WARPAGE SIMULATION OF INJECTION OVER-MOLDING PLASTICS ON CONTINUOUS FIBER REINFORCED COMPOSITES Zhihao Zuo, Zhiliang Fan, Franco Costa, David Astbury Autodesk Australia Pty Ltd, Moldflow R&D Center, 259 Colchester Rd, Kilsyth VIC 3137, Australia Abstract Injection overmolding of thermoplastic over a continuous fiber reinforced composite is one of the new manufacturing approaches for automotive lightweighting which is emerging as a potential way to increase vehicle fuel economy. It not only takes advantage of excellent strength and stiffness properties of continuous fiber reinforced composite, but also has the advantage of forming complex and intricate functional shapes with the injection molding process. Warpage simulation of injection molding helps designers optimize the part and mold design, material choice and processing parameters, in order to meet tight dimensional tolerances for assembly purposes. In this paper, we extend our warpage simulations to account for the effects of orthotropic and fully anisotropic mechanical properties of continuous fiber reinforced composite inserts. The feature enhancement includes buckling and large deflection analyses of overmolded plastic components. The numerical results from Autodesk Moldflow simulation for two plastic parts injection over-molded onto continuous fiber reinforced composites are presented. Introduction Lightweighting is a major trend in automotive industries for reducing weight and environmental costs while increasing performance. Recent developments in low-cost advanced lightweight composite materials offer promising potential in final product implementation. The field of automotive composites is expanding rapidly to exterior, interior, and under-the-hood structural, semi-structural, non-structural and mechanical applications. Figure 1 shows an example of a plastic part injection-overmolded onto a thermoformed continuous fiber reinforced sheet. Continuous fiber reinforced composites offer advanced mechanical properties such as strength and stiffness, however, they cannot be used for intricate functional shapes such as ribs, bosses, bolt locations, etc. Addressing this shortcoming, interest is rapidly growing in technologies such as multi-material injection molding or overmolding. In addition, overmolding is an excellent approach to producing lightweight technical parts and can reduce production and assembly costs. Applications of over-molding plastic components help meet the current demanding requirements from modern industries including automotive and consumer products; further exploiting this technology makes it possible to incorporate innovative design features that are not possible with traditional materials. Figure 1. An injection overmolding plastics. One of the key aspects of the production of automotive components via the injection molding technology is the need to meet tight dimensional tolerances for assembly purposes. Towards the aim of optimizing designs for improved product quality, warpage simulation predicts the final shape of an injection molded part, and thus allows designers to avoid potential problems early in the design stage. To fully consider the complex effects introduced by inserts in an injection overmolding process, a truly threedimensional finite element solution is required [1, 2]. In this paper, we extend our warpage simulations to account for orthotropic and fully anisotropic mechanical properties of continuous fiber reinforced composite inserts. This feature enhancement considers buckling and large deflection analyses of overmolded plastic components. Simulation results are presented for two plastic parts injection-overmolded onto continuous fiber reinforced composites. Warpage with Inserts For injection molded parts, warpage is caused by variations in shrinkage throughout the part, which is largely determined by the varying pressure and temperature histories coupled with the frozen layer growth SPE ANTEC™ Indianapolis 2016 / 1249 [1, 2, and 3]. The presence of continuous fiber reinforced composite inserts could significantly affect the warpage in four ways: Firstly, it could affect the temperature and pressure fields thus affecting the shrinkage of the injected polymer. A continuous fiber reinforced composite insert acts as an insulator and delays the cooling and solidification of the injected polymer in the cavity. Secondly, the injected plastic can solidify around the insert while still being hotter than the insert. The subsequent differential shrinkage between the injected polymers and anisotropic composite inserts becomes a source of warpage for injection over-molded parts. Thirdly, continuous fiber reinforced composite inserts will provide significant resistance to part warpage due to their great stiffness property. Lastly, if the insert is thermoformed by the closing action of the mold, residual stresses in the insert itself may influence the final part shape after ejection. In addition to many other assumptions associated with general injection molding simulations [1, 2, 3], an additional assumption used for over-molding with inserts is that the continuous fiber reinforced composite inserts and the injected polymer are well bonded. Fiber Composite Properties The mechanical properties of a continuous fiber reinforced composite insert have a great influence on its behavior under thermal expansion and forces/pressures during the injection molding process, and thus have a great impact on the shrinkage and warpage simulation of an injection overmolded part. There are many forms of common raw continuous fiber composites, such as pultruded rods, woven mats and unidirectional tapes; each consists of the reinforcing fibers and a base matrix material (thermoplastics, thermosets etc.) In most cases, it is likely that the fibers embedded in the composite component are strongly oriented. As fibers normally have very different physical properties than the matrix materials, the fiber orientation introduces significant anisotropy in the thermo-mechanical properties of the composite fiber-matrix system. As a result, isotropic constitutive models are not valid for continuous fiber reinforced composites. Not to lose generality, the full Hooke’s Law under a spatial Cartesian coordinate system can be described in the following form. σ ij = Cijkl ε kl contracted notations (subscripts 1 to 3 indicate normal directions, and 4-6 the shear terms) as follows ⎧ε 1 ⎫ ⎡C11 ⎪ε ⎪ ⎢ ⎪ 2⎪ ⎢ ⎪⎪ε 3 ⎪⎪ ⎢ ε kl → ⎨ ⎬ Cijkl → ⎢ ⎢ ⎪ε 4 ⎪ ⎢ ⎪ε 5 ⎪ ⎢ ⎪ ⎪ ⎪⎩ε 6 ⎪⎭ ⎢⎣ , C12 C 22 Sym C13 C 23 C14 C 24 C15 C 25 C33 C34 C 44 C35 C 45 C55 C16 ⎤ C 26 ⎥⎥ C36 ⎥ ⎥ C 46 ⎥ C56 ⎥ ⎥ C66 ⎥⎦ (2) For a fully anisotropic material without any plane of symmetry, e.g. combinations of multiple fiber sheets with random alignment angles, all the 21 elastic constants in the above stiffness matrix are independent of each other and potentially non-zero. For an orthotropic material that has at least two orthogonal planes of symmetry, the shear-normal terms (e.g. C14 and C36) and the off-diagonal shear-shear terms (e.g. C45 and C56) in the elastic matrix vanish, which leaves only nine independent elastic constants. Alternatively, the orthotropic properties can be described by three Young’s moduli, three Poisson’s ratios and three shear moduli, each defined on a principal direction. Orthotropy can be found in fiber reinforced composites with fiber layup in orthogonal directions, a typical example is the continuous carbon fiber reinforced thermoplastic composite tape. Unlike for the transversely isotropic material model commonly used for polymers, the coefficient of thermal expansion (CTE) for an orthotropic fiber composite can be different in all three principal directions. Consequently, there are three independent CTEs for an orthotropic composite. As for a fully anisotropic composite, the number of CTEs becomes six, i.e. each corresponds to one strain term as follows (3) ε kl = α kl ΔT It is worth pointing out that an injection overmolded design can involve fiber reinforced composite inserts of very complex geometries and double curved surfaces, e.g. from a thermoforming or draping process. As a result, the fiber sheet direction and thus the composite principal directions vary according to the insert surface normal and any in-plane shear which has occurred during forming. It is critical to define the local material axis correctly for use in simulation. Simulation Technology (1) Mesh Bonding where the tensors of stress/strain σij/εkl and the elastic stiffness Cijkl can be further order-reduced with the SPE ANTEC™ Indianapolis 2016 / 1250 For numerical implementation and accuracy purposes, an aligned mesh would be preferable at the interface of the cavity and continuous fiber reinforced insert; this way, the nodes on the cavity and insert interface would be perfectly matched so that displacements can be shared on both meshes of the cavity and insert. However, to enforce perfect matching requires a significant effort during meshing and may force the use of highly distorted elements. An alternative and more practical solution is to mesh the insert and cavity independently, and then force consistent physical (e.g. displacement) results on the interface. This latter approach was adopted in this work. used which combines the AMG-CG equation solver with the subspace eigenvalue iteration algorithm. To bond the cavity-insert interface together in the warpage solution, the analysis starts by identifying the interface area between the insert mesh and the cavity mesh. Based on the relative geometric position and the displacement interpolation function of the element, the relationships between the degrees of freedom at the nodes of the cavity mesh and the nodes of the insert mesh can be established. The relationships are typically described in multi-point constraint (MPC) equations. The insert nodes on the interface are regarded as the master nodes, and the cavity nodes as the slave nodes. It is recommended that the density of the insert mesh be coarser than that of the cavity mesh. With these MPCs, the structural behavior is expected to be identical to that of a perfectly matched mesh of the combined system of inserts and cavities. Various algorithms are available for dealing with MPC equations in finite element methods. The elimination method based on the Lagrange multiplier formulations is used in our simulation. In this method, the constrained degrees of freedom and multipliers are eliminated, thereby yielding a lower order matrix equation for the unconstrained degrees of freedom. Case 1 Numerical Examples Injection overmolding on orthotropic and fully anisotropic inserts has been implemented in the Autodesk Moldflow simulation package. Simulation results of two examples with anisotropic continuous fiber reinforced composite inserts are presented in the following to demonstrate the effectiveness and efficiency of the implementation. The first case addresses a demonstration model shown in Figure 2. The simple model consists of two overlapping Lshaped meshes, with one mesh representing the plastic part (in green), and the other representing the continuous fiber reinforced composite insert (in grey) onto which the plastic part is overmolded. The insert material is an E glass fabric with fibers at 0/90°. The material properties of the isotropic polymer (a relatively flexible PFA polymer) and the orthotropic composite insert are listed in Table 1. Significant orientation can be seen from the composite properties, which are locally defined, i.e. along the local axes following the surface direction change. A practical example of this type of insert would be a glass fiber angle created from a peel ply fabric. Buckling and Large Deflection A typical plastic component injection overmolded onto continuous fiber reinforced composites will be thin-walled in nature. Buckling may occur with the injection molding process. When the part is ejected from the mold, in-mold residual stresses which depend on the time, temperature and pressure history of the entire part [4, 5], have to be rebalanced. Consequently the part may buckle. In our simulation, the handling of anisotropic continuous fiber reinforced composite inserts has also been extended into the buckling and post-buckling warpage analyses [6, 7]. Classical linear buckling analysis generally yields a sufficiently accurate buckling indicator for most practical injection molding problems, where only small displacements and deformations occur before an instability point is reached. To deal with the large scale threedimensional finite element models of real-world injection overmolded plastics parts, a fast parallel Eigen-solver is Figure 2. An L-shaped part with a part insert. Figure 3 shows the temperature history of the process on a longitudinal cut section, including the progression of the injected melt front. While surface temperature quickly cools to the mold temperature, internal regions are seen to cool much more slowly. Figure 4 shows the final warped shape. For comparison purpose, two further tests were run where the insert material was set in turn to a soft unfilled polymer and a different fiber composite (see material properties in Table 1). The analysis results are shown in SPE ANTEC™ Indianapolis 2016 / 1251 Figure 5. Note that the stronger orthotropic insert restrains the part from warping, when comparing cases (a) and (b). On the other hand, when comparing cases (a) and (c), we observe a larger warp angle and greater shrinkages in the Z direction depending on the different fiber layup orientations. (a) T = 0.17s T = 0.56s (b) T = 2.98s T = 30.64s Figure 3. Temperature results at different times. (c) Figure 5. Warped shape and Z direction deflection with the insert set as three different materials: (a) E glass fabric, fibers at 0/90°, (b) PI X05, BIP-PF, (c) E glass fabric, fibers at +/-45°. Figure 4. Final warped shape of the part. SPE ANTEC™ Indianapolis 2016 / 1252 Case 2 The second case shown in Figure 6 addresses the simulation of a seat model with a PP material (in green) over-molded onto a continuous fiber reinforced sheet (in grey) and a polymer cushion (in yellow). The fiber orientation of the insert is deliberately made nonorthogonal, i.e. with fibers at 0/45° to demonstrate anisotropic behaviors. The insert has greater diagonal terms in the elastic matrix than the overmolding polymer, in order to demonstrate distinguished structural performances between the two different components. Note that the fibers of the insert run in compliance with the curved surface. T = 20.93s T = 50.77s Figure 6. An overmolded seat model. Figure 7. Molding and insert temperature results at various times during processing. Figure 7 shows the temperature distribution on a crosssection of the polymer part and inserts at the end of filling and at the ejection time. Figure 8 shows the predicted warpage from a large deflection analysis. Since overall the composite insert has much less shrinkage than the polymer, it is forced to bend outwards. Consequently, the angle between the seat back and bottom becomes larger after the warp deformation. It is also noticed that, despite the symmetry of the geometry and injection location, the deformed shape of the insert shows an asymmetric pattern about the part center line as shown in the contour plot in figure 9. This is caused by the anisotropic mechanical properties which result from the non-orthogonal fiber layup orientations. Figure 8. Final deformation of the warped part (large deflection analysis). SPE ANTEC™ Indianapolis 2016 / 1253 fiber orientation effect. Meshes of the cavity and inserts are bonded with MPC in the finite element formulation. Large deflection analyses were employed to predict part warpage. Numerical results show that the present 3D simulation technology is capable of predicting the warped shape of injection overmolded parts with continuous fiber reinforced composite inserts. References 1. 2. 3. Figure 9. Contour plot of insert deflection showing deflection asymmetry. 4. Summary 5. We have extended the Autodesk Moldflow 3D warpage simulation of injection overmolded parts to consider continuous fiber reinforced composite inserts. Orthotropic and fully anisotropic constituent relations can be specified for the composite inserts in order to take into account the 6. 7. P. Kennedy, R. Zheng, Flow Analysis of Injection Molds (2nd edition), Hanser Gardner Publications, New York (2013) Z. Fan, C. Kietzmann, S. Ray, F. Costa, P. Kennedy, SPE ANTEC Proceedings, 51: 568-572, Boston (2005) S. Ray, F. Costa, SPE ANTEC Proceedings, 49: 632636, Nashville (2003). R. Zheng, P. Kennedy, N. Phan-Thien, X-J. Fan, J. Non-Newtonian Fluid Mech. 84: 159-190 (1999). A. Guevara-Morales, U. Figueroa-Lopez, J. Mater. Sci. 49: 4399-4415 (2014). Z. Fan, D. Astbury, H. Wang, C. Friedl, SPE ANTEC Proceedings, 57: 1584-1588, Boston (2011). Z. Fan, H. Yu, J. Xu, D. Astbury, SPE ANTEC Proceedings, 59: 1383-1388, Ohio (2013). Table 1. Material properties of the polymer and insert (Moduli E/G, Poisson’s ratio v, thermal expansion coefficient a, thermal conductivity k). Polymer: PI X05, BIP-PF Insert: E glass fabric, fibers@0/90° Insert: E glass fabric, fibers@+/-45° E1 = E2 = 600, E1 = E2 = 25000, E3 = 1200, E1 = E2 = 12200, E3 = 1200, G12 = 204.08 G23 = G13 = 600, G12 = 4000 G23 = G13 = 600, G12 = 8000 v v12 = v23 = 0.45 v23 = v13 = 0.4, v12 = 0.2 v23 = v13 = 0.4, v12 = 0.2 α (1/C deg.) a1 = a2 = 0.00013 a1 = a2 = 7.4×10-6, a3 = 0.00012 a1 = a2 = 1.0×10-5, a3 = 0.00012 κ (W/mK) 0.19 0.3461 0.3461 E, G (MPa) SPE ANTEC™ Indianapolis 2016 / 1254