November 1, 2001
This series of articles is designed to help molders understand how a few analytical tools can help diagnose a part failure. Michael Sepe is our analyst and author. He is the technical director at Dickten & Masch Mfg., a molder of thermoset and thermoplastic materials in Nashotah, WI. Mike has provided analytical services to material suppliers, molders, and end users for 15-plus years.
A lifetime ago I sat in my first Glenn Beall seminar. Glenn has always been that proverbial voice in the wilderness calling for what ultimately ends up being the most efficient of all approaches, making it right while things are still on paper or the tube.
On this particular day, Glenn was making a point about the different scenarios under which tool changes are made and how much they can cost. The slides that still stick with me showed the contrast provided by the scale of the project at hand. One slide showed a very small mold that could be removed by hand from the press and broken down on the bench in a matter of minutes. The next slide showed a man standing next to a huge molding machine. His head barely reached the lower tiebar. In the press was a massive tool that would take hours to pull, would likely need to be transported to a dedicated facility to be disassembled, and would be out of service for days for even the most minor of revisions or maintenance.
Many years later the case for smart mold design is at least understood and accepted if not universally practiced. The idea of material analysis as part of the product development process is a decidedly less popular notion. This arises largely from the fact that analytical testing services come from outside the molding and moldmaking facility and are therefore a visible cost.
Machine time, while of essential value to a profitable molding operation, is not thought of in the same way because the molding firm does not have to write a check at the end of the day to pay for the machine time it uses. It does, of course, have to cover the costs associated with running that machine. But psychologically this is absorbed into the overall fabric of the operation. Perhaps if processors actually did buy machine time the same way they buy outside services like testing, the true cost of trying to solve a material problem solely by making press adjustments would be more apparent.
Troubleshooting at the Press
We can, however, look back after a problem is solved and account for the resources expended to come to a conclusion regarding the best strategy for solving a difficult processing problem. The following case study will attempt to capture the costs related to a particularly troublesome program launch and contrast those costs with the dollars spent in the lab to actually find the root cause of the problem.
This particular program involved four molds; we will focus on one of the four for which the best history can be assembled. All of the parts were large, running in presses ranging from 1000 to 3000 tons. The largest of the parts, and the one we will focus on, weighed almost 9 lb and was produced in a 90-second cycle, or 40 shots/hr.
The mold, like all the others in the program, used a hot runner system with multiple gates. The material was a dark gray polycarbonate with a nominal melt flow rate of 22 g/10 min that, due to the considerable volume of the project, was priced competitively at $1.50/lb. The problem on initial startup was an apparently incurable splay.
Material analysis should be part of the development process.
When large parts running in tools of such complexity develop problems, the possible causes that need to be evaluated seem endless. The manifold system had more than 20 different zones of temperature control. Overheating in any one of them could cause the type of problem that the processor was seeing. The drying unit is always cause for concern. In recent years some processors have put moisture monitors in place in an attempt to gain an increased level of understanding about the drying process. However, most of these systems are simple loss-in-weight devices that can capture anything that might boil off of the material at the chosen test temperature.
In this case a moisture monitor was being used but the results were ambiguous. Regardless of what readings the instrument gave, the splay problem did not abate. As is almost always the case, the backdrop for this problem-solving exercise contained a customer that was an ever-decreasing number of days away from a critical product launch. In such cases, action is usually favored over logic. Consequently, a team of people with substantial skills in processing and tooling were working almost around the clock on the problem.
Step 2: Material Analysis
After about a week of at-the-press troubleshooting, one of the members of the team, who had a background in the tools of analysis, sent us several samples. Among these were a sealed jar of raw material that came directly from the resin dryer, a larger sample of raw material for a melt flow rate test, and a very large molded part showing the cosmetic problem.
An assessment of the moisture content of the material from the dryer using a chemical method known as Karl-Fischer titration showed that it was, in fact, dry. This was a surprise to everyone since the actual metrics by which a dryer is often judged, such as dewpoint and airflow rate, were suspect. Nevertheless, the moisture content of the material was just less than .019 percent, below the required .02 percent.
But the melt flow rate comparison of the raw material to the molded part clearly showed that the material was being degraded during the molding process. The melt flow rate was rising from 26.8 g/10 min in the pellets to 43.7 g/10 min in the molded partâ€”an increase of 63 percent. This is outside the boundaries of good molecular weight retention and usually points to a processing problem. Impact tests on coupons cut from the part showed a tendency for brittle failure that would not be expected from polycarbonate. But was it the process?
The results of the melt flow test shifted attention away from the dryer and towards the melt temperature control. When a hot runner system is involved this is not a trivial investigation. The screw, barrel, and nozzle system must be assessed. But even after these are given a clean bill of health, the hot runner itself can be a problem. In molds of this size, pulling the mold and putting it on a bench for disassembly and inspection of the paths in the hot runner is a mammoth undertaking. Pulling the screw from the barrel is likewise a task of Herculean proportion. But a thorough inspection of the entire melt handling system turned up only a questionable screw design.
Breakthrough at a New Location
In desperation, the mold was moved to a different facility. Different machine, different drying system, and different screw design produced the same result and more lost time and expenditure of talent. Then a breakthroughâ€”the same grade of polycarbonate compounded in a lighter shade of gray was processed and ran flawlessly. In fact, attempts to upset the system and provoke the splay problem could not produce a bad part. As a scientist these are the moments you live for. It seemed almost certain at this point that it was a material problem, but we were still a long way from establishing the exact nature of the problem.
The first test we wanted to perform, and ultimately the only one that was needed, was a melt stability test. This is a variation on the standard melt flow rate test where the melt flow rate is evaluated using a standard preheat time (usually 5 to 6 minutes). The test is then repeated using a longer preheat time. The purpose of the test is to assess the thermal stability of the material. It is a test we discussed about a year ago with respect to some problems that arose with specific colors of polycarbonate that could not be controlled during the molding process.
Table 1. Melt stability results, g/10 min
Table 1 shows the results of the melt stability test on the light and dark gray materials. The light gray material behaved in a manner consistent with historical data on polycarbonate: The shift was very slight as a function of the extended exposure time. The dark gray material shifted by 63 percentâ€”as much as it had during the molding process. In other words, the dark gray material lacked the inherent thermal stability needed to survive the injection molding process.
The results on this second sample of raw material suggested that the problem also affected the raw material. The natural polycarbonate was a grade with a nominal melt flow rate of 22 g/10 min. A typical range of results around such a nominal result would be 18 to 26. The first sample of dark gray had been more than 26 but not far enough to raise genuine concern. But the second sample was near 31, indicating that the process control over the color compounding step was showing signs of the difficulty that would become manifest during molding.
Once the problem was understood it was quickly traced to a stabilizer package that was supposed to be incorporated during color compounding but had been missed in the recipe. The problem was uncovered with less than a week remaining in the product launch schedule and provided the compounder with just enough time to make a new lot of material with the appropriate chemistry and ship it to the molder so that the molder could deliver parts for the first product build.
This problem-solving process had huge implications for the success of the program. Starting with a variety of possible causes for the problem, it was narrowed down to the raw material with less than two days in the analysis lab at a cost of less than $1500. Most people, when they consider this kind of a cost associated with testing, do what most of the people reading this article just didâ€”they cringe. Testing costs are seen as an expense with no return.
But consider the costs that were incurred in trying to do it all by the press. Even if we consider just this one mold, the processing and equipment assessments that were being performed tied up almost two weeks of machine time. The exact rate on a 3000-ton machine obviously varies, but if we assume a revenue-generating capability of $250/hr and estimate that the molder might have used half of that machine time to generate revenue on other work, the cost of the machine time alone was $42,000. It probably cost more to pull and set the mold once than it did to perform the tests.
Then there is the material. When a material like polycarbonate undergoes the type of change in molecular weight that our tests showed, it cannot simply be reground and put back into the process. The damage is irreversible and the material value is lost. For an estimated 10 hours a day over a period of two weeks during the troubleshooting process, the largest mold in the program was processing 360 lb/hr at $1.50/lb or $540.00/hr. The estimated cost in spoiled material was $64,000.
And of course there is the human cost of assembling the type of team required to work on such a project under a tight timeline. Five people from various disciplines were all putting everything else that they normally do during the day on hold. The direct expense alone was estimated at $11,500, but anyone who has been involved in such an 11th hour effort knows that the real cost of putting everything else aside to put out this type of fire is actually much higher.
Add in the incidental costs of shipping huge molds between locations and resetting them, and it is not a stretch to put the total cost of the effort on the shop floor at $125,000. And that covers just one of the four molds! Conservatively the total program costs could have easily run to $250,000. If the material testing could have prevented even half of the at-the-press troubleshooting, then the testing cost is approximately one cent for every dollar spent in traditional shop floor problem solving. Most people in business would gladly allocate money on an activity where they could save $100 for every dollar they spent.
And that is from the processor's perspective. At the customer level the consequences of failure are always much higher. This particular customer placed the value of the product being rolled out at $60 million. Taken to that scale, each dollar of testing was the leverage for $40,000 in product sold! The reluctance to stop molding parts comes from the notion that as long as someone is pushing buttons on the molding machine, progress is being made.
In this case, and in many others like it, the problem cannot be solved at the press. This is not to discount the value of information gathered in the shop. If the contrast in processing between the light and dark gray materials had not been noted, the analytical investigation would have taken longer and been more expensive. And that is exactly the point. Testing is not done in a vacuum and it does not replace practical experience on the floor. It is part of it.
You May Also Like