Anda di halaman 1dari 4

Modelling of Abrasive Waterjet Machining: A New Approach

M. EITobgy, E-G Ng, M.A. Elbestawi(1)


Machining System Laboratory, McMaster Manufacturing Research Institute, McMaster University,
Dept. of Mechanical Engineering, Hamilton, ON, L8S 4L7, Canada

Abstract
Abrasive waterjet (AWJ) machining is one of the recent non-traditional methods starting to be used widely in
industry for material removal of different materials. The cutting performance of AWJ is achieved by a very
high speed, small-scale erosion process. In this paper, a modified form of Finnie's model for erosion is
developed for application to AWJ. This modified form is able to deal with curved surfaces rather than flat
surfaces only. Furthermore, the new modelling approach is capable of simulating multiple particle erosion.
This approach uses standard material properties and requires no calibration constants. The modelled results
agreed well with both experimental and analytical data.
Keywords;
Abrasive, Water, Simulation

1
INTRODUCTION
Abrasive waterjet (AWJ) machining is one of the recent
non-traditional methods that have been used widely in
industry for parting cuts on ductile materials such as
aluminium, brass, steel, titanium, and nickel based alloys
as well as brittle materials like glass, stone, and ceramics
[ I ] . In addition, AWJ machining can be used in a variety of
applications such as drilling, polishing, turning, and milling.
Abrasive waterjet cutting is a process that depends on
erosion as the main mechanism for material removal.
Material removal in the process is due to a small volume
removed with each particle impacting the surface at high
velocity. The numbers of particles impacting the surface
are in the order of tens of thousands per second.
The erosion of materials caused by the impact of hard
particles is one of several forms of material degradation
generally classified as wear. Erosion was defined by Bitter
[2] as "material damage caused by the attack of particles
entrained in a fluid system impacting the surface at high
speed", while Hutchings [3] defined erosion as "an
abrasive wear process in which the repeated impact of
small particles entrained in a moving fluid against a
surface results in the removal of material from that
surface". Solid particle erosion is a serious problem in gas
turbines, rocket nozzles, cyclone separators, valves,
pumps and boiler tubes. However, solid particle erosion
can be utilized in manufacturing processes such as
abrasive waterjet cutting.
In erosion, material removal occurs through the process of
micro-plastic deformation and/or brittle fracture. For ductile
materials such as pure metals and alloys, the impact with
the hard particles causes severe localized plastic strain on
the impact points of the surface. For brittle materials, such
as ceramic and intermetallic compounds, the force of
impacting particles causes localized cracking at the
surface. With subsequent impact events, these cracks
propagate and eventually link together, and as a result
material becomes detached from the surface [4].
Both analytical and empirical methods have been used to
model the AWJ machining process [5]. Researchers have
also employed finite element methods [6], fuzzy logic
algorithms [7] and reaction kinetics [8], to model this highly

complex manufacturing technique, which is fundamentally


based on an ultra high speed erosion process. A recent
approach based on using a coherent set of models was
presented by Hoogstrate et al. [9]. A review of the
published models for the AWJ cutting shows that all the
existing analytical
models require experimentally
determined constants, which are process dependent. Both
empirical models and heuristic approaches also
necessitate extensive experimentation, which is a major
drawback due to time expense and limited applicability.
A review of the basic erosion models can be found in [lo].
Based on the methodology employed, the published
erosion models use two different solving approaches: a) by
solving the equation of motion of single particles, or b)
based on an energy equation. Finnie [ I I ] and Hashish [I21
employed the former method whereas Bitter's model
employed the energy equation [2]. Currently, the existing
erosion models for ductile material do not account for the
simulation of multiple particIe erosion.
In this paper, a new approach for modelling the abrasive
waterjet machining process is presented. This new
approach is based on a modified version of the erosion
models, originally developed by Finnie [ I I ] and Hashish
[12]. This modification allows for modelling the multiple
particles erosion set-up. Following on from here, the
modified erosion model is employed to model the AWJ
machining process with surface regeneration capabilities,
where a new surface is generated after the erosion of each
single particle, and a different particle starts eroding the
regenerated surface. The numerical solution will proceed
until the particle either exits the cut or the particle velocity
changes its direction. With this approach, modelling the
AWJ machining process with thousands of particles is
feasible and computational time will be acceptable even
when using a Pentium based type processor.
2
MODIFIED EROSION MODEL
The modified erosion model is based on the particle
equation of motion and is also able to deal with curved
surfaces rather than flat ones only. Furthermore, this
modelling technique is capable of simulating the effects of
multiple particle impact. The model assumes that the

workpiece material is homogeneous, the material


behaviour at the micro scale level is similar to that found in
meso II scale, and is insensitive to strain rate and
temperature. An orthogonal cutting model was employed
to calculate the forces generated during the erosion
process. The depth of cut is dependent on the particle
velocity direction.
Figure 1 illustrates a particle eroding a curve surface with a
striking velocity v ( u ) . This particle is free to move in the
x,y coordinate and rotate in the x-y plane. Cutting forces
fv and fvnare generated when the particle impacts the
surface. fv and fvn are forces acting parallel and
perpendicular to the cutting direction respectively. By
applying the simplified form of orthogonal cutting force
model, fv and fvncan be calculated using Equations (1)
and (2) respectively, where K, is the specific cutting
resistance of the workpiece material, t is the depth of cut
perpendicular to the cutting velocity v , b is the width of
cut and C is a constant. Equation (1) is valid only when
fv is parallel to v as shown in Figure 1.

14 = K,btr[cos(Q)+ c sin(Q)]
k=x+r#

(9)

I = y+r#
(10)
The central difference algorithm shown in Equation (11)
was used to solve the single particle erosion problem.

MU = [f]
The numerical solution starts by initialising the
displacement, velocity and acceleration of the particle. The
velocity and direction components were computed using a
mathematical loop. A point on the existing surface that
satisfies the criterion that the angle generated by the tip of
the particle and the velocity direction shall be
approximately 90 degrees is found by incorporating a
special search algorithm. Once this criterion is met, the
depth of cut and the vector direction from the particle tip to
the point on the surface can be computed. Following on
from here, the predicted depth of cut and the direction of
the impacting particle are incorporated in Equations (1) to
(4) to calculate the cutting forces. In the next step, the
acceleration, velocity, and displacement of the particle are
modelled using Equations (12), (13) and (14) respectively.
The new position of the particle tip and the new surface
location are then calculated in a new iteration. This loop is
discontinued either when the particle exits the cut or the
particle velocity changes direction.

Figure 1: Particle eroding a curve surface


fv= K,bt

(11

fvn
= Cf,

The cutting forces f, and fvFcan be discrete into x and y


direction as shown in Equations (3) and (4) respectively,
where 8 can be determined with Equation (5).
(3)
Fy = Fv sin@)+ Fvn COS(Q)

(4)

Q = arctan(vx/vy)
Further, it is assumed that the stiffness effect is combined
with the cutting forces and the damping is equal to zero.
The equations of motion for the particle in the x and y
direction are shown in Equations (6) and (7) respectively,
while the,rotation motion is given in Equation (8), where x ,
y and # are the accelerations at the centre of gravity and
r is the radius of rotation. The locations k and I of the
particle tip can be calculated using Equations (9) and (10)
respectively. Unfortunately, Equations (6) to (10) cannot be
solved in a closed form, however numerical methods can
be used to obtain the solutions.
mx = K,bt[cos(Q)+csin(Q)]

(6)

my

(7)

= K,bt[sin(Q)+ccos(Q)]

Figure 2 shows the effect of the particle's attack angle on


the modelled surface generation. The initial surface before
impact was curved. Four different attack angles were
modelled. The modelled results show the surface shape
after the impact with a single particle at the four different
angles. When the attack angle was 15 deg, the erosion
depth in the Y-direction was lower than that generated at
the 45 deg attack angle. This was because with a lower
attack angle (15 deg), the particle momentum before
striking the surface in the y direction was lower than the
momentum in the x direction, therefore the penetration
depth was lower. This demonstrates the capability of the
modified erosion model to predict material removal for nonflat surfaces. However, no available data or models can be
found in academic literature to compare the results of the
model.

Figure 2: Effect of particle's attack angle on modelled


surface generation.

Figure 3 details two different modelling circumstances; in


the first one the particle is striking a perfectly flat surface
while in the other a sinusoidal surface is modelled. The
particle striking the sinusoidal surface shows a deeper
penetration depth when compared to the one striking the
flat surface. This was likely due to the fact that with the
initial sinusoidal surface, the initial Y-direction depth was
already lower than on the flat surface.

water and along the orifice wall. cd is calculated by


dividing the actual measured flow rate by theoretical flow
rate, which is usually in the range of 0.6 to 0.8 [13].

3.2
Numerical Simulation
The proposed method for simulating the AWJ process is
based on the modified erosion model and the numerical
solution of single particle erosion problem. The
incorporation of the modified erosion model allows for the
simulation of cumulative effect of multiple particles. In
order to achieve the simulation results in a reasonable time
frame, the computational code has to minimise
computational time. Currently, a simulation run with 10,000
particles imparting on the surface requires 60 minutes to
complete when using a Pentium 4, 1.8 GHz clock speed
processor. The assumptions made in this simulation are:
a) size and velocities of the particle are assumed to be
constant and b) the directions of the particles leaving the
nozzle are assumed to be inline with the nozzle. Figure 4
shows the results of the advancement of the cutting front,
using the modified model.

Figure 3: Effect of particle striking a flat surface and a


sinusoidal surface on modelled surface generation.
3
MODELLING OF AWJ MACHINING
Fundamentally, modelling of the AWJ process is based on
a surface generation technique where a new surface is
generated after the erosion of each single particle and a
different particle starts eroding the regenerated surface in
a continuous process. With this concept, the simulation of
thousands of particles is feasible in an acceptable time
frame, using average computational resources. The model
depends only on the workpiece material properties and
does not require any experimentally determined, or
calibration constants.

3.1
Jet Generation
In order to model the jet generation process, v , the
abrasive particle velocity, has to be a function oywater
pressure P , abrasive mass flow rate mA , and water mass
flow rate m,, as detailed in Equation (15). In addition,
other variables such as nozzle geometry and abrasive
particle size need to be taken into account. The constant
qt is a function of several process parameters that include
pump pressure, abrasive particle characterization, nozzle
orifice, and focus geometry. Values of qt range between
0.73 and 0.94 [13].

m,
By employing Bernouli's law, vo can be determined from
Equation (16), where coefficients c, and y are obtained
experimentally by measuring the actual velocity of the high
pressure water and compared with those acquired
theoretically. Typical values for c, and y range between
0.83 and 0.93 [13].
vo = c v Y m

The actual volumetric flow rate of the water, Qa can be


determined using Equation (17), where cd is a nondimensional number which considers the reduction in the
water mass flow rate, m,, due to the sudden change in
the fluid mechanic conditions on the orifice outlet as well
as the reduction in jet velocity due to friction between

Figure 4: Advancement of the cutting front using the


modified model.
4
RESULTS AND DISCUSSION
Figure 5 and 6 compare the effects of water delivery
pressure on penetration depth using data acquired from
Hashish's analytical model [12], experimental results [ I 31,
and the newly developed modelling method, at nozzle
velocities of 7.6 mm/s and 5.1 mm/s respectively. The
workpiece material modelled was AlSl 1035 mild steel.
The K, value employed was 50 Joules per cubic mm,
which is similar to that used for grinding process [14]. This
K, value is 30 to 40 times higher when compared to metal
cutting. Hashish first introduced his model for AWJ cutting,
based on Finnie's erosion model, in 1984 [15]. He later
modified the erosion model to simulate the AWJ cutting
process [12]. The penetration depth can be determined
using Hashish's model, as detailed in Equation (18).

h
-= 0.282C

di

The process parameters employed by the experimental


analysis, Hashish's model and the newly developed model
are listed in Table 1. When employing Hashish's model,
the constants E and c are acquired by assuming that C,
and v, are equal to zero (as suggested by Hashish).
When the nozzle transverse speed modelled was
7.6 mm/s, both Hashish's and the newly developed model
showed that with larger water delivery pressure, the
penetration depth increases, which agreed well with the
experimental results, as shown in Figure 5. However, as

however it is modified to enable the simulation of erosion


generated by multiple particle impact. The model also has
the capability to predict the volumetric material removal
rates and new surface generation resulting from multiple
particle impact on curved surfaces. Moreover, the newly
developed erosion model is successfully employed to
model the highly complicated AWJ machining process,
taking into consideration the flow characteristics of the
abrasive particles. The main benefit of employing this
approach is that no process specific experimental
constants are required, thus the model is independent of
operating parameters, although it is dependent on
workpiece material mechanical properties. The new model
shows similar trends as those predicted by Hashish's
model [12, 141, and agrees even better with the
experimental results.
AC KN0WLE DGM ENTS
The authors would like to take this opportunity to thank
Materials and Manufacturing Ontario (MMO), NCEAUTO21, and McMaster Manufacturing Research Institute
for their financial support.

Figure 6: Effect of water delivery pressure on penetration


depth at a 5.1 mm/s nozzle transverse speed.
u (mm/s)
7.6
5.1

172
0.75
1.74

207
0.83
2.32

P (MPa)
24 1
1.11
3.00

276
1.18
3.79

310
1.65
4.18

Table 1: Process parameters employed to compare


experimental and modelled results.
shown in Figure6, when the transverse speed was
reduced to 5.1 mm/s, the penetration depth predicted by
the newly developed model agreed with experimental
results more closely than Hashish's model. Recalculation
of the calibration constants employed in Hashish's model
will improve its accuracy, however, extensive experimental
work needs to be carried out to obtain the calibration
constants. This is one of the main limitations when
employing such modelling methods, while with the newly
developed method, no calibration constant is required with
a change in nozzle transverse speed.
5
CONCLUSIONS
A new approach, for modeling the AWJ machining process
is presented. This technique is fundamentally based on
Finnie's model of erosion adopted by many researchers,

REFERENCES
Siores E., Wong W.C.K., Chen L., Wager J.G., 1996,
Enhancing Abrasive Waterjet Cutting of Ceramics by
Head Oscillation Techniques, Annals of the CIRP,
45/1: 327-300.
Bitter J.G.A., 1963, A Study of Erosion Phenomena,
Part 1, Wear, 6:5-21.
Hutchings I.M., Winter R.E., 1974, Particle Erosion of
Ductile Metals: A Mechanism of Material Removal,
Wear, 27/1: 121-128.
Alman D.E., Tylczak J.H., Hebsur M.G., 1999, Solid
Particle Erosion Behaviour of a Si3N4-MoSi2,
Materials Science and Engineering, A261: 245-251.
Venkatesh V.C., 1984, Parametric Studies on
Abrasive Jet Machining, Annals of the CIRP,
33/1: 109-1 12.
Hassan A.I., Kosmol J., 2001, Dynamic ElasticPlastic Analysis of 3D Deformation in Abrasive
Wat e rjet Machining , Jo urna I of MateriaIs Processing
Technology, 113:337-341.
Kovacevic R., Fang M., 1994, Modelling the Influence
of the Abrasive Waterjet Cutting Parameters on the
Depth of Cut Based on Fuzzy Rules, International
Journal Machine Tools and Manufacture, 34/1:55-72.
Momber A.W., 1995, A Generalized Abrasive Water
Jet Cutting Model, Proceedings of the 8th American
Water Jet Conference, 8:359-371
Hoogstrate A.M., Karpuschewski B., van Luttervelt
C.A., Kals H.J.J, 2002, Modelling of High Velocity,
Loose Abrasive Machining Processes, Annals of the
CIRP, 51/1:263-266.
Kovacevic R, Hashish M., Mohan R., Ramulu M, Kim
T.J., Geskin E.S., 1997, State of the Art of Research
and Development in Abrasive Waterjet Machining,
ASM E Transact ions, Journal of Manufacturing
Science and Engineering, 119/4B: 776-785.
Finnie I., 1960, Erosion of Surfaces by Solid
Particles, Wear, 3/2:87-130.
Hashish M., 1989, A Model for Abrasive Water Jet
Journal
of
Mach ining, ASM E Transactions,
54-162.
Engineering Materials and Technology, 111:I
Hashish M., 1993, Pressure Effects in Abrasive
Water Jet (AWJ) Machining, Journal of Engineering
Materials and Technology, 111:221-228.
Oberg E., Jones F.D., Horton H.L. and Ryffel H.H.,
2000, Machinery's Handbook, 26th Edition, Industrial
Press, New York
Hashish M., 1984, A Modelling Study of Metal Cutting
with Abrasive Water Jet, ASME Transactions, Journal
of Enaineerina
106:88.
u
u Materials and Technoloav.
u,
I

Anda mungkin juga menyukai