Injection molding is one of the most used processes for mass producing plastic parts. It can produce complex parts with high dimensional stability within a few seconds/minutes. However, injection molding requires a high investment on the mold. The molds are usually made out of steel, which have a fabrication time some weeks depending on the design of the part.
In the past years, injection molding manufacturers have used a replaceable aluminum insert to create prototypes, or to produce customized parts that will not be mass produced. Nevertheless, these inserts need to be machined. Nowadays, additively manufactured plastic inserts are replacing the metal counterpart; the 3D printed plastic inserts require less time to manufacture for a fraction of the cost of metal tooling.
By using 3D printed inserts, the injection molding process and final part properties can change. First, the cooling of the part will be different due to polymer inserts lower thermal conductivity. The cooling time will increase and affect the quality and properties of the final part. Lastly, the number of cycles before the plastic insert fails will be significantly lower than the metal inserts. Therefore, the objective of this project is to study the number of cycles performed before insert failure, and the cooling of the parts and their effect on the mechanical properties. Different processes and resins will be used and compared to determine the most efficient and optimal injection molding process.
1. Design and Development of Experiments
The set up for the experiment is simple. It consists of using a conventional injection molding machine, in which a 3D printed insert with a tensile specimen cavity is fixed in an aluminum plate (Further details will be shown in section 1.1). It is imperative to remark that in this project the goal is to find the optimal performance of a specific polymer material used for the printed insert.
Furthermore, characterization tests of the insert will be performed. Characterization methods include: differential scanning calorimetry (DSC) to determine the heat capacity, physical measurements to obtain the density, laser flash analysis (LFA) to yield thermal conductivity of the insert, and tensile tests to determine the elasticity modulus. These characterizations are pivotal for the simulation part of our experiment; once all the information about the insert-material is gathered, accurate simulations regarding the injection molding 3D printed insert-cavity can be performed. Figure 1 summarizes the work that has been done and future work.
Figure 1. Design and development of experiments
1.1 Cavity Insert Design Considerations, Measurements, and Modeling
For this project, an aluminum plate was given to be attached to the injection molding machine. The aluminum plate had a portion cut out of it where the additively manufactured cavity could be inserted to hold the injection mold print (Figure 2).
Fig. 2: Aluminum Plate with inserted additive manufactured cavity
The plate and mold insert combination was placed in between the pre-existing metal mold plates used in the injection molding machine. The insert cavity faced towards the metal plate attached to the injection molding screw, replacing the mold geometry of the plate it’s covering (Figure 3a). The mold plate it faced has half of the runner system; to optimize the flow of the material, the runner system was mirrored in this mold. This is best achieved by copying the cavity and runner geometry of the covered plate (Figure 3b). Unfortunately, the company who manufactured the metal mold did not release public CAD files.
Figure 3. (a)Mold Plates with and without the custom cavity and (b) Shaded geometry of the mold plate
Two major design considerations were to fixate our aluminum plate to the left mold plate and to implement the injector pins into our process. For fixing the mold, carving out slots in the aluminum mold and the metal plate and joining them together with an adhesive was considered, but that seemed impractical. The other concern was the injector pins. Normally injector pins on the left mold plate are retracted, and when the part is finished molding, they come out and push out the part from the mold. If the aluminum plate could be fixated to the left mold plate, then the injector pins could potentially push out our polymer insert if the injection molded part was stuck in the cavity.
Fortunately, after running the machine to test out those processes, neither of those concerns were an issue. During the process, when the plate was clamped in between the two mold plates and then released, it surprisingly stayed facing the right mold plate and hung on to guide pins. Since the part was stuck on the right side, it seemed easy to simply remove the injection molded part manually, and forgo the ejector process and need to fix it to the left mold plate. Due to the option to run the machine in a semi-automatic mode, the testing process was still streamlines as long the aluminum plate was hung on the right mold plate, and the molded part was manually removed in between shots.
Measuring the dimensions of the mold directly was a challenge due to the mold plate being securely fixed in the machine. Instead, a trace method was used, where the mold plated was covered with a piece of paper and a pencil was used to roll over the mold and shade the geometry onto the paper (Figure 3b). With the shaded geometry of the mold completed, a caliper was used to measure the dimensions of the runner system and the test specimen.
Being that the aluminum plate was previously manufactured, the dimensions of the cut out were used to model the lengths, width, and heights of the cavity insert. The runner system needed to be precise to match the plate the insert was facing. While it is possible to use a half circle geometry for the runner system, utilizing only the pre-existing plate, it was important to confirm there were no issues with filling the cavity to make a properly-working runner system. By measuring the shaded drawing along with some parts injected with the metal molds, the dimensions of the runner system were obtained. The gate dimensions were approximated using the same methods as the runner system; and all that was left was the part.
While the shape of the cavity did not need to match the metal mold’s because there were no constrictions from the mold the insert was clamped against, the same geometry was used to conduct stress tests on the part.
Additionally, some of the dimensions were hard to measure, most notably the radius of the neck region. After searching online for similar geometries, an article was discovered where there was a drawing with dimensions similar to our specimen [1]. Double checking the measured values against the drawing showed that the geometry was nearly identical; therefore, those dimensions and tolerances were used (Figure 4).
Figure 4 : Tensile Specimen dimensions from Journal of Rehabilitation Research & Development
With all of the dimensions needed to construct the part, the insert was modeled using SolidWorks. The idea was to model a filled in runner system and cavity and then subtract that out from the entire filled insert. The placement of the cavity and runners on the insert was determined by measuring the distance of the part from the edges of the plate using the shaded geometry and then comparing it to the known dimensions of the aluminum plate and insert. The following Figures 5 and 6 show the finished CAD models for the part.
Figure 5: Model of the filled runner system and cavity
Figure 6: Runner system and cavity subtracted out from the insert model
One design consideration implemented into the model was the addition of a small fillet around the edge of the tensile specimen, which was indented into the cavity (Figure 7). This was present on the parts created by the metal molds, but it could not be accurately measured. Having slightly rounded edges allows for easy extraction of the part from the cavity compared to parts with sharper edges. In addition to this, a draft angle of 2 degrees was implemented in the cavity to allow easy extraction of the part.
Figure 7: Fillet on side of tensile specimen
Before the initial printing and testing of the mold, the accuracy of the cavity and runner geometries were double checked. One of the features of SolidWorks allows to import an image as a sketch plane to be used to superimpose on the modeled part. The shaded geometry was uploaded to SolidWorks as the superimposed image to confirm accuracy of the modeled part (Figure 8). There were distinctly small differences in the cavity, and the cold slug well at the top was shallow, but all within tolerance. Overall the geometry looked trustworthy, and the first prototype was printed.
Figure 8:Modeled part superimposed on shaded geometry
1.2 Using FFF with PLA to Define the Geometry of the Insert
The first printing test was done to check if the geometry of the insert was good enough to fit in the aluminum mold, and to test if the placement of the cavity and runner system lined up with the fixed metal mold. For this, the accuracy of the cavity and the usage of multiple cycles was ignored, and the initial print was done with Polylactic Acid (PLA) through Fused Filament Fabrication (FFF). With this combination of material and printing, a cheap print was created that was sturdy enough to be clamped in the machine. After modifying the STL file to print at 20% infill, the cost of the print ended up being less than two dollars and a print time of 4 hours.
While testing out the insert geometry of the first print, it was noticed that it did not easily fit into the aluminum plate. The width and length of the print seemed to be fitting properly, but the insert did not fall neatly into the aluminum plate. Further inspection showed that it was the corners of the print that were overlapping with the aluminum plate; therefore, for future prints the radius of the rounded edges was increased (Fig 9). Since the goal was to test the insert in use, the edges and corners were sanded down evenly until the insert fit properly.
Figure 9: Radius of corner was increased
The first trial run of the first print was successful. The mold was hung from the right mold plate, and ran through a single automated run of the machine. After the aluminum plate was taken out, it was revealed that the material went through the runners correctly and filled up most of the cavity (Figure 10). The reason the cavity was not fully filled was due to the occurrence of flash through the entirety of the runner and cavity. The main reason why flash occurred is because the insert was not high enough; it was slightly sunken into the aluminum plate. This caused there to be a small gap between the insert and the fixed mold plate, which allowed material to leak through the parting line. The runner system for the insert was also slightly shorter than the runner system of the fixed mold plate, so a small adjustment was needed. Overall, the changes made to the model were increasing the radius of the insert edges, increasing the insert height, and increasing the length of our runners.
Figure 10: Print Trial 1, fist injection test
Another important concern stemmed from removing the injection molded part for the insert. When looking at the face of the insert, the runner and cavity were severely warped due to the heat of the material flowing through (Figure 11). Although our insert was tough enough to withstand the clamping force of the plates, it did not have the heat resistance to survive past the first test. While it was unfortunate that the insert had started to melt, the indent of the flow could easily be seen etched into our cavity.
Figure 11: Print Trial 1 with melted geometry
For the second print trial, the geometry of the inserted was still tested using PLA with a 20% infill, acknowledging it may melt. With the added changes to the insert edge radius, the insert easily fell into the cavity and stayed in fairly well. The changes to the insert’s height and slight runner shift also eliminated the flash that occurred compared to the first print trial (Figure 12). With this trial, two parts were printed before the melting of the insert occurred, and the the material started to flash through.
Figure 12: Print Trial 2 printing with no flash
With the mold geometry looking nearly perfect, an insert with greater infill was tested to determine if the infill correlated with the amount of runs. For the third print trial, the insert was printed using PLA with an 80% infill. Once again, the insert fit well in the aluminum mold and printed with minimal flash. The mold ended up melting still and the flash was likely due to the melted insert and rough curvature of the runner system. The surface for the FFF print is not ideal since it is extremely rough (Figure 13 and 14). For a more precise and smoother print, the method was switched to Stereolithography (SLA) with materials known to withstand higher temperatures. Satisfied with the insert, runner, and cavity geometry, SLA was used
Figure 13: Print Trial 3 close up of surface and melted spots
Figure 14: Print Trial 3 close up of molded part. Small bits of flash and rough surface finish
1.3 Using SLA with High Temp resin:
As described above, once the geometry of the insert was well defined and no flash from the runner and part was observed, then the switch to another additive manufacturing technique that was considered more suitable for the experimental environment was made. We excogitate that a very high accuracy and a smooth surface finish would be attained.
In this case, SLA was used due to its higher quality and smooth resolution compared to FFF. Another factor considered was the SLA resin. After doing some research and interchanging ideas with project group members, it was concluded that the SLA resin should be the high deflection temperature (HDT) resin, because it would allow for the use of higher mould temperatures during processing [3]. This resin has a HDT of 238?, with an elongation at break of 2.3%.
Initially, a clamping force of 80 KN was used, but the insert-cavity broke after the first attempt of using it. Therefore, the clamping force was lowered to 10 KN for the next print. Moreover, the elongation at break was not of focus for this trail because the focus was on abiding by the HDT requirements. A resin with a superb HDT will guarantee that no melting in the 3D printed insert-cavity will occur during the injection molding process.
Similarly, there were no problems observed after printing the part for the second time until run number 11. There was a clearly observed internal crack in the high temperature resin mould. The decision the change the material to increase the number of cycles of made after noticing that significant crack in the mould.
The reason the high temperature resin broke is due to the lack of elongation at break. There was an observed 2.3% elongation at break, compared with the tough material, which have 24% elongation at break [3]. The discrepancy between the two materials is easily noticeable. That was the first reason the decision to change the material to be test in the experiments was made. Using the high temperature resin, there was no observation of melting in the cavity; therefore, impact resistance was the biggest concern. Figure 15 shows how the two inserts made out of High Temp material cracked and failed.
Figure 15. High Temp inserts
1.4 Using SLA with Tough resin:
Once again, SLA was chosen for the trial 6 runs; however, the tough SLA resin was used. Tough resin was considered, because it was predicted to last through more cycles than the HDT resin previously used, due to its better mechanical properties, and still maintain some thermal properties. The tough resin material was printed with supports that were removed hours after the print was completed. After the removal, some sanding was needed to be able to insert the mould in the aluminum plate.
The clamping force used in trial 6 was a range between 10-15 KN with a cooling time of 30 seconds. The importance of this trial was to test how long the tough resin part would last compared to the HDT resin.
For this trial, many of the injected parts yielded significant amounts of flash (Fig 17). This was caused by over sanding after removing the mould from the support structure, and material leakage through the cavity. As seen in a previous trial, these factors caused the insert to be unaligned with the metal mould’s, and slightly lower in the aluminum plate.
At run number 20, there was a noticeable hole that went straight through the cold slug well of the mould (Fig 18). It was concluded that this hole was a result of lack of depth in the mould. For the next trial, an updated CAD stl was used with deeper tensile specimen and runner geometries in the mould. Although the part broke at just run 20, the mechanical and thermal properties of the mould were better than all the previous trials. Figure 16 shows how the two inserts made out of High Temp material cracked and failed. 
Figure 16. Tough inserts
2. Simulation:
For the initial simulations, a control simulation was used were the entire mold was made out of P20 steel, using the automated parameters Moldex3D. The simulations of the mold insert were compared to the control simulation. The mesh control are shown below. Figure 17 illustrates a simple mold made entirely of one material, and an insert that surrounds the part at half of its z-axis.
Figure 17. Left: metal mold. Right: polymer insert
The material used for the mold insert was PMMA. Moldex3D only allows the user to use a metal as a mold insert, but it does allow the creation of your own materials. Therefore, the properties of PMMA were manually imputed into the software as a sold onto the mold cavity in the simulation. The following parameters were used for both the simulation with a mold insert and without a mold insert.
| Maximum Injection Pressure | 250 MPa |
| Maximum Packing Pressure | 250 MPa |
| Filling Time | 0.15 Seconds |
| Packing Time | 4.27 Seconds |
| Eject Temperature | 90 °C |
| Cooling Time |
12.9 Seconds |
The results conclude that the major problem which will be encountered is cooling. The thermal conductivity of the PMMA is around 20 times less than that of P20 Steel, which is used for injection molding. A lower thermal conductivity does not allow the part to cool efficiently (Figure 18). which shows the temperature of each part after cooling time of the part without the insert and with the mold insert respectively.
Figure 18: Left: Temperature after cooling without insert . Right: Temperature after cooling with insert
It is clearly seen that the difference in temperature occurs where the mold insert begins. For the non-insert part, the exterior temperature was uniform with a temperature close to 40 °C; it took 19.03 seconds to reach ejection temperature, and had a max temperature of 105 °C after cooling time. For the part that used the mold insert, the temperature was not uniform, showing half of the part with temperatures in the 120 °C-140 °C range and the other half having a temperature close to 40 °C. It took 33.4 seconds to reach ejection temperature, and had a max temperature of 264 °C after cooling time. The effect that all of these results have on the part, are that they all lead to an increase in warpage. While both parts had significant warpage, the amount of warpage experienced by the part made with the mold insert was 10 times larger than that of the control part (Figure 19).
Figure 19: Left: Warpage of control simulation. Right: Warpage using mold insert
The major challenge in modeling this process was to imitate all the aspects that come from having a mold cavity made from plastic onto the simulation. An additional challenge was the cooling channels of the mold could not be seen visually; thus, the cooling channels were updated to represent the actual existing channels. Furthermore, the simulation will be used to determine a process that can output a part of the same quality as the control process using the mold insert.
3. Summary and Future work:
Different materials and additive manufacturing processes were studied. First, the design of the cavity and runners were tested with an FFF insert. Since the material used here was a thermoplastic (PLA), the insert started to melt. To improve the thermal properties we moved to SLA. The first material used in SLA was a High Temp resin which showed better thermal properties but the mechanical properties were not a fit for this case. Finally, we tried a Tough resin which displayed excellent mechanical and thermal properties. In addition to this, multiple clamping force and cooling time were put to test. Table 4.1 summarizes the results obtain.
| Trial | Material and Process | Number of Cycles Before Failure | Clamping Force
(KN) |
Cooling Time (sec) | Observation |
| 1 | PLA – FFF
20% infill |
1 | 20 | 20 | Runner dimensions were a little bit off. Therefore, we had to adjust the runner. |
| 2 | PLA – FFF
20% infill |
1 | 20 | 20 | The runner and gate melted. |
| 3 | PLA – FFF
80% infill |
3 | 20 | 20 | The melted polymer penetrated the cavity. |
| 4 | High Temp – SLA | 1 | 80 | 20 | It broke right away. |
| 5 | High Temp – SLA | 9 | 10 | 20-30 | In the 9th cycle time, we noticed an internal crack. |
| 6 | Tough – SLA | 20 | 10-15 | 30 | In the 20th run, we saw a hole through the cold slug well. |
| 7 | Tough – SLA | 40 | 10-15-20-25 | 30-60 | In the cycle 40th we noticed a crack and in the next cycle broke |
As seen in the table above, the Tough resin showed the best results. Therefore a more detailed study of different injection molding processing parameters will be done using this insert. Future work for this project will include:
- Testing a different material for the insert: Digital ABS manufactured by Stratasys using PolyJet technology
- Characterizing the insert to have a more accurate simulation in Moldex3D
- Improving the cooling of the inserts by coating the insert with CPU Silver Paste
- Studying the cooling in more detail looking for ways to improve it
- Studying the properties of the part being injected such as shrinkage and mechanical properties
4. References
[1] Gerschutz, Maria J., et al. “Journal of Rehabilitation Research & Development (JRRD).” Tensile Strength and Impact Resistance Properties of Materials Used in Prosthetic Check Sockets, Copolymer Sockets, and Definitive Laminated Sockets, 2011, www.rehab.research.va.gov/jour/11/488/page987.html.
[2] Stratasys. “ Technical Application Guide: PolyJet for Injection Molding”. Retrieved April 4th, 2019 from http://usglobalimages.stratasys.com/Main/Files/Technical%20Application%20Guides_TAG/TAG_PJ_InjectionMolding.pdf?v=635923370695739650.
[3] “Functional Prototyping Materials for Engineers.” Formlabs, http://formlabs.com/materials/engineering/#high-temp-resin.