What Casting Quality Does to a Ball Valve: Porosity, Inclusions and Field Failure

Why stainless valve bodies are cast at all

A ball valve body has internal geometry that is expensive to reach with a cutting tool: seat pockets, a stem bore intersecting the flow bore, a cavity that has to be concentric with both. Investment casting produces that geometry close to net shape, which is why ASTM A351 grade CF8M, the cast equivalent of 316, is the standard body material rather than bar stock 316.

The trade is that a machined bar-stock body inherits the soundness of rolled material, while a cast body inherits the soundness of a pour. Everything below follows from that.

The grade matters for a second reason that returns later. Under ASTM A351, CF8M carries molybdenum at 2.0–3.0 percent alongside 18–21 percent chromium and 9–12 percent nickel. CF8, the cast equivalent of 304, is specified to a maximum of 0.50 percent molybdenum. That single element is the difference in chloride pitting resistance, it is the difference in alloy cost, and it is measurable in seconds with a spectrometer.

How gas porosity forms

Hydrogen dissolves readily in liquid steel and far less readily in solid steel. During solidification of austenitic stainless the solubility drops sharply, and the great majority of the dissolved hydrogen is rejected at the advancing solidification front.

If the melt is still liquid around it, that gas escapes. If the solidifying shell advances faster than the gas can move, it is trapped as a rounded void. The result is gas porosity, and it is distributed by thermal history rather than randomly.

Shrinkage porosity has a different cause and a different signature. Steel contracts as it freezes, and any section that solidifies last without a feed path draws a void into itself. Shrinkage tends to appear in thick sections and at junctions, which on a valve body means flange hubs, the boss around the stem bore, and the transition into the seat pocket.

Inclusions are a third family: slag, refractory particles or oxide films entrained during pouring. They are not voids, but mechanically and electrochemically they behave much like them.

All three are governed by melting practice, thermal control, mould quality and part design. They are process outcomes, which is the reason they are controllable and the reason they recur when the process is not controlled.

What a defect does to a finished valve

This is where casting quality stops being a foundry concern and becomes a valve problem.

It becomes a corrosion initiation site. Porosity and inclusions break the passive film that gives stainless its corrosion resistance. Under a stagnant or chloride-bearing fluid, a pore is a crevice, and crevice and pitting attack begin at exactly such discontinuities. In a closed cooling loop or a potable water line, this is the mechanism that turns a valve that passed every factory test into a weeping valve two years later.

It becomes a leak path across the seat. Machining removes material, and a void that was subsurface in the raw casting can be opened at the seat pocket face. A seat needs a continuous bearing surface; a pore in that surface is a bypass that seat load cannot close.

It reduces the pressure boundary. A cluster of shrinkage in a flange hub or wall is material that is not there. ASME B16.34 sets a minimum wall thickness for the pressure boundary, and porosity is material missing from inside that wall.

Why the pressure test does not catch it

This is the point most often misunderstood on both sides of a purchase order.

An API 598 shell test proves that the pressure boundary holds at test pressure, on the day it is tested, at ambient temperature, with clean water. A subsurface void with sound material over it does exactly that. The valve passes, and the record says so.

The shell test is run above the valve's rating, at 1.5 times the rating at 38°C under API 598, and held for a defined period. That is a real proof of leak tightness at delivery. It is also a short hydrostatic load applied once, against a mechanism that operates over years of service.

What the test cannot demonstrate is that the pressure boundary is sound throughout, that no discontinuity will be exposed by erosion or corrosion, and that the seat face has no pore under the seat ring. "100 percent pressure tested" is a true and useful statement about leak tightness at delivery. It is not a statement about casting integrity, and treating it as one is how casting quality disappears from a specification.

What detects each defect

Spectrometry deserves its own line because it is regularly confused with soundness testing. It confirms that the body is CF8M rather than CF8, which is a chemistry question. It says nothing about whether that CF8M is sound.

What a buyer can specify

The base pressure-temperature standard does not require volumetric examination. ASME B16.34 distinguishes a Standard Class from a Special Class, and the Special Class carries additional nondestructive examination requirements and correspondingly higher permitted ratings. A valve bought to Standard Class has not been radiographed unless radiography was separately specified.

That leaves four things worth writing into a specification, in rough order of value:

1. Acceptance criteria, referenced to a standard. Radiographic severity is judged against reference radiographs, and which standard applies depends on section thickness. Each grades discontinuities by category, including gas porosity and sand or slag inclusions, at severity levels 1 to 5, where 1 is the least severe. Surface irregularity is judged separately against MSS SP-55. Specifying "no porosity" is not purchasable; specifying a category, a severity level and a section is.
2. Which parts, and at what rate. Radiographing every body is uneconomic for most commercial valve programmes. A defined sampling rate, or radiography restricted to critical sections such as the flange hub and seat pocket region, is the usual middle path.
3. Weld repair policy. Foundries repair castings by welding, and the question is not whether it happens but whether it is disclosed, procedure-qualified and re-examined. An undisclosed repair in a pressure boundary is a materially different product from the one specified.
4. Material traceability. EN 10204 3.1 for the heat, with the heat number traceable to the specific casting, so a defect found in the field can be tied back to a pour rather than to a shipment.

The three standards overlap slightly in severity levels for comparable categories, so a specification that names a thickness range as well as a level removes the ambiguity.

The sourcing question underneath

Every control listed above sits in the foundry, not in the valve assembly shop. Melting practice, deoxidation, pouring temperature, gating and feeding design, mould quality, and the decision to scrap rather than repair are all made before a valve body reaches a lathe.

That makes the ownership question a practical one rather than a marketing one. A valve maker with an in-house foundry sets those parameters directly and holds the melt records. A valve maker buying castings specifies them to a supplier and verifies on receipt, which can be done well, and which places the melt record in another company's system.

Both models produce good valves. They differ in how a problem is traced. When a body weeps in year two and the root cause is a shrinkage cluster at a flange hub, the in-house case can pull the heat, the pour record and the thermal data for that specific casting. The bought-in case can pull a certificate and start a conversation.

The question to ask a supplier is therefore not "do you have a foundry" but "for a valve you shipped me last quarter, can you produce the melt record for that body's heat number, and the examination record for that lot?" The answer, and how long it takes, describes the system. A capability statement describes an intention.

Frequently Asked Questions

Published by LINS Valve Industrial Co., Ltd., Taichung, Taiwan. Last Updated: 2026-08-26