The Project: The sealing of O-rings inside of a component to a blood diagnostic device was resulting in leaks.
The Solution: A simulation service was used to determine how to best make the necessary modifications for a successful result.
Design and development of a medical device is an exacting process, to say the least, and it assumes even more complexity because it almost always extends across years that encompass multiple iterations of the device. As a result, small things that were satisfactory in the prototype and first production run can grow to be problems that must be solved as production volume rises.
That is the situation that the medical contract manufacturer Medbio Inc. faced with a blood diagnostic device when production volume was ramping up. Naturally, that was good news for Medbio’s client, but it also meant the time had come to take care of what had been a minor leaking problem in sealing O-rings. Medbio took advantage of plastics MoldFlow simulation services provided by CAE Services Corp. to solve the problem.
Medbio began work on this device with its client about seven years ago when it was still a lab product. The diagnostic device is a platform that performs a wide variety of analyses on blood specimens contained in a molded cartridge that is inserted into the main device. The O-rings in question were used to seal the channels in the cartridge. As part of the client’s original concept, they were molded of EPDM [ethylene propylene diene monomer (synthetic rubber)].
There were 21 of the 3.5-mm O-rings used per cartridge, which are individually attached with cyanoacrylate adhesive to the molded body of the cartridge. Medbio’s engineering manager Joe Szyperski said, “There was tremendous variability in the EPDM O-rings. For instance, they could be either too dry or too sticky, which meant that the valve plate could not move as designed. Often, it would bind, which was obviously unacceptable.” Sometimes, an entire lot of the rings was unusable.
At one point, Medbio began molding the O-rings in-house using a TPE (thermoplastic elastomer) material it recommended to replace the EPDM. Szyperski said Medbio deepened its molding skill molding the rings and was able to overcome the surface defects that were problematic in both assembly and product functionality. However, the company knew that this was only an interim step to an optimized solution.
Founded in 2004, Medbio is relatively young as a company, however, each of its founders and top managers is long on specific experience in medical production. Szyperski, for example, has over 20 years of engineering and manufacturing experience, almost all of it working with medical products and components for products such as defibrillators and cardiac surgical devices. Szyperski’s last 11 years have centered on manufacturing and assembly of injection-molded medical device components such as the cartridge in this case.
That experience enabled Medbio to recommend to its customer that the O-ring seals be overmolded onto the body of the cartridge, which Medbio was already molding. Szyperski said the client quickly understood that eliminating the assembly of 21 individual ring seals made economic sense and would eliminate many potential quality issues. Tooling was ordered and Medbio evaluated a number of materials that would be better suited to solid adhesion between the overmolding and the cartridge body. The client decided on one—another TPE.
Shortly after the overmolding started, said Szyperski, a problem arose. It was a slight variability in production, but big enough that very small leaks were being detected during leak testing. The leaks, which were localized in the end-of-fill area for the sealing rings, were creating some fairly high fallout in some lots. Even under magnification, it was hard to really see the problem; the rings were very small and the material was transparent. But finally, the inspection revealed knit lines in the end-of-fill area that would cause a crease in the material and allow the leaks.
Szyperski describes the molding process as very sensitive, to the point where Medbio had to ask its molding machine supplier—Arburg—to change the settings so that molding could be done using injection pressures that are well below normal. Szyperski points out that all of these cavities are instrumented with Kistler pressure transducers—the one-millimeter, flush-mount type. “We are able to drive a very consistent process, but even so, there was variability in the way the cavity filled because we are overmolding both sides of the part at the same time.”
There was a gasket being overmolded on one side of the cartridge body while TPE material was being run through the cartridge body to overmold the gasket on the opposite side. Szyperski explains, “Whether it was subtle changes in the material lot or in temperatures, for some reason, some runs would fill better than others. We had variability, so we started looking at how we could process this differently. We added overflow channels thinking maybe we’re just not getting enough material through.”
The first question an experienced molder or mold maker would ask is, “Why are those overflow channels going out at a 90º angle from the flow?” Szyrpeski says, “If we had the real estate, we would've gone straight through. It’s the optimum solution, but there is a barrier there—a raised surface on the part being overmolded—so we put the overflow at a 90-degree angle.” As can be seen in Figure 1, that overflow channel design did not solve the problem.
As mentioned previously, medical devices are developed over time. When the cartridge was first designed, the O-rings were mounted on the molded cartridge body individually. Space for an overflow channel was not even a consideration. In fact, the space between the end-of-fill cavity and the barrier Szyperski mentions was about 0.0035 inch, not even close to what’s necessary for material overflow (Figure 2).
It was at this point that Medbio knew they wanted a moldfilling simulation and analysis, specifically a Moldflow analysis from their long-term supplier CAE Services. Szyperski explains that there are times when a part is different enough or has some aspects that concern them. They will then go to CAE Services and ask for input on best gate locations.
“CAE’s engineers can simulate different gate locations and we can see what will work best. Moldflow is a good tool to use for a variety of things. It’s helpful to see what will happen if you put the gate here versus another location.”
The first thing CAE generally does with a problem-type mold is to run a baseline simulation that determines what’s going on with the current design. The company’s engineers want to see if the analysis predicts the same kind of problem that is actually occurring. For the overmolded O-ring, CAE incorporated the existing part and tool design, along with Medbio’s process settings into a simulation model, then ran an analysis to see if anything was identified that would red-flag a leak-type issue.
All Moldflow analyses are finite element analyses. The part geometry in this case necessitated a true, 3D finite element analysis, as contrasted to the traditional thin-shell, midplane type of mesh. The latter type can be used for thin-wall parts, but when the parts are “thick and chunky,” the true 3D model is needed. Even though the cartridge was small in size, it fit the “thick and chunky” description.
CAE’s baseline study confirmed the problem. The air trap visible in Figure 1 was well predicted by the software; it made sense to CAE’s analysts and it was in the area where Medbio said the problem was located. The next step was to determine what could be done to solve the problem within the constraints of the mold design and/or the part design, and to present alternatives.
For CAE, it was clear that, due to part geometry, moving gate locations would not solve the problem. Its engineers focused on changing the design and location of the overflow well, moving the entrance to the well closer to the air trap, and routing the overflow channel to avoid that wall feature of the cartridge (Figure 3).
Szyperski said that Medbio, the toolmaker, and the customer looked at the alternatives, made their decision, and within one to two days, had a proposal that CAE could analyze. The total time needed to find the solution was about three to five days. Once CAE had all the information, the meshing process was under a day. Additionally, results are typically running within one or two days.
Of course, what mattered to Medbio and Joe Szyperski are results, and those were excellent. “It pretty much cleared out all the problems in that area of the mold, which was about 90% of the total issues with the mold. Remaining issues are in other areas.”
Szyperski stresses that Medbio does not want to live with a marginal process, and will change the process, the tool, and if need be, even the part design, though that is a last resort. When CAE’s analysts have a Moldflow simulation that reveals some possible changes that will result in a more robust or larger processing windows, they will report it.