- Specifying the right leakage class for a valve is one of the most consequential decisions in valve engineering — over-specifying tight shut-off drives up valve cost, increases operating torque, shortens seat life, and creates maintenance headaches, while under-specifying allows leakage that can cause process contamination, safety incidents, and environmental violations.
- ANSI/FCI 70-2 defines six leakage classes (I through VI) for control valves, while API 598 covers on/off isolation valve testing — these standards use different test methodologies, different acceptance criteria, and are not interchangeable, a distinction that trips up even experienced engineers when specifying mixed valve populations.
- Metal seat valves cannot achieve the same leakage rates as soft seat valves under standard test conditions, but metal seats are essential for high-temperature, high-pressure and fire-safe service — understanding the trade-off between leakage performance and seat durability is fundamental to intelligent valve specification.
Why Leakage Class Matters More Than Most Engineers Appreciate
Ask a process engineer what leakage class they want on a valve and the instinctive answer is almost always “bubble-tight” or “zero leakage” — the tightest available. This sounds prudent but is actually poor engineering. Specifying tighter shut-off than the application requires costs more money (tighter leakage classes require lapped metal seats or precision-moulded soft seats with tighter tolerances), reduces valve life (tighter seats wear faster under cyclic loading and are more susceptible to damage from particles in the flow stream), increases operating torque or thrust (particularly for ball valves and gate valves where seating load determines seat tightness), and can actually compromise process performance when the harder seating creates valve control problems.
The correct approach is to determine the maximum acceptable leakage rate for each application and then specify the lowest leakage class that satisfies that requirement. This requires understanding what each leakage class actually means in practice — not just a label on a data sheet, but a quantified rate of fluid passing through a nominally closed valve under defined test conditions.
ANSI/FCI 70-2: The Control Valve Leakage Standard
ANSI/FCI 70-2 (Control Valve Seat Leakage) defines six leakage classes for control valves and establishes the test procedures for determining which class a valve meets. The standard applies primarily to globe-style and rotary control valves, though the classes are frequently referenced in other contexts.
Class I is the no-test class — it simply means that the manufacturer and purchaser have agreed that no seat leakage test is required. This is appropriate for certain service conditions where leakage through the closed valve is acceptable or where the valve will never be in a fully closed position in normal service. It does not mean the valve has any particular leakage rate, only that it was not tested.
Class II allows leakage of 0.5 percent of rated valve Cv at full open, measured with the test fluid (air or water) at the rated differential pressure. To give this some physical meaning: a 50 mm globe control valve with a full-open Cv of 30 US gpm per psi^0.5 can leak up to 0.15 Cv worth of flow in Class II. In water service at 1 bar differential, that is approximately 0.15 litres per minute. For most continuous process applications where the valve is a flow control element and is rarely at fully closed position, Class II is perfectly adequate and is the default for many cast body globe valves with metal seats.
Class III tightens this to 0.1 percent of rated Cv — an order of magnitude improvement over Class II relative to valve capacity. Class IV is 0.01 percent of rated Cv. These classes require closer manufacturing tolerances and more careful seat finishing but are still achievable with metal seats in globe valves and are appropriate for flow control service where the valve is occasionally closed and leakage through the closed valve has a measurable process impact.
Class V is defined differently: it specifies an absolute maximum leakage of 5 x 10^-4 mL per minute per mm of seat diameter per bar differential, with water as the test fluid and at the maximum rated differential pressure. This is an extremely tight requirement. For a 50 mm seat diameter at 40 bar differential, Class V allows approximately 1 mL per minute maximum — achievable only with precision-lapped metal seats or high-quality soft seats and requiring careful manufacturing quality control. Class V is specified for isolation duties within process systems where leakage would cause product contamination, batch quality failures, or where the valve serves as a secondary containment barrier.
Class VI — bubble-tight — is the tightest classification and is achievable only with soft-seat valves (PTFE, elastomeric, or similar compliant seat materials). The test is performed with air as the test medium at 3.5 bar (50 psi) differential, and the acceptance criterion is expressed as a maximum number of bubbles per minute from a calibrated bubble tube, dependent on valve size (ranging from 1 bubble per minute for 25 mm valves to 45 bubbles per minute for 200 mm valves). Class VI is the default specification for on/off process valves in gas service, chemical dosing, and any application where leakage through a nominally closed valve is simply unacceptable.
API 598: Valve Testing and Inspection for Isolation Valves
API 598 (Valve Inspection and Testing) is the standard that governs leakage testing for gate, globe, ball, plug, check, and butterfly valves used in the refining and petrochemical industries. It is not a leakage class standard in the same sense as FCI 70-2 — instead it specifies what tests must be performed and what the acceptance criteria are for production acceptance testing of individual valves.
API 598 requires two types of test for each valve. The hydrostatic shell test confirms the pressure integrity of the valve body and bonnet — the valve is filled with water (or air or inert gas for certain valve types) and pressurised to 1.5x the rated pressure at 38 degrees C, and there must be no visible leaks from the body or bonnet. The seat (or closure) test checks that the valve seals when closed. For resilient-seated (soft seat) valves, the acceptance criterion is no visible leakage at the test pressure with air or water as the test medium. For metal-seated valves, API 598 allows a defined leakage rate that varies with valve type and size.
The critical distinction is that API 598 metal seat acceptance criteria are substantially more lenient than FCI 70-2 Class IV or V. For a metal-seat gate valve tested to API 598 in water at the rated differential pressure, the allowed leakage is measured in drops per minute or millilitres per minute, and for large bore valves can be several hundred millilitres per minute. For a metal-seat ball valve, API 598 allows up to 10 mL per minute per inch of nominal bore at the rated pressure for liquid test, and double that for gas. These are not tight shut-off rates by most process standards.
This is why specifying “test to API 598” for a valve that is intended to provide tight process isolation against a flammable or toxic fluid is insufficient. API 598 compliance confirms that the valve meets the minimum acceptable leakage rate for its type — it does not guarantee Class IV, V, or VI performance. If you need tighter shut-off on a valve that is tested to API 598 as a production test, you must specify the additional leakage requirement separately.
Metal Seats vs Soft Seats: The Leakage Performance Trade-Off
Soft seat valves — those with PTFE, elastomeric, or other compliant seating elements — can achieve Class VI (bubble-tight) leakage routinely. The compliance of the seat material allows it to conform to minor surface irregularities on the mating closure element, providing a hermetic seal even with moderate surface finish imperfections. This is enormously useful in manufacturing and quality control terms: soft seat valves can be produced to Class VI at relatively low cost because the seat geometry and surface finish tolerances do not need to be as tight as for metal seats.
The price for this leakage performance advantage is temperature and mechanical durability limitations. PTFE seats become soft and begin to creep above approximately 200 degrees C; at 260 degrees C they are too soft to maintain reliable seating. Elastomeric seats (EPDM, Viton, Buna-N) are generally limited to 180 degrees C and lower in aggressive chemical service. In fire-safe design, soft seats are assumed to be destroyed in a fire scenario — the fire-safe test (API 607, API 6FA) assesses the metal-to-metal leakage performance of the valve after the soft seat materials have been burned away.
Metal seat valves — hardened stainless, Stellite, Inconel, ceramics — can operate at temperatures up to 650 degrees C and above, handle abrasive media that would destroy soft seats rapidly, and survive fire scenarios with their sealing integrity substantially intact. The trade-off is leakage: metal-to-metal seating requires tight dimensional tolerances and surface finish (typically Ra 0.4 to 0.8 micrometres on both mating surfaces, with lapping to achieve matching profiles) to achieve leakage rates equivalent to Class IV or V. Class VI is generally not achievable with metal seats under standard test conditions, though some manufacturers claim Class V equivalent performance on hardened metal seat ball valves with precision-lapped mating surfaces.
Specifying Leakage Class for Different Applications
The starting point for leakage class specification is the consequence of leakage through the closed valve. Consider the following application categories.
For process isolation valves on flammable hydrocarbon service where leakage through a closed valve could contribute to a fire or explosion risk, tight shut-off is justified. Soft seat Class VI if temperature permits; metal seat API 598 leakage class with additional requirement for Class V equivalent performance by agreement if temperature or fire-safe requirements preclude soft seats. The consequence of leakage justifies the additional cost.
For flow control valves in continuous duty service, the valve is rarely fully closed and leakage through the closed valve is not the primary design concern. Class II or III metal seat is typically appropriate and allows the use of cost-effective cast body globe valves with standard seat finishing. Specifying Class V or VI on a flow control valve in this service is over-engineering.
For steam isolation valves, metal seats are mandatory above about 200 degrees C, and the leakage specification depends on the consequence: a steam supply valve to a critical process can justify Class IV metal seat with certified test results; a drain valve or non-critical isolation can be specified to standard API 598 metal seat acceptance criteria.
For double-block-and-bleed arrangements in custody transfer metering, flare isolation, and toxic service, the combined leakage of both block valves and the bleed valve must be considered. Both block valves must provide Class V or VI performance to ensure that the bleed cavity between them can be opened safely. This is a situation where the leakage specification interacts with the system design and process safety assessment.
Testing Procedures and What They Actually Measure
Understanding what the leakage test actually measures is essential to interpreting the results correctly. FCI 70-2 Class V and VI tests are performed at specific conditions: Class V at maximum rated differential pressure with water; Class VI at 3.5 bar (50 psi) with air. The test condition is not necessarily the operating condition. A valve that achieves Class VI at 3.5 bar test pressure may or may not achieve Class VI at your operating pressure of 30 bar — the seating geometry and seat stress at higher differential pressures are different from the test condition.
For critical applications, specifying that the leakage test be performed at the actual operating differential pressure — not just the standard test pressure — gives more meaningful results. This is more expensive (requires higher-pressure test rigs) but eliminates the uncertainty of extrapolating low-pressure test results to high-pressure operating conditions.
The test medium also matters. Air and water tests do not always give equivalent results. Gas leakage through a valve seat is governed by different physics from liquid leakage — a seat that shows no visible leakage in a water test can still pass measurable gas leakage when tested with air or nitrogen. For gas service isolation valves, specifying a gas seat leakage test (pneumatic test) rather than a liquid test gives more relevant results. API 6D (Pipeline Valves) requires both liquid and gas seat tests for pipeline ball valves specifically for this reason.
Common Specification Mistakes
Specifying “zero leakage” on a metal seat valve is a common and problematic practice. Metallurgically, zero leakage through a metal-to-metal contact is not achievable — there will always be some molecular transport through the contact zone, particularly in gas service. The meaningful specification is the maximum acceptable leakage rate expressed in quantified terms. “Zero leakage” on a specification document creates a contractual problem when the supplier delivers a valve that meets Class V and the purchaser insists it should be “zero” — the argument consumes time and money and usually ends in a compromise that satisfies nobody.
Specifying Class VI for a valve that will never be in fully-closed service is another common error — expensive soft seats on a flow control valve that throttles between 20 and 80 percent open, never going to zero. The soft seat provides no process benefit in this service but increases valve cost and reduces trim life under cyclic throttling service.
Confusing FCI 70-2 and API 598 on the same specification document — referencing both standards without clarifying which governs for which test — creates ambiguity that manufacturers will resolve in the cheapest direction. A clear specification states: “Seat leakage test to API 598 for isolation valves; seat leakage test to ANSI/FCI 70-2 [Class IV/V/VI] for control valves; test medium [water/air/gas]; test differential pressure [bar].” This removes ambiguity and ensures the valve you receive matches the performance your application requires.
Future Developments: Digital Leakage Monitoring
Acoustic emission monitoring and ultrasonic through-valve testing are increasingly used in modern plants to detect valve leakage in service without requiring the valve to be isolated and physically tested. These techniques detect the characteristic acoustic signature of turbulent flow through a leaking valve seat — distinguishable from background noise by frequency analysis. They do not give absolute leakage rates directly, but trend monitoring can detect the onset of seat degradation before leakage reaches levels that affect process quality or safety.
For critical isolation valves on dangerous fluid services — HIPPS block valves, ESD valves, toxic isolation valves — partial stroke testing combined with acoustic seat leakage monitoring provides a powerful condition monitoring system that can verify valve operability and seat integrity without removing the valve from service. This is the direction the industry is moving: from scheduled strip-and-inspect maintenance to continuous condition monitoring that allows maintenance to be targeted where it is genuinely needed.