Troubleshooting unstable SewerGEMS and CivilStorm model results using the Explicit SWMM Solver

Applies To 
Product(s): CivilStorm, SewerGEMS
Version(s): CONNECT Edition, V8i
Area:  Layout and Data Input
Original Author: Jesse Dringoli, Bentley Technical Support Group

Problem

When using the Explicit (SWMM) numerical solver ("Active Numerical Solver" set to "Explicit (SWMM)"), the Calculation Summary indicates a high "Not Converging" (or a high "Flow Continuity Error" in the Routing Summary tab) and/or hydraulic results appear to be unstable and otherwise questionable.

Solution

Check data entry

There are many things that can cause a high continuity error or instability with the SWMM solver. First, review all your data input for errors and review the User Notification list. Errors in data entry such as elevations, including the elevations of composite outlet structures, can cause problems with the calculations which can possible impact the entire model.

You can also use the Hydraulic Reviewer to pinpoint specific areas with high "deviation" (mass balance error) to help locate areas to check more closely.

Adjust Routing Step

Next, try reducing the "Routing Time Step" in the calculation options. Reduce it gradually until results are better. For example, you may try a routing time step of 30 seconds, 10 seconds, 5 seconds, 2 seconds, down to a minimum of 0.1 seconds for very difficult models. Smaller routing time steps are often needed in models with fast-changing conditions such as pumps that cycle and sensitive control structures (especially weirs) and flow splits.

Note that for issues related to excessive flooding or surcharging, a smaller Routing Step may not be best and a value of 5 seconds or greater may be best.

Adjust Advanced Calculation Options

In more recent versions of SewerGEMS and CivilStorm, you can adjust addition calculation options:

  • The Max Trials per Time Step and Head Convergence Tolerance can be adjusted for better convergence. For the head convergence tolerance, if the difference in computed head at each node between successive trials is below this value the flow solution for the current time step is assumed to have converged. The default tolerance is 0.005 ft (0.0015 m), which can be increased for a better chance of convergence, at the cost of accuracy. You can increase the Max Trials per Time Step to allow the solver to try additional trials to attempt to converge, which can have an impact on performance.

  • You can view pipe convergence statistics in the output report, just below the mass balance section. In the CONNECT edition version (10.XX.XX.XX) of the software on the calculation summary go to the Report tab and review the different reports to determine, which elements needed to be reviewed for troubleshooting.

  • The Use Bentley transition equation option is related to when pipes first become surcharged. Using this option enables a new algorithm for a smoother transition from gravity to near-full and to full surcharged.  If you have trouble with the starting water surface elevation for ponds when using this option (mostly for versions above 10.03.XX.XX), try setting it back to False and adjust other calculation option- parameters mentioned here to help with stability.
  • Also related to surcharging pipes is the Surcharge Method SWMM calculation option. Either try "Extran" or "Slot". The differences are explained here: What is the difference between the Slot and Extran option for the Surcharge Method SWMM calculation option?
  • The Minimum Surface Area option is an EPA SWMM stability enhancer, representing the minimum surface area at nodes when computing changes in water depth. If 0 is entered, then the default value of 16.67 square feet is used. This can increase stability, at the potential cost of accuracy.
  • The "Use Variable Time Step?" option can be used if in doubt the smallest times step is being used. It is computed for each time period to prevent an excessive change in water depth at each node. The value entered is applied against the variable time step and is automatically applied to shorter pipes to preserve stability.
  • The "Time Step Multiplier" option is used when "Use variable time step" is set to True and it enables you to enter a safety factor (between 10 and 200%) to be applied against the variable time step as automatically derived to preserve the Courant stability criterion. A typical adjustment factor would be 75% to provide some margin of conservatism.
  • The "Time Step for Conduit Lengthening" property of the calculation options can also be used if in doubt the smallest time step is a trouble factor. This property allows the conduit to be artificially lengthened during the analysis such that travel time is equal to or greater than the time step.
  • Define Supercritical Flow: It is recommended that you keep this as the default setting of "Froude and Slope"
  • Try changing the Solver Compatibility calculation option in case a different solver version handles your network better. See more in the "Solver Compatibility" section of this article: EPA SWMM solver versions used by SewerGEMS and CivilStorm
  • If using Elevation as your Pump Control, make sure the pump OFF elevations are NOT set equal to the wetwell bottom.  Make them 0.5 to 1 foot higher, and also make sure the ON elevation is well above the OFF elevation (try something like 8 feet higher if there seems to be plenty of room before hitting the max elevation of the wetwell).

Other Techniques

  • Export the file to SWMM by going to File > Export > EPA SWMM. Open the file in the latest version of EPA SWMM, then compute the model and troubleshoot the messages in the status window.

  • If you're using pumps in your model, check the on and off elevation range, pump curves and adjacent pressure pipes. With the Explicit solver, better stability is typically achieved when using a virtual pipe on either side of the pump. For example, the pipe between the wetwell and the pump - set "is virtual?" to "true". Also, a particularly small routing step may be needed. Additionally, try using Multi-point as the pump curve type, as this can typically yield better results. If you still encounter stability problems (especially with complex manifolds), consider using the Implicit solver which tends to be more stable with pumping. If you are modeling a pressure sewer consisting mostly of pump stations, consider the GVF-Convex (SewerCAD) solver.

  • Review storm data and to make sure it's entered correctly and there is a label associated with each storm event.

See Also

Troubleshooting unstable SewerGEMS and CivilStorm results using the Implicit solver

What is the difference between the Slot and Extran option for the Surcharge Method SWMM calculation option?

Using the Hydraulic Reviewer

Recommended
Related