[Scott La Vanne]

Chris,

How do you know whether a steady-flow optimization is diverging or converging? Is there an option to review the iterations?

Thanks

Scott

[Scott La Vanne]

I’ve only ever used the NOAA HM reports to develop a PMP. I’m not aware of other sources.

[Scott La Vanne]

Thanks for your feedback West. These sounds like good approaches. I’ll try this out and post back on here if I learn anything worth sharing.

Thanks again!

[Scott La Vanne]

For anyone else interested in this problem, below is the solution which Cameron was so gracious to help me discover.

The fix:

You have two cross-sections right next to each other (2042.102 and 2040.102) which are right next to each other, but the reach lengths for 2042.102 is over 2000 feet (should be 2 feet??). This reach length is the biggest culprit, others include the initial flow for the 1D areas and lowering the time step. I got it to run with 20 cfs and 100 cfs without a problem one I changed the reach length to 2 feet and adjusted the time step.

This reach length error creates a very long completely flat area within the reach, causing the high water surfaces. As I copied this from a purely 1D model before adding the 2D area, HEC-RAS must have recalculated reach lengths. The take home lesson from this is to verify your data hasn’t changed unexpectedly when copying forward data from another model, which I suppose is always good practice regardless of which program you’re using. I guess I was just too distracted with all this shiny new 2D stuff!

[Scott La Vanne]

Cameron – I have tried using the 1d/2d iteration options. It did not occur to me to use the full momentum solver until you mentioned it. I’ve now ran the model with the full momentum equations and something still isn’t right with the downstream reach. I’d be happy to share the data though if you’re willing to take a look. I’ll send you an email shortly.

[Scott La Vanne]

For anyone else who may come along later with the same error this is what I have learned.

1. the geometry worked in a steady state simulation so I suspected that a geometric element that is only active during unsteady simulations was to blame.

2. I isolated the problem element to a single lateral structure that connected two parallel reaches. Deleting the structure allowed the unsteady simulation to run. Despite many attempts to revise the structure and channel geometry I could not get the simulation to run with the structure present. There are multiple other very similar lateral structures present in the model that do not create this problem. I don’t know what made this structure unique.

I eventually decided that the structure wasn’t critical to the results I was interested in, and not worth investigating further. If anyone comes along with further insight I’d still be interested to hear your thoughts.

[Author]

Posts