Mold Imbalance Goes With the Flow...Literally

Almost any molder who has used an eight-cavity mold has probably run into this predicament: parts molded in the inside cavities are always heavier than parts produced in the outside cavities. Conventional wisdom for years has blamed it on a higher heat load from the plastic occurring in the center of the mold. The inside cavities stay hotter longer, decreasing viscosity and packing more resin into the part. The other culprit, it's been said, is mold deflection.

And for a long time John Beaumont believed the conventional wisdom. Beaumont is an assistant professor in plastics engineering technology at Penn State at its Erie campus. Prior to that, he put in his time at a variety of small and mid-size molding shops before working as the technical manager for Moldflow's U.S. operations.

Not long ago, Beaumont was working with a molder running an eight-cavity mold processing a self-compounded, magnetized material for parts with extremely tight weight specs. Despite good, direct-insert cooling, parts from the inner cavities were routinely heavier than parts from the outside. In fact, the parts were out of spec, and no amount of process fudging could bring them into balance.

"I said, 'Let's solve this problem by creating separate cooling circuits,'" says Beaumont. So they did. In fact, the molder went so far as to chill the inside cavities, but run with no mold temperature control at all on the outside cavities. Still, the parts were out of balance and out of spec. "It forced me," Beaumont says, "to stop and really think about what the real problem was."

The solution to the real problem came to Beaumont in the car during a long holiday road trip from Massachusetts. "It was one of those deals where you're just staring at the horizon ahead of you, and it hits you," he says. He played with the concept in his mind, and when he got home he modeled his idea in clay. "Sure enough," he says, "it seemed to make sense."

The Theory
The "real" problem, he theorized, has nothing to do with mold temperature control. It has a lot to do, however, with the distribution of shear across the melt as it moves through the runner. The concept, put simply, is this: Viscosity of plastic is affected by temperature and shear. In a runner, shear is greatest next to the outer wall, generating higher temperatures and decreasing the viscosity of the melt. Material in the center of the flow, however, experiences less shear and, therefore, is cooler and more viscous.

The profile of the melt looks like a doughnut, with a ring of high-temperature, low-viscosity material surrounding a core of low-temperature, more viscous material. Beaumont theorized that when this flow hits a runner intersection, it essentially splits itself down the middle. The low-viscosity, high-temperature material tends to follow the inside wall around the corner (Figure 1), while the high-viscosity, low-temperature material tends to follow the outside wall.

The new melt profile consists of half high-flow material on the inside wall and half low-flow material on the outside wall. In a four-cavity mold with a balanced "H" runner system, this isn't a problem, since there's only one intersection. Each runner contains material with flow properties in the same proportion.

Consider, however, an eight-cavity (or more) mold that has more than one intersection. At those second and third intersections, the flow splits again, sending the high-temperature, low-viscosity material to the inside cavities, and the low-temperature, high-viscosity material to the outside cavities (Figure 2). Call it shear segregation. Beaumont thought that if this was correct, the inside cavities would fill sooner than the outside cavities, resulting in a generally heavier part with associated variations in shrinkage and dimensions.

Proving the Point
Before running the tests that would ultimately prove his hypothesis, Beaumont persuaded Jack Young, an associate professor in mechanical engineering on campus, to attempt to model the phenomenon using Fidap (Fluid Dynamics Analysis Package) flow simulation software made by Fluent Inc. (Lebanon, NH). It's Beaumont's contention that current mold filling analysis programs do not model such behavior. Beaumont says that "for practical reasons, the software developers have made simplified assumptions in their modeling in order to reduce modeling requirements and gain speed. These are very practical concerns, and in most cases, these simplified assumptions work well."

Young and a student of his spent the summer of 1997 developing the programming and runner models required to mimic the flow behaviors Beaumont thought he was dealing with. When he was through, Young approached Beaumont with data that, at the time, seemed in error. "He came to me and said, 'Gee whiz, John, according to Fidap, about 70 percent of the flow is going to the inside cavities,'" Beaumont recalls. The variation, he thought, would be more subtle than that. He was wrong.

To find out for sure what was happening, Beaumont set up an experiment molding a variety of materials through two different runner systems. He also tested the solution, the one that came to him in the car. A modular test mold was built for the test. It is a four-cavity mold, but the runner system is designed to replicate the flow conditions of an eight-cavity mold. The cavities themselves are referred to as "purge molds." They're oversized, rectangular cavities that allow the melt to flow freely through the runners while minimizing flow resistance from the cavities themselves. This was done to isolate the effect of the runner and simulate steady flow conditions. Beaumont also tested two runner diameters: .125 inch and .25 inch.

Six different materials were used in the tests: ABS, acrylic, nylon, PBT, LDPE, and PP. All of the materials were run at five fill rates: 5, 3, 1.5, .5, and .25 seconds. The shot size was adjusted so that none of the cavities would fill completely. Doing this permitted determination of flow into the cavity prior to pack and hold. Molded parts were weighed and recorded.

Beaumont discovered that the mass fraction of material going to the inner cavities was greater than 50 percent for all the materials under all conditions (Figure 3), except for the polyolefins (polypropylene and LDPE). The mass fraction going to the inner cavities dropped to just under 50 percent at the slowest flow rate for polypropylene and LDPE. Remember, ideally and theoretically, each cavity should receive half the material flowing to its side of the mold. What Beaumont found was that the Fidap simulations were more correct than he thought. With the .125-inch-diameter runner, the mass fraction of flow to the inner cavities ranged from 60 to 95 percent, for an average of 65.2 percent. That number improved somewhat with the .25-inch diameter runner, dropping to an average of 56.8 percent.

So, it appeared that Beaumont had found the smoking gun. Flow imbalance in geometrically balanced runner systems is a result of nonsymmetrical shear distribution within the melt as it travels through the runner system. Now for the solution.

The Runner Flipper
Whatever the solution was, Beaumont knew it had to be located within the runner system, it had to be ejectable, and it couldn't precipitate injection pressure loss. "I expected the solution to be very complicated," he says. What he came up with is relatively uncomplicated, and it meets the criteria he set out. He calls it the Runner Flipper. It's a runner insert that takes the melt through a series of dips, twists, and turns to reorient the shear distribution (Figure 4). "It rolls the melt stream 90°; that's what it really does," says Beaumont. "I probably should have called it the Runner Roller."

The idea is to turn the melt stream so that the high-temperature, low-viscosity material rests on the bottom half of the runner, with the cooler, high-viscosity material on top. That way, when the material hits an intersection and splits down the middle, the mix of high- and low-viscosity melt remains proportional. Flow, then, remains virtually constant through each tertiary runner as the material enters each cavity. Although he's not yet conducted tests for pressure loss, Beaumont says the Runner Flipper does nothing to restrict flow. "In the runner system, I can't believe I'd increase pressure more than 2 or 5 percent," he says.

To test his device, Beaumont ran the same materials again through the .125-inch runner system with the Runner Flipper inserts installed at the intersections. The Runner Flipper, according to his results, virtually eliminates the imbalance at all flow rates. Without the Runner Flipper, the mass fraction of flow to the inner cavities averaged 65.2 percent. With the Runner Flipper, the flow to the inner cavities ranged from 49 to 52.9 percent, for a 50.2 average--almost ideal.

The Next Step
Beaumont is in the early stages of manufacturing and marketing his invention. Clearly, he admits, the Runner Flipper is not for every molder, but he hopes to see it become a standard in runner design. He imagines high-precision molders making weight- and dimension-sensitive parts will benefit most from what the flipper has to offer. "Now," Beaumont says, "an eight-cavity mold can perform just like a four-cavity one." In fact, Beaumont believes the imbalance phenomenon was forcing many molders to mold with four cavities exclusively, using smaller presses in the process.

He says he wants to make licensing of the Runner Flipper as simple as possible, and costs will probably range from about $1000 to $2000 per mold. Licensing right now is handled by Beaumont through an on-campus business he's established. The business will provide mold design analysis, mold filling analysis, insert design and supply, and licensing on a per-mold basis.

For larger scale plantwide applications, Beaumont will direct potential customers to Penn State's intellectual properties office. Further details on Beaumont's experiment and solution are being presented as a paper at ANTEC '98, April 26-30 in Atlanta.