Analyzing plastics with FEA: Part 6

Sinks and voids can be the nemesis of an otherwise well-designed part. But how does a designer know beforehand that the part he or she is creating on a CAD screen will cause major headaches at the molding press?

According to Shrikant Oak, an analyst at injection molding process simulation specialist Feamold Inc. (Troy, MI), the creation of sinks and voids during the molding process is a 3-D phenomenon that can be detected using the finite-element method. However, few analytical studies are published in this area. Why is it so difficult to simulate voids? All analysis packages assume that flow in the thickness direction is minimal, so they don't account for it. If they did, compute times would be excessive. Unfortunately, the pressure causing sinks and voids is in the thickness direction. So to minimize sink and void creation, analysts at Feamold track temperature across various points of the part, then find out how to keep the flow path and gate unfrozen for the longest time.

IMM asked Oak to share a recent project for Siemens Electric Corp. that brings this type of analysis to light. He explains how the project began: "Siemens' Ben Reginella came to Feamold with a slight problem. Prototype tools for a resonator, molded from glass-filled polypropylene, were turning out parts with voids at a thick T-section supporting a locator tab," he recalls. "Typically, analysts have been unable to capture sink and void problems using CAE, but with the help of C-Mold filling analysis software and our own in-house expertise, we were eager to try."

According to Chrysler's John Van Hout, project engineer for Large Car Platform, the resonator is a portion of the air induction system that helps attenuate noise generated by the engine. This part experiences short-term temperatures up to 250F under the hood. "The initial concern," says Van Hout, "involved the structural integrity of the locator tab. We performed an FEA to verify that the design would meet our requirements, then asked our supplier Siemens to implement that design."

During the prototype stage, however, Siemens found voids at the critical section that weakened the tab itself. That's when Feamold received the project. "Our objective was to evaluate options to minimize the potential for voids at the base of the thick locator tab, and to determine the optimum gating design for minimum part warpage," adds another analyst, Ross Nordin.

Main criteria used to compare part warpage characteristics included bowing in of the part rim and lifting or sagging of the part away from and toward the mating half. Analysis was performed using two gates, each 60 mm wide, for a total width of 120 mm along one edge of the part.

Feamold analyzed several different designs, varying the gate location along the long edge of the part and along the locator tab flange. However, the design already in production at Siemens ultimately prevailed to avoid the cost of recreating prototype tools.

Siemens provided part geometry to Feamold in the form of an IGES file. A wireframe model of the part was created, and topological attributes of thickness and shape factor were assigned to it. The design incorporates two tunnel gates at two ribs along one side of the part (Figure 1). Pressure required to fill the part was approximately 9000 psi.

Process Specifics

In the final, optimized design, nominal wall thickness of 2.5 mm was increased to 3.5 mm close to the gates for better packing of the locator tab base, ultimately reducing the potential for voids. In this case, Oak says, the uneven wall stock did not have an adverse effect on part warpage.

Feamold used the following conditions for the analysis, which mimicked conditions during mold trials:

The shrinkage level (magnitude) observed in the analysis was higher than ideal. However, the shrinkage gradient was low, indicating that the part would shrink uniformly. Uniform shrinkage results in uniform deformation, that is, less warpage. Shrinkage can be reduced further by increasing packing pressure. The packing pressure used for the analysis was 4000 psi, resulting in a maximum shear stress in the part of approximately 28 psi, well below the acceptable limit of the material. Oak adds, "For this particular part, packing pressure can be increased up to 8000 psi, without seriously over-stressing the material."

Evaluating the Design

Two criteria were used to determine the effects of gate design and location. First, analysts checked these factors for their potential to create voids at the locator tab base. Second, they were analyzed for their contribution to part warpage.

The base of the locator tab (where it joins the part wall) has very thick sections for structural integrity. The part wall leading to this thick section originally had a nominal wall of 2.5 mm. The thinner part wall would freeze early, thereby cutting off the (packing) pressure to the thick section, while the core of the thick section remained molten. This would allow the thick section to solidify under low or zero pressure, resulting in voids in the section.

Analysis showed that the void problem could be eliminated by allowing longer packing time and/ or increasing packing pressure for the thick sections. Moving the gate close to the thick section increases the packing pressure at the thick section.

Increasing the thickness of the part wall between the gate and the thick section also allows the part wall to remain fluid longer, and therefore allows the thick section to be packed longer. Cooling time is proportional to the square of thickness. Therefore a small increase in the part wall thickness allows it to remain fluid for a significantly longer time.

Oak explains that the potential for sink and voids cannot be explicitly identified by the analysis software. Instead, the temperature traces were examined at various points along the flow path leading to the thick sections, as well as sampling points in the thick section itself. The temperature traces at these sampling points are color coordinated; the trace color and the marker color are the same. In the first iteration, a single gate was located along the center line of the short edge of the part.

Void Creation

To understand how voids are created in thick sections, Oak offers the following explanation.

During the injection phase, plastic temperatures are lowest at the mold walls and hottest at the center. More of the cross section begins to freeze and shrink as pack and hold pressure is applied. This pressure should compensate for shrinkage by pushing more material into the areas that are contracting. But if pressure is too low or is not applied long enough, sinks and voids are created. Packing pressure has its limits, however, and cannot go beyond a safe range for the material. Packing time can be varied more easily.

Melt velocity also has a significant effect on temperature. The velocity determines how much convective and shear heating will be experienced. Also, if velocity decreases, the residence time may increase, thereby allowing greater cooling due to conduction. When melt velocity drops to zero (at freezeoff), there is no more positive contribution to the heat load; instead, the negative contribution is increased, resulting in a drastic change in the thermal balance and a sharp drop in the temperature trace at that location.

Both pack and hold pressures and residence time contributed to voids in the resonator. As the path feeding the thick section froze off early, the thick sections were allowed to cool under no or low pressure. Cooling occurs from outside-in, that is, the outer skin freezes first and the core is the last to cool. As the outer layers cooled, they shrank, pulling the fluid core towards them, resulting in voids.

Packing Out Voids

Several additional iterations were performed to analyze design options to pack out the base of the locator tab. Optimum results were achieved when a single gate was located along the horizontal flange of the locator tab. However, this gate location was not practical, since it could cause leakage or maintenance problems in that region. Instead, a long edge gate along the short edge of the part was chosen, which reduced the distance from the gate to the locator tab to transmit greater packing pressure.

In this case, Oak recalls, the reduction in flow length alone was not sufficient, because it did not eliminate the problem of early freezeoff of the flow path leading to the tab. To allow the flow path to remain open, the part section in that area was thickened iteratively. Optimum results were achieved when the green area was thickened to 3.5 mm.

A single long edge gate required a thick secondary (feeder) runner tapering down into a thinner film gate, along one edge of the part (Figure 1, p. 37). Based on feedback received from Siemens' Reginella, analysts felt that the presence of the thick secondary runner would increase the heat load in that area, and that this was undesirable. To eliminate the need for a secondary runner, a split runner scheme was utilized (Figure 1, p. 37).

Temperature traces taken at the sampling locations, using the final

recommended design, showed that the flow path leading to the base of the locator tab remained fluid much longer, thereby allowing better packing.

Taming Warpage

Feamold compared warpage characteristics for three resonator designs. To do this, a reference (datum) plane was first defined:

Assume that the blue and green outlines represent the original and deformed part shapes, respectively. To measure part warpage, the deformed part could be placed on a flat surface, and the distance from the flat surface of any point on the part can be measured as displacement or deformation.

If the deformed part is kept on the table, so that any one of the corners of the deformed part coincides with the original location of this corner, then this corner acts as the local origin for reference, and the flat surface defines the reference plane.

The anchor plane is defined by three nodes (Figure 2, p. 38):

To measure deformation characteristics, sampling points were chosen by the analysts along the rim of the part, and the nodal displacement values at these locations were recorded.

The first criterion was to measure the tendency of the part (top) edges to bow in. All designs had comparable warpage characteristics. There was no significant difference between the three designs, indicating that the proposed design had ample tolerance to allow minor changes if required during tooling.

Secondly, Feamold measured the displacement in the Z direction, the tendency of the part to lift away from (positive) or sag toward (negative) the mating half. Again, all three designs showed similar results, as shown in Figure 3.

In terms of absolute deflection, the bowing in of the long edge was the greatest deflection observed in the part. Therefore, analysts concluded any improvement in this characteristic would result in the greatest improvement in part quality. - Michelle Maniscalco

Statistical Know-How

When it comes to interpreting analysis results for temperature and stress, Oak explains, single point minimum or maximum values can be misleading. For example, consider a case where 99 percent of the part is at a uniform temperature, say 500F, and only one tiny point may be at 200F. Typically, such local discrepancies may be ignored, since they do not affect the process as a whole. Rather than basing your results on these single points, the following data are also useful: 95th percentile value.This is the value associated with 95 percent (by volume) of the part.5th percentile value.This is the value associated with 5 percent (by volume) of the part.Average.The numerical average. RMS deviation.Root Mean Square deviation from the above numerical average.

The characteristic of a good design is that the 95th percentile value should be within the acceptable processing/design range, and the RMS deviation should be as low as possible, indicating low variation or gradients in the part.