AlAA 93-0076 Effect of Micron-Sized Roughness on Transition in Swept-Wing Flows

R.H. Radeztsky Jr., M.S. Reibert, W.S. Saric Arizona State University Tempe, AZ

S. Takagi National Aerospace Lab0 rat0 ry Tokyo, Japan

31 st Aerospace Sciences Meeting & Exhibit

January 11-14, 1993 / Reno, NV For permission to copy or republish, contact the Amerlcan Institute of Aeronautics and Astronautics 370 L'Enfant Promenade, S.W., Washington, D.C. 20024

Effect of Micron - Sized Roughness on Transition in Swept - Wing Flows

Ronald H. Radeztsky, Jr., Mark S. Reibert, and William S. Saric Mechanical and Aerospace Engineering

Arizona State University Tempe, Arizona, 85287-6106

Shohei Takagi National Aerospace Laboratory

Tokyo, Japan

ABSTRACT

Boundary-layer transition-to-turbulence studies are conducted in the Arizona State University Unsteady Wind Tunnel on a 45-degree swept airfoil. The pressure gradient is designed so that the initial stability characteristics are purely crossflow-dominated. How visualization and hot-

’-’ wire. measurements show that the development of the crossflow vortices is influenced by roughness near the attachment-line. Comparisons of transition location are made between a painted surface (disaibuted 9 pn peaks and valleys on the surface), a machine-polished surface (0.5 pn rms finish), and a hand-polished surface (0.25 rms finish). Then, isolated 6 pn roughness elements are placed near the attachment line on the airfoil surface under conditions of the final polish (0.25 pn rms). These elements amplify a centered stationary crossflow vortex and its neighbors, resulting in localized early transition. The diameter, height, and location of these roughness elements are varied in a systematic manner. Spanwise hot- wire. measurements are taken behind the roughness element to document the enhanced vortices. These scans are made at several different chord locations to examine vortex growth.

NOMENCLATURE

c chordrm] D diameter of roughness [mml -1

k height of roughness [pn] N Re, chord Reynolds number Ra, roughness Reynolds number xu/c transition location x distance. along chord [m] y normal-to-surface coordinate [ml z distance along span [ml a angle of attack

w travelling-wave frwluency

1. BACXGROUM) AND MOTIVATION

linear stability amplification factor: h(u’/u’d, = N

stationary crossflow vortex wavelength

The present investigation is a continuation of previous efforts to determine the fundamental nature of the crossflow instability that is characteristic of the breakdown to turbulence in three-dimensional boundary layers that are characteristic of swept wing flows. Reviews of the current literature and problems are given by Reed and Saric (1989). Arnal(1992). and Saric (1992).

This experiment uses the NLF(2)-0415 airfoil (Somers and Horstmann, 1985) as the basic test configuration. The model has a pressure minimum on the upper surface at approximately x/c = 0.71 (see Figure 1) and thus is an ideal crossflow generator when swept Specially designed wall liners within the wind-tunnel nozzle and test section produce an infmite-swept-wing flow with no spanwise gradients (Dagenhart, 1992). Standard boundary-layer codes (Kaups and Cebeci, 1977)

are used for this configuration. The predicted pressure fields and undisturLw3 boundary-layer profiles are in agreement with the measurements.

With a 45" sweep and a small negative angle of attack, the favorable pressure gradient produces a boundary-layer flow that is sub-critical to Tollmien- Schlichting (T -S) wave formation at moderate chord Reynolds numbers, but produces considerable crossflow. This permits the isolated examination of the crossflow stability problem. The instability at a = -4' was analyzed using the MARIA and SALLY stability codes. At a chord Reynolds number of Re, = 3.8 x 106. the maximum predicted N-factor is N = 16.0 for f = 200 Hz. Dagenhart (1981) indicates that transition may be expected for N- factors in the range from 9 to 11. Thus, for sufficiently high Reynolds numbers (less than 3.8 x le), the transition due to crossflow vortex amplification is expected to occur well ahead of the pressure minimum at XIc = 0.71. The fundamental h e a r stability results for this configuration are complete and include the wavelengths, local growth rates, frequencies, and evolution patterns of the crossflow vortices in the linear range (Saric et al. 1989; Dagenhart et al. 1989, 1990, Dagenhart 1992). In this work, the frequencies of the crossflow instability are observed to establish the existence of the theoretically-predicted moving crossflow vortices in addition to stationary crossflow vortices.

The nonlinear processes that control transition are documented by Kohama et al. (1991). In this work, the stationary crossflow vortex is shown to be the harbinger of transition by causing a high-frequency secondary instability which results from distortions in the steady boundary-layer flow.

One of the key missing ingredients in aJl 3-D boundary-layer experiments is the understanding of receptivify. Here, receptivity refers to the mechanisms that cause environmental disturbances to enter the boundary layer and cause unstable waves (e.g. Saric et al. 1991). Receptivity has many different paths through which to

inuoduce a disturbance into the boundary layer. They include the interaction of freesheam turbulence and acoustical disturbances (sound) with model vibrations, leading-edge curvature, discontinuities in surface

curvature, or surface roughness. Any one or a combmdon of these may lead to unstable waves in the boundary layer. If the initial amplitudes of the disturbances are small they will tend to excite the linear normal modes of the boundary layer.

'd

Whereas linear Stability theory predicts that the travelling crossflow waves are more amplified than the stationary crossflow waves, many experiments observe stationary waves. The question of whether one observes stationary or travelling crossflow waves is cast inside the receptivity problem. Bippes and Muller (1990) and Bippes (1991) describe a serjes of comparative experiments in a low-turbulence tunnel and a high-turbulence tunnel. Their results show that travelling crossflow waves are observed in the high-turbulence tunnel rich in unsteady freesheam disturbances and the dominant structure in a low- turbulence tunnel is a stationary crossflow vortex. Since the flight environment is more benign than the wind tunnel, one expects the low-turbulence results to be more important.

Our previous work showed that detailed measurements within the boundary layer can reveal both the travelling and stationary waves according to h e a r theory. However, our recent work (Kohama et al. 1991) showed that even though the travelling wave is more highly amplified. the mean flow distortion due to the stationary vortex induces a high-frequency secondary instability that leads to transition. Thus, crossflow transition is determined by the characteristics of the stationary wave.

The receptivity problem for stationary vortices is cast withii sources of sheamwise vorticity. We have shown (Saric et al. 1990) that even in the Stokes-flow h i t , the flow around a small roughness withiin a shear layer separates and is a source of vorticity. Thus the present work concentrates on the role of surface roughness in influencing swept-wing transition.

2. RESULTS

2.1 DisPibuted Micron-Sized Rouehness due t~ Surface Finish

The results from this part of the study are obtained

W

2

from naphthalene flow visualization. This technique relies d on the sublimation patterns of the naphthalene to identify

It may be just a coincidence, but the stability amplitude ratios are in the proportion 1 : 18 : 40 while the

'd

-J

the stationary crossflow structure and the location of transition. This technique was calibrated in previous experiments by the use of surface-mounted hot films (Dagenhart et al. 1989: Mangalam et al. 1990) as shown in Figure 2. Surface roughness measurements were done by taking castings of the surface with a dental pattern resin which created a rigid sample suitable for profilometer measurements. Figure 3a shows the surface characteristics of the painted model. This is the surface of all of the previous experiments. The peak-to-peak roughness in this case is around 8 to 10 pm. The paint was removed and the surface was sanded and machine-polished to the level of Figure 3b. Here the surface finish is 0.5 pn rr The model was then subjected to a systematic hand-poLah and the surface finish was reduced to 0.1 pm rms in the mid- chord region and 0.25 pm rms near the attachment line as shown in F i m 3c. ..

Transition measurements are conducted at a number of different chord Reynolds numbers, Re, under the conditions of different surface finishes described above. In Figure 4, the bansition location, x,,/c, is shown to increase with subsequent levels of polishing of the model. The painted-smface results are from Dagenhart et al. (1989). The polished-surface results show quite dramatically the effect of surface fmish and the sensitivity to even small changes in surface roughness. Table 1 is an example from one typical chord Reynolds number.

Table 1. Transition location as a function of roughness height. k is the nominal roughness height in microns, x,Jc is transition location, N(o = 0) is the amplification factor for most amplified stationary crossflow wave, and N is amplification factor for the most amplified travelling wave. Rek is the roughness Reynolds number based on roughness height and the velocity at that height. Re, = 2.7 x IO6

k WI X,/C N(w = 0) N

9 0.40 5.5 7.1 0.30

0.5 0.61 8.4 11.0 < 10-3

0.25 0.68 9.2 11.8 <lW3

initial amplitudes of the surface roughness are approximately in the proportion of 1 : 1/18 : 1/36 near the leading edge. This initially suggests that a qmtitative measure of stabiIity/transition N-factor may be determined from surface roughness.

2.2 lated Roughness Elemen6

The roughness elements are obtained by applying "rub-on" symbols h m transfer sheets commonly used in the graphic arts. Geographics Inc. manufactures a large variety of symbols which includes a convenient range of sizes of circular dots. The dimensions of the applied elements are wefully calibrated by repeatedly applying them to an aluminum block. The applied elements are carefully measured using a dial-indicator displacement gauge and a profdometer. The average height of the elements is 6 pm. Note that earlier reports indicated a height of 9 pn, based on preliminary measurements. The surface of the elements is uneven, with thickness variations on the order of 20%. Taller elements are created by stacking the 6 pm dots. Careful measurements indicate a mean thickness of 6 pn per layer, with very little compression of the lower layers. The sides of the elements are not perfectly straight because the pressure applied during installation spreads the material slightly at the edges.

When a 6 pn roughness element (3.7 mm diameter) was inaoduced at x/c = 0.023, local transition returned to its original upstream location near x& = 0.40. Preliminary reports on this work contained a misprint which identifed the roughness location as x/c = 0.04. The valuexlc = 0.023 reported here is the correct position.

The naphthalene flow visualization reveals that the early transition occurs in the vicinity of the particular stationary crossflow vortex which passes near the roughness location. Roughness elements were placed along the chord direction and it was found that transition was not influenced Thus, the roughness effect seems to be confined to a small region near the attachment line where crossflow disturbances first begin to be amplified (this is discussed below). This justifies a posteriori the use of naphthalene in the mid-chord region.

3

2.2.1. Rounhness Diameter sensitive to the roughness position. For the data shown, - In order to fully characterize the effect of isolated

roughness near the attachment line, a variety of roughness sizes are introduced at x/c = 0.023. Roughness diameters, D, ranged from 0.3 mm to 3.7 mm, which corresponds to 0.08 to 0.5 times the stationary crossflow vortex wavelength, kF. Heights varied between 6 pm and 36 pm. The roughness Reynolds number, Reb based on roughness height and local velocity ranged from 0.12 to 4.5. Flow-visualization measurements reveal a suong dependence of transition location on both the roughness height and the roughness diameter as shown in Figure 5. The transition location moves forward for the larger diameters and heights, and moves downstream as the roughness size is reduced. However, Figure 5 also shows that once the roughness diameter becomes smaller than about 10% of &, the transition location returns to the x/c = 0.7 (pressure minimum) location, and is not influenced by the roughness, even as the height is increased. This may be because at these small diameters, the roughness induces no net streamwise vorticity on the scale of the stationary vortices.

The effect of roughness diameter on the uansition location is repeated at three different unit Reynolds numbers. The results are. shown in Figures 6 and 7 . Figure 6 shows transition location vs. roughness diameter. while Figure 7 shows transition Reynolds number versus roughness diameter. It is again apparent that for small roughness diameters, the transition Re returns to the clean- surface value. In Figure 7 , it is clear that the dependence of transition on Re, is not really a unit-Reynolds-number effect, but instead is an Rekeffect.

These measurements illustrate the importance of spanwise scale, vis A vis with &J, in assessing the role of surface roughness. Note that these conclusions are for Ret < 4.5. Obviously, small diameter but larger roughness heights (like paint specks) could cause early transition.

2.2.2. 3 -0 Roughness Location

The effect of roughness location is determined by systematically applying single 6 pm elements at chord locations between x/c = 0.005 and x/c = 0.045. Figures 8 and 9 clearly show that the transition location is very

- the attachment l i e is a t d c = 0.007 and the neunrll point is at x/c = 0.02. When the roughness is located upstream of the neutral point, the crossflow modes influenced by the roughness will initially decay. so the influence of the roughness is minimized. For roughness too far downstream of the neutral point, the influence is also minimized, with roughness beyond x/c = 0.04 having no effect on the transition location at least for Rek c 4.5. This quantitatively justifies the early comments about the role of roughness in the mid-chord region.

2.2.3. 2 - 0 Roughness

When a long (100 mm) spanwise strip of 6 p n roughness is applied at x/c = 0.023, early transition wedges appear downstream only on the stationary vortices which originate near the ends of the roughness strip. Thii confms the idea that the most important effect of the roughness is to seed the crossflow vortices by injecting streamwise vorticity as the separated region wraps around the edges of the roughness. The wake behind the center regions of the strip is stable and has no influence on the transition location.

2.3 LJ

Insensitivitv to Freemeam Sou nd

Sound, along with 2-D and 3-D roughness (following Saric et al. 1991 for 2-D boundary layers), has no observable effects on transition. Sound pressure levels of up to 95 LB are introduced at all important frequencies, e.g. the frequency of the maximum amplified travelling crossflow wave, the frequency of a destabilized T-S wave, the frequency of the secondary instability, as well as broad-band sound, with negative results in each case. Thii further substantiates the important role of the stationary crossflow wave in causing transition in this flow condition.

2.4 m t m n

Detailed hot-wire measurements are made downstream of the isolated roughness element in order to further quantify the behavior of the roughness-induced vortices. Because the stationary vortices are very weak, a direct measurement of the v’ and w’ velocities is impossible. However, when acting over large distances, these weak vortices produce strong distortions (10% to

b--

4

50%) in the streamwise velocity profile, and this can be The qualitative description of this

roughness. Small discrepancies in the streamwise growth are due to a slight variation of the probe altitude in the 4 ' e a s i l y measured.

follows Kohama et al. (1991). In regions of v' > 0, low momentum fluid near the wall is entrained into the upper regions of the boundary layer. In regions of v' < 0, high momentum fluid is driven downward. Thus, very small v' can produce O(1) changes in mean streamwise velocity. This behavior is responsible for the spanwise periodic streaks in the naphthalene and for the secondary instability leading to transition (Kohama et al. 1991).

Four different chord locations and three different heights in the boundary layer are chosen for hot-wire scans. These sensitive (and tedious) measurements require very long run times. The ASU Unsteady Wind Tunnel does not have an active cooling system, so hot-wire temperature drift could be a potential problem. However, this is obviated by applying a software temperature compensation technique (Radeztsky and Reibert, 1993) during the test runs, and also during calibration of the Wires.

Figure 10 displays the results at xlc = 0.35, and shows a modulation in the streamwise velocity over the entire scan and a strong modulation in the region influenced by the roughness element. Tracing the streamline downstream from the roughness element would locate it at z = 25 mm in Figure 10. The strongest gradients in u' occur well up in the boundary as documented by Kohama et al. (1991). The spatial power spectrum given in Figure 11 shows a vortex wavelength of about 8 mm (when projected from the constant chord line at 45 degrees to the direction perpendicular to the crossflow vortex axis). This agrees with the vortex spacing observed with flow visualization.

various scans. Measurements of the vortex directly behind the roughness element show a significant enhancement of the u' fluctuations. In Table 2, this is seen as a difference in %dmS between the "Enhanced" and "Unenhanced" columns.

The streamwise growth of the vortices can also be seen in Figure 12, which shows a 3-D representation of the hot-wire data taken at ulV, = 0.5. The streamline of the roughness element is centered at z = 25 mm. At xlc = 0.25, the streamwise velocity distortions are already sizeable and the enhanced vortices behind the roughness can already be seen near z = 25 mm. The distortions grow dramatically back to xlc = 0.4, which is near the start of the uansition wedge. One also sees an excitation of neighboring vortices as predicted by Mack (1985).

The eegree of enhancement increases dramatically with the roughness diameter as indicated in Table 3. This is consistent with the variation of transition location with roughness size. These data indicate that the isolated roughness elements directly influence the stationary crossflow modes by providing higher initial amplitudes, leading to earlier local transition.

2.5 Co muan 'son with 2-D Boundm Lavers

In the case of 2-D boundary layers (either unswept wings or swept wings in the mid-chord and weak dpldx) small 3-D roughness does not directly influence the T-S instability. Here, 2-D roughness is more important as a receptivity mechanism (Saric et al. 1991). The von Doenhoff and Braslow (1961) collection of data refers only to cases where the 3-D roughness directly hips a transition wedge. As such, it is basic-state independent Table 2 shows that the streneth of the velocitv

1

fluctuations grows in the saeamwise direction. These values were obtained by subtracting the mean streamwise velocity from each scan, and then computing the rms of the resulting fluctuation data. In Table 2, the "Enhanced" column shows t h e r m fluctuation level for the fmt 50 mm of the scan. This region contains the vortices which originated near the roughness element The "Unenhanced" column shows the rm fluctuation level for the remainder of the scan, i.e. in the region that was unaffected by the

%./

since transition is occurring due to a wake instability of the flow behind the roughness element (they refer to this as T-S but it is neither T-S nor crossflow). This mechanism is operative at the "high Rek values ( x 100). Hence, because much of the data are taken from 2-D boundary layers, only the "direct trip" mechanism is recorded and any influence on the linear stability part of the process is by-passed. The recent experiments of Juillen and Amal (1990) confi i the von Doenhoff and Braslow (1961)

5

Figures 5 , 6, and 7). This is consistent with our original conjecture i.e. as the diameter, D, becomes a small fraction of the wavelength, L-p. the stationary crossflow vortex sees a net zero vorticity from the roughness. If D& exceeds 0.08, we can see the change regardless of roughness heighht An increase in height decreases N.

'e'

.

We can conclude the following: In a low- disturbance environment, stationary crossflow vortices dominate transition via a high-frequency secondary instability (Kohama et al. 1991). The initial conditions (receptivity) are determined by surface roughness and roughness influences linear stability (bence N-factors). Isolated roughness influences transition until the diameter is less than 8% of the crossflow wavelength. Ret in the range of 1 to 4 can change N by 3 or 4. Distributed roughness (random) appears to have the dangerous scales that effect aansition. In a high-disturbance environment, travelling crossflow waves dominate transition (Bippes and Muller. 1990, Bippes, 1991) and the saw-tooth pattern of transition is not smng. One may conjectm that roughness is not as imporrant in this case.

ACKNOWLEDGEMENTS Ld,

The experimental work was supported by NASA- Langley Research Center Grant NAG1-937. Mr. Radeztsky was supported by NASA-Langley Research Center Fellowship NAGl-1111 and Mr. Reibert was suppolted by the Office of Naval Research Fellowship Program. The invaluable help of Mr. D. Clevenger, who polished the model and prepared the roughness samples, and Mr. DJ. On, who measured and analyzed the roughness samples, is gratefully acknowledged.

REFERENCES

Amal, D. 1992. Laminar-Turbulent Transition: hediction, Application to Drag Reduction. AGARD Reporf 786 (Special course on skin friction drag reduction) VKI, BNSS~S.

Bippes, H. 1991. Experiments on Transition in Three- Dimensional Accelerated Boundary Layer Flows. Proc. RAS. Boundary k y e r Transition and Control, Cambridge, UK, ISBN 0 903409 86 0.

Bippes, H. and Muller, B. 1990. Disturbance Growth in an Unstable Three-Dimensional Boundary Layer. Numerical and Physical Aspects of Aerodynamic

..-/

correlation for large roughess in 3-D boundary layers. In contrast, the present results show how roughness influences the linear stability part of crossflow-vortex amplification.

3. CONCLUSIONS

3.1 S ~ e c ific Conclusions From Present ResulQ

We have shown that micron-sized roughness can strongly influence crossflow-dominated transition and the effect of roughness height and roughness diameter is quantitied. This effect is confined to roughness near the attachment line and is not influenced by sound. Moreover, there exis& a critical diameter below which the roughness has no effect.

3.2 Ge neral Co nclusions on Cross flow-Dominata

The a = - 4O condition of the present experiment is completely subcritical to T-S wave growth. The growth is stretched to 71% chord (the pressure minimum) so that the boundary-layer thickness is large (3 mm). On an actual flight airfoil, the pressure minimum would, of course, be forward of this. However, the basic idea has always been that linear-stability phenomena are dependent on Reynolds number and are independent of fetch. Thus, we can compare transition at 40% chord in these experiments with 15% chord (say) in flight because the N-facfors would be the same. Therefore, the final interpretation of our data should be in terms of the change in N and not in the change in location. The utility of these experiments is that we can measure sensitive phenomena in a quantitative way and be able to see the difference between 0.25 pn and 0.5 pm (at the same time of course, we can see that the 9 pn roughness changes N by 3).

We have shown before that when nansition is crossflow dominated, the sfationmy crossflow (C-J?) wave is the dominant instability that causes transition even though theory may predict that travelling crossflow waves may be more amplified (Kohama et al., 1991). The role of 3-D roughness is to provide a source of streamwise vorticity that influences the initial amplitude of the stationary C-F wave. We have found that the small roughness (6 - 9 micron) influences transition until the roughness diameter is 8% of the crossflow wavelength (see

6

Flows, Vol. IV, Ed. T. Cebeci, Springer-Verlag.

Dagenhart, J.R. 1981. Amplified Crossflow Disturbances in the Laminar Boundary Layer on Swept Wings with Suction. NASA TP-1902.

Dagenhart, J.R., 1992. Crossflow Disturbance Measurements on a 45 Degree Swept Wing. PhD Thesis, Virginia Polytechnic Institute and State University, December, 1992.

Dagenhart, J.R., Saric, W.S., Mousseux, M.C., and Stack, J.P. 1989. Crossflow-Vortex Instability and Transition on a 45-Degree Swept Wing. AIAA Paper No. 89-1892.

Dagenhart, J.R., Saric. W.S., and Mousseux, M.C. 1990. Experiments on Swept-Wing Boundary Layers. Laminar-Turbulent Transition, Vol III. Ed. D. Amal, Springer-Verlag.

von Doenhoff, A.E. and Braslow, A.L. 1961. The effect of distributed surface roughness on laminar flow. Boundary Luyer Control, Vol. II. Ed. Lachmann, Pergamon.

Juillen, J.C. and Arnal, D. 1990. Etude exp€rimentale du dklenchement de. la msi t ion parrugositeS et par rainer sur de bord dattaque d'une aile en flkhe en

-d Bncoulement incompressible. CERT/ONERA Rapport Final No 51/5018.35,1990.

Kaups, K. and Cebeci, T. 1977. Compressible laminar boundary layers with suction on swept and tapered wings. J. Aircraft, Vol. 14, p 661.

Kohama, Y., Saric, W.S., and Hoos, J.A. 1991. A High- Frequency Secondary Instability of Crossflow Vortices that leads to Transition. Proc. Roy. Aero. SOC.: Boundnry Luyer Transition and Corurol, Cambridge, UK, ISBN 0 903409 86 0. Submiaed to J. Fluid Mech.

Mack, L.M. 1985. The Wave Pattern Produced by a Point Source on a Rotating Disk. AIAA Paper No. 85- 0490.

Mangalam, S.M., Maddalon, D.V., Saric, W.S., and Agarwal, N.K. 1990. Measurements of Crossflow Vortices, Attachment-Line Flow. and Transition Using Microthin Hot Films. AIM PaperNo. 90-1636, June 1990.

Radeztsky, R.H., Jr. and Reiben, M.S. 1993. A software solution to temperature-induced hot-wire voltage drift. Proc. 3rd Internalional Symposium on Thermal Anemometry, ASME-FED, June 1993.

4 Reed, HL. and Saric, W.S. 1989. Stability of Three-

4/ 1990. Dimensional Boundary Layers. Ann. Rev. Fluid Mech. vol. 21,235.

Saric, W.S. 1992. Laminar-Turbulent Transition: Fundamentals. AGARD Report 786 (Special course on skin friction drag reduction) VKI, Brussels, Mar 1992.

Saric, W.S., Dagenhart, J.R., and Mousseux, M.C. 1989. Experiments in Swept-Wing Transition. Numerical and Physical Aspects of Aerodynamic Flows, Vol. IV, Ed. T. Cebeci, Springer-Verlag. 1990,359 (fist appeared Jan. 1989).

Saric, W.S., Krutckoff, T.K., and Radeztsky, R.H. 1990. Visualization of Low-Reynolds-Number Flow Fields Around Roughness Elements," Bull. Amer. Phys. Soc.. vol. 35.2262.

Saric, W.S., Hoos, J.A., Radeztsky, R.H., and Kohama, Y. 1991. Boundary-layer Receptivity of Sound with Roughness. Boundary Layer Stability and Tranrition to Turbulence, FED-Vol. 114, E& D.C. Reda, H.L. Reed, R. Kobayashi, ASME.

Somers, D. M. and Horstmann, K-H. 1985. Design of a Medium-Speed Natural-Laminar-Flow Airfoil for Commuter Aircraft Applications. DFVLR-IBI29- 85/26.

-

7

i i lU. cz 0.25 ' Location Enhanced Unenhanced -

%chord ii/Ue & a/Ue % dms ' ___- -~ 25 0.25 5.9 0.23 4.6 30 0.21 4.8 0.17 4.2 35 0.26 5.9 0.23 3.0 40 0.27 8.7 0.20 2.8 -

Vortices downstream of applied roughness.

Diameter r / c = 0.30 mm ?i/& %hS 2.0 0.50 4.4 2.5 0.52 5.7 3.0 0.48 5.6

r/c=0.35 - ii/G %U;ms 0.53 6.7 0.53 7.6 - 0.54 8.9

ii/U, c 0.50 Diameter Z/C = 0.30 z/c = 0.35

2.0 0.48 3.9 0.52 3.7 2.5 0.50 3.6 0.51 3.2 3.0 0.45 3.8 0.54 3.2 3.5 0.43 3.4 0.53 3.1

mm iilu, %U;ms n/u, You',,,

, ~~ ~~ n 30 I 0.47 I 6.4 I 0.43 I 3.4 H

Location %chord

_ _ _ _ - 1 - I -

35 1 0.57 I 11.2 1 0.53 I 3.1 40 I . 0.57 I 11.8 I 0.57 I 4.9

Enhanced 1 Unenhanced ii/& I % Gms 1 iT/Ue I % Gms

Location Enhanced Unenhanced %chord z / U e %Gms ii/ue % qms

25 0.74 5.4 0.73 2.4 30 0.72 8.0 0.69 3.0 - 35 0.76 11.5 0.72 3.2 40 0.78 9.4 0.73 5.2

3.5 I 0.47 1 6.4 1 0.57 I 11.2

2.6 x 10'. Roughness height = 6pm.

Undisturbed vortices.

i i /U . c 0.75 Table 3. Variation of u' with roughness diameter. Re, =

W

Table 2. Streamwise variation of u'. Re, = 2.6 x lo6. Roughness element with D = 3.7mrn. H = 6pm is at x / c = 0.023.

8

CP

-2.0 .5 .o .5 0 .5

1 .o 1.5

X/C

Figure 1. NLF(2)-0415 pressure distribution in wind "le1 at m = -4". Hollow symbols indicate uppa-surface pressures.

c 0

3

0.80

0.60

0.40 T A

o.20L--- 0.00 1 .s 2.0 2.5 3.0 3.5

R,x 10.

Figure 2. Calibration of naplhalene flow-visualization technique.

J 9

5

E 5 0

-5

-10

- 2.

-1 5 E 500 1wO 1500 2000 25w

Figure 3a. Profilometer measurement of NLF(2)-0415 sur- faceroughness. Paintedsurface-filtered2Opm to 1500pm, rms = 3.30pm.

-20 0

2 ( P I

-5

-1 5

500 1000 15w 2000 25w

Figure 3c. Profilometer measurement of NLF(2)-0415 sur- face roughness. Hand polish - filtered 20pm to 1500pm. rms = 0.121un.

-20 0

z frm)

10 "i 5

-10 ;-:m -1 5 - 2wO 2500 0 5w 1wO 15M)

-20

2 (pml

Figure 3b. Profilometer measurement of NLF(2)-0415 sur- face roughness. First polish - filtered 20pm to 1500pm, rms = 0.509pm.

0.9 .L./ y..,...,.. , I . . , , , . . : - Final Pdish (0.25pm) : - First Polish ( 0 . 5 ~ 1 ) : 0.8

73

0.5 - 2

0.4 -

Figure 4. Napthalene visualization results showing aver- age transition location vs. chord Reynolds number for three different surfacefinisha. Paintroughnessis measured peak- to-peak. Polishedroughness is rms.

0.9 - I r - I I , 8 , I , , , , I , , I , 3 . , I " . ' I - " ' I - ' I 1 " " l " " :

H ( W ) H/6,, Re, 0.8 - 6 .0068 .12

A 12 0.014 50 8 v 18 0.020 1.1

+ 24 0.027 2.0 0 30 0.034 3.1 0 36 0.041 4.5

-

0.7 - -

- E 0.6 -

0.5 - - Y s

-

A v A

rn m -

+ I i v 0.4 -

0

0.3 - -

0.2 " ~ ~ 1 " ' " ' " " " ' " 1 " " 1 ' ~ " 1 ' ~ ' ' 1 " " 1 1 " '

0.90 . . . . , . , . . , . . . . / . . . . , . . . . I . , , ' ,

0.80 - - Ree=3.4x10'

2'oo 1.90 7 1 .a0

1.70

1 .so 9 1.50

5 1.40

120

1.10

1 .a0

0.90

0

E 1.30

0 8

Wl Figure 7. Variation of transition Reynolds number with roughness diameter at three chord Reynolds numbers.

11

E 0.60:

s - 0.50 - 0.40 -

0.30 -

0.20 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

%d(")

Figures. Variationof transitionlocation withroughness x/c at three chord Reynolds numbers.

--C ReC=3.4x10' -c- Re,=3.0x10e + Re,=2.6x1O0

u., . , I , ..- 0.0 O S 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Xe(%C)

Figure 9. Variation of transition Reynolds number with roughness z/c at three chord Reynolds numbers.

0.001 " " I ' ' ' I " " ' ' ' I 0 25 50 75 100

2 (mm)

Figure 10. Spanwise hot-wire scans at x/c = 0.35. Constant-y scans at G/U, of 0.25.0.50.0.75. A single roughness with D = 3.7mm, H = 6pm is at x/c = 0.023.

element

12

xi (mm-')

Figure 11. Normalized spatial power spechum of streamwise velocity from spanwise hot-wire scan.

Figure 12. Spanwise hot-wirescans showing streamwise vortex growth. Constant-y scans at G/Ue = 0.50. A single roughness element with D = 3.7mm, H = 6pm is at z/c = 0.023.

13