Dear all! I am desiging the steam pipe to a turbine. As you know the inlet valve closes in a very short time, so steam hammer analysis is recommended. I know that the best way to perform it is using a fluid transient analysis and getting the force vs. time function in each node and use it in CAESAR dynamic analysis. As far as I know to get those functions a fluid transient software is necessary and as I don't have any so I am looking for any simplified analysis. Do you know of any document, paper, book, etc where I can find a simplified calculation method? (perhaps static instead of dynamic).

Thanks for any help. best regards.

Hi Carletes,
You can carryout static analysis. I shall send you a MS Excel file for calculating steam hammer load. Please let me know your email.
Regards,
Manoj Sarkar
Email: manoj_sarkar@rediffmail.com

Dear Carletes,

Be careful with a static approach in this case. Especially steam lines are very flexible due to
the support concept based on the required thermal expansion. Keeping this in mind, these lines can be regarded as more vulnerable to dynamic loads than piping located on pipe racks or in process plants.

I don’t think an excel spreadsheet will generate correct loads for you. Using static equivalent forces, you might have to think about an additional dynamic load factor for each pipe segment.

In my opinion, the dynamic piping loads caused by a steam hammer should be computed using an appropriate fluid dynamic software AND by experienced people.

Nevertheless, I am of course interested in the Excel files of Manoj Sarkar.

All the best,
Veit Bockemühl
esn - engineering services network - gmbh
Fischmarkt 16
D-22767 Hamburg / Germany
vb@esn-gmbh.com

Dear Mr Veit Bockemühl,
You are correct, calculated steam hammer load is always multiplied with a suitable DLF to take care of dynamic nature of the load. Anyway, I am forwarding the said excel file to you.
Regards,

The problem with doing this analysis as a static equivalent is that you are being ultra-conservative because you will be considering the steam hammer forces to act throughout your piping system at all bend nodes simultaneously and persistently. This of course does not reflect the actual situation. The comments regarding the DLF are certainly prudent. Since the maximum DLF is 2.0 I would double all the calculated steam hammer forces and then input those at each bend node in the system. Be sure you use the proper load cases for the occasional load evaluation, namely those for nonlinear restraints. Using F1 to represent steam hammer loads at each bend or tee use load cases similar to these:

1) W+T1+P1 (OPE)
2) W+T1+P1+F1 (OPE)
3) W+P1 (SUS)
4) L1-L3 (EXP)
5) L2-L1 (OCC) segregation of steam hammer loads
6) L3+L5 (OCC) use ABS or Scalar combination method under the load case options tab.

By the time you have completed all this work it seems you could have already performed the dynamic analysis.

To do this dynamically you can assume a linear ramp-up to maximum force equal to the closing time of the valve and the same for the ramp-down profile. The only thing missing then is the duration of the unbalanced force that acts on the bend. This is equal to the length of the pipe leg that follows the bend divided by the velocity of the pressure wave, which is equal to the speed of sound in the steam. If you have the temperature and pressure of the steam it should be straight-forward to calculate the speed of sound from a = (Gamma*R*T)^.5 where Gamma is the ratio of specific heats, R is the Gas Constant, and T is the temperature (in R or K, not F or C).

Of course there is no perfect substitute for a good fluid-hammer computer program and there are several on the market. But failing that the above approach will give you a fair representation of the piping response in most cases.

Neither the static equalvent loads nor the force response spectrum method considers the timing interaction between elbows - where two individual events at different locations and times combine to push one or more modes to an even greater response than a simple summation (where the resulting DLF is greater than 2.0). Maybe it doesn't happen in your system, but only time history analysis can account for it.

I have been following all the discussion about water-hammer and unfortunately instead of helping Carletes with his problem I want to make a question. The answer anyhow could help him in understanding better the problem.
Now I’m dealing with this 26” transfer line and the closure of a shut-down valve. The formulas for determining if there is a slow T>2*L/c or a rapid T<2*L/c closure of the valve. Where T = valve time closure, L = distance from the valve to a big big change in diameter, for example from the valve to the tank, and c = speed of sound in the fluid. Both formulas have their respective formulas to calculate the delta H [meters] and consequent hammer loads at the valve.

Slow closure : delta H = 2*L*v / g* T , Michaud

v = fluid velocity
g = 9.81 m/s^2

Rapid closure : delta H = c * v / g , Allievi

There in the CAESAR manual page 7-29 says that if the situation is that of a slow closure the tendency of a water hammer will be probably abated. Here the words “tendency” and “probably” mean that the problem doesn’t finish there. I would like to know if there is another formula or analysis to demonstrate that there in the system will not be a water hammer. I mean, I went further and did the dynamic analysis using the DLF. From this analysis the dynamic loads in the some restraints are much much higher that the ones we were expecting, and there the delta in the pressure will be “probably” dispersed in the system before any of this loads is generated.

best regards.

Santiago Naranjo

I have very basic knowledge of fluid mechanics and following is my own understanding of the problem

Here there are two types of DLF are involved. One is hydraulic and the second is mechanical.

When pressure is generated in the system it travels untill it reaches a point of reflection where the pressure wave is reflected back. This reflected wave may superpose the original wave to produce pressures at certain points, which are higher than the original value. A proper fluid dynamic analysis will give the pressure history in the system, which depend upon the hydraulic properties of the system. A hydraulic DLF in this sense relates the peak system pressure to the pressure that was originally generated.
Now unbalanced pressure in the piping will generate forces on the system. Since this force is applied dynamically, the system response will be higher that the response observed when the same force is applied statically. A mechanical DLF therefore relates the peak system response to the response observed by same amount of static force.

If the valve closure time is higher than the critical time then it means that there will not be hydraulic amplification and a hydraulic DLF of 1 can be taken, without doing detailed fluid dynamic calculation, to find the peak pressures. If valve closure time is lower than the critical value then there will be dynamic amplification and a proper fluid dynamic analysis is required to find the peak pressures.
Since this peak pressure, in both cases, is not static but dynamic in nature, we can't ignore the mechanical impacts of this unbalanced pressure. We will have to do piping dynamic analysis applying the impact force calculated on the basis of above peak pressures.

Whoa,

Lets settle down a tad. First unsteady fluid flow is a science unto itself. We are engaged in the field of structural analysis not unsteady fluid flow.

As such let me adress the simple formula T>2*L/c this is the classic formula for critical closure time. However similiar to the B31 codes simple Dy formula for flexibilty analysis it is extremely simple and prone towards misapplication. Non-linear valves, such as butterfly valves etc. for instance are not adressed by this. Nor are vapor pockets etc.

So rather than extending our small bit of knowledge too thin I advise seeking the help of an expert in unsteady fluids.

Dynamic or Static.... well to me static has always been fraught with hazrds, probably because when I was a youngster the Tacoma narrows bridge disaster was pounded into me. http://www.nwrain.net/~newtsuit/recoveries/narrows/gg.htm , http://www.ketchum.org/tacomacollapse.html Dave is correct 2.0 may not be the maximum and if the system is critical dynamic is the preferred discrete method.

Dear all,

Thank you all for your help. I see that some of you consider that a dynamic analysis is completely necessary, for which I think a specialized software is necessary.
Which one is in your opinion a good (easy to use , reliable and economical if possible) software? Is there any which can use the CAESAR files directly? I would use it just for fluid hammer calculations in power plants (cooling water and main steam systems, mainly).

Thanks again and best regards,