This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Problem with model results and negative pressure

hello: i have a water pumping system with 5 reservoirs, 1 supply and 4 receiving  at branched elevations of 97m,102m,100m & 93m. My pump head is set to 113m. I can not seem to get results every time I simulate, at one time I got negative pressures at n

FOUR RESERVOIRS PUMPED SYSTEM HEAD PROBLEMPDF

Parents
  • Hello Xolani,

    What program are you using? The drawing doesn't look like it came from a WaterCAD or WaterGEMS model. 

    Assuming it is WaterCAD or WaterGEMS, what appears to be the issue? Are you getting an error message, like network unbalanced

    If there are pumps and valves in the model, you will want to check the settings to make sure they are accurate. For instance, a pressure reducing valve with a setting of zero will likely give inaccurate results. Overall, more information on what is happening may be needed.

    Regards,

    Scott

  • Hello Scott

    I am sorry for the late reply. I used Pipe Flow expect for the model but I will upload a model from WaterCAD with this reply.  The problem is I can not seem to build a model that can deliver water to all the tanks, I get negative pressures at some nodes resulting in flow direction changing. The data in the model (WaterCAD) is as it is on site, with deference in the pump and control valve. Pump info is a KSB WKLn 65/4 stage pump impeller dia 192 mm driven by motor with speed min: 2955-3555. the control valve is a Bermad Booster pump control and pressure sustaining valve pilot controlled. I do not know at what pressure to set the control valve so it can control the system

    Corana Pumping Scheme.wtg 

  • Thanks for providing the model file. You may want to review the settings of the valves in the system. For the VLAs, I don't often see these used in a WaterGEMS or WaterCAD model. In WaterGEMS and WaterCAD, the VLA is treated as a GPV with a headloss curve based on the valve's discharge coefficient. Basically, it would introduce a headloss based on the discharge coefficient settings. You may be better off changing this to another valve type.

    As for the PSV, this link has more informaition on how it works: How does the Pressure Sustaining Valve (PSV) work in WaterGEMS and WaterCAD?. You will want to make sure the intial setting of the pump is accurate. You may want to review the information here and in the help documentation to make sure this is the correct valve type for what you are modeling.

    The negative pressures you see are all on the suction side of the pump. This is occurring because the reservoir and the pump have the same elevation. Is the elevation for these actually zero? If no, an accurate elevation at these elements is recommended.

    The negative pressures are not the reason that the pump is not flowing to all of the downstream reservoirs. Instead, it may be related to the setup of the system.

    The first thing I would do is make sure that the elevation at and near the pump are correct, as well as making sure the pump curve is accurate. The next thing I would do is make sure the valves and valve settings are correct.

    Once this is correct, if there is no flow into all of the reservoirs, or flow from one reservoir to another, it is likely related to the elevations and assumptions on how pumps operate in WaterCAD and WaterGEMS. Pumps add enough head and the corresponding flow rate, which is defined by the pump definition, to overcome the losses and elevation differences between your upstream and downstream boundary conditions. Essentially, the pumps will deliver enough head so that after headlosses, the hydraulic grade at the reservoir is right at the reservoir elevation.

    In this model, the elevations of the reservoirs vary from 95 meters to 104 meters. This is causing an issue for the pumps. If the pump were to add enough head to reach 104 meters at the highest reservoir, this would be too much head for the reservoirs with a lower elevation, at least with the current setup. Essentially, the network would be unbalanced. As a result, the pump is adding enough head for the lower reservoirs, which is being supplemented by flow from higher elevation reservoirs.

    As a test, I also put check valves on the pipes connected to the four downstream reservoirs by opening the pipe properties and setting "Has check valve?" to True. This would stop flow from coming out of the reservoirs. In this case, there was flow into only the lower elevation reservoirs and no flow into the ones with a higher elevation. The reasoning for this is the same as above.

    If this is an actual system and you do see flow only into the reservoirs, you may need to calibrate the model. Calibrating may end up changing the headlosses and possibly the overall results. If the flow into the reservoirs is representative of demands, it may be better to model these as demand nodes instead. The pump flow will then match the demands.

    Regards,
    Scott

  • The VLAs are actual gate and globe valve, I did not know what to choose on waterCAD to model them. 

    The PSV,  is it possible to model this system without it? On site it is there and the Bermad expert informed me that it is a Bermad 700 series, model 743 (bosster pump control and pressure sustaining valve)

    Yes the elevation of pump and reservoir is 0.0 m or less than 1 m (see picture attached) but water level on the reservoir is around 2.2m  so I am not sure the reservoir model is done correctly. 

    Yes this is an actual system, I will attach a google earth link (for info). System starts at Ncipizweni pump station and ends at Gxulu reservoir as labelled  

    https://earth.google.com/web/@-31.59627048,28.85130381,763.30136399a,5318.41778411d,30y,0h,0t,0r/data=MicKJQojCiExQzBfZUxURkgyeGZfVjRLTDZ3OElQSVk4YS1uR29SWXQ

    Your answer is very informative, I will be working on the changes you suggest to see if I can get the system to pump to all reservoirs

  • Hello Xolani,

    We usually recommend modeling what is in the system. If the PSV needs to be represented, you may want to include it. However, you may want to review the pressure setting.

    For the VLAs, the globe valve is a valve type included in a certain type of valve. A GPV may be better suited here, though you will need a GPV curve that represents the headloss. If the healoss is minimum, you may be able to omit these from the model.

    Regards,

    Scott

    Answer Verified By: Xolani Tswina 

  • Hi Xolani,

    I took the liberty of downloading your model files (apologies if I shouldn't have - in any event I will delete all the files later).

    I have a couple of questions and some comments.

    1. Are you certain that you want to model R-2, R-3, R-4 and R-5 as reservoirs and not "tanks"?  There is a huge difference between how Watercad and the "real world" defines a reservoir.  From the looks of your system I would guess that you want to model them as tanks.  Modeling a storage vessel as a tank has several advantages over modeling them as reservoirs; a reservoir has a fixed hydraulic grade whereas a tank does not (the water level can fluctuate between the minimum and maximum).

    2. Are you positive that all of the control valves and check valves you have in the model are needed?  You typically do not need a check valve in a pump discharge as Watercad will not allow reverse flow through a pump.

    3. I did not see any demands at any junctions.  Do you have any idea what the demands are?

    I modified your model to change R-2 through R-5 from reservoirs to tanks.  I set the Maximum water level to the same elevation as the reservoir, and added a 2 m difference between Max and Min levels.  I also set the Base elevation 0.5 m below the minimum water level. 

    I also set the model up to run as an EPS (just for the heck of it).  With no demands, the system responds as you would expect - all of the tanks fill up and then the built-in altitude valve feature of the tanks close as each tank fills, and pump output drops to 0 l/min after all tanks are full.  In the EPS run, all of the tanks are full after about 18 minutes, then all flow in the system stops.

    As Scott mentioned in his reply, the only negative pressure occurs at the pump suction.  It is highly unlikely that you can eliminate this negative pressure condition.  Your alternatives are: 1) build a higher supply reservoir, 2) physically lower the pump, or 3) make the suction line larger to lower head loss.

    Your system as I have it modified runs perfectly fine with results as I would expect to see.

    If you are interested I will zip the files and send them to you.

    Good luck!

Reply
  • Hi Xolani,

    I took the liberty of downloading your model files (apologies if I shouldn't have - in any event I will delete all the files later).

    I have a couple of questions and some comments.

    1. Are you certain that you want to model R-2, R-3, R-4 and R-5 as reservoirs and not "tanks"?  There is a huge difference between how Watercad and the "real world" defines a reservoir.  From the looks of your system I would guess that you want to model them as tanks.  Modeling a storage vessel as a tank has several advantages over modeling them as reservoirs; a reservoir has a fixed hydraulic grade whereas a tank does not (the water level can fluctuate between the minimum and maximum).

    2. Are you positive that all of the control valves and check valves you have in the model are needed?  You typically do not need a check valve in a pump discharge as Watercad will not allow reverse flow through a pump.

    3. I did not see any demands at any junctions.  Do you have any idea what the demands are?

    I modified your model to change R-2 through R-5 from reservoirs to tanks.  I set the Maximum water level to the same elevation as the reservoir, and added a 2 m difference between Max and Min levels.  I also set the Base elevation 0.5 m below the minimum water level. 

    I also set the model up to run as an EPS (just for the heck of it).  With no demands, the system responds as you would expect - all of the tanks fill up and then the built-in altitude valve feature of the tanks close as each tank fills, and pump output drops to 0 l/min after all tanks are full.  In the EPS run, all of the tanks are full after about 18 minutes, then all flow in the system stops.

    As Scott mentioned in his reply, the only negative pressure occurs at the pump suction.  It is highly unlikely that you can eliminate this negative pressure condition.  Your alternatives are: 1) build a higher supply reservoir, 2) physically lower the pump, or 3) make the suction line larger to lower head loss.

    Your system as I have it modified runs perfectly fine with results as I would expect to see.

    If you are interested I will zip the files and send them to you.

    Good luck!

Children
  • Hi Gary

    It's perfectly fine to download the files, no problem at all. On your questions and comments:

    1. No I am not certain that R-2 to R-5 must be modeled as a tank or reservoir. I only chose reservoir as there is no pressure they put on there system. I consider a tank as being under pressure so I don't know if the software works that way as well.

    2. I do not actually need the valves in the model as I only used them to help represent head loss in the pipes. I can do without them and add them as equivalent length. But the PSV i am of the idea that it plays an important wrong in the system pressure upstream and downstream and I am struggling to model the pressure setting. I wanted to compute the system without the PSV at first (if it fills up all the tanks) then use the pressures to set the PSV pressure.

    3. No I do not know what the demands are but if volume flow like litres/day required, I can estimate. Should I add demands? This is a rural area system, once the water is in the reservoir it is distributed by gravity to stand alone taps. Reservoirs can take a couple of days to drain when full.

    I am very interested on the modified model, yes you can Zip it and send, thank you for that, my understanding will grow 

  • Xolani -

    1. Regarding R-2 to R-5:  From you last sentence ("Reservoirs can take a couple of days to drain when full.").  That tells me that, from a WaterCad perspective, you have "tanks" and not "reservoirs".  Modeling them as "tanks" should do what you want them to do.  I know the terminology can be confusing, but that's just how it is.  Regarding "... a tank being under pressure...", these as hydropneumatic tanks - they are indeed under pressure, but you do not have any of these in your model. Should you want to add hydropneumatic tanks in the future, WaterCad can indeed handle these just fine.

    2. Adding valves in a model just to add head loss is not the ideal way to model your system.  Question: How do you know how much head loss you need to add using the valves?  You want the model to mimic the real-world system as closely as possible.  If it were me modeling this system I would start off by entering the actual in-field pipe lengths, sizes and materials (which I believe you have already done).  Then adjust your piping "C" values to match any testing you have done.

    Why do you need the PSV?  I guess I don't understand how you are trying to operate the system.  Again, if it were me, assuming the pump is properly sized, I would eliminate the PSV.  The pump would then ride up and down it's curve according to the system head it sees at any particular point in time.

    3. Regarding demands:  Yes, I would add demands, as that is what this software does best (it is demand-driven, that is, it will always satisfy demands regardless of pressures or flows). I would add junction nodes at the appropriate places in the system (and add pipe as needed), then assign demands to the nodes.  A very useful feature of WaterCad is the ability to modify demands based on usage patterns, but you may want to use patterns only after you get your model up and running properly.

    A benefit of using the EPS option is that you can look at how each system element responds over time; i.e. - how long it takes tanks to empty/fill, how much water your pump is delivering, how your system pressures vary, etc.  I rarely use the static calculation option just so I can see how my systems respond over time.

    Attached is a zip file of the modified system. Your "base scenario" remains as you set it up; I added a couple of alternatives and a new child scenario to show how I modified the system.

    Good luck!

    Corana Pumping Scheme-mod.zip

    Answer Verified By: Xolani Tswina 

  • Gary

    You are correct with the valves, I wanted to mimic the real system. I will remove then now and in future models, I will just use the equivalent length (L/D)  calculation to add extra length on pipe.

    Thank you for the modified model, I will familiarize myself with it and also modeling on EPS 

  • Xolani  -

    One thing I did not do was enter the correct diameter for the tanks (old R-2 thru R-5).  I also gave this some more thought and have these questions:

    1. How are the pumps controlled? On/off by hand, or some type of automatic control?

    2. Since R-2 thru R-5 apparently have no altitude valves, the pipe feeding each individual tank will need to be able to be opened & closed as they reach maximum level.  Do R-2 thru R-5 have any type of inlet control valve?  If not, you will need to provide these.  Without these the tanks at the lower elevations will overflow and the highest one will never fill completely.  The simplest would be some type of valve with a mechanical float that shuts the valve as the tank fills. These are available from many manufacturers (Bermad, Watts, OCV, etc) and require no external power to operate (they are line pressure driven).  I don't see any reason to actually include these in the model as their operation mimics (the closing, at least)  that of the WaterCad built-in altitude valve, but they (or something like them) will be needed in the physical system.

    Good luck.

  • Gary

    Diameter of the tanks is 5m

    1. Pumps have a 3 phase motor connected to a control panel that is controlled on/off by hand.

    2. There is a ball valve on the end of the pipe that shuts off when the tank fills