System Curve for Closed System

[NTaylor]

Hello,

I am in need of any insights for producing a System Curve for a closed system. I have everything setup and when I attempt to run the System Curve on a closed system the model returns a number of errors but it will run in the normal Steady State mode instead of attempting to run a System Curve.

I have read here that emitters are used to help simulate flow through a system to produce a System Curve in the following link -
http://ftp2.bentley.com/dist/collate...urnal-awwa.pdf

When I use emitters and I play with the emitter coefficients according to the demand (average day, peak hour, etc..) I get curves that have a wide range but don't seem to correspond. As I mentioned before, the system is closed. I am running a scenario where I need to generate a system curve for a pump that will be operating to keep the system at a minimum of 40 psi at the highest location and know that I need to have 150.3' of head to keep that minimum of 40psi.

Thanks in advance if anybody has any insights,
Noah

[Patrick Moore]

Noah,

System curves for closed systems are kind of challenging to model. Traditional tools to calculate a system curve (like what the Innovyze software does) to create a system curve is to convert the pump station to a series of two junctions. One on the suction side (modeled as a positive demand), and one on the discharge side (modeled as a an equal and opposite negative demand (a source)). The normal process is to vary the flow through the pump station by running a series of model runs where the demands on the two nodes are varied form zero up to a maximum value. So if the suction node has a 1000 gpm demand the discharge node will have a -1000 gpm negative demand (and all the pumps are off). The head difference between the suction and discharge node represent the head gain the pump would need to push that amount of water into the system.

The problem first arises when you run the simulation at zero supply. As expected, because this is a closed zone with supply only from the pump station, the model will typically issue a disconnected node warning and the simulation will fail. As an added difficulty for this type of scenario, the pump station supply needs to exactly equal the system demand in order for the software to properly find a solution and this makes it difficult to cycle through varying amounts of supply when the demand and the supply have to match exactly.

This is in essence what I believe in the article you posted why they are using emitters as they adjust the flow to match the system head conditions but wont use more or less than what is supplied.

However, one other method we often recommend is to place a small tank at the high point in your system connected to the system by a very small pipe. Set the tank elevation at the HGL desired minus 10 ft, make the tank 20 ft tall and have an initial level of 10 ft. The Tank diameter can be small like 10 ft. Then create a pattern that will vary the system flow form zero to your desired maximum flow using multiples of the current base demand in the model for the closed zone. Thus if the sum of all base demands in the zone are 100 GPM to have a 50 gpm flow the multiplier would be 0.5, and for 200 gpm the multiplier is 2. Set all of the pumps at the facility to initially off and have all controls disabled. Set the positive and negative demands set to the current demand for the closed zone (in this example it would be 100 gpm- suction and -100 gpm on the discharge).


Here is an example of a pattern used previously (click for larger image if necessary)
closed zone system curve pattern.jpg

Run and EPS run of length equal to the number of pattern values to get the full range of results. Backcheck each hour to make sure the flow in or out of the tank is essentially zero. Once verified the difference between the head of the discharge and suction nodes is the pump Head required at the specified flow. This is your system curve in head vs flow curve you can overlay with your specific pump curves.

One last thing, make sure you check the rest of the system so that it will account for the high and low static lift for the pump station as well. Tank levels on the suction side of the pumps will impact the static lift the pump faces and may require more than 1 run to get the maximum static lift and the minimum static lift the pump will be expected to face.

Please give us a call on the main Support line at 626 568 6869 or email us at support@Innovyze.com if you need any further assistance.

Patrick Moore

[Vasevida]

Hi Patrick - is this the only way to model closed pumped pressure systems? I am trying to calibrate our municipal model and run an extended analysis and am having no luck. I have tried using regular controls and VSP controls and still the model does not balance. I have a couple neighborhoods like this in my model and would help any tips in getting the model to balance and run. I am asking if you suggest I develop system curves as per the above methodology or are there any other ways to model distribution systems pressurized by pumps?

[Patrick Moore]

Vasevida (James Clark?),

There are several ways to model pumps that supply a closed zone. The best way to model any system depends generally on how the actual system is operated as usually the best method is the one that most closely resembles actual system operation.

Understanding what the problem is in your model is difficult to say without a bit more information. Set the run report to FULL and make sure it has the checkbox for warning messages checked in the report options. That report usually helps the modeler hone in on where the problems are. Worst case you can always use a negative demand on the discharge and a positive demand on the suction of the pump station similar to what is done in a system curve to see what the head difference between the discharge and suction nodes is to compare against what is used in the pump definition. That is a little trick you can use to get results in a pinch to help evaluate if your pump definitions need adjustment.

If you are having difficulty with the model converging there are several tips that may help posted in this thread that may be of assistance: http://forums.innovyze.com/threads/1...lance-Question

What may be the easiest is to have you send an email to support@innovyze.com and one of our support team can help you troubleshoot your questions in a quick WebEx session.

Hi Patrick - is this the only way to model closed pumped pressure systems? I am trying to calibrate our municipal model and run an extended analysis and am having no luck. I have tried using regular controls and VSP controls and still the model does not balance. I have a couple neighborhoods like this in my model and would help any tips in getting the model to balance and run. I am asking if you suggest I develop system curves as per the above methodology or are there any other ways to model distribution systems pressurized by pumps?

[Vasevida]

I have the reporting on full and I cannot get the model to balance in steady state, so no results. I will give it a bot more of an effort to run in steady state before I go to innovyze support with my (I am sure ) many extended scenario questions. I have a system that has considerable topography and some very old components run manually. I am getting a VSP error of 413, and cannot find what that means however. The system will not balance and goes through PRVs and Pumps switching them on and off and does not ever balance and converge. I have tried increasing the trials, and decreasing the accuracy of the model, tried any number of manual pump controls, and am now trying putting in reservoirs at the pump locations in hopes of determining the problem area(s). Any suggestions would be helpful. Thanks Jamie

[Patrick Moore]

Jamie,

EDIT: Error 413 means the pump cannot meat the discharge pressure setting even when the pump is at its specified maximum speed. Generally this means either the suction head is getting depressed or that the discharge pressure is to high for that pump.

One thing that can occur is if the setting on the VSP control would cause the pump shutoff head at that speed to fall below the static lift, the model could crash.

Try putting in a positive and negative demand on a steady state run with the pump off like show here:

(click for larger image if necessary)
System Curve Analysis.jpg


If the discharge is a closed zone make sure the positive and negative demands equal to expected downstream demand.

What this will do is allow you to see what discharge head would be required by the pump station to run at that flow. By subtracting the discharge node head form the suction node head that is the required head of the pump at that flow.

If that discharge pressure is below your VSP control it is likely that the model could not satisfy system demands at that setting. Further, as noted above, if the required TDH to get that specific flow would cause the pump curve shutoff at that reduced speed to fall below the system curve head at zero flow, the pump would not have the ability to meet the conditions specified. You could also try manually running the pump at different speeds and see what the discharge pressure is. At some point you may find that either the discharge setting may not have been ideal for that pump or that the pump is not ideal for that operation.

Can you run a system curve on that pump? Overlaying the pump curve and the system curve is the best way to evaluate the pump selection for that operation. If you have a suite tool with the pump station analyst, this is very easy to do and you can evaluate changes in speed easily.
example PSA (click for larger images if necessary)
PSA results-Individual.jpg
PSA results-Individual-VSP.jpg

I have the reporting on full and I cannot get the model to balance in steady state, so no results. I will give it a bot more of an effort to run in steady state before I go to innovyze support with my (I am sure ) many extended scenario questions. I have a system that has considerable topography and some very old components run manually. I am getting a VSP error of 413, and cannot find what that means however. The system will not balance and goes through PRVs and Pumps switching them on and off and does not ever balance and converge. I have tried increasing the trials, and decreasing the accuracy of the model, tried any number of manual pump controls, and am now trying putting in reservoirs at the pump locations in hopes of determining the problem area(s). Any suggestions would be helpful. Thanks Jamie

[Patrick Moore]

We have also created a system curve presentation that may help users better understand how they are created and hydraulic issues that should be considered when making system curves so as to have the greatest confidence in the results developed by the model. Closed zone system curves are a special case and this file includes instructions on how to create these type of system curves using the model software:

Link for System Curve Presentation: http://innovyze.com/files/2016/InfoW...asics_v1.3.pdf
Cover Photo(click if need larger image)
system_curve_presentation_cover.jpg
Please feel free to contact us at support@innovyze.com if you would like a copy of an example excel file used to create the demand pattern for system curve analysis and how to calculate the system curve from model output data as well. Just request this file and we can send it along as an example file to help guide you in how to get the data needed to calculate the closed zone system curve.

Patrick Moore