Software helps qualify new molds fast.

Getting good parts out of a new injection mold can take time - sometimes days of trial-and-error molding followed by weeks of re-cutting steel. Even a more systematic approach to tool qualification can easily bog you down with complex statistical exercises. To make the whole qualification process faster and easier, GE Plastics, Pittsfield, Mass., recently developed "QTIP."

Short for Quick Tool Introduction Program, QTIP is a package of proprietary software and related process monitoring know-how that together help molders identify the relationship between molding conditions and critical part characteristics. "QTIP creates a unique fingerprint of a mold's optimal processing conditions without regard to the machine that mold will run in," says program leader Tom O'Connor. With this "fingerprint" in hand, molders can recreate those optimal conditions on whatever machine ends up running the job.

Using terminology more familiar to molders than statisticians, QTIP's proprietary software first walks users through design-of-experiment (DOE) trials. The same software also helps users optimize process conditions once the tool has been proven capable of meeting part-quality requirements. QTIP also includes guidance in how to apply process-monitoring tools to tool-qualification tasks.

Automated DOE

Running on just about any Windows-based PC, QTIP can set up either "full factorial" or "central composite" experimental designs. Users need only enter the process variable and part-quality attributes to be examined. The program then displays a suitable DOE matrix. In all, users have to navigate only three drop-down menu categories to use the software.

Unlike the many generic DOE programs available from software companies, the GE package uses common molding terms. "We've put everything in molding rather than statistical language," says O'Connor. "The program doesn't ask for 'factors' and 'responses.' It asks for 'process conditions' and 'customer requirements.'"

Easy optimization

Once the DOE has been set up and performed to qualify the tool, QTIP software then generates a "process characterization model." Better known as a processing window, this model graphically relates key process variables to one another and to the part-quality requirements. This model, in turn, enables users to optimize their processes so as to balance the inevitable trade-offs in a satisfactory manner. "You can experiment on the computer, rather than on the press," O'Connor says.

According to O'Connor, QTIP helps users optimize their process in two ways: First, it can display complex process-characterization models as a simple, color-coded graphic, making clear all the interactions between process variables and quality. Second, the software can recommend optimization strategies based on user-specified criteria. "You can find the optimal point based on Cpk, ppm, or yield," he says.

The current QTIP software handles up to 11 process-variable inputs and any number of inputs that capture the customer's quality requirements. In test applications for GE customers, QTIP typically required only three or four inputs to optimize the process.

O'Connor notes that process optimization can target a wide range of customer-specified quality requirements - not just the obvious dimensional attributes. In one test application, GE used QTIP to maximize a pipe manifold's burst strength by identifying the best mold temperature, injection time, and packing pressure.

QTIP may also aid in machine selection. For example, "You might not need to run a tight-tolerance job on your best press after all. Having a large process window allows you to run on a less controlled press and still meet specs," O'Connor notes.

Fingerprinting a Tool

QTIP relies on extensive process monitoring to get the data needed to take a mold's performance "fingerprint." Those monitoring procedures, which will be passed on to users by GE field-service engineers and also through a forthcoming written guide, would be familiar territory to current users of process-monitoring systems. What's different, says O'Connor, is that the monitoring is not intended as a check on machine performance. Instead, it helps characterize a process before it goes into production. "We're applying existing monitoring tools in a proactive way," O'Connor explains.

At present, the QTIP program has been conducted only with process-monitoring systems from RJG Technologies of Traverse City, Mich., and Nicollet Process Engineering of Minneapolis. O'Connor expects that GE will work with most other monitoring systems as QTIP matures.

GE has yet to decide whether to sell the QTIP software and related process-monitoring guidance or to offer it to its customers as part of resin-purchase agreements.