Dear all,

I have a couple of questions I would like to ask the forum regarding PSV closed discharge systems.

First of all, is it sensible to consider that the reaction force from the PSV is acting just at the point of discharge or would it be more accurate to consider also forces acting on each change of direction such as bends or Tees up to the header where the discharge system ends? According to my understanding, based on the relevant appendix of B31.1, there's no need to do so, but if this is not the case, how can one define the values for such forces?

Additionaly, assuming that I wanted to simulate a closed discharge system using the response spectrum method, how should I proceed? In volume 26 of mechanical engineering news I found a quite detailed analysis on how to do so in case of an open discharge system. Would the approach be the same in case of closed systems as well, or not?

Any help would be greatly appreciated!

You may consider my opinion on subject.
It is in this post

best regards

There are several programs such as PIPENET-Transient and BOS-Fluids to calculate fluid dynamic forces. You can import force spectrums at any piping leg into CAESAR II dynamic input with the time lag for time history analysis. I personally used these programs with CAESAR II and got satisfactory results. In addition, you can import CAESAR II piping model directly into BOS-Fluids and perform the analisys.

Thank you very much for all the valuable information!

From my understanding the time history method would be the most sensible approach to the solution of such problems.

However, how should one proceed in case he/she wanted to analyse a closed discharge system, using the response spectrum method, assuming that a force with the magnitude of the PSV's reaction force acts on each piping leg?

Regards,

I'll quote from Coade #26/ Dec 98 you mentioned:

"You can also build your own response spectrum. CAESAR II makes this easy by providing a processor that converts an event time-history into a response spectrum. To complete a time-history of the event, we will have to know not only the opening time of the valve but also the overall duration of the event."

So you have to start defining the time-history and that means to define a wave traveling through your system, "leg by leg". You decided that the wave "amplitude" is the PSV reaction force, it remains to describe "how" the wave travels.
If you intend to do manually, you would study the "dynamics" examples/files Intergraph gave you with software as well Caesar's documents. Or you may follow Farhad's suggestions.

To "assembly" all the stuff in Caesar please study CAESAR II documentation. I think one clue taken from c2TR is "In the simplest form of force spectrum loading there is only a single shock component in the load case, i.e. there is only a single line of input on the load case screen. When there are multiple lines of input on the load case screen, as when the user is analyzing a traveling pressure wave that impacts different elbow-elbow pairs, there can be many components to the shock load case."

best regards

Dear Mr. Mariog ,

I have seen your responses regarding PSV analysis and they are all well explained and show your in depth knowledge on this complicated subject. I am confused in finalizing the time history definition for each leg of the PSV closed system.I am explaining my problem by following example.

One of the large leg(L=25 meter,Node 30-40) of PSV system which is oriented in Caesar X axis.sonic velocity & reaction force is C=250 m/s . F= 18 KN respectively.Typical PSV valve opening and closing time are respectively 80 ms and 250 ms. I have assumed overall duration of valve operation is 2000 ms. Then the time required for traveling the wave/flow from the node 30 to 40 will be 100 ms. The wave/flow reaches at node 30 at 150 ms (L/C) after PSV pop up. As per my understanding ,we can make time history profile of Impact load for this leg in two ways

(A) If we consider the water hammer type wave/flow then force (F=18KN) being applied in +X direction only at node 30 for elbow pair 30-40. The time history for this leg will be
Time (ms) Force Factor (N)
150 0
(150+80) 1
(150+100) 1
(150+100+250) 0

I have a doubt on this approach because i have not considered the Total overall duration (i.e. 2000 ms) for making the Time history definition(Force vs Time Profile)for leg.

(B) As a second option we can apply equal and opposite direction loads at node 30 & 40 of the elbow pair with time lag. This time leg equal the time used in traveling from point 30-40 i.e. 100 ms. The time history for this case is

a)Load (F=18 Kn)at node 30 in +X direction with time history as


Time (ms) Force Factor (N)
150 0
(150+80) 1
(150+2000) 1
(150+2000+250) 0

b)Load(F=18KN) at node 40 in –X direction with time history as

Time (ms) Force Factor (N)
(150+100) 0
(150+100+80) 1
(150+100+2000) 1
(150++100+2000+250) 0

Load time history a & b both are same except a time lag of 100 ms to account for unbalanced load.In this case the PSV overall duration profile is considered for analysis.

In my opinion method provided in (B) is more correct but on the same time, it takes more time especially when a lot of elbow pairs are present in the closed PSV discharge piping.Please correct me if i am wrong.Can i use method A for building the time history definition. If both methods are not correct then could you please tell me the right method to build the time history definition.

Please respond.

Thanks in advance

For me makes more sense a procedure similar with Goodling's paper on Steam Hammer that CraigB posted long time ago. Thanks Craig!
As I said in another post, Goodling’s paper is not referring to PSVs, but he presented his paper as valid for both traveling waves of increasing and decreasing pressure. In my opinion, it would be engineering satisfactorily to consider such approach for either expansion or compression waves.

So I suggest you to study the article and to post your conclusions/ opinion.

Thanks for your words, but I'm sure I don't have a "depth knowledge on this complicated subject", probably is fair to say I haven't a deep ignorance...
About the subject it's true I'm rather skeptical about the existence of real troubles in closed systems (which is also API skepticism); however I have no problem to follow any imposed Client specification or to consider any stupid Vendor value as Reaction force...

My best regards.

Dear Mr. Mariog ,

Thanks for your response and suggestion. I have read the goodling paper , in this paper he has considered a wave which is traveling with sonic velocity.The impulse function he has considered as either triangular or rectangular.I think in my example of previous post Impulse function A is based on the principle of wave propagation except in my example i have considered trapezoidal impulse function.Apart that Goodling has mentioned many important assumptions and suggestions in his paper.

However my Time history profile(B) was based on the assumption that same impulse function profile and magnitude force is applied at each elbow of a elbow pair in opposite direction . In this a time lag is incorporated to get the effect of unbalnced force on an elbow pair . APart that the PSV operational profile is used as a impulse function for this method.
But this method B involves a lot of man hour and time so i will go with method A for building the time history definition for PSV closed system.

Please correct me if i am wrong.

Thanks & Regards

I can give you a long answer but I would prefer to say my opinion is that.... you decided you know a lot about the wave propagation and just try to put it in math terms.

So it seems that you decided that there is an unbalanced force of 18 KN (probably based on a postulated equivalence of the "unbalanced force" with the free jet "reaction force") on each leg and the concern is how to define "more realistic" the propagation of this wave. If I understand well, in the A case you try to describe directly the "unbalanced" force, while in case B you tried to describe more elaborate the wave propagation and you expect the math will get enough information on how the "unbalance" appears.

The question would be focused on physics rather on math, so to be "sure" the math is describing the correct physics (something as in that T-shirts joke where the math of a famous formula is "E=ma2, E=mb2, E=mc2"...)

The math description should be just a written tentative to detail the physics of wave propagation, so practically if you can detail in common language the propagation of wave you can also put all the stuff in math terms. And the opposite is also true; having a math description- as is given clear in Goodling's paper, you can try to narrate in common language how Goodling considered the perturbation travels (which is not so clear in that paper).
You may do this and I'm sure this exercise will be not a fruitless effort.

My best regards.