Difference in peak flow

We would need a copy of your model, to see how the model is set up.

There are two options for sharing your model on Communities, whichever you choose please be sure to zip your files first. The first option is to attach the zip file containing your model to your reply on the forum using the Advanced Reply Editor (you'll find the link below and to the right of the reply box). If your data is confidential please use the instructions in the link below to send it via Bentley Sharefile. Files uploaded to Sharefile can only be viewed by Bentley employees. Please be sure to reply on this thread with the name of the file after it has been uploaded.

http://communities.bentley.com/help/w/be_communities_help/7079.be-communities-secure-file-upload

I've uplaoded my file (2416-10-03-Regulateur_Saint-Louis.stsw.zip), Thank you!

Hello Sushma,
I my model there is 2 pump station and one votex chamber. In all scenarios (Actuel_Qmoy, Emissaire_autoroute, Bassin_retention, TPU_Ferme, Actuel_Qmax, Statuquo) PMP-4 doesn't start (very small watershed) and PMP-3 starts at least 4 times. For example, when I use the scenario Statuquo, and add fixed inflow (10 L/s) at node 123-005 and run the model the flow in conduit ES-21129 is 175L/s. If I delete the fixed inflow at node 123-005 and add 10 L/s at node 124-818 and run the model again the flow is not the same (178L/s) in conduit ES-21129. Does it help to understand my problem?

Thank you

Hello Sushma,
I just want to know if I explained my problem well enough and if you an idea why my model reacts like this?

Thank you and have a nice day

Hello Marie-Joelle,

Sushma was out of the office yesterday (holiday). I went to take a look for you but unfortunately the file has expired and is no longer available for me to download. I apologize for the delay and inconvenience.

Here are some general comments that might help as you wait for Sushma's response. You can also re-upload the model and I will plan to take a look tonight if possible.

1) If you're using the Implicit or Explicit solvers, or if you're using an EPS with the GVF Convex solver, flow will change over time. You refer to a single flow value observed at a conduit - does this refer to a peak flow, or are all the inflows fixed? If there are "sharp" peaks in your hydrograph due to upstream pump cycles, the way in which the flows combine could be sensitive.

2) The differences in flow is relatively small and may be within tolerance of the hydraulic convergence. Meaning, slight differences in convergence between the two different flow entry points may result in small differences in flow downstream. You might consider these within the tolerance of a dynamic hydraulic simulation.

3) If your continuity error is high, the model may be especially sensitive to changes. After computing the model, note the "continuity error" shown in the calculation summary and total trials - compare between the two different flow entry points. If the continuity error is high, you may need to adjust the advanced calculation options. Some guidance on that can be found here:

communities.bentley.com/.../21403.troubleshooting-unstable-sewergems-and-civilstorm-results-using-the-implicit-solver

communities.bentley.com/.../21286.troubleshooting-unstable-sewergems-and-civilstorm-model-results-using-the-explicit-swmm-solver

Thank you I'll have a look at all this! I've uploaded my model : 2416-10-03-Regulateur_Saint-Louis.stsw.zip
Thanks!

Thanks for sending the model and for your patience.

Here is a screenshot of the hydrograph through conduit ES-21129, in SewerGEMS 08.11.05.113. The blue line is for the case where the 10 l/s inflow is added close to the pipe where the flow is observed, and the red line is for the case where the 10 l/s is added far upstream.

Intuitively, you might expect that the red line (inflow far away) would be less than the blue line (inflow close) due to attenuation, or equal to it due to the inflows in question being fixed. In the first couple hours, and at around 7-8 hours, this does happen. (flow is lower in the far-away inflow case) The case that you point out where the opposite occurs (178 vs 175) occurs at the peak, around 5-5.5 hours. However, viewing the two graphs together you can see that despite this, the lines are actually relatively very close. Considering the accuracy of other parts of the model (data input assumptions vs what's actually in the real system), you may consider this to be within acceptable tolerance for a hydraulic model.

Small differences in convergence between the two cases can also cause slightly differences in the results, especially if you have a high continuity error (as reported in the calculation summary) or instability. In this case, we can see that in the case of the inflow being far away, continuity error was 1.5%, and in the case where the inflow was close, it was 1.2%. Both are good, but slightly different. Further, there are some user notifications about potential problems, all of which could potentially cause the model to be more challenging to solve, be more unstable, and thus more prone to differences such as the one you see.

With that said, I took a look at your model using the Explicit solver (you were using the Implicit) and notice fairly similar results, with a very slightly higher peak flow around 5 hours in the case of the inflow entry point far upstream.

Here's what I think may be causing this small difference.

In the case of the "near" flow entry point just upstream of the pipe where flow is observed, that fixed inflow almost immediately enters the pipe in question, then quickly exits the system not far downstream at the outfall. On the other hand when the inflow is added very far upstream, it has to travel through nearly 5 km of pipes before it gets to the same point of observation.

In both cases, at ~5 hours into the simulation, the 10 l/s inflow has reached the point of observation. However in the case where the inflow is added much further upstream, the volume of extra water in the network is much higher, because that 10 l/s is passing through all of the pipes between the entry point and the point of observation. I suspect that the effect that this has on the system may be what is influencing the peak flow at the point of observation. Perhaps the extra upstream volume's (the 10 l/s passing through 5 km of pipe) causes an increase in the flow depth/HGL (you can see this in a manhole on the flow path between the far-upstream inflow entry point and the pipe in question), which then has an impact on the momentum and hydraulics of the flow at the pipe in question.

It's worth noting that, in the case of the inflow entering far upstream, although the peak flow is slightly higher the volume appears to be about the same. Looking at the graph, you'll notice that on the receding limb of the hydrograph (~7-8 hours), the flow is less.

Thanks for sending the model and for your patience.

Here is a screenshot of the hydrograph through conduit ES-21129, in SewerGEMS 08.11.05.113. The blue line is for the case where the 10 l/s inflow is added close to the pipe where the flow is observed, and the red line is for the case where the 10 l/s is added far upstream.

Intuitively, you might expect that the red line (inflow far away) would be less than the blue line (inflow close) due to attenuation, or equal to it due to the inflows in question being fixed. In the first couple hours, and at around 7-8 hours, this does happen. (flow is lower in the far-away inflow case) The case that you point out where the opposite occurs (178 vs 175) occurs at the peak, around 5-5.5 hours. However, viewing the two graphs together you can see that despite this, the lines are actually relatively very close. Considering the accuracy of other parts of the model (data input assumptions vs what's actually in the real system), you may consider this to be within acceptable tolerance for a hydraulic model.

Small differences in convergence between the two cases can also cause slightly differences in the results, especially if you have a high continuity error (as reported in the calculation summary) or instability. In this case, we can see that in the case of the inflow being far away, continuity error was 1.5%, and in the case where the inflow was close, it was 1.2%. Both are good, but slightly different. Further, there are some user notifications about potential problems, all of which could potentially cause the model to be more challenging to solve, be more unstable, and thus more prone to differences such as the one you see.

With that said, I took a look at your model using the Explicit solver (you were using the Implicit) and notice fairly similar results, with a very slightly higher peak flow around 5 hours in the case of the inflow entry point far upstream.

Here's what I think may be causing this small difference.

In the case of the "near" flow entry point just upstream of the pipe where flow is observed, that fixed inflow almost immediately enters the pipe in question, then quickly exits the system not far downstream at the outfall. On the other hand when the inflow is added very far upstream, it has to travel through nearly 5 km of pipes before it gets to the same point of observation.

In both cases, at ~5 hours into the simulation, the 10 l/s inflow has reached the point of observation. However in the case where the inflow is added much further upstream, the volume of extra water in the network is much higher, because that 10 l/s is passing through all of the pipes between the entry point and the point of observation. I suspect that the effect that this has on the system may be what is influencing the peak flow at the point of observation. Perhaps the extra upstream volume's (the 10 l/s passing through 5 km of pipe) causes an increase in the flow depth/HGL (you can see this in a manhole on the flow path between the far-upstream inflow entry point and the pipe in question), which then has an impact on the momentum and hydraulics of the flow at the pipe in question.

It's worth noting that, in the case of the inflow entering far upstream, although the peak flow is slightly higher the volume appears to be about the same. Looking at the graph, you'll notice that on the receding limb of the hydrograph (~7-8 hours), the flow is less.