Académique Documents
Professionnel Documents
Culture Documents
510516
Abstract
Velocity is a critical test variable in erosion, and can easily overshadow changes in other variables, such as target material, impact
angle, etc. The effect of velocity on erosion rate was studied in 7030 brass cold worked and annealed. and FeC martensite
as-quenched and tempered. and it was found, as previously shown, that erosion rate is dependent on velocity by a power law, given by
ER s kV n. However, the velocity exponent n was found to be target material independent and is governed by test conditions, including
particle characteristics and the erosion test apparatus. In addition, n is not dependent on the erosion mechanism. Results from tested
as-quenched martensite and tempered martensite showed that the exponent is approximately 2.9 for both materials, even though
martensite eroded by a brittle cracking mechanism, while tempered martensite eroded by a plastic deformation mechanism. No difference
in the erosion rate relationship was found between the fully annealed and the 70% cold worked brass. The exponent n was found to
change over time with nominally the same erosion test conditions, indicating that n is very sensitive to slight changes in erodent particles
andror the test apparatus, and that it must be measured periodically if erosion results generated at different times are to be compared. 1
q 1999 Published by Elsevier Science S.A. All rights reserved.
Keywords: Solid particle erosion; Particle velocity; Velocity exponent; Erosion test methodology
1. Introduction
The definition of erosion by the American Society for
Testing and Materials is the progressive loss of original
material from a solid surface due to the mechanical interaction between that surface and a fluid, a multi component
fluid, or impinging liquid or solid particles w1x. The
erosion rate of a material is dependent on several factors,
including target material properties and erosion test conditions of angle, velocity, temperature, particle flux, and
erodent. Erosion rate has been shown w25x to follow an
empirical power law relationship with velocity:
Erosion Rate s kV n ,
where V is velocity, k is a constant, and n has values
between 2 and 3.5 for metallic materials. Brittle materials
tend to have a larger n range, from 2 to 6.5 w2x. McCabe et
al. w4x have shown that an n value of 2 fits erosion data for
various steel morphologies, including martensite and
Corresponding author
Prime Novelty: The velocity exponent n is independent of both target
material and erosion mechanism and is highly dependent on erosion test
conditions for the materials tested.
1
0043-1648r99r$ - see front matter q 1999 Published by Elsevier Science S.A. All rights reserved.
PII: S 0 0 4 3 - 1 6 4 8 9 9 . 0 0 0 8 5 - X
regimes, i.e., erosion dominated, erosioncorrosion dominated, and corrosion dominated, as the surface changed
from a ductile alloy to a brittle scale. However, the
effects of corrosion go beyond the scope of this paper and
will not be considered further.
The velocity exponent n is dependent on test conditions
and particle variables. Sundararajan and Shewmon w8x
tabulated results from a number of researchers and found
that, for a given researcher, the velocity exponent increased as particle size increased. As the particle size of
the glass erodent being used to erode Al alloy 6061-T6
was increased from 230 mm to 550 mm and 650 mm, the
value of n increased from 2.3 to 3.0 and 3.3, respectively.
Another researcher found that as the size of quartz erodent
being used to erode 11% Cr steel was increased from 25
mm to 60 mm and 200 mm, the n value increased from 2.0
to 2.1 and 2.3, respectively. The same researcher also
found that a titanium alloy and a nickel based superalloy
eroded with the same sized quartz erodent 150 mm. had
the same velocity exponent 2.3.. It was also found that
1095 steel with a hardness of 66 HRC, 63 HRC, and 30
HRC had an n value of 3.0, 3.0, and 2.9, respectively,
when eroded with 273 mm quartz, and that there was
minimal effect of hardness or target mechanical properties
on n. These results indicate that n is dependent on particle
variables, and not necessarily on target material variables.
However, the effect of different test apparatus on n has not
been established.
The objective of this paper is twofold: First, to develop
a test methodology that can precisely measure and control
particle velocity and minimize differences in impact conditions, and secondly to determine what factors control
andror influence n. Since erosion rate is dependent on a
power law relationship with velocity, it is critical that not
only V be measured precisely, but that n is well defined.
Once these two variables are determined for a particular
test set-up, subsequent studies can be conducted, such as
determining the effect of target material microstructure and
mechanical properties on erosion rate.
2. Experimental procedure
An Fe0.6%C binary alloy and 7030 brass were used
to study the effect of material variables on the velocity
exponent. Samples of the ironcarbon alloy were austenitized at 8508C for 45 min and water quenched. One sample
was sectioned and metallographically prepared to ensure
martensite had formed. Some samples were tested in this
conditions, while other water quenched samples were tempered at 6908C for 6 = 10 5 s, which resulted in a
spheroidized structure with an average carbide size of 1.85
mm. The tempered samples were encapsulated to prevent
decarburization. The 7030 brass samples were used in
two conditions: annealed and 70% cold worked. The ero-
511
512
Fig. 3. Plot of the particle velocity and data rate across the erodent
stream. The vertical lines indicate the distance across the width and the
diagonal of the erosion sample.
Fig. 4. Erosion weight loss vs. time plot for the WQ Fe0.6%C alloy.
Erosion conditions: velocity: variable; impact angle: 908; temperature:
258C; erodent: 355 mm Al 2 O 3 ; feed rate: 90 grmin.
513
Fig. 6. Erosion rate vs. particle velocity for the Fe0.6%C alloy in the
WQ and WQ&T conditions. Erosion conditions: impact angle: 908;
temperature: 258C; erodent: 355 mm Al 2 O 3 ; feed rate: 90 grmin.
WQ. and water quenched and tempered WQ & T. conditions. Fig. 4 shows the weight change versus time for the
WQ condition. An increase in slope in the weight loss vs.
time plot can be seen for increasing velocities. A similar
result can be seen in Fig. 5, which gives the weight change
results for the WQ & T material. The standard deviation in
the slope of the lines was found to be a maximum of 3%
for the 12 tests run, with most having a standard deviation
of less than 1.5%. The slopes were then plotted versus
velocity on a loglog scale to determine the effect of
velocity on the erosion rate See Fig. 6.. A power law
relationship is evident for both materials, and the exponential variable n is equal to 2.85 " 0.2 and 2.94 " 0.15 for
the WQ and WQ & T conditions, respectively. No significant difference in slope for the erosion ratervelocity equation logErosion Rate. s n logVelocity. was found between the two sample conditions. Recall from the literature
w25x that values between 2 and 3.5 are common. However, the erosion rates for the WQ & T samples were
always higher than the WQ samples, and this difference
Fig. 5. Erosion weight loss vs. time plot for the WQ&T Fe0.6%C alloy
with an average carbide size of 1.85 mm. Erosion conditions: velocity:
variable; impact angle: 908; temperature: 258C; erodent: 355 mm Al 2 O 3 ;
feed rate: 90 grmin.
Fig. 7. Erosion rate vs. particle velocity for WQ Fe0.6%C and brass
with 0% and 70% cold work. Erosion conditions: impact angle: 908;
temperature: 258C; erodent: 355 mm Al 2 O 3 ; feed rate: 90 grmin.
514
brass sample testing had been used previously for a different erosion test. While the particles are re-sieved after use
to account for particle fragmentation and change in size, it
does not address changes in particle shape. While this flaw
in experimental procedure was discovered after testing
erodent is no longer reused for erosion testing., it pointed
out that the n value is very much dependent on test
conditions. It has been shown that particle shape affects
erosion rate w6x, but its effect on n is unknown. If slight
changes in one test rig can produce changes in n of 0.2, it
is not hard to imagine what the effect of a completely
different test apparatus would be utilizing different erodents. This may begin to explain why a whole range of ns
have been found for similar materials w25,7x.
3.3. Erosion mechanism
The FeC samples were cross-sectioned and prepared
to reveal the erosion surface. Fig. 8 shows the surface
appearance for the WQ sample at 30 and 64 mrs. The WQ
samples are characterized by a martensitic structure, which
Fig. 10. Knoop microhardness profiles for the 0% and 70% cold worked
brass samples, tested at 20 and 50 mrs. Erosion conditions: impact angle:
908; temperature: 258C; erodent: 355 mm Al 2 O 3 ; feed rate: 90 grmin.
515
4. Discussion
Given the power law relationship of velocity on erosion
rate, velocity measurement must be precise and the test
apparatus must be capable of reproducing a test velocity
with little error. Ideally, velocity would be measured during each test, which an LDV system allows the operator to
do. Also, the system should be designed so the particles all
reach the intended velocity, regardless of size. This will
narrow the velocity distribution and make measurement
and testing more accurate. Finally, the erosion sample
should be located within the particle stream and confined
to the part of the stream with a nearly constant velocity.
This should eliminate errors from different particle velocities and those from different impact angles if a crater is
formed. Once these conditions are met, it is critical that the
erosion rate measurement be properly made by avoiding
the effects of incubation time. Several tests at different
times should be run to determine the end of incubation and
to measure the steady state erosion rate. This is especially
important when evaluating the effects of velocity, since
incubation time is related to particle velocity. It can be
seen from Fig. 5 that at a velocity of 20 mrs, incubation
time is approximately 30 min, whereas at 64 mrs, the
incubation time is nearly zero, so an erosion rate measurement that does not account for incubation will lead to
lower erosion rates at the lower velocities and an n value
that is incorrect.
As stated earlier, erosion rate is given by kV n , where n
can vary from 2 to 6.5. This is expected, since the particle
kinetic energy is 1r2 mV 2 . If the kinetic energy term is
separated out, the resulting equation is:
ER s kV 2qm. ,
where k is a variable that contains a target material
dependent term, V is the particle velocity, and m s n y 2.
is test conditionsrtest apparatus dependent. It has been
shown that m is dependent on the test conditionsrtest
apparatus in the comparison of the two WQ samples run at
different times, and by others w7x investigating the effect of
particle size. Since m is determined by the erosion test
apparatus and the erodent particle, m should be checked
periodically to ensure the test conditions have not changed,
especially if the erosion results tested at different times are
to be compared.
Once m or n. has been well defined, material comparisons can be made. This is a time consuming process, and
can be avoided if a well conceived test matrix is used in
which all samples are tested together. By testing every
sample in the test matrix for a given time interval, and
repeating this process until enough data points are gathered
to establish the erosion rate, it is known that all of the
samples are tested under the same erosion conditions. This
type of matrix was used for each set of experiments in this
study. While this is useful for single sets of experiments
516
5. Conclusions
Velocity is a critical test variable in erosion, as the
effects of different velocities easily overshadow changes in
the target material. A change in erosion rate of one and a
half orders of magnitude was achieved by changing the
velocity from 20 to 64 mrs. These tests clearly demonstrate the importance of particle velocity and the ability to
measure it during the erosion test, especially if one wishes
to study the microstructural effects of a material. The
significant findings of this work are as follows.
1. The velocity exponent n is related to test conditions,
and not target material properties, for the alloys tested.
When different materials brass vs. water quenched Fe
0.6%C alloy, WQ vs. WQ & T Fe0.6%C alloy. were
tested on the same date, the value of n was the same,
thereby being material independent. When the same material WQ Fe0.6%C alloy. was tested at different dates
under slightly different test conditions, the value of n
changed. This indicates that n is dependent on test conditions.
2. The velocity exponent n is not dependent on the
erosion mechanism for the materials tested. Results from
the testing of as-quenched martensite and quenched and
tempered martensite showed that the exponent is approximately 2.9 for both materials, even though as-quenched
martensite eroded by a brittle cracking mechanism, while
quenched and tempered martensite eroded by a plastic
deformation mechanism. However, the erosion rate is dependent on the erosion mechanism, as martensite was more
erosion resistant at each velocity as compared to the
tempered structure.
References
w1x G40-92, Annual Book of ASTM Standards, Vol. 3.02, ASTM
1992. 160.
w2x C.M. Preece, N.H. Macmillan, in: R.A. Huggins Ed.., Erosion,
Annual Review of Materials Science 7 1977. 95.
w3x T.H. Kosel, Solid Particle Erosion, ASM Handbook, Vol. 18, ASM
International, 1992.
w4x L.P. McCabe, G.A. Sargent, H. Conrad, Wear 105 1985. 257.
w5x T. Foley, A.V. Levy, Wear 91 1983. 45.
w6x M.M. Stack, F.H. Stott, G.C. Wood, Journal de Physique IV 3
1993. 687.
w7x S. Bahadur, R. Badruddin, Wear 138 1990. 189.
w8x G. Sundararajan, P.G. Shewmon, Wear 84 1983. 237.
w9x B. Lindsley, K. Stein, A.R. Marder, Measurement Science and
Technology 6 1995. 1169.
w10x B. Levin, J. DuPont, A.R. Marder, Wear 181 1995. 810.
w11x Y. Oka, I.M. Hutchings, Corrosion Engineering 39 1990. 677.