Blog

“What can I trust when it comes to injection molding simulation?” That’s the title, slightly rephrased, of an eye-opening (and perhaps throat-gripping) presentation at our Molding 2021 Conference last month in Rosemont, Ill. It poses a vital question for molders in this era of “smart manufacturing.” And the answer, according to the speaker, Jennifer Schmidt, senior Moldflow instructor at the American Injection Molding (AIM) Institute in Erie, Pa., is, “It depends …”—on an awful lot of things.

Schmidt took the real-life example of a simulation that was performed for a client by a service bureau. She painstakingly critiqued the simulation results report—ripped it to shreds, actually. She identified at least six tires to kick when evaluating such a report (something she has ample experience in), what questions to ask about each, what revealing symptoms to look for when—as in this case—direct answers to those questions are not provided in the report, and tips on best practices that should be used in any competent analysis.

Question 1: What software and version is being used? Is that software up to date? Although the report stated that a current software release was used, Schmidt pointed out some “tells” (as they say of poker players) that indicated otherwise.

Question 2: What type of mesh was used? Mid-plane, dual-domain or 3D? And not just for the part but for the runner as well (special mesh types are available for runners). Is the mesh type appropriate for the part geometry? Is it fine enough in critical areas to capture important details?

To answer those questions, you must make sure the report gives you access to a display of the mesh itself. “Why hide it?” Schmidt asked. And even without access to the mesh, she showed how the simulation results can reveal insights into the mesh quality—things like the filling animation, weld lines and sink marks.

If all that seems too deep in the weeds for you, Schmidt showed how simulations of the same part with the same material and same mesh density, but different mesh types for the part and runner, produced wildly different results for pressure at the fill-to-pack switchover point—ranging from as little as 10,000 psi to as much as 26,000 psi. That could make quite a difference in what happens in actual molding.

Question 3: What was the material data in the simulation based on? Was data on the actual material available, or was data on a substitute used—a resin of the same generic family but perhaps different molecular weight and flow characteristics, and perhaps without the same level of fillers, reinforcements or other additives used in the actual molding. And what was the level of material characterization? How much data was available? And how recently was that data generated? Might the formulation have changed since then? The bottom line is that an accurate molding prediction requires good material data. Without it, and it’s like you were test driving a different make and model before buying a car.

Question 4: What was modeled? What analysis sequences were run? In this case, the report provided warpage results, but none of the inputs to it. No data on time to freeze, volumetric shrinkage, cooling, etc. So how could you evaluate the credibility of the warpage prediction?

Question 5: What processing conditions were used—fill time, mold temp, melt temp, packing pressure, packing time, cooling? And how were those conditions determined? “Beware of default settings!” was Schmidt’s cautionary note.

Question 6: Who is the analyst? Are they certified? What is the level of experience or plastics background? And something else to give you pause: How often do they do simulations like this? “You can’t just do this once a month and keep up your skills,” warned Schmidt.

The highly suspect simulation report that Schmidt used as an example was full of soothing phrases like, “Flow pattern is uniform, no shortshot or hesitation, flow is balanced.” Don’t be fooled by bland clichés. You need someone like Schmidt to go over the report with a fine-toothed comb to ensure it is credible.

At the end of her talk, I thanked Schmidt for her thoroughly terrifying presentation. More like an autopsy. I don’t know if I would be more terrified if I were the customer or the analyst who did the report, knowing someone like her would give it such an uncompromising critique.

Don’t get me wrong: Molding simulation is undoubtedly one of the most important developments in plastics of the last 50 years. Used correctly, it is little less than miraculous. It has so changed the culture of injection molding that the enormous waste of time and expense in trial-and-error mold development is no longer tolerated. But, like any other tool, it must be used correctly. And Schmidt’s presentation is a valuable reminder that we can’t relax into complacency that we can let the computer do our thinking for us.