Hi John,
There are a couple of steps to this, but it doesn't take too long.
First run the Initial Condition analysis and record the velocity in the pipe downstream of the check valve. Let's assume it is 4 ft/second.
Next, run the transient simulation and open the Transient Results Viewer. Plot a time history graph of flow in the pipe downstream of the check valve and measure the time from when the pump turns off until the flow curve crosses the x-axis (i.e. when flow reaches zero). Let's assume it is 2 seconds.
So the deceleration of the water column is 4 ft/second / 2 seconds = 2 ft/s^2.
This is an average deceleration rate, but from what I have seen that generally seems to be what the valve manufacturers quote in their literature.
Also, I will see if we can get an actual deceleration rate result field into HAMMER at some point in the future.
Regards,
Mal Sharkey Product Manager Bentley
Mal,
Thanks for the help. I have one more question.
I want to model different types of check valves in a pump / pipe system. I have Reverse Velocity vs. Deceleration curves from various valve manufactures. How can I model these valve curves in Bentley Water Hammer.
Thanks
That type of analysis requires a bit of judgement and interpretation. Let me highlight some of the issues...
Those reverse velocity vs. deceleration curves don't tell you much about the relationship between flow and headloss in the check valve, or the rate the valve is closing (is it a linear closure, or does it start closing slowly and then accelerate?). What they do give engineers is a flow velocity 'delta' to use in Joukowsky's Equation, dH = dV . a / g (where dH is change in head, dV is change in velocity, a is wave speed and g is gravitational constant - see Bentley's Advanced Water Distribution Modeling & Management book, section 13.3 for details) so you can do a quick check of the change in head in the pipe as the check valve slams shut. So I would say these curves are a bit more of a 'rule of thumb' than actual physical valve parameters and should be used with appropriate caution - but they are usually the best information available from manufacturers on dynamic behavior of check valves.
Now HAMMER V8 XM currently assumes that check valves close instantly upon sensing reverse flow (i.e. reverse velocity = 0) and it sounds like this is not really what you want (unless the reverse velocity is small). The next HAMMER release (due in a few weeks) will allow users to specify a check valve closure time and opening time (over which the check valve is assumed to close linearly), and this will make the modeling of dynamic check valve behaviour easier. In that case you would set different closing times and run the model, then review the results and look at the flow through the valve as it shuts. You might need to iterate a couple of times until the reverse flow corresponds to your reverse velocity for the valve.
In the meantime in HAMMER V8 XM, you should use a regular valve if the reverse velocity is high as these allow you to control the closure time (you just need to be careful that the system pressures don't dictate that a check valve would open again. If they do, you would open the valve again at that time). If the reverse velocity is not high, then it should be ok to model them as regular check valves that close instantly. You might ask what I mean by a 'not high' velocity, but I think that's an engineering judgement call that could vary between models.
In the future, something we could do for check valves is allow users to enter a reverse flow velocity directly. We would probably need to assume that the valve stays fully open until the reverse velocity is reached, then slams shut instantaneously. This would be a conservative assumption, but that's probably ok. Is that something you would be interested in seeing in HAMMER?
Mal Sharkey
Product Manager Bentley
Mal Sharkey: This is an average deceleration rate, but from what I have seen that generally seems to be what the valve manufacturers quote in their literature.
Hi Mal,
This is an interesting topic and one we have been struggling to answer using the transient results. Perhaps using examples from our case would help enrich the discussion. Please find below flow deceleration and dv/dt estimates for a single pipeline where water is pumped from about 50 m to 250-280m via a single pump station near the source. We have two flow cases to consider in our surge analysis and found dv/dt estimates changes dramatically if we literally look at the changeover point vs. a representative dv/dt before flow reversal occurs. One thing to note in both flow cases is there is a prolonged duration of low flows (ignoring the flow direction) before a sudden/sharp drop in flowrate occurs. I say sudden because the flow drop dv/dt=15 m/s2 estimate is based on occurs in one time step according to the results from Hammer.
Also, one manufacturer curve I have indicates, as I understand, the lower the dv/dt ratio the longer it takes for a check valve to close and therefore, the worse the transient pressures will get. Is that true? Although the manufacturer curves include Reverse Velocity as well, closing time is the only parameter with which we can test the sensitivity of results in Hammer. There seems to be a correlation between Reverse Velocity and dv/dt that I do not fully understand.
Regards
Sukru
Hi Sukru,
I think that what you see in your results comes back to the fact that water column deceleration in a real water system (as opposed to an ideal one) is often not constant or regular. That's why I mentioned in a previous post that the charts you see from valve manufacturers showing max. reverse velocity vs. deceleration are more like 'rules of thumb' than explicit valve properties, because I believe that those charts are mostly created in laboratories using simple, ideal systems. I expect that if you tested those same valves in a real system (like the one you are currently analyzing), the maximum reverse velocities could differ.
There is a good discussion about dynamic check valves in Wylie and Streeter's 'Fluid Transients in Systems' (1993, pg. 252). It shows a chart of maximum reverse velocity versus mean deceleration |dV / dt| for a number of valve types.
As I understand it, the relationship between the max. reverse velocity and mean deceleration is as follows:
- flow initially travels forwards through the checkvalve, generally up hill in the case of a pumped main; - if the pump stops, gravity causes the column of water in the pipe to decelerate; - if the pipe has a steep incline the flow will decelerate quickly. If the incline is not steep the flow will not decelerate as quickly. Either way though, the flow will soon reverse and start to come back down the hill; - the check valve begins to close as the flow reverses but this doesn't happen instantaneously (i.e. for swing type valves the disc must travel a certain distance to close off the valve and that takes time. There might also be a damping mechanism attached which slows the disc movement, etc.) - Once the flow reverses, the flow starts accelerating down the hill and it will reach a certain reverse velocity before the check valve has time to close and completely stop the flow. If the pipe has a steep incline, the reverse velocity will get relatively high (since the flow will accelerate quickly down the hill). If the incline is less steep, the reverse velocity will be lower.
This relationship is complex though because faster reverse velocities will impart more force on the check valve, influencing the time to close. However, in answer to one of your questions, it always seems to be the systems with the highest mean deceleration that result in the highest reverse velocities, which in turn result in the highest transient pressures as a result of check valve closure.
This discussion doesn't really help you find the right check valve 'time to close' parameter though, but I think that will come down to engineering judgement, as well as a few sensitivty studies covering a range of likely check valve closure times. Using the valve manufacturers 'rule of thumb' curves as a guide, the 'time to close' will be approximately equal to the max reverse velocity divided by the mean deceleration.
In the future we are considering one or more of the following enhancements in HAMMER:
1. Allow users to enter a max. reverse velocity instead of a time to close at a check valve. Once the max. reverse velocity is reached the valve will close instantly
2. Allow users to enter a mean deceleration vs. max. reverse velocity curve and possibly also a sampling period for computing the mean (i.e. compute mean deceleration over the last X timesteps, where X is user-defined)
3. Detailed physical modeling of the check valve (i.e. analyze the different torques acting on the valve and compute its opening / closing rate accordingly). The problem with this approach is that it is hard to generalize (because there are many different types of check valves), and also it is hard to get detailed information on things like spring constants. disc weights, etc.
Hope that helps.
Mal