Wenjun 2010

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Int J Adv Manuf Technol (2011) 53:247–253

DOI 10.1007/s00170-010-2836-7

ORIGINAL ARTICLE

Numerical simulation for abrasive water jet machining based


on ALE algorithm
Gong Wenjun & Wang Jianming & Gao Na

Received: 22 January 2010 / Accepted: 6 July 2010 / Published online: 21 July 2010
# Springer-Verlag London Limited 2010

Abstract In dealing with fluid impact and large deformation 1 Introduction


problems by traditional Lagrange grid, calculation failure
often happens due to grid distortion. An abrasive water jet The utilization of abrasive water jet (AWJ) machining is
machining model is created to simulate the whole stage by growing. Comparing with most traditional machining
software LS-DYNA from the jet into the nozzle to the technologies, AWJ cutting technology has a number of
workpiece material removal process using ALE (Arbitrary distinct advantages, such as no thermal distortion, high
Lagrange–Euler) algorithm. The mesh for the abrasive and flexibility, good surface quality, and low machining force
water is based on the ALE formulation, while the target mesh etc. [1–3]. However, the mechanics of AWJ cutting is
applies the Lagrange formulation. The effect of jet penetration complex, and the material removal process is not fully
is implemented by coupling the grids of ALE and Lagrange. understood yet. The choices of the process parameters and
The jet traverse speed is achieved by definition of the its arrangements have a great effect on the cutting results,
movement of ALE grid to reduce the mesh domain. The and it is difficult to describe this machining processing
abrasive constitutive equations are also presented in this paper. using precise mathematical model [4, 5].
The uniform mixture for abrasive and water is achieved by The abrasive water jet researches focus mostly on the
definition of volume percentage of the two materials in the experimental fields or theories [6–9], but simulation
initial ALE elements. Simulation results give the relationships researches are relatively few and mostly limited to discuss
between processing parameters and the cutting depth. The the flow fields in nozzle with Euler algorithm [10–12]. As
good agreement between simulation results and experimental the simulation studies of workpiece removal process, most
data verifies the correctness of the simulation. of existing researches deal with pure water jets machining
or single abrasive particle impacting on the workpiece [13,
Keywords Abrasive water jet . Abrasive constitutive 14]. The pure water simulations use Euler algorithm; the
model . ALE . Fluid–solid coupling abrasive particle and the traverse speed are not considered
[15]. Single abrasive particle model uses Lagrange grid,
which ignores the impact role of water completely [16].
In this paper, the ALE algorithm is applied to build a
numerical model to simulate a whole machining process of
abrasive water jet by software LS-DYNA, which uses
G. Wenjun (*) : W. Jianming : G. Na
School of Mechanical Engineering, Shandong University,
Lagrange grid to modeling workpiece material and uses
Jinan 250061, China ALE grid to model water, abrasive, and air medium
e-mail: gwjagassi@yahoo.com.cn materials. Fluid–solid coupling is calculated by coupling
W. Jianming the ALE and the Lagrange grids in order to simulate the
e-mail: wangjianming@sdu.edu.cn penetration of abrasive water jet. The Euler grid is given
G. Na certain movement to define water jet traverse so as to
e-mail: gaoxiaodou0116@hotmail.com.cn reduce the initial Euler mesh domain.
248 Int J Adv Manuf Technol (2011) 53:247–253

The good agreement between simulation results and Stress tensor σij can be expressed as the following
experimental data verifies the correctness of the model and equation:
simulation results. It provides a powerful tool to the

simulation research of abrasive water jet machining and s ij ¼ pdij þ m vi;j þ vj;i ð4Þ
its process parameter optimization.
These equations can be solved by the following
boundary conditions and initial conditions:
2 ALE algorithm and fluid governing equations
vi ¼ Ui0 ; in the domain Γ 1 ð5Þ
2.1 ALE algorithm description

The Lagrange grids are fixed on the material and move with s ij nj ¼ 0; in the domain Γ 2 ð6Þ
it. The Euler grids are immovable in the space, but in the
description of ALE algorithm, a reference coordinate where, Γ 1 [ Γ 2 ¼ Γ ; Γ 1 \ Γ 2 ¼ 0, Γ is the complete
system is introduced, which is movable but asynchronous boundary of the calculate domain, and Γ1 and Γ2 are essential
with the material. and natural boundary conditions relatively. Superscript
indicates the initial specified value of this parameter, and ni
@f ðXi ; t Þ @f ðXi ; t Þ @f ðxi ; t Þ is the unit vector of outside normal of domain surfaces. δij is
¼ þ wi ð1Þ
@t @t @xi the Kronecker function.
where 3. Energy conservation equation
Xi The Lagrange material point coordinates @E @E
xi The Euler space point coordinates r ¼ s ij vi;j þ rbi vi  rwj ð7Þ
@t @xj
wi Material movement velocity relative to the reference
coordinate system The items in the above formula containing wj are
When using Lagrange grid to model, the movement of the collectively referred to as convection items, which need to
reference coordinate system and material point is synchronous calculate the material's transport through the grid, and it
without calculating the transport of material through the grid, does not appear in the Lagrange algorithm. Therefore,
that is wi is zero. So, its calculation amount is relatively small calculating ALE equations is more difficult than Lagrange
and suitable for small deformation analysis of solid. equations.
When using Euler grid to model, the coordinate system
is fixed in space. It is a special case of the ALE algorithm.
It needs to calculate the transport of material through the 3 Material models
grid, and its calculation amount is larger than Lagrange
grid. So, it is suitable for large deformation analysis of fluid In this paper, the pressures of fluid materials including
dynamics. water and abrasive are calculated by their state equations.
The purpose of using ALE algorithm to model is that the Generally, state equations are semi-theory and semi-
jet traverse speed can be defined into the movement of ALE empirical formula, and their parameters are determined by
grid, so that the mesh domain will be reduced relative to experiments.
Euler algorithm.
3.1 Material model for water
2.2 Fluid governing equations in ALE algorithm
During the process of the water jet impact on the
1. Mass conservation equation workpiece, water is in a compressed state, and the change
of density generates pressure. Chisum and Shin provided an
@r @vi @r approach to transfer Gruneisen state equation into a
¼ p  wi ð2Þ polynomial equation [17]. In the compressed state,
@t @xi @xi

2. Momentum conservation equation P ¼ a1 m þ a2 m 2 þ a 3 m 3 þ b0 þ b 1 m þ b 1 m 2 E ð8Þ

@vi @xi where a1 =2.19 Gpa, a2 =9.224 Gpa, a3 =8.767 Gpa, b0 =


n ¼ s ij;j þ rbi  rwi ð3Þ
@t @xj 0.4934, and b1 =1.3937.
Int J Adv Manuf Technol (2011) 53:247–253 249

Table 1 Parameters for numeral simulation

p Water pressure 150–350 Mpa


d1 Diameter of entrance 1 mm
d2 Diameter of mix exit 0.3 mm
ma Abrasive mass flow 0.45 kg/min
v Traverse speed 140–380 mm/min
α Nozzle taper 13 mm
a Length of workpiece 4 mm
b Width of workpiece 4 mm
c Height of workpiece 10 mm
L Total length of nozzle 11 mm
L1 Stand off 3 mm
L2 Length of air domain 13 mm
Fig. 2 Grids coupling for Euler and Lagrange

3.2 Material model for abrasive where


ρa The density of solid abrasive
The P vs. ρ relationship is defined as
ρair The density of air
P ¼ C02 ðr  rmix Þ ð9Þ Vair The volume of the air included in the initial
incompact abrasive
where V The total volume of the initial incompact abrasive
C0 Speed of sound of solid abrasive
ρmix The initial incompact abrasive density which Defining the change rate of density μ by the following
includes some air formula,
ρ The current abrasive density which includes some air r
m¼ 1 ð12Þ
and rmix

rmix ¼ ð1  a0 Þra þ a0 rair ð10Þ Substituting Eq. 12 into Eq. 9, we can gain the final linear
polynomial equation of state for the abrasive:
Vair
a0 ¼ ð11Þ P ¼ C02 rmix m ð13Þ
V
The value of C0 is 9.03 km/s obtained from [18].

3.3 Material model for workpiece

Workpiece applies C-17 steel, its density is 7,800 kg/m3,


elastic modulus is 200 Gpa, Poisson's ratio is 0.24, yield
Fig. 3 Grids coupling profile

Fig. 1 Numerical simulation model of abrasive water jet


250 Int J Adv Manuf Technol (2011) 53:247–253

Fig. 4 Different phases of


AWJ machining. a Jet initial
inflow. b Jet flow taper pipe.
c Jet filling nozzle. d Jet
machining

(a) Jet initial inflow (b) jet flow taper pipe

(c) Jet filling nozzle (d) jet machining

strength is 235 Mpa, and its material model uses the implemented by coupling the grids of ALE and Lagrange,
movement hardening model. so C and D grids need to overlap in this simulation, as
Its hardening tangential modulus is 350 Mpa, its failure shown in Figs. 2 and 3.
is determined by its material failure strain, and the value is As a result of penetration of the high pressure, large
0.3 in this paper. deformation only occurs near the contact area of jet and
workpiece. Therefore, it is necessary to apply more elements
in the central area of domain D to improve the calculation
4 Modeling for numerical simulation

A simulation model for a whole machining of the abrasive


water jet is established using LS-DYNA software. The
simulation model parameters (see Table 1) come from
existing experimental data [19] to compare both results
conveniently.

4.1 Modeling description

The whole model is divided into four domains, A, B, C,


and D, as shown in Fig. 1. Domain A is the jet source
including abrasive and water; domain B is the nozzle part,
which is the empty material domain at the beginning, and
abrasive and water will flow through this domain during the
simulation; C represents the air domain to define air
medium; and D is the workpiece domain. A, B, and C
domains are meshed by ALE grids. D domain is meshed by Fig. 5 The relationship between pressure and cutting depth (v=
Lagrange grids. The impact of jet to workpiece is 220 mm/min)
Int J Adv Manuf Technol (2011) 53:247–253 251

accuracy. Since an internal shock wave will occur and


transfer to the material during penetration, this will cause
discontinuities in pressure, density, energy, and acceleration.
Hence, the artificial viscosity is necessary to be introduced
into the calculation to reduce the numerical oscillation.

4.2 Traverse speed

Giving a speed to ALE grid domains (i.e., domains A, B,


and C) to define the traverse speed v can avoid using much
more Euler grids to improve the computational efficiency.

4.3 Fluid–solid coupling

The Euler grids in domain C will couple with the Lagrange


grids of domain D to achieve the coupling of structure and
Fig. 6 The relationship between traverse and cutting depth (p= fluid by means of the keyword *CONSTRAINED_
200 Mpa)
LAGRANGE_IN_SOLID in LS-DYNA [20], in which
workpiece and fluid material are defined as master material
and slave material relatively, as shown in Figs. 2 and 3.
During the numerical simulation, the program will auto-
matically determine whether elements in the workpiece fail or
Fig. 7 Process of workpiece
erosion

t=40us t=120us

t=180us t=240us
252 Int J Adv Manuf Technol (2011) 53:247–253
Fig. 8 Workpiece profile in
different times

t=40us t=120us

t=180us t=240us

not in accordance with its failure strain and delete those failure 5 Analysis of numerical results
elements. When associated elements around a node fail, the
node will be deleted from the calculation model. An abrasive water jet machining model is created to
simulate the whole processing. Figure 4 indicates the flow
4.4 The initial conditions of jet resource and mixing ratio status of abrasive water jet in the nozzle at different times,
and the abrasive water jet forming process is reflected by
Jet resource A is used to form continuous inflow of water and the changes of material density.
abrasive. The water pressure is applied to the top surface of the Some factors (or parameters) which influence on jet
jet source, and its elements are defined to pressure inflow type machining performances are studied by this simulation.
using the keyword *SECTION_SOLID_ALE, which reduce the
initial domains of the water jet. Therefore, the simulation time
is not controlled by the initial domains of the water jet [20].
The mixing ratio of abrasive and water is achieved by
the keyword *INITIAL_VOLUME_FRACTION, which
could define the respective percentage of abrasive and
water volume fraction in the initial Euler grid, and their
volume fractions are based on the mass flow of abrasive
and water. Thus, it breaks through the limitation for one
element can only contain one kind of material.

4.5 Boundary condition

The bottom of workpiece domain D is fixed displacement


boundary, and its four sides apply non-reflect boundary in
order to simulate unlimited boundary and reduce the initial
material mesh domain.
The sidewall of nozzle domain B is zero velocity
boundary so that it can meet no slip conditions. Fig. 9 Fitting curve for cutting depth and width
Int J Adv Manuf Technol (2011) 53:247–253 253

Traverse speed, cutting depth, pressure, and water cushion The good agreement between simulation results and
effect will be discussed in the following section. experimental data verifies the correctness of the simulation.
The influence of pressure on the cutting depth is shown
in Fig. 5. In order to compare the simulation results with Acknowledgements This paper is financially supported by the Nature
Science Foundation of Shandong province in China (number Y2007A07).
experimental data, both results are drawn on the same
figure. From it, we can see that under certain traverse
speed, the cutting depth through workpiece has a linear
relationship with pressure. References
When the water pressure increases from 150 to 350 Mpa,
cutting depths change from 0.15 to 9.55 mm at the traverse 1. Van Luttervelt CA (1989) On the selection of manufacturing
speed of 220 mm/min, and the simulation results agree well methods illustrated by an overview of separation techniques for
sheet materials. CIRP Ann Manuf Technol 38(2):587–607
with the experiment data. 2. Bostjan J, Daniel B, Mihael J (2004) Monitoring of abrasive water
When target distance is a constant, the higher the water jet (AWJ) cutting using sound detection. Int J Adv Manuf Technol
pressure is, the greater the abrasive kinetic energy will be. 24:733–737
So, the ability of jet cutting will increase to cause deeper 3. Ramulu HM, Kunaporn S, Arola D, Hashish M (2000) Water jet
machining and peening of metals. Press Vess Technol 122(1):90–95
cutting depth. 4. Parikh RJ, Lam S (2009) Parameter estimation for abrasive water
The cutting depth vs. traverse speed has an exponential jet machining process using neural networks. Int J Adv Manuf
decay relationship as shown in Fig. 6. In this figure, the Technol 40:497–502
traverse speed increases from 140 to 380 mm/min, while 5. Kovacevic R, Fang M (1994) Modeling the influence of the
abrasive water jet cutting parameters on the depth of cut based on
cutting depth has a sharp decline from 7.55 to 1.85 mm at fuzzy rules. Int J Mach Tools Manuf 34(1):55–72
the water pressure of 200 Mpa. Simulation results also 6. Momber AW (1995) A generalized abrasive water jet cutting
agree well with the experimental data. model. Proceedings of the 8th American Water Jet Conference,
The slower traverse speed is, the higher the energy Vol. 8, Houston, pp 359–371
7. Parikh PJ, Lam SS (2009) Parameter estimation for abrasive water
concentrates and deeper the cutting depth will be. With the jet machining process using neural networks. Int J Adv Manuf
speed increases, the largest cutting depth reduces significantly, Technol 40:497–502
and the kerf surface quality will also be worse rapidly. 8. Bitter JGA (1993) A study of erosion phenomena: part I. Wear
Figures 7 and 8 show workpiece erosion and its profile 6:5–21
9. Wang J (1999) Abrasive water jet machining of polymer matrix
during the simulation, and Fig. 9 shows the fitting curves composites cutting performance, erosive process and predictive
for cutting depth and width. Under certain water pressure models. Int J Adv Manuf Technol 15:757–768
and traverse speed, the maximum cutting depth in the 10. Johnson GR, Bsissel SR (1996) Normalized smoothing functions for
workpiece does not increase over time. When the compu- SPH impact computations. Int J Numer Math Eng 39:2725–2741
11. Johnson GR, Stryk RA, Beissel SR (1996) SPH for high velocity
tation time reaches a certain value, the cutting depth tends impact computations. Comput Methods Appl Mech Eng 139:347–373
to be stable. This is because the penetration takes place in a 12. Liu H, Wang J, Kelson N, Brown R (2004) A study of abrasive
very short time, it is easier to gather water in the bottom of water jet characteristics by CFD simulation. J Mater Process
the groove, and that water would play like a water cushion Technol 153:488–493
13. Li M, Rong HB, Yi MG (2008) Water jet penetration simulation
and weaken its erosion capacity. by hybrid code of SPH and FEA. Int J Impact Eng 35(9):1035–
1042
14. Eltobgy MS, Elbestawi MA (2005) Finite element modeling of
6 Conclusions erosive wear. Int J Mach Tools Manuf 45:1337–1346
15. Kyriaki M, Thomas K (2007) A finite element based model for
pure water jet process simulation. Int J Adv Manuf Technol
An abrasive water jet machining model and abrasive 31:933–940
constitutive model are created to simulate the whole stage 16. Paul S, Hoogstrate AM, Van L, Kals HJJ (1998) Analytical and
from the jet into the nozzle to removal of the workpiece experimental modeling of abrasive water jet cutting of ductile
materials. J Mater Process Technol 73:189–199
material in this paper. According to the simulation results, 17. Liu GR, Liu MB (2003) Smoothed particle hydrodynamics: a
we obtain the following conclusions: meshfree particle method. World Scientific Publishing Co.,
Singapore
1. Cutting depth has a linear relationship with the jet pressure. 18. Gwanmesia GD, Zhang J, Darling K, Kung J, Li BS, Wang LP,
2. Cutting depth has an exponential decay about traverse Neunille D, Liebermann RC (2006) Elasticity of polycrystalline
pyrope (Mg3Al2Si3O12) to 9 GPa and 1000°C. Phys Earth Planet
speed increasing.
Inter 155(3-4):179–190
3. Due to the water cushion effect, when the computation time 19. Hassan SI, Chen C, Kovacevic R (2004) On-line monitoring of depth
reaches a certain value, the cutting depth tends to be stable. of cut in AWJ cutting. Int J Mach Tools Manuf 44(6):595–605

You might also like