15 replies to this topic

[jprocess]

Dear All,
I have some questions about this important scenario and I should be grateful if you would answer to these questions:
1.API 521-1997 EDITION in page 23,speak about 2/3 rule.But in 2007 EDITION we can not find anything about "2/3" value.Only page 11, part 4.3.2 contains an example that implicitly speak about a value of 130% for low pressure side test pressure.So they omitted the 2/3 rule and also did not speak explicitly about 10/13 value.So what approach the designer should follow?
2.Which value is more reliable?2/3 or 10/13?
3.API 521-2007 EDITION in page 53,part 5.19.2 state:
"The use of maximum possible system pressure instead of design pressure may be considered as the pressure of the high-pressure side on a case-by-case basis where there is a substantial difference in the design and operating pressures for the high pressure side of the exchanger."
An example for these situations where the difference between the design and operating pressure is large can be the settling pressure for which the interstage coolers in multistage compressors are designed. It might happen that, for some of the coolers the operating pressure of the process side is less than the cooling medium pressure itself and yet the design pressure is higher.For this case what approach should be followed?
4.Is there any similar scenario for plate and frame heat exchangers?
5.Finally API 521-2007 EDITION in page 54,part 5.19.4 speak about "shock wave":
"It may be necessary to locate the relieving device to be located either directly on the exchanger or
immediately adjacent on the connected piping. This is especially important if the low-pressure side of the exchanger is liquid-full. In this case, the time interval in which the shock wave is transmitted to the relieving device from the point of the tube failure increases if the device is located remotely."
Does this mean that the PRV for the other failure cases open under static pressure, whereas in tube rupture, it is by the pressure wave?! If so,have spring loaded safety valves the ability of fast response against this shock wave?
Thanks in advance.
Cheers.

[pleckner]

1. First, let's understand where these values, 2/3, 10/13 come from.

In the beginning, ASME required that equipment and piping be tested at 150% of stated design pressure. Saying this another way, if my equipment design pressure is 100 psig, then my test pressure must be 150 psig; so 100/150 = 2/3. API said that if the design pressure of the low-pressure side was within 2/3 of the design pressure of the high pressure side, then you would not have to include tube rupture as a credible relieving scenario. They figure that in all practicality you would not have a catastrophic failure of the tube if their design pressures were so close. By the way, API extended this "rule" beyond just the heat exchanger in determining if you have a credible scenario or not. They included the entire system as needing to be within the 2/3 rule, not just the shell and tube of the exchanger. That is, if there is a vessel after the exchanger tied directly into the low-pressure side of the exchanger, then its design pressure must also be within 2/3 of the high-pressure side of the exchanger.

Today ASME only requires the test pressure of equipment to be 130% of the stated design pressure. Saying this another way, if my equipment design pressure is 100 psig, then my test pressure must be 130 psig; so 100/130 = 10/13. Now to avoid a tube rupture as being a credible scenario, the design pressure of the low-pressure side has to be within 10/13 of the design pressure of the high-pressure side, including the system. Again, the test pressure of equipment is allowed to be tested at 130% of the stated design pressure but piping must still be tested at 150%. Also, nothing says you can't still test even equipment at 150% of the stated design pressure.

So the bottom line, if you test at 150%, then you can still use the 2/3 rule to determine if tube rupture is a credible scenario and if you test at 130%, then you need to be using 10/13.

But saying all this, the latest version of API Standard 521 states the issue such that there seems to require a lot more thinking. They give you many more "ways out" if you want to spend the effort to go through a lot more analysis. I would just go with the 2/3 or 10/13 ratios and be done with it; it just isn't worth the engineering effort to try to avoid the issue any other way.

2. I believe I addressed your question #2 with my discussion above.

3. I would be very careful how you try to interpret API's statement in this case. They don't really tell you how much "substantial" is now do they? You must use the maximum operating pressure that you could ever expect to see when comparing to the design pressure in these cases.

4. No.

5. No, a relief device opens on the static pressure. The concept of the pressure wave is that it is so fast that a PSV will probably not be able to respond fast enough to actually protect the equipment. We should consider rupture disks or buckling pin devices instead.

[fallah]

Dear Pleckner
What is the mean of "the design pressure of the low-pressure side was within 2/3 of the design pressure of the high pressure side"

For example: If the design pressure of high pressure side is 100 psig,then what is the range of design pressure of low pressure side,based on the above statement?

Thanks in advance
Fallah

[gvdlans]

Thank you Phil, for your excellent reply in this interesting topic!

With respect to the transient effects of tube rupture I still feel that the picture is not very clear, although there has been some improvement since the previous edition of API 521. I have the impression that the high peak pressures have thus far only been shown by dynamic modelling, and that these peak pressures could be the result of the assumption used that the rupture is instanteneous (meaning full rupture in 0 milliseconds). I have never heard of any accidents/incidents or experiments where these effects have occurred.

Furthermore, the guidelines on how to do these dynamic simulations are quite obscure (references [64] and [65] in API STD 521 (2007)).

Also, the assumption that rupture disks open very rapidly has not been proven, as indicated in the report on http://www.hse.gov.u...02/oto02023.pdf (this is reference [119] in API STD 521 (2007)) and http://www.hse.gov.u...00/oto00130.pdf.

Phil, what is your opinion on this? Do you actually carry out dynamic simulations for this scenario?

[pleckner]

@gvdlans,

Yes, I've read that study and it has created doubt in my own thinking. I think this is also why API is pretty vague on the issue. I still think a rupture disk is better than a PSV if you expect such transients but I guess I too should re-think this. I also agree that experience shows that tube rupture is rarely if ever an abrupt break. And no, I won't spend the engineering manhours doing dynamic simulations. I have a hard enough time trying to convince clients when they have 2-phase relief that I require more sophisticated calcualtions (read more time and money needed)!

After saying all this, let me add that this is NOT how I would approach the design in the first place. I will do everything I can to eliminate the scenaro entirely. Since I am the Process Design Engineer, the heat exchangers are always designed so that the tube rupture scenario never becomes credible. I think this is easy to do and is easy to convince my client that this is the only safe way to proceed.

Bottom line, Process Engineers should not be worrying about protecting against the tube rupture scenario, they should be making it go-away! There is too many unknowns to make this a really safe system.

And this leads me to answering @fallah:

If the high pressure side design pressure is 100 psig, then based on the 2/3 "Rule" the design pressure of the low pressure side should be 66.67 psig or greater. I would make it a minimum of 70 psig. Nothing says you can't make the two design pressures the same either. It is a matter of economics and in all practicality, I don't think you'll see much of an economic impact on making the low pressure side design pressure 100 psig in this case.

[gvdlans]

Phil,

Thank you for the interesting points made!

You wrote "I will do everything I can to eliminate the scenaro entirely. Since I am the Process Design Engineer, the heat exchangers are always designed so that the tube rupture scenario never becomes credible." I assume you mean here that you will always design the system so that it meets the "design pressure/test pressure rule" and/or by assuring that an open flow path can pass the tube rupture flow.

As per the first paragraph of section 5.19.2 (in API Std 521 of course), designing according to the design pressure/test pressure rule does not eliminate the possibility of a tube rupture but makes it highly unlikely that there will be loss of containment to atmosphere in case of a tube rupture.

In the last sentences of section 5.19.2 it is stated that "Modelling has shown that, under these circumstances, transient conditions can produce overpressure above the test pressure, even when protected by a pressure-relief device. In these cases, additional protection measures should be considered." With "these circumstances" they mean a wide difference in design pressure between the 2 exchanger sides (e.g. 70 bar or more), especially where the low-pressure side is liquid-full and the high pressure side contains a gas or a fluid that flashes across the rupture.

It is my interpretation that in case there is a wide difference in design pressure between the two-exchanger sides (e.g. 70 bar/1000 psi or more), additional measures should be considered to protect against transient conditions, even when the system has been designed according to the design pressure/test pressure rule.

So in case the design pressure of the tube side is 300 barg, the design pressure of the shell side should be at least 230 barg. These pressures may seem remote, but I can imagine that such pressures occur in natural gas handling installations (e.g. gas compression for gas injection, where pressures can be several hundreds of bars. Compressor interstage cooling with cooling water would be an example). Here the costs of increasing the design pressure of the shell side (system) could become an issue...

The reports that I linked to in my previous post indicate that the UK HSE Executive has commissioned several studies into this subject, so hopefully in a few years from now we will know more...

[pleckner]

@gvdlans

Yes, I will always design the system so that it, as you say, "meets the "design pressure/test pressure rule" and/or by assuring that an open flow path can pass the tube rupture flow" and therefore I don't have to deal with the tube rupture scenario; as given by API.

Sorry but I need to correct some inaccuracies in your post. The paragraph you quote on "modeling" and "transient pressures" is at the bottom of paragraph 5.19.3, not 5.19.2. I don't interpret it the same way you do. The discussion on "modeling" and "transients" and wide pressure differeneces as written in paragraph 5.19.3, discusses "determining the required relief rate", not in determining if you even have a relieving scenario. The correct wording is:

"Modelling has shown that, under these circumstances, transient conditions can produce overpressure above the test pressure, even when protected by a pressure-relief device [64], [65], [66]. In these cases, additional protection measures should be considered."

You interpret this as saying, "... even when the system has been designed according to the design pressure/test pressure rule." I don't see this.

Once you follow the criteria we've been discussing, the chances of a catastrophic break that would even produce a transient problem becomes remote. This is why API gives us the "way out". It's nice to see they get practical at times.

Now, I will agree with this statement by API in the next paragraph (5.19.4):

"It can be impractical to protect some heat exchangers (and associated piping) by relief devices alone e.g. if there is a high pressure difference between the shell and tube sides. In these cases, different layers of protection, such as improved metallurgy, more frequent inspection and increasing the design pressure of thelow pressure side (including upstream and downstream piping until the pressure is dissipated), can be necessary."

I agree with you in that this area needs much more study and the good news is that it is a recongnized problem in the pertinent industries.

[gvdlans]

Phil,

You were right about the paragraph numbering, I missed one paragraph header. I also think that the text in 5.19.4 is important and an improvement over the previous edition.

My interpretation was mainly based on an official Technical Interpretation from the API committee (see attached file). There was a question asked about the 3rd edition of API 521. The question asked was what the basis was for the 2/3 rule as found in this 3rd edition. The official reply was as follows:

"Increasing the low pressure side design pressure will have little impact on the likelihood of a tube rupture. However, by raising the low pressure side design pressure to a point where the low pressure side test pressure is equivalent to the high pressure side design pressure will ensure that damage to the exchanger due to a tube rupture is extremely unlikely. [...]"

So even when the design meets the design pressure/test pressure rule the possibility of a tube rupture remains, and this may result in transient effects (if there is a wide difference in design pressure etc.) that can produce overpressure above the test pressure (as written in paragraph 5.19.3).

Attached Files

•  API_521_technical_interpretation.pdf   54.75KB   669 downloads

[pleckner]

It is important to realize from whence (I always wanted to find a way to use that word) and where things come from. The interpretation you refer to is an interpretation from the 1990 edition of API RP521. As we all know, a new edition of API 521 has finally been released and written by a more representative group, the international community.

Again, as written in Paragraph 5.19.2 Pressure Considerations:

"Pressure relief for tube rupture is not required where the low-pressure exchanger side (including upstream and downstream systems) does not exceed the criteria noted above. The tube rupture senario can be mitigated by increasing the design pressure of the low-pressure exchanger side (including upstream and downstream systems), and/or assuring that an open flow path can pass the tube rupture flow without exceeding the stipulated pressure, and/or providing pressure relief."

The way I read it, the portion of this paragraph I underlined completely contradicts that which was published as an Interpretation, "Increasing the low pressure side design pressure will have little impact on the likelihood of a tube rupture."

What is still valid is that there is still a possibility of a tube rupture but one more related to small leaks, not to catastrophic failure that could induce severe pressure spikes and the like.

I go back to what you and I completely agree upon, further study of this is badly needed.

[gvdlans]

Phil,

We are getting to the point where we may be reading more in the wordings than what was actually meant... I am not a native English speaker, but according to my American Heritage Dictionary, "to mitigate" means "to moderate (a quality or condition) in force or intensity; alleviate; to become milder".

As a safety/risk consultant I always use "to mitigate" to mean "reduce severity of consequence". As you know risk is a combination of likelihood of an outcome in combination with the consequence of that outcome. When you take mitigating measures to reduce the risk this means that you reduce the effects (e.g. by providing a fire extinguishing system), but you do not reduce the likelihood of the event (e.g. the fire).

So when I read "the tube rupture scenario can be mitigated ..." I understand this as "the consequences of the tube rupture scenario can be made less severe" (since there will not be a loss of containment/release to atmosphere). So I don't see a contradiction between the Technical Interpretation and the latest revision of API 521.

I just sent a Technical Inquiry to the API about this. If I get a reply I will post it in this forum. According to the API.org website this may take up to 12 months...

[pleckner]

I think we are in basic agreement.

Yes, you make the tube rupture scenario less severe and to me that means what I last posted in my second to last paragraph, "What is still valid is that there is still a possibility of a tube rupture but one more related to small leaks, not to catastrophic failure that could induce severe pressure spikes and the like."

But I can't see where the API interpretation saying "Increasing the low pressure side design pressure will have little impact on the likelihood of a tube rupture." is the same as saying "The tube rupture senario can be mitigated by increasing the design pressure of the low-pressure exchanger side..." And I again point out the statement that opens the paragraph, "Pressure relief for tube rupture is not required where the low-pressure exchanger side (including upstream and downstream systems) does not exceed the criteria noted above."

I agree it may be just a matter of wording and how one interprets that wording but then isn't that always the problem with Codes and Standards?

So what exactly did you ask API?

[gvdlans]

Phil,

I also get the feeling that I am starting to repeat myself...

Basically I see the case where the pressure difference is more than 70 bar as an exception to the general statement in 5.19.2. In general you will prevent release to atmosphere in case of a tube rupture by increasing the design (and test) pressure of the low pressure side to the design pressure of the high pressure side, but the exception is the case where the transient effects are such that the test pressure may be exceeded.

Here comes my e-mail to API:

"API Standard 521 Pressure-relieving and Depressuring Systems (5th edition (2007)), paragraphs 5.19.2 and 5.19.3

As written in Paragraph 5.19.2:

"Pressure relief for tube rupture is not required where the low-pressure exchanger side (including upstream and downstream systems) does not exceed the criteria noted above. The tube rupture senario can be mitigated by increasing the design pressure of the low-pressure exchanger side (including upstream and downstream systems), and/or assuring that an open flow path can pass the tube rupture flow without exceeding the stipulated pressure, and/or providing pressure relief."

In Paragraph 5.19.3:

"Modelling has shown that, under these circumstances, transient conditions can produce overpressure above the test pressure, even when protected by a pressure-relief device. In these cases, additional protection measures should be considered."
"These circumstances" refers to a wide difference in design pressure between the two exchanger sides (e.g. 70 bar or more), especially where the low-pressure side is liquid-full and the high pressure side contains a gas or a fluid that flashes across the rupture.

My question is: if the design of my system meets the criteria of 5.19.2 (low pressure side test pressure at least as high as high pressure side design pressure), do I still have to protect against possible transient effects in case the design pressure difference between the two sides is more than 70 bar?"

[pleckner]

I am most curious to see the API answer.

Interesting discussion.

[MarcoProcess]

Hi. First post here.

I have a question.

When we're talking about heat exchanger with cooling water on the low pressure side it is not economically possible to increase all the cooling water system design pressure up to 10/13 of the highest "other side" design pressure.

I think that one could design the cooling water side of the heat exchanger and associated piping up to and including the block valves at 10/13 of the "other side" design pressure.

I'm thinking about an open cooling water system - cooled by cooling towers.

This way only a TSV will be required.

Do you agree with this design?

[Nirav]

Overall, good discussions in this topic....

apparently, we did some dynamic simulation for the 'tube rupture' scenario. It was found during very later stage of project that we need to do such study. Otherwise, what Phil told is correct that we don't waste time during early or grass root design for dynamic simulation in such scenario. We try to design system in such a way that tube ruputre can be avoided as possible scenario. The study we carried out concluded that due to large pressure differences between high pressure & low pressure side of heat exchanger, it happens to be a "choke flow" condition at a specific pressure level. Therefore, pressure won't go above that level on low pressure side.

I think that one could design the cooling water side of the heat exchanger and associated piping up to and including the block valves at 10/13 of the "other side" design pressure.

I'm thinking about an open cooling water system - cooled by cooling towers.

This way only a TSV will be required.

Do you agree with this design?

Macro, I do agree with your approach. It is the case for most of designs we performed.

[fallah]

Dear Nirav

If the lower pressure side be the shell, choked flow occurs in tube rupture case (constant mass flow) toward the shell as long as the shell pressure be below the critical pressure and seems it can increase the shell pressure. Would you please explain why" pressure won't go above that level(spesific pressure level) on low pressure side."?

Regards