HI Scott,
Thanks for the responses. I believe that following your response I managed to achieve a good progress with model. I've conducted the following analysis.
a) I ran the model without any surge protection (Scenario All Pump Trip_No Protection_CVs). The results showed high cavitation as expected.
b) I added air valves at many locations where I found vapor cavities (Scenario All Pump Trip_Avs). I successfully managed to reduce the cavities to almost zero in the whole model, but there were still negative pressures in the system, so I decided to put a surge vessel.
c) I then added a surge vessel d/s the pumping station (Scenario All Pump Trip_Avs_Vessel). I manipulated the inputs until the negative pressure d/s the pumping station almost disappeared. However, I then noticed from (Profile-1) animation that the remaining small negative pressure occurs because as the simulation progresses, the HGL keeps falling, endlessly as it seems. To confirm this, I increased the simulation time from 200s to 500s, and I saw that the HGL kept dropping progressively.
Question 1: I'm wondering is this the case because I'm using emitters, or is there another reason?
d) I then ran the analysis with the surge vessel but without the air valves (All Pump Trip_Vessel). I was surprised to see that the results were identical to the scenario described in c).
Question 2: This last result suggests that air valves in a system that uses a surge vessel do not make any difference. I expected that I needed a smaller surge vessel in the system with air valves, because air valves will help reduce the negative pressures. What's your opinion on this one please?
e) I then tried to setup the model to run an (All Pumps Start) scenario. I followed Bentley's technical note "Modeling A Pump Start-Up Transient Event In Bentley HAMMER V8i [TN]", but I couldn't get the model to run. It seems that because the pumps are initially set to "OFF", the model won't run. Could you please help with this one?
I'll upload the model. Please note that because I placed the surge vessel immediately on the pumping main, the scenarios above that don't use surge vessel can be run only if the surge vessel is deleted.
Thanks and Regards,
Fadi Sirrieh
Hello Fadi,
As these are a different set of questions, I have split this thread into a new forum post. Whenever there are new questions, it is best to start a new thread, as this will allow future users to better search the forum for answers to similar questions.
I do not believe that this is related to the user of emitter coefficients. The emitter coefficient is basically calculating the demand. The only thing of note in regards to demands in HAMMER is that demands are considered to be pressure dependent (though only in the transient solver). The transient solver uses the demands calculated in the initial conditions as the starting point, and then adjusts this demand as needed based on the change in pressure. So as the pressure decreases, the demands at a node decrease as well. That might explain why the flow from the hydropneumatic tank appears to be slowly descreasing with time, as shown in a Time History graph of the results at the tank.
As for the hydraulic grade dropping, the hydropneumatic tank is sustaining the flow, but it is draining. As the hydraulic grade drops in the hydraulic grade, the pressure drops in the rest of the system. This in turn causes the demand to decrease. With less demand, the flow decreases.
The tank is the only source of flow in the system and because of the demands, it will continue to drain. It is not the expected result, you may need to look into the setup of the model in order to try simulate what is actually happening in the system.
Regards,
Scott Kampa
Bentley Technical Support
Answer Verified By: FADI SIRRIEH
If you look at the time history graph for the air valves in the scenario "All Pump Trip_Avs_Vessel", the air valves are contributing. You can see the air volume increasing with time. However, air valves in HAMMER give protection right at the air valve. Moreover, it appears that the air valves are not even active until the pressures drop late in the model run.
If you look at an animation of the profile, when the pumps shut down, the hydraulic grade near the pumps drops in the first few time steps, but recovers because of the hydropneumatic tank. However, when the pressure drops, it remains positive. with the pump off, the only source of flow in your model in the hydropneumatic tank. As stated in the previous post, the hydropneumatic tank is slowly draining and the hydraulic grade is decreasing. Eventually, the hydraulic grade drops enough that it is below the high point in the model profile. The air valve starts introducing air into the system at this point, but that is not going to keep the pressure from becoming negative at all points in the model. The introduction of air will keep the pressure at zero at the air valve, but it will not solve the issue.
You may want to consider looking at the modeling case you are using. For instance, the pumps are shutting down, but is there a time when the pumps can be expected to turn back on? Obviously, this would increase the hydraulic grade downstream and keep negative pressures from occurring.
You could try modeling the air valves in line instead of at a tee as well. This still gives viable results and would allow you to better visually show that the air valves to contribute. However, the issue will still be there. The pressures will decrease as the hydraulic grade in the hydropneumatic tank.
The likely cause of the issue is that the demands in the system are disconnected from a source. With the pumps turned off, the reservoir is unavailable to supply the demands in the system. Since you are using emitter coefficients, the pressures are zero (or negative) and the demands used by the transient solver are zero as well.
There are two issues that can occur because of this. First, the demands in the transient analysis will not be representative of the demands in your system. The transient solver would be using zero demands instead of the demands as calculated from the emitter coefficients. This can skew the results, as the demands in the system will not be representative of the system.
Second, you can see cases where the pressure is at the vapor pressure in the ititial conditions. That is happening in your model when you try to compute the transient solver. Basically, there is already air or vapor in the system before the transient solver even runs. This is not allowed by HAMMER. See the following solution for some additional information on this: communities.bentley.com/.../9699.aspx
Some adjustment to the model will be necessary in order to model this case. First, you could add another source, such as a tank or reservoir, that would allow demands to be satisfied. You could then get viable results out of the system. Second, you could model the junctions with the demands as Discharge to Atmosphere elements. This allows you to enter a typical flow and presssure drop, essentially creating a curve that allows HAMMER to adjust the flow at the element. This curve works in the transient solver and could allow for better results presentation in relation to demands.
Hi Scott,
I got nowhere with this. Why would I need to add a source that in reality will not be there? Moreover, when the pumps start, the tank or reservoir will then become either a permanent source or demand that does not exist, and therefore will affect the results.
I didn't try discharge to atmosphere because this requires global changes to the model. I will give it a go, but I'm not convinced that this is the right way to resolve the issue.
In any pump start-up model, the initial demands will be more than zero and the pumps will be off. The model engine should be taking this issue into consideration.
If I set the pumps initially to "OFF", then the initial condition won't run because the system is disconnected from the pumps. If I set the pumps initially to "ON", I get an error message that says the pump start-up analysis cannot have steady flow more than zero. Nowhere to go!
My real question is, why wouldn't the model work, although I've set it up according to Bentley's TN "Modeling A Pump Start-Up Transient Event In Bentley HAMMER V8i"? Does the TN need an update?
The TechNote assumes that the demands in the system can be met. In the TechNote, it states that the program "assumes a steady-state analysis for the initial conditions and that you have storage downstream of the pump in question, or other pumps, either of which could supply the demands you have entered when the pump in question is off." In other words, if there are demands in the system, they still need to be satisfied in the initial conditions in order for the system to run.
You're modeling case is a little more complicated than that. You have emitter coefficients instead of a demand applied to a node. So when the pump is off, the demands are zero. After computing the initial conditions in the current setup, take a look at the results in the junction table. For all junctions, the demand is zero. This the demand value that is being passed to the transient solver, since the initial considers results are used as the starting point for the transient analysis. For that reason, when the pumps are turned on, the demands at all junctions will be zero. That will tend to skew the results if some demand is expected at these junctions.
Pressure is also an issue. After computing the initial conditions, many of the pressure values at the junctions are zero or negative. This is not unusual when using emitter coefficients with no source of flow. However, starting a negative pressure in the initial conditions is not a viable starting point in HAMMER. Moreover, some of the pressures are less than the vapor pressure. Essentially, you have vapor or air in the pipes before you ever run the transient analysis, which is not recommended; pressures below vapor pressure are not allowable at all.
The reasons above are why one possible solution would be to have a downstream source in the model. That would allow for the demands to be satisfied and the pressure to remain positive in the initial conditions.
Based on your message, that is not possible in your case. Another possible solution would be to go about the pump startup in a different way. Instead of starting with the pumps off, start with them on. In order for this to work, you would need to use a different Transient Pump Type: "Variable Speed/Torque". You can set up an operating rule to slowly turn the pump off, allow the system to reach a new equilibrium, then turn the pump back on. This may take some trial and error, and you will need to be aware of the additional issues with the high point in the system (as noted in earlier responses to your question). However, this could allow you to see what happens when the pumps turn on without worrying about demand and pressure results being incorrect in the initial conditions.
If this doesn't help, if may be useful to know more about what is expected from the system, in particular, what is expected when the pumps are off. What kind of initial results should occur? Are the demands met by some other means, or are they assumed to be zero? Is there still positive pressure in the system? Any additional information related what you are expecting in the current setup of the model may be useful.
Thank you.