683

Journal of Engineering Sciences

Assiut University
Faculty of Engineering
Volume: 42
No. 3
May 2014
Pages: 683-702

INVESTIGATION OF CAVITATION DAMAGE PROGRESS IN THE

INCUBATION PERIOD USING STEPWISE EROSION AND IMAGE
PROCESS TECHNIQUES
M. Osman 1 and S. M. Ahmed 2,*
1, 2
Mechanical Eng. Depat. College of Engineering, Majmaah University, Saudi Arabia
Received 20 March 2014; accepted 15 April 2014

ABSTRACT
Cavitation damage process of stainless steel during the incubation period was detected and
evaluated using stepwise erosion technique and image processing. The damage experiments were
conducted using the vibratory test rig. The scanning electron microscope images at different
locations on eroded surface using stepwise technique were presented. In addition, the surface
roughness was measured with the testing time. Discrete wavelet transforms and fractals had been
adopted to extract the relevant features. It was found that the feature parameters: Shannon entropy,
energy loss, fractal dimension and intercept show a fairly linear increase with the testing time. It
was also observed that the feature parameters have reasonable correlations with the roughness
parameters.
Keywords: Cavitation erosion, electron microscopy, surface analysis, stainless steel

1. Introduction
Cavitation is defined as the repeated nucleation, growth, and violent collapse of vapor
cavities, or bubbles, in a liquid [1-3]. These collapses exert intense stress pulses, damage
and cause material loss and erosion on nearby solid surfaces. Cavitation damage is one of
the major problems confronting the designers and users of modern high-speed
hydrodynamic systems as reported by Hammitt [1]. It is a serious problem in hydraulic
turbines, pumps, valves, control devices, hydraulic structures, sluices, energy dissipators,
ship propellers, hydrofoils, bearings, diesel engine wet cylinder liners, aircraft engines,
sonar domes, acoustic signal devices, and processing and cleaning equipment [2]. This
damage represents a significant problem related to the operating life and economic values
of these components [4]. One of the most common methods for predicting the amount of
cavitation erosion is based on measurement of mass loss from materials. However, this
method has limitations, because the weight loss is global while the cavitation erosion is a
local phenomenon. Besides, this method overlooks the early stages of the effects of the

* Corresponding author.
E-mail address: shemy2007@yahoo.com
684
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
cavitations and sums only the overall resulting material erosion. However, under cavitation
damage, the material surface may be deformed and cracked causing design changes of this
surface even before the material begins to have a detectable loss. Therefore, it is essential
to characterize the surface change in the early stage of cavitation. The eroded features such
as plastic deformation, fracture initiation and its progress are dependent on the load history
and surface properties under cavitation erosion. Hence, of particular importance in the
recent researches is the need to develop numerical methods providing a reliable and
accurate characterization of surface topography with the change in the time.
Literature reports several types of devices and techniques used in laboratory tests for
collecting data of cavitation erosion. The ultrasonic vibratory test rigs are the most
commonly used to conduct the accelerated tests, allowing a large quantity of data to be
accumulated over a relatively short period of time using a wide variety of test conditions,
liquids and materials. As reported by Hammitt [1] most of the data obtained from such
tests are very helpful in clarifying the cavitation erosion mechanism. A consensus showed
that material removal in multiple–impact situations (e.g. cavitation erosion, liquid-droplet
erosion, and of solid–particle erosion) is not a result of single impulses or impacts, but
rather, it is damage accumulations over thousands of impacts before a particle is dislodged
as discussed by Preece and Hansson [2].
It is well established that the rate of erosion is a function of the exposure period, where
this rate increases from negligible values, reaches a maximum, then decreases and levels
off to a steady value [5]. Thus, cavitation erosion history of a material can be divided into
four periods according to ASTM standards G-32 [3]: an incubation period; an acceleration
period; a steady-state period and attenuation period. Among these periods, incubation
period gained a large share of the research as it represents the importance of the process of
cavitation erosion in general due to the existence of plastic deformation, pit formation and
surface hardening. The spread of cracks, which control removals often occur in successive
periods. Some authors [6-9], in their experimental works found that the pits have a
minimal effect in developing erosion. From the analysis of the eroded surface and the
dislodged particles for different test conditions, it was found [7, 10-12] that the plastic
deformation development and the surface hardening lead to the fatigue failure. However,
there is little research [13] that tries to evaluate the erosion resistance of metallic materials
by analyzing the plastic deformation behavior of metallic materials during the incubation
period. As the cavitation erosion is locally distributed, the surface profile technique may
not be a representative to the whole damage on the surface.
Recently, the image analysis technique is considered a worthwhile tool for being able
to perform analyzes of fast, inexpensive and non-destructive for many processes. Image
analysis is efficient in many tribological applications, e.g. wear debris morphology, texture
classification and surface quality in machining as well as it is an appropriate tool to
qualitatively and quantitatively characterize the early stage of cavitation erosion process
[11, 12, and 14]. The topographical images of cavitation erosion surfaces at water and oil-
in-water emulsions were quantified using fractal and wavelet techniques [15, 16]. The
image feature parameters of both techniques and the weight loss decrease with adding oil
to water. Also, it was found that the variation of image feature parameters versus
concentration of oil-in-water emulsions had a general trend that did not depend on
magnification factor.
 685
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….
In the present work, fractal and wavelet analysis of the stepwise erosion combined with
relocation scanning electron microscopy were used to delineate the real erosion processes
within the incubation period in considerable details. The surface roughness evolution
within the incubation period was also measured for the stepwise erosion.

2. Image analysis methods

Various image processing tools can be used to analyze damage surface images and
extract the relevant features. Wavelet transforms and fractals were adopted in this study,
and are described briefly below and elsewhere [15, 16].
2.1 Wavelet and wavelet packet decomposition
The main advantage of using wavelets is that it provides multi-resolution analysis to
improve the image quality obtained from microscopy techniques, such as SEM and others.
The different wavelet transform functions filter out different range of frequencies (i.e. sub
bands) by decomposing the image into low frequency and high frequency sub band
images. When applying the wavelet transform, original image can be decomposed into four
sub-band images. The decomposition process is recursively applied to the low-frequency
sub-band to generate the next level of the hierarchy. That is, the two-dimensional discrete
wavelet transform (2D DWT) [17] decomposes an image into several sub-bands according
to a recursive process as shown in Fig 1. These include LH1, HL1 and HH1 which
represent detail images and LL1 which corresponds to an approximation image, which is
then decomposed into second-level approximation and detail images. This process is
repeated to achieve the desired level of resolution. The obtained coefficients for the
approximation and detailed sub-band images are useful features for texture categorization.

Fig. 1. Pass-band structure for a two dimensional sub-band transform with three levels.
As an extension of the standard wavelets, wavelet packets represent a generalization of
the multi-resolution analysis and use the entire family of sub-band decompositions to
generate an over complete representation of images [18]. There are many outstanding
properties of wavelet packets that encourage research in many widespread fields.
An image is decomposed into one approximation and three detail images in 2-D
discrete wavelet packet transforms (2-D DWPT). The approximation and the detail images
are then decomposed into a second-level approximation and detail images, and the process
is repeated. The wavelet packet tree for 3-level decomposition is shown in Fig. 2. The
standard 2-D DWPT can be implemented with a low-pass filter h and a high-pass filter g
[19]. The 2-D DWPT of an N x M discrete image A up to
686
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702

level (P+1 (P £ min(log2 N + log2 M)), is recursively defined in terms of the

coefficients at level p as follows:
c4PK1,(i , j )  m n h(m) h(n) cKP ,( m2i ,n2 j )
(1)
cP1
4 K 1,( i , j )  m n h(m) g (n) c P
K ,( m2i ,n2 j )
(2)
c4PK12,(i , j )  m n g (m) h(n) cKP ,( m2i ,n2 j )
(3)
c4PK13,(i , j )  m n g (m) g (n) cKP ,( m2i ,n2 j )
(4)
c0
Where 0 is image A and K is an index of the nodes in the wavelet packet tree
denoting each sub-band; h and g are the filter coefficients of low-pass and high-pass filters,
respectively. Supposing that Haar basis has been used, h={- 0:7071; 0:7071}, and g =
P
{0:7071; 0:7071}. At each step, the image cK is decomposed into four quarter-size
c P1 , c P1 , c P1 , c P1
images 4 K 4 K 1 4 K 2 4 K 3 . The capital letters (N or M) are maximum constants
defined by the image size. However, small letters (m or n) are defined at each step. For
example, when P = 5, P can be 1, 2, 3, or 4 and so on.
The Shannon entropy in different sub-bands is computed from the sub-band coefficient
matrix as:

Entropy P (k )  
2 2
i
cP
j k ,( i , j )
log ckP,(i , j )
(5)
Energy P (k )  
2
i
cP
j k ,( i , j )
(6)
(k) (k)
Where Energy p and Entropy p are the energy and entropy of the image projected
onto the subspace at node (p, k). The entropy of each sub-band provides a measure of the
image characteristics in that sub-band. The energy of each sub-band provides a measure of
the image characteristics in that sub-band. The energy distribution has important
discriminatory properties for images and as such can be used as a feature for texture
classification. From the equations above, it follows that the wavelet entropy is minimum
when the image represents an ordered activity characterized by a narrow frequency
distribution, whereas the entropy is high when an image contains a broad spectrum of
frequency distribution.

Fig. 2. Wavelet packet tree decomposition.

 687
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….

2.2 Fractal analysis

Fractal geometry is a well-known non-traditional method which has found many
applications in science and engineering. It is common knowledge that many objects in
nature are of irregular form which cannot be described by Euclidean geometry. These non-
Euclidean objects are called fractals, and can be described by non-integer numbers. These
non-integer numbers define the fractal dimension (FD) of an object. The main concept of
fractal geometry analysis is that a fractal dimension can be considered as a quantitative
measure of object surface heterogeneity because of its inherent self-similarity features. In a
simplified representation, one could interpret the fractal dimension as a measure of
heterogeneity of a set of points on a plane, or in space. Many works were reported on the
application of fractal geometry to describe complex physical phenomena such as
turbulence, brittle fracture of materials, machining, and tool wear [20, 21]. Fractal
characterizations have also been used to describe complex two or three-dimensional
surfaces, such as deposited surfaces [22], wear-erosion surfaces [23, 24], and wear
particles [25-28]. Eroded surfaces are likely to be multi-fractal because they are produced
by the accumulative effects of multiple impacts, i.e., rough surfaces. Fractal dimension at
low frequency describes raw structures, while fractal dimension at high frequency
represents texture. Textural fractal should define very fine feature in object boundary,
surface, area, or volume. Structural fractal should describe an overall object morphology.
Therefore the distinction between structural and textural fractals allows investigators to
characterize microstructure and macrostructure effects on the surface behavior.
Various fractal analytical techniques are applied to micrograph images of eroded surfaces
such as Fourier, Kolmogorov, Korcak, Minkowski, root mean square, Slit Island, etc. These
techniques differ in computational efficiency, numerical precision and estimation boundary.
The most efficient procedure for measurement of the fractal dimension FD of eroded
surfaces, and one which allows characterization of anisotropic surface as well, seems to be
through Fourier analysis [29, 30]. Therefore, Fourier analysis is adopted to estimate fractal
values in this work. For a surface image represented by the function I(x, y), the power
spectral density (PSD) is equal to the square of the Fourier transformation F (u, v) of the
surface function I(x, y). The power spectral density function is defined as;
S (u, v)  F (u, v)
2

(7)
Where u and v are the spatial frequencies (number of waves per unit wave length) in
the x and y directions respectively. The PSD is converted to the polar coordinate system S

(f), such that f  u  v . The value of S(f), at each radial frequency f, is averaged over
2 2

angular distributions. For a fractal surface, Fig.3 (a), the power spectrum shows a linear
variation between the logarithm of S(f) and the logarithm of the frequency as shown in
Fig.3 (b).The slope of the linear regression line  is related to FD by equation [13, 31]:
8
FD  (8)
2
It is reported in literature that fractal dimension and intercept are significant fractal
parameters that describe the irregularity and complexity of the surfaces. Moreover, the
688
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
intercept correlates well with the overall magnitude of roughness of the observed texture
appearance of the surface images. In the present study, fractal analysis is going to be used
to assign numerical values to indicate the development of stepwise erosion for cavitation
eroded surfaces.
Log Log plot of Magn. vs Freq.
26
data 1
y = - 2.5*x + 26 linear
24
data 2

22

Log Magnitude
20

18

16

14
(a) (b)
12
0 1 2 3 4 5 6
Log Frequency

Fig. 3. (a) A surface image of polished surface, and (b) Power spectrum of its
Fourier transform, plotted as directionally averaged of log (PSD) versus log
(frequency).

3. Experimental procedure
The test material was austenitic stainless steel SUS 304, whose nominal mechanical
properties and chemical composition are shown in Table 1. To ensure metallurgical
uniformity, all test specimens were cut from single round rod of the material. The round
specimens were machined to flat-surfaced discs of diameter of 15.9 mm and a height of 11
mm with threaded sections for attachment to the cavitation vibratory horn. . Initial surface
roughness of the specimens is known to be of a considerable importance in creating pits
[7], to facilitate crack initiation [32], to develop the fracture and to increase the rate of
linear particle dislodgment [33]. To minimize these problems in the present study, the
working faces of specimens were highly polished with grade 3000 dry emery paper.
Cavitation erosion experiments were performed using ultrasonic cavitation test
apparatus shown in Fig. 4, which is in accordance with ASTM standard G32-06 [3]. The
apparatus consists of a quartz transducer oscillating at a resonant frequency of 20 kHz,
attached to an exponentially shaped prop, which amplifies the amplitude of vibration. The
double amplitude (peak to peak) was 50 µm. This alternating sound field generates
pressure fluctuation with the horn-tip immersed in liquid. During one half of each vibration
cycle, a low pressure is created at the test specimen surface, producing cavitation bubbles.
During the other half of the cycle, bubbles collapse at the specimen surface.
The specimens were immersed in 1000 ml open glass beaker, containing 700 ml of tap
water. Since the test liquid temperature markedly affects the degree of erosion [34,35], the
test water temperature in the beaker was controlled by cooling water circulation to be
within 22±1 °C , as shown in Fig.4.
Four specimens were tested at successive intervals within incubation period, two
specimens for capturing the eroded surface image and the other two for surface roughness
measurement. Such short experimental time (i.e.: elapsed time is within 5 min.) was
 689
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….
scheduled to reduce other damage effect on the surface, such as corrosion. The surface
roughness was 2-dimensionally measured by a stylus profilometer in the direction of lay
and in perpendicular direction to the lay. Then the average value of the surface roughness
parameter was estimated for the two directions. The average value of roughness parameters
Ra (mean surface roughness) and Rq (Root-mean-square roughness) was calculated from
five random profiles of sampling length 2.5 mm at each test time. The duration of each
interval time must be sufficient to produce gradually change in the features of the damage.
Therefore, a 5 min. was decided for a total of 45 min. of testing on each specimen. The
specimens were ultrasonically cleaned in acetone, dried in open air, and weighed by a
digital balance with a sensitivity of 0.01 mg.
It is known that cavitation erosion is localized and unevenly distributed. So in order for
the stepwise erosion results to be representative, photographs were taken at four locations
on the surface of the sample, with two near the center and the other two near the edge. For
comparative examination, capturing an image of the same area on the surface of the sample
after each test requires relocation of the sample in the microscope, which has been done
through the rotation of the sample and observing the lay of specimen using small
magnification and comparing it to the previous images.

Fig. 4. Schematic view of test apparatus.

690
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702

Table 1.
Mechanical properties and chemical composition of SUS 304
Density 8027 kg/m3
Yield stress 517 MPa
Tensile strength 758 MPa
Elastic modulus 193 x 103 MPa
Hardness, BHN 240
Elongation percent 60
Nominal chemical C 0.08 max, Mn 2max, Si 1 max, P 0.045max
composition S 0.03 max, Cr 18 – 20, Ni 8 – 12

4. Results and Discussion

4.1 Morphologies of eroded surfaces
Figure 5 shows the sequence of images before and after testing for times of 5, 10 …and
35 min. It is clearly seen on the images of the edge and center positions at t = 0 min., there
are a number of horizontal parallel polishing lines. A number of black spots seen on the
edge images, at t = 5 min., represents contaminations by dust particles that do not appear
on the subsequent images. Therefore, the lines and the spots are not related to the
cavitation. It was reported [6, 7] that when specimens of ductile austenitic stainless steel
were exposed to cavitation, local plastic deformation of its surface was observed. This
deformation was in the form of surface undulations with the appearance of defects, such as
grain boundaries and slip bands as well as the formation of pits. With the development of
testing time, these undulation changes in width and height, but generally they do not
change in shape and number. It is noted in these images for plastic deformation slip bands
are at least of 40 µm in length.. Such long undulations reveal that the shock pressure
developed by the bubbles collapse plays a vital role in developing the erosion [9]. The
images also show that the loss of significant particles in the evolution of the erosion
process began along the slip bands as it is clear from the images of the eroded surface at
the edge of time of 20 to 25 min. The particles were fallen off due to fatigue. Once the
work hardening capacity of the surface reaches a critical level, fatigue cracks initiate
plastically in the slip bands at microstructural discontinuities. The latter, being
inhomogeneous and "condensed" plastic strains, changes locally and unevenly the
microgeometry of the sample surface (micronotches), also causes local stress
concentrations.
Another feature of eroded surface is the pit formation encircled in the images of Fig. 5.
The size of these pits is in the order of µm. Such a tiny pit arises from the exposure to
highly erosive microjets of pressures larger than the strength of the tested material. It is
clearly seen from the sequence of images that the shape and diameter of pits do not change
at all. Based on this observation on the pits feature and elsewhere [5], it can be concluded
that the pits scarcely play an important role in the development of erosion.
 691
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….

Fig. 5. Sequences of images show the progression of damage with time at two
positions on the sample: (a) the center, (b) the edge.
692
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702

Fig. 5. Continue
 693
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….

4.2 Change in surface roughness

The results of the average roughness and the root mean square roughness with
the test time are plotted in Fig. 6. These data can be divided into two stages based on the
incubation period: stage I for t ≤ 15 min., and stage II for t > 15 min. In both stages, the
roughness values increase linearly with the time, where this rate of the increase is larger in
the second stage. In both stages I and II, the material surface is plastically deformed leading
to a uniform rate of roughness. Meanwhile, the increase in the rate of roughness in the
second stage II is caused by the initiation of cracks as shown at the edge of the image for 20
min. Many results about the measurement of roughness in the incubation stage and
subsequent stages were reported in the literature [7, 12, 36]. However, the present
measurements show slightly different results compared to the reported results due to the fact
that they are the average value of the measurement of two profiles, one in the direction of lay
and the other in perpendicular direction to it. The assigned profile includes the whole eroded
surface. Similar results have been reported by Yabuki et al. [37] but with different scale for
the roughness measurement. They proposed "surface increment percentage" index to
represent the surface modification during the incubation period. The increment of the surface
area was determined from the material surface profile data, measured through roughness
meter. According to this index, the incubation period of cavitation erosion is to be divided
into two stages. They found that, in the first stage, the rate of surface increment is related to
the hardness of the metallic materials while in the second stage this rate is related to the
strain hardening exponent. Therefore, Yabuki et al. [37] concluded that surface increment
index can be applied to the rapid determination of the level of erosion resistance, which in
turn can lead to the rapid prediction of the service life of material. Recently, Matsmura [38],
in his analysis for this methodology, saw that this method is inevitable for the determination
of cavitation intensity in hydraulic machines.

Fig. 6. Change of average Ra and RMS roughness Rq with test time t.

4.3 Feature extraction
In this section, the features parameters of images such as wavelet entropy, wavelet
energy, and fractal dimension and intercept values are extracted. Then, these parameters
are correlated to the surface roughness parameters. A set of 8 images (4 images per sample
at different locations) were collected to represent the whole surface damage. Before any
data processing was made, these raw images were visually assessed to avoid any effect on
694
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
the parameters of image like change in dimension, change in structure value due to
contaminations. Therefore the data at test time of 5 min is omitted due to contamination as
reported in Sec. 4.1.
The results of the feature extraction, performed by using 2D-DWPT and fractal analysis
described in Sec.2 are given in Figs 7-10. Shannon entropies and energy of wavelet packet
decomposition of image were calculated with the testing time from all the selected sub-bands
of the image as shown in Figs. 7 and 8. Figures 9 and 10 show the fractal dimensions and
intercept for the image of eroded surface at different times. Fitting results between the
specimen's image parameters and exposure time and the correlation coefficient, R2 , are
shown in these figures. Entropy has the best linear correlation with time with adjusted R2
value of 0.98. Then, followed by intercept, fractal dimension and energy with R2 values of
0.90, 0.90, and 0.89 respectively. From these figures, it is shown that the slope of the line of
entropy and its intercept increase with time, while it decreases for energy. Tao et al. [39]
used the image energy loss D as a characteristic to estimate the corrosion loss. The image
loss D was calculated as follow:
Dj = E0 - Ej (9)
Where j means the exposure time and is the energy of uneroded specimen. The
energy loss versus time is plotted in Fig. 11.
11
x 10
0
y = 6.2*109*x - 7.1*1011
-2
R2 = 0.976
Shannon entropy

-4

-6

-8

-10

-12
0 5 10 15 20 25 30 35
Test time (min)

Fig. 7. Change of Shannon entropy with test time t.

7
x 10
6
y = - 1.5*105*x + 3.8*107
R2 = 0.8867
5
Energy

2
0 5 10 15 20 25 30 35
Test time (min)
Fig. 8. Change of wavelet energy with test time t.
 695
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….

3.5
y = 0.027*x + 1.9
R2 = 0.9005

Fractal dimensional
3

2.5

1.5
0 5 10 15 20 25 30 35
Test time (min)

Fig. 9. Change of fractal dimension with test time t.

38
y = 0.16*x + 30
R2 = 0.9025
36
Fractal intercept

34

32

30

28
0 5 10 15 20 25 30 35
Test time (min)
Fig. 10. Change of fractal intercept with test time t.
6
x 10
18
16 y = 4*105*x + 2.5*106
R2 = 0.886
14
12
10
Energy loss

8
6
4
2
0
0 5 10 15 20 25 30 35
Test time (min)

Fig. 11. Change of wavelet energy loss of eroded surface image with the time.
696
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
From this figure, it can be seen that the energy loss increases with time in a similar
manner to the behaviors of entropy, fractal dimension and intercept with time.
Each feature corresponds to a visually recognizable property of the image described in
Section 4.1. That is, for uneroded surfaces with homogeneous textures the results show
relatively low values of entropy, energy loss, fractal dimension and slope. While, these
extracted features have high values for eroded surfaces due to the coarseness of the surface
textures. To verify these results, Figs. 12 & Fig.13 examines the possible correlation
between feature parameters extracted from images and corresponding average roughness,
and root mean square roughness , given in Fig.6. It is clear that the correlations of
feature parameters with the roughness parameters are similar to the correlations with the
testing time. It is worth noting that Shannon entropy exhibits the best correlation with
and with correlation coefficient 0.98. It can be concluded that the image texture
analysis by discrete wavelet analysis and fractal analysis of the cavitation eroded surface
can be used to predict the nature variation of cavitation damage.
11
x 10
-3.5 y = 6.5*1011*x - 7.2*1011
R2 = 0.967
-4
Shannon entorpy

-4.5
-5
-5.5
-6
-6.5
-7
-7.5
0 0.1 0.2 0.3 0.4 0.5 0.6
R (µm)
a

(a)
6
x 10
18
y = 2.1*107*x + 2.6*106
16
R2 = 0.8636
14
12
Energy

10
8
6
4
2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ra (µm)

(b)
 697
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….

3.2
y = 1.5*x + 2
3 R2 = 0.8848

Fractal dimensional 2.8

2.6

2.4

2.2

2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ra(µm)

(c)
37
y = 8.3*x + 31
R2 = 0.8937
36
Fractal intercept

35

34

33

32

31
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ra (µm)

(d)
Fig. 12. Feature parameters: Shannon entropy (a), energy (b), fractal dimension
(c), fractal intercept (d) versus .

5. Conclusions
Image processing based on discrete wavelet transform and fractal analysis was used to
characterize the cavitation damage images. Those images were captured using the stepwise
erosion technique. The following results can be drawn:
(1) The extracted feature parameters; Shannon entropy, energy loss, fractal dimension
and fractal intercept change linearly with exposure time.
(2) The relationships between feature parameters and roughness parameters were linear.
The best correlation was for Shannon entropy.
(3) The results indicate that the image analysis procedures rep promising techniques
since they are effective in characterizing the changes of surface topography with
exposure time.
698
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
(4) The surface topography with test time showed the plastic deformation development
that leads to the fatigue
11
failure was the predominant mechanism.
x 10
-3
y = 8.4*1011*x - 7.2*1011
R = 0.9672
-4
Shannon entorpy

-5

-6

-7

-8
0 0.1 0.2 0.3 0.4 0.5 0.6
R (µm)
q
(a)
6
x 10
y = 2.8*107*x + 2.5*106
15 R2 = 0.8681
Energy

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Rq (µm)

(b)
3.2
y = 2*x + 2
R2 = 0.8804
Fractal dimensional

2.8

2.6

2.4

2.2

2
0 0.1 0.2 0.3 0.4 0.5 0.6
Rq(µm)

(c)
 699
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….
37
y = 11*x + 31
36 R2 = 0.8912

35
Fractal intercept
34

33

32

31

30
0 0.1 0.2 0.3 0.4 0.5 0.6
R (µm)
q
(d)
Fig. 13. Feature parameters: Shannon entropy (a), energy loss (b), fractal
dimension (c), fractal intercept (d) versus .

REFERENCES
[1] Hammitt FG. Cavitation and multiphase flow phenomena. New York: McGraw- Hill;
1980.
[2] Preece CM, Hanssson ILH. A metallurgical approach to cavitation erosion, Advances
in the Mechanics and Physics of Surfaces. Latanision, R. M. and Courtel, R. J., Ed.,
Harwood Academic 1; 1981, p.199-253.
[3] ASTM Stand-G32-06. Standard test method for cavitation erosion using vibratory
apparatus. In: Annual book of ASTM standards, Part 03.02.
Philadelphia, PA: ASTM; 2002. p. 98-112.
[4] March P, Hubble J. Evaluation of relative cavitation erosion rates for base materials,
weld overlays, and coatings, Tennessee Valley Authority Engineering Laboratory,
Norris, TN, Report No. WR28-1-900-282, September 1996.
[5] ITTC. The Specialist Committee on Cavitation Erosion on Propellers and Appendages
on High Powered/High Speed Ships: Final Report and Recommendations to the 24th
ITTC. In: Proc. 24th International Towing Tank Conference, 2005; 2:509-542.
[6] Abouel-Kasem A, Ezz El-Deen A, Emara KM, Ahmed SM. Investigation into
cavitation erosion pits. ASME, J Tribol 2009; 131: 031605-1-031605-7
[7] Ahmed SM, Hokkirigawa K, Oba, R. Fatigue failure of SUS 304 caused by vibratory
cavitation erosion. Wear 1994; 177:129- 137.
[8] Ahmed SM, Hokkirigawa K, ItoY, Oba R, Matsudaira Y. Scanning electron
microscopy observation on the incubation period of vibratory cavitation erosion. Wear
1991;142: 303-314.
[9] Miyazaki K, Ahmed SM, Oba R. High speed observations of the vibratory cavitation
accompanying hard erosion, JSME international journal. Ser. B, Fluids and thermal
engineering 1993; 36(4): 511-516.
[10] Abouel-Kasem A, Ahmed SM. Bubble structures between two walls in ultrasonic
cavitation erosion. ASME, J Tribol 2012; 134(2): 02172-1-9.
[11] Abouel-Kasem A, Emara KM, Ahmed SM. Characterizing cavitation erosion particles
700
JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702
by analysis of SEM images. Trib Int 2009; 42(1); 130-136.
[12] Abouel-Kasem A, Saleh B, Ahmed SM. Quantitative analysis of cavitation erosion
particle morphology in dilute emulsions. ASME, J Tribol 2008; 130: 041603-1-6.
[13] Pohl M, Stella J. Quantitative CLSM roughness study on early cavitation-erosion
damage. Wear 2002; 252: 501–511.
[14] Szkodo M. Relationship between microstructure of laser alloyed af C45 steel and its
cavitation resistance, Proceedings of the 13th Scientific International Conference
"Achievements in Mechanical and Materials Engineering” AMME’2005, Gliwice-
Wisáa, 2005: 643-646.
[15] Abouel-Kasem A, Alturki FA, Ahmed SM. Fractal analysis of cavitation eroded
surface in dilute emulsions. Trans. ASME, J Tribol 2011;133(4): 041403-1-9.
[16] Alturki FA, Abouel-Kasem A, Ahmed SM. Characteristics of cavitation erosion using
image processing techniques. J Tribol 2013; 135: 041403-1-7.
[17] Lewis AS, Knowles G. Image compression using the 2-D wavelet transform, Image
Processing. IEEE Trans. 1992; 1(2): 244 -250.
[18] Huang K, Aviyente S.Wavelet feature selection for image classification. IEEE Trans.
Image Process 2008;17(9): 1709–1720.
[19] Mallat S. A wavelet tour of signal processing. New York :Academic; 1999
[20] Russ JC. Fractal Surfaces, New York: Plenum; 1994.
[21] Kassim AA, Zhu M, Mannan MA. Texture analysis using fractals for tool wear
monitoring. IEEE ICIP 2002: III - 105- III-108.
[22] Guessasma S, Montavon G, Coddet C. On the implementation of the fractal concept to
quantity thermal spray deposit surface characteristics. Surf Coat Technol 2003; 173(1):
24–38.
[23] Ribeiro LMF, Horovistiz AL, Jesui'no GAO, Hein LR, Abbade NP, Crnkovic SJ.
Fractal analysis of eroded surfaces by digital image processing. Mater Lett 2002;
56:512–517.
[24] Rawers J, Tylczak J. Fractal characterization of wear-erosion surfaces. J Mater Eng
Perform 1999; 8(6): 669–676.
[25] Yuan CQ, Li J, Yan XP, Peng Z. The use of the fractal description to characterize
engineering surfaces and wear particles, Wear 2003; 255: 315–326.
[26] Stachowiak GW, Podsiadlo P. Characterization and classification of wear particles and
surfaces. Wear 2001; 249:194–200.
[27] Podsiadlo P, Stachowiak GW. Scale-invariant analysis of wear particle morphology-a
preliminary study. Tribol. Int. 33, 289–295(2000).
[28] Stachowiak GW. Numerical characterization of wear particles morphology and
angularity of particles and surfaces. Tribol. Int. 31(1–3), 139–157(1998).
[29] Russ JC. Fractal analysis. Encyclopedia of Material: Science of Technology, Elsevier
Science Ltd 2001: 3247–3254.
[30] Zhang, J. Detection and monitoring of wear using imaging methods. Ph.D. thesis, The
Netherlands:, University of Twente, Enschede; 2006.
[31] Babadagli T, Develi K. Fractal analysis of natural and synthetic fracture surfaces of
geothermal reservoir rocks. In: Proc. World Geothermal Congress 2000 Kyushu -
Tohoku, Japan, May 28 - June 10; 2000.
[32] Ahmed SM, Hokkirigawa K, Oba R, Kikuchi K. SEM observation of vibratory
cavitation fracture–mode during the incubation period and the small roughness effect.
JSME Int J 1992; 34(3): 298–303.
[33] Ahmed SM, Hokkirigawa K, Kikuchi K, Higuchi J, Oba R. SEM studies of particles
produced by cavitation erosion. JSME Int. J. Ser. B 1993; 36(4): 517–523.
[34] Ahmed SM. Investigation of the temperature effects on induced impact pressure and
cavitation erosion. Wear 1997; 218: 119–127.
[35] Saleh B, Abouel-Kasem A, Ezz El-Deen A, Ahmed SM. Investigation of temperature
 701
M. Osman and S. M. Ahmed, Investigation of cavitation damage progress in ………….
effects on cavitation erosion behavior based on analysis of erosion particles.Trans
ASME, J Tribol 2010; 132: 031601-6.
[36] Ahmed SM, Hokkirigawa K, ItoY, Oba R, Matsudaira Y. Developing stages of
ultrasonically produced cavitation erosion and corresponding surface roughness. JSME
Int. J. Ser. II 1990; 33(1): 11-16.
[37] Yabuki A, Noishiki K, Komori K, Matsumura M. The surface behavior of metallic
materials during the incubation period of cavitation erosion. ASTM STP 2001;
1339:357-369.
[38] Matsumura M. Erosion-corrosion: An Introduction to Flow Induced Macro-Cell
Corrosion. Bentham e Books; 2012.
[39] Tao L, Song SZ, Zhang XY, Zhang Z, Lu F. Image analysis of atmospheric corrosion
of field exposure high strength aluminum alloys. Appl Surf Sci 2008; 254: 6870-6874.
‫‪702‬‬
‫‪JES, Assiut University, Faculty of Engineering, Vol. 42, No. 3, May 2014, pp. 683 – 702‬‬

‫"التحقيق في تقد ضرار ت كل التك ف باستخدا‬

‫الت كل التدريجي تقنيات تح يل الصورة"‬
‫الم خص العربى‬
‫ﻟ ﺻد ﺧ ﻓﺗر ا حتضا ﺑ ﺳﺗﺧدا ﺗ ﻧﯾ‬ ‫ﺗ اﻟﻛﺷف اﻟت ييم ﻟعم ي اﻟت ﻛل اﻟن ﺗج عن اﻟت ف ﻟ فو اﻟﻣ‬
‫اﻟت ﻛل اﻟت رﯾجي ﻣع ﻟﺟ اﻟﺻ ر‪ .‬ﺟرﯾ اﻟﺗﺟ ر ﺑ ﺳﺗﺧدا اﻟج اﻟ ا ﻟتوﻟي اﻟت ف ﺗم اﻟت ا ﺻاور‬
‫ﻟ ت ﻛااال اﻟتااا رﯾجي ﺑ ﺳااات ا اﻟﻣﺟ ر اﻹﻟﻛﺗر ﻧي اﻟﻣ ﺳﺢ ﻓي ﻣ اﻗﻊ ﻣﺧﺗ ف ع ﻰ اﻟسااااﺢ اﻟمت ﻛااال ﺑ ﻹﺿ ﻓ إﻟﻰ‬
‫اﻟموﺟااا اﻟم ااا د اﻟ ن ﺳااا‬ ‫ﻟ ‪ ،‬ﺗ ﻗﯾ ﺧﺷ ﻧ اﻟﺳ ﺢ ﻣﻊ ﻗ ا ﺧت ااا ر‪ .‬ﻗااا ﺗااام اﺧتيااا ر ﻣ اااو‬
‫اﻟ سي ﯾ ﺳت ا اﻟمع م (اﻟ اﻣت ا ) اﻟ ﺻ ﺑت ك اﻟصور‪.‬‬
‫ﯾا د ﺧ ﯾ إﻟﻰ حد‬ ‫ﻣع ﻣ اﻟصور ﻣ ل‪ :‬اأﻧت ﺑي ﻟش ﻧو ‪ ،‬ﻓ دا اﻟ ﻗ ‪ ،‬اﻟﺑعد اﻟ سي ي ﺗظ‬ ‫ﺗﺑﯾ‬
‫ﻣع ﻟ ﻣﻊ ﺧﺷ ﻧ اﻟساﺢ اﻟمت ﻛل‪.‬‬ ‫ﻣ ﻣﻊ ﻗ ا ﺧﺗﺑ ر‪ .‬ﻟ حظ ﯾﺿ ﻣع ﻣ اﻟصور ﻟدﯾ ا ارﺗﺑ‬