At the time of posting this I will have sent the file to you guys. I am having trouble modeling this system the way its supposed to be working. I'm not sure it is important for the type of analysis im running though. Currently The distribution system works exactly how I expect it to except i cannot get the booster pump station to work. In order for the booster pump to be modeled in the system, I have the Reservoir setup to have a constant hydraulic grade. In practice, this is not a reservoir and is actually a tank. The booster pumps are set to maintain pressure in the distribution system. This pressure is controlled by a transducer that is located in the tank. This pressure is 82ft above the bottom of the tank or Elev, 344ft.
My question is, is it accurate to model the system as a reservoir with constant head (that constant head being the head the booster pumps are set to maintain). I am only interested in fire flow results (ie pressures, pipe velocities and total flow). In a scenario where a snapshot is taken (like a fire flow) I don't see why this modeled scenario would be considered inaccurate. In the case where a 24 hour period needs to be tested, this is not an accurate representation because pressures will fluctuate and flow.
In a fire flow scenario, I anticipate all 3 wells and the booster pumps will be in operation.
I may need to further clarify things, so if that is the case please let me know.
Current scenario is "Well Pumps On"
T. pump = 300gpmc. pump = 200gpmg. pump = 50gpm
Total source flow = 550gpm
Another question is J-28 is connected to a 6inch AC pipe. I put a TCV on that line to limit the flow through that pipe. The field hydrant test resulted in much lower values than what the model was originally predicting. P-229 is also inactive. I was wondering if there was a better way of "calibrating" this portion of the system?
If running the "SE Hydrant Scenario" and putting a fire flow at H-24-SE the flow says 1096gpm. @ 20psi. Field tests show the value should be 1123gpm. So i would say this is an accurate calibration but I'm not sure if there is a better way of doing this.
My next question is, without setting the maximum allowable fire flow to some obsurd value (like 5000gpm) how do I show what the model predicts the flow is going to be at 20psi?
Thanks in advance!
For further Clarification:
The "b.pumps" is actually 3 pumps in parallel. they are a VSP that are set to maintain pressure in the distribution system.
The well pumps turn on when the water level in the tank drops. However with a constant head reservoir that will never happen. So i have them modeled to flow at their respective values as seen in practice.
Hello Nick,
For the booster pumps, if these are "b pumps" this is set as being inactive in the "Well Pumps On" scenario, meaning that it will be ignored. If it should operate during this scenario, you will want to make this part of the system active. ls this pump actually a variable speed pump? If not, what are you trying to get this pump to do?
For the calibration, the TCV has a headloss coefficient of zero, which may not be correct if you are trying to apply a headloss. For calibration, you may want to explore adjusting the roughness coefficients in the pipes instead of adding valves that may not exist in the system.
For the results in the SE Hydrant scenario, adjusting the roughness values may help with calibration here as well. If you have WaterGEMS, you can also use Darwin Calibrator to help with this.
Finally, to see what the fire flow results would be if the pressure reached 20 psi, I am not seeing a field like this. Is there a reason that this is required?
Regards,
Scott
For the well pumps on scenario i have the booster pump turned off. I have previously turned it on to attempt to get it to work properly. I have muddled around with the pump controls etc. It just wont work. When I set the control node as the "tank" It says tanks with separate inlet outlet cannot be the pumps control node. I have an idea why that is the case, but how this system works hydraulically, it actually gets around the pump directly pumping back into the tank. I'm struggling with determining how best to put that booster pump station in the system. How to exactly model it because of how complex this particular system is.
Im not entirely sure what I would adjust the roughness coefficient to on that pipe. I dropped it down to almost C=25 and I was still above the flow we saw. The only way to deal with that issue was for me to limit the flow through the valve.
For the pressure at 20psi, typically when the fire department respond to a fire they draw whatever water they need. There is a minimum flow required at 20psi residuals. As for the flow at 20psi, I used this to determine the model was calibrated or not. Because hydraulically the system can meet the required flow without a fire truck pump. I wanted to know how close the values we saw in the field were to the values the model predicted.
In order to use a tank as the control node for a VSP, the tank would need two pipes connected to it, one for inflow and one for outflow. The VSP will try to maintain the elevation in the tank by adjusting the relative speed factor. There are two inactive tanks in the model. Were one of these the tanks used as the control node. Though the booster pump is inactive, it currently uses a junction as the control node.
Adjusting the roughness of a single pipe will likely have little impact on the results. Pipes of a similar age and material will like age a the same rate, so you would like see these similar pipes all change in roughness. Darwin Calibrator works on an assumption like this, but using the global edit feature to adjust the roughness values on multiple pipes, which may improve the results.
For the flow at 20 psi, does this mean you let as much water out as possible until you reach a value of 20 psi? I don't know that you would necessarily need to have the pressure drop all the way to 20 psi to have a useful flow test for calibration. No matter the flow and pressure, if you can calibrate the model so the results at this location are similar to the flow test, that should be useful.
If you are using WaterGEMS, you may want to take a look at Example5.wtg (C:\Program Files (x86)\Bentley\WaterGEMS\Samples). There is an example Darwin Calibrator study there which you can use to see example flow tests used in this study.