I am attempting to set up a model to demonstrate how a RPZ (Reduced Pressure Principle Backflow Preventer) will respond to Hammer.
For those unfamiliar with the operation of an RPZ, I have attached a document which outlines this very well.
It is basicly a double check with an atmosperic vent between the two checks. This releases water to atmosphere to keep the pressure between the two checks lower than the pressure upstream of the RPZ (upstream is the municipal distribution system).
I will describe below what is occuring in the distribution system, which should explain why I am setting up this model:
A municipal distribution system is experiencing water hammer throughout one of its zones. This zone is served by a pump and a couple tanks. The utlimate source of the hammer is uknown. The pump and tank operations have been thoroughly analyzed and it is generally believed they are not the cause. Its most likely a hydrant being opened or closed too quickly. During the Hammer, the ground shakes around hydrants, 4" service connections dance all over the place , people all over the pressure zone call asking if there's an earthquake, PRV's on the system have blown discs. It would be expected that this would damp out after a few minutes. This lasts for 45 minutues to 2.5 hours. It happens every few weeks or so (highly variable), but it tends to be in the morning on weekdays (pump on the system not running, tanks full).
Five buildings in a development have RPZ's. When these RPZ's are isolated from the distribution system the hammer stops. During the hammer events water shoots out of the RPZ's (from the atmospheric vent between the two checks) and the valves sound like a machine gun (constant opening and closing). It is theorzied that the action of the RPZ's propagates the hammer.
What I would like to know is what is the best way to model this to either prove or disprove that the RPZ's are propagating the hammer. I do have an idea:
Run a model with a fast hydrant closing to start a hammer event to get the graph of the surge (I have done this and the graph of pressure shows the pressure flucuating very quickly above and below HGL (maybe 100 times a second). Add two checks and a TCV in between wherever I want an RPZ. Have the TCV vent to a resevoir at an elevation to match atmospheric pressure (or a tank). Setup the throttling of the TCV to match the graph of the surge (with the TCV opening when pressure is below HGL in the distribution system, which is when the RPZ discharges to atmosphere). I will have to get the closing times of the checks from the RPZ manufacterer or make some assumptions.
Any idea? And if anyone has any other ideas as to what could be causing the hammer to continue for so long, I would appreciate it as well (I know I said the hammer stops when the RPZ's are isolated, but those who say that have some stake in it being the RPZ's fault, so I take it with a grain of salt)
Thank you
I too would think it would be unusual for RPZs to be the source of the hammer, but wouldn't discount them playing a part in the water hammer event.
I would try to look at this through empirical evidence to start with. Is it possible to procure some pressure transient loggers (eg. Radcom units)? Then you can get an idea of exactly when the event starts, and what measurable affect different isolations/changes might have on the events.
We've hunted down/investigated some water hammer events with the help of our pressure monitoring equipment. With pressure logger units similar to above, we even traced down some serious hammer and main bursts being caused by a customer's newly installed, large fire pumps. On that line, is it possible to talk to each of the property managers in the area to find out if they've changed anything on their premises eg. Fire equipment?
I will definately look into data logging.We do lack good analytical evidence.
Currently, in terms of records, all we have are years worth of circular pressure charts for four of the buildings. The pressure is read after the jockey pumps, which pressurize the sprinkler system (usually ~160 psi). Also, there are some issues with these records: charts not changed regularily, ink smudged everywhere when charts changed, fire pump tests, etc. I have gone through these charts and all I really get out of it is, is a general idea of how long the hammer events last (when I can pick them out of the noise on the charts).
The five buildings in question have fire pumps, and like all the other buildings in town with fire pumps, they are tested yearly. The largest in town is 3000 gpm (not in 5 buildings in question). We know and have witnessed the fire pump test procedure for the five buildings in question. We are in the process of collecting information on the rest of the fire pumps in town. We do plan to inspect them all. We have been told that there are no other RPZ's in town, but we plan to confirm this. We are going to get the dates of all fire tests and try to match them up to reports of hammer.
In fact, the annual fire pump tests for the five building in question were a week ago. Two went fine, the third caused a hammer event, the RPZ's in all buildings were isolated and it stopped. The final tests were performed without incident with the RPZ's in the buildings isolated. Also, the start of the town hydrant flushing program seemed to cause hammer problems (the program was suspended due to this). The rate of increase and decrease of flow in the fire pump tests were slow (maybe not slow enough?). All water department and fire department employees know how to operate the hydrants and valves, of course it is not known exactly what goes on in the field.
The town water department says that there were no surge problems until the buildings with the RPZ's were constructed.
The working theory is that fire pumps, hydrants operations, etc cause a small hammer, which may otherwise go unnotived, but the RPZ's cause it to be worse and continue longer than they otherwise would. I would like to model this to see if it is a possibility.
Thanks for your reply and would appreciate any other ideas you may have.
It's true that there is no element in HAMMER that represents an RPZ exactly. Your workaround might be a little tedious to set up, but I think it would give you a reasonable idea of whether the RPZ is having a major influence (or not) on the transient event. Another (admittedly fairly crude) option you could try is to model this as a Surge Anticipator with a checkvalve on the downstream side. The SAV can be set to open and close based on pressure, and will discharge to atmosphere just like the RPZ.
Set the SAV to open when the pressure is say 5psi lower than the steady state pressure at that location, set the SAV Closure Trigger to "Threshold Pressure" and set the Time for SAV to Open / Close to some small number (this would represent the time it takes for the PRZ relief valve to open and close). You will also need to enter a Discharge Coefficient, which you should be able to approximate or get from the manufacturer (this is the discharge coefficient for flow that is discharged to atmosphere when the PRZ relief valve is open. Set it very high to get high flow rates through the relief valve).
In the attached figure, assume that steady state HGL is 100 psi. After the initial transient (from fire hydrant closure) starts, the pressure at the SAV will most likely rise and fall. If it falls below 95 psi the SAV will open fully over a time of 0.1 seconds and discharge 1 cfs / (ft head)^0.5 (as per the discharge coefficient). When (if) the pressure rises above 95 again the SAV will start to close, and will close fully over 0.1 seconds and this will hopefully get you the sort of rapid fire opening and closing that you see in the real RPZs. This should be pretty quick to set up...and it might tell you if this is enough to sustain the transients for such a long time (2.5 hrs is one long transient!). However, as you can see, it isn't modeling the RPZ behavior exactly so you should interpret any results accordingly.
Note: I only had one check valve in this diagram because I wasn't whether the second checkvalve would prevent the SAV from opening (or closing) . However you could experiment with the second checkvalve (perhaps give it a longer time to close so that it doesn't isolate the SAV from the rest of the system instantaneously). It could be that the slamming of either (or both) of the checkvalves is also having an impact?
We have listed the RPZ valve element as a HAMMER feature request for further consideration. I would be very interested to hear if you figure out whether these valves are indeed having a big influence on the transients in your system (as this would certainly have change the priority we would give to adding this element to HAMMER).
Good luck!
Regards,
Mal Sharkey Product Manager Bentley
I tried your method out. One problem I am having with it (and not completely unexpected since it is just an approximation) is that the SAV opens to the atmosphere when preassure is below a set limit at the SAV itself . However, in an RPZ the relief valve in between the two checks only opens to atmosphere when the pressure in between the two checks exceeds the pressure upstream of the device. Also, the first check cause a ~8 psi drop. What is occuring in my model with the SAV is that after around 30 seconds pressure drops to 0 at the SAV and remains there. The RPZ responds to a changing pressure. The pressure at which the SAV responds has to be locked in. I see how the SAV is a decent approximation since it responds to low pressures just like the RPZ responds to relatively low pressures upstream of the device. The lack of the response to dynamic pressure change seems to cause the SAV to help damp out the transients instead of propagate them.
It is a possibilty that the slamming of the checks is causing or contributing to the problem.
Thanks for your help
You actually seem to have done a fair amount of background investigation, which is great!
The most likely culprit probably is with the property fire systems, and as you're saying, it is perhaps the slamming of the check valves on the RPZs that exarcebate the hammer.
Just some more food for thought, but some other events I've run into:
I've only thrown that one in because you're suggesting this equipment should only operate once a year but you are getting weekly hammer events?