Seal material selection for solenoid valves: NBR, EPDM and FKM

Seal material is usually picked from a chemical compatibility table, and that table is the weakest of the three inputs. Temperature and compression set decide whether the seal still pushes back after a thousand hours, and a 70 Shore A NBR seal that is perfectly compatible with the fluid will still weep at 110 °C because it has taken a 30 % set and lost most of its squeeze.

Choose the compound for the temperature first, then for the fluid at that temperature, and only then check the extrusion gap. Doing it in the other order produces a seal that passes the compatibility test and leaks in the field.

Why this happens

A valve seal has three jobs at once. It has to stay elastic, it has to survive the media, and it has to resist being pushed into the clearance between the seat and the body. Each job is governed by a different property of the compound, and the properties trade against each other in every formulation.

Elasticity is measured as compression set, not as hardness. The standard test compresses a button by 25 % and holds it at temperature for 70 hours, then releases it and measures how much of the deflection it failed to recover. A compound with 20 % set returns 80 % of its deflection. Take a 2.0 mm cord sitting in a 1.6 mm groove, which is 0.4 mm of squeeze or 20 %. A 30 % set means the seal permanently loses 0.12 mm of that 0.4 mm, and the remaining squeeze is 0.28 mm, or 14 %. That still seals a static joint at low pressure, and it is already marginal for gas at 6 bar.

Temperature limits are not single numbers. NBR runs continuously to about 100 °C and briefly to 120 °C. EPDM reaches 150 °C. FKM reaches 200 °C, with short excursions to 230 °C. Life falls steeply with temperature: as a rule of thumb for elastomers, every 10 K above the rated temperature halves the service life. Running NBR at 110 °C instead of 100 °C therefore takes roughly half the life, which turns a 5 year seal into a 2 year seal without any visible change in the material.

Fluid resistance is temperature dependent, and the tables are written at room temperature. NBR is good in mineral oil, grease and cold water, and poor in ketones, esters, ozone and brake fluid. EPDM is excellent in water, steam, glycol and brake fluid, and poor in mineral oil. FKM is good in acids, fuels and hot oils, and poor in hot water, steam, amines and ketones. A rating of “good” measured at 20 °C says very little about the same pair at 90 °C, where reaction rates are several times higher.

Swell is the second-order effect that decides the fit. Immersion testing usually reports volume swell after 70 hours at a set temperature. Below 10 % swell the seal behaves as designed. Between 10 % and 20 % the groove has to have room for it. Above 20 % the compound is wrong for the duty, and the swelling shows up as a plunger that will not move rather than as a leak.

The extrusion gap converts pressure into a cutting force. Pressure acts on the seal and pushes it toward the clearance. A 70 Shore A seal at 10 bar tolerates a gap of about 0.2 mm. At 100 bar the same seal needs a gap below 0.05 mm, or a backup ring, or a harder compound. This is why high-pressure valves use 90 Shore A seals with a PTFE backup ring, and why the same seal that works at 6 bar fails at 60 bar with no change in chemistry.

Check these in order

1. Measure the seal temperature, not the fluid temperature. A coil at 80 °C inside a 40 °C cabinet heats the body, and the seal sits 20 mm from the winding. Thermocouple the seat area with the valve energised for its full duty cycle. The number is routinely 15 K to 25 K above the fluid.

2. Set the temperature class from that number. Up to 100 °C, NBR is the default and the cheapest correct answer. Between 120 °C and 150 °C, EPDM or HNBR. Above 150 °C, FKM. Below -20 °C, check the low-temperature grade rather than assuming the family covers it, because standard FKM stiffens near -20 °C while NBR still flexes at -40 °C.

3. Check compatibility at the service temperature. Ask for the immersion data at the temperature the valve runs at, not the 20 °C column. If the supplier only has room-temperature data, run a 70 h immersion at the duty temperature; it costs one seal and one week.

4. Quantify the swell and convert it to squeeze. A 2.0 mm cord swelling 10 % by volume grows about 3.2 % linearly, which adds 0.06 mm and takes a 20 % squeeze to 23 %. That is usually harmless. A 25 % volume swell adds 0.15 mm and takes the squeeze to 27 %, which raises seat load and can hold a small plunger shut.

5. Check compression set at the duty duration. The 70 h test is a screen. At 1000 h, set values commonly run 1.5 to 2 times the 70 h figure on the same compound. A seal with 18 % set at 70 h can be at 33 % after a season, and that is the number that decides whether the valve still seals in month eight.

6. Check the extrusion gap against the pressure. Measure the actual diametral clearance hot, because an aluminium body and a brass seat grow differently. At 10 bar, 0.2 mm is tolerable for 70 Shore A. At 100 bar, budget 0.05 mm or add a backup ring.

7. Check groove fill. The seal volume plus its expected swell should stay below about 85 % of the groove volume. Above that, thermal expansion has nowhere to go and the seal behaves like a solid, which transfers load into the seat and the plunger.

8. Check the counterface finish. A dynamic seal wants 0.2 µm to 0.4 µm Ra. Polishing to 0.05 µm removes the micro-pockets that hold lubricant, raises friction, and produces a seal that is leak-free on the bench and sticky after a week.

What actually to change

If set is the failure, change the temperature or the class, not the grade within the class. Moving from one NBR compound to another buys perhaps 5 K of rating. Moving the coil 10 mm away from the seat, or potting the body with a metal path to the frame, buys 20 K. Measure the seat temperature before and after, because the second option is usually cheaper than an FKM seal and does not change the chemistry.

If swell is the failure, change the family or enlarge the groove. Reducing squeeze to make room for a swelling seal trades an external leak for an internal one. A groove 0.2 mm deeper and 0.1 mm wider costs nothing in a machined body and buys 15 % of volume, which covers a 12 % swell without touching the compound.

If extrusion is the failure, change hardness before pressure rating. A 90 Shore A seal in the same groove, with a PTFE backup ring on the low-pressure side, moves the working pressure from 10 bar to 100 bar. The seat load rises with hardness, so check that the return spring still opens the valve at the minimum supply voltage.

If cold start is the failure, the compound is only half the answer. NBR at -30 °C stiffens enough to take 5 times the force to deform. A low-temperature NBR or a silicone seal helps, and so does a seat geometry that gives the seal somewhere to move rather than requiring it to flex in place. Silicone is the widest temperature range available and the weakest mechanically, with low tear strength and poor oil resistance.

If attack appears only above 80 °C, treat it as a temperature problem with a chemical symptom. NBR hydrolyses in hot water, and the visible result is a cracked, hardened seal that looks like a chemical incompatibility. FKM is not the fix in water: it dehydrofluorinates in hot water and steam above roughly 100 °C. EPDM is the compound for hot water, and the design then has to live with its poor oil resistance.

When it IS the harder problem

The valve sees hot water above 90 °C and oil mist in the same housing. No single elastomer covers both. EPDM swells in oil, FKM degrades in hot water, and NBR fails both at that temperature. The workable answers are a split seal arrangement, with an EPDM media seal and an NBR or FKM actuator seal separated by a diaphragm, or a PTFE seal with an elastomer energiser. That is a body redesign, not a compound change, and it should be decided before the housing drawing is released.

The seal is also the seat. Then the material has to seal statically and survive a million impacts on the same face. Rebound and tear strength matter as much as set, and a 70 Shore A NBR seat will take a permanent flat at the contact line within 200,000 cycles. A 90 Shore A compound, or a separate bonded seat in PEEK or PTFE, is the usual answer, at the cost of higher seat load.

The media changes in service. A valve that runs product at 20 °C, then caustic at 80 °C, then a rinse, sees three different environments in one shift. Testing each fluid separately misses the sequence effect: a seal that swells in one medium and dries in the next develops surface cracks along the compressed face. Run the sequence, 200 cycles of it, and inspect the seal under 10x magnification.

The customer names a compound but not a grade. Two NBR compounds can differ by a factor of two in compression set at 100 °C. Ask for the data sheet with the set figure at the service temperature and the swell figure in the actual medium, and treat a missing number as an unknown rather than as an average.

FAQ

Is FKM always better than NBR? No. It costs three to five times as much, stiffens near -20 °C, and degrades in hot water and steam. It wins in acids, fuels and hot oils, which is a narrower set of duties than its reputation suggests.

Can EPDM be used in hydraulic oil? No. Mineral oil swells EPDM severely, and a 70 h immersion can produce volume swell above 100 %. The seal then fills the groove, the plunger cannot move, and the failure looks mechanical.

What compression set is acceptable? Below 25 % at the service temperature for a static seal, below 15 % for a dynamic one. Both figures must be quoted at the duty temperature and at a duration near the service life, not at 70 h and 20 °C.

Does a harder seal seal better? It resists extrusion better and needs more squeeze to conform. A 90 Shore A seal commonly needs 25 % squeeze against 15 % to 20 % for 70 Shore A, so the actuator and spring have to supply that extra load at the minimum supply voltage.