	Applies To
	Product(s):	StormCAD, SewerCAD
	Version(s):	All
	Area:	Output and Reporting

Problem

My scenario is not converging when computed - the following message appears in the calculation summary: "Convergence is NOT achieved."

Background

The calculation solver was not able to converge on a solution given the number of traversals of the network and/or other unresolved user notifications. This can be caused by a number of reasons

Solution

Try to resolve all user notifications that are pertinent to the network functioning properly

In the user notifications dialog box you can double click on each notification and if it relates to a single element you will be brought directly to that element in the drawing pane. If there are many of the same type, right click on one of the user notifications and choose "select all elements with message id...."
Read the user notification carefully and make the necessary changes to the element properties based on the notification details
Run the Validate tool to potentially identify additional problems that need to be addressed.

Check for flooding

If you're experiencing Non-convergence and your model is severely undersized, with significant flooding occurring, please note that this can often be the cause of non-convergence. This condition makes it more difficult for the solver to converge. More recent versions of the product handle this situation a bit better when increasing the max trials option mentioned further below, but it is worth considering the limitations regarding flooding with the GVF-Rational numerical solver. The HGL will be reset to the rim elevation and the flow will continue in the pipe, whereas in the real system some of the flow will be lost. See this article for more.

Increase the number of network traversals in the calculation options

Open the calculation options for the current scenario (Analysis > Calculation Options)
Try increasing the number of network traversals. Start with a large number like 30. If that doesn't work try 100. If that doesn't work then there may be data entry issues or something else wrong with the model. If careful review does not surface the root cause, contact technical support or post on the Hydraulics and Hydrology forum. If 100 does work then increase the number until you find the lowest number that still allows the model to converge.
See more about the effect of increasing the max network traversals in the article shown below under See Also. Note that is is best to find the root cause of the excessive iterations rather than simply increasing the max trials.

See also: Understanding how Maximum Traversals help with non converging models

Adjust the Average Velocity Method

In the Calculation Options, try changing the Average Velocity method. If your model has particularly high velocities, check to make sure the slope/invert elevations are correct.

Make sure you have the latest version and/or patch set installed

For subscribers, the latest version can be downloaded by following these instructions: Downloading OpenFlows Software

Contact Technical Support for patches. See:Cumulative patch set frequently asked questions

Remove very short pipes

Pipes that are very short can tend to cause problems like this. Either correct the length, combine multiple pipes in series, or manually enter longer lengths. This can be done for example by filtering the conduit and/or channel flextable on Length (Unified) less than 5 ft, the global edit "Has User Defined Length?" to True, then global edit the user defined length field to 5 ft.

Pressure networks in SewerCAD (GVF-Convex solver)

If the non-convergence is associated with a pressure subnetwork, click the "Pressure Summary" tab of the Calculation Summary and review the related subnetwork . You may notice a specific timestep(s) that are not converged (as indicated by a "False" in the "Balanced?" column). Click the timestep in question and click the "status messages" tab at the bottom for potential clues. Next, use the Time Browser to locate the timestep in question and review what is happening in the model at that time. Here are a few possible reasons why the model may struggle to converge:

Excessive intermediate timesteps: for pressure networks, intermediate timesteps can be inserted if controls would trigger between the major hydraulic calculation timesteps. Make sure you have the "reporting time step type" set to "<All>" in the calculation options, otherwise you may be missing out on results (such as in graphs) between the major timesteps. Check to see if the pumps are cycling rapidly and if so, check and fix the controls to turn them on or off. See: Pump On/Off Elevations vs. Logical Controls vs SWMM Control sets
Air valve usage - if you have placed air valve nodes at high points within the pressure network and left the "treat as junction?" property as "false", then the numerical solver can have a harder time trying to solve the network as these are treated as PSVs internally. See more here: Troubleshooting Air Valves at High Points with the GVF Convex (SewerCAD) Solver
Check valve orientation - for pressure pipes with "has check valve" set to True, check to make sure the pipe's start/stop orientation is correct.
Variable speed pumps - If any pumps have "is variable speed pump?" set to True, confirm that the settings are correct and that you only have one such pump connected to the discharge side network (as multiple VSPs could "fight" against each other)
Adjust convergence calculation options - As a quick check, you can trial increasing the maximum Trials calculation option (for example 400 instead of the default of 40) or increase the Accuracy (for example 0.01 instead of 0.001) to help achieve a balanced run. It is generally better to locate the cause of the instability, but in some hydraulically challenging situations such adjustments may be necessary.
Zero flow conditions - during timesteps when a large portion of the pressure network has zero flow (for example all pumps are off), it can sometimes be difficult to converge, since the convergence checks the relative flow change between numerical trials. If you've determined that unbalanced timesteps occur during static/zero flow conditions, try increasing the "accuracy" calculation option (for example to 0.01). This enables the solver to balance with a larger change in flow between trials. Also if you have a very large pressure network with a large number of manifolded pumps discharging into it and the problem occurs during timesteps when all pumps are off, try inserting a small inflow into the upstream-most pressure junction. In the junction properties, expand the Inflow Collection and add a fixed load of 1 gpm for example. This small flow can help the model achieve convergence during times that would otherwise experience zero flow. Make sure that you place the inflow on a junction at the upstream end of the pressure network so that it results in a large portion of the network with the small flow.
Multi-point pump curves with sharp changes - if one or more pumps are using the multi-point pump curve type and your pump curve has sharp changes, consider setting "Use Linear Interpolation for Multi Point Pumps" to "false" or manually smooth out the sharp changes by adding more points to the curve.

See more in the following article (note that most are related to WaterCAD/WaterGEMS but some information applies to SewerCAD/GVF-Convex Solver): Troubleshooting the Network Unbalanced

Utilize the Explicit Depth and Slope equations Calculation Option

In the calculation options, try setting "Use Explicit Depth and Slope Equations" to "True". This can help with circular pipes that have trouble converging. For more details on what this does, please see the Help.

Complex diversions

Check the model for complex diversion situations ("Is Diversion Link?" = "True"). For example multiple diversions in series. Reduce complexity and excessive diversions wherever possible.

Convergence NOT achieved Message when computing StormCAD or SewerCAD model