Sunteți pe pagina 1din 7

Engineering with Computers (2006) 21: 289–295

DOI 10.1007/s00366-006-0018-x

O R I GI N A L A R T IC L E

Geng Tie Æ Li Dequn Æ Zhou Huamin

Three-dimensional finite element method for the filling simulation


of injection molding

Received: 28 December 2004 / Accepted: 26 January 2006 / Published online: 20 May 2006
 Springer-Verlag London Limited 2006

Abstract With the development of molding techniques, thermal with the position of the moving flow front [1–3].
molded parts have more complex and larger geometry Because of these inherent factors, it is difficult to analyze
with nonuniform thickness. In this case, the velocity and the filling process. Therefore, simplifications are usually
the variation of parameters in the gapwise direction are used. For example, in traditional middle-plane model
considerable and cannot be neglected. A three-dimen- and dual-domain model [4, 5], the Hele–Shaw approxi-
sional (3D) simulation model can predict the filling mation [6] is used. So both of these models are 2.5D
process more accurately than a 2.5D model based on the models. In 2.5D model, the velocity and the variation of
Hele–Shaw approximation. This paper gives a mathe- pressure in the gapwise direction are neglected except
matical model and numeric method based on 3D model that the temperature is solved by FDM, and the filling of
to perform more accurate simulations of a fully flow. a mold cavity becomes a 2D problem in flow direction
The model employs an equal-order velocity–pressure and a 1D problem in gapwise direction. As most of the
interpolation method. The relation between velocity and injection molded parts have a sheet-like geometry in
pressure is obtained from the discretized momentum which the thickness is much smaller than the other
equations in order to derive the pressure equation. A 3D dimensions of the part, these models have been generally
control volume scheme is used to track the flow front. successful in predicting the advancement of melt fronts,
During calculating the temperature field, the influence of pressure fields, and temperature distribution.
convection items in three directions is considered. The The interest in 3D simulation of injection molding has
software based on this 3D model can calculate the increased tremendously and some progress has been made
pressure field, velocity field and temperature field in [7–9] in the past few years. One reason is the processing of
filling process. The validity of the model has been tested large and complex parts. With the development of
through the analysis of the flow in cavities. molding techniques, more and more molded parts have
thick or nonuniform thickness, such as those encountered
Keywords 3D Æ Equal-order interpolation Æ in gas-assisted injection molding. In these cases, the
Injection molding Æ Simulation velocity and the changes of parameters in the gapwise
direction are considerable and cannot be neglected. On
the other hand, the requirements on the performance of
1 Introduction injection molded items have been ever increasing. Several
situations occurring during mold filling which cannot be
During injection molding, the rheological response of accurately predicted using the Hele–Shaw approximation
polymer melts is generally non-Newtonian and noniso- need to taken into account nowadays, such as the fluid
behavior at the free surface (flow front), the fluid behavior
near and at the solid walls, the phenomenon occurring at
G. Tie Æ L. Dequn Æ Z. Huamin merging of two or more fluid streams (weldlines), and the
State Key Laboratory of Mold & Die Technology,
Huazhong University of Science and Technology, kinematics in areas where shear and extensional defor-
Wuhan, Hubei 430074, People’s Republic of China mations contribute significantly to the stress field (gates,
ribs, etc.). A 3D simulation model should be able to gen-
G. Tie (&) erate complementary and more detailed information re-
Machine and Electric Engineering College,
Henan University of Technology, lated to the flow characteristics and stress distributions in
450052 Zhengzhou, Henan, People’s Republic of China molded parts. This will be particularly important when
E-mail: gengtie2000@sina.com dealing with multicomponent mold filling and with
Tel.: +86-0371-67758626 molding of fiber-reinforced systems.
Fax: +86-372-3932808
290

This paper presents a 3D finite element model to deal g0 ðT ; P Þ


with the 3D flow of injection molding. In this model, the g¼ . ;
velocity in the gapwise direction is not neglected and the 1 þ ðg0 c_ s Þ1n
pressure also varies in this direction. An equal-order
velocity–pressure formulation method [10–12] is em- where n, c_ ; s* are non-Newtonian index, shear rate and
ployed, and the relation between velocity and pressure is material constant, respectively.
obtained from the discretized momentum equations. A Because there is no notable change in the scope of
3D control volume scheme is introduced to track the flow melt temperature during filling, Arrhenius model [13] for
front. During calculating temperature field, the influence g0 is employed as following:
 
of convection items in three directions is considered in Tb
order to get more exact results and to apply to the wider g0 ðT ; P Þ ¼ B exp expðbP Þ;
T
range of parts. Finally, the validity of the model has been
where B, Tb, b are material constants.
tested through the analysis of some cases.

3 Finite element calculations for the pressure field


2 Governing equations
3.1 Velocity–pressure relation
The pressure of melt is not very high during filling the
cavity, so the melt is considered incompressible. Inertia and In a 3D model, since the change of the physical quan-
gravitation are neglected as compared to the viscous force. tities are not neglected in the gapwise direction, the
With the above approximation, the governing equa- momentum equations are much more complex than
tions, expressed in Cartesian coordinates, are as fol- those in a 2.5D model. It is impossible to obtain the
lowing: Momentum equations: velocity–pressure relation by integrating the momentum
        equations in the gapwise direction, which is done in a
@ @u @ @v @u @ @w @u
2g þ g þ þ g þ 2.5D model. The momentum equations must be first
@x @x @y @x @y @z @x @z
discretized, and then the relation between velocity and
@ðP Þ pressure is derived from it. In this paper, the momentum
 ¼0
  @x       equations are discretized using Galerkin’s method with
@ @v @u @ @v @ @w @v bilinear velocity–pressure formulation. The element
g þ þ 2g þ g þ equations are assembled in the conventional manner to
@x @x @y @y @y @z @y @z
: form the discretized global momentum equations and
@ðP Þ
 ¼0 the velocity may be expressed as following:
@y
        @P @P @P
@ @w @u @ @v @w @ @w ui ¼ ~ui  Kiu vi ¼ ~vi  Kiv ~ i  Kiw
wi ¼ w ; ð4Þ
g þ þ g þ þ 2g @x @y @z
@x @x @z @y @z @y @z @z
@ðP Þ where
 ¼0 !
@z X
ð1Þ ~ui ¼  Axij uj  Bxij vj  Cijx wj =Axii ;
i6¼j
Continuity equation: !
X
@u @v @w ~vi ¼  Byij vj  Ayij uj  Cijy wj =Byii ;
þ þ ¼ 0: ð2Þ i6¼j
@x @y @z !
X
Energy equation: ~i ¼
w  Cijz wj  Azij uj  Bzij vj =CiiZ ;
i6¼j
   
@T @T @T @T @ @T
qCP ¼ qCP u þv þw þ K the nodal pressure coefficients are defined as:
@t @x @y @z @x @x 0 1
  ; Z
@ @T @ @T
þ K þ ðK Þ þ g_c 2
Kiu ¼ @ Ni dV A=Axii ;
@y @y @z @z
V
ð3Þ 0 1
Z
where x, y, z are 3D coordinates and u, v, w are the Kiv ¼ @ Ni dV A=Byii ;
velocity components in the x, y, z directions. P, T, q and V
g denote pressure, temperature, density and viscosity, 0 1
Z
respectively.
Kiw ¼ @ Ni dV A=Ciiz ;
Cross-viscosity model has been used for the simula-
tions: V
291

where Axij, Bxij, Cxij, Ayij, Byj , Cyij, Azij, Bzij, Czij represent global velocity field obtained by solving momentum equations
velocity coefficient matrices in the direction of x, y, z does not satisfy continuity equation. The velocities are
coordinate, respectively. Kui , Kvi , Kwi denote the nodal updated using the following relations:
pressure coefficients in the direction of x, y, z coordinate, Z Z
respectively. The nodal values for Kui , Kvi , Kwi are ob- 1 @P 1 @P
ui ¼ ~ui  x N dV vi ¼ ~vi  y N dV
tained by assembling the element-by-element contribu- Aii @x Bii @y
V V
tions in the conventional manner. Ni is element Z
interpolation and i means global node number and j is, 1 @P
wi ¼ w~i  z N dV :
for a node, the amount of the nodes around it. Cii @z
V

3.2 Pressure equation 3.5 The tracing of the flow fronts

Substitution of the velocity expressions (4) into discret- The flow of fluid in the cavity is unsteady and the
ized continuity equation, which is discretized using position of the flow fronts varies with time. Like in 2.5D
Galerkin method, yields element equation for pressure: model, in this paper, the control volume method is em-
Z      ployed to trace the position of the flow fronts after the
@Ni @Nk @Ni @Nk FAN (flow analysis network)[14]. But 3D control vol-
Nj Kju Pk þ Nj Kjv Pk
@x @x @y @y ume is a spacial volume and more complex than the 2D
V
  control volume. It is required that 3D control volumes
@Ni @Nk of all nodes fill the part cavity without gap and hollow
þ Nj Kjw Pk dV
Z @z @z  space. Two 3D control volumes are shown in Fig. 1.
@Ni @Ni @Ni
¼ Nj ~uj þ Nj~vj þ Nj w~ j dV :
@x @y @z
V 4 Finite element calculations for the temperature field
The element pressure equations are assembled in the
The temperature field plays an important role during
conventional manner to form the global pressure equa-
injection molding process. Because the viscosity of the
tions.
polymer varies with its temperature, so the variation of
the temperature of polymer will have important influ-
ence to the injection molding process. Only after the
3.3 Boundary conditions
temperature field during filling has been calculated ex-
actly, the simulations for packing and cooling are
In the cavity wall, the no-slip boundary conditions are
meaningful. In 2.5D model, though the variation of the
employed, e.g.,
temperature in the gapwise direction is solved by FDM,
u ¼ v ¼ w ¼ 0; ~u ¼ ~v ¼ w
~ ¼ 0; Kiu ¼ Kiv ¼ Kiw ¼ 0 the model is based on the Hele–Shaw approximation,
which supposes the injection-molded parts are thin. As
on an inlet boundary, shown in Fig. 2, in 2.5D model, the triangular elements
u ¼ v ¼ w ¼ given Kiu ¼ Kiv ¼ Kiw ¼ 0: are meshed in the gapwise direction by creating finite
difference grids, and the temperature in flow plane is
represented by linear interpolation, and the tempera-
3.4 Velocity update tures in the gapwise direction are represented by FDM.
In 2.5D models, the velocity in the gapwise direction is
After the pressure field has been obtained, the velocity neglected, so only the thermal conduction item is con-
values are updated using new pressure field because the sidered in the gapwise direction. This paper gives a 3D

Fig. 1 3D control volumes.


a Control volume of an internal
node and b a boundary node
292

Thermal convection item and viscous heat item are


anisotropic and has to do with the direction of flow. To
keep the numerical stability, the upwind method is em-
ployed to handle the convection item and viscous heat
item, e.g., only the contributions of the upriver elements
from the nodes are considered when the convection item
and viscous heat item are calculated.
In the above equations, the time T is discretized using
a forward-difference method:
nþ1
@Tj Tj  Tjn
¼ ;
@t Dt
where Dt denotes time step.
The element temperature equations are assembled in
the conventional manner to form the global temperature
equations. The overall procedure for pressure and tem-
perature calculations is relaxation iterative. Because the
pressure, velocity and temperature influence each other
during the calculation, the temperature and pressure are
Fig. 2 Illustrative finite difference in the gapwise direction coupled during the procedure.

model for calculating the temperature field which con- 5 Results and discussion
siders the influence of convection items on three
dimensions and suitable for the wider range of parts and The first test cavity and dimensions are shown in
has more exact results compared with the 2.5D models. Fig. 3a. The meshed 3D model of cavity is shown in
According to the energy equation (3), by the use of Fig. 3b. The selected material is ABS780 from Kumbo.
Galerkin’s method, the equation for the temperature The parametric constants corresponding to the n,s*,
field can be expressed as following: B,Tb and b of the five-constant cross-type viscosity
Z Z    model are 0.2638, 4.514 · 104 Pa, 3.13198043 ·
@T @T @T @T
N qCp dV ¼ N  qCp u þv þw 107 Pa S, 1.12236 · 104 K, 0.000Pa 1. Injection tem-
@t @x @y @z
perature is 45C, mold temperature is 250C, injection
V   V     
@ @T @ @T @ @T flow rate is 44.82 cu cm/s.
þ K þ K þ K þ g_c2 dV : ‘‘Fountain flow’’ is a typical flow phenomenon during
@x @x @y @y @z @z
filling. It has to do with the fluid near the center moving

Fig. 3 The test cavity. a The


cavity dimension and b the
meshed cavity

Fig. 4 Comparison between


predicted shapes of flow front
based on present 3D model (a)
and based on 2.5D model (b).
a Shape of 3D flow front and
b shape of 2.5D flow front
293

Table 1 Material properties

Index Material Unit Reference


property value

1 Density kg/m3 968.6


(q)
2 Specific J/kg K 1.70 · 103
heat (Cp)
3 Thermal W/(m K) 0.140
conduc-
tivity (K)
4 Cross-type
viscosity model
N 0.3783
B Pa s 1.0527 · 103
Tb K 9.3841 · 103
Fig. 5 The example cavity
b 1/Pa 0
s Pa 1.955 · 103

faster than the average across the thickness and upon


catching up with the front, deflecting to move toward
the walls, so the shape of the flow front is round like the
fountain. In 2.5D models, the convection effects in the
fountain region cannot be represented and the details of
the fountain region are also lost, as shown in Fig. 4b. In
presented 3D model, this fountain flow phenomenon can
also be simulated. The round shape of the flow fronts at
three filling times is illustrated clearly in Fig. 4a.
Another example is typical of an industrial
application as shown in Fig. 5. The outline dimensions
of the cavity are 63.3 · 43.4 · 24.2 mm3 with a thick-
Fig. 6 The meshed cavity ness 4 mm. ‘‘ ’’ represents the location of entrance.

Fig. 7 The flow front at four


different filling times.
Time = 0.08, 0.36, 0.65 and
0.80 s
294

Fig. 8 Temperature field on the


plane Z = 13 in four filling
times. Time = 0.08, 0.36, 0.64
and 0.82 s

The meshed cavity is as in Fig. 6. The injection tem- especially for these cavities with complex geometry and
perature is 250C, mold temperature is 45C, injection thick walls. Compared with the 2.5D model, which can
time is 0.82 s. The selected material is PS ASAHIPS 408. only simulate the flow of melt in the surface of the
The material properties and the parametric constants cavity, 3D simulation model is suitable for the wider
corresponding to the five-constant cross-type viscosity range of cavities and has more exact results.
model are specified in Table 1. Figure 8 shows the temperature distributions on the
Figure 7 shows the locations of flow fronts in four plane Z = 13 in four filling times. It can be seen that
different filling times. A complex 3D flow field develops there is a higher temperature in the interior of the cavity
in the cavity and a rounded free surface is clearly seen. It and the lower temperature near the cavity walls. The
can be seen that the filling process of the melt in the temperature near the entrance is even higher than the
interior of the cavity can be predicted in the 3D model, injection temperature due to the viscous heating of the
and it is crucial for predicting more exactly the locations melt: Figure 8 shows that a thermal layer is presented in
of the weldlines, the possibility of the air entrapments as the filled portion of the cavity, in which there is a vari-
well as the pressure and temperature distributions, ation of temperature from wall to the interior of cavity.
295

It can be seen that the heat transfer is mainly driven by 2. Schlichting H (1968) Boundary-layer theory. McGraw-Hill,
convection and the conduction is rather small, so the New York
3. Chen BS, Liu WH (1989) Numerical simulation and experi-
thermal layer is very thin. It can be seen that the tem- mental investigation of injection mold filling with melt solidi-
perature distributions on the arbitrary section plane can fication. Polym Eng Sci 29:1039–1050
be seen clearly in the 3D model. On the contrary, in the 4. Dequn L (2002) New progress of flow simulation for plastic
2.5D model, only the average temperature in the thick- injection molding. China Int Forum Die Mould Technol 5:47–
48
ness direction is shown on the middle-plane or the sur- 5. Huamin Z, Dequn L (2002) Computer filling simulations of
faces of the cavity, and for the thick or nonuniform- injection molding based on 3D surface model. Polym Plast
thickness parts, which are not, suitable for the Hele– Tech Eng 41:91–1021
Shaw approximation, the results from 2.5D model have 6. Hiebert CA, Shen SF (1980) A finite-element/finite-difference
much error and even are mistakes. simulation of injection molding filling process. J Non-Newto-
nian Fluid Mesh 7:1–32
7. Inoue Y, Higashi T, Matsuoka T (1996) Numerical simulation
of 3-dimensional flow in injection molding. ANTEC Proc
6 Conclusion 1:744–748
8. Pichelin E, Coupez T (1998) Finite element solution of the 3D
filling problem for viscous incompressible fluid. Comput
A numerical model to simulate the filling of injection Methods Appl Mech Eng 163:359–371
molding based on a 3D finite element model is presented 9. Hwang CJ, Kwon TH (2002) A full 3D finite element analysis
in this paper. The 3D model uses the equal-order of the powder injection molding filling process including slip
velocity–pressure formulation method and a 3D control phenomena. Polym Eng Sci 42:33–50
10. Prakash C, Patankar SV (1985) A control volume-based finite-
volume scheme is adopted to track the flow front. element method for solving the Navier–Stokes equations using
During calculating temperature field, the influence of equal-order velocity–pressure interpolation. Numer Heat
convection items in three directions is considered in or- Transfer 8:259–280
der to get more exact results and to apply to the wider 11. Prakash C (1986) An improved control volume finite-element
method for heat and mass transfer, and for fluid flow using
range of parts. Two parts have been employed as equal-order velocity–pressure interpolation. Numer Heat
example to test the validity. It has been seen that 3D Transfer 9:253–276
simulation model is suitable for the wider range of parts 12. Rice JG, Schnipe RJ (1986) An equal-order velocity–pressure
and has more exact results compared with 2.5D models. formulation that does not exhibit spurious pressure modes.
Comput Methods Appl Mech Eng 58:135–149
13. Hieber CA (1987) Injection and compression molding funda-
Acknowledgements The authors would like to acknowledge finan-
mentals. Marcel Dekker, New York
cial support from the National Natural Science Foundation
14. Tadmor Z, Broyer E, Gutfinger C (1974). Flow analysis
Council of the People’s Republic of China, under Grant 20490220
method for solving flow problems in polymer processing.
and Research Foundation for PhD Candidates of Universities of
Polym Eng Sci 14:660–665
the People’s Republic of China (20020487032).

References

1. Gotoh Terumasa, Iizuka, Miyamoto Masayuki et al (1986)


Simulation of polymeric flows in the cavity filling process of
injection molding. Sharp Tech J 34:63–70

S-ar putea să vă placă și