Comments on Pressure Relief System Design

I would like to draw your attention to a few points on the November 2008 article, Pressure Relief System Design (pp. 40–45):

1. Pressure drop and pressure loss: The inlet pressure “drop” that the author refers to under Relief system piping is non-recoverable pressure loss, and does not include a “drop” in pressure due to elevation increase. It should be labeled as a pressure loss rather than a pressure drop as the author has specified elsewhere. It may sound trivial, but such errors are not uncommon. The 3% rule is code-mandated and should be followed. However, any excess above this can still be defended in some circumstances in the court of law, but only by an expert analysis.

2a. Bursting of rupture disk: In the penultimate paragraph, when a rupture disk is used in series with a relief valve, the author writes: “… the disk would never burst”. The rupture disk (as well as a conventional relief valve) is a differential pressure device. So the disk will burst when the difference between the upstream pressure and the downstream pressure of the disk equals the burst pressure of the disk (or the set pressure of a conventional relief valve).
The other point to remember is that the pressure developed in the space between the disk and a relief valve in a combination system caused by a pinhole in the disk is equilibrated with the pressure in the protected equipment under normal range of pressure variation. But under emergency condition, the rate of pressure rise is too high to reach an equilibrium between the pressure in the equipment and that in the space between the disk and the relief valve. As a result, a differential pressure is developed until either the protected equipment fails or the differential pressure reaches the burst pressure resulting in the breakage of the disk.

Let me illustrate this with an example. Suppose the protected equipment has design pressure of 100 psig. The disk burst pressure is 90 psig and the relief valve set pressure is 100 psig. The normal operating pressure is 60 psig. A pinhole develops and the space between the rupture disk and the relief valve reaches a pressure of 60 psig. Due to some contingency, the pressure rises rapidly in the protected equipment and by the time the pressure in the space between the disk and the relief valve reaches 70 psig, the pressure in the equipment reaches 160 psig. Since the differential pressure is 90 psig (i.e. 160 – 70), the disk will burst. However, depending on the temperature developed, factor of safety involved in the design, material of construction, available thickness of material, the equipment may retain its shape, undergo plastic deformation, or fail catastrophically under this contingency of scenario. Please remember that there is a margin, which varies with the governing code, between the allowable stress and the yield point of the material of construction. This margin allows an equipment to be stressed up to the proportional limit without a permanent deformation.

2b. Pressure indicator between a rupture disk and relief valve: The second point in this subject is the illustration of Figure 5 which is backed by the author’s recommendation: “whenever a rupture disk is installed upstream of a relief valve, it is important to have a pressure indicator in the section between the two.” If the author is following the ASME VIII, Div. 1, the paragraph UG-127 stipulates the following: “the space between a rupture disk device and a pressure relief valve shall be provided with a pressure gage, a try cock, free vent, or suitable telltale indicator. This arrangement permits detection of disk rupture or leakage.” Because of the presence of a serial comma in the phrase in italic letters, a grammatical parsing of the phrase accepts the author’s recommendation to include just a pressure gage to be in technical compliance with the code. However, the spirit of intent of the code will not be followed by such design. In fact just the installation of a pressure gage is not enough and it is unsafe to do so.

The reason behind this is that it requires monitoring of the pressure gage, and the proposed design does not have a vent valve to bleed the pressure between the disk and relief valve. This is typically done by installing an excess flow valve at the downstream of the disk. The excess flow valve in essence is a ball check valve, which allows the leaked fluid go past the ball into atmosphere in the case of a pin hole, but creates enough resistance to show up pressure in the telltale indicator. In the event of a sudden breakage, the ball seals the flow to outside by closing the port. The gage must be of telltale type. When the space cannot be vented due to toxicity of the fluid or other reason, a bleed valve is still needed, but the bleed valve is kept car-sealed closed, and a control-room connected pressure switch with a local indicator and  high pressure alarm with appropriate safety integrity level should be installed in the space between the disk and the valve. Finally, the rupture disk can be substituted by a breaking pin device, which facilitates in-place replacement of the pin without shutting down the system.

3. K.O. drum: By showing the K.O. drum in strategic position, the author implies two-phase flow possibilities. The DIERS (Design Institute for Emergency Relief Systems) does not recommend the use of conventional relief valves for applications with potential two-phase flows. The pressure drop calculations for two-phase flows are more complex than the methods used by the author.

The API standards cited by the author are outdated and the recent ones address the two-phase issues. The author is also advised to refer to the following: 1) Emergency Relief System Design Using DIERS Technology, 2) Guidelines for Pressure Relief and Effluent Handling Systems, 2nd edition, to be published in 2010 (first edition available now), both published by CCPS of AIChE, 3) A dynamic simulation software, SuperChemsTM (ioMosaic Corp., Salem, NH,USA) for dynamic simulation of emergency relief systems and 4) ChemCAD  software (Chemstations, Inc.; Houston, Tex.)

Author replies

1. Pressure drop and pressure loss: Having worked in various projects worldwide, I had a look at process data sheets for the terminology used for permanent pressure loss. In many cases, I found that the term “pressure drop” has been used to denote permanent pressure loss, such as in flow-orifice data sheets. The following books also use the term “pressure drop” instead of “pressure loss”:
• “Control Valve Handbook”, 4th edition, by Emerson
• “Applied Instrumentation in the Process Industries” Vol III (Andrew WG/Williams HB) and Vol IV, (Leslie M Zoss)
• “Perry’s Chemical Engineers’ Handbook” uses the term frictional losses in some cases. However, at many places, it also uses the term “pressure drop” for two-phase flow, single-phase flow, flow in spirals, annular flow, flow through beds of solids and so on
• Even the API 521 (5th Ed, January 2007) uses the term “pressure drop” at many places
The term “pressure drop” may be interpreted by some as a drop in pressure due to elevation increase. However, references show that in several cases, the term “pressure drop” has been used to denote frictional losses.

2a. Bursting of rupture disk: When I mentioned “rupture disk would never burst”, what I meant was “rupture disk would never burst at its burst pressure”. An example is illustrated in a rupture disk with a burst pressure of 90 psig. The protected equipment design pressure is 100 psig. A pinhole develops and the space between the rupture disk and the relief valve reaches a pressure of 60 psig. Now the contingency develops. The protected equipment reaches 90 psig (the burst pressure of the disk). Downstream of the rupture disk, the pressure is 60 psig. At this stage, we would expect the rupture disk to burst, but it will not, because the differential pressure is only 30 psig (90 – 60). Now the protected equipment reaches 100 psig (its design pressure). Even now the disk will not burst. This is what I meant by my statement in the article. The pressure in the protected equipment now further develops until it reaches 160 psig. Downstream of the rupture disk, the pressure has reached 70 psig. Only at this stage will the rupture disk burst. But the damage to equipment may have been done. It may be argued that there is a margin of safety in the vessel. However, this is not what the design was intended for. The disk should burst at its burst pressure.

2b. Pressure indicator between a rupture disk and relief valve: The pressure indicator between the rupture disk and the relief valve, of course, needs monitoring. We have detailed designs of combinations of rupture disks and relief valves with a pressure switch on the line and an on/off valve, which depressurizes the system once the pressure reaches a certain set value. Detailed schemes such as these could be the subject of another article.

3. K.O. drum: Yes there are two-phase flow possibilities. But two-phase flow will not necessarily occur all the time. Single phase flows are also possible. Please note that Figure 1, where the K.O. Drum is shown has no direct relation with the illustrated example. The example illustrates a single-phase flow calculation to show how discharge-side piping headers and branch lines are sized. The basic theory and equations for single phase flow will not change, whether it is the more recent API 521 (January 2007) or the relatively older API standard cited in the article. For two-phase flow, there are several references in the literature. But it was not the intent here to illustrate a two-phase flow calculation.