You are on page 1of 7

Wear 258 (2005) 545–551

Rain erosion resistance characterizations


Link between on-ground experiments and in-flight specifications
A. Déom∗ , R. Gouyon, C. Berne
Office National d’Etudes et de Recherches Aérospatiales, 29 Avenue de la, Division Leclerc, 92320 Châtillon, France

Received 22 December 2003

Abstract

With the increasing velocity of vehicles and the fact that the vehicles must fly even in rainy conditions, environmental problems such as
rain erosion become more important. Precise specifications are now being set for the rain erosion of materials in the vehicle design.
Different attempts have been done to predict the rain erosion resistance of materials from their mechanical properties but this will give
only the order of magnitude of the resistance, which is generally insufficient. Thus this resistance must be known more accurately, that
is possible only by experiments. To perform this, different apparatus are existing: rotating arms, linear tracks, water jet generators. Each
apparatus present advantages and disadvantages. For example, with rotating arms the advantage of using real water droplets is counter-
balanced by the disadvantage of the existing tangential force which can affect the measurements. With tracks, very severe vibrations are
induced. With water jet generators, there is no tangential force, but the use of a jet instead of a droplet necessitates the knowledge of
equivalence laws. Each apparatus has its own characteristics: water droplet diameter, water concentration, which cover a part of possible
rain specifications. In-flight specifications are characterized by generally one rain intensity, a time duration, and a droplet diameter distribu-
tion. The link between on-ground conditions and in-flight specifications must be established to determine, from the rain erosion resistance
measurements performed on specimens of the material, if the structure built with this material will withstand the in-flight rain erosion
specifications.
We will present the results of our recent studies concerning the possibilities of linking on-ground characterizations obtained with water-jet
apparatus to in-flight specifications.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Rain erosion; On-ground experiments; Water-jet experiments; In-flight specifications

1. Introduction than before to a very small change of thickness as it can be


produced by erosion. These new specifications make nec-
The possibility of in-flight damage due to rain erosion essary new studies in the field. Rain erosion in-flight spec-
is an important point taken into account since many years. ifications for a system (aircraft, missile, . . .) are given by
We can see for example Figs. 1 and 2 photographies of rain a required time of resistance of a part of the system (win-
damaged equipment published in the proceedings of a rain dow, dome, . . .) when it encounters at a given velocity water
erosion conference dated on 1965 [1]! Since this date, the ve- droplets coming from a rain characterized by its rain inten-
locities of the aircrafts and missiles increased largely, what sity in mm/h. Furthermore, this rain is composed of a large
increased dramatically the possibility of erosion damage. number of droplets of different diameters. The distribution
Moreover, new equipment such radomes are more sensitive in diameter of the droplets is dependent on the intensity
of the rain.
∗ Corresponding author. Tel.: +33 1 46 73 48 76; fax: +33 1 46 73 48 91. It is very difficult and expensive to make in-flight experi-
E-mail address: deom@onera.fr (A. Déom). ments to characterize the rain erosion resistance of a system.

0043-1648/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2004.09.053
546 A. Déom et al. / Wear 258 (2005) 545–551

Fig. 1. Rain damaged navigational light of Javelin aircraft.

One of the main reason is the difficulty to measure the char- 2. Link between on-ground experiments and in-flight
acteristics of the rain (distribution of the droplets in diameter specifications
and water concentration) and the intrinsic variability of such
a meteorological phenomenon. 2.1. Analysis of the distribution of droplets in a natural
On-ground rain erosion characterizations are more conve- rain
nient but whatever the apparatus used, whirling arm, water-jet
generator, . . . the experiments are made with given diameters Real rain is characterized by a specific drop diameter dis-
or with diameter distributions of the droplets largely different tribution which can be described approximately by the for-
from those of actual rains. mula of Marshall and Palmer:
Our attempt was to try to link the measurements from an
on-ground apparatus: the water generator MIJA with in-flight Nd = 8000 exp[−4.1(d/I 0.21 )] (1)
rain specifications. where Nd is the number of drops per unit of volume and unit of
diameter interval δd, d, diameter of the spherical drop (mm),
and I intensity of rain (mm/h)
The product Nd δd is then the number of drops in 1 m3 of
air with a diameter between d and d + δd. Fig. 3 shows the
distribution of drops versus drop diameter for rain intensities

Fig. 3. Comparison of drop distribution for rain rates of 100 and 10 mm/h
Fig. 2. Sandwich type radome of Hawker Hunter after flight in rain. following the Marshall and Palmer law.
A. Déom et al. / Wear 258 (2005) 545–551 547

each diameter can be calculated:


πDi2
Ndr = Nd × × Vg × t (2)
4
Table 1 shows an example of calculation of the actual
number of drops encountered by a 10 mm diameter surface
going through a rain of 100 mm/h rain intensity at 300 m/s
during 10 s. The step of diameter δd used for the computation
is 0.25 mm.
According to Field [2], there is a relation between the
diameter of the drop and the velocity for different impacts
leading to the same degradation of an infrared material:
Vd2 /Vd1 = (d1 /d2 )1/3 (3)
Fig. 4. Comparison of the volume filled by the drops of different diameters,
for both rain rates. Field quantifies the degradation using the measurement of
the damage with a microscope at a perfectly defined magni-
of 100 and 10 mm/h. Fig. 4 shows a comparison of the volume fying power. We will make the hypothesis that this relation
filled with the drops for both rain intensities. Increasing the also applies to the damage criterion we are using: a 10% loss
intensity increases of course the quantity of water impinging of optical transmission. Relation (3) assists in the conversion
on the material but also increases the relative part of large from a distribution of drops with different diameters but at
diameter droplets. the same velocity to a distribution of drops with the same
Palmer’s formula gives for each drop diameter the number diameter, but at different velocities. One can see, Table 1, in
of drops contained in 1 m3 of rain. Knowing the diameter Di column “equivalent 2 mm drop velocity” the result of such a
of the impacted surface, the velocity Vg of the object and the calculation in the case of the use of the water jet generator
rain exposure duration t, the actual number Ndr of drops of multiple impact jet apparatus (MIJA).

Table 1
Results of the computation of the equivalent rain erosion experiment, using an apparatus with only one droplet diameter, with an actual rain
dmin dmax Number of Volume filled by the Actual number Equivalent 2 mm Number of 2 mm
droplets/ m3 drops (mm3 /m3 ) of drops droplet velocity (m/s) droplets
0.00 0.25 2000 2 471 1 0
0.25 0.50 1355 37 319 150 2
0.50 0.75 917 117 216 189 9
0.75 1.00 621 218 146 216 15
1.00 1.25 421 314 99 238 20
1.25 1.50 285 388 67 256 22
1.50 1.75 193 433 45 273 23
1.75 2.00 131 451 31 287 23
2.00 2.25 89 445 21 300 21
2.25 2.50 60 420 14 312 19
2.50 2.75 41 384 9.6 323 16
2.75 3.00 28 342 6.5 334 14
3.00 3.25 19 297 4.4 343 11
3.25 3.50 13 254 3.0 353 9
3.50 3.75 9 213 2.0 362 7
3.75 4.00 6 176 1.4 370 6
4.00 4.25 4 144 0.9 378 5
4.25 4.50 3 116 0.6 386 4
4.50 4.75 2 93 0.4 393 3
4.75 5.00 1 74 0.3 400 2
5.00 5.25 1 58 0.2 407 2
5.25 5.50 1 45 0.1 414 1
5.50 5.75 0 35 0.1 420 1
5.75 6.00 0 27 0.1 427 1
6.00 6.25 0 21 0.0 433 1
6.25 6.50 0 16 0.0 439 0
6.50 6.75 0 12 0.0 444 0
Total 5122 1459 235
Rain intensity I (mm/h): 100; impacted surface diameter (mm): 10; droplet velocity (m/s): 300; time duration (s): 10; first diameter (mm): 0.00000001; diameter
step (mm): 0.25; power velocity dependence: −7.
548 A. Déom et al. / Wear 258 (2005) 545–551

Bench MIJA generates successive water jets which hit at It then remains to add the number of impacts of 2 mm
a great speed the object for which one needs to measure the drops at 300 m/s to be performed by MIJA.
resistance to rain erosion. The water jets are sent randomly
onto the object so as to simulate the random distribution of 2.2. Experimental validation
water drops in real rain. A full description of MIJA can be
found in [3]. Comparative measurements with trials on the 2.2.1. Principle
whirling arm at SAAB-SCANIA facility help to calibrate Two steps would be necessary in order to verify the previ-
MIJA with a more conventional facility and also to give the ous analysis. The first one would consist in a comparison of
correlation of the drop speed equivalent to the MIJA jet, for the degradation of two samples, the first one exposed at suc-
a given drop diameter (in our case 2 mm) [4]. cessive trials with different diameters of droplets at the same
Simulating real rain could be obtained by using MIJA in velocity and the second one exposed to successive trials with
successive trials with a different number of impacts at dif- MIJA at different velocities.
ferent velocities. This fastidious trial can be avoided if one Referring to Table 1, we must compare, at a given ve-
knows the dependence of the degradation on the impact veloc- locity of 300 m/s, the degradation of the first sample after
ity, which is the case if the material has already been charac- 319 impacts of droplets with a diameter between 0.25 and
terized by rain erosion measurements at different velocities. 0.50 mm, followed by 216 impacts of droplets with a diam-
Following our know-how, curves of dependence of the rain eter between 0.50 and 0.75 mm followed by . . . 1 impact
exposure time to get 10% of transmission loss can generally of droplet with a diameter between 4.0 and 4.25 mm with
be written as follow: the degradation of the second sample after 319 impacts with
MIJA at a velocity of 150 m/s, followed by 216 impacts at
t0.1 = aV n (4) a velocity of 189 m/s, followed by . . . 1 impact at a velocity
of 378 m/s.
with n the power velocity dependence for the material, which This first step which requires experiments either in
can vary from 3 to 15, depending largely on the material. flight with perfectly known rain parameters or to use an
For a given drop diameter, the number of impacts needed on-ground apparatus in which samples can hit or be hit
to obtain a given degradation at a given velocity, can be cal- with spherical water droplets with a diameter changing
culated from the number of impacts necessary to get the same from 0 to 4 mm at a velocity of 300 m/s has not been
degradation at another velocity with the formula: performed.
NV2 = NV1 (V2 /V1 )n (5) The second step consists to compare, always referring to
Table 1, the degradation of a sample hit with MIJA with 319
In the case of the example of Table 1, this permits the impacts at 150 m/s, followed by 216 impacts at 189 m/s, fol-
calculation of the number of impacts to launch with MIJA at lowed by . . . 1 impact at the velocity 378 m/s, with the degra-
300 m/s for each set of drops at the equivalent 2 mm droplet dation of a sample after 235 impacts at 300 m/s. This step can
velocity V. be performed.

Fig. 5. Variation of the normalized transmission in the 3.5–6 ␮m range of the sample of glass 9754 no. 101 with the increase of the normalized cumulative rain
exposure time (Ved = 295 m/s).
A. Déom et al. / Wear 258 (2005) 545–551 549

Table 2
Calculated distribution of droplets of a natural rain corresponding at a rain intensity of 19.43 mm/h and computation of the number of 2 mm droplets to simulate
the erosion of a 12 mm diameter surface of a glass 9754 sample hit at 295 m/s during 30 s
dmin dmax Number of Volume filled by the Actual number Equivalent 2 mm Number of 2 mm
droplets/ m3 drops (mm3 /m3 ) of drops droplet velocity (m/s) droplets
0.00 0.50 4000 33 4002 1 0
0.50 1.00 1332 294 1333 186 26
1.00 1.50 444 454 444 234 62
1.50 2.00 148 414 148 268 65
2.00 2.50 49 293 49 295 49
2.50 3.00 16 178 16 318 31
3.00 3.50 5 98 5 338 17
3.50 4.00 2 50 2 355 9
4.00 4.50 1 24 1 372 4
4.50 5.00 0 11 0 387 2
5.00 5.50 0 5 0 400 1
5.50 6.00 0 2 0 413 0
6.00 6.50 0 1 0 425 0
Total 1859 6000 267
Rain intensity I (mm/h): 19.43; impacted surface diameter (mm): 12; droplet velocity (m/s): 295; time duration (s): 30; first diameter (mm): 0.00000001;
diameter step (mm): 0.5; power velocity dependence: −8.52963315.

Table 3
2.2.2. Experimental results MIJA experiments performed on sample no. 104 to simulate the erosion of
To perform the experiments, we choose an infrared mate- a 12 mm diameter surface of a glass 9754 sample hitting at 295 m/s during
rial also transparent in the visible, in order to follow easily the 30 s a precipitation of 19.43 mm/h intensity
optical degradation of the material. This material is a glass Number of impacts to Equivalent 2 mm Equivalent 2 mm drop
with the reference 9754. deliver by MIJA droplet velocities to velocities really
The first step was the measurement of the dependence on generate (m/s) generated (m/s)
velocity of the characteristic optical time t0.1 for this material. 1333 186 188
Let us remember than t0.1 represents the rain exposure time 444 234 236
148 268 270
for which a 10% loss of optical transmission can be observed. 49 295 295
The normalized rain exposure time is calculated from the 16 318 312
number of impacts N delivered by MIJA using the following 5 338 328
relation: 2 355 359
1 372 356
106 N (Dg )3
tnorm = (6)
1.5 (Di )2Ved
Table 4
Calculated distribution of droplets of a natural rain corresponding at a rain intensity of 19.43 mm/h and computation of the number of 2 mm droplets to simulate
the erosion of a 12 mm diameter surface of a glass 9754 sample hit at 450 m/s during 1 s
dmin dmax Number of Volume filled by the Actual number Equivalent 2 mm Number of 2 mm
droplets/ m3 drops (mm3 /m3 ) of drops droplet velocity (m/s) droplets
0.00 0.25 2000 2 102 1 0
0.25 0.50 1154 32 59 225 0
0.50 0.75 666 85 34 283 1
0.75 1.00 384 135 20 325 1
1.00 1.25 222 165 11 357 2
1.25 1.50 128 174 7 385 2
1.50 1.75 74 166 4 409 2
1.75 2.00 43 147 2 430 1
2.00 2.25 25 124 1 450 1
2.25 2.50 14 100 1 468 1
2.50 2.75 8 78 0.4 485 1
2.75 3.00 5 59 0.2 500 1
3.00 3.25 3 44 0.1 515 0
3.25 3.50 2 32 0.1 529 0
3.50 3.75 1 23 0.0 542 0

Total 241 13
Rain intensity I (mm/h): 19.43; impacted surface diameter (mm): 12; droplet velocity (m/s): 450; time duration (s): 1; first diameter (mm): 0.00000001; diameter
step (mm): 0.25; power velocity dependence: −8.52963315. with Dg is the equivalent droplet diameter (=2 mm), Di the
550 A. Déom et al. / Wear 258 (2005) 545–551

Fig. 8. Front face of sample no. 104 after experiments using MIJA with
Fig. 6. Front face of sample no. 101 after rain erosion using MIJA
successive trials at different velocities to simulate 30 s of erosion at 295 m/s
(Ved = 295 m/s, number of impacts = 245, tnorm = 30.8 s).
in a 19.43 mm/h rain intensity.

diameter of the impacted surface, and Ved equivalent droplet the velocities to perform and the velocities really performed,
velocity (m/s). particularly for the experiments with a small number of hits.
One can see for example in Fig. 5 the decrease of the nor- Fig. 8 shows a photography of the front face of sample no. 104
malized transmission of a 9754 glass sample with the increase at the end of the experiment. The degradation of the sample in
of the cumulative normalized rain exposure time for the ve- the visible range is comparable to the degradation of sample
locity of 295 m/s. t0.1norm is found equal to 30.8 s. Fig. 6 shows no. 101 tested at the unique velocity of 295 m/s. The loss of
a picture of the eroded front face of the sample at the end of transmission of sample no. 104 is equal to 8%, which is very
the trial. Three velocities were used to determine the depen- comparable to the 10% loss of transmission of sample no.
dence of t0.1norm on velocity: 246, 295 and 349 m/s (Fig. 7). 101.
The power velocity dependence is found equal to −8.53. The replacement of the characterization at different veloc-
Let us calculate the distribution of droplets of an actual rain ities being successful at 295 m/s, that is to say for a velocity
for the parameters used in one of the previous experiments: in the middle range of the characterization of the power de-
I = 19.43 mm/h, Di = 12 mm, Ved = 295 m/s and for a time of pendence on velocity we used, it was interesting to test the
exposition to the rain equal to the duration of the test, about method for a higher velocity. Let us consider the velocity
30 s (Table 2). of 450 m/s. Considering the power velocity dependence of
From the previous calculation, the experiment to perform the material, we could obtained a 10% loss of transmission
with MIJA to verify if one can replace several successive in 0.93 s corresponding to about 18 MIJA impacts. Table 4
trials at different velocities by one experiment at a unique
velocity, can be deduced. One can see Table 3 the trials to
realize and the experiments really performed with MIJA on
sample no. 104. A little discrepancy can be observed between

Fig. 9. Front face of sample no. 106 after experiments using MIJA with
Fig. 7. Dependence on velocity of the characteristic optical time for glass successive trials at different velocities to simulate 1 s of erosion at 450 m/s
9754. in a 19.43 mm/h rain intensity.
A. Déom et al. / Wear 258 (2005) 545–551 551

shows the results of the computation of the number of 2 mm ments. Supposing the influence of droplet diameter follows
droplets to perform with MIJA and the corresponding droplet the Field law, we experimentally show that it is possible to
distribution of a natural rain. Fig. 9 shows a photography of simulate the action of the droplets of different diameter of a
the front face of sample no. 106 hit with the method of the natural rain by the action of droplets with a unique diameter
successive trials at different velocities. A loss of optical trans- as produced by the water jet generator MIJA. MIJA can be a
mission equal to 6–7%, instead of 10% was measured at the very useful tool to characterize the rain erosion resistance of
end of the experiment. This little discrepancy can come from materials within in-flight requirements.
the extrapolation to high velocities of the power velocity de-
pendence of the characteristic optical time. Effectively, an
experiment at the velocity of 450 m/s gives a value of 2.1 s References
instead of 0.93 s for t0.1norm obtained for 26 impacts instead
18. [1] A.A. Fyall, in: A.A. Fyall, R.B. King (Eds.), Proceedings of
the Rain Erosion Conference held at Meersburg, West Germany,
5th–7th May, Royal Aircraft Establishment, Farnborough, UK,
1965.
3. Conclusions [2] R.J. Hand, J.E. Field, D. Townsend, J. Appl. Phys. 70 (11) (1991)
7111–7118.
There are not too many difficulties in performing rain ero- [3] C. Seward, J. Pickles, J.E. Field, Window and Dome Technologies
sion tests when the goal of the experiments is only the com- and Materials II, Proceedings of the SPIE’s 1990 Tech. Symposium,
vol. 1326, 1990, pp. 280–290.
parison of one material to another. The difficulties are largely [4] A. Déom, D. Balageas, Proceedings of the 8th European Electromag-
increasing when the objective of the experiments is to com- netic Structures Conference, Nottingham, 6–7 September, 1995, pp.
pare the resistance of the materials to given flight require- 87–94, TP ONERA1995-169.

You might also like