Check valve slamming Vs. Dynamic characteristic curves
If you have a deceleration of pipeline system as 3m/s2 during the pump trip, based on the dynamic characteristic curves, which valve would you expect to do the red colour visualized measured slamming? https://www.dhirubhai.net/posts/mina-milad-02658231_lockdowngames-pumpsystems-pumpindustry-activity-6699195043258892288-O-Up
#pumpsystems #pumpindustry #NRV #checkvalve
This was a question I dropped on the timeline and I need to answer it. I hope it will help you to understand the dynamic characteristic curves of NRV and to give you feedback regarding its reliability
First of all, you need to know how the dynamic characteristic curves are generated, how to apply it, then you can decide based on the observation results weather it is reliable or not. then we can discuss the better solutions
How the dynamic characteristic curves are generated
the figure shows the concept of the test station, you generate reverse flow to oppose the NRV and to close it. the deceleration (dec) is calculated and adjusted and the resulting pressure spike is measured and calculated in reverse using Joukowsky equation to estimate the reverse velocity (Vr). the same concept is used whatever the test station details are. then they plot the dec and the Vr, with or without dimensions to give a handy and simple indicator of slamming. the resulting graphs are classifying the slamming as 3 categories (no slam, mild slam, or severe slam).
The disadvantages of the dynamic characteristic curves:
- the graphs are generated for a specific valve type and size and settings. so if there is a different counterweight or assisting spring modification, the results are not valid.
- some manufacturers are using this single test as an indicator to their whole line of valve type sizes which is misleading.
- some valve manufacturers are adding their curves without proper testing
- the generated data has no input to the behaviour of the NRV (shattering or reopening as per the site conditions)
- some cases the same as we have here the graphs were misleading
- It needs a high Frequency pressure sensors in order to capture the slamming even if you could feel it or see its consequences
How to apply the dynamic characteristic curves.
Usually, engineers are estimating the system dec by hydraulic modelling or by site measurement, then they go to the graph applying the dec value to find the Vr and whether it is in the safe category or not. in our case we can enter the dec as 3m/s2, the NRV was one of the top 4 non-slamming NRV which are all close and safe at the defined dec.
the expected resulting Vr is smaller than 0.1m/s which is on graph safe!
The observation in an actual case
while in real life we had a sever slamming with a 6 bar magnitude circa!
To translate the result we need to know that the constructions of the station were shaking and the slam was so loud, and if we analyze the slamming we will find a rapid impulsive load caused by applying 6 bar on a 400mm disc.
Which has the effect of 631000*(PI*0.2^2) = 79293.8 N
Equivalent to an 8.08 ton falling on each SB NRV disc.
You shall multiply the force by 4 (total number of NRV 3Duty +1standby) hence we can imagine equivalent load of 24-ton falling.
The better solutions
the Ideal non-slamming check valve which has a 0 Vr is a unicorn, and the approach of using very fast closing NRV is not safe always and more dangerous with bigger valve diameters. at the end, we will have a dynamic range of dec for each system depending on how many pumps are running as duty and if it is a trip or ramps down. the current ideal non-slam NRV can match one case if it could but it needs to be tested under the operating conditions which is unsafe and could cause a total loss for a pump station.
Even doing the 3D CFD analysis is not a solution, it is takes time and it will be valid for the fixed condition you study. can't be the general solution. if we assume it will get a proper results without readjusting the model based on site measurements
some NRV developers are using fixed time hydraulic assisted NRV which could match a singular condition and trouble all other. while few developers have developed a proper fixable solution with limitations, while others are trying the best with armature mechanisms which need site adjustments (dangerous).
Conclusion:
I can say that the Dynamic Characteristic curves are not reliable for me to select an NRV as a specialist engineer after practicing the analysis and following it up in sites.
If you need any help with an NRV selection or site slamming, let me know if I can help
If you are interested in developing a test station or a new type of non-slamming NRV I could help you.