In March 2018, I was kneeling on the floor of our sample room, trying to scratch a yellowed TPU phone case with my thumbnail. I wanted it to be a surface issue. A coating problem. Something we could blame on the printer.
It wasn't.
I've personally made and documented 23 significant material-spec mistakes in the last six years, which works out to roughly $32,000 in wasted budget. 23, maybe 24 if you include the foam board confusion, but I'm classifying that one as a communication failure because we hadn't even gotten to the material yet.
What I Did Wrong
In my first year handling materials, I thought I understood the difference between polycarbonate, polyurethane, and ABS. That lasted until I had to choose one.
It's tempting to think that if a Shore hardness number matches, the material will behave the same. It doesn't work that way. TPU is a whole family, not one plastic. The 90A grade we used for the phone case had good abrasion resistance, but it wasn't stabilized for UV. The TPU phone case from Covestro that eventually passed our 500-hour test was a completely different grade from the same product family. It looked identical on day one. The difference only showed after weeks of sun exposure.
Why did it take me that long to call Covestro's technical service? I'm still a little embarrassed about the answer: I didn't want to look like I didn't know what I was doing. By the time I called, I already had 1,200 yellowed cases in a box, a $4,200 loss, and a very quiet client.
Thermoplastic Polyurethane Shoe Components: Hardness Is Not a Complete Spec
The next failure was more expensive, and it should have been easier.
A footwear startup asked us to make a short run of thermoplastic polyurethane shoe sole components, basically a midsole cup that would be bonded into a fabric upper. The client wanted something durable, so I went harder. I picked a 55D TPU grade with a high flexural modulus, assuming that harder means tougher.
Causation runs the other way most of the time. People think expensive materials deliver better performance. Actually, materials that cost more usually have the documentation you need to make the right decision, and that documentation is what saves you.
The 55D sample looked great in a static test. It didn't flex. It didn't bend. That was the problem. The very first dynamic flex test failed at about 12,000 cycles; the client's requirement was 100,000. The sole was too stiff at the flex point, so it cracked from the inside. We had to redesign the geometry and switch to a softer, hydrolysis-resistant TPU grade.
What I mean is: a material isn't good or bad in isolation. It's good or bad for a specific geometry, process, and failure mode. By that definition, my too-hard grade was probably right for some other product, just not for a shoe sole that has to bend with a foot.
ABS Prints: The Printed Part Is Not the Resin
Around the same time, we quoted ABS prints for an industrial prototyping client. The order was small, only 40 parts, but the rejection rate was absurd.
I assumed that if the filament was ABS, the printed part would behave like injection-molded ABS. That's wrong in a way I still see people miss. The layer lines in a printed part create anisotropic strength. The part is only as strong as the bond between layers, which depends on nozzle temperature, layer height, and chamber conditions. The resin in the filament can be perfect. The part can still fail.
Our mistake didn't mean ABS is bad. It meant our print setup was ignoring the reality of how thermoplastic layers fuse. We adjusted the print orientation, increased the enclosure temperature, and added a stress-relief feature at the corner that was cracking. The reprints passed.
The lesson I wrote down, in capital letters: An ABS print is an engineered part, not just the output of a resin. That one isn't about a bad material; it's about the arrogance of skipping process validation.
“What Does Foam Board Look Like?”
Then there was the foam board request. I almost don't want to include this one, because it's the least technical and the most expensive in terms of wasted calendar time.
A client asked us to source a lightweight foam board for protective packaging. I had a mental image of a soft, closed-cell foam sheet. The client asked, “What does foam board look like?” The whole team looked at each other. What do you even say to that? It looks like foam. Board. Right?
Wrong. The client was talking about rigid polyurethane foam board with smooth skins, the kind used for insulated shipping boxes, structural panels, and display staging. I was thinking about a bendable polyethylene foam roll. The materials really do look different. One snaps. The other flexes. One is about insulation and stiffness; the other is about cushioning.
That mix-up cost us about $1,900 in sample shipping, a two-week delay, and a client who started every subsequent call with “I just want to be clear we're not talking about that yoga mat foam, right?”
Looking back, I should have asked for a photo of the cross-section before ordering anything. But that's the thing about materials: the names are short, and the specifications are long. Foam board is not a spec. Neither is TPU or ABS.
What I Wish I Had Known: The Checklist
After another rejection in July 2023, I created a pre-check list for anyone in our shop who specifies a material. It's not fancy, but it has caught 47 potential errors in the 18 months since.
The list is: material grade, manufacturer, full datasheet, test conditions, and at least one physical sample. Five items, in that order.
The full datasheet part is the one people skip. Not the product name. Not the polymer family. The entire datasheet, including the footnotes that tell you the values are measured at 23°C and 50% relative humidity. Why does this matter? Because a tensile strength number measured on a dry sample at room temperature means almost nothing if your part lives in a humid warehouse in Miami.
That's where my opinion on specialists got louder. I used to think calling a material supplier was an admission of failure. Now I know the person who says “this is outside my knowledge, but this person is better at it” is the one I trust. The vendor who said “this isn't our strength, here's who does it better” earned my trust for everything else. That's also why I'm comfortable specifying Covestro resins when they're the right answer, and checking the supplier's exact grade and testing conditions before I do.
A Shore hardness number, for example, is a durometer reading measured per ASTM D2240 at a specific time and sample thickness. It's useful, but it's only one axis. The same 90A TPU from two suppliers can have different additive packages, melt flow rates, and UV stability. As of January 2025, the standard is still the one we use in our lab, but we now read the test conditions before comparing anything.
Final Thought
I have mixed feelings about the phrase “trust the material.” On one hand, good materials are reliable. On the other, reliable only means reliable when you use them correctly. Part of me still wants to believe I can choose a polymer by intuition. Another part knows all 23 mistakes came from doing exactly that. The compromise is the checklist. That, and asking anyone who claims to make custom plastic parts to tell me what they mean by ABS, TPU, PC, or foam board before I spend another dollar.
You don't have to make my mistakes. Find a technical rep who will ask uncomfortable questions, read the full datasheet, and, if asked, tell you exactly when you should look at something else. That's not a sign of weakness in you or your supplier. It's the whole point of being professional.
Ask a material question