Pump pressure

Hi Ashraful,

Are you saying that you are running the transient simulation without any pump shutting down, or are you saying that one pump is shutting down while the other one's time delay is set beyond the end of the transient simulation duration?

When using the "shut after time delay" transient pump type in HAMMER, a constant applied torque is assumed until the time specified as the delay until shut down, at which point the torque instantly drops to zero. Since the torque is constant before the pump shuts down, the speed can vary if the conditions warrant. So, a "shut after time delay" pump with a shutdown delay beyond the end of the transient simulation duration will not really be operating at a constant speed during the transient simulation. As conditions change in the system (from the other pump shutting down for example, or from tanks filling/draining, etc), the pump speed will change, hence the results will vary. This is also explained in the following article from our Wiki: Pump speed not constant before emergency pump shutdown transient event

So in short, if you want to model a pump with a constant speed during the transient simulation, the "shut after time delay" will not do that and instead you should try using the "constant speed" transient pump type. There are two "constant speed" transient pump types, labeled as follows in the latest version of HAMMER:

1. "Constant Speed - Four Quadrant Characteristics" - uses the 4-quadrant curve specified in the pump definition so that the pump could potentially operate in any of the four quadrants (while still applied a constant speed)
2. "Constant Speed - Pump Definition" - strictly uses the standard pump definition which limits operation to the first quadrant only, but ensures that the pump curves will be followed as it is entered in the pump definition (as opposed to the four-quadrant curve which is based on the initial flow/head along with the predefined shape of the 4-quadrant curves based on the selected Specific Speed)

See more in this wiki article: Differences between the Constant Speed transient pump types

If you are using the "shut after time delay" transient pump type with the duration set beyond the simulation duration, with no other transient event configured in the model, the changes you see in the model may be within the tolerance of "numerical noise", or you might have a "spurious surge" caused by a discrepancy between the initial conditions and the first transient timestep. There are a number of reasons for this, as described in the below article. First though I would try the constant speed pump type, as that would be more appropriate anyways, for a case where you want to assume a constant speed during the whole transient simulation.

Troubleshooting an unexpected Initial Surge during a transient simulation

If this does not help, we'll need to take a look at your model: Sharing Hydraulic Model Files on the OpenFlows Forum

Hi Jesse

Thanks for your quick and detailed reply. I am running transient simulation without creating a transient event. As things are not changing, I expect to see a steady or nearly steady flow variables. I have tried two options - constant speed with pump curve and shutdown after time delay (delay time longer than run duration). One pump running (VSP with a target flow). There are d/s valves, but I kept those "inactive". For both cases, pressure and flow time histories at the pump node are attached here. For constant speed scenario, there is a significant change in flow and pressure. Variables also change in the other scenario but change is less than the constant speed scenario. Something is wrong here.

Ashraful, it looks like an "initial surge" situation - see the article link in my previous reply. If this does not help, please provide a copy of the model file, then reply here with the file name.

Also, what is the purpose of running the transient simulation without any transient event occurring? How would the model results help you in this case?

Jesse

In some scenarios, I will model one pump shutdown while others are in operation. So I wanted to make sure that flow conditions at the pumps are steady before triggering a transient event. If the results change so much without a transient event, how can we justify the results? With the constant speed option, pump flow starts with about 30,000 gpm and changes to some 40,000 gpm without having any external disruption. I have added the model file (pump_startup.zip)

Pump-startup.zip

Hi Ashrafal,

Thanks for providing a copy of the model. I see the same results you see, when using the "shut after time delay" and "constant speed - pump definition" for the transient pump type. However, if you choose the "constant speed - four quadrant characteristics" option, the simulation is quite steady. In profile view, there is no noticeable difference (see screenshot below) perhaps aside from that which results from the tank level changing. The flow as observed from the same point as your screenshots varies by only about 2%.

The "noise" here when attempting to look at graphs with pumps operating at a constant speed can also be sensitive to the Specific Speed (when using the "constant speed - four quadrant characteristics" or "shut after time delay"). For example if you use the equation in the help and in this article to estimate the specific speed, for me it came out to 2169 (US), and the closest choice from the available default specific speeds is 2155 (US). When choosing that, the flow varies by less than 1% compared to the 2% with the 1935 (US) specific speed that you had selected. I realize the specific speed you selected might be best for you based on your judgment and other factors, but wanted to illustrate the sensitivity to the specific speed.

Furthermore, this noise can also be sensitive to the accuracy of the initial conditions. The initial conditions solver will converge on a balanced solution based on the "accuracy" and maximum trials calculation options. In some cases there can be slight inaccuracies in the results, even though they may be within the specific tolerance for flow convergence. For example if you change the Accuracy (steady state/EPS calculation options) from the default of 0.001 to 0.0001, the transient simulation results flow graph will vary by an even smaller percentage when using the "constant speed - four quadrant characteristics" transient pump type.

Lastly, since water level in tanks will change (albeit by a small amount) within the 300 second simulation duration, as an illustration of how you can get the constant speed pump transient results to be even more steady, if I morph both the upstream and downstream boundary tanks into reservoirs (fixed HGL) with elevation set to the tanks' original Initial Setting, the transient results flow graph varies by less than less than 0.01%.

Hi Ashrafal,

Thanks for providing a copy of the model. I see the same results you see, when using the "shut after time delay" and "constant speed - pump definition" for the transient pump type. However, if you choose the "constant speed - four quadrant characteristics" option, the simulation is quite steady. In profile view, there is no noticeable difference (see screenshot below) perhaps aside from that which results from the tank level changing. The flow as observed from the same point as your screenshots varies by only about 2%.

The "noise" here when attempting to look at graphs with pumps operating at a constant speed can also be sensitive to the Specific Speed (when using the "constant speed - four quadrant characteristics" or "shut after time delay"). For example if you use the equation in the help and in this article to estimate the specific speed, for me it came out to 2169 (US), and the closest choice from the available default specific speeds is 2155 (US). When choosing that, the flow varies by less than 1% compared to the 2% with the 1935 (US) specific speed that you had selected. I realize the specific speed you selected might be best for you based on your judgment and other factors, but wanted to illustrate the sensitivity to the specific speed.

Furthermore, this noise can also be sensitive to the accuracy of the initial conditions. The initial conditions solver will converge on a balanced solution based on the "accuracy" and maximum trials calculation options. In some cases there can be slight inaccuracies in the results, even though they may be within the specific tolerance for flow convergence. For example if you change the Accuracy (steady state/EPS calculation options) from the default of 0.001 to 0.0001, the transient simulation results flow graph will vary by an even smaller percentage when using the "constant speed - four quadrant characteristics" transient pump type.

Lastly, since water level in tanks will change (albeit by a small amount) within the 300 second simulation duration, as an illustration of how you can get the constant speed pump transient results to be even more steady, if I morph both the upstream and downstream boundary tanks into reservoirs (fixed HGL) with elevation set to the tanks' original Initial Setting, the transient results flow graph varies by less than less than 0.01%.