"Overrule" Safety Automation; Minimum Control Valve Size

A Reader Asks Our Experts to Explain "Underwater Nuclear Reactors" and "Overrule Safety." Plus, What's the Minimum Control Valve Size in an Oil Pipeline?

By Bela Liptak

2 of 3 1 | 2 | 3 View on one page


A:  I do not believe that the requirement of API 553 is intended to apply only to control valves. It seems to me that in order to guarantee the mechanical strength of all pipelines, all weak points should be eliminated, no matter if they are caused by inserting excessively small valves or any other undersized in-line device (flowmeter, filter, etc.). I believe that the standard should say: "Inline equipment size must not be smaller than two sizes below the size of the pipeline."

Before accepting the pipe size, you should consider the possibility that it is oversized and should be reduced. In the past, I have often found that when I sized a valve, and it came out to be more than two sizes below that of the pipe, that was a hint that somebody goofed and oversized the pipe. Therefore, you should also check if the pipeline was properly sized (not oversized). In other words, check not only the maximum flow, but also the pressure drop assigned to the valve, which, if unnecessarily high, not only reduces the valve size, but also wastes a lot of valuable energy.

Béla Lipták.

A: I do not have a copy of API 533, but my interpretation is that API is concerned about the risk of creating a weak point in the piping system if the valve is too small. This can happen with high-pressure-drop control valves. The sizing requirements for the valve can result in a valve which is significantly smaller than the adjacent pipe. It is not unusual to have a required valve size of only 4 in. to control flow in a 12-in. pipe, or to control the flow of gas/steam in a 16-in. pipe.

Installing a 4-in. valve in a 12-in. pipe system will result in very high bending loads at the connection between the valve and the pipe and on the valve itself. The consequences can include flange leakage, premature failure of weld joints or severe sticking of the valve due to distortion of the valve body.

ASME B31.1 provides methods for calculating pipe loads. Pipe engineers and valve engineers can use these calculations to assess mechanical integrity. There are a couple of options for solving problems:

  • Make the valve larger, i.e., provide a 12-in. valve for 12-in. pipe.
  • Use a reduced trim size inside a larger valve; i.e., 4-in. trim inside a 12-in. valve. Small trim size may be needed in some cases to optimize flow control and turndown.
  • Apply a special valve design that incorporates suitable pipe expansions and wall thickness reinforcement to improve mechanical integrity; i.e., use 4-in. trim inside an 8-in. valve body with 8-in. x 12-in. expanders that are of heavier schedule than adjacent pipe. In my experience, this is quite common in steam letdown and steam conditioning systems.

Stephen Freitas

A: I was a member of the API Committee on the Refinery Equipment Subcommittee on Instruments a long time ago, and 553 sounds familiar. It is probably a recommended practice. You have a valid question and one not answered in the standard or specification.

These standards and recommendations advise against control valves much smaller than the pipe size. The reason is that the body of such a valve can be expected to experience mechanical stresses far beyond what it can handle, and this can result in body failure. (Installers are notorious for casually and commonly forcing pipe into alignment with little concern for attached equipment).

The usual reason for discovering that a small valve is sufficient is because the pipeline is far oversized.

For many styles of valves, the manufacturers can provide the required reduced trim size in larger valve bodies and solve this problem. This cost is usually far less than replacing the large pipe. It also might be a good idea to verify the provided flow data before purchasing the valve. People do make mistakes. Density and specific gravity are often confused, dimensions copied in error and so on.

Cullen Langford

A: To provide an explanation, let me assume that your pipe line size is 12 in. (DN300), and you intend to use a 6-in. (DN150) valve. Let us also assume that the velocity of liquid in the pipeline is 2 m/sec. If you install a reducer DN300 x DN150, the velocity in the reduced pipe section (DN150) will be approximately 5 m/sec. If it is a ship-loading or unloading pipe, then normal velocities will be much higher (say 4 m/sec) and, hence, the velocity in the reduced pipe section will be ~ 18 m/sec. This is far too high a velocity for a pipe, and the reducer/welded section will be eroded by the liquid velocity, thus weakening the mechanical integrity of the pipe. This is why API 553 does not allow more than two pipe sizes reduction.

Raj Binney

A: The requirement is a valid one. Going down more than two line sizes will mean that the weakest part of the pipework is at the control valve attachment. The pipe size has been predicated on a velocity constraint, and reducing the attachment from, say, NB6 to NB3 (two sizes down) will give four times the velocity in the attachment. NB6 to NB2 (three sizes down) will be roughly nine times the velocity. Erosion in pipework is a high power function of velocity (perhaps 7th order). Therefore, the requirement to keep body size up is valid.

Exit velocity from a control valve body in liquid service is normally recommended to be no higher than 10 m/s (better, 5 m/s). The trim velocity to remove excess pressure is independent of the body velocity. Anti-cavitation trim, with multiple pressure drop stages to avoid dipping into the cavitation region, has very high velocities and requires special materials.

2 of 3 1 | 2 | 3 View on one page
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


  • <p>I almost always agree with Bela Liptak, but I must take exception to one of his "overrule safety" solutions to the nuclear power plant problem. You cannot isolate a nuclear power plant from ANY external data communications. I seem to recall an NRC requirement for "remote operation" of a nuclear power plant in case the local control center becomes damaged or is otherwise inoperable. The requirement was for that plant to be operated from a distant location sufficient to regain control and safely operate it or shut it down in an orderly manner. This does not require an Internet connection, but it is a communications line out of the plant.</p> <p>I have often heard people exclaim that there should be no internet connections to the process control network, as a solution to the potential for control systems being "hacked." Well, that didn't protect the Iranian uranium enrichment plant from the Stuxnet virus that was probably planted into the operating system of the Siemens System 7 at least a year before it was shipped. These days, it is unrealistic to insist on NO internet connection for any process control system. There are too many necessary vendor support services connected via the Internet that are necessary to keep a modern process control system and the attached smart instrumentation in good repair and fully operational. As always, the Internet connection must be secure an allow only previously authorized connections. It's not impossible to achieve protected access, and all communications must be encrypted to prevent damage and covert data transmission. I didn't say it was easy, and it is usually not fast, but protected Internet connections must be allowed.</p>


RSS feed for comments on this page | RSS feed for all comments