The book by ARD Thorley "Fluid Transients in Pipeline Systems" published by D & L George contains charts of vasrious types of check valves and their operating times. The book may be out of print so check your local university or water authority library.

manufacturers are alittle useless on such matters.

Sharing knowledge is a way to immortality

check out http://www.adamsvalves-usa.com/images/adams/filemanager/Marketing/Selection%20Guide.pdf

Sharing knowledge is a way to immortality

MortenA

Speed of responce is never a real issue. The real issue is the associated pressure surge on liquid applications. For gas I don't believe that speed of responce is ever a real issue! If anybody has other thoughts about it ... let me know. I like to learn

If it is a liquid application drop me the process data and I can calculate (and graph) the closing speed using the formulae of the dimensionless dynamic characteristics.
(robert_verwey@hotmail.com)

I disagree with RobV on the question of speed of response of check valves not being an issue.

The relation ship of dv/dt is paramount in the system response to a closing check valve.

If speed of response is not an issue why do Mokveld, Noreva and Mannesmann all make valves that close in 0.3 secs to reduce surge? Why do other companies have fast to close and dashpot to finish closing valves? Why are counterweights added to valves?

Suggested reading is Thorley Fluid Transients in Pipeline Systems. Streeter & Wylie Fluid Transients, Boldy et al, Zhou, Hicks and Steffler.

Paper from the 5th International Conference on pressure Surges Hannover 1986, Dynamic Behaviour of Large Non Return Valves covers this subject well.

Sharing knowledge is a way to immortality

stanier

I might not have been clear. Generally one shouldn't  concentrate of the speed of closure of the check valve but should concentrate of the (unacceptable) high SURGE PRESSURE accossiated with the closure of the check valve. It's the pressure surge that needs to be avoided.
If these surge pressures can be kept to acceptable levels with a 'slow' check valve and the check valve reliability is not negatively influenced by this then there is no real reason to have a sofisticated quick closing check like like the Mokveld TKZ-Y.
If on the other hand a the pressure surge created by a 'slow' check valve is excessive for the design of the system, than a fast acting valve will reduce the surge pressures to acceptable levels.

So concentrate on the surge pressure. When the surge is acceptable for the systems design than you have a suitable check valve with regards to speed of closure. (this doesn't ofcourse cover all aspects of selecting the proper check valve, there are many more issues involved.

P.S. The paper referred does very accuratly cover the subject.

RobV

Im working on a pressure surge analysis and i dont have data for the checkvalves used - so i try go get a general idead of the closing time. The program that i use (flowmater) requires this info.

By the way: I newer said gas - actually its stabilised crude.

Best Regards

Morten

The reference Thorley gives the industry data on the dimensionless relationship of deceleration vs maximum reverse velocity. These data can be used to provide the data required for computerised surge analysis programs such as AFTs Impulse. The raw data has to be generated by the check valve designer/manufacturer. Unfortunately the less sophisticated check valves have not been tested.

I disagree with RobV about the importance of the check valve closing time. The surge pressure depends upon this characteristics of the check valve . Ergo one has to model various different check valves to ensure that the valve selected has characteristics that present a maximum reverse velocity below that which the system can tolerate without check valve slam.

Sharing knowledge is a way to immortality

MortenA,

We recently performed a pressure surge analysis involving check valves and ran into the same problem you are facing -- it is difficult to obtain data on check valve closing time, yet the pressure surge analysis is highly dependent on this parameter.

So, we simply assumed a "fluid float" model in which the valve face moves at the fluid velocity.  Thus, at the instant of fluid reversal (i.e., the zero flow condition) the check valve is still full open.  As the flow reverses and accelerates, the check valve begins to close.  We further assumed a characteristic closing length of one pipe diameter.

We ran the transient calculation twice.  The first time, we performed the calculation with out any check valve to determine the dv/dt of the fluid.  We then integrated the fluid acceleration to determine how long it would take for the fluid to travel one pipe diameter in the reverse direction.  This is the time we assumed for check valve closure in the second iteration of the transient calculation.

Based on other comments I have read, the check valve would likely close faster than predicted with the fluid float model.  Thus, the fluid float model should provide a conservative estimate of the valve closing time since the longer it takes for the valve to close, the higher the fluid velocity at the time of closure and hence the larger the pressure surge.

We have not benchmarked this model against pressure surge data for closing check valves, but would like to do so.  If any one has access to such data, we would be happy to benchmark the model.

Tremolo.

Tremelo,

Your approach sounds very sound. The only rider I would add is that with spring assisted valves such as split disc and nozzle type the valve will close more quickly.

So you could argue that this gives you a conservative number. What do you do if the conservative number results in too higher pressure or indication of valve slam. Would you invest in surge mitigation devices to dampen the transient?

The approach in Thorley gives dimensionless responses for various type of check valves based upon industry data. This can be used to aid check valve selection to avoid slamming. This will give you the most economic selection. Otherwise you could go straight to a nozzle type that you may not need.

Sharing knowledge is a way to immortality

Stainer:

unfortunately i dont have time to access a technical library / buy books for this job But if its a good reference for pressure surge i would maybe buy it.

Best Regards

Morten

Stanier,

Thanks for your feedback.

If our conservative model produces unacceptably high pressure surges, then we would have to sharpen our pencil rather than recommending purchase/installation of new equipment.

The Thorely reference sounds interesting and would be useful.

Here's another reference that may be of interest.  It's a paper published in the 7th NRC/ASME Valve and Pump Symposium, July, 2002.  You can access the PDF file of the conference papers at:

http://www.nrc.gov/reading-rm/doc-collections/nuregs/conference/cp0152/v4/

There are 4 pdf files.  Open the second file (Labeled Part 2) and see the article on pages 2A-77 through 2A-99.  This article presents test results of axial flow check valve closing time in the form of "performance graphs".  The article also provides a method for calculating water hammer pressure due to closing check valves.

Tremolo's approach may lead to over-conservative reverse velocities, peak pressures and finally a more expensive design. The best (only) way to model check valve behaviour correctly is to use the dynamic characteristic of the check valve (reverse velocity versus fluid deceleration).

The dynamic characteristic of check valves can be measured at Delft Hydraulics (see www.wldelft.nl/facil/nonret/index.html for test facility details)

Regards, IPSYS

ipsys,

Thank you for your feedback.  I am familiar with Delft Hydraulics Lab and have used some of their experimental data to benchnmark my fluid models.  

The Delft papers I am familiar with have been well written and have contained thorough descriptions of the test configuration and test results.  The papers have been quite interesting and useful for model benchmarking purposes.

I will visit their web site to see if any publications are available for download related to check valve testing.

Tremolo.

ipsys,

I was able to download the following article from the Delft web page:

Lavooij, C.S.W.;Koetzier, H.;Kruisbrink, A.C.H., Dynamic behaviour of large non-return valves., 5th International Conf. on Pressure Surges, Hanover, F.R. Germany, 22-24 September, 1986. (co-authors), 1986.

This paper provides "dynamic characterisric" curves which plot the maximum reverse fluid velocity against the average fluid deceleration during check valve closure.  Once the maximum reverse velocity is known, the waterhammer pressure can be calcualted using the Joukowsky equation: dP = rho*u*a, where dP is the waterhammer pressure rise, rho is the fluid denisty, u is the fluid velocity, and a is the sonic velocity in the fluid.


I will compare results from the "fluid float" model I mentioned previously against the experimetntally determined dynamic characteristic curves presented in the Delft paper.  I will provide another post with the results of this comparison.

Tremolo.

Tremolo,

I am looking forward to the comparison.