Instabilities in unsteady model - I don't know where to begin to debug

[Mike_H]

Dear all

I'm a fairly new user of HEC-RAS, and I've built several steady-state models. However, I'm currently working on an unsteady model of a watercourse with 5 tributaries. Some of the survey data is less than perfect. The model has stability problems, with errors ranging from having exceeded the maximum number of iterations, to matrix errors.

The problem I'm having is that I've no idea where to begin to debug these errors. I've seen the page on the HEC-RAS website regarding the possible sources of error, and I had some success in trying to build the model up successively, reach by reach, by altering theta, adding cross-sections, altering the downstream boundary, but eventually, HEC-RAS defeated me. Any success I have had seems to be haphazard, to the extent that I wouldn't have any confidence in the final modelled ouput.

I've tried to look at the output log, but I don't even know where to begin to spot where the error is. Does anyone have any tips? (The HEC-RAS website suggests patience is important)

Thanks in advance

Michael

[MattZ]

I would model each tributary as a separate channel first. Once you have all of them up and running, then connect them one at a time.

Debugging is a skill that is hard to teach over a forum, so I probably can't be of too much help.

[cary.grant.jr]

In my experience initial conditions play a much more important role in unsteady simulations than in steady ones. I would try to find stable steady situations first for the range of flows you're simulating, so that you can track if the unstabilities are related to particular flow situations, and move to the unsteady version afterwards. Don't forget that friction changes with flow, your Manning's M-s (or n-s) might cause some trouble if you keep them steady while changing flow.
I also experienced that starting with a high flow (and corresponding high water level throughout the reach) is more likely to result in a stable situation, than low flow. Starting with high flow, one can alter Q & h gradually to reach more realistic values (adding artificial series before your real series)