Material and velocity effects on cavitation erosion

pitting
Jean-Pierre Franc, Michel Riondet, Ayat Karimi, Georges L. Chahine

To cite this version:

Jean-Pierre Franc, Michel Riondet, Ayat Karimi, Georges L. Chahine. Material and velocity effects on
cavitation erosion pitting. Wear, 2012, 274, pp.248-259. �10.1016/j.wear.2011.09.006�. �hal-00905642�

HAL Id: hal-00905642

https://hal.science/hal-00905642
Submitted on 29 Oct 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License

 Material and velocity effects on cavitation erosion pitting

Jean-Pierre Franc a,∗ , Michel Riondet a , Ayat Karimi b , Georges L. Chahine c

a
Grenoble University (LEGI), Grenoble, France
b
Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland
c
Dynaflow, Inc., Jessup, MD, USA

Cavitation erosion during the incubation period was investigated via pitting tests conducted on three different materials: an Aluminum alloy, a Nickel
Aluminum Bronze alloy and a Duplex Stainless Steel. Pitting tests were conducted in a cavitation tunnel in the velocity range 45–90 m/s at a constant
cavitation number. The test section was made of a straight nozzle 16 mm in diameter discharged into the radial 2.5 mm space between two flat walls.
Cavitation appears in the form of a toroidal cavity attached to the nozzle exit and damage on the samples facing the nozzle is concentrated in a circular
ring centered in the cavity closure region. The exposure time was adjusted to avoid pit overlapping. The material surface was examined using a
conventional contact profilometer which allowed us to identify the pits, count them, and measure their main characteristics such as depth, surface area,
and volume. From these the pitting rate, the coverage rate, and the depth of deformation rate were defined. Pits were classified according to their
diameter. For all materials and operating conditions, pitting rate appears to follow an exponential law in relation to the pit diameter. This law depends
upon two parameters only, which were identified as the coverage time  (i.e. the time required for the surface to be covered by erosion pits) and a
characteristic pit diameter ı, which corresponds to the pits whose contribution to the coverage process is the highest. Scaling laws for pitting were
derived accounting for both material properties and flow velocity, and a procedure to make pitting test results non-dimensional is proposed. The
influence of the material on pitting test results was analyzed. It is shown that the damage is not correlated in simple terms with the elastic limit
determined from conventional tensile tests and it is conjectured that other parameters such as the strain rate might play a significant role and should be
included in the analysis. The effect of flow velocity on both parameters  and ı was analyzed and a classical power law was found for the influence of the
flow velocity on pitting rate for all three materials. Finally, some analysis and discussion is given concerning distributions of pit volume and pit depth.

1. Introduction loads due to bubble collapses, at least the most intense ones. Pitting
tests are an alternative to the use of conventional pressure sensors
This paper is devoted to the incubation period of the cavitation which may be damaged by cavitation and which do not necessar-
erosion process which precedes material removal and mass loss. ily meet the required conditions for an accurate measurement of
During incubation, damage is characterized by small isolated plas- the impact loads in terms of rise time and resonant frequency in
tic indentations on the material surface. Each pit is expected to be particular.
produced by a cavitation bubble collapsing close to the wall. It is Although pitting tests appear as an attractive option to quantify
well-known since Lord Rayleigh’s work in 1917 that a vapor bub- the impact loads, no validated procedure is available yet. Estimat-
ble can actually generate very high loads capable of damaging solid ing loads from pitting tests requires accurately understanding the
walls when collapsing [1]. material response to the impact load and, by an inverse technique,
Pitting tests have been recognized very early as a helpful tech- infer the characteristics of the loads from the measured geometric
nique to estimate the “Cavitation Intensity” as introduced first by characteristics of the pits. The high strain rates involved with a pit
Knapp in the 1950s [2,3]. The idea behind pitting tests is to use formation together with the most likely triaxial loading makes this
the material itself as a kind of sensor which will reveal the impact inverse procedure quite difficult. In addition, the material response
at the microscopic scale of the collapsing bubble may be affected by
multiple parameters such as microstructure, grain size, modes of
∗ Corresponding author. Tel.: +33 4 76 82 50 35; fax: +33 4 76 82 52 71. plastic deformation, etc. The use of pitting tests for the quantifica-
E-mail addresses: [email protected] (J.-P. Franc), tion of cavitation intensity is still the subject of investigations and
[email protected] (A. Karimi), [email protected] (G.L. Chahine). the present work is a contribution to this general objective.

1
 Nomenclature

D pit equivalent diameter based on pit flat surface area

D* non-dimensional pit equivalent diameter (Eq. (11))
Dmax maximum equivalent pit diameter (see Table 3)
e engineering strain
E modulus of elasticity
h pit depth
KY parameter of the Ramberg–Osgood constitutive
equation
mY parameter of the Ramberg–Osgood constitutive
equation
N pitting rate per unit exposure time and unit surface
area
N* non-dimensional pitting rate (Eq. (10))
n probability density function (Eq. (5))
ˇ coverage rate
ˇ* non-dimensional coverage rate (Eq. (14))
depth of deformation rate
ı characteristic pit equivalent diameter (Eq. (4))
ε true strain
ε̇ strain rate
εe elastic deformation
εp plastic deformation
εU ultimate strain Fig. 1. Sketch of the test section.
 true stress or cavitation number
U ultimate tensile strength
Y elastic limit is also investigated at constant cavitation number, i.e. for the same
 coverage time cavitation extent in the loop.
 pit volume This work is part of an ongoing effort between the laborato-
ries of the authors and two US Navy laboratories (NSWCCD and
NRL) to derive a procedure for estimating impact loads from pitting
tests. The cavitation intensity, which can roughly be defined as the
The incubation period and associated pitting has been studied
spectrum of impact loads, should obviously be independent of the
by many investigators using various experimental devices. Cavi-
material used for pitting tests, even though the use of soft materials
tation tunnels [4,5] produce cavitation erosion in a flowing liquid
probably makes it possible to extend the spectra towards loads of
whereas vibratory systems [6–12] generate cavitation erosion in
small amplitude. The analysis of pitting tests on different materials
an almost stagnant liquid. Other devices have been developed for
carried out in this paper is an important step in the development
special purposes such as vortex cavitation generators [13,14], the
of a technique to assess the cavitation intensity from pitting tests.
Hopkinson bar type impact device [15], the magnetic impact test-
ing machine [16], the rotating disk cavitation apparatus [17,18] or
submerged cavitating jet devices [19–22]. Some of these techniques 2. Experimental facility and procedures
were standardized and resulted in American Society for Testing and
Materials (ASTM) Standards such as G-32 “Test Method for Cavita- 2.1. Test section
tion Erosion Using Vibratory Apparatus” and G-134 “Test Method
for Erosion of Solid Materials by a Cavitating Liquid Jet”. Most of Experiments were conducted in a cavitation flow loop described
these techniques produce a large spectrum of bubbles of various in details in Ref. [25]. The test section, also described in [25], is
sizes, at various distances from the eroded sample with possible axisymmetric and made of a straight nozzle 16 mm diameter which
collective effects which may affect the erosive potential of each generates a high velocity flow. As illustrated in Fig. 1, the flow is
individual bubble due to its interaction with the neighboring ones. deflected by the sample to be eroded which is set at a distance
Detailed experimental studies have also been made on a single of 2.5 mm from the nozzle exit, and propagates in between two
bubble collapsing in a static fluid in order to analyze the basic mech- parallel circular plates formed by the plane of the sample and that
anisms of cavitation damage such as the formation of a high-speed of the nozzle exit orifice. Cavitation takes the form of a toroidal
liquid jet and the emission of shock waves. The resulting damage cavity attached to the nozzle orifice. Cavitation erosion is observed
can be much more complex than a unique circular pit because of the in the closure region of the cavity in the form of a circular ring
formation of the re-entrant jet and a torus like structure which may whose mean diameter is of the order of 50 mm. The radius where
break up into smaller bubbles whose collapse may cause additional maximum damage occurs can easily be determined from mass loss
pits of smaller size [23,24]. tests and the detection of the radius where the erosion depth is
In the present work, pitting is generated in a cavitation loop. maximum. The flow remains confined in opposition to a cavitating
The loop is operated at high enough velocity and pressure in order submerged jet.
to be able to investigate cavitation erosion on resistant materials The maximum operating pressure of the tunnel is 40 bars, which
within reasonable exposure times. Because of the large number of corresponds to a maximum velocity of 90 m/s. Pitting tests were
bubbles produced simultaneously, the investigation is based on a conducted at different velocities between 45 m/s (upstream pres-
statistical analysis of erosion pits. Pitting tests have been made on sure of 10 bar) and the maximum velocity of 90 m/s. For all tests,
three different materials in order to investigate the influence of the ambient pressure was adjusted such that when the flow veloc-
the material on pitting results. The influence of the flow velocity ity was changed the cavitation number remained constant. This

2
 Fig. 3. Stress–strain curves of the three tested materials deduced from conventional
Fig. 2. Stress–strain curves of the three tested materials deduced from conventional
tensile tests – enlargement of the elastic domain – the open circles indicate the yield
tensile tests.
strength given in Table 1.

ensured that the cavity length was almost the same for all tests and
The main properties of the tested materials are shown in Table 1.
that a similarity existed between the different flows at different
The elastic limit  Y corresponds to the critical stresses where the
velocities. The velocity effects discussed in this paper are then pure
stress/strain curve (ε) departs from the linear elastic relationship:
velocity effects which do not include any cavitation number effect.
For each material and each flow velocity, the test duration was  = Eεe (2)
adjusted in order to avoid any overlapping of pits which would
where E is the modulus of elasticity. The criterion used for estimat-
result in measurement errors. The exposure time was only a few
ing the elastic limit corresponds to a ratio εp /εe of the permanent
seconds for the tested alloys of Aluminum and Nickel Aluminum
plastic deformation εp to the elastic deformation εe equal to 5%
Bronze at the maximum velocity and was about 15 min for the
[30]. A magnification of the elastic region is shown in Fig. 3 together
tested Stainless Steel at the smallest velocity. In order to limit the
with the elastic limits represented by open circles. The elastic limit
effect of transient flow especially for tests with a short exposure
which is expected to be the critical stress threshold for the forma-
time, a preliminary test was conducted with the purpose of finding
tion of pits will be further considered in Section 5 devoted to the
out the appropriate settings, i.e. the pressurization level and the
effect of material on pitting.
operating flowrate. Then, the sample was changed, the pressuriza-
Table 1 also contains the values of the best fit of parameters KY
tion level was adjusted and the pump was started. Flow rate was
and mY of the Ramberg–Osgood constitutive equation [30]:
increased until cavitation inception was identified acoustically. The
  − Y
 mY
facility was run at a point just below inception for a sufficient time

ε = εe + εp = + (3)
in order to stabilize the flow. Then, the flow rate was almost instan- E KY
taneously increased from inception to the final operating point. The Tensile tests have been carried out at a velocity of 30 ␮m/s which
increase in flow rate is very rapid and has no effect on the pressur- corresponds to a strain rate ε̇ of 4 × 10−4 s−1 considering the length
ization level. This special procedure minimizes the time needed to of the samples of 70 mm. However, it is well-known that the char-
reach the operating point. The samples were mechanically polished acteristic strain rate during the cavitation erosion process is much
using a semiautomatic polishing machine (Logitech PM%) to obtain higher, typically 103 s−1 . In order to have indications on the effect
a metallographic surface of roughness <0.1 ␮m. of strain rate, tensile tests were also conducted at a higher strain
rate of 1.5 s−1 . The corresponding results are presented as dashed
2.2. Materials lines in Figs. 2 and 3. In this range of variation of strain rate, Al
and NAB appear to be only a little sensitive to strain rate. How-
Three materials were selected for comparative tests at LEGI and ever, the Stainless Steel properties including its elastic limit appear
Dynaflow under this program: Aluminum alloy 7075-T651, Nickel to be significantly affected by the strain rate (see also [31]). These
Aluminum Bronze alloy, and Duplex Stainless Steel 2205, named conclusions on the influence of strain rate need to be confirmed by
respectively Al, NAB and SS in the present paper. The materials were tests at higher strain rate, typically 103 s−1 , more representative of
characterized by conventional tensile tests. Figs. 2 and 3 compare the actual strain rate encountered in cavitation erosion.
the stress–strain curves of the three materials. Three tests were
conducted for each material under quasi-static loads and a small 2.3. Pit analysis technique
strain rate. Al and NAB show much better reproducibility than SS
that might contain more heterogeneities or inclusions. Various techniques can be found in the literature to analyze pits
The stress and strain values presented in Fig. 2 are the true such as optical profilometry [32,33], laser profilometry [4], scan-
stress  and true strain ε related to the engineering stress S and ning electron microscopy [34]. In the present work, a conventional
engineering strain e by [26–28]: contact profilometer using a stylus with a tip radius of 2 ␮m was
used to measure the surface of the sample after pitting. The mesh

ε = ln(1 + e)
(1) size for describing the surface was 1 ␮m × 1 ␮m and several sur-
 = s(1 + e)
faces of 2 mm × 4 mm were analyzed (typically 3) in order to have
No correction was applied after necking since the diameter a large enough number of pits for the statistical analysis. It has been
reduction and radius change in the necked region were not mea- estimated that a total number of pits between 200 and 300 is gen-
sured [26,29]. Necking was important only for SS. erally sufficient [25]. All measurements were made in the region of

3
Table 1
Tested materials and main properties.

Material Aluminum alloy 7075 T651 (Al) Nickel Aluminum Bronze alloy (NAB) Duplex Stainless Steel 2205 (SS)

Elastic limit,  Y 530 MPa 150 MPa 230 MPa

Ultimate tensile strength,  U 645–660 MPa 590–610 MPa 765–810 MPa
Modulus of elasticity, E 72 GPa 122 GPa 186 GPa
Ultimate strain, εU 0.086–0.095 0.13–0.15 0.23–0.26
Ramberg–Osgood parameter, KY 447 MPa 1210 MPa 910 MPa
Ramberg–Osgood parameter, mY 1.98 2.07 3.2

maximum pitting damage, i.e. at the same distance from the center surface where the identified pits have been circled is also shown
of the sample but at different azimuthal positions. for three different values of the depth threshold. As expected, the
In order to identify the pits, a cut-off depth was applied during coverage rate (as well as the pitting rate and the depth of defor-
post-processing of the measured surface data. For all results pre- mation rate not presented here), increases when the cut-off depth
sented in this work (unless otherwise stated), the cut-off depth was decreases. In the opinion of the authors, the most serious difficul-
chosen at 0.5 ␮m below the original material surface. This choice ties when selecting the cut-off depth are to avoid (i) the detection
was considered to give a satisfactory account of the pitted surface of erroneous pits which would lead to an over-estimation of the
for all three materials. pitting rate and (ii) the merger of a pit with neighboring pits or
As an example of the influence of the cut-off depth, Fig. 4 shows with small surface defects which would alter pit size estimation.
how the coverage rate is affected by the choice of the threshold. The This is obviously the case for the smallest cut-off depth of 0.1 ␮m
coverage rate is defined in Section 3.2 and represents the fraction of considered in Fig. 4. On the contrary, identification of pits appears
surface covered by the pits per unit exposure time. The image of the to be satisfactory for the two other cases although the shallow pits

Fig. 4. Influence of cut-off depth on the identified pits and on the coverage rate. The three surfaces presented at the top of the figure are 2 mm × 4 mm. The gray level is an
indicator of the local depth.

4
are more and more discarded when the cut-off depth increases. The
cut-off depth of 0.5 ␮m selected in this work is considered as a good
compromise.

3. Distribution of pits with diameter

The pit analysis technique used here allows us to identify indi-

vidual pits and measure their surface areas and volumes. It is then
possible to determine the number and the cumulative surface and
volume of all the identified pits. It can be expected that pitting data
will be proportional to the exposure time as long as no overlapping
between pits occurs. However, this requires that the cavitating flow
is statistically stable in time and that the material properties remain
constant during exposure to cavitation. Both points are discussed
below.
Even though the operating conditions are kept constant, cavi-
tating flows are by nature unsteady and involve periodic shedding
of bubble clouds. Shedding generally does not affect pitting tests
because the exposure time is usually much longer than the shed-
ding period. Therefore, a large number of bubble clouds are shed Fig. 5. Cumulative pitting rate as a function of pit diameter for different values of
during the exposure time and the cavitating flow can be considered the upstream pressure and a constant cavitation number  = 0.9 on Duplex Stain-
less Steel 2205 (cut-off depth: 0.5 ␮m). Straight lines correspond to an exponential
as stable on average.
distribution.
Material properties may change over time during exposure to
repeated loading because of work hardening and this change could
affect the material response during exposure to cavitation. Con-
sidering that the plastic zone around a pit is significantly larger
than the pit itself, it is important that pits remain sufficiently sep- function of pitting rate defined as the number N of pits per unit time
arated and in any case do not overlap. This guarantees that each and unit area counted with a diameter larger than a given value, D.
new impact falls on a virgin area where the material has not been The distribution of pitting rate can be approximated, reasonably
hardened by a previous impact. If so, no aging effect occurs on the well, by an exponential law, so that the following mathematical
material side and the deformations caused by similar impacts are form can be assumed:
similar during the exposure to cavitation.
Under such conditions, the number, cumulative surface and
cumulative volume of pits can be considered as proportional to 8
N= e−(2D/ı) (4)
the exposure time. In the same way, these data can be considered ı2 
as proportional to the analyzed surface. This requires mainly that
the region of analysis be sufficiently small in comparison to the
ı and  are two fitting parameters whose physical interpretation
overall pitted region. If so, the number of pits is constant on equal
will be given later. ı has a length dimension and  has a time dimen-
surface areas and then proportional to the analyzed area. In the
sion so that N has actually the proper dimension of a pitting rate per
present work, we essentially considered the region of maximum
unit surface area. The factor 8/ is introduced to simplify further
damage and did not investigate the variation of pitting damage on
interpretations.
both sides of the maximum.
According to Fig. 5, the approximation by an exponential dis-
Provided the previous conditions are satisfied, densities can be
tribution is not so good for large values of the diameter which
defined by dividing the measured data by the exposure time and
correspond to rare events, not necessarily correctly captured during
the analyzed surface. This is done for the three data measured here,
a limited exposure time.
i.e. the number of pits, their projected surface and their volume,
The probability density function n is obtained by differentiation
which leads respectively to the pitting rate, the coverage rate and
of Eq. (4):
the depth of deformation rate, as indicated in Table 2. Let us observe
that each of these quantities can be defined either for all measured
pits or, for a more detailed analysis, only for those pits whose diam- dN 16 −(2D/ı)
eter is larger than a given value and then become a function of this n=− = e (5)
dD ı3 
threshold value. Distribution functions with diameter can then be
derived and the contribution of each class of size analyzed.
This equation gives the number of pits of diameter D per unit
3.1. Pitting rate bandwidth in diameter. This is also an exponential law where
n continuously increases with decreasing diameter, expressing
The counted pits can be classified according to their diameter. that the number of pits regularly increases when the pit size
Fig. 5 presents a typical example of the cumulative distribution decreases.

Table 2
Measured and computed data.

Measured data Computed data Name Unit

number of pits
Number of pits N= exposure time×analyzed surface
Pitting rate Pits/m2 /s
surface of pits
Surface of pits ˇ= exposure time×analyzed surface
Coverage rate s−1
volume of pits
Volume of pits = exposure time×analyzed surface
Depth of deformation rate m/s (or ␮m/h)

5
 The coverage rate is plotted in Fig. 6 in the case of Stainless Steel
and for different operating conditions corresponding to different
cavitation intensities. The intersection with the X-axis, i.e. for D = 0
corresponds to the total coverage time for all measured pits what-
ever may be the size, whereas for any non-zero value of D, a partial
coverage time can be defined corresponding to the contribution of
the only pits whose diameter is larger than D. This figure confirms
the asymptotic behavior of the coverage rate when pit diameter
approaches zero, i.e. when all pits are taken into account. This
proves that small pits, even though they are the most numerous,
do not contribute significantly to the surface coverage.
It appears then that each class of pit diameter does not con-
tribute equally to the coverage process. The probability density
function of the coverage rate defined in Eq. (6) is given by:

dˇ 4
− = 3 D2 e−(2D/ı) (7)
dD ı 

This density does not vary in a monotonic way with diameter but
presents a maximum, which occurs for a pit diameter equal to
Fig. 6. Cumulative rate of coverage as a function of pit diameter for different values ı. This gives a physical interpretation to the second parameter
of the upstream pressure and a constant cavitation number  = 0.9 on Duplex Stain- of the exponential distribution law of pitting rate. As a conse-
less Steel 2205 (cut-off depth: 0.5 ␮m). Solid curves correspond to the exponential
quence, the major contribution to surface coverage is due to pits
distribution plotted in Fig. 5.
whose diameter lies in a bandwidth around ı, skewed towards
the larger diameter pits. Larger pits contribute less because of
their lower frequency, whereas smaller pits do not contribute
3.2. Coverage rate
appreciably either, because of their small size even though the asso-
ciated frequency is relatively high. This trend can also be found in
It is then possible to integrate the probability density function
[25,33].
over the area of all pits of diameter larger than a given value D to
Fig. 7 gives typical examples of these probability density func-
obtain the parameter ˇ:
tions for the pitting test results presented in Figs. 5 and 6. The
∞   characteristic diameter ı appears to increase with upstream pres-
D2 1 2D 2D2 sure (or equivalently with flow velocity) and our estimates for
ˇ= n dD = 1+ + 2 e−(2D/ı) (6)
4  ı ı the present tests on Duplex Stainless Steel 2205 lie in the range
D 35–57 ␮m according to the flow velocity. The influence of flow
velocity will be discussed in more details in Section 6. It is impor-
From a dimensional viewpoint, ˇ has the unit of the inverse
tant in practice that pit analysis focuses on this range of diameter
of time. It can be interpreted as the coverage rate, i.e. the frac-
which appears to be the most damaging during the incubation
tion of surface covered by all pits larger than D per unit exposure
period in terms of covered surface. It is conjectured that this
time to cavitation. Eq. (6) shows that, when D approaches zero,
range of size is also the most damaging in terms of material fail-
the coverage rate ˇ tends to 1/. Therefore, the parameter  can
ure during the more advanced stages of erosion where mass loss
be interpreted as the time required for the surface to be entirely
occurs.
covered just one time by erosion pits. Note that the factor 8/ was
introduced in order to remove any multiplication factor in Eq. (6).
The coverage time  is expected to play a major role in cavitation
erosion, not only in the initial pitting regime but also in the more
advanced mass loss regime. The erosion model developed in [25]
predicts that (i) the incubation time is proportional to this coverage
time and (ii) the erosion rate in the steady state regime of ero-
sion which follows the incubation period, measured by the mean
depth of penetration rate (MDPR), is inversely proportional to .
Even though the incubation time and the erosion rate depend upon
several other parameters including material properties, the cover-
age time  appears to be a key parameter which strongly depends
upon the flow via the frequency of occurrence of bubble collapse
and the characteristic size of the cavitation bubbles.
Let us observe that the so-defined  parameter suffers from a
certain imprecision since it depends upon the selected cut-off depth
used in the pit analysis (see Section 2.3). Moreover, in terms of
impact load, it is likely that the load is non uniform on the pit sur-
face. Then, the definition of the pit surface from the cut-off depth
without any reference to the impact load distribution over the
material needs to be improved. Even though a more precise def-
inition of pit size is needed to derive a more objective parameter,
Fig. 7. Estimated probability density function of coverage ratio as a function of pit
the general concept of coverage time and coverage rate appears
diameter for different values of the upstream pressure and a constant cavitation
to be essential in the physics of cavitation erosion. The concept of number  = 0.9 on Duplex Stainless Steel 2205 (cut-off depth: 0.5 ␮m). The curves
coverage can also be found in [35]. have been obtained by differentiation of the solid lines plotted in Fig. 6.

6
Fig. 8. Cumulative rate of deformation depth as a function of pit diameter for dif- Fig. 9. Ratio of the deformation depth rate by the coverage rate ˇ for differ-
ferent values of the upstream pressure and a constant cavitation number  = 0.9 on ent values of the upstream pressure and a constant cavitation number  = 0.9 on
Duplex Stainless Steel 2205 (cut-off depth: 0.5 ␮m). Duplex Stainless Steel 2205 (cut-off depth: 0.5 ␮m). This ratio can be considered as
an estimate of the average pit depth.

3.3. Deformation depth rate

of them simultaneously. The two different situations are identified
Assuming that a relation between pit volume  and pit diameter by subscripts 1 and 2. Pitting rates are written in the form (cf. Eq.
is available, it is possible to compute the deformation depth rate, (4)):
, as a function of pit diameter using the following equation: 8
N1 = e−(2D/ı1 )
∞ ı21 1

8 (9)
= n dD (8) N2 = e−(2D/ı2 )
ı22 2
D
These equations are represented by straight lines in a semi-log
As shown in Section 7, pit volume varies like  ∼ = 0.47D2.1 graph. Hence, they can be superimposed by a series of two pro-
(cf. Fig. 18). Because of the non-integer value of the expo- portional transformations. To transpose pitting test 2 into pitting
nent, integration of Eq. (6) using the previous equation for  is test 1, Eq. (9) show that diameters have to be multiplied by the
not straightforward and involves a complex Kummer’s confluent ratio ı2 /ı1 and number densities by (ı1 /ı2 )2 ( 1 / 2 ). Provided that
hypergeometric function. In order to easily extract a general trend, both transformations are applied, the histogram for pitting test 2
we observe and use the fact that the exponent is actually close to 2. reduces to that for pitting test 1.
The rate of deformation depth should then approach a linear rela- Fig. 10 gives an example of this scaling for two different materi-
tionship with the coverage rate ˇ as it can be concluded from Eqs. als, Aluminum alloy and Stainless Steel, and two different operating
(6) and (8) given by /ˇ ∼ = 0.6 ␮m if it is assumed that  ≈ 0.47D2 . conditions. It shows that this transformation works well in a large
In order to evaluate the validity of this estimate, the ratio /ˇ domain of pit size and looses accuracy only for the largest pits
is plotted as a function of pit diameter in Fig. 9. The figure shows whose density suffers from a larger uncertainty due to their rare
that /ˇ does not vary much and lies mainly between 0.8 ␮m and occurrence.
1 ␮m. The /ˇ values are more scattered for the largest sporadic
pits and are no longer almost constant in the case of the lowest flow
aggressiveness corresponding to an upstream pressure of 10 bar.
In other words, the behavior of the rate of deformation depth
is generally close to that of the coverage rate ˇ and the comparison
of Figs. 6 and 8 confirms the similar evolution of both variables with
pit diameter.
It was observed previously that there is a class of pit diameters
which contributes predominantly to damage in terms of surface.
The similar behavior of ˇ and ensures that the same class of pit
diameters also contributes primarily to cavitation damage in terms
of volume of deformation.
The ratio /ˇ can be interpreted as an average pit depth. The
range 0.8–1 ␮m obtained here is consistent with the results pre-
sented in Fig. 20.

4. Scaling laws for pitting tests

In order to analyze scaling laws applicable to pitting tests, let us

consider two different pitting tests. The difference can be either a Fig. 10. Principle of the scaling of histograms of pitting rate. Transposition requires
difference in material or a difference in operating conditions or both two proportional transformations, one on diameters and one on densities.

7
 This scaling law proves that pit size is scaled like the parame-
ter ı. This parameter is then relevant for characterizing pit size. It
will systematically be used below to analyze material and velocity
effects on pit size. As for pit density, it is scaled like ı−2  −1 as also
suggested by a purely dimensional analysis.
In other words, if we introduce the non-dimensional pitting rate
defined by:

N ∗ = ı2 N (10)

and the non-dimensional diameter defined by:

D
D∗ = (11)
ı
the cumulative histogram of pitting rate takes the following non-
dimensional form:
8 −2D∗
N∗ = e (12)

As for the non-dimensional coverage rate defined by:
∗
ˇ = ˇ (13)
Fig. 12. Comparison of the cumulative coverage rates on three different materials
it is given by the following non-dimensional form of Eq. (6): (Aluminum Al7075, Nickel Aluminum Bronze and Stainless Steel SS2205) for two
different values of the upstream pressure (10 bar and 40 bar) and a constant cavita-
∗
ˇ∗ = (1 + 2D∗ + 2D∗2 )e−2D (14) tion number  = 0.9 (cut-off depth: 0.5 ␮m). Note that the exposure time for pitting
test on Aluminum alloy for the most aggressive situation (40 bar) is very small (2 s
with a limit value equal to 1 when diameter approaches zero. These only).
non-dimensional equations give a useful basis to interpret scale
effects on pitting tests.
Whatever may be the variable used for characterizing the
erosion damage among pitting rate, coverage rate or rate of defor-
5. Effect of material properties
mation depth, the most damaged material during the incubation
period was Al, followed by NAB and finally SS. It is surprising to
In order to evaluate the influence of material properties, pit-
observe that Al was more damaged than SS and NAB whereas its
ting tests were conducted on three different materials (cf. Table 1):
elastic limit evaluated from conventional tensile tests is signifi-
an Aluminum alloy, a Nickel Aluminum Bronze alloy and a Duplex
cantly larger (see Section 2.2). This point is further discussed below.
Stainless Steel.
It is generally assumed that the material has a filtering effect on
Figs. 11 and 12 present a comparison of the distributions of
the histogram of impact loads with a threshold value often consid-
cumulative pitting rate and cumulative coverage rate versus diam-
ered as the elastic limit [36]. The material is expected to deform
eter for two different operating conditions corresponding to the
elastically when the load is smaller than its elastic limit and to
lowest and the highest speeds. For both cases, the cavitation num-
return to its original state after unloading. If the load exceeds the
ber was kept constant. (Results on the rate of deformation depth are
elastic limit, the material is supposed to experience plastic defor-
not presented since they are very similar to those on the coverage
mation and a pit is formed.
rate as explained in Section 3.3.)
This simple but a priori realistic reasoning should then lead to
the smallest pitting rate for Al as schematically shown in Fig. 13. If
the expected ranking of SS and NAB with respect to pitting damage
is qualitatively in agreement with experimental data, the prediction
concerning Al does not account for the relatively poor resistance of
Al to pitting shown by experiments.
Several reasons can explain this. Firstly, the actual response of
the material to cavitation impact loads may be more complex than

Fig. 11. Comparison of the cumulative pitting rates on three different materials
(Aluminum Al 7075, Nickel Aluminum Bronze and Stainless Steel SS2205) for two
different values of the upstream pressure (10 bar and 40 bar) and a constant cavita-
tion number  = 0.9 (cut-off depth: 0.5 ␮m). Note that the exposure time for pitting
test on Aluminum alloy for the most aggressive situation (40 bar) is very small (2 s Fig. 13. Schematic illustration of the filtering effect of the load by materials with
only). respect to their elastic limit.

8
a simple binary response with the elastic limit as a threshold. It is Table 3
Maximum pit size Dmax and ratio of maximum pit size to characteristic pit size Dmax /ı.
also possible that the elastic limit is affected by the strain rate. The
previous discussion on material ranking was based on the values Operating Material Maximum pit size ı Dmax /ı
of the elastic limit determined under quasi-steady conditions by conditions measured, Dmax
conventional tensile tests. It has been shown in Section 2.2 that the 40 bar Al 203 ␮m 88 ␮m 2.3
properties of Stainless Steel are strongly affected by the strain rate. NAB 224 ␮m 116 ␮m 1.9
Even at the relatively small strain rate of 1.5 s−1 , the elastic limit SS 144 ␮m 57 ␮m 2.5

of Stainless Steel is significantly increased and approaches that of 10 bar Al 137 ␮m 60 ␮m 2.3
Aluminum which is not significantly affected by the strain rate. It NAB 112 ␮m 61 ␮m 1.8
SS 73 ␮m 38 ␮m 1.9
can then be expected that the ranking of the three materials with
respect to their elastic limit changes with the strain rate, particu-
larly if high strain rates of the order of 103 to 104 s−1 are considered.
The measurement of material properties at high strain rate using and the diameter of the indentation increases consequently. As for
for instance a Hopkinson bar system should allow us to figure out pitting tests, Al is taken as a reference and the diameter of the
what is the influence of the strain rate. Another advantage of such a indentation on the Aluminum alloy is plotted on the horizontal axis.
test is to provide data under compressive stress analogue to cavita- The different diameters on the x-axis correspond to different loads.
tion induced stresses and not under uniaxial tensile stress. It is well The same tests are performed on SS and NAB and the diameters
known that, unlike an Aluminum alloy, NAB and SS2205 behave of SS and NAB indentations are evaluated at the same loads as Al.
differently under tensile and compressive loadings [37]. Finally, let Note that the indentation diameter is estimated from the measured
us recall that the influence of the cut-off depth on pit detection penetration depth on the basis of a simple geometric analysis.
remains unclear. The cut-off depth, which has been chosen con- The comparison of pitting data and nanoindentation data shows
stant and independent of the material in the present study, might similar qualitative trends. Pit size and indentation size at equal load
introduce a bias in the pit counting technique since it leads us to are smaller for SS than for Al and NAB, which are both comparable.
discard the shallowest pits. However, nanoindentation tests apparently fail to quantitatively
As explained in Section 4, the factor ıSS /ıAl can be used to eval- predict the ratio between SS and Al which is around 0.8 for inden-
uate the ratio of pit size between Stainless Steel and Aluminum tation tests and 0.6 for pitting tests. The analysis of pitting tests
alloy. This ratio together with the similar ratio ıNAB /ıAl for NAB is using nanoindentation tests needs to be further investigated.
plotted in Fig. 14 as a function of Al pit size taken as a reference. Figs. 11 and 12 show that the maximum pit size significantly
The characteristic pit size used on the horizontal axis is actually the depends upon the material. Maximum pit size is difficult to capture
size parameter ıAl . The different values of ıAl correspond to differ- because of the small probability of occurrence of the largest pits.
ent operating conditions. It will be shown in Section 6 that pit size Nevertheless, Table 3 presents the diameter of the largest measured
increases with flow velocity so that different values of ıAl on the pits on the three materials for two different operating conditions
horizontal axis correspond to different velocities. and the selected threshold of 0.5 ␮m.
Pit size for NAB appears to be very similar to that for Al with an The ratio Dmax /ı has an average value of roughly 2, for all three
average value of the ratio of 1.1. Conversely, pits on SS are signifi- materials at all operating points. Hence, in spite of the difficulty
cantly smaller than on Al and NAB and the ratio is about 0.6. to accurately determine the maximum pit size Dmax , it appears that
In addition to pitting test results, Fig. 14 presents nanoindenta- the maximum pit size can reasonably well be predicted by the same
tion test results obtained with a spherical indenter of radius 50 ␮m. scaling law, i.e. varies approximately like the ı parameter and more
During a nanoindentation test, the load is progressively increased precisely like 2ı.

6. Effect of flow velocity

Tests were carried out at different upstream pressures between

10 bar and 40 bar, corresponding to flow velocities between 45 and
90 m/s. As discussed in Section 2.1, pitting test results depend upon
only two parameters, namely the characteristic diameter ı and the
coverage time . The effect of flow velocity on both parameters and,
as a consequence, on the whole distribution function of pitting rate
is investigated in the present section.
Fig. 15 presents the effect of flow velocity on the pit character-
istic size ı. Let us recall that this characteristic size measures the
pit diameter which contributes the most to the total area of defor-
mation (see Section 3.2). It appears that pit size increases with flow
velocity with a power smaller than 1. The increase is similar for all
three materials and the exponent is estimated at about 0.75.
As for the coverage time , Fig. 16 shows that it decreases very
rapidly with the flow velocity. For all three materials, the coverage
time exhibits a decrease with a power estimated at about 6.5.
Both parameters are combined in the unique parameter 1/(ı2 )
which has the same unit as the pitting rate. This parameter is the
reference pitting rate to be used for making the actual pitting rate
Fig. 14. Comparison of size of indentation (pit or nanoindentation) on Aluminum, non-dimensional (see Eq. (10)). In other words, it is the scaling
Nickel Aluminum Bronze and Stainless Steel. Open symbols correspond to nanoin- parameter which allows the estimate of the actual pitting rate from
dentation tests with a spherical indenter of 50 ␮m in diameter. Comparisons of
nanoindentation diameters are made at equal load. Closed symbols correspond to
the non-dimensional pitting rate given by Eq. (12).
pitting tests at 10, 15, 20 30 and 40 bar and a constant cavitation number  = 0.9 Fig. 17 shows that this reference pitting rate increases roughly
(cut-off depth: 0.5 ␮m). like V5 without any significant difference for the three materials

9
 tested here. This dependency is in agreement with trends gen-
erally reported in the literature. As an example, Stinebring et al.
[38] observed that the pitting rate on gives constructed of pure
annealed aluminum and tested in a high-speed water tunnel fol-
lows a sixth-power law curve in a range of velocities between 14.9
and 59.3 m/s.

7. Depth and volume of pits

The volume of pits is correlated to the equivalent diameter of

the pit plane surface area as demonstrated by Fig. 18. Let us recall
that pit volume is calculated using a fully 3D approach and that no
assumption is made on the shape of the pits which, in particular,
are not assumed axisymmetric for volume computation. Each point
in Fig. 18 represents one pit and a total number of 1258 pits are
plotted corresponding to various values of the upstream pressure.
This figure shows that the volume of pits follows a power law with
the equivalent diameter which does not depend significantly upon
the operating conditions in the present range of investigation. The
exponent is slightly larger than 2 and is estimated close to 2.1 so
that pit volume  appears to vary like  ∝ D2.1 .
Fig. 15. Influence of flow velocity on the pit characteristic diameter ı. If it is assumed that pit shapes are geometrically similar on the
average whatever may be their size, pit volume should vary like
 ∝ hD2 . Then, pit depth should vary like h ∝ D−2 ∝ D0.1 and rel-
ative pit depth should vary like h/D ∝ D−0.9 . Fig. 19 confirms this
behavior and consequently validates the assumption of geometric
scaling for pit shape. It is important to note that the geometric simi-
larity of pit shapes applies only to the average. Fig. 20 clearly shows
that there is not a unique pit shape but a large variety of shapes
since depth can vary significantly for a given equivalent diameter.
The similarity law on pit shape is then valid only from a statistical
viewpoint for the average. Significant deviations can appear around
the average shape.
Note that the same pits have been considered in the plots of
Fig. 18 for the volume, in Fig. 19 for the relative depth (ratio depth
to diameter) and in Fig. 20 for the depth. The apparent difference
in scattering is due to the fact that volume varies over four decades
whereas depth varies over one decade only.
In order to check the influence of the cut-off depth, several anal-
ysis were conducted with different cut-off depths between 0.3 and
0.7 ␮m. Fig. 21 shows that a decrease in cut-off depth essentially
leads to a translation of the curves towards larger diameters but

Fig. 16. Influence of flow velocity on the coverage time .

Fig. 18. Pit volume as a function of pit diameter for different values of the upstream
pressure and a constant cavitation number  = 0.9 on Duplex Stainless Steel 2205
(cut-off depth: 0.5 ␮m). The solid line corresponds to the equation  = 0.47D2.1 where
Fig. 17. Influence of flow velocity on the reference pitting rate defined by 1/ı2 .  is pit volume in ␮m3 and D pit diameter in ␮m.

10
Fig. 19. Relative pit depth as a function of pit diameter for different values of the
Fig. 22. Pit volume as a function of pit diameter for Aluminum Al 7075, Nickel Alu-
upstream pressure and a constant cavitation number  = 0.9 on Duplex Stainless
minum Bronze and Stainless Steel SS 2205 (upstream pressure 20 bars, constant
Steel 2205 (cut-off depth: 0.5 ␮m).
cavitation number  = 0.9, cut-off depth: 0.5 ␮m).

the different curves remain parallel. It is then concluded that the

previous trend on the evolution of pit volume with pit diameter is
independent of cut-off depth.
Moreover, this trend is the same for the three tested materials
(Aluminum Al 7075, Nickel Aluminum Bronze and Stainless Steel
SS2205) as shown in Fig. 22.

8. Conclusion

The present work shows that the distribution of pits with diam-
eter can be approximated by an exponential law. This law depends
upon two parameters. One is the characteristic diameter ı of the
pits which contribute mostly to the total area of erosion damage.
Another is the coverage time  which is the time required for the
surface to be entirely covered just one time by the erosion pits.
For each pitting test, the distribution of pits with diameter is fully
described by these two parameters.
Scaling laws for pitting tests can be analyzed on the basis of both
parameters. Pit size is scaled like ı and pitting rate like ı−2  −1 .
Fig. 20. Pit depth as a function of pit diameter for different values of the upstream
Using these two scaling laws, a non-dimensional distribution of
pressure and a constant cavitation number  = 0.9 on Duplex Stainless Steel 2205.
The cut-off depth of 0.5 ␮m corresponds to the lower limit of the figure. pits can be obtained which appears to be universal for the cavita-
tion configuration and geometry under consideration here and in
particular independent of material and operating conditions. These
scaling laws were derived from pitting tests conducted on three
materials, at different flow velocities in the range 45–90 m/s and at
equal cavitation number.
The characteristic pit equivalent diameter ı depends upon the
material. Comparison of nano-indentation tests conducted at equal
loading appears to give a good qualitative prediction of the effect
of the material on pit size.
On all three materials, pit equivalent diameter increases slightly
with flow velocity, approximately like V0.75 . As for the coverage
time , it strongly decreases with the flow velocity with an expo-
nent of the order of −6.5. This effect is probably due to both an
increase in the frequency of bubbles collapsing and an increase of
the pressure pulse heights with flow velocity, whereas the thresh-
old for pitting can reasonably be expected as almost constant and
only material dependent. Both reasons concur to make the cover-
age time strongly increase with flow velocity at constant cavitation
number. As for the scale factor on pitting rate ı−2  −1 , present pit-
ting tests show that it increases with the fifth power of the flow
Fig. 21. Influence of cut-off depth on the relationship between pit volume and pit
diameter (Aluminum alloy Al 7075, upstream pressure 15 bar, constant cavitation velocity, which confirms the commonly accepted strong influence
number  = 0.9). of the flow velocity on pitting damage.

11
 The present work shows that the erosion damage (permanent [14] M.A. Dominguez-Cortazar, J.P. Franc, J.M. Michel, The erosive axial collapse of
deformation) during the incubation period and in particular the pit- a cavitating vortex: an experimental study, Journal of Fluids Engineering 119
(3) (1997) 686–691.
ting rate is not correlated in a simple way to the elastic limit of the [15] M. Futakawa, H. Kogawa, R. Hino, H. Date, H. Takeishi, Erosion damage
materials determined from conventional quasi-steady tensile tests. on solid boundaries in contact with liquid metals by impulsive pres-
Further investigations are needed to clarify this issue. In particular, sure injection, International Journal of Impact Engineering 28 (2) (2003)
123–135.
it would be necessary to better quantify the strain rate effect and [16] M. Futakawa, T. Naoe, H. Kogawa, C.-C. Tsai, Y. Ikeda, Pitting damage formation
try to correlate pitting test results with material properties at high up to over 10 million cycles off-line test by MIMTM, Journal of Nuclear Science
strain rates more representative of the actual loading conditions and Technology 40 (11) (2003) 895–904.
[17] C. Haosheng, L. Yongjian, C. Darong, W. Jiadao, Experimental and numerical
during the cavitation erosion process. investigations on development of cavitation erosion pits on solid surface, Tri-
bology Letters 26 (2) (2007) 153–159.
Acknowledgements [18] C. Haosheng, Iridescent rings around cavitation erosion pits on surface of mild
carbon steel, Wear 269 (7–8) (2010) 602–606.
[19] T. Momma, A. Lichtarowicz, A study of pressures and erosion produced by
This research was conducted under a NICOP project funded by collapsing cavitation, Wear 186–187 (Part 2) (1995) 425–436.
the Office of Naval Research. The authors wish to thank Dr. Ki- [20] G.L. Chahine, P. Courbière, Noise and erosion of self-resonating cavitating jets,
Journal of Fluids Engineering 109 (4) (1987) 429–435.
Han Kim from the Office of Naval Research (ONR) and Dr. Richard
[21] G.L. Chahine, P. Genoux, Simulation of the pressure field due to a submerged
Vogelsong from the Office of Naval Research Global (ONRG) who oscillating jet impacting on a solid wall, Journal of Fluids Engineering 106
supported this work. They are also very grateful to Dr. Farrel Mar- (1984) 491–496.
tin (Naval Research Laboratory) and Dr. Martin Donnelly (NSWCCD) [22] G.L. Chahine, V.E. Johnson, Mechanics and applications of self-resonating cav-
itating jets, in: International Symposium on Jets and Cavities, Miami, Florida,
for fruitful discussions. The authors would like to thank Dr. Michel ASME, WAM, 1985.
Suery and Mr. Charles Josserond (from SIMAP, the Materials and [23] Y. Tomita, A. Shima, Mechanisms of impulsive pressure generation and dam-
Processes Science and Engineering Laboratory of the Grenoble Insti- age pit formation by bubble collapse, Journal of Fluid Mechanics 169 (1986)
535–564.
tute of Technology) for offering us the opportunity to complete the [24] A. Philipp, W. Lauterborn, Cavitation erosion by single laser-produced bubbles,
tensile tests. Journal of Fluid Mechanics 361 (1998) 75–116.
[25] J.-P. Franc, Incubation time and cavitation erosion rate of work-hardening
materials, Journal of Fluids Engineering 131 (2) (2009) 021303.
References [26] J. Choung, S. Cho, Study on true stress correction from tensile tests, Journal of
Mechanical Science and Technology 22 (6) (2008) 1039–1051.
[1] L. Rayleigh, On the pressure developed in a liquid during the collapse of a [27] Y. Ling, Uniaxial true stress–strain after necking, AMP Journal of Technology 5
spherical cavity, Philosophical Magazine 34 (1917) 94–98 (Series 4). (1996) 37–48.
[2] R.T. Knapp, Recent investigations of the mechanics of cavitation and cavitation [28] G.E. Dieter, Mechanical behavior of materials under tension, in: Mechanical
damage, Transactions of the ASME (October) (1955) 1045–1054. Metallurgy, McGraw-Hill, New York, 1976, pp. 329–348.
[3] R.T. Knapp, Accelerated field tests of cavitation intensity, Transactions of the [29] G. La Rosa, G. Mirone, A. Risitano, Postnecking elastoplastic characteriza-
ASME (January) (1958) 91–102. tion: degree of approximation in the Bridgman method and properties of the
[4] R. Fortes Patella, J.-L. Reboud, A. Archer, Cavitation damage measurement by flow-stress/true-stress ratio, Metallurgical and Materials Transactions A 34 (3)
3D laser profilometry, Wear 246 (1–2) (2000) 59–67. (2003) 615–624.
[5] M. Dular, B. Bachert, B. Stoffel, B. Sirok, Relationship between cavitation struc- [30] J. Lemaitre, J.L. Chaboche, Mécanique des matériaux solides, Dunod, 1996.
tures and cavitation damage, Wear 257 (11) (2004) 1176–1184. [31] R.W. Armstrong, S.M. Walley, High strain rate properties of metals and alloys,
[6] T. Okada, Y. Iwai, K. Awazu, A study of cavitation bubble collapse pressures and International Materials Reviews A 53 (3) (2008) 105–128.
erosion. Part 1: A method for measurement of collapse pressures, Wear 133 [32] C. Pieralli, G. Tribillon, Traitement d’images 3D appliqué à la profilométrie
(1989) 219–232. optique pour l’étude du phénomène d’érosion de cavitation, Journal of Optics
[7] A. Abouel-Kasem, A.E. El-Deen, K.M. Emara, S.M. Ahmed, Investigation into 18 (1) (1987) 9–18 (Paris).
cavitation erosion pits, Journal of Tribology 131 (3) (2009) 031605. [33] B. Belahadji, J.P. Franc, J.M. Michel, A statistical analysis of cavitation erosion
[8] M. Dular, A. Osterman, Pit clustering in cavitation erosion, Wear 265 (5–6) pits, Journal of Fluids Engineering 113 (4) (1991) 700–706.
(2008) 811–820. [34] S.M. Ahmed, K. Hokkirigawa, Y. Ito, R. Oba, Scanning electron microscopy obser-
[9] Y. Tomita, T. Inaba, R. Uchikoshi, T. Kodama, Peeling off effect and damage vation on the incubation period of vibratory cavitation erosion, Wear 142
pit formation by ultrasonic cavitation, in: The International Conference on (1991) 303–314.
Hydraulic Machinery and Equipments, Timisoara, Romania, 2008, pp. 19–24. [35] H. Soyama, M. Futakawa, K. Homma, Estimation of pitting damage induced
[10] C. Haosheng, L. Shihan, Inelastic damages by stress wave on steel surface at the by cavitation impacts, Journal of Nuclear Materials 343 (1–3) (2005)
incubation stage of vibration cavitation erosion, Wear 266 (1–2) (2009) 69–75. 116–122.
[11] C. Haosheng, W. Jiadao, C. Darong, Cavitation damages on solid surfaces in sus- [36] J.P. Franc, J.M. Michel, Fundamentals of cavitation, in: R. Moreau (Ed.), Fluid
pensions containing spherical and irregular microparticles, Wear 266 (1–2) Mechanics and its Applications, Springer, 2004.
(2009) 345–348. [37] R. Rodríguez-Baracaldo, J. Benito, J. Cabrera, Tensile and compressive test in
[12] C. Haosheng, L. Jiang, C. Darong, W. Jiadao, Damages on steel surface at the nanocrystalline and ultrafine carbon steel, Journal of Materials Science 45 (17)
incubation stage of the vibration cavitation erosion in water, Wear 265 (5–6) (2010) 4796–4804.
(2008) 692–698. [38] D.R. Stinebring, J.W. Holl, R.E.A. Arndt, Two aspects of cavitation damage in the
[13] A. Karimi, Cavitation erosion of a Duplex Stainless Steel, Materials Science and incubation zone: scaling by energy considerations and leading edge damage,
Engineering 86 (1987) 191–203. Journal of Fluids Engineering 102 (December) (1980) 481–485.

12