Most spec engineers learn leakage classes the hard way — usually by writing “tight shutoff” on a data sheet, getting a valve that passes Class IV, then discovering the process actually needs Class VI. The standard for control valve seat leakage is ANSI/FCI 70-2 (mirrored internationally as IEC 60534-4), and it defines six allowable leakage classes — Class I being the loosest, Class VI being effectively bubble-tight. Understanding what each class means in practical terms is the difference between a working valve and an expensive RFQ to replace it.
Here’s what each class actually allows, how each is tested, and how to pick the right one for what your process needs.
What the standard actually measures
ANSI/FCI 70-2 measures seat leakage — the amount of fluid that passes through a fully-closed valve under specified test conditions. It’s not the same as fugitive emissions (stem leakage to atmosphere, covered by ISO 15848) and it’s not the same as shell leakage (body integrity, covered by API 598).
Three things drive which class a valve can achieve:
- Seat material. Soft seats (PTFE, elastomers) achieve much tighter classes than metal seats because they conform to imperfections in the plug or disc.
- Seat geometry and contact force. A heavier actuator pushing harder on the seat reduces leakage. So does a wider seating surface, up to the point that contact pressure drops.
- Test fluid. Water and air give different numbers. The standard specifies which to use for each class.
The six classes, in plain English
| Class | Max allowable leakage | What it means in practice |
|---|---|---|
| Class I | By agreement — no defined maximum. | “Whatever the buyer and manufacturer agree to.” Rarely specified intentionally. Often means “we didn’t pressure-test.” |
| Class II | 0.5% of rated valve capacity (at full open). | Loose. Acceptable for many on/off bypass services where some leakage is fine. |
| Class III | 0.1% of rated capacity. | Still loose by control-valve standards. Mostly historical. |
| Class IV | 0.01% of rated capacity. | The standard ANSI metal-seated control valve. Good general industrial service. |
| Class V | 5 × 10-4 ml/min per inch of port diameter per psi differential. (Water-test.) | Tight metal-to-metal. Used where Class IV leakage is unacceptable but soft seats can’t handle the temperature or media. |
| Class VI | Defined as bubble count per minute per inch port size, tested with air at low ΔP. Bubble-tight at the lab scale. | Effectively zero leakage. Requires a soft seat (elastomer or PTFE). The standard for clean shutoff, isolation, and downstream protection. |
How each class is tested
The test conditions are part of the spec — not just the leakage allowance. Don’t accept a “passes Class IV” certificate without confirming the test method matched the application.
Classes II, III, IV
Tested with clean water at 50–125°F. Maximum service ΔP or 50 psi, whichever is lower. The valve is closed against the rated actuator output. Leakage is collected downstream and measured by volume per unit time.
Class V
Tested with water at the maximum service ΔP (up to design rating). The result is normalized to ml/min/inch of port diameter/psi of differential. This is the only class where the test directly reflects service conditions — for the others, the test is a proxy.
Class VI
Tested with air at low ΔP (typically 50 psig, but down to 10 psig for some valves). Leakage is measured by counting bubbles in a water column downstream. The allowance scales with port size — a 4″ Class VI valve is allowed slightly more bubbles than a 1″ one. Air test is used because soft seats leak more air than water, so this is the stricter test.
Picking the right class for your service
Don’t default to Class VI “just to be safe.” You’re paying for soft seats, lower temperature ceiling, and shorter trim life. Match the class to the actual consequence of leakage:
| Your situation | Right class |
|---|---|
| Throttling control where small leakage is irrelevant (cooling water, recirculation) | Class II or III |
| General industrial on/off, metal-seated, no critical isolation requirement | Class IV |
| Tight metal-to-metal isolation at high temperature where soft seats fail (steam, hot oil) | Class V |
| Process isolation where leakage means lost product, contamination, or safety risk | Class VI |
| Anti-surge, ESD, or block valves where leakage is a safety-system input | Class V or VI per the safety case |
| Steam isolation for boiler tie-in, double-block-and-bleed, or chemical injection | Class VI (soft seats) if temperature allows, otherwise Class V with redundant block |
Common spec mistakes
- Writing “tight shutoff” instead of a class number. Different vendors interpret this differently — some default to Class IV, some Class V. Always specify a class.
- Specifying Class VI on hot service. The PTFE or elastomer seat may not survive the operating temperature. Verify the seat material’s rating, not just the class.
- Assuming Class VI = forever. Seat life is finite. For long-cycle applications, ask about expected leakage at end of seat life, not just at commissioning.
- Confusing seat leakage class with API 598 shell test. Two different tests for two different things. A valve can pass API 598 shell test and fail Class IV seat leakage.
- Specifying Class VI for a valve that physically can’t reach it. A metal-seated globe or ball valve typically maxes out at Class V. A soft-seated knife gate or resilient-seated butterfly can hit Class VI. Match the class to a valve type that can actually deliver it.
Class VI doesn’t mean zero leak forever
This is the single most common misconception. Class VI is a test condition, not a service guarantee. Soft seats wear. Cycle count, thermal cycling, particulate, and chemistry all degrade them. A Class VI valve that’s been in cycling service for two years may be effectively Class IV.
For applications where leakage truly cannot drift, the right answer isn’t just Class VI — it’s Class VI plus a maintenance program that includes scheduled seat replacement, OR a double-block-and-bleed arrangement where two valves in series with a bleed line in between let you verify each one is sealing.
How to write a clean spec
A clean leakage spec on a data sheet includes four elements:
- The class number (e.g., “Class V per ANSI/FCI 70-2”).
- The test fluid and conditions if non-standard (“tested at maximum ΔP with water at 200°F”).
- Required test certificate on a per-valve basis vs. type-test acceptance. Type-test acceptance is cheaper; per-valve testing is what you want for critical service.
- Re-test interval if the application is critical (“Class VI verified at commissioning and every 24 months thereafter”).
That four-line spec eliminates 90% of the leakage-class confusion that shows up later on RFQs.
The bottom line
Leakage classes aren’t marketing tiers — they’re specific, testable allowances with specific test methods. Class IV is a perfectly good metal-seated standard for most on/off industrial service. Class V buys you tighter metal-to-metal sealing at the cost of higher actuator torque. Class VI buys you bubble-tight performance but requires a soft seat that imposes temperature and life limits.
Pick by consequence-of-leakage, not by reflex. And when you do specify, write the class number, the test conditions, and the certification requirement on the same line.
If you’ve got a control or isolation application where leakage class matters and you want a second set of eyes on the spec, send us the conditions. We’ll cross-check the class against the available seat materials, temperature, and cycle requirements before the valve ships.