CIM: brings out the best in thermoforming.

Few thermoformers have their process as well wired as this Midwest firm. Its success at raising quality and cutting costs shows other sheet formers that it's time to take a close look at CIM.

It is still quite rare to hear of a thermoformer with experience in computer-integrated manufacturing (CIM). But American Sheet Extrusion gained enough confidence from operating a CIM system on its three ABS and HIPS sheet lines in Evansville, Ind., to put a separate CIM system to work monitoring seven of its 11 thermoformers. Those custom-built, roll-fed machines form up to 39 different refrigerator door panels for the company's main customer, Whirlpool. In three years, real-time process monitoring with CIM helped American Sheet Extrusion reduce average thermoforming cycle times by about 2 sec, or nearly 10%.

DATA SHARING ON A NETWORK

American Sheet Extrusion's CIM system gathers process data from each thermoformer with a PLC 5 programmable controller from Allen-Bradley Co. The data are transmitted over an Allen-Bradley Data Highway Plus network to several personal computers in plant offices.

The monitoring programs are customized from a PC-based CIM software platform called Paragon TNT from Intec Controls Corp. American Sheet Extrusion developed around 50 proprietary screens with the software. Thirty-four are process related, eight of them specifically for time and heat settings, and 26 others are used for SPC applications like trend charts and histograms. Eight other screens are for production status and alarms, and another eight are for machine maintenance.

Paragon allows American Sheet Extrusion to import machine-monitoring data from the Allen-Bradley PLCs into standard spreadsheet programs to generate reports for use by the purchasing, financial, or customer-relations departments. For example, the financial department's IBM AS400 midrange computer can read those spreadsheets in order to measure real-time production data against expected costs. The AS400 system also contains order and inventory data, so it is a useful tool for informing customers of the exact production status of their orders.

Via a modem, the Paragon software automatically dials up a pager carried by a quality technician. The technician is thus alerted to make a quality check at a particular machine based on the number of parts produced since the last check, rather than simply on a time basis.

Approximately 140 points of data from each machine are logged on each cycle. Ninety of those points are temperature related, the most critical variable in producing a good part. Readings from the quartz heaters in the oven, as well as the devices that adjust them, are collected automatically for each cycle. Seven sensors in each mold measure sheet temperatures and provide feedback for automatic heater adjustment. Also monitored are data on the accuracy of press movements and the hydraulics operating the trim die.

Machine operators can display all 140 points of process data on each cycle at their touchscreen terminal through ControlView software from Rockwell Software Inc. Only 10 or so data points (mainly process speeds and temperatures) are monitored remotely. Machine operators do not have access to all of the trend screens that are available on the remote CIM terminals. Operators also have limited ability to change setpoints of the various parameters. In fact, the system has six levels of security. This precaution trims one of the biggest sources of process variation, human intervention, says Ron Rust, engineering manager of the sheet and thermoforming divisions.

REMOVING VARIABILITY

American Sheet Extrusion's initial investment for its CIM system was roughly $15,000 to $25,000 per machine. It has paid for itself in numerous ways.

In the past, putting the correct amount of heat into the sheet was the company's biggest problem. Before integrating CIM into its thermoforming process, the company relied on guesswork by the operator, says Howard Dibble, engineer in the thermoforming division. "It was very difficult to keep the heating consistent," Dibble says. The operator might vary the cycle time by as much as 1 to 1.5 sec while hunting for an optimum heating cycle.

With CIM, monitoring of critical process variables from shot to shot has helped hold cycle times to 22.5 sec and cut variability to [+ or -]0.3 sec. CIM also helped reduce rejects by 30% through monitoring oven and sheet temperatures. The company also reduces the chance of generating scrap by monitoring the sheet index system's movement to [+ or -]0.02 in. vs. 0.250 in. previously.

Better control has allowed the company to trim sheet thicknesses by 20% and hold part dimensional tolerances to within 0.1% versus 0.24% without CIM. Another big savings came from trimming the number of personnel at a line to six per shift from 20 before CIM, Dibble says.

CIM AS SLEUTH

CIM prompted American Sheet Extrusion to monitor data points that help identify the root causes of processing anomalies. One problem area was the vacuum box. The company had previously used vacuum gauges, but the data were never recorded, Dibble says. He designed a program that records the peak vacuum level during a cycle and how long it took to reach it. "CIM tells us if we have a good seal on the pressure box and indirectly gives us an idea of the sheet's perimeter temperature. If the plastic is too cold or too hot it won't seal properly," Dibble adds.

As one example of applying SPC data, Dibble keeps an eye on the Cpk (process capability index) for the peak vacuum level. If he sees the Cpk is at least 2, he judges that variable to be in good control. If the Cpk is lower, Dibble then takes a closer look at the monitoring data to find out why.

In one case, the company started to see a fluctuation in part definition. The possible source of the problem could have been the sheet, the operator, or some element of the machine. Dibble looked over the vacuum data logged over a period of time through its CIM program and discovered a trend in the vacuum readings toward the minimum tolerance limit. The cause turned out to be plugged vacuum lines, which adversely affected part cooling. The vacuum trend data can reveal hose leaks as well. The company programmed an alarm to signal whenever vacuum-gauge readings show a trend toward the tolerance limits.

Monitoring of data trends helped American Sheet Extrusion to notice when a change in mold or oven temperature wasn't registering fast enough. "When the sensors go bad they become real slow to change," says Dibble. Usually a broken sensor will remain "stuck" at some temperature reading, regardless of what the process is actually doing. Without the trend evidence, it could appear that all processing parameters are fine, even though bad parts are being produced.

In another instance, CIM helped track down a puzzling oven disturbance that came from outside the process. The CIM records showed frequent temperature alarms and readjustments of heater setpoints by the temperature controllers in an attempt to compensate for the upsets. Physical inspection indicated that the quartz tubes were in fine condition. Was something wrong with the controllers?

Dibble developed a program to report the electrical current running through the heaters, which revealed a curious ebb and flow of current consumption. "At first we though it was a neighbor starving us for power," he says. Bob Knowles, maintenance supervisor, noticed that a newly installed extruder was ramping up to speed at the same time that the thermoformers were experiencing this problem. Comparing the data logs from the extruder and the thermoformers showed that the oven problem was related to a brown-out effect whenever the extruder's 500-hp drive changed speeds. The extruder was fed from the same high-voltage transformer as the thermoformers.

American Sheet Extrusion also monitors the position of the upper and lower platens at the beginning and end of each cycle in order to recognize press drift before it starts to affect production. The positions are normally accurate to 0.1 in. "This information helped us schedule cylinder rebuilds and reduced unplanned downtime," Dibble notes.

The company also created a CIM screen that charts the time it takes for the top platen to rise. It shows the effects of not cooling the hydraulic fluid. "As the oil becomes warmer the platen takes longer to rise due to leakage. Measuring just how much longer it takes helped us to make an informed decision on the purchase of heat exchangers," Dibble reports.

FURTHER INTEGRATION AHEAD

American Sheet Extrusion plans to monitor more process variables (such as unwind speed) in the future for even tighter control. Another goal is to link the results of quality-control checks to process data. Eventually, the company also plans to deliver job schedules directly to a screen at the machine. American Sheet hopes that one day CIM can be used to automatically start and warm up the machines before an operator arrives.