FCV downstream of source (water treatment plant) error

[Sara]

I am trying to model the flow coming out of a water treatment plant using fixed head reservoir and a downstream flow control valve (FCV). I set a flow pattern to the FCV to match observed flows.

FCV_pattern.PNG
However, I see reverse flow- from FCV to the reservoir rather than the opposite. I tried increasing the reservoir head but nothing seems to work.
WTP_FCV_pumps_PRV.PNG
I have a PRV to control discharge pressures. Error report is as below:

"0:00:00: Unbalanced after 41 trials (flow change = 0.679477)
Total Supplied: 335414.57 gpm
Total Demanded: 446142.08 gpm
Total Stored: 743059.05 gpm
0:00:00: Reservoir R1 is closed
0:00:00: Reservoir R2 is closed
0:00:00: Reservoir R3 is closed"

Thanks for the help.

[Patrick Moore]

Sara,

A FCV can allow reverse flow if the downstream head exceeded the upstream head. In the EPANET engine used by InfoWater (SA and Pro too) and H20NET there is no check valve feature on a FCV. However you did have an incomplete solution when it says "unbalanced after X trials" such that the results you have may not even be completely reliable either.

I would check the following:

1) Check the FCV direction: Are the pipes connected to the FCV drawn in the desired direction of flow? The pipe direction if both pipes is the direction the model uses for the valve flow.

2) Report Options - Make sure you are using the Full Hydraulic Status run report with Warning messages generated. This is the most helpful when troubleshooting run issues.

(Click if need larger image)
troubleshoot image1.jpg

3) Simulation Options for convergence: These settings can't guarantee everything will converge, but these settings often resolve many convergence issues and may be worth investigating

General Tab Settings
Trials: 200 (This is the maximum trials to reach convergence)
Accuracy: 0.001
Unbalanced: Continue
Extended Run: 50

Advanced Tab Settings
Status Check Frequency: 2
Max Status Check Iteration: 10
Relaxation Factor: 0.6
Damping Limit: 0.01 (should be set to 10 X the current accuracy used)

Notes:
• 200 trials – – If a model will not converge in 200 trials there is generally an issue (controls,
setpoints, or model setup issue) that needs to be resolved
• Unbalanced – Continue – By continuing you can get results to help troubleshoot, but need to watch out for warning messages
• Extended Run – set to 50 this allows additional runs with certain conditions to find a solution
• Advanced parameters

• Relaxation Factor : 0.6 This is the most helpful if models are having difficulty converging. (1
solves faster but can have more difficulties with instability at times)
• Damping Limit – Set to 10X the accuracy. If accuracy is 0.001 then damping Limit is 0.01.

Screen Captures:
General Tab Simulation Options
(Click if need larger image)
Sim Options 1.jpg


Advanced Tab Simulation Options (with notes)
(Click if need larger image)
Sim Options 2.jpg

4) Rerun the model: Using the report and sim options can you rerun and verify what is occuring?

Patrick Moore

[Sara]

Thanks Patrick. The model runs now with the suggested changes. Now I see that the are times where the required flow (se flow at FCV) is not always met- I tried with higher reservoir heads assuming the head is not sufficient but didn't work. Could this be due to the flow I am trying to push through the FCV is higher than the demands? There are multiple water treatment plants in the system which could be taking over part of the service area.
Capture.PNG

[Patrick Moore]

Sara,

Using a FCV in the layout you has in kind of artificially constraining the supply. You generally do not want to do this unless the system is known to have something controlling flow in this way.

In addition, from what I recall, the GLWA system (I actually modeled that system for a few years early in my career so am somewhat familiar with it) is essentially a large "pumped only" system without floating storage in it. While there are some ground storage tanks that can be filled and then pumped out, these are essentially isolated from the pipes supplying demand and make where the demands are occurring to be in areas supplied only by direct pump discharge. This means that supply and demand need to be perfectly balanced at all times and would make adding a FCV mathematically problematic under certain conditions in this model.

A FCV essentially works like this:

1) it determines what downstream head of the valve is necessary to push the flow setting flow into the network. In essence this is exactly what a negative demand also does). We will call this required Head as "Head A". The difference form a negative demand is that a FCV then checks the "Head A" value against the available head upstream of the valve, while a negative demand will not. If the head upstream of the valve is "Head B", then if Head A > Head B, there is not sufficient head available to push the flow as requested. The valve will issue a warning and set the valve status to OPEN and allow the available head (Head B) to supply however much water it can , but the flow will be under the flow setting. If the Head B > Head A, the valve will induce headloss so that the head downstream is exactly "Head A" and the flow through the valve is exactly the setting requested.

So because there is no tank for extra supply to fill in your network, if the demand is less than the FCV setting and this station is the only supply for the area, then there is no way to supply more water than can get consumed and you would likely get an "Open but cannot deliver Flow" error on the FCV because you are trying to send more water than it can use.

FCV in a model are kind of very constraining to the results and in a system like yours are kind of superfluous unless you have either multiple points of supply to a region and the demand is exceeding the requested FCV setting. Even then, in your network you are trying to exactly match both values which is inherently challenging. I suspect perhaps you are trying to fix the supply to match calibration results perhaps, but the flow from the station will basically equal the downstream demand plus any pump station flow out of the direct service area. As such, using a FCV will generally cause more difficulties that assist you. In addition if you want to restrict station flow it is often better to put the FCV after then pumps rather than before the pumps as well.

But as noted, I would use caution in trying to hyper constrain a model by adding FCVS especially in a system like for the GLWA system where there are no tanks to allow flexibility of either taking excess water or adding water when supply and demand are not balanced.

Recall your demands are likely "averaged" demands and may slightly different from demands on a given day and the demand patterns may also vary. This means pump station flows will generally be directly correlated to both the system demands and downstream pump station flows. Adding anything that would try to constrain supply in a system like this is likely only to cause model challenges and errors in most cases and should be used with caution. Similarly adding anything that is not actually existing in the real system, should also be added with much caution as you may be trying to mask other model issues by doing so. That is generally something most modelers want to generally avoid as it can add more uncertainty to a model as well as make it have a harder time solving. So please use caution when doing so.

The flow from the station will need to satisfy demands and downstream pump station flows but when there are multiple points of supply would also somewhat depend on the discharge head of the station. If you had 2 stations feeding an area, the flow from each would somewhat depend on the discharge head of each as the higher discharge head will often supply more than one discharging at a lower discharge head. Essentially the system balances out the supply from each station so that the headloss to each demand junction is the same, so if a flow is off it can either mean i have a pump curve or supply head issue, or I am seeing different headloss on the paths from the stations to the demand. But when it is one supply with no tank, then the flow is to meet the demand and or any downstream pumping out of the zone. In those cases, constraining the flow out is not reasonable and should be avoided as if the flows are off it is because the demands or pump flows out of the zone are off.

Anyway, I hope this better helps explain how FCV's work and why you want to use caution when adding them unless there is something restricting the discharge flow that the FCV is meant to represent. If flows are off, then look at what else is causing it and fix the problem where it is actually occuring is generally preferred.


Please feel free to respond if you need further assistance.

One last note: You can also reach us at support@innovyze.com if you have a support question as well. We can often discuss questions via email, phone or via a WebEx session in more detail that we can here in the User Forum, so feel free to keep that in mind in case a call would help clear up any remaining questions.

Patrick Moore

Thanks Patrick. The model runs now with the suggested changes. Now I see that the are times where the required flow (se flow at FCV) is not always met- I tried with higher reservoir heads assuming the head is not sufficient but didn't work. Could this be due to the flow I am trying to push through the FCV is higher than the demands? There are multiple water treatment plants in the system which could be taking over part of the service area.
Capture.PNG