When I analyze a piping system with a relief valve that discharges to the atmosphere it is pretty straight forward to calculate the thrust load and apply it to the analysis.

My question is: What to do when I have a relief valve that discharges into a closed system such as a flare header. Are there any loads that need to be added to the analysis?

I would appreciate any advice or insights on this topic.

Hello,

Regarding relief valve systems:

You will want to refer to ANSI/ASME B31.1, nonmandatory Appendix II (In B31.1, Roman numeral appendices are nonmandatory). B31.1 gives an approximate methodology for calculating blow-down forces in this appendix. You will also find help with these systems in API RP-520. API RP-520 addresses open to atmosphere systems and closed or "header" type systems. There are proprietary software systems which address "flare" type systems - it might be productive to search the web.

Of course in a closed system you would want to address the loading on the vent pipe (and resulting stresses) as a function of time(as the acoustic wave changes direction at elbows et. al.). Loadings of this type can be analyzed with Caesar II and perhaps one of our friends at COADE will give you some direction in that.

Ben Nottingham has taken the rules of B31.1, Appendix II (for relief valve blow-down forces) and put them on a spreadsheet template (ExCel) and this streamlines that whole procedure. Actually Ben's "PipingOffice" suite of piping design aids is MUCH more help than just this one application. He takes a lot of the simple but time consuming tasks and automates them. I think Ben's e-mail address is

bnottingham@chartertn.net

And his home is:

Ben A. Nottingham
509 Kilkenny Rd.
Kingsport, TN 37664

This is not intended as an endorsement, but you might want to look into it.

Best regards, John




[This message has been edited by John Breen (edited April 23, 2001).]

Mr. Stanky

For the closed discharge system in safety and relief valve, the forces acting on piping system during steady state flow may be considered as self-equilibrium. This means the forces do not create large moment or stresses on piping system. However, unbalanced transient force will act on piping system by traveling of pressure wave during transient flow. So, your analysis need to include these transient forces applied on each elbow points.
Generally the safety/relief valve opens very short time, within few millisecond and the fluid through the valve orifice will be chocked condition with acoustic velocity(Mach=1). And the flow of downstream will be supersonic or subsonic flow, and sometimes may generate shock wave. These flow conditions will depends on the condition of downstream such as back pressure, pipe size and length, frictional effects,

In my experience, this transient force is not significant for the flare systems in chemical and petrochemical plant. Therefore; I add just some guides and stoppers on the system from safety valve to flare header without detail calculation or analysis.
However, if you want to check it more detail, it will be very difficult and tedious job to get reasonable transient forces without the special software. I had used the SRVA which was developed by Sargent & Lundy to generate time history data for the main steam line safety valve system of nuclear power plant about 6 years ago. This analysis requires many physical properties and valve information. Also, Caesar II can generate time history data for safety and relief valve system and it would be helpful to you, but I have never used this module because I just check this force by adding supports without detail analysis.

The best way, I think, is to check it by conservative methods with simple hand calculation or the practices in Code and Standards as Mr. John Breen mentioned. Also, I recommend one article, Analysis of Power Plant safety and Relief Valve Vent Stacks, Transactions of the ASME, October 1975, G.S.Liao. I remember Caesar II also refer to this article in Safety/Relief Valve Analysis.

Joe; sometimes, when a valve is opened quickly in a closed psv system, you need to look at the P+deltaP x A loads on the elbows on either side of the valve.
There is a fluid hammer type of effect that can exist, where an unbalanced load occurs via a P+deltaP*A load applied at one point, while a downstream elbow can see P*A.
This may not be of concern for you if the load duration is very short, but it is worth looking at. With a gas, the pressure information should travel downstream at sonic velocity, which will assist you in calculating the duration of the load.

Hi Joe,

You have asked a question which is more difficult to answer than perhaps
you realize. First, before reading the rest of the post, the question(s)
I have for you are... "How important is the relief system? How large are
the forces involved? What flow regime is the flow in?"

Easy, least accurate/reliable (perhaps)approach...
1) Determine the discharge force for an open relief system. This can be
done with the CAESAR II relief synthesizer or some of the other
tools my colleague John Breen has mentioned.

2) Take this force, multiply by the dynamic load factor* of your choice
and apply it along elbow pairs in the correct vector one at a time
(you'll need multiple models).

*The use of a DLF in static structural analysis of a dynamically applied
impact load is widely published in various engineering texts. The value
commonly used is X 2. I have found that in some cases this is woefully
inadequate.

Harder, but a bit more
realistic...
Using the force number in 2 above use one of two methods to perform a
dynamic analysis. The two types of dynamic analysis possible were
described in an article written in COADE's newsletter. I believe that
this is available on this web site.

Hardest, most reliable answer
1) Get a copy of the DIERHS handbook from the AIChE organization.

2) Pour through the book and look over some very interesting math!

3) Realize that you may need some help go find a qualified, sympathetic,
Chem E

4) Chem E will then develop time-based data for momentum/ pressure loads
at elbows throughout the system.

5) Summate the data to form dynamic time/force data sets in Excel.

6) Use this data in a detailed system model of the system in CAESAR II. This model should incorporate large masses and structural K rates. Analyze and design using this data. **(See
Illustration)

7) Collect fee from owner at the end, and rest comfortably knowing that you did the best you could.

Some facts:
The unbalanced loads are applied on short lengths of pipe for a short time, conversely the long lengths have these unbalanced force applied longer. This means you need restraints on the longer runs.

Ironically everybody likes to clamp down the piping at the relief valve
and ignore the long runs!

Structural participation of even a few milliseconds load can be great or small and depends on the relative stiffnesses of the piping and the structure as well as large masses. An imposed load can be magnified much more the x2! I have had to resort to sping loaded sway struts to cut these enormous types of load down to size.

That's it, the questions you have to answer are the hardest. I have found on a critical relief system, when the owner has great concerns, that the best answers are readily paid for.

------------------
Best Regards,

John C. Luf

Do we need to multiply by DLF (dynamic load factor)to force if calculated as mentioned in API 520 ? I had a feeling that only if force was calculated by B31.1 then this multiply by 2 is necessary.

If you are going to run the analysis "statically", then yes you should multiply the load by the DLF value.

If you are going to run the analysis "dynamically", then the software will automatically handle the DLF.

I agree to all of Mr. Luf's suggestion.I have a small question for the stress community.The general tendency is not to consider relief loads for closed systems because of the "static balance " concept which is very inappropriate for long runs as pointed out by Mr. Luf.At the same time there is a general tendency to use the relief loads ( computed by any means) at points of sudden enlargement like entry to a header for such closed systems.

I personally feel it is due to the backflow as a result of adverse pressure gradient (dp/dx>0) at the opening.

Is that the reason or else what are the reasons for this consideration?

A.Bhattacharya

Stress Analyst

Bechtel Corporation

For Calculation of Reaction Force of Relief valve as per API 520,

F=(W sqrt((kT)/(k+1)M))366 + A *P

and P is defined as static pressure within the outlet at the point of discharge, psig.

so, if
Ps = Set Pressure = 165 psig,
Pi = Inlet Pressure (Relief valve inlet) = 150 psig
Po = Outlet Pressure (Relief Valve discharge) = 15 psig
Pa = Atmospheric Pressure.

Then my question is what is this 'P' in the calculation.
I practice of taking this as Po-Pa.
But I have found some engineer's considering as Ps-Pa.
Which of the above is correct?

John, can you kindly clarify what is it "DIERHS handbook from the AIChE organization." I can't find anywhere on a web and I asked chemical engineer, she doesn't know it either.
Thank you

[QUOTE]Originally posted by John C. Luf:
[QB] Hi Joe,

From the AICHE web site.....


Emergency Relief System Design Using DIERS Technology


OSHA (29 CFR 1910.119) has recognized AIChE/DIERS two-phase flow publications as examples of "good engineering practice" for process safety management of highly hazardous materials. The prediction of when two-phase flow venting will occur, and the applicability of various sizing methods for two-phase vapor-liquid flashing flow, is of particular interest when designing emergency relief systems to handle runaway reactions. This comprehensive sourcebook brings together a wealth of information on methods that can be used to safely size emergency relief systems for two-phase vapor-liquid flow for flashing or frozen, viscous or nonviscous fluids. Design methodologies are illustrated by selected sample problems. Written by industrial experts in the safety field, this book will be invaluable to those charged with operating, designing, or managing today's and tomorrow's chemical process industry facilities.

Special Details: Hardcover


Publish Date: 1992
Pages: 538pp
Price: $175.00
ISBN No: 0-8169-0568-1
Pub No: X-123

Thank John

Additional information:

http://www.iomosaic.com/diersweb/htdocs/publications/publications.html

Leonard@thill.biz
www.thill.biz

Formula mentioned by Ladi (member 907)is only for compressible fluids i.e. gases and vapours as per API 520. Whenever PSV is located on liquid lines say for example in discharge of positive displacement pump and outlet of PSV connects in pump suction line i.e. it is closed loop, then
which is the approch to find the reaction force.

As per discussed,
at which point does the time imbalance in these forces become important?

ie for V=300m/s through pipe 10m long,
then duration of imbalanced force on elbow = 0.03s which is negligable,
however if the said pipe = 300m long then the duration of imbalance = 1s
which i would be worried about.

what are the norms used for when these time imbalances becomes important.?

or, at what point using simple static techniques
does the system fall out of equilibrium.?

I also assume that if t=0 then energy =0?

Best regards