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

Issue with pump shut down-Bentley Hammer

Hi ,

I am performing the pump shut down at a pumping station. The issue that I have is the transient pressure fluctuations do not reflect what is expected. There is no high pressure/ pressure spike in the result. I have sent you the model package. Would you please take a look and advise accordingly.

  • Hello Vali,

    Which scenario are you looking at? Is it "LIWA EAST WEST_OP-2 (peak)"? Why do you expect a high pressure spike from this transient event? (has one been observed in the real system if it is built already?)

    This is an extremely large HAMMER model which will likely take several hours to run with the selected timestep and duration. I would recommend skeletonizing the model and/or focusing on the area of interest that is likely to be impacted by the transient event being modeled. See:

    Preparing an existing model for a transient analysis in HAMMER

    With such a large model, you may also want to consider a more reasonable calculation timestep such as 0.01 seconds and some adjustments to the reporting options as mentioned here: Improving performance of transient simulations (recommended reporting calculation options)

    Without running the model yet (I would need to do this overnight), I can offer some potential insight into why you are not seeing a "high pressure" spike. Your transient event appears to be configured as an emergency pump shutdown of four pumps in parallel, which appear to feed a large portion of the model consisting mostly of demands and one single tank. I extended your existing profile to go from the pump station in question (left side in screenshot below) to the single downstream tank (right side). Most of this section of piping has an initial pressure over 150 m which is very high, and even over 70 m at the highest elevation point before the tank on the right side below. 

    Initial conditions hydraulic grade and physical elevation profile between pumps and tank:

    When the pumps on the left side shut down, a "downsurge" (low pressure) wave will travel toward the right side. The magnitude of the downsurge will depend on a few factors such as the velocity, wave speed and pump inertia (set in the Transient tab of the pump definition, see this). Since the initial pressure is already very high, most likely the pressure from this downsurge will not reach zero or the vapor pressure limit and thus no vapor will form. Vapor pockets from low pressure transients often cause the most severe upsurge (high pressure) transients when they collapse, but if the vapor pocket does not occur or if it does not collapse before the end of the simulation, you will not see a high pressure spike from it. If you are observing vapor pocket formation but not collapse within the transient duration, you may need to simulate the pumps turning back on (see guidance here) Once the pumps turn back on and pressure returns, the vapor pockets will collapse and you may see a severe upsurge transient from that.

    Once the downsurge wave reaches the tank location (and other endpoints from other downstream paths), it will reflect back a certain way and that reflection can sometime return as a positive (high pressure) surge but the system may be dampening out those effects in this case. You will need to animate the profile in the Transient Results Viewer (I recommend making a longer version than the one you had).

    Here are a few other observations:


    Regards,

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

  • Thanks Jesse for the prompt reply,
     
    The scenario that I working on is the one that you mentioned ( "LIWA EAST WEST_OP-2 (peak)").
     I have adjusted some setting as you suggested, they are helpful, so thanks for notifying me for them.
    The model is large and this is what was provided to us by another consultant, so we may use the skeletonized later. I provided the packed zip including the results so that you do not have to resimulate this . Just a quick background: the model includes 8 pumps which 4 of them supply the west side and the other 4 supply the east side.
     
    The main issues that I have:
     
    I simulated all pumps shut down after time delay. It seems so the pump rundown is well working see the snapshot below. At the discharge side we expect to have the static pressure (or very close to that one with small fluctuations) once the transient pressure is damped (say a few minutes after pump shut down). However, this is not the case here. The snapshot below shows the transient pressures at J23709 (the discharge point for the west side), as you see the transient pressure intends to be stabilized around 40 m while the final pressure should be close to 7 m (157 (water level at reservoir-at suction side)-150 (the node elevation at discharge side)). This is not due to instability in results or short time simulation. I got pretty the same results with time step 0.005 and  the duration of 500 s. This my main concern  as we know that the transient pressures should be reached to static pressure after pressure dampening for the scenario of pump shut down. Do you have a reason for that?

     
     

     
     
     
    2- As you see from the snapshot above, we do not have too much fluctuations at the discharge point. I understand that the pump inertia may be high or we have too much pressure as the initial conditions, but we do not have the pressure spike even the same as steady state pressure?
     
    Would you please pass on your comments?
     
    Many thanks,
    Vali Ghorbanian, Ph.D, P.Eng
    Senior Hydraulic Engineer
    PARSONS
    Office:9059173296
    Cel:6478877154
     
  • The snapshot below shows the transient pressures at J23709 (the discharge point for the west side), as you see the transient pressure intends to be stabilized around 40 m while the final pressure should be close to 7 m (157 (water level at reservoir-at suction side)-150 (the node elevation at discharge side)). This is not due to instability in results or short time simulation. I got pretty the same results with time step 0.005 and  the duration of 500 s. This my main concern  as we know that the transient pressures should be reached to static pressure after pressure dampening for the scenario of pump shut down. Do you have a reason for that?

    It sounds like you might be thinking about the differential head and not the pressure. After the pumps shut down, the check valve in the pumps closes since the downstream tank is at a higher elevation than the upstream reservoir.

    If you graph hydraulic grade instead of pressure you will see that it settles at around 200 m (physical elevation of the junction of 150 m plus pressure head of 50 m), which is roughly equal to the hydraulic grade at the downstream tanks (the boundary conditions connected to the junction after the pumps shut down and the check valve closes.)

    2- As you see from the snapshot above, we do not have too much fluctuations at the discharge point. I understand that the pump inertia may be high or we have too much pressure as the initial conditions, but we do not have the pressure spike even the same as steady state pressure?

    This screenshot shows the trend of pump impeller rotational speed. When the check valve closes, the flow and speed will be zero.

    On the topic of pressure returning to steady state pressure - this will only happen if the pumps turn back on, or if an upsurge occurs from vapor pocket collapse or wave reflection.


    Regards,

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

  • Thanks Jesse for the prompt reply,

     

    The scenario that I working on is the one that you mentioned ( "LIWA EAST WEST_OP-2 (peak)").

     I have adjusted some setting as you suggested, they are helpful, so thanks for notifying me for them.

    The model is large and this is what was provided to us by another consultant, so we may use the skeletonized later. I provided the packed zip including the results so that you do not have to resimulate this . Just a quick background: the model includes 8 pumps which 4 of them supply the west side and the other 4 supply the east side.

     

    The main issues that I have:

     

    • I simulated all pumps shut down after time delay. It seems so the pump rundown is well working see the snapshot below. At the discharge side we expect to have the static pressure (or very close to that one with small fluctuations) once the transient pressure is damped (say a few minutes after pump shut down). However, this is not the case here. The snapshot below shows the transient pressures at J23709 (the discharge point for the west side), as you see the transient pressure intends to be stabilized around 40 m while the final pressure should be close to 7 m (157 (water level at reservoir-at suction side)-150 (the node elevation at discharge side)). This is not due to instability in results or short time simulation. I got pretty the same results with time step 0.005 and  the duration of 500 s. This my main concern  as we know that the transient pressures should be reached to static pressure after pressure dampening for the scenario of pump shut down. Do you have a reason for that?

     

     

     

     

     

     

     

    2- As you see from the snapshot above, we do not have too much fluctuations at the discharge point. I understand that the pump inertia may be high or we have too much pressure as the initial conditions, but we do not have the pressure spike even the same as steady state pressure?

     

    Would you please pass on your comments?

  • Hello Vali, please see the reply I just posted a few minutes ago. Here it is again:

    The snapshot below shows the transient pressures at J23709 (the discharge point for the west side), as you see the transient pressure intends to be stabilized around 40 m while the final pressure should be close to 7 m (157 (water level at reservoir-at suction side)-150 (the node elevation at discharge side)). This is not due to instability in results or short time simulation. I got pretty the same results with time step 0.005 and  the duration of 500 s. This my main concern  as we know that the transient pressures should be reached to static pressure after pressure dampening for the scenario of pump shut down. Do you have a reason for that?

    It sounds like you might be thinking about the differential head and not the pressure. After the pumps shut down, the check valve in the pumps closes since the downstream tank is at a higher elevation than the upstream reservoir.

    If you graph hydraulic grade instead of pressure you will see that it settles at around 200 m (physical elevation of the junction of 150 m plus pressure head of 50 m), which is roughly equal to the hydraulic grade at the downstream tanks (the boundary conditions connected to the junction after the pumps shut down and the check valve closes.)

    2- As you see from the snapshot above, we do not have too much fluctuations at the discharge point. I understand that the pump inertia may be high or we have too much pressure as the initial conditions, but we do not have the pressure spike even the same as steady state pressure?

    This screenshot shows the trend of pump impeller rotational speed. When the check valve closes, the flow and speed will be zero.

    On the topic of pressure returning to steady state pressure - this will only happen if the pumps turn back on, or if an upsurge occurs from vapor pocket collapse or wave reflection.


    Regards,

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