Negative pressure at high points

Question

Hello 
 How can i resolve a problem of negative pressure in water hammer intial condation?? i am delling with a system of 16 km leangth & 2 working pumps with 162m3/h @ 98.8m each at 14 Km i have some high points where gravity flow for 1.2 Km has been considered from the most higher point (310.15m) to the discharge tank(MWL =298m) the pump head has been calculated based on the assumption of gravity flow. hence, water to be pumped only to 310.15m. 
 i tried many time to insert som air valves but still the problem as it is 
 please helpe me 
 HGL for the system is in shown below:

Jesse Dringoli · Accepted Answer

Ali Khalfan said: I tried to put air valves at high points and treat as junction" set to false but still the same problem of negative pressure is there!! 
 Can you provide a screenshot? If the negative pressure occurs on the downstream side during initial conditions, this is expected and is interpreted as part-full flow as seen in this article . As discussed, ending the system at the air valve with the D2A element would be needed for a transient simulation. 
 
 Ali Khalfan said: when i am saying that system is pressurized up to high point I mean that the pump is designed to left water to that point then water will flow by gravity for 1.2 Km. this case only when 2 pumps are in operation while when 1 pump is operated before the second one swiched on the first pump designed to derive wated to the discharge tank not only to the highest point( no garvity flow in this case).hope this is clear now sir?? 
 Are you saying that you expect part-full flow (with the air valve open) *only* when two pumps are running, and that when only one pump is running, you expect that the high point will be pressurized, with full pressure flow all the way to the actual end of the system ("discharge tank" with water level 290.2 m?) If so, my response to this is that I do not see how that could happen unless the single pump added an extremely large head. See diagram further below. I estimate that the pump would have to add roughly 300 m of head in order to keep the hydraulic grade about the high point (and thus keep it pressurized). 
 If you really do have such a condition where the high point is under pressure, then you would want to set up a separate scenario using active topology to make the downstream piping active, since the transient pressure waves would then travel all the way to the end (and you would likely have low pressure problems at the high point with air being admitted into the system from the air valve). You could then set up a separate scenario for the case where part full flow occurs downstream of the high point, by making the elements inactive in that scenario, and making the D2A and connecting pipe active. 
 If you run a steady state with only one pump running and the system modeled all the way out to the end, with the air valve at the high point, do you really see the HGL as the red line below? (pressurizing the high point)

Jesse Dringoli · Answer

Hello Bushra, 
 When using the air valve element, you will need to set the "treat as junction" property to "false" in order for the initial conditions calculations to "see" it and have the upstream pump know that it needs to add enough head to overcome the high point. See more in these articles from our Wiki: 
 Why do I get a negative pressure at a high point in my system? Shouldn't the pump add enough head to push the water over the hill? 
 Modeling Air Valves At High Points in WaterCAD or WaterGEMS 
 However, this doesn't work well for a transient run because the real system would have part-full flow in the piping downstream of the high point when the air valve is open, and there would be an air gap that the transient wave wouldn't pass through. An option to consider here is to replace the high point with a Discharge To Atmosphere element and delete all the piping downstream of it. See the section called "Transient Simulation Implications in HAMMER" in the second article above.

Jesse Dringoli · Answer

Yes, if the high point experiences part-full flow under normal operating conditions (and in the initial conditions for your transient event), then the D2A approach would be a better model representation of the conditions, and should yield better results than if you tried modeling it as an air valve. Meaning, if your air valve is normally open, with some air in the downstrream side, then there would be an air gap and any transient in the upstream pipeline would not pass further. It would act as a free discharge into the part-full pipeline and some wave reflection would occur - hence the D2A. I made a little diagram for you, and have added it to the related articles:

Regards,

Jesse Dringoli
Technical Support Manager, OpenFlows
Bentley Communities Site Administrator
Bentley Systems, Inc.

Jesse Dringoli · Answer

Please see the link in my first post: 
 Modeling Air Valves At High Points in WaterCAD or WaterGEMS 
 If you add an air valve element at the high point and set the "treat as junction?" property to "false", the upstream pump will "see" it as a hydraulic boundary so that it adds enough head to reach the high point, with part-full flow downstream.

Jesse Dringoli · Answer

Sorry, looking back at the thread I recall now that you are using HAMMER. Have you tried the D2A approach as suggested? 
 The air valve node with "treat as junction" set to false will yield the "correct" pump head since the pump will need to "lift" the water to the high point (if you are not seeing this, then there may be a problem with the model configuration that we would need to look into). However it is best used with a steady state or EPS in WaterCAD or WaterGEMS. For a transient simulation in HAMMER, consider using the discharge to atmosphere (D2A) element at the high point location as previously mentioned, since the transient waves will not travel further into the part-full flow portion. The pump head should also be "correct" in that case because it will also need to add enough head to "lift" the water to the D2A elevation. 
 The PSV + tank approach would be a bit overly complicated and you would need to consider how a wave reflects off of those boundaries vs. the free-atmosphere D2A boundary. Using the D2A should be a simpler, cleaner approach that should yield acceptable results if the D2A is configured correctly to represent the outflow characteristics at the pressure-to-gravity transition.

Jesse Dringoli · Answer

Ali Khalfan said: I have tried to change the D2A setting but i could not get 324m3/h. How can I get this value?? 
 The D2A establishes a pressure dependent demand based on the entered "typical flow" and "typical pressure drop". So, if the pressure at the D2A is above the setting, the outflow will be above the flow setting. You cannot force it to provide a specific flow -instead, you should enter the "typical flow" and corresponding "typical pressure drop" based on the characteristics of the opening to the atmosphere. One way to estimate this would be to assume an estimated pressure (likely near-zero), then use the orifice equation to calculate what the flow would be at that pressure, using the diameter of the pipe, then use the resulting flow as the "typical flow" at the D2A. (see the note about this in the D2A wiki article ) Even if this value is different from what you want the flow to be, the hydraulics would dictate what the outflow becomes, based on the D2A characteristics as well as the system conditions and the pump characteristic curves. (so you may find it could be closer to what you expect. 
 Another approach would be to run a steady state using a reservoir at the high point location and elevation, compute a steady state, and see what the flow to the outfall is. Then, morph the reservoir into a D2A and use a near-zero value for the "typical pressure", and the observed flow for the "typical flow". 
 Ali Khalfan said: also it is noticed that each pump give different flow & head values where they are identical with same suction & dischrage pipes proparties !! 
 Are the pump definitions and relative speed factor the same? Are you sure that there is no difference in the suction or discharge pipe length, minor loss coefficients, diameter, or roughness? To assist further with this, we'd need to see a copy of the model: Sharing Hydraulic Model Files on the OpenFlows Forum 
 Ali Khalfan said: As well as when the transinet sumilation is done after Shut Down After Time Delay has been selected the velocity, Flow & speed cuves shown as constant line?? 
 This could be due to a number of reasons. Check things like: 
 
 Make sure the pump initial status is On and that the value entered for the "time (delay until shutdown)" is less than the simulation duration 
 Check the pump "pump valve type" and "time (for valve to close)". If you do not have a check valve in your pump, set the valve type to Control Valve and the time delay to a large number such as 9999 seconds 
 Check the settings in your pup definition's "transient" tab to make sure that an appropriate specific speed and rotational speed are entered 
 Check the transient calculation options and make sure the duration is set appropriately (something greater than the pump shutdown delay), and that the setting "specify initial conditions" is set to "false" (so that it uses the computed initial conditions) 
 
 If this does not help, please provide a copy of the model (see link above) 
 Ali Khalfan said: if D2A has been swiched by tank or resirvoir at high point what will be the effect?? 
 The transient waves will reflect differently with a storage boundary vs. a D2A. With the tank or reservoir, it would act as if there is a submerged body of water at the high point whereas with the D2A it will act as if there was an opening to the atmosphere. See more under the section "boundary considerations" in this article .

Jesse Dringoli · Answer

Ali Khalfan said: Is this approach work?? 
 The model will compute with a reservoir at the end of the system, at the air valve location, but you will need to use your engineering judgment along with the knowledge of transients and HAMMER assumptions, to decide if that is the best approach. As mentioned in my previous post, using a reservoir will set a fixed HGL and the model will solve around that (and you may or may not get the desired flow, but it will be the flow needed for the pump to lift water to that reservoir), but the transient waves will reflect off the reservoir differently than it will reflect off a D2A, demand, or other approach. The degree to which this difference impacts the transient results you are interested in may be different for every modeling situation. 
 It might be best to try both approaches and compare the transient response (try the reservoir, then try converting the reservoir to an equivalent D2A per my previous post, compare the transient min/max envelope)

Jesse Dringoli · Answer

Ali Khalfan said: how can I ensure that no negative pressure will accur in that deleted part of 1.2 Km during transient ?? 
 I assume there is no valve open/closure or other change in velocity occurring in that downstream portion. 
 If you are confident that there is part-full flow (and an air gap) at the high pint during the conditions that you are analyzing the transient simulation for, then the transient wave from the pump shutdown would not pass across the air gap. Based on the screenshot of the profile you originally provided, I don't think it is possible that the high point would ever be under pressure (the hydraulic gradient could not be steep enough for the HGL to be above the entire pipeline). In that case, the downstream pipeline would always be flowing by gravity, with part-full flow on the upstream end as water spills into the partly-full pipe at the high point. 
 Perhaps you could put together some figures to illustrate this, along with explaining the situation like above? For example you could include a screenshot of the profile of the model with the air valve included, showing part-full flow downstream of the high point. 
 Here is an diagram i had previously attached:

Jesse Dringoli · Answer

Where exactly is this second pump? Is it in parallel with the first pump, on the far upstream side of the system? 
 When you say "the pipe line will be fully filled with water", are you referring to the entire pipeline, or the pipeline just up to (upstream of) the high point in question? Based on the original profile provided, even if pump was properly adding enough head to raise the HGL up to the high point in question, the slope of the HGL would be such that the downstream pipeline would have to be flow part-full (meaning, an air gap / pipeline only partly full of water), which means the transient wave would not pass through and down into the last (1.2 km?) pipeline at the downstream end. 
 Please clarify using screenshots/diagrams if at all possible.

Jesse Dringoli · Answer

Ali Khalfan said: based on software assumptions in calculating desired head it takes the static diffirence between the suction & discharge tanks Plus pipes losses neglicting the high point. that is why negative pressure is shown in my first uploded profile. 
 Yes, that is how it works unless you model the high point as an air valve as previously discussed. (at least for the initial conditions / steady state/EPS) 
 Ali Khalfan said: in reility the designed head is enough while in the software my pump curve is not working since the gravity portion is not considered in the design. 
 The pump head will account for the high point if you model it as an air valve with "treat as junction" set to false, or if you cut off the system at the high point using a reservoir or D2A (it will add enough head to "lift" the water to the high point.) 
 Your most recent concern was about negative pressures in the 1.2 km pipeline downstream of the high point. Based on your description, the air valve at the high point (which I assume you will have as vapor would likely form if you attempt to model it as a siphon) would be open, allowing air into the system and causing there to be part-full flow in the downstream pipeline. In that case, the transient would not pass over the air gap and thus there would be no concern with transient pressure problems in the downstream pipeline. 
 After I first made this point, you replied to say that the system will be fully pressurized. I took this to mean that you do not think that there would be an air gap/part-full flow at the high point, but I do not see how that could be the case, based on the system profile. In your most recent reply, you then stated " pressurized only tell that point.", which sounds to me like you are saying that only the upstream portion of the pipeline will be under pressure, and that there would in fact be part-full flow (in which case it goes back to my previous point about how the transient wave would not pass across the part-full flow section). 
 Can you clarify what your current concern is? Apologies, but I am having difficulty understanding.

Jesse Dringoli · Answer

Ali Khalfan said: in regards of single pump operation please note that it will be operated at lower speed with duty point of 162m3/h @ 59.4m head this head has been calculated based on static difference between high point 311 & suction tank 271.7m pluse losses inside all pipes up to discharge tank (21m). so the system will be fully pressurized in this case sine pump head can over come the high point level along with pipes losses. 
 If the pump is only able to add enough head to overcome the high point, since the slope of the HGL is less than the slope of the pipeline from the high point to the downstream boundary tank, then you will have part-full (partially non-pressurized) flow in that downstream section. This is what a steady state run will show you, with the air valve added. This means that there would be an air gap and the transient would not propagate past it, into the last, 1.2 km section. 
 Please see the diagram in my previous reply - in order to pressurize the high point, the pump would need to add around 300 m of head - enough so that the hydraulic gradient is steeper than the slope of the pipeline downstream of the high point (as seen in my diagram). Or, there would have to be a PSV upstream of the final tank, to keep the HGL at the downstream end high enough so that HGL stays above the high point.

Jesse Dringoli · Answer

I have split your new question off into a separate thread. 
 Ali Khalfan said: Now it is clear thanks alot but what about if i placed a PSV to rise the HGL to help in computing intial condation!!! is this approach will led to correct results?? 
 I'm not quite sure I understand what you are asking. If you use the PSV approach, the results will be "correct" if you will actually have a PSV there. My suggestion was not meant as a way to "trick" the model to get you the results you want to see. it is always best to set up the model to match the real conditions, while keeping in mind software limitations. Again, if the pump can only add enough head to lift it to the high point, then based on the topography provided, there will be part-full flow in the downstream piping. If a PSV is installed at the end of the pipeline, the HGL could be kept above the high point and pressurize the downstream pipe, but I'm not sure that this is any more desirable. Meaning, it might be OK to leave things as-is and have an air valve at the high point, with part-full gravity flow downstream. (unless you have some local requirement that prevents you from using air valves or having part-full flow)

Negative pressure at high points

Top Replies