Chemical Compatibility and the Checks Worth Running

Why this matters

Most of what gets called a compatibility check in the field is not one. A quick spot test, a pressure test after assembly, a hardness reading, a glance at the part after a month: all of them can pass on a pairing that is going to fail, and three of them pass most reliably in exactly the cases where the failure is worst. This card is a short list of checks that actually produce evidence, and a longer explanation of the plausible-looking checks that come off the list, because knowing what not to trust is what stops a false pass from ending the investigation.

Two different questions get called compatibility

Degradation asks whether the fluid attacks the material: swelling, softening, hardening, cracking, dissolving, extraction of an ingredient.

Permeation asks whether something passes through the material while the material stays perfectly intact. A polymer can be fully compatible with a fluid by every degradation measure and still let gas or liquid migrate straight through its wall.

These fail in opposite directions and they need different checks. Degradation shows up on the part. Permeation shows up somewhere else entirely, on a component the tubing never touched, which is why it goes undiagnosed for years. The worked case below is one of those.

The checks worth running

1. Name the material class precisely, then verify it. Not "PVC" but which compound classification, not "stainless" but which grade, not "nitrile" but which supplier's compound. Ratings are published per compound and the spread within a family is wide enough to invert a decision. This is a check, not an assumption, and it means finding the marking or the documentation.

2. Build the whole fluid list, not the process fluid. Cleaners, flushes, assembly lubricants, primers, leak-check sprays, and whatever was in the system before the retrofit. The procedure for this and for the acceptance gate that follows it lives in the companion HowTo; the check here is simply that the list exists and has more than one line.

3. Take the temperature at the material, not in the fluid. A pipe wall against a hot surface, a seal near a bearing, a gasket next to a combustion path, and a fitting in direct sun all run above the fluid temperature. Chemical attack rate roughly doubles for every 18 degrees F, so a 30 degree F error in where you took the reading is not a rounding error, it is close to a factor of three on the rate.

4. Check whether concentration inverts the answer. This is the check most people do not know exists. Some pairings are worse dilute than concentrated. Carbon steel is the standard example: concentrated sulfuric acid forms a protective film on steel and is routinely stored in steel, while dilute sulfuric attacks steel aggressively. A tech who knows the concentrated case and assumes the dilute case is milder has the relationship exactly backwards. Never interpolate a rating downward in concentration without checking; look up the dilution you actually have.

5. Check the vapor space separately from the liquid. The headspace above a liquid is often the more aggressive location, because condensing vapor is effectively distilled, it carries dissolved oxygen, and it wets and dries cyclically at the liquid line. A gasket at the top of a vessel and a gasket at the bottom are in two different services even though they are in one tank.

6. Check what the material puts INTO the fluid. Compatibility runs both directions. A material can survive a fluid perfectly while contaminating it: plasticizer migrating into a process stream, metal ions into water, extractables into anything potable. Where the fluid is drinking water, this is a hard gate rather than a judgment call, and the certification to look for is NSF/ANSI 61.

7. Check permeation as its own question wherever a gas is involved. Oxygen into a closed water loop, refrigerant through a hose wall, fuel vapor through a polymer line, moisture into a sealed enclosure. Ask what is on the other side of the wall and whether it minds.

The checks that come off the list

Each of these looks like diligence and produces a false pass. The reason each one comes off is different, and the reason is the useful part.

The flame or smoke test comes off on hazard alone. Fluoroelastomers and PTFE release hydrogen fluoride and other toxic decomposition products when overheated, and the reason you are testing is that you do not know which material you have. There is no version of this test that is safe on an unknown sample, and no result it produces that is worth the exposure.

The spot test comes off because a pass from it means nothing. Putting a drop of the fluid on the material and watching for a reaction runs at ambient temperature, for minutes, on an unstressed sample. Those are precisely the three conditions under which slow attack and environmental stress cracking do not appear. A fail from a spot test tells you something you would have predicted from the chart anyway. A pass tells you nothing at all, and its real damage is that it ends the investigation.

The post-assembly pressure test comes off because it is anti-correlated with the failure. A seal that is swelling in an incompatible fluid gains squeeze before it loses integrity, so it seals better during the early swelling phase than a correct seal does. The joint that passes its pressure test most convincingly can be the one that is being destroyed. A pressure test proves you assembled the joint correctly. It cannot say anything about the material.

The durometer reading comes off because it lags. Hardness moves late, well after absorption or extraction has begun, and general-purpose seals across every family cluster around the same hardness anyway, so the reading has little room to discriminate. Hardness change is useful as one output of a controlled immersion test with a matched control sample. It is not useful as a field screen.

Family-level lookup comes off because it is not a check. Reading a chart row labeled with a polymer family or an alloy family and applying it to the specific compound in your hand is the same act as assuming, performed with a document open. If the chart does not name the compound, the chart has not answered your question.

"It has been fine so far" comes off whenever anything changed. A system that has run for years on one water chemistry, one cleaner, and one temperature is evidence about those conditions only. A softener installed, a chemical switched, a setpoint raised, or a component replaced resets the evidence to zero, and the fact that it ran fine before is the reason nobody looks.

Worked case: the loop that ate its own iron

A closed hydronic loop was extended during a remodel using flexible polymer tubing. Two years later the circulator seized, the strainer was packed with fine black sludge, and the ferrous components in the loop showed general rust with no leak anywhere in the system.

Getting into it came first, in a fixed order. A closed loop is pressurized and hot. Isolate, kill the heat source, let the loop cool, relieve pressure to zero and confirm at the gauge rather than by feel, then drain. Where the isolating valves or the boiler disconnect could be operated by someone else while the loop is open, lock or tag them and verify the zero-energy state before breaking any joint, per 29 CFR 1910.147.

The obvious compatibility question passed, and it was the wrong question. Was the tubing compatible with the water in the loop? Yes, completely. No swelling, no cracking, no softening, tubing in perfect condition after two years. Every degradation check available would have passed it, which is why the tubing was cleared early and the investigation went looking at the circulator instead.

The permeation question had never been asked. The tubing installed was a non-barrier product. Oxygen diffuses through polymer tubing wall from the room air into the loop water, continuously, whether or not the tubing itself is affected. In a closed loop, dissolved oxygen is consumed by the ferrous components: cast iron circulator bodies, steel tanks, steel piping. Consumed oxygen means consumed iron, and the product of that reaction is the fine black magnetite sludge that was packing the strainer.

What the standard says, and what it tells you about the size of the problem. DIN 4726 sets the oxygen-barrier requirement for tubing in closed heating systems at not more than 0.1 grams of oxygen per cubic meter of system water per day at 40 degrees C. Put that against this loop's roughly 40 gallons of water. Forty gallons is about 151 liters, or 0.151 cubic meters, so a compliant barrier product may admit up to 0.151 times 0.1, which is about 0.015 grams per day, or about 5.5 grams per year.

Read that number the right way. It is the amount a compliant system is allowed to admit, and closed-loop systems are designed to tolerate it. The point is not that 5.5 grams is a lot. The point is that a barrier standard exists at all, that non-barrier tubing makes no claim against it, and that the corrosion in this loop consumed ferrous metal continuously for two years with no leak, no visible tubing damage, and nothing on any degradation chart that would have flagged it.

The correction: replace the non-barrier tubing with a barrier product for the closed loop, or isolate the ferrous side behind a heat exchanger so the oxygenated water never contacts iron. Flush the loop and replace the sludge-packed components; the sludge is consumed iron and it does not go back.

What would have caught it at install: one line in the material list asking what the tubing lets through, not just what it survives. The tubing was compatible. The system was not.

How to verify you got this right

• The material class in your record is specific enough that somebody could reorder it. If it says a family name, the check has not been done.
• Both questions are answered for every polymer in the assembly: does the fluid attack it, and does anything pass through it into somewhere that minds.
• No conclusion in the record rests on a spot test, a pressure test, a durometer reading, or a family-level chart row.
• Where a system changed, the record says what changed and on what date, so the next person knows which era the "it has been fine" evidence covers.
• Where a fluid touches potable water, the certification is on paper, not inferred from the material type.

References

• DIN 4726 for the oxygen permeation limit applicable to tubing in closed heating systems
• NSF/ANSI 61, Drinking Water System Components - Health Effects, for materials in potable water contact
• 29 CFR 1910.1200 for the safety data sheet format, including Section 10 stability and reactivity, which names incompatible materials
• 29 CFR 1910.147 for lockout/tagout and verification of zero energy before opening a pressurized, heated loop
• See related: How to Check a Material Against the Fluid It Will Touch; The Elastomers and What Attacks Each One