Here's a universal scale for grading your machine capability.

The current striving for "world-class" quality in plastics processing is somewhat hampered by the lack of a simple, commonly understood means of describing machine repeatability. How repeatable is "highly repeatable"? And how precise is a "precision molding" machine?

One answer to those questions was offered at last May's SPE ANTEC meeting in Detroit. Denes Hunkar, CEO of Hunkar Laboratories Inc., Cincinnati, proposed a universal "quality standard yardstick" for characterizing injection molding machine repeatability. He also has devised different versions of the same concept for blow molding machines. Since then, his concept has been put into action: Several processors and a major rebuilder of injection and blow molding machinery have adopted it as their machine quality standard.

A NINE-LEVEL SCALE

The injection machine quality standard proposed by Hunkar is a nine-level scale, each level defining an acceptable tolerance band for 19 process parameters. This scale was derived from a statistical database of millions of cycles performed by over 1800 injection machines in actual production. Hunkar divided this machine population into nine more or less equal subgroups according to their repeatability. "Class 1 represents the best machines in operation today," says Hunkar, "and TABULAR DATA OMITTED Class 9 represents the worst machines that can still produce acceptable quality with highly tolerant tooling and material combinations." Accordingly, the tolerance "window" for each variable in Class 9 is wider than for Class 1.

Hunkar does not claim that his scheme of nine classes is the only one possible. "What is important is that a yardstick exists to which all machines can be compared."

PUT INTO PRACTICE

Dale Werle, manager of the Contract Services Business for Cincinnati Milacron's U.S. Plastics Machinery Div. in Batavia, Ohio, says his group has used the quality standard yardstick for about a year on its rebuilt injection and blow molding machines. "It's a super sales tool for us," says Werle. "Customers are becoming more and more aware of statistical techniques for analyzing process variability."

Milacron typically does a before-and-after analysis of a rebuilt machine. First, a Contract Services technician brings a portable machine analyzer to the customer's plant to determine the quality class rating of the machine in operation. The machine is checked again after rebuilding. "We guarantee Class 2 or better performance on remanufactured machines," says Werle. "And we can go back to the customer's plant and check the machine again a year or two later to see that it's maintaining that level of performance." Werle notes that some major processors themselves use the quality yardstick as the specification when ordering machinery rebuilds.

HOW IT'S DONE

Applying the machine capability yardstick is relatively simple. It requires properly instrumenting the machine with sensors to pick up time, temperature, pressure, and position data. You'll also need a high-speed means of recording those data in real time. You can print out and analyze the data by hand, but some type of statistical analysis software would make it easier.

To compile the data, the molding machine should be running on a stable cycle. Then gather data on the 19 parameters in Fig. 1 over the course of a statistically significant number of shots--at least 32-50 but preferably 100, according to Hunkar.

The next step is to analyze the machine performance for each of the 19 parameters. Starting with the first parameter on the list, find the mean value over the total number of shots, and then determine the difference from the mean for each shot. Are any of those differences from the mean greater than the Class 1 "acceptable tolerance" for that parameter shown in Fig. 1? If the answer is no, then you have obtained Class 1 performance for that parameter. If the answer is yes, you may still have Class 1 performance if the number of out-of-tolerance deviations is too small to be statistically significant. Hunkar defines "statistically significant" as a number greater than the square root of the total shots sampled. Also, disregard large deviations that resulted from an identifiable process interruption such as a setpoint change.

After such "editing," if you're still left with a statistically significant number of deviations greater than standard, then move down to Class 2 and compare your data with the acceptable tolerances there. Keep moving on to the next quality class level until no statistically significant number of exceptional cycles remains.

Follow the same procedure for all the other parameters on the list. The overall quality class level for the machine is the same as the "worst" quality class level determined for any of the 19 parameters. In other words, a "Class 4" machine must achieve Class 4 or better performance on all parameters.

To make all this easy and almost automatic, Hunkar offers a suitcase-sized Portable Machine Analyzer to take readings from sensors or the machine process controller. It uses Hunkar SPC-PRO software to collect the data and then provide the machine capability analysis. Or, if the machine is equipped with a Hunkar DAT- or DAC-type data-acquisition terminal, the data can simply be downloaded from the terminal to a PC equipped with SPC-PRO software for the analysis. What's more, users of Hunkar's CIM system can do the analysis at PC user terminals, drawing on the centrally collected data from all machines.

As a result of a recent alliance with Hunkar Laboratories, Milacron's advanced-level Vista controls on new and rebuilt machines are available with software that gives them capabilities similar to those of a Hunkar DAT or DAC terminal. The data can thereby be downloaded from the controller to a PC running the SPC-PRO program.

With SPC-PRO, the machine capability analysis is said to take only about 3 min. As shown in Fig. 3, the software provides a table of the mean, minimum and maximum values observed for each variable; the upper and lower specification limits (USL and LSL) that define the acceptable processing window for that variable; and the USL/LSL Count of cycles that exceeded those limits. The display automatically highlights any USL/LCL Counts that are statistically significant. An editing function allows you to "clip out" any "bad" cycles caused by a setpoint change or other process interruption.

The quality class levels are shown in a pull-down window. When you pick any level from 1 to 9, the software automatically adjusts the USL/LSL limits accordingly. The best class level for which no statistically significant USL/LSL counts are highlighted equals the quality class for the machine.

GOOD DATA ARE ESSENTIAL

The accuracy of pressure readings is crucial to the machine rating process. For that reason, Milacron recommends Hunkar's "Gold Standard" pressure sensors and instrumentation to customers who are seriously interested in machine capability studies. That's because of the unusual quality-assurance procedures used in their manufacture.

Hunkar tests all its transducers by comparing them under actual operating conditions with one of its own transducers whose calibration accuracy has been certified by the National Institute of Standards and Technology (NIST) in Gaithersburg, Md., or one of NIST's authorized agents. Likewise, all Hunkar amplifier instrumentation is tested for accuracy with an NIST-certified sensor simulator.