New or existing mold trials need to produce a production-worthy process that will run 24/7.

Click Image to Enlarge

If you want a process that will mold good parts consistently, don’t accept the first trial parts that “look good.” You need to do some testing to establish the process window and ensure it encompasses normal material variation.

Recently I had a few calls with the same theme: “We made good parts during the initial mold trial, but now that we are in production the dimensions are not consistent.” In such situations, I have found that the mold startup “process” sometimes consists of quickly shooting a few parts that “look okay,” then walking away believing you are ready for production. On other occasions molders have actually done their homework, conducted a design of experiments (DOE), ran some trial runs, etc., yet the parts still lack consistency.

New or existing mold trials need to produce a production-worthy process that will run 24/7. Yes, I know the mold may have come in three weeks late and somebody wants parts shipped the next day, making everybody frantic. I recognize you do not always have the time to do the full studies—or maybe you do have time and you do the studies and still production is poor or there are customer returns. You have been doing this for umpteen years and things are getting worse. Certainly a case can be made for trying new tactics.

First, get everybody on same page. Explain that you need to put together five critical components for a successful molding application. They are, as formulated by Glenn Beall and Plastics Technology columnist Mike Sepe: piece part design; resin selection and handling; tool design and construction; processing; and testing.

Most molders cannot afford nor need in-house expertise for all of the above elements. Arrange for an outside specialist in the area(s) needed. Remember to have all tool drawings double-checked. The time to find problems is at the drawing stage, not after steel is cut.

Let’s move on to mold delivery and developing a molding process. Most molders would like (and even pay for) someone to write down a stepwise procedure that quickly yields a capable production process. That is the ISO 9000 or TS mentality. Chuck it. The fact is, you are not slinging hamburgers, and no one can write a foolproof procedure for something as complex as molding. Worse, these stepwise procedures kill creativity and critical thinking. Every mold and resin combination has its own oddities. Finding them early is the challenge and following a set procedure is not going to get you there.


Guidelines, not strict procedures, are okay with me. You must allow the operator to think! You also need trained, proficient operators to follow these guidelines. They should be instructed to use their talents to interpret results of various experiments to define any weakness in any of the five components. Their goal is to find and define problems, not process around them. Problems will show up in production and the competitive molder knows that finding them before production will cost less than afterwards.

My guideline experiments include cooling; viscosity curve with visual part inspection at each velocity; velocity-to-pressure switchover response, including momentum; pressure-loss analysis; short-shot study; gate seal; melt temperature; part temperature via infrared imaging (not expensive or time-consuming); second-stage pressure range; screw recovery; and cycle time. All this should take about 2 hr, and if it takes much longer it’s a sign of significant production troubles ahead.

Other experiments may be called for if something odd turns up. I never know beforehand. There is no race to make a part that “looks good.” You need a series of experiments to define the critical parameters for this mold and resin combination. Essentially these experiments are about range-finding—i.e., defining a process window. You want to establish a base process with a list of concerns to be addressed. Do a DOE if necessary. The base process derived from these experiments establishes the levels for the variables (factors), as well as which variables are critical. This often speeds the DOE, as you have data to establish levels and factors rather than guesses.

Developing a process does not start with shooting plastic in the mold. It starts with ensuring the mold is set up properly in an appropriate machine that works. For example, cooling is 90% of your cycle time, and it is often not done right. Most of the time, somebody hooks up a half dozen or more hoses to the mold from a manifold and walks away. Document that each coolant channel is running turbulent flow (Reynolds number of at least 5000) and is regulated such that the flow (gallons or liters/min) can be duplicated from run to run.

Once you have a base process, make some parts for testing to move the process to the center of the part specs. Do not send or show parts to the customer yet, even if they “look good.” What if good parts are scarce when you actually run the first order?  Better run for cover; your boss and client will want to know why. If you do make a part that seems pretty good right off the bat, hide it or push it under the forklift. There is testing to be done, including melt flow rate (MFR) on pellets and parts, thermal cycling, and any other tests that show the part meets the critical requirements.

During this testing phase you will learn how to center the process to the part specifications. Often this requires tool modifications. During this phase you will be told these modifications can’t be done, or there is no time or money to do them, so you better find a way around the problem. Don’t do it, or you and the company will be fighting that problem for the life of the job, killing the profit margin and convincing the client you are less than competent.

Once the operator feels this process is stable and initial part testing shows promise, you must challenge the process. Is it truly a stable process? Are fill time, pressure at transfer, cushion, recovery time, cycle time, and part temperature within an acceptable range as you start up, shut down, and have somebody new start it up again?

If so, the next step is to force a viscosity change. This is critical. It will happen anyway sooner or later, so it’s best to know what is going to happen and how to deal with it. Force a viscosity change by testing a different lot, or wet resin, or 100% regrind, or different colors, or whatever you can think of that will mimic the variables in 24/7 production. Are these material variations reflected in the process outputs? Is first-stage consistent as the material variations run through the process? Can you accommodate these changes by changing the second-stage pressure?

This is perhaps the most critical test for a true production process—forcing a viscosity change and seeing if it can be accommodated. It is rarely done. All the work goes into testing one lot of material, but when 24/7 production arrives, you have to process different lots. You cannot blame the resin supplier. As Dr. Deming taught, all processes have normal variations. Your process has to deal with them. Bottom line: If you have tested only one lot of material you do not have a production-capable process.