Variable Machine Conditions Produce Critical Medical Part With Consistent Quality

September 08, 1998

A few years ago, injection molder DTM Products in Niwot, CO inherited a one-cavity mold for a 4-liter polycarbonate canister used to filter blood during heart bypass operations. Other molders who'd had the tool before DTM struggled mightily to push viscous polycarbonate to the end of the cavity through the canister's long and relatively thin walls. The result: vacuum voids at the end of fill, producing astonishing reject rates of 18 to 22 percent. Such vacuum voids between the canister rim and the lid at the joint lead to welding problems that might not be airtight, says Carson Payne, vice president, operations.

Complicating matters was the fact that the tool had a programmable valve gate retrofitted to eliminate vestige on the canister. The resulting pressure loss made it that much harder to fill the tool. "That put us directly into a pressure limited situation," Payne reports. "It limited the viscosity of the resin we could use to a very narrow range of melt flow index."

With such a sensitive part requiring precision molding, DTM decided to try to optimize the part using cavity-pressure sensing. To prove out the technique, DTM ran a trial with the mold in a 400-ton press using two processes. The first process assessed part quality under constant machine conditions by transferring from first-stage fill to second-stage pack and hold on screw position. This process used "traditional" molding techniques where optimum machine conditions are established for the tool using a fixed hydraulic profile.

The second process transferred from the first-stage fill to second-stage pack and hold based on cavity pressure under variable machine conditions where the hydraulic profile was allowed to vary to meet the set cavity pressure. The goal was to let the machine adjust itself to produce optimum part conditions. And DTM worked with its material supplier to receive grades of polycarbonate at or above a certain melt flow index value. This was necessary to prevent large variations in material viscosity batch-to-batch and help to ensure full shots each run.

Figure 1. Nozzle pressure at transfer, constant machine conditions.

To start, a strain gauge load cell pressure transducer was retrofitted into the mold, about 3.5 inches from the gate, which is at the bottom of the canister, and 9 inches from the end of fill. Five lots of clear polycarbonate were selected with varying melt viscosity ranges within the agreed-upon tolerance. In the first 220 shots of the trial, the five materials were fed in consecutive order while the peak hydraulic pressure at transfer from first to second stage was set at a constant 21,200 psi. Under these constant machine conditions, the peak cavity pressure was measured for all 220 shots.

The hydraulic pressure curve for the trial (Figure 1) shows that the press held a steady line at about 21,250 psi. In the mold, however, peak cavity pressure fluctuated from a peak of about 8000 psi down to about 4250 psi as material viscosity changed from lot to lot (Figure 2, p. 92). The result of the trial was a 15 percent reject rate. And even parts that looked OK lacked consistency. "They may look similar," says Payne, "but they're actually packed at different pressures." This became apparent during Emabond welding of the lid to the canister, where the high heat remelt temperature stress relieves the canister lid joint.

Figure 2. Peak cavity pressure, constant machine conditions.

Next, DTM used the same five lots of material to run another 220 shots. This time transfer from first-stage fill to second-stage pack and hold was triggered based on cavity pressure. Using scientific molding methods and traditional screw position transfer, the ideal peak cavity pressure was established at 5900 psi. The switchover cavity pressure needed to maintain that 5900-psi peak was determined to be 5280 psi. By basing the material processing on conditions in the mold, it would become the machine's responsibility to adjust itself to meet that 5280 psi.

The results from the second process were promising. Unlike the first process where melt pressure held steady, in the second process it varied 1200 psi, ranging from 20,800 psi to 22,000 psi (Figure 3). Peak cavity pressure, on the other hand, remained closer to the targeted 5900 psi, with some variation down to about 5600 psi (Figure 4). More important, the reject rate on the canister dropped from 15 percent to 2 percent, saving DTM about $80,000 a year on material and machine time.

Figure 3. Nozzle pressure at transfer, variable machine conditions.

Payne acknowledges that the inconsistent hydraulic pressure required to produce consistent parts batch-to-batch runs counter to everything molders have learned during the past 30 years. Conventional wisdom says that to mold quality parts, there must be constant machine conditions. "It's very unnerving to see a machine with peak hydraulic pressure all over the map," Payne notes. "And when you tell machine manufacturers that you want to transfer on cavity pressure, they say, 'OK, but it will make my machine's SQC chart look out of control.'" But it doesn't matter how inconsistent your machine's hydraulic pressure is, as long as it produces good parts with a consistent cavity pressure batch-to-batch.

Similarly, traditional-method molders balk at the cost required to achieve consistent cavity pressure shot-to-shot. Payne says the cost to retrofit one press can run as high as $10,000. And although such optimization is not required for every mold, converting an entire plant to cavity-pressure sensing can get expensive.

Figure 4. Peak cavity pressure, variable machine conditions.

By transferring on cavity pressure you produce less start-up scrap, more good parts, and consume less energy. "The question is," says Payne, "which blood filter would you rather have used in your heart operation? One molded using traditional, fixed-machine conditions or one molded using variable machine conditions?"

Comments (0)

Please log in or to post comments.
  • Oldest First
  • Newest First
Loading Comments...