Seat leakage Class IV vs V vs VI — what FCI 70-2 actually tests
A requisition goes out asking for Class VI, or saying "tight shut-off", or citing ANSI B16.104 because the datasheet template still carries it. The quotation comes back from a valve shop in Korea reading Seat leakage: ANSI/FCI 70-2 Class IV. Somebody has to decide whether that difference is worth sending the enquiry back.
It usually is, but not for the reason people reach for first. The class is not a grade of valve quality and it is not a shut-off rating. It is one bench test on a valve not yet installed and not yet hot, run at a temperature the standard fixes and, for every class but V, a pressure it fixes too. Read it as a statement about your service and you size a downstream drain on a number nobody measured under your conditions.
Why your spec says B16.104 and the quotation says FCI 70-2
They are the same document under two names. The foreword says it was "formerly known as ASME B16.104" and was adopted as FCI 70-2 on 24 November 1970; older requisitions still write it ANSI B16.104. Nothing is lost in the handover: the numerals carried straight across, and a shop reading a spec that cites B16.104 will quote and test to the current FCI edition.
What does move is the edition. It was revised in 1998 to line up with the IEC work and again in 2003 to allow a low-pressure gas test for Class V; the current one is ANSI/FCI 70-2-2021. The clause numbers and figures below are from the editions now in circulation; confirm them against the edition your specification cites before the order goes out.
What each class is measured against
The six classes are not six rungs of one ladder. They are measured against three different things.
Class I is the odd one: a Class II, III or IV valve whose design intent is unchanged but where, by agreement between user and supplier, no test is required. Classes II through VI are tested on every valve, so a measured figure exists for each tag even when nothing is issued with it. Class I is where it does not.
Classes II, III and IV are percentages: 0.5%, 0.1% and 0.01% of rated valve capacity, which the standard defines as the test fluid that would pass at rated travel under the stated pressure conditions. The allowance scales with the valve: Class IV on a 300 mm body permits far more absolute leakage than on a 50 mm body. Two lines of one requisition carrying the same three words are not asking for the same leak rate.
Class V is an absolute rate tied to geometry and pressure: 5 × 10⁻⁴ ml per minute of water per inch of seat diameter per psi of differential, or 5 × 10⁻¹² m³ per second per mm per bar. A gas alternative allows 4.7 standard ml per minute of air per inch of orifice diameter. It is the class whose allowance is written directly in terms of the differential you specify, rather than through a capacity figure.
Class VI is a table, not a formula: 0.15 ml/min at seat diameters up to 25 mm, 0.90 at 76 mm, 4.00 at 152 mm, 28.4 at 400 mm, with bubble counts beside the smaller sizes.
The test conditions are most of the answer
The procedures differ more than the numbers do. Type A covers Classes II, III and IV: clean air or water at 10–51 °C, at 3–4 bar (45–60 psig) or within ±5% of the maximum operating differential, whichever is less. A valve destined for a 90 bar differential is seat-tested at roughly 3.5 bar, and clause 5.1.5 allows no credit for the extra seat load the actuator develops precisely because the test differential is the lower of the two.
Type B, the Class V water test, is the exception: the test differential is within ±5% of the maximum service pressure drop across the plug, capped at the body's room-temperature rating per ANSI B16.1, B16.5 or B16.34. Type B1 is the Class V gas alternative, air or nitrogen at 3.5 barg. Type C, for Class VI, is air or nitrogen at the maximum rated differential or 3.5 bar, whichever is least. Temperature is 50–125 °F throughout — written as 10–51 °C in the Type A clause and 10–52 °C in the others — nothing here is run hot. The standard says so of itself: the classes "cannot be used as a basis for predicting leakage at conditions other than those specified".
Reading the leakage line on a quotation
Each numeral buys something different.
| What the line says | What was actually tested | How to confirm | If you read it wrong |
|---|---|---|---|
| Class I | Nothing. Design intent of a II, III or IV valve, agreed between the parties | Ask whether a record exists for the tag at all | The shipment waits at FAT for a certificate the class never called for |
| Class II / III / IV | 0.5 / 0.1 / 0.01% of rated valve capacity, air or water, about 3.5 bar, ambient | Ask for the rated capacity figure and test pressure used, then convert to your units | A percentage reads as small, and the drain gets sized for a flow that is real on a large body |
| Class V, procedure not named | Water at your maximum service differential (Type B) or air at 3.5 barg (Type B1) | Name Type B or B1, and the differential the certificate is to be written under | Both procedures satisfy Class V: Type B1 runs on air at 3.5 barg and says nothing about behaviour at your service differential |
| Class VI, or "bubble tight" | Air or nitrogen at 3.5 bar or the rated differential, whichever is least, against the ml/min table by seat diameter | Check the seat diameter against the table row; confirm the seat material suits your fluid and temperature | "Bubble tight" reads as zero, and as a permanent property rather than a bench result |
| "Tight shut-off" or "zero leakage" | Nothing; neither phrase exists in FCI 70-2 | Replace with a class, a procedure type and a differential, or specify an isolation standard | Both parties sign a requirement with no test behind it |
Where the class stops being the right document
The standard is direct about its own boundary. During the canvass of the 2003 revision, one respondent asked that on/off valves used for tight shut-off be excluded. FCI's reply: 70-2 has always been intended for control valve seat leakage, and where line isolation or absolute tight shut-off is a normal expectation, another standard such as API 598 should be specified instead.
If the valve's job includes holding a line while someone works downstream, the leakage class is not the document that governs it; that is a separate valve or a separate test standard, on a different line of the requisition. FCI says it plainly in its own material on 70-2: control valves are not isolation valves, and they will leak. The class is a benchmark measured on a new valve at the bench, not a property it keeps in service.
Where the duty is genuinely a control valve held closed against a high differential for long periods — the case Class V was written for — the answer moves into trim and seat construction rather than into the standard. Kent Introl works at that end of the range.
Turning a class into something a Korean shop can quote and test
A Korean specification may cite KS C IEC 60534-4 where a US one cites ANSI/FCI 70-2. The class numerals line up, but the two documents are not identical — FCI 70-2 makes no provision for Class IV-S1 and does cover valves below a Cv of 0.1 — and they revise on separate clocks, so say which line the certificate is to be written under. Past that, what is worth writing down is the part the standard deliberately leaves open — five items:
- The class, and which line you mean — ANSI/FCI 70-2 or IEC 60534-4 / KS C IEC 60534-4 — with the edition if your spec pins one.
- For Class V, the procedure type. Type B and Type B1 are both compliant and are not the same test.
- The test differential in bar. For Type B the default is your maximum service pressure drop; section 5.2.2 also allows a lower differential by agreement between the parties, so write the figure you want tested into the requisition; it is one of the few numbers the standard leaves to the two parties.
- Whether the test is witnessed, and by whom. This moves the schedule more than the other four — it fixes a date the valve must be ready by.
- What comes back: a record per tag with serial number, class, procedure type, test medium and measured value. The measurement is made on every Class II to VI valve, so the figure exists; what decides whether it ships with the valve is the documentation schedule, so list it there.
Which classes a given body and trim can reach is a model-level answer; valve pages such as KOSO 501G and 550G are where that starts. For an installed base, the class on the original FAT certificate describes a new valve; after an overhaul it has to be tested again to be claimed, which is why a seat leakage record belongs in the acceptance documents for repair work.
The misread that costs the most
It is not the one about test pressure. It is treating I to VI as a single ladder and climbing it when something goes wrong.
A valve passes more than the process can live with, so the next requisition asks for Class VI where the last one said Class IV. Class VI is the class the standard associates with resilient seating — single-seat trim with O-rings or similar gapless seals — although section 2.1 says class selection is not restricted by valve design. In practice a shop meeting the Table 2 figures at a large seat diameter will usually reach for a soft seat, and that is where the change starts. The trim now carries a temperature ceiling and a fluid compatibility list the metal-seated Class IV did not have, and if the valve throttles at low opening the soft element sits in the fastest part of the flow path. The certificate says Class VI. Eighteen months later it passes more than the Class IV it replaced.
The second version is quieter. Class V is not a property of body and trim alone: the Type B procedure is run with the net actuator thrust set at the specified maximum, and anything above that is expressly not allowed. The class travels with the actuator and its supply pressure. Install the valve on an instrument air header running lower than the datasheet figure and the bench result is not reproducible on site — the seat is fine, the thrust is not. Check the assumed supply pressure on the actuator datasheet against what the plant delivers, on KOSO Hammel Dahl valves as on any other, before anyone concludes the seat failed.
Send a tag list with the maximum shut-off differential and the class you need, and you get back which models can be built to it and which procedure the certificate would be written under — get in touch.