The email I get twice a quarter
I'm the quality compliance manager at a contract manufacturing company that specifies silicone materials for construction, automotive, and medical device clients. I review roughly 400 incoming material lots a year. In 2024, I rejected 17% of first deliveries — and almost none of those rejections were for the reason you'd guess from looking at a data sheet.
Last month a procurement lead sent me a side-by-side comparison: same durometer, same elongation, same temperature range, 40% less than what we currently pay for Dow Corning silicone sealant 789 and Dow Corning silicone 791. (We still call them dow-corning products internally, even though the brand has been part of Dow since that 2016 restructuring. Old habits.) Can we switch, she asked.
I've fielded some version of that question every quarter since I took this role. For the first two years, I gave the standard answer: if the data sheet matches, run a trial and check the results.
Then Q3 2024 happened.
We lost 11 days on a commercial glazing project because a batch of "equivalent" sealant didn't cure within the window printed on its own data sheet. The crew showed up, found the joints still tacky, and waited. Twice. Third-party testing later confirmed the material was technically compliant — its published cure time was validated at 23°C and 50% relative humidity. The job site that week was running 31°C and 82% RH. Nobody lied on the data sheet. The material just wasn't tested under conditions that resembled ours.
That delay cost $22,000 in crew standby and a missed completion bonus. The sealant we'd switched away from would have cost about $1,800 more across the entire job.
I didn't fully understand the difference between spec compliance and application performance until that project. Now I do, and it's changed how I review every incoming lot.
What a data sheet actually measures
A data sheet is a snapshot taken under controlled lab conditions. It tells you what the material did once, in one place, under one set of circumstances. It doesn't tell you what it'll do on a humid Tuesday in July when your crew is already three days behind and the substrate is 15 degrees warmer than the lab standard.
For sealants specified to ASTM C920 — the standard covering elastomeric joint sealants — the published numbers (tack-free time, full cure, adhesion) are useful starting points. But the variables that actually determine success on a job site are substrate temperature, surface moisture, joint movement during cure, and ambient humidity. A sealant that hits its numbers at 23°C/50% RH can behave very differently at 31°C/82% RH. That's not a defect. It's chemistry.
The problem is that when you compare two data sheets side by side, you're comparing lab results, not field behavior. It's tempting to think "same durometer, same performance." But durometer is a single point on a curve that changes with temperature, time, and cure state. The gap between the data sheet and the job site is where projects go sideways.
The consistency problem nobody specs
Here's a thing I've learned the hard way: "within spec" doesn't mean "every batch behaves the same."
A durometer range of 40–50 Shore A means any batch landing in that window passes. But if your process was tuned for a batch that measured 42 and the next lot comes in at 49, you may see different flow behavior, different compression set, different adhesion. The spec allows both. Your application may not.
This matters most where the material is doing something precise. O-rings are the classic example. If you're sizing a gland with an o-ring gland calculator, the math assumes a specific compression ratio and a specific material response. But the calculator doesn't know whether your silicone batch is at the soft end or the hard end of its spec range. And it definitely doesn't know how that batch will behave after 1,000 hours at 100°C. Compression set — the property that quietly determines whether your seal still seals at month eighteen — is usually reported as a single number from a test like ASTM D395, run for a fixed duration at a fixed temperature. It's a snapshot taken under conditions your application may never repeat. And because compression set varies with batch and cure state, the data sheet number tells you less than you'd think about the number that matters later.
Same story in construction. People ask "can you stack foam board insulation" and focus on R-value, thickness, and fasteners. But the joints between layers are where performance actually gets lost, and that's usually a sealant decision. A sealant that's technically within spec on adhesion but slightly stiffer than your process expected may not accommodate the movement between stacked boards across a freeze-thaw cycle. The data sheet says it passes. The wall assembly says otherwise.
Medical applications are even less forgiving. Silicone components in breast pump assemblies — diaphragms, valves, and seals that contact expressed milk — need to meet biocompatibility requirements per ISO 10993 and extractables limits batch after batch. A "spec-equivalent" material that passes on paper but varies in post-cure volatile content isn't equivalent at all when you're filing for regulatory clearance or answering a customer audit. And in assemblies where silicone seals against PP (polypropylene) housings, the interaction between the two materials' thermal expansion matters as much as either material's individual spec.
What it actually costs when "equivalent" isn't
Here's the surprise, at least for me: the price difference on the material is almost never the expensive part. I pulled three years of lot rejections and field failures traceable to material behavior, and the pattern held across construction, automotive, and medical projects:
- Rework, standby labor, missed deadlines, and expedited replacement shipping ran 3–8x the material price difference.
- Failures clustered around time-dependent properties — cure speed, compression set, adhesion after aging — not the headline numbers on the data sheet.
- Every single failure involved a material that met its published spec.
And when you're on a deadline, the math gets even starker. A rush replacement order of a reliable product costs a premium. A rush replacement order of an unreliable one costs a premium plus another delay. In March 2024, we paid $400 extra for guaranteed next-day delivery because the alternative was missing a $15,000 event window. The same logic applies to materials. In a time-constrained project, "probably on time" and "probably within spec" are the most expensive phrases in the procurement vocabulary.
What I do differently now
I still review every incoming lot. But the criteria have shifted.
First, I ask for application-specific data, not catalog data. For sealants, that means cure behavior at the temperatures and humidities we actually work in — not just the lab standard conditions in the product literature. For O-rings, it means compression set curves over time and temperature, not a single pass/fail number. For medical-grade silicone, it means batch-level extractables data and a documented post-cure process. As of January 2025, we've got fourteen months of lot-level data on this approach, and the rejection rate on first deliveries has dropped to under 6%.
Second, I treat batch consistency as its own specification. If a supplier can't show me lot-to-lot variance data, I assume the variance exists and plan for it. That might mean adding a buffer to cure schedules or specifying a wider joining tolerance. Either way, it's cheaper than pretending the variance doesn't exist.
Third — and this one took me the longest to accept — I budget for certainty. When a project has a hard deadline or a high cost of failure, I'd rather pay 10–15% more for a material with a verifiable track record than save that amount and carry the risk myself. That isn't brand loyalty. It's math.
The procurement lead from last month didn't love my answer. But she also didn't have to explain a second 11-day delay to a client. In my experience, that's the trade-off that actually holds up.
Leave a Reply