Three Hidden Secrets to Speeding up Your SWMM6 Model

Three Hidden Secrets to Speeding up Your SWMM6 Model

1. Optimize Time Step Range:

   • Secret: The fastest simulations have a moderate time step range, where the minimum and maximum time steps aren't drastically different from the average time step.
   • Action:
     • Decrease the maximum time step: This reduces the average number of iterations needed per time step, leading to a faster run. Aim for an average iteration count below 4-5.
     • Increase the minimum time step: Techniques like "equivalent conduit lengthening" (which essentially simplifies the model hydraulically without significantly altering the results) can help achieve this. This will improve stability.
     • Monitor the Statistics: Keep an eye on the minimum, average, and maximum time steps reported in the simulation output. Experiment to find the "sweet spot" for your model.
   • Why it's hidden: Many users focus on just one time step parameter (often the routing time step) and don't realize the interplay between the minimum, maximum, and average values is crucial.
   • Example: Your provided example shows a maximum time step of 30 seconds and an average of 9.071 seconds, with an average of 4.821 iterations. This suggests that decreasing the maximum time step might further speed up the simulation.

2. Fine-tune Convergence Parameters:

   • Secret: Adjusting the stopping tolerance (also called convergence tolerance or error tolerance) and the maximum number of iterations allowed per time step can influence speed and accuracy.
   • Action:
     • Increase Stopping Tolerance: A slightly larger stopping tolerance means the model will accept a solution that's a bit less precise, but it might converge much faster. Be cautious: If the tolerance is too large, your results might be inaccurate.
     • Decrease Maximum Iterations: Lowering the maximum number of iterations prevents the model from getting stuck in very complex calculations. If the model often hits the maximum iterations, you might need to simplify the model or adjust other parameters.
   • Why it's hidden: The default values often work reasonably well, so users might not explore these settings. Understanding what they mean and how they affect the simulation is key.
   • 
     
     
     

3. Reduce Graphical Output for Continuous Simulations:

   • Secret: Writing large amounts of output data to the hard drive takes time, especially for long-duration continuous simulations.
   • Action:
     • Minimize Reported Data: Only report the variables you truly need for analysis.
     • Increase Reporting Time Step: If you don't need very high temporal resolution for your output, use a larger reporting time step (different from the routing time step).
     • Only report specific nodes or links: Report only the data you need from your network elements.
   • Why it's hidden: It's easy to default to generating a comprehensive report, but the impact on simulation time can be significant, especially for long runs.

Additional Important Tips  

• Utilize Multi-Core Processing: If your computer has multiple cores, configure your SWMM software to use them. This can dramatically speed up simulations, especially for large, complex models.
• Check for Model Issues:
  • High number of iterations or very small time steps often indicate underlying problems in the model setup (e.g., very short, steep conduits or instability at nodes).
  • Significant continuity errors can indicate problems that should be addressed to improve speed and accuracy.

In essence, optimizing a SWMM model for speed involves a balance between accuracy and computational efficiency. By understanding these hidden secrets and carefully adjusting the relevant parameters, you can significantly reduce simulation time without sacrificing the reliability of your results.