Buckling Analysis in 3D Warp Validation Report
Buckling Analysis in 3D Warp Validation Report
Buckling Analysis in 3D Warp Validation Report
2014
Contents
Introduction ...................................................................................................................... 3
Buckling Algorithm .......................................................................................................... 3
Buckling for Injection Over-molded parts ...................................................................... 5
Graphical User Interface Design for Large Deflection Analysis ................................... 6
Numerical Examples and Discussion ............................................................................. 6
Summary ........................................................................................................................... 8
References ........................................................................................................................ 8
2
BUCKLING ANALYSIS IN 3D WARP
Introduction
As typical polymer injection molded parts have two distinctive properties: low Young’s
moduli of plastics (that are often two orders of magnitude less than steel), and thin-walled
in nature, buckling may occur during the injection molding process. The molded part may
buckle on ejection because of the in-mold residual stresses, which are dependent on the
time, temperature and pressure processing history of the entire part. These stresses have
to be rebalanced, causing the buckling. Buckling analysis is useful for optimizing part and
mold design, material choice, and processing parameters to help control part deformation.
Classical linear buckling analyses generally yield sufficiently accurate results for most
practical injection molded problems, in which only small displacements and deformations
occur before an instability point is reached.
The 3D flow, cooling and warpage simulation technique was developed several years ago
[1, 2, 3]. The high gradient variation of velocity and temperature in gap-wise directions,
and the limitation of maximum element aspect ratio, necessitate the use of many elements
across the part thickness. Consequently, the number of elements for a 3D analysis has
becomes very large. The issue becomes worse in the warpage simulation as the
tetrahedral elements must be upgraded from first-order elements to second-order
elements to get reliable warpage simulation results for thin-walled injection molded parts.
Although the number of nodes per element is increased only from 4 to 10, the total
number of nodes in a 3D mesh will be increased by around 6 times. This leads to an
excessive memory requirement and computation time, particularly when geometric
nonlinearity is included in the warpage analysis [4]. To overcome this, a very fast and
robust parallelized AMG-CG equation solver is used whose convergence rate is
approximately independent of mesh resolution [5].
The buckling analysis has also been extended to the injection insert over-molding process
and two-shot over-molding process. The buckling analysis results for two injection
simulation cases are presented.
Buckling Algorithm
The aim of a buckling analysis is to find one or more of the lowest positive eigenvalues
and corresponding eigenvectors for the following generalized eigen-system.
Where is the linear stiffness matrix, is the geometric or initial stress matrix. and
are the eigenvalue and corresponding eigenvector respectively.
The Lanczos method and subspace iteration method (or variants of these two iterative
schemes) are very widely used for the solution of large eigenvalue problems in finite
3
BUCKLING ANALYSIS IN 3D WARP
The subspace iteration algorithm starts with the initial matrix X1 for the requested
eigenvectors. It projects the large eigenvalue system into a much smaller subspace. The
main iteration steps include:
2. Find the reduced linear stiffness matrix and initial stress matrix,
3. Use the generalized Jacobi iteration method to solve the eigen-equation of the
projected matrices for its eigenvalues and eigenvectors
The most time-consuming linear equation solution step is traditionally solved using the
direct solver. Its distinct advantage is that the repeated linear equation solution during the
subspace iteration is just a backward substitution once the sparse direct factorization of
is available. However, as we discussed in the introduction, the number of unknowns
tend to be very large in the 3D injection molding warpage simulation. In some real-world
industrial applications, the number of unknowns can reach several millions or even tens of
millions. Therefore, performing the sparse direct factorizations could become prohibitively
expensive. The excessive memory requirement and extreme computational time for the
large-scale eigen-system makes the direct solver an unrealistic option even on modern
powerful workstations. A feasible approach is to replace the sparse direct method with an
AMG-CG equation solver within the subspace iteration algorithm.
As remains unchanged in the solution process except that the shifting procedure is
applied, the preconditioning calculation in AMG-CG is done once only during the entire
subspace iteration. On the other hand, on modern cache-based machines, data locality is
often the most critical factor affecting performance. The linear equation solution of multiple
4
BUCKLING ANALYSIS IN 3D WARP
iteration vectors in step 1 of the subspace iteration method enables efficient parallelization
implementation by exploiting the data locality [7].
The negative eigenvalue has no physical meaning in the warpage analysis. If negative
eigenvalues occur, the shifting procedure is automatically activated for getting the lowest
positive eigenvalues.
Us =BUm
Where Us is the slave displacement vector, and Um is the master displacement vector. B is
the master-slave displacement relationship matrix. With these constraints the structural
performance is expected to be identical to the composite structure with perfect bonding
between inserts and cavities.
The elimination method based on the Lagrange multiplier formulation is used for
handling MPC equations, in which the constrained degrees of freedom and multipliers are
eliminated, thereby yielding lower order matrix equations for unconstrained degrees of
freedom [8]. The buckling load and corresponding buckling modes for injection over-
molded parts can be obtained by solving the following generalized eigen-problem.
Where , , and are the master zone, slave zone and coupling zones of
the linear stiffness matrices respectively. , , and are the master zone,
slave zone and coupling zones of the initial stress matrices respectively.
5
BUCKLING ANALYSIS IN 3D WARP
There are two choices in the Solver parameters dialog: Number of eigenvalues to
output and Convergence tolerance for eigenvalue calculation. The default settings
are normally good enough. However, you can output more eigenvalues to see several
buckling load and buckling modes. In some special cases, the lowest positive eigenvalues
may be close to each other.
The implementation has been verified by comparing the predicted buckling load levels and
buckling shapes with the analytical solutions for several simple thin plates under different
external loadings. Figure 2 shows the 3D buckling analysis results of a simply supported
plate subjected to an in-plane compressive loading. Both the predicted buckling load level
and buckling shape are consistent with the analytical solution [9].
6
BUCKLING ANALYSIS IN 3D WARP
The simulation results of two injection molded parts are presented to validate the
algorithms. The first model is a tray model, as shown in Figure 3. This is a very thin-walled
part with an average wall thickness of 0.80 mm. Although a midplane shell solution would
be the best option for this uniform wall thickness geometry, it is analyzed using a 3D
simulation to validate the solution. 169,017 4-node tetrahedral elements are used in the
flow simulation, and all elements have been upgraded to 10-node tetrahedral elements for
the warpage analysis. The number of nodes used in the flow and warpage analysis was
31,047 and 236,421 respectively. The second-order tetrahedral element is used for
avoiding the notorious locking issues in the first-order tetrahedral element [8].
The buckling analysis predicts a buckling load factor of 0.9013. This means that the part
buckles at 90.13 per cent of the in-mold residual stresses. Figure 5 shows the buckling
mode, which is consistent with the experimental results in Figure 4. To get the final post-
buckling shape, a large deflection analysis is required.
The second case is an over-molded mobile phone cover, as shown in Figure 6. Since the
insert has much less shrinkage than its surrounding injected polymer, it is forced to bend
upwards slightly. The buckling analysis predicts the lowest eigenvalue of 7.647, and the
corresponding buckling mode is shown in Figure 7. It indicates that the part will not buckle
after ejection from the mold.
7
BUCKLING ANALYSIS IN 3D WARP
It is worth mentioning that the 3D buckling solution developed in this paper has been
successfully used for warpage simulation of large-scale finite element models of real-
world injection molded parts. With the current approach, we solved the buckling problem
of a large finite element model of over 6.5-million degrees of freedom on a HP Z800
workstation within 5 hours.
Summary
A buckling feature for 3D warpage simulation of injection molded components has been
developed using a new fast parallel eigen-solver which combines the AMG-CG equation
solver with the subspace eigenvalue iteration algorithm. The buckling analysis has also
been extended to the insert over-molding and 2-shot over-molding processes. The
presented method has provided an efficient tool for predicting if the injection molded parts
buckle after ejection from the molds.
References
1. K. Talwar, F Costa, V Rajuplaem, L. Antanovski and C Friedl, Three Dimensional
Simulation of Plastic Injection Molding, SPE ANTEC Proceedings (1998).
8
BUCKLING ANALYSIS IN 3D WARP
9. W.C.Young, ROARK’s Formulas for Stress & Strain, 6th Edition, McGraw-Hill Book
Company(1989)
Autodesk and Moldflow are registered trademarks or trademarks of Autodesk, Inc., and/or its subsidiaries and/or
affiliates in the USA and/or other countries. All other brand names, product names, or trademarks belong to their
respective holders. Autodesk reserves the right to alter product and services offerings, and specifications and
pricing at any time without notice, and is not responsible for typographical or graphical errors that may appear in this
document.