be answered for heavy gas flows are: how do the viscous processes of transition, separation, and shock-boundary layer interaction differ from air and what methods can be used to provide viscous similarity? These questions must remain the subject of future studies. References [l] Mack, M. D., and McMasters, J. H., “High Reynolds Number Testing in Support of Transonic Airplane Development”, AIAA Paper 92-3 982, June 1992. Fig. 9. Computation of displacement thickness growth in air and SF,, Re,=30x106, MSF6=0.72, [2] Huber, P. W., “Use of Freon-12 as a Fluid for M,,=0.70. Aerodynamic Testing”, NACA TN 1024, 1946 . The above calculation of the displacement [3] Schwartzberg, M. A., “A Study of the Use of thickness growth assumes that both boundary layers Freon-12 as a Working Medium in a High-speed are fully turbulent. However, in reality, it is certainly Wind Tunnel”, NACA RM L52J07, 1952. possible that the transition process is different between the two gases. Unfortunately, transition was [4] vonDoenhoff, A. E., Braslow, A. L., and not measured in the present experiment, and one can Schartzberg, M. A., “Studies of the Use of Freon-12 only speculate regarding it’s effect. Early transition as a Wind Tunnel Test Medium”, NACA TN 3000, for SF, could offset, or even reverse the results 1953. shown in figure 9, and/or it is possible that the response of the boundary layers to the adverse [5] Chapman, D. R., “Some Possibilities of Using pressure gradient generated by the shock is Gas Mixtures Other Than Air in Aerodynamic somewhat different. In order to answer these and Research”, NACA TN 3226, 1956. other questions regarding the viscous flow characteristics of heavy gas, detailed boundary layer [6] Pozniak, 0. M., “Investigation Into the Use of investigations need to be conducted examining the Freon12 as a Working Medium in a High-speed viscous phenomena of transition, separation, and Wind Tunnel”, The College of Aeronautics, shock-boundary layer interaction. Cranfield, Note No. 12, 1957. 9 American Institute of Aeronautics and Astronautics [7] Treon, S. L., Hofstetter, W. R., and Abbott, F. T., “On the Use of Freon-12 for Increasing Reynolds Number in Wind-Tunnel Testing of Three- Dimensional Aircraft at Subcrtical and Supercritical Mach Numbers”, AGARD CP-83, 1971. [8] Yates, E. C., and Sanford, M. C., “Static and Longitudinal Aerodynamic Characteristics of an Elastic Canard-Fuselage Configuration as Measured in Air and in Freon-12 at Mach Numbers Up to 0.92”, NASA TN D- 1792, 1963. [9] Anderson, W. K., “Numerical Study of the Aerodynamic Effects of Using Sulfur Hexafluoride as a Test Gas in Wind Tunnels”, NASA TP-3086, 1991. [lo] Bonhaus, D. L., Anderson, W. K., and Mavriplis, D. J., “Numerical Study to Assess Sulfur Hexafluoride as a Medium for Testing Multielement Airfoils”, NASA TP-3496, 1995. [ll] Anders, J. B., “Heavy Gas Wind Tunnel Research at Langley Research Center”, ASME Paper 93-FE-5, 1993. [12] Mineck, R. E., and Hill, A. S., Calibration of “ the 13 by 13-inch Adaptive Wall Test Section for the Langley 0.3-Meter Transonic Cryogenic Tunnel”, NASA TP 3049, 1990. [13] Mears, W. H., Rosenthal, E., and Sinka, J. V., “Physical Properties and Virial Coefficients of Sulfur Hexafluoride”, J. Phys. Chem., 73, 2254, 1969 [14] Jenkins, R. V., “Program to Calculate Isentropic Flow Properties of Sulfur Hexafluoride”, NASA TM 4358. 1992 [15] Anderson, W. K., and Bonhaus, D. L., “An Implicit Upwind Algorithm for Computing Turbulent Flows on Unstructured Grids,” Computers and Fluids, Vol. 23, NO. 1, 1994, pp. 1-21 [16] Liepmann, H. W., and Roshko, A., “Elements of Gasdynamics”, John Wiley & Sons, Inc., c.1957 10 American Institute of Aeronautics and Astronautics