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

Energy Grade Line Drops Going Upstream

Hi all,

I'm working with SewerGEMS for the first time ever and am having an issue with the EGL dropping upstream (or jumping downstream) at a manhole structure. Can anyone explain this? For reference there is a large basin discharging roughly 80 cfs into this same manhole at this period in time. I would appreciate any help.

Thank you,

Kevin

Parents
  • Hello Kevin,

    Can you clarify where in the screenshot you are referring to? Which numerical solver are you using? (Implicit, Explicit, GVF-Rational or GVF-Convex)

    If the element between stations ~250 ft and ~750 ft is a pond, the velocity in a pond is zero hence the velocity head is zero.

    Energy grade discontinuity can be seen with the GVF-Rational solver as mentioned in this article but I do not think this is what you're referring to.

    If you're referring to the element at around station 1300 ft, check the velocity result fields in that element and the adjacent conduits.

    If this does not help, please send a copy of the model for review: Sharing Hydraulic Model Files on the OpenFlows Forum


    Regards,

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

  • Hi Jesse,

    Thanks for the quick response. I meant to include more detail in my post but had to rush out of the office. My model is using the Implicit numerical solver, it was the default when I created a new model so I left it. I was indeed referring to the element around station 1300 ft. I compared velocity in and velocity out at the peak time step and it shows 19.5 ft/s in and 22.3 ft/s out. I'm not convinced my model is functioning the way it should be so I will send in a copy for review.

    Regards,

    Kevin

  • Hi Kevin, I had sent you a reply to your private message earlier - I am working on a response to your concerns but it would be best to post here in the forum thread. If possible can you reply here with the questions/concerns that you had sent me via PM?


    Regards,

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

  • Sure thing Jesse.

    I just uploaded my SewerGEMS files via the Bentley Secure upload portal. I wanted to provide some context for the project. Our municipal client is currently approving plans for and soon to begin construction on a large 84" RCP culvert underneath an embankment (STA 100). This culvert will discharge into a downstream ravine (shown here as a pond from STA ~250 to STA ~750) that will outflow to a nearby river (STA ~1500). There is roughly 750 LF of existing stone box culvert (3'Wx4'H) running from the ravine out to the river (with one manhole, DMH-7 in the model, along that run at STA ~1300) that will need to be replaced with larger pipes. These replacement pipes are P-1 and P-2 in the model. The SewerGEMS model I created has several physical alternatives with varying pipe sizes for the length of pipe running to the outfall but initial results indicate the 72" RCP alternative might be the best option.

    I am currently assessing the model without an outlet structure in the ravine as a baseline model. Once I am satisfied that is running properly I will test out different outlet structure configurations. In its final form, the ravine will likely feature check dams and water quality features along its length to treat low flow storm events. The ravine is in rough shape, so the tradeoff we are analyzing is the cost of reinforcing/armoring the ravine (depending on how high the water stages up) vs. the cost of upsizing pipes downstream.  

    The ravine is modeled from its existing contours (from online GIS data) taken from AutoCAD using the Stage Storage command. I noticed that the results of the model are wildly different when I use the Elevation-Area definition vs. the Elevation-Volume definition. Is there a reason for this?

    The network upstream of this model was modeled in HydroCAD. I copied the output hydrographs from HydroCAD and used them as wet inflows into the most upstream structure (STA 0, DMH-2 in the model) for the respective storms. Was this the correct way to model the upstream system without re-creating it in SewerGEMS?

    Furthermore, you will notice in the model that the surrounding basins discharge directly into the ravine (basins E-1, E-2, and E-3) and into DMH-7 (basin E-4). The client so far has not provided me with any data regarding the surrounding drainage infrastructure so I can't model the lateral pipe systems or catch basins. Instead of many smaller basins we are left with several much larger ones. I have modeled each basin with its actual area and SCS CN and estimated the Tc. Will this provide a more or less conservative result as compared to if I could properly model the surrounding area's drainage pipes and catch basins?

    I understand this is a lot of information but the project has been anything but straight forward. I am also picking it up from someone else's initial designs that have changed many times over the years. I was unhappy with HydroCAD's UI and the way the results were displayed (I love the profiles in SewerGEMS). It made it extremely difficult to relay the results of the model to my boss, who is more of a potable water design guy than a stormwater design guy. As mentioned above I am new to SewerGEMS and have almost certainly made some mistakes in my model. If you see anything out of the ordinary please let me know. 

    Thanks again,

    Kevin

  • The ravine is modeled from its existing contours (from online GIS data) taken from AutoCAD using the Stage Storage command. I noticed that the results of the model are wildly different when I use the Elevation-Area definition vs. the Elevation-Volume definition. Is there a reason for this?

    If the curves are set up to be equivalent then the storage should be the same. Can you explain more about how you are setting up the elevation-area curve versus the elevation-volume curve? Using a feature available in PondPack (where you can generate the computed elevation-volume curve used by the solver based on the entered elevation-area curve), I computed the equivalent elevation-volume curve based on the elevation-area curve seen in your model, and it produces similar results. (I can send it to you via private message if needed) However I see if I switch to elevation-volume in the model you sent, there is some information that you had entered previously, and although it is not exactly the same as the curve I calculated, it seems to produce fairly similar results when the model is computed (compared to the elevation-area curve you had entered). Can you describe in more detail the "wildly different" results you see?

    The network upstream of this model was modeled in HydroCAD. I copied the output hydrographs from HydroCAD and used them as wet inflows into the most upstream structure (STA 0, DMH-2 in the model) for the respective storms. Was this the correct way to model the upstream system without re-creating it in SewerGEMS?

    Yes, that sounds like a reasonable approach.

    I have modeled each basin with its actual area and SCS CN and estimated the Tc. Will this provide a more or less conservative result as compared to if I could properly model the surrounding area's drainage pipes and catch basins?

    What you say generally makes sense; the detailed drainage network would likely attenuate the catchment runoff so omitting it would indeed likely result in a conservative inflow hydrograph. However, you might want to try to confirm that the catchment runoff does indeed ultimately discharge to the same element that you had assumed in the model. For example are you sure that all the runoff from catchment "E4" will enter the manhole "DMH-7"? If some of that flow goes to the pond, or further downstream, that could significantly impact your model results. Yes the model might be conservative but it might run the risk of being over-designed. Ultimately it will be up to your engineering judgment.

    As for the EGL issue in question - during the storm peak, it appears that the two pipes are flowing full/near-full and since you have additional flow entering the manhole, the flow in the downstream pipe increases, so the velocity increases, and therefore the velocity head increases (and therefore the energy grade line increases.) As an example if you direct the outflow of catchment E4 to the pond instead, you'll notice the jump in EGL does not occur since the flow (and velocity) in the downstream pipe does not increase. Was there something else specifically that you saw that made you feel that the results were not correct?


    Regards,

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

  • Hi Jesse,

    I went back and looked at my stage storage output from AutoCAD Civil 3D and it looks like I missed elevation 25, which messed up my Elevation-Volume table for pond PO-2. I have attached the updated stage storage table that I used to update the pond model. I'm still seeing discrepancies between the two methods, the only change I made between the two batch computes was changing Elevation-Volume to Elevation-Area for pond PO-2. I've attached screen grabs showing the maximum ponding volume results of the two methods. Any idea what might explain this?

    Also regarding basin E-4 that you mentioned, I am fairly confident in my delineation. We have a printout from a GIS model from the client showing the network of the nearby collection system. Pairing that with the contour data helped me delineate the basin. The client hasn't provided the actual GIS data so we're not holding our breath. 

    Thanks again,

    Kevin

    PDF

  • Hi Kevin,

    Thanks for the clarification. I've taken a look at the differences in results between the E-A and E-V cases and found that, although the max storage result field appears to be significantly different between the two cases, other results such as pond outflow do not have as large of a difference. I set up a separate scenario with a new child physical alternative to observe the differences between E-A and E-V for the 100 year event, so that the pond outflow can be graphed to visualize this. You can also see the mass balance for the pond in the hydraulic reviewer, which indicates only a very small differences.

    So, I believe that the difference in results may indicate that the model is very sensitive, so a small change like this exhibits a relatively large change in some results. This can happen more often in models with sharp hydrograph peaks and certain pond geometries where conditions change quickly. This is also especially the case with the Implicit solver, which tends to require smaller timesteps and calculation option adjustment to achieve stable results with ponds. On the other hand, the Explicit (SWMM) solver tends to handle pond models better. I noticed if I use the Explicit solver in your model with a timestep of 0.025 hours, the storage results are almost identical between the E-A and E-V cases. I would recommend using the Explicit solver with this pond model.


    Regards,

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

Reply
  • Hi Kevin,

    Thanks for the clarification. I've taken a look at the differences in results between the E-A and E-V cases and found that, although the max storage result field appears to be significantly different between the two cases, other results such as pond outflow do not have as large of a difference. I set up a separate scenario with a new child physical alternative to observe the differences between E-A and E-V for the 100 year event, so that the pond outflow can be graphed to visualize this. You can also see the mass balance for the pond in the hydraulic reviewer, which indicates only a very small differences.

    So, I believe that the difference in results may indicate that the model is very sensitive, so a small change like this exhibits a relatively large change in some results. This can happen more often in models with sharp hydrograph peaks and certain pond geometries where conditions change quickly. This is also especially the case with the Implicit solver, which tends to require smaller timesteps and calculation option adjustment to achieve stable results with ponds. On the other hand, the Explicit (SWMM) solver tends to handle pond models better. I noticed if I use the Explicit solver in your model with a timestep of 0.025 hours, the storage results are almost identical between the E-A and E-V cases. I would recommend using the Explicit solver with this pond model.


    Regards,

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

Children