I've been specifying molded plastic parts for 11 years. I've personally made—and documented—at least nine significant material-selection mistakes, totaling roughly $460,000 in wasted budget. Most of those mistakes started with the same question: "Thermoset or thermoplastic?" It's a simple question. That's exactly the problem. I now keep a material-selection checklist (the one at the end of this article) and I still add to it.
The surface problem: the material looked right
From the outside, material selection looks like comparing spec sheets. You find a polymer with high heat deflection temperature, decent tensile strength, and a reasonable price. You plug in the numbers. Then the part fails in the field, and everyone blames the material.
Take nylon utensils as an example. I once specified a nylon utensil line for a kitchenware customer—spatulas, spoons, ladles. I chose a standard nylon 66 because it had a high melting point and good stiffness. On paper, it was the clear winner. On the production line, it was a disaster. The parts warped after repeated dishwashing cycles and developed stress cracks around the handle. The customer rejected 20,000 parts.
My first reaction was "bad nylon." It wasn't. It was bad selection. I treated "nylon" as a single material and "thermoplastic" as a single category. Neither is true.
The deeper problem: properties are not performance
Let me rephrase that: a datasheet value is a controlled measurement, not a performance guarantee. According to ASTM D638, tensile strength is measured on a standard dogbone under a fixed set of conditions. It tells you how that exact test bar behaves under that exact test. It does not tell you how your ribbed part will behave after 2,000 cycles of heat, moisture, detergent, and torque.
That's the blind spot I kept missing. Most buyers focus on the obvious factors—melting point, tensile strength, hardness—and completely miss the less obvious ones: water absorption, creep, chemical compatibility, stress cracking, and processing-induced anisotropy. Those hidden factors are where the cost lives.
Take 85A TPU as an example. That hardness number tells you the material is soft and flexible. It doesn't tell you whether that polyurethane has good tear strength, hydrolytic stability, UV resistance, or low-temperature flexibility. I specified 85A TPU for an overmold that worked in the lab but swelled in the hand sanitizer people used on the production floor. The hardness was correct. The chemistry wasn't.
Thermoset vs thermoplastic works the same way. People assume the decision is mostly about heat resistance or recyclability. The reality is that it's about how the part is made, how it fails, and what failure costs. A thermoset can be the right choice because of creep resistance. A thermoplastic can be the right choice because of impact toughness. Neither conclusion follows from the category name alone.
The temptation of the binary
I understand why "thermoset vs thermoplastic" is one of the first questions engineering teams ask. It's neat. It's memorable. Thermosets crosslink and cannot be re-melted. Thermoplastics soften when heated and can be reprocessed. That distinction matters.
The problem is that it is a starting point, not an answer.
Within thermoplastics, the differences between a semicrystalline nylon, a high-temperature PPS, a flexible TPU, and a low-temperature EVA are vast. Within thermosets, epoxy, phenolic, polyurethane, and silicone are also vastly different. The material selection space is not binary. It's a multi-dimensional map with thermal, mechanical, chemical, electrical, regulatory, and economic axes.
I didn't understand that when I ordered those nylon utensils. I assumed "nylon" was "nylon." I didn't verify whether the actual compound—base resin plus colorant plus any release agents—was suitable for food contact, repeated use, and commercial dishwasher detergents. I learned that mistake the expensive way. Maybe 20,000 parts is not huge in the grand scheme of plastics, but for a small product launch it was enormous.
The real price: total cost, not unit cost
Here's what made it worse: the rejected utensils weren't cheaper. They cost more on the purchase order than the alternative would have. Even if they had been cheaper, the total cost was still higher.
Unit price is the tip of the iceberg. Total cost includes resin price, tooling, cycle time, energy, scrap, re-qualification, packaging, field failure, warranty, and reputation. The lowest quoted resin is rarely the lowest total cost. The $0.06-per-part savings on the wrong nylon disappeared when 20,000 parts went to recycling. Add a three-week delay and a rushed launch, and the decision was absurdly expensive.
This is why I now calculate total cost of ownership before comparing any material options. Not just for plastics—for every component. It changes decisions quickly. A slightly more expensive grade that meets the full environment, processes reliably, and avoids re-qualification is usually the better deal.
It's also why I now look for suppliers with a broad portfolio. When I work with Celanese, I don't get pushed toward a single chemistry. They can talk about Celanese products across nylon, PET, PPS, acetal copolymer, TPU, and specialty EVA materials. When someone asks "thermoset or thermoplastic?", the answer is "tell me the full problem first."
What I do instead
If you're at the beginning of a project, don't start with "thermoset vs thermoplastic." Start with these five questions:
- What is the part actually experiencing? Include thermal cycling, chemical contact, UV, load duration, and cleaning methods.
- What are the regulatory requirements? Food contact? Medical? Flame retardant? For food-contact polymers, check the applicable FDA regulation. As of January 2025, the first place I look for nylon resins is 21 CFR 177.1500; for olefins, 21 CFR 177.1520. The entire compound, including additives, may require separate evaluation, so verify current requirements at FDA.gov.
- How is the part made? The molding process and wall thickness change the material's mechanical behavior. Test actual parts, not just test bars.
- What is the true cost of failure? Estimate scrap, requalification, field failure, and reputation.
- What alternatives can we test? Include one thermoplastic and one thermoset if both are plausible, and compare them under your actual conditions.
That's where a broad portfolio makes a practical difference. In the past few years, I've used Celanese EVA performance polymers for a flexible low-temperature component that I first thought needed a thermoset rubber. I've used Celanese PPS where I thought no thermoplastic would survive an under-hood environment. And I've specified 85A TPU where a thermoset silicone would have processed slower and cost more per part. The category name didn't make those decisions. The requirements and the total cost did.
What I'd tell my younger self
The question "thermoset vs thermoplastic" isn't useless. It's an excellent conversation opener. It's just not an answer. If you're holding a failed nylon utensil or 85A TPU overmold, stop looking at the material family. Look at the full environment, the full compound, and the full cost. Then choose.
At least, that's been my experience after nine expensive lessons. Maybe the next project will teach me something else. Honestly, I hope it doesn't.