Document
NASA Technical Memorandum 87316
Turbulent Dispersion of the Icing
Cloud From Spray Nozzles
Used in Icing Tunnels
fNASft-TH-87316) 1UHBU1EH1 DISPERSION OF THE N86-31588 ICING CLOUD FEOB S P B A Y NOZZLES OSED IN ICING CSCL 21E T U N N E L S ( N f t S A ) 11 P „ , Unclas G3/07 U3538 C. John Marek and William A. Olsen, Jr.
Lewis Research Center
Cleveland, Ohio
Prepared for the
Third International Workshop on Atmospheric Icing of Structures
Vancouver, Canada, May 6-8, 1986
NASA
TURBULENT DISPERSION OF THE ICING CLOUD FROM SPRAY NOZZLES USED IN ICING TUNNELS John Marek and William A. Olsen, Jr.
National Aeronautics and Space Administration Lewis Research Center Cleveland, Ohio 44135 ABSTRACT 1. The air jet used to atomize the water into acceptably small droplets.
2. The wake of the spray bar.
To correctly simulate flight in natural icing 3. The turbulent flow from upstream of the conditions, the turbulence in an icing simulator must be as low as possible. But some turbulence is spray bars.
required to mix the droplets from the spray nozzles and achieve an icing cloud of uniform liquid water The goal of any icing simulator is to obtain the content. The goal for any spray system is to obtain widest possible spray cloud for the lowest possible the widest possible spray cloud with the lowest pos- turbulence. The nozzles are spaced so that the sprays sible turbulence in the test section of a icing overlap to produce a fairly large uniform LWC.
tunnel.
The experience at the NASA Lewis Research Center This investigation reports the measurement of in the Icing Research Tunnel (IRT) was with a short- turbulence and the three-dimensional spread of the chord blunt-trailing edge spray bar. The horizontal cloud from a single spray nozzle. The task was to spray bars and spray nozzles of the IRT were developed determine how the air turbulence and cloud width are in 1950. In the new AWT, low loss coefficient aero- affected by spray bars of quite different drag coef- dynamic spray bars were proposed, it was expected ficients, by changes in the turbulence upstream of that the large scale turbulence originating off the the spray, the droplet size, and the atomizing air. separated wake of the IRT spray bar would result in larger amounts of lateral mixing and was required to An ice accretion grid, located 6.3 m downstream produce a wide dispersion of the spray. In addition spray dispersipn data taken in the IRT showed a strong of the single spray nozzle, was used to measure cloud spread. Both the spray bar and the grid were located effect of the tunnel velocity on spread. The spread in the constant velocity test section. Three spray of the wake from an airfoil is not a function of the bar shapes were tested: the short blunt spray bar tunnel velocity, therefore, we wanted to study the used in the NASA Lewis Icing Research Tunnel, a thin spray/wake interaction.
14.6 cm chord airfoil, and a 53 cm chord NACA 0012 airfoil. The AWT was proposed to be a Mach 1 tunnel. It contains a six to one contraction to the test section At the low airspeed (56 km/hr) the ice accretion so that for a given test section velocity, the veloc- r pattern was axisymmetric and was not affected by the ity across the spray bars located upstream of the shape of the spray bar. At the high airspeed contraction is more than double that in the IRT tunnel (169 km/hr) the spread was 30 percent smaller than at with its 14:1 contraction. Because of the high veloc- the low airspeed. For the widest cloud the spray ities, the pressure loss across the spray bars was of bars should be,located as far upstream in the low concern, and it will be necessary to make them as velocity plenum of the icing tunnel. streamlined'as possible without adversely affecting the mixing of the spray.
Good comparison is obtained between the cloud spread data and predictions from a two-dimensional The goals of this test were to: (1) examine the cloud mixing computer code using the two equation effect of spray bar shape on spray dispersion, (2) turbulence (keg) model. examine the effect of upstream turbulence on disper- sion, and (3) compare the experimental results with numerical predictions of the cloud spread.
Introduction Three spray bar shapes were tested: the short (14.6 cm) blunt IRT spray bar, a thin 14.6 cm chord Refurbishment of the NASA Lewis Altitude Wind airfoil, and a 53 cm chord NACA 0012 airfoil. All Tunnel (AWT), proposed for completion in the early spray bars were tested at airspeeds of 56 and 1990's, was planned to include the capability of con- ducting icing research along with aerodynamics, pro- 169 km/hr and two median volume droplet sizes, 12 and pulsion and acoustic studies. Since ice accumulation 22 urn. The IRT spray bars, upstream of the test sec- tion, were used to increase the upstream turbulence on aircraft and engine surfaces can seriously degrade level by flowing air from the IRT spray nozzles.
performance, icing tests are an important aspect of Longitudinal and transverse turbulence intensity the development and verification tests of aerospace measurements were made with single and cross hot flight systems. The ultimate goal of a ground based test facility is to effectively simulate actual icing wires, respectively.
conditions encountered by an aircraft in flight.
A spray bar and nozzle system is required to Numerical Computer Model produce a uniform cloud. Consideration needs to be given to the spacing of the nozzles, the spacing of To predict the dispersion of the spray a computer code which was developed for gas turbine combustors the spray bars, and the shape of the spray bars.
Turbulence in an icing simulator must be as low as was used. The code had to take into account the effects of air turbulence on the spray dispersion.
possible. But some turbulent mixing of the droplets Many techniques exist in the literature, but the one from the spray nozzles is required to achieve an icing used here was developed by Shuen, Solomon, Zhang, and cloud of uniform liquid water content (LWC). Figure 1 Faeth (Ref. 1), and is based on the model of Gosman shows a nozzle and a spray bar in a tunnel flow. The turbulence arises from three sources: and loannides (Ref. 2). The model, called the where Up is the droplet velocity. The droplet is Stochastic Separated Flow (SSF) model, resulted in assumed to interact with the same eddy as long as the the best agreement with experimental data presented time is less than t or t .
in Ref. 1. i e p The gas phase velocity and turbulence levels were Droplet trajectories were determined using a computed using the 6ENMIX computer program of (Ref. 5).
Lagrangian formulation of the governing equations.
Source terms were included to consider droplet-gas The droplet momentum equation is coupling using the droplet source in cell approach of (Ref. 6). A simpler model assuming constant turbul- dU 3pC D
} U - U 1=1, 3 ence properties was tried in this work but with unsat-
isfactory.results. The interaction of the droplets
-
with the atomizing air jet is important and is com- puted with the GENMIX program.
where The assumed initial conditions were: a sonic Cn, drag coefficient velocity for the atomizing air for all pressures above g gravity the choked condition and a droplet velocity equal to the liquid exit velocity. The turbulence properties dp droplet diameter were taken to be those of case 2 from Ref. 1 which was for the exit conditions of a particle-laden tube.
p droplet density Experimental Facility Up-j droplet velocity in ith direction The tests were conducted in the test section of Up total droplet velocity the NASA Lewis IRT. The IRT is a closed loop refrig- erated atmospheric total pressure wind tunnel. The p density of air test section is 1.83 m high and 2.75 m wide. The maximum test section velocity is 482 km/hr. Natural U\ instantaneous air velocity in ith cloud conditions were simulated by an array of 77 direction air-atomizing nozzles located upstream of the contraction.
total air velocity U Cloud spread was established by measuring the ice where the droplet motion is based on the instantaneous shapes accreted on a pair of 0.32 by 2.54 by 75 cm velocity of the continuous phase determined below from bars mounted perpendicular to each other with the the turbulence properties.
0.32 cm edge facing into the flow. The ice thickness at each bar position after a period of time is Droplets were tracked through the airstream directly proportional to the average cloud liquid eddies in a stochastic manner, interacting with a water content at that cloud location.
single eddy for the shorter of two times - (1) the eddy life-time t and (2) the residence time tp e for the passage through the eddy. In the calculations The experimental configuration for the dispersion presented here, two thousand droplets of a given size studies is shown in Fig. 2. The test spray bar was were tracked and statistically averaged to obtain the located at the start of the tunnel test section and the ice accretion crossbars were located at the end of mean spray properties.
the test section. In this way the spray cloud was The mean turbulence was determined using the two located entirely within the test section so that the equation turbulence (keg) model of Launder and air velocity around the entire spray cloud was con- Spalding, Ref. 3. The velocity u' of each eddy at stant, and the air turbulence was at a minimum. At the start of a droplet-eddy interaction was determined 169 km/hr the velocity across the spray bar was high enough to simulate the AWT spray bar velocity at the by choosing a Gaussian random number with a standard 1 2 deviation of (2k/3) ' and mean velocity Ui. The maximum proposed test section velocity.
characteristic size L of an eddy is given as e The IRT spray bars, located 11 m upstream of the L = Cp3/4 3/2 e k /e test section entrance, were used to increase the upstream air turbulence intensity by flowing air from the IRT spray nozzles. The ice accretion crossbars where were located 6.32 m downstream of the single nozzle.
This was the maximum distance downstream within the k turbulent energy = 3/2 U' constant velocity section.
e turbulent dissipation The crossbars were supported by a grid composed of 0.32 cm thick by 5.0 cm deep bars spaced 15.25 cm where Cy = 0.09. The eddy lifetime t is computed e apart in both the horizontal and vertical directions.
with This grid was used for locating the center of the spray and for supporting the ice accretion cross made up of 0.32 cm thick by 2.54 cm deep bars. After an 1/2 icing run the cross was removed and the ice accretion thickness measured at locations 2.54 cm apart along the vertical and horizontal directions.
The droplet residence time through an eddy is After all of the icing tests were completed for each spray bar, the support grid was removed and the traversing aerodynamic probes were installed to measure the velocity and turbulence profiles across the center of nozzle jet. Measurements were made at a tunnel temperature of 10 °C without water spray or the atomization air and the shape of the spray bar heating of the atomization air. The traversing probes does not influence the spreading. This result was consisted of a total and static pressure probe, a surprising in that it was expected that the large thermocouple, a single hot wire and a cross hot wire scale turbulence orginating from the separated wake of the IRT spray bar would result in larger amounts to measure the axial and normal components of veloc- ity. The first station was 0.6 m downstream of the of lateral mixing. The wide dispersion found in the IRT was thought to be produced by the blunt trailing nozzle and the probes were traversed horizontally.
The second station was 5.81 m downstream of the edge of the IRT spray bar.
nozzle. The traversing mechanism was manually moved between stations one and two after a test series. In Fig. 7(b) there appears to be a large'differ- ence in the center!ine accretion level between the A photograph of the thin 14.6 cm chord spray bar airfoils and the IRT spray bar, but the spray is not mounted in the tunnel is shown in Fig. 3. The thick- much wider for the IRT spray bar. Because of the ness of this airfoil was 3.2 cm which was the smallest circular pattern, differences in ice thickness at the size that could contain the nozzle air, water and larger radius greatly affects the centerline values.
steam lines.
For example an increase of 0.5 mm at 30 cm would require a decrease of 1.5 mm at the center to maintain the same total mass of ice accreted.
Photographs of all three spray bars are shown in Fig. 4. The NASA IRT spray bar was 7.62 cm thick with At 169 km/hr the spray dispersion was greatly a chord length of 14.6 cm. It consisted of a shaped skin which was wrapped around the thin 14.6 cm air- reduced. The ice accretion profiles for the 12 ym foil. The shape of the IRT spray bar is shown in volume median drop size are shown in Fig. 8. The Fig. 1. The trailing edge was a quickly tapered 30° spray pattern for the IRT spray bar was elliptical half-angle which resulted in flow separation. However but again not in the direction expected. At the the IRT spray bar does not cause strong backflow that higher velocity the spreading was greater along the would cause spray droplets to impinge upon the spray IRT spray bar rather than in the normal direction bar and freeze. None of the three spray bars tested indicating that the wake was producing a sheltered caused ice to form on the spray bars. The NACA 0012 zone. The spreading along the IRT spray bar was the airfoil was chosen to have low aerodynamic drag and same as the lower velocity case. The spreading in to have the same chord length as the proposed AWT the normal direction was nearly the same as the air- foils. The spray pattern for the airfoils appeared spray bar design (53 cm).
axisymmetric. The wake of the IRT configuration allowed increased spreading along the spray bar and The NASA Lewis standard icing spray nozzle is shown in Fig. 5. This nozzle produces a narrow spray did not increase the spreading normal to the spray bar. At a tunnel velocity of 169 km/hr the spreading angle of finely atomized droplets. The drop sizes reported were obtained from previous measurements was reduced in the normal direction by 30 percent taken with laser diagnostics. The atomizing air is from the lower velocity case for all three spray bars.
introduced at a pressure above the choking condition for the exit orifice. The air leaves the nozzle and The 22 m volume median drop size data are shown u in Fig. 9. The curves are similar to the 12 pm data.
interacts with the water jet at sonic speeds. The The IRT spray bar curves are very elliptical. There atomizing air interacts both with the water spray and is some variation in the peak levels between the NACA the tunnel air in the wake of the nozzle face to form a complicated flow configuration. The droplets 0012 and the 14.6 cm airfoil. Some of the error is caused by the ice accretion cross not being exactly in breakup by going through the violence of a normal shock at the exit of the nozzle. the center of the pattern. For the steep gradients obtained, an error of 4 cm in position can produce a difference of 1 mm in ice accretion. The conclusions Results stated in the above paragraph apply to this data also.
Tests were completed to determine the spray dis- Cloud spread was much wider at the low airspeed persion width as a function of droplet size and tunnel of these tests than at the high airspeed. Therefore velocity, and to evaluate the effect of the spray bar for the widest cloud of uniform LWC and the lowest shape and turbulence on dispersion. The ice accretion turbulence, spray bars should be located as far on a grid was measured in order to determine the dis- upstream as possible in the low velocity plenum persion of the spray. Then the velocity and turbul- chamber of an icing tunnel.
ence distribution were measured. Comparisons were In order to determine the effect of upstream made between numerical predictions of dispersion using a Monte Carlo model and the experimental data. turbulence on spray dispersion, the IRT strut atomiz- ing air was used which flowed out of 77 IRT nozzles The ice accretion profiles for the IRT spray bar upstream of the bellmouth. The ice accretion curves both along the spray bar (vertical) and normal to it are shown in Fig. 10. The catch was reduced and more (horizontal) are shown in Fig. 6 for a tunnel velocity frost was produced outside the main area. The amount of 56 km/hr. The spray pattern was axisymmetric so of frost produced, however, did not account for the the distribution and spread are very close in both decrease in catch. It is not known whether the directions. The spray with the larger (22) median increased turbulence affects the catch efficiency enough to alter the mass balance. The data for all volume drop size did not disperse as much as the 12 ym the spray bar shapes produced similar results. The droplets. The data have not been corrected for spread of the spray is not affected by the upstream changes in catch efficiency because it should not turbulence. Some data at the higher velocity were affect the indicated spread of the cloud. The catch taken but the reduction in catch was very high and is efficiency increases by 10 percent for the change from 12 to 22 m drops. not presented.
v Turbulence and velocity measurements were taken The two airfoils also produced axisymmetric ice with a single and crossed hot wires. The probes at accretion patterns at the 56 km/hr tunnel velocity.
the 0.6 m position could be traversed out of the spray The data for the three spray bars are shown in Fig. 7.
At the low tunnel speed, the mixing is dominated by bar wake to sample the upstream (freestream) condi- The interaction of droplets with a turbulent x tions. The value of the turbulence is the RMS veloc- airstream is very complex which emphasizes the need ity divided by the local mean velocity. Figure 11 for a numerical model. The numerical results have presents the effect of IRT strut air on tunnel turbu- been plotted along with the experimental data for the lence. At 56 km/hr the upstream turbulence increased 56 km/hr velocity on Fig. 17. The agreement with the from-0.6 to 3.5 percent. At 169 km/hr the tunnel experimental data is good for both droplet sizes.
turbulence increased from 0.8 to 1.2 percent. The The numerical predictions have been normalized by the data from the single and cross wires agreed. The peak of the experimental curve. In the numerical axial and vertical turbulence levels agreed well calculations only a single dropsize was used, whereas indicating that the fluctuations were nearly iso- in the experimental data the spray contains a wide tropic. A spectrum analysis of the signals indicated range of drop sizes. All 'three spray bar shapes that no large waves were present. gave the same dispersion at low velocities so that the spreading was governed by the atpmization air.
At higher tunnel velocities the contribution of No attempt was made to adjust the initial conditions the strut-air to the total turbulence was much lower. to determine the sensitivity of the results. The Assuming that the magnitude of the velocity fluctua- measurement station was located 1975 nozzle diameters - tions from the IRT strut air remains the same, downstream of the nozzle exit. Because the GENMIX increasing the velocity by a factor of three would code is a forward marching (parabolic) code, it could decrease the turbulence intensity by the same factor. not handle the recirculation zones behind the IRT : The effect of tunnel velocity on turbulence intensity spray bar. The calculations were performed assuming is shown in Fig. 12. At a tunnel velocity of a positive freestream velocity with a 5 percent tur- 300 km/hr the contribution of the atomizing air to bulence level.
the turbulence intensity is insignificant, but at low The comparison between the numerical results and speeds (below 150 km/hr), the increase in turbulence could significantly affect the aerodynamic measure- the experimental data at 169 km/hr is shown in ments. The difference in contraction ratio between Fig. 18. The numerical results agree very well with the IRT and the proposed AWT tunnels would also affect the data for the airfoils. Again the numerical, pre- .
the results. dictions were normalized this time by a value between the peaks of the airfoils. Reduction in the profile for the IRT spray bar was a three-dimensional effect Measurements of the velocity and turbulence at a and because the code was two-dimensional, it was not distance 0.6 m downstream of the experimental IRT able to predict this. The numerical code does predict spray bar is shown in Fig. 13. This distance was the collapse of the spray with increased tunnel veloc- chosen to be far enough away from the trailing edge to ity. This effect is computed for a jet in Abramovich, have pressure recovery of the flow. As the atomizing air pressure is increased, the wake of the spray bar Ref. 6.
is filled in the neighborhood of the nozzle, making the configuration a propelled body with a negative drag coefficient. It is interesting that the spread Conclusions: of'the spray bar wake equals that of the atomizing air jet at this condition. 1. Cloud spread was much wider at the low air- speed of these tests than at the high airspeed.
The wake turbulence level for the IRT spray bar Therefore for the widest cloud of uniform liquid water is high with a,peak of 13 percent. But with atomizing content (LWC) and the lowest turbulence, spray bars air the turbulence is dominated by the jet reaching a should be located as far upstream as possible in the level of 20 percent for air pressures required for low velocity plenum chamber of the icing tunnel.
large drops and 45 percent for air pressures required for the small ones. 2. At the low airspeed (56 km/hr) the ice accre- tion pattern (i.e., cloud spread) from the sing.le nozzle was axisymmetric and not affected by the shape A comparison of velocity profiles and the turbu- lence distribution for the three spray bars is shown of the spray bar. This result was true for the 12 and 22 vim droplet spray tests. The dispersion was domi- in Fig. 14. The velocity and turbulence profiles were much wider for the IRT spray bar. The wake tur- nated by the atomizing air jet.
bulence was three times larger, which should result in greater mixing. The drag coefficient of the IRT 3. At the high airspeed (169 km/hr) the ice accretion pattern was not axisymmetric. The cloud spray bar was 18 times that of the NACA 0012 airfoil.
spread more along the spray bar than perpendicular to it. Furthermore, the cloud spread along the IRT spray An unusual result was that the thin 14.6 cm air- bar which is blunt with a high drag coefficient, was foil produced a larger drag coefficient than the much broader than along the low drag spray bars. These longer 53 cm NACA 0012 airfoil. Although the 14.6 cm results were true for both droplet sizes tested. This airfoil was not separated, there was more flutter in does not impact the IRT tunnel operation because the the attached tufts indicating possible transition.
maximum airspeed at the spray bars is only 35 km/hr.
The turbulence measurements at 5.81 m are shown in Figs. 15 and 16 for the IRT and NACA 0012 spray bar 4. At the low airspeed, the upstream turbulence could be increased from 0.5 to 3.5 percent by turning shapes respectively. No velocity data are presented because a significant velocity difference in the wake on the air to the IRT spray bars. In spite of this large increase, the cloud spread did not change was not seen at this axial position. At low velocity for the IRT spray bar with no atomizing air the tur- noticeably. The catch however was reduced. This demonstrates that the turbulence generated upstream bulence wake appears almost doubly humped. As the .
atomizing air is increased the dominance of the jet was not important to mixing.
is seen. Surprizingly at 169 km/hr velocity, the 5. Upstream turbulence did not affect the cloud atomizing air jet was dissipated by 5.81 m and the profiles are nearly identical. This occurred for spread.
both spray bars.
6. The predictions of the stochastic separated 3. Launder, B.E.; and Spalding, D.B.: Mathematical flow (SSF) computer model compared well with the Models of Turbulence. Academic Press, 1972.
experimental data.
4. Pantankar, S.V.; and Spalding, D.B.: Heat and Mass Transfer in Boundary Layers. 2nd ed., REFERENCES International Textbook Company, London, 1970.
1. Shuen, J.S., et al: A Theoretical and Experi- 5. Crowe, C.T.; Sharma, M.P.; and Stock, O.E.: The mental Study of Turbulent Particle-Laden Jets, Particle-Source-in Cell (PSI-CELL) Model for Gas- NASA CR-168293, November 1983.
Droplet Flows. J. Fluids Eng., vol 99, no. 2, June 1977, pp. 325-332.
2. Gosman, A.O.; and loannides, E.: Aspects of Computer Simulation of Liquid-fueled Combustors. 6. Abramovich, G.N.: The Theory of Turbulent Jets.
J. Energy, vol. 7, no. 6, Nov.-Dec. 1983, MIT Press, 1963.
pp. 482-490.
7. Boldman, D.: Personal Communication, NASA Lewis Research Center, 1985.
INITIAL PARTICLE MOMENTUM ^TURBULENCE FROM UPSTREAM Figure 1. - Flowfield-droplet interactions to consider.
0. 3C. II JM ^ B •" U-0.6 m
-
TUNNEL SPRAY !AY SUPPORT t : BARS USED TO TO GRID C5 3.
% 1.83m — f=| - CONTROL UP- O STREAM -ICE ACCRETION
"*
TURBULENCE E CROSS BARS
I
O \ \ \ \\ ' HOT WIRE AND TOTAL O PRESSURE TRAVERSING 14:1 CONTRACTION Figure 2. - IRT Tunnel with single nozzle mounted in test section and ice accretion grid. Tunnel is 1.83 m by 2.75 m.
ORIGINAL PAGZ ij OF POOR QUALITY Figure 3. - Photograph of airfoil mounted in tunnel looking upstream.
ORIGINAL
OF POOR "OUAUTY Figure4. - Photograph of the three airfoils tested-IRT, 0.146 meter, andthe NACA0012.
HIGH PRESSURE ATOMIZING AIR ORIFICE 0.32 cm DIAM AMBIENT AIR 5UU - 1UUU K \ \ \tl\l ENTRAINMENT \ \ WATER TUBE 0.066 cm ID -«\ P\ \
\ K
^7=7^ —* — -H.-'-f/W/l"*^' WATER
9®&!&
£^Z^ — — / L.
WATER TUBE -/ DROPLETS -« 5 nrm Figure 5. - NASA - Lewis standard icing spray nozzle.
,_ ALONG SPRAY BAR NORMAL TO SPRAY BAR o o LU O -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 NORMAL TO SPRAY BAR (a) 12 Micrometers dropsize.
(b) 22 Micrometers dropsize.
Figure 6. - Spray dispersion from IRT spray bar at 56 km/hr. Tunnel temper- ature, -13 °C; 3 min. sprays.
14.6cm NACA 0012 IRT ct: o LU O -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 POSITION ALONG SPRAY BAR, cm (a) 12 Micrometers droplet size. (b) 22 Micrometer droplet size.
Figure 7. - Comparison of spray dispersion from three spray bars at 56 km/hr Tunnel temperature, -13 °C; 3 min. sprays.
14.6 cm NACA 0012 4r- o: o o O -40 -20 0 20 40 -40 -20 0 20 40 POSITION ALONG SPRAY BAR. cm Figure 8. - Comparison of spray dispersion from 3 spray bars at 169 km/hr. Drop size, -12 urn; tunnel temperature, -13°C; 1-min sprays.
NACA 0012 14.6cm IRT o; o -60 -40 . -20 0 20 40 60 -60 -40 -20 0 20 40 60 POSITION ALONG SPRAY BAR, cm Figure 9. - Comparison of spray dispersion from 3 spray bars at 169 km/hr drop size, -22 urn, tunnel temperature, -13 °C; 1-min sprays.
NO UPSTREAM AIR PRESSURE 0.6% TURBULENCE 544 kPa AIR PRESSURE, 3.5% TURBULENCE 4r— E 3 o; o -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 POSITION ALONG SPRAY BAR, cm (a) IRT spray bar. (b) NACA 0012 airfoil.
Figure 10. - Effect of upstream turbulence on spray dispersion at 56 km/hr.
Dropsize -12 micrometers. Tunnel temperature, -13 °C, 3min. sprays.
•4- ' NO STRUT AIR CD O 544 kPa STRUT AIR g.
, LU OQ o; I I I 20 40 -60 -40 -20 0 20 40 -60 -40 -20 0 POSITION NORMAL TO AIRFOIL, cm (a) Tunnel velocity, 56 km/hr. (b) Tunnel velocity, 169 km/hr.
Figure 11. - Effect of upstream spray bar flow on tunnel turbulence at 0.6 m downstream of NACA 0012 airfoil.
O NO UPSTREAM STRUT AIR D 544 kPa STRUT AIR OPEN-DATA OF THIS REPORT CLOSED - DATA FROM ref. 7 \ \ LU \ O CO Qi 100 200 300 400 TUNNEL VELOCITY, km/hr Figure 12. - Test section turbulence.
AIR CD PRESSURE, kPa n (v>(\ U. UtO -0.272 A/I A ? 167 040 L. 101 | 120 A ft 1 i I i — i J - 40 \ \ 100 — ' S \ i
i "
— \ 2L 30 80 , \ ' ' uJ- AM ° 60 ff\ \ - It \ 1 ° // S\ m
\X i <h\ \
"10
- fV-
\ - — \-J 1 K-1
1 1 1 1 o
0 -20 0 20 40 -4 0 -20 0 20 41 -4 POSITION NORMAL TO SPRAY BAR, cm Figure 13. - Effect of nozzle air pressure for IRT spray bar at 0.6 m downstream of trailing edge at 56 km/hr.
AIRFOIL CD 14.6cm 0.007 NACA 0012 0.006 IRT 0.112 — <u — o
n
E O 160 ~ f '
a
3 Z3 UJ CO — 0£.
1 1 1 1 i
i?n -40 -20 0 20 40 -40 -20 0 20 400 POSITION NORMAL TO SPRAY BAR, cm Figure 14. - Effect of airfoil shape on tunnel flow 0.6 m downstream of the trailing edge, 169 km/hr - no nozzle air pressure.
AIR PRESSURE, c kPa o l_ o> — 0 a.
LJ-T O
a
CO -40 -20 0 20 40 -40 -20 0 20 40 POSITION NORMAL TO SPRAY BAR, cm (b) Tunnel velocity, (a) Tunnel velocity, 169km/hr.
56km/hr.
Figure 15. - Effect of nozzle air pressure on turbulence for IRT spray bar at 5.81 m downstream.
AIR PRESSURE, kPa - 0 o> o.
O CQ o: I -40 -20 0 20 40 -40 -20 0 "20 40 POSITION NORMAL TO AIRFOIL, cm (a) Tunnel velocity, (b) Tunnel velocity, 56km/hr. 169knWhr..
Figure 16. - Effect of nozzle air pressure on turbu- lence for NACA 0012 airfoil at 5.81 m downstream of the trailing edge.
NORMAL TO SPRAY BAR ALONG SPRAY BAR NUMERICAL MODEL o O -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 POSITION, cm (a) 12 Microns diameter. (b) 22 Microns diameter.
Figure 17. - Comparison of numerical model with experimental spray dis- persion for IRT spray bar at 56 km/hr. Tunnel temperature, -13°c, —3-min-Spr.ays.
NACA 0012 14.6cm I— IRT NUMERICAL o o o -40 -20 0 20 40 -40 -20 0 20 40 POSITION NORMAL TO SPRAY BAR, cm (a) 12 Micrometers drop (b) 22 Micrometers drop size. size.
Figure 18. - Comparison of numerical results with experimental data at 169 km/hr.
2. Government Accession No. 3. Recipient's Catalog No.
1. Report No.
NASA TM-87316 5. Report Date 4. Title and Subtitle
Turbulent Dispersion of the Icing Cloud From Spray
Nozzles Used 1n Icing Tunnels 6. Performing Organization Code
None 8. Performing Organization Report No.
7. Author(s)
.C. John Marek and William A. Olsen, Jr. E-3047
10. Work Unit No.
9. Performing Organization Name and Address 11. Contract or Grant No.
National Aeronautics and Space Administration
Lewis Research Center
Cleveland, Ohio 44135
13. Type of Report and Period Covered Technical Memorandum 12. Sponsoring Agency Name and Address
National Aeronautics and Space Administration
14. Sponsoring Agency Code
Washington, D.C. 20546
15. Supplementary Notes Prepared for the Third International Workshop on Atmospheric Icing of Structures, Vancouver, Canada, May 6-8, 1986.
16. Abstract
To correctly simulate flight 1n natural 1c1ng conditions, the turbulence.1n an
1c1ng simulator must be as low as possible. But some turbulence 1s required to
mix the droplets from the spray nozzles and achieve an 1c1ng cloud of uniform
liquid water content. The goal for any spray system 1s to obtain the widest
possible spray cloud with the lowest possible turbulence 1n the test section of
a 1c1ng tunnel. This Investigation reports the measurement of turbulence and
the three-dimensional spread of the cloud from a single spray nozzle. The task
was to determine how the air turbulence and cloud width are affected by spray
bars of quite different drag coefficients, by changes 1n the turbulence upstream
of the spray, the droplet size, and the atomizing air. An 1ce accretion grid,
located 6.3 m downstream of the single spray nozzle, was used to measure cloud
spread. Both the spray bar and the grid were located 1n the constant velocity
test section. Three spray bar shapes were tested: the short blunt spray bar
used 1n the NASA Lewis Icing Research Tunnel, a thin 14.6 cm chord airfoil, and
a 53 cm chord NACA 0012 airfoil. At the low airspeed (56 km/hr) the 1ce accre-
tion pattern was ax1symmetric and was not affected by the shape of the spray
bar. At the high airspeed (169 km/hr) the spread was 30 percent smaller than at
the low airspeed. For the widest cloud the spray bars should be located as far
upstream 1n the low velocity plenum of the 1c1ng tunnel. Good comparison 1s
obtained between the cloud spread data and predictions from a two-dimensional
cloud mixing computer code using the two equation turbulence (keg) model.
17. Key Words (Suggested by Authors)) 118. Distribution Statement
Sprays Unclassified - unlimited
STAR Category 07
Dispersion
Turbulence
Modeling
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of pages 22. Price-
Unclassified
Unclassified
*For sale by the National Technical Information Service, Springfield, Virginia 22161 National Aeronautics and SECOND CLASS MAIL Space Administration Lewis Research Center ADDRESS CORRECTION REQUESTED UVMAIL Cleveland. Ohio 44135 Official Business Penalty for Private Use $300 Postage and Fees Paid National Aeronautics and Space Administration NASA-4S1