Professional Documents
Culture Documents
Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
I. FINNIE
University of California, Berkeley, California (U.S.A.)
(Received July 2, 1971)
SUMMAkY
The factors which may influence the erosion of ductile metals are listed. It is
shown that the effect of some of these variables may be predicted on simple fundamen-
tal grounds. On this basis, quantitative predictions may be made for the erosion of
ductile metals by hard abrasive grains which strike at grazing angles. Other aspects of
erosion which are not as well understood at the present time, are also discussed.
INTRODUCTION
1. Angle of impingement.
2. Particle rotation at impingement.
3. Particle velocity at impingement.
4. Particle size.
5. Surface properties.
6. Shape of the surface.
7. Stress level in the surface.
8. Particle shape and strength.
9. Particle concentration in the fluid stream.
10. Nature of the carrier gas and its temperature.
Not all of these factors are controllable or even easily measurable during a test but we
should at least attempt to estimate their relative importance.
Compared to brittle materials, it is relatively easy to see how one might start
analyzing ductile erosion because material removal should occur by a cutting or
displacing process as in metal cutting or grinding. This led the author’* to write the
equations of motion for a rigid abrasive particle striking the surface of a ductile metal.
The analysis will be outlined here with some corrections and emphasis upon the
physical assumptions involved.
We consider the two dimensional case shown in Fig. 1 with an idealized particle
of unit width although the treatment can be extended to a particle of arbitrary shape’.
The volume displaced by this idealized particle is approximately the integral of yt dx,
(where x,, yr are the coordinates of the particle tip) taken over the period in which
Fig. 1. Idealized picture of an abrasive particle striking a surface and removing material
cutting occurs. So, we have to determine the trajectory of the tip of the particle and
estimate when cutting stops.
For simplicity, the following assumption were made:
(1) The displaced volume is the volume removed by the idealized particle.
(2) The particle is rigid and doesn’t fracture.
(3) No initial rotation of the particles, i.e., q$,= 0. In a sense, this is an average
condition. If the initial angular velocity distribution were known, it could be incor-
porated into the analysis.
(4) Rotation of the particle is small during the cutting period. This can be
confirmed’ and for polyhedral particles, such as shown in Fig. 1, implies that :
xtz.%+r6, Yt”yG
(5) The conhguration of the particle and of the deformed material is assumed
to be geomet~cally similar during the cutting process. Thus we take the ratio of the
vertical force to the horizontal force on the particle as a constant K. Based on grinding
tests15 using abrasive grains and measurements made with single grains a reasonable
value is Kc= 2, i.e. a coefficient of friction = 0.5.
(6) Since large strains will be reached even at the beginning of the cutting
process, we would expect, based on metal cutting tests 16,that the plastic flow pressure
between particle and metal will be essentially constant. We denote the horizontal
component of this pressure by p.
(7) A final assumption, based on metal cutting observations, is that the area
over which the metal contacts the particle is about twice that given by the depth of cut.
That is, I N 2 y, in Fig. 1.
With the preceding assumptions the equations of motion may be written and
solved for x,, yr, As has been shown”i4 this leads to
c MU2
v=
where
V = volume removed from surface,
M = mass of eroding particles,
m = mass of an individual particle,
Z = moment of inertia of particle about its center of gravity,
r = average particle radius,
tt = angle of impact,
U = particle velocity,
p = horizontal component of flow pressure,
c = fraction of particles cutting in idealized manner,
a!; = horizontal velocity of tip of particle when cutting ceases.
However, the particle may leave the surface while the tip is still moving horizontally
and in this case we need to determine 2: for yt = 0. It may be shown that this is given by :
2u
fj=Ucosci - --sinol
P
where P = K + (1 + mr2/1)
Thus we obtain
c MU2
v= 2 [cos2a]
T----l
4p 1 + y--
cMU2 2
v=
mr2 F
a ; 2_:for Yt=O
4p 1+7
T-----l
The maximum volume removal occurs at tan 2a = P while the two expressions are
equal at the slightly higher angle given by tan a=P/2. Typically, i-mr2/3 and for
# N 2, P LZ:0.5 and so the m~mum erosion should occur at about a = 13”. Figure 2
shows experimental data as well as the predicted behavior. It is seen that the agreement
is excellent at low values of a but becomes poorer as CIapproaches 90’. This dis-
crepancy is to be expected for the idealized cutting mechanism on which the analysis is
based can hardly be applicable when the particle has no velocity component tangential
to the surface.
Mech~isms which might be invoked to explain the erosion of ductile metals
at angles near a = 90’ are :
(a) Once the surface is roughened, particles strike the surface locally at a
variety of angles and at grazing angles volume is removed.
(b) Particles follow the air flow and actually strike at grazing angles.
(c) The grains have initial rotation which leads to volume removal even at
a = 90’.
(d) The grains fracture on impact and by moving radially outward remove
surface material.
(e) Multiple impacts, battering the surface back and forth, eventually produce
fracture by low-cycle fatigue.
(r) The surface workhardens and eventually fails in a “brittle” manner.
Explanation (a} does not readily lend itself to analysis but would not appear
to be a major factor in producing erosion at angles near 90”. Explanation (b) can
easily be tested by calculating or observing the deflection of particles by a fluid stream
and in the present experiments it may be discounted. Explanation (c), as we will show,
does indeed lead to volume removal at large values of o(.However, it appears unlikely
that the particle angular velocities could be large enough to explain all of the observed
erosion at or near a = 90’. Explanation (d) has been studied by Tilly and Sage l2 and is
undoubtedly a major factor in influencing the form of the V-a curve when the
eroding particles are of a weak and friable nature. However, it appears unlikely that
particle fragmentation was a major factor in the tests shown in Fig. 2. Similar curves
0 I I I I I I I
0 30 60 90
ANGLE OF IMPINGEMENT , 0’
Fig. 2. Predicted and observed values for erosion of commercially pure aluminum, by 120 mesh Sic particles
at 500 ft./set. The scale off (a) is chosen arbitrarily so that its maximum value’coincides with the experi-
mental data. (From Sheldon and Finnie”).
were obtained when other ductile metals of varying hardness values were eroded by
silicon carbide grains. Explanation (e) has, to our knowledge, not been offered before.
However, it is clear that repeated impacts in adjacent areas will subject the surface to
alternating plastic strains which introduces the possibility of fracture by a low-cycle
fatigue mechanism. Explanation (l) has been given a number of times and usually it is
coupled with the assumption that a certain amount of energy is required to remove
surface material in perpendicular impact. That this explanation is, at best, a crude
approximation to actual behavior can be seen by eroding gold This material is
capable of absorbing enormous amounts of energy when it is beaten into thin sheets
of foil but shows the same type of volume removal Versus angle plot as other ductile
metals. In the present case it appears that factors (a) and (e) may be the main reason for
erosion at a = 90” with contributions also from (c) the initial rotation of the particles.
A number of authors have attempted to explain the discrepancy between the
solid and dashed lines in Fig. 2 by invoking two simultaneous mechanisms for volume
removal. For example, Bitter2 took the approach, which has been followed by a
number of other workers, of dividing erosion into “cutting wear” and “deformation
wear”. For “cutting wear” he took an expression similar to that derived here and in
earlier work14. A certain amount of the input energy is assigned to volume removal
by “deformation wear” which is taken as
V= E (U sin a-K)’
where ICand E are disposable constants. The mechanism by which this “deformation
wear” occurs is not at all clear but it is assumed that this is the mode by which volume
removal occurs in brittle solids. In fact, this type of analysis would appear to be
merely curve fitting. It is incapable of explaining the velocity or particle size dependence
of erosion rate in brittle solids and cannot predict the transition from a brittle to
Wear, 19 (1972) 81-90
86 I. FINNIE
ductile type of erosion that occurs in certain brittle solids when the grit size is very
small 13. In general, energy considerations appear to have been used very loosely in
the literature on erosion, and other types of wear. Of course, one can make use of the
specific energy (energy expended +- volume removed) to characterize, in an approxi-
mate manner, various types of wear or shaping processes. However, this approach
sheds little light on the physical processes involved in volume removal and is unlikely
to lead to precise predictions of the role of the various variables.
I.2
IO
4’
&
g 0.8
w OMEGA DISTRIBUTION
z
3 0.6
P
0
0 10 20 30 40 50 60 70 80 90
ANGLE OF IMPINGEMENT
Fig. 3. Iniluence of rotation on weight loss-angle relation. The assumed ~stribution for the dimensionless
rotation parameter a= &r/V is also shown.
Quite independently of whether the particles have initial rotation, we can see
in the equations for volume removal that it is important to allow for rotation in writing
the equations of motion. Typically, 1-S mr* so about $ of the initial kinetic energy
of translation is converted into rotational kinetic energy during impact.
Particle velocity
The equations predict T/cc U2 and this is observed experimentally to a first
approximation. However, more careful observation *J~*‘~ shows the relation to be
more nearly V4 in many cases. The reason or reasons for this discrepancy are not
clear. The explanations offered have been particle fragmentation at higher velocities’
and size effect*’ (to be discussed later in connection with particle size). Curiously for
brittle solids, with no scatter in strength, the predicted relation&pi3 is VK U2.4.
Particle size
One of the most intriguing aspects of erosion is that the volume removed by a
given mass of abrasive grains is independent of particle size for particles larger than
about 100 pm. For particles below this size, the erosion process becomes less and less
efficient as the particle size is decreased I7 . This is a fortunate aspect of erosion because
as particles become smaller it becomes more difficult to separate them from the fluid
stream.
The physical reasons for this size effect are still not clear. Some of the factors
which one might consider are :
(a) Fragmentation of larger particles leading to more efficient cutting than
with smaller particles.
(b) Grain size of the metal eroded.
(c) Oxide coating on the metal eroded.
(d) Change in the geometry of the cutting process as smaller particles are used.
(e) A true physical size-effect such that regions below a certain size show an
increase in strength values.
Explanation (a) has been offered by Tilly and Sage12. However, it can be
discounted by the observation that the same size effect is observed in abrasion tests
on soft metals at very low speeds and in this case preferential fracture of large grains
could hardly be a signi~cant process. Explanations (b) and (c) appear to be possible
reasons for the size effect but were dismissed after a series of erosion tests were carried
out on single crystals of copper and on pure gold, which has no oxide coating. Expla-
nation (d) also appears plausible but was dismissed after comparing the tests on
alum~ium shown in Fig. 4. The curves of weight loss versus angle for erosion by 127
pm and 9 pm particles are so similar (except for the scale factor of 4) that it appears
34
t
::
- ;
Le 2E m
2c1. I
B $
P
2 4 I I-
0
z
z z
fl? P
0 OW
0 30 60 90
PARTICLE APPROACH ANGLE , o”
Fig. 4. Weight removed (&g of abrasive particles) when commerciaIIy pure alu~num is eroded by Sic
particks of two sizes at 500 ft./set. (From SheIdon and Finnie”).
that the same process is occurring in both cases. We are left only with the explanation
(e) that there is a true physical size-effect. In this connection, it is interesting to note
that indentation tests made with tiny needles, and observed with the scanning electron
microscope show in some cases about a three-fold increase in hardness relative to
conventional tests’*.
IO 100 IO00
VICKERS HARDNESS, kg per sq mm, VHN
Fig. 5. Volume removed (mm3/g abrasive) as a function of Vickers Hardness when annealed metals are
eroded by 60 mesh SIC at o!= 20’ and U = 250 ft./set. (From Finnie, Wolak and Kabil”).
CONCLUSIONS
The model presented for ductile materials describes many features of the erosion
process. However, a number of aspects remain to be explained. Among these is the
ACKNOWLEDGEMENT
REFERENCES