Why setup sheets, DOEs, validations, etc. may not yield the results they should.

Click Image to Enlarge

In this test, the only machine parameter that was changed was injection velocity. But note the difference in part dimensions even though shot size was not changed at all.

In our business, an extraordinary amount of time, money, and effort is spent to develop, check, and document reproducible 24/7 processes for making parts and assemblies.  The automotive market has its Production Part Approval Process (PPAP), medical OEMs have the FDA Validations, and most of us are familiar with Design of Experiments (DOEs). 

All of these are performed with the best of intentions. Yet despite all this effort, bad parts are still molded and the struggle continues to transfer a process from one machine to another and make identical parts.  The time and expense involved are huge—in the hundreds of thousands of dollars.  Perhaps worse is the acceptance by many that we need to go through the above exercises just for the sake of doing so.  Many times I’ve heard, “It is the price of doing business, it is required.”  I disagree. Spending this human, financial, and other resources must yield appropriate benefits and dollar value if our critical manufacturing base is to remain competitive, survive, and grow.  If we are going to spend these resources we need to maximize their impact and value. How do we get on track to get the desired results?

While validation/PPAP is a huge topic, there are some basic issues that will significantly improve results and save time and money.  First on the list is to always work with the plastic variables—not the machine variables.  That sounds straightforward enough but it’s not readily understood or even used within our industry.  Proper documentation is critical for validations, PPAP, DOEs and repeating processes to make identical parts. 

To put this in perspective, let us say there are 8000 molding facilities in the U.S.  Each needs to document the processes run on its machines. Common practice is for each shop to use its standard “setup sheet.”  This is where things start to go awry.  If you examined each shop’s setup sheet you’ll find 8001 different ones. Does this make sense to anyone? Not to me. It clearly reveals that we molders do not have our act together. What’s more, these varied setup sheets don’t always yield the desired result of making identical parts.

Most setup sheets document what are known as machine parameters, but not plastic parameters.  What is the difference?  While it might seem logical to duplicate setpoints on your machine controller from one run to another, this will not duplicate the plastic parameters, and it is these parameters that are truly responsible for making your part.  Your controller lets you establish setpoints, but they do not describe or identify what is really happening to the plastic forming the part. 

Start with melt temperature. How is temperature documented in your plant?  Most likely, your operators will transcribe numbers from the controller screen onto the setup sheet.  They may even calibrate the thermocouples, but keep in mind that thermocouples are probes that make very little contact with the steel of the barrel; nobody even bothers to put heat transfer paste in the thermocouple wells to provide better heat transfer so they read more accurately.  But in any case, it’s best to document the actual melt temperature, not the steel temperature. 

Some brave souls interrupt the cycle, which is not good for steady-state processing, to purge plastic out of the barrel and make an attempt to take a melt-temperature measurement.  This is not easy to do and you cannot get five people to get the same temperature that way.

Bottom line: Setpoints are machine parameters and actual temperatures are the plastic parameters that need to be duplicated or used for PPAP etc. We could take this to the next step and get into questions like, “What is the actual melt temperature in the barrel, nozzle body, hot runner, and hot tip?”  Granted this is complicated. but we have to recognize that the number on the screen of the controller is probably wrong.  Does anybody want to argue that actual melt temperature is not important? 

Next is velocity or injection speed.  Again, most people like to write down the setpoints shown on the controller and then try to duplicate the process by replicating those setpoints. But once more, these are machine parameters not plastic. Sure, the computer considers them as the same, and signals are sent to valves to move or open up; but do the valves open up the same every time, or do they get varnish buildup and do amplifier cards burn in with time?  Voltages can vary and valves can stick. 

Some molders try to combat this by calibrating the machine every year.  Good money is spent on this, but most calibrations on velocities check voltages going to the valves, not actual velocities. Some will argue that on closed-loop systems the computer checks to make sure it is going at the set speed.  This seems to work but are you getting the set speed?  A better way to duplicate injection speed is to document the fill time for the volume of material injected, and allow the molder to use whatever velocity setpoint is needed to match fill time. There is still the issue with different machines having different accelerations, but it is a step forward. 

This relates to duplicating velocity, but what about using velocity as variable/factor in a DOE?   You can select any number of factors for a DOE, but many share the same problem: Change one machine parameter, such as velocity, and you change more than one plastic parameter.  This fouls up your DOE results and makes interpretation impossible.  How?  Review the sequence of parts in Fig. 1.  This is a series of parts where the velocity was changed seven times, yielding seven parts made at different fill times.

The only change to the machine’s controller was for injection speed.  Shot size was not changed, yet the parts are not the same size. The machine was working properly.  Nearly all machines will reproduce this variance in part size from changing injection speed.  This is done with fill only (no pack pressure), but leaving the normal pack or second-stage time the same as if you were making full parts. This verifies that one machine parameter adjustment changed at least two plastic part parameters—fill time and shot size. Shot-size setpoints were identical for each of these seven shots. The intention was to change one variable or factor, but at least two were changed.  In order to get DOE data that are comparable, the shot size needs to be adjusted so that the identical amount of plastic is injected into the part at the different injection speeds. Then you could see and properly evaluate the influence of velocity on your part. 

The point is, if you want proper documentation and good data, you must understand that setpoints are just numbers on the controller. What’s happening to the plastic is often different.

ABOUT THE AUTHOR: John Bozzelli

John Bozzelli is the founder of Injection Molding Solutions (Scientific Molding) in Midland, Mich., a provider of training and consulting services to injection molders, including LIMS, and other specialties. E-mail john@scientificmolding.com or visit scientificmolding.com.