Material characterizations were carried out at the start of the project and the corresponding material cards for thermal, compression molding, and warpage simulation were generated on this basis. The viscosity, the pvT behavior, the thermal properties (heat capacity and heat conductivity), the crystallization kinetics, and the viscoelastic properties were measured for the LFT. In contrast, a reduced characterization was carried out for the tape so that only the crystallization kinetics and the viscoelastic properties were measured. 

Results from viscosity (left) and pvT (right) testing. 

To produce an LFT-tape-sandwich component, the individual components are placed on top of each other. For this purpose, the lower tape laminate is taken from an oven at a defined homogeneous temperature. Next, the LFT strand is transferred from the extruder on top before the second layer of tape laminate follows. Finally, the stack of LFT and tape is transferred into the mold. The resulting transfer and dwell times lead to cooling of the individual materials, which depends on the local stacking (sandwich or pure tape).  

The thermal behavior is predictable via Finite Element Analyses (FEA). First, a full-field 3D thermal model was set up in Abaqus. Such a model, however, is time-consuming in model setup and computation. 

Result of the thermal simulation (temperature) with Abaqus. 

For an efficient analysis of thermal behavior, a local 1D thermal simulation was conducted with SimuTherm. It is observed that both results are identical if the boundary effect between material regions can be ignored. SimuTherm delivers these results in real-time via an easy-to-use web interface and therefore provides the opportunity for efficient thermal analyses. 

On the basis of thermal analyses, different handling concepts and transfer times can now be compared and evaluated to identify an optimal process. Moreover, the predicted thermal fields of the tapes can be used as initial fields for both methods in the mold-filling simulation, which increases the prediction quality. 

Comparison of thermal simulations with Abaqus and SimuTherm.

The components investigated in the protECOlight project have ribs on the upper side of the sandwich, which are filled through the tape. Two variants were carried out during the compression molding tests: with and without flow aids. Flow aids are small slits or holes in the tape so that the LFT can flow directly through it. If no such flow aids are used, the tape needs to be damaged locally through suitable process control and geometric design of the ribs. 

To take rib filling into account in the mold filling simulation, simulation routes with and without flow aids were conducted. In both cases, thermal fields predicted by SimuTherm are integrated. The flow aids can be modeled in simulation directly. Therefore, corresponding models were set up and the rib filling was compared to experimental tests using different press force levels. The simulation showed excellent agreement with experimental tests with a slightly conservative tendency.

Unfilled areas observed for 8640 kN in experimental tests (left) and simulation (right) 

For the simulation without flow aids, a multi-stage simulation is adopted. Initially, a mold-filling simulation is carried out in which the rib filling is neglected. This involves predicting the pressure field, which is then exported and transferred to a forming simulation using SimuDrape. In this forming simulation, the forming of the tape into the ribs is analyzed as a function of the rib geometry and process control. By correlating the resulting deflection and curvature to the experimental results for tape damage under various processing conditions, a processing map is generated. Based on this, the likelihood of local tape damage and thus filling of the ribs becomes predictable. These results are then transferred back into a mold-filling simulation, in which the local damage is integrated via flow aids into the model to analyze the rib filling.

Forming simulation results for the curvature and deflection in a reduced model. 

In addition to molding simulation, analyses on warpage prediction were carried out on the demonstrator. Here, the Simutence approach SimuWarp, an add-on for Abaqus, was compared against a competitor product as a result of a combined filling, packing, and warpage analysis. In both cases, the thermal simulations described above served as input for the molding simulation. The simulation results were then compared to experimental results, which were digitized using a 3D scan measurement.  

The results reveal that the warpage pattern is correctly predicted by SimuWarp. In contrast, a competitor tool with the same data basis predicts a different warpage pattern. In both cases, differences in the warpage magnitude are observed. SimuWarp underestimated the warpage, whereas the competitor product significantly overestimated the warpage. 

Forming simulation results for the curvature and deflection in a reduced model. 

When considering possible causes for the underestimation of the magnitude in the warpage simulation with SimuWarp, two possible causes were identified: A temporal and inhomogeneous temperature distribution in the tool (also applies to the commercial tool), as well as possible inaccuracies in the modeling of the ribs, as it is not known exactly when these are filled. The further development of SimuWarp to overcome this is currently in progress. Stay tuned!