An injection mold for a water manifold was experiencing deflection on an extremely long core, causing a high defect rate on the finished product. In addition, inserted small plastic capillaries were being crushed if the mold parameters were outside a narrow window. By taking the pressure output from a filling simulation and using it as the boundary condition for a stress analysis on the mold itself, excellent correlation was achieved with the existing molding problems. This led to insight into the true cause of the core deflection along with a successful redesign of the tool to produce consistent, quality parts.
Original Part Overview
The water manifold consists of a tube that is 47 mm in diameter and 680 mm in length. The wall thickness is 2.5 mm.. Along one edge of the manifold is a series of 50 small tubes that are overmolded onto a blanket of mating capillaries. The overall assembly is used as part of a solar water heating system.
The manifold is made of 40% GF polypropylene. The overmolded tubes are composed of a 10% GF PP header fastented over an extruded PP core. The material that was used was reprocessed, and thus varied from lot-to-lot.
The primary core of the manifold was made of two interlocking steel cores. During molding these cores were deflecting, resulting in wall thickness variation up to the full wall thickness, which resulted in an unacceptable number of manufacturing rejects.
In addition, if the process parameters varied outside of a very narrow window the overmolded capillary tubes were being crushed close by the pressure of the manifold material. Unfortunately reducing the injection pressure or lengthening the cycle time resulted in short shots in the cavity. The processing window was exacerbated by the fact that regrind material was used. Each time a new material lot was used, a lengthy trial-and-error startup period was required to get the settings correct.
When tooling was started for a next-generation manifold we were hired by the manufacturer as consultants to perform simulations to improve the performance of the mold and reduce the incidence of manufacturing rejects.
Simulation of Core Deflection
Our strategy for the core analysis was to perform a mold filling simulation through the pack and hold phase to determine the pressure on the core, and feed those pressures into a linear structural analysis. Our original hypothesis was that the core deflection was being caused by the initial injection of molten plastic into the cavity as it emerged from the gate and hit the core. We believe that once the plastic had fully encapsulated the core the forces would be relatively balanced.
The original gate design had the material entering the cavity at two points along the long edge in a balanced fashion. The manufacturer was using intrusion injection to try to reduce the overall injection pressure and stabilize the core, and we simulated this through profiling the injection phase. The resultant hydrostatic pressure contours at an early point in the fill are shown in Figure 1. The key figure for the pressure is not the absolute pressure at each node, but the relative pressure between opposite sides of the core.
Figure 1: Pressure Contours
Not surprisingly, the highest pressures were determined to be at the points closest to the gates. However it turned out that this was also where the highest differential pressures were. Figure 2 is a graph that shows the pressure at the two gate nodes (454 and 1052) and a node opposite the gate (380). The most interesting feature of this graph is to note that the differential pressure between the point on the core closest to the gate and the point on the opposite is constantly increasing as the part fills – in other words the core deflection will actually become worse and worse as the fill progresses. Even though there is material on the opposite side of the gate fairly early on in the fill, it provides little support against the deflection. The maximum pressure differential varies from 0.5 to 1.3 MPa.
Figure 2: Pressure Profiles
This is supported by viewing the simulation’s prediction of the percentage of material that is frozen at the moment of maximum pressure. The relatively large wall thickness results in the vast majority of the plastic remaining molten until well into the packing phase.
Transfer of Pressures to Linear Static
The next step was to take the pressure contours from the filling simulation and feed them into the linear static analysis. We did this through a manual calculation method, doing a best fit through the pressure data at the worst point and transferring nodal pressures onto the cylindrical core model. Pressure differences varied from 0.5 to 1.3 MPa. The deflection results are shown in Figure 3. The simulation predicted a deflection of 3.4mm which is slightly larger than the wall thickness. This is consistent with the movement of the core fully against the opposite side. This gave us confidence in our methodology.
Figure 3: Core Deflection
Methodology Deficiencies
One deficiency with our analysis is that by breaking it down into two analyses (filling and structural) it did not allow for any time-varying conditions. For example, while maximum differential pressure on the core occurs at the end of the fill, there will be some deflection throughout the fill. This deflection will create a change in the wall thickness space that the plastic can flow through. In a non-newtonian fluid like a thermoplastic, the wall thickness has a big effect on how easy it is for the material to flow. It is commonly known that in injection molding plastic will move through thicker areas first. So as pressure forces the core towards the rear, plastic will preferentially fill the front of the core first, as there is more room there, and the area behind the core becomes restricted. This will increase the differential pressure, leading to even more deflection, and more preferential packing of the front of the core.
This positive feedback loop in the flow path is not captured by our methodology. However we tried to simulate it though a ‘snapshot’ model, by looking at the deflection of the core based on the early pressure differential, and applying those new wall thicknesses back to the original flow model. This did result in slightly higher pressure differentials that the original constant wall-thickness model.
Another deficiency is that we did not model the compressibility of the thermoplastic. As the core moves towards the back surface it will compress the material that is already there, especially since most of the movement occurs later in the filling cycle. This material, especially if it is frozen, will present resistance to the core motion that is not capture in the analysis. We believe this is the primary reason we are overpredicting the core deflection.
Simulation of Capillary Tube Crush
The second facet of the simulation was modeling the small overmolded capillary tubes. As stated earlier, there was a ‘blanket’ of approximately 50 tubes that were made of extruded 10% glass-filled polypropylene. During molding these tubes were sometimes crushed close. If too many were forced close the product would be rejected.
The simulation of this aspect of the design was simpler than the core deflection. The tubes are all located along the long edge of the core, exactly opposite the gate locations. This means that the pressure around each individual tube is, for all practical purposes, consistent top and bottom, and that the more dynamic flow features that were discussed above during the core evaluation do not come into play.
So all that needed to be done was to take the maximum pressure from the fill/pack phase and apply them to a model of the capillaries.
Structural Results
The result of this is shown in Figure 4. The maximum stress is 62 MPa compressive at the sharp inner corners. The average tensile yield stress for this material is 52 MPa. Compressive yield is typically double tensile. However there are also elevated temperature effects. Material data indicated a 50% reduction in tensile yield at 100 degC.
Figure 4: Capillary Tube Stress
While the exact temperature of the overmolded tubing at maximum pressure is not known, it is clear that 62 MPa is very close to the ultimate yield failure point of the capillary tubes. This is confirmed by the ‘on-the-edge’ nature of the actual molding process – slight variations result in total crush of the capillary tubes.
Alternate Gating Scenarios
Various alternatives to the original gate scheme were tested.
The first is to add additional gates along the edge, going from two gates to four. However, while the overall injection pressure was reduced, the differential pressure from front to rear still remained at approximately 0.5 MPa, resulting in similar core deflection to what was seen originally. This situation persisted regardless of how the gates were positioned, even when more advanced solutions such as valve gating were evaluated.
Of all the possibilities reviewed the only one that dramatically reduced the differential pressures was to gate on the end of part, front and rear as shown in Figure 5. This resulted in the flow front resembling a ‘ring’ that progressed down the length of the manifold. Because of this ring the pressure was equalized around the entire core.
Figure 5: Final Gate Configuration
There are two possibilities when gating on the end – either gating on a single side, or on both ends. Simulations indicated that gating on a single end would fill the part in spite of the long flow path, and would also have the advantage of having no weld line at the center. However it resulted in even higher pressures around the capillary tubes than was seen in the original design. Thus crushing was a big concern and killed this idea.
The alternative was to gate at both ends. This resulted in a sharp weld at the center of the part, however review of the filling analysis results showed that the material temperature was fairly high at the junction point, giving us some confidence that while aesthetics might suffer, the integrity of the weld should be good.
The pressure around the capillary tubes was slightly lower than in the original design, however we were still concerned that there would be the possibility of crush. In order to alleviate this problem the capillary tube itself was redesigned to give an adequate safety margin under the expected pressures during molding.
Conclusion
Based on this analysis, new tooling was constructed and the capillary tubes were redesigned to improve their strength. The modifications resulted in a very successful production startup, as the reject rate went from 15% to under 1%. Short shots, cross-sectioning and other measurements validated the results that were obtained from the analyses.
This project demonstrates that mold filling simulation can answer more questions than the typical ‘how does it fill’. The output from a fill/pack simulation can be used to determine the impact on inserts in the mold or the mold steel itself.
The main deficiencies with this type of coupled analysis is the difficulty in using the output of one analysis as the input of another, both in the simple transfer of boundary conditions (pressure contours in this case) and the more complex dynamic coupling between mold steel movement and wall thickness change. While specialized in nature, it might be an additional capability that software vendors may consider adding in the future.
Keywords
Finite Element Analysis
Filling Analysis
Injection Molding
Case Study
