Investor and application news

I Rejected 37 Adhesive Batches in Four Years. Cost Per Kilo Was Never the Real Problem

Document disciplineSDS-firstApplication evidence

A batch that passed every test and still got rejected

The batch arrived with a certificate of analysis that looked clean. Lap shear strength was in spec against ASTM D1002, the standard tension-loading test for metal-to-metal adhesive joints. Viscosity matched the TDS. Pot life matched the TDS. I rejected it anyway, because the substrate listed on the certificate was anodized aluminum and the parts moving down our line that week were cold-rolled steel.

That gap—between what the document proves and what the part actually needs—is where most adhesive decisions go sideways. I've worked quality and brand compliance on assembly lines for four years. I review every adhesive and sealant specification before it reaches purchasing: roughly 260 submissions so far, and about 14% of first deliveries never make it to the production floor. Very few of those rejections were about bad chemistry. They were about chemistry that was perfectly fine for a different job.

The two chemistries that land on my desk most often for structural and semi-structural work are epoxy and polyurethane. Both are specialty polymer products sold on spec sheets that look almost interchangeable at first glance, which is exactly why the comparison keeps going wrong. Both work the same way at the interface: the adhesive wets the substrate, cures into a solid, and transfers load across the joint. The difference is what happens during cure and what the cured polymer does under heat, load, and time. That difference is where the money hides.

Here's the framework I use on every comparison sheet. Four dimensions, all measured in total cost of ownership rather than unit price: cost per bonded part, dispensing and handling, service temperature, and shelf life. If you've ever signed off on a "cheap" adhesive and then watched the scrap rate climb, you know why the first number on the quote is the least useful one.

Dimension 1: Cost per kilogram is a meaningless number

You don't buy adhesive. You buy bonded parts. A quote of $11.40 per kilo means nothing until you know grams per part—and for structural grades, grams per part for epoxy and polyurethane usually land in the same narrow band, somewhere between 0.4 and 1.8 grams on the assemblies I've reviewed. On a $9 bracket, the adhesive line item is pennies either way. The spread between the two chemistries is noise.

What actually diverges is the dispensing. Epoxy is almost always two-component, and two-component means metering and mixing: either a benchtop dispenser you tear down and clean every shift, or a meter-mix machine with a maintenance contract and a spare-parts budget. Polyurethane is often one-component (moisture-cure) or a hot-melt, which can run on considerably simpler equipment.

I learned this the expensive way. In 2022 I approved a two-part epoxy for a low-volume bracket line because the per-kilo quote beat the polyurethane alternative by 18%. Nobody priced the dispenser, the purge solvent, the nozzle consumption, or the two days of operator training. First-year cost of ownership came in about $7,800 above the polyurethane path that had looked more expensive on paper. The adhesive was cheaper. The program wasn't.

Dimension 2: Cure speed and the hidden cost of fast

Polyurethane generally cures faster—handling strength in 5 to 30 minutes for many structural grades—while epoxy typically needs 30 minutes to several hours, sometimes a full day at room temperature before it takes load. On a high-throughput line, that gap is real money in work-in-process and floor space.

But fast cure is not automatically cheaper, and this is the part I got wrong first. Short open time means a short window to position the part correctly. Every second of hesitation becomes rework, and rework on a cured structural bond usually means scrap, not repair.

We didn't have a formal work-instruction process for the fast-cure line when we first switched. Cost us when a new-hire shift misaligned 60 parts inside the open-time window and nobody caught it until after full cure. That was a $9,200 scrap event, and it had nothing to do with the adhesive's quality. It was a process gap, and the adhesive was simply the thing that exposed it. The third time anything like that happens, you stop blaming the material and start writing instructions.

The supply line nobody puts in the cost model

Feedstock geography belongs in a TCO comparison, and it almost never appears there. Methanol sits behind a lot of polyurethane chemistry, and its global capacity is concentrated:

Global methanol capacity is on the order of 110 million metric tons per year, with China accounting for more than half. — Methanol Institute capacity data, 2024

Epoxy's chain—epichlorohydrin and bisphenol-A—is geographically broader, though nobody who bought resin in 2021 would describe it as stable. The practical effect: when methanol pricing moves, polyurethane quotes tend to move faster and lead times stretch first. I'm not arguing that makes polyurethane the wrong pick. I'm arguing that a twelve-month supply plan built on the spot price of one feedstock is a guess dressed up as a number. In our 2023 sourcing review, the polyurethane line's landed cost moved 9% across the year against a nominally stable contract price, purely from allocation and expediting. That never shows up on the quote sheet.

Dimension 3: Service temperature is a gate, not a trade-off

This is where the comparison stops being a comparison. Epoxy's glass transition temperature—the point where the cured polymer softens and load transfer collapses—commonly falls between 80°C and 150°C in standard grades, and higher with specific curing agents. Polyurethane's is lower, usually in the 60°C to 100°C range for structural grades, with select high-performance formulations pushing past that.

If the part goes through a powder-coat or e-coat oven, that's typically 30 minutes at 180°C or above. No standard polyurethane survives that as a structural joint. Not because it's inferior. Because it's the wrong material for that process. There is no cost trade-off to make here. It's a hard gate, and treating it as a soft one is how programs fail.

I found that gate in 2021, the hard way. They warned me to verify service temperature before the first powder-coat trial. I didn't listen. Forty-two brackets came out of the oven with adhesive soft enough to peel off by hand. About $6,300 in scrap plus a missed shipment, and it permanently changed how I read every TDS sheet since. If you ask me, service temperature should be the first filter on any comparison table—before price, before cure speed, before anything.

Dimension 4: Shelf life is a warehouse problem, not a chemistry problem

Epoxy generally holds 12 to 24 months at controlled storage. Polyurethane—especially moisture-cure formulations—runs 6 to 12 months and is far more sensitive to humidity and heat. A warehouse that hits 38°C in July will quietly shorten that number without telling anyone.

I only believed this mattered after comparing two quarters side by side: same supplier, same product code, same incoming inspection, different storage building. First-pass yield on the polyurethane line dropped six points in Q3 against Q1, and every rejection traced back to partially cured adhesive in drums that had spent the summer in the warmer building. The adhesive wasn't defective. The storage spec was somewhat optimistic is the polite way to put it.

After I built the rotation checklist—FIFO by drum date, temperature log posted at the door, monthly shelf-life sweep—yield came back. It also changed how I read TCO. A cheaper adhesive with a shorter shelf life is only cheaper if you turn it fast enough. On a slow-moving part number, six months of shelf life is a red flag, not a bargain.

Where each one actually earns its place

There's no overall winner here, and that's not a hedge. Four years of rejected batches teaches you that the right answer depends on what the part does after it leaves your floor. If I compressed it into a decision rule, it would look like this:

  • Choose epoxy when the assembly sees sustained load above 80°C, when you're bonding metals or high-surface-energy plastics, when shear and peel strength drive the design, and when you already have—or can justify—metering equipment. Araldite grades from Huntsman are one of the lines we've qualified in this role, alongside several other suppliers.
  • Choose polyurethane when substrates are flexible or dissimilar, when impact or peel dominates over static shear, when service temperature stays comfortably under 90°C, and when line speed matters more than ultimate strength. Huntsman's Irostic thermoplastic polyurethanes are one option in that space; there are plenty of others.
  • Choose neither when the joint has to come apart for service. Both are structural adhesives and neither is designed for disassembly. I've watched a maintenance team destroy a $400 casting trying to defeat a $2 bond, and that's a no-brainer argument for mechanical fastening in that specific application.

The bottom line: cost per kilogram is the number that gets you into the meeting. Cost per bonded part, cost of the dispensing line, cost of a failed oven trial, and cost of the drum that expires at the back of a hot warehouse—those are the numbers worth arguing about. If you're on the fence between these two chemistries, pull your last twelve months of scrap and rework data and look at what actually failed. It's rarely the adhesive that cost the most per kilogram.