Unsteady model error - "Matrix Solution Failed"

[Lih]

I am trying to run a model using unsteady flows, and it is blowing up straight away with the error "Matrix Solution Failed" and "Minimum error exceeds allowable tolerance at 29DEC2005 2400"

what do these errors mean and what are likely causes/solutions?

I have trawled through the forums for potential solutions and have tried the following:

*I thought it might have been due to initial conditions. I matched up my initial flows to te start of the flow hydrographs.

*I incorporated pilot channels along pretty much the whole model length

*I tried to create a hotstart file, using steady flows in an unsteady flow file + initial water levels at each XS for the corresponding flows, sourced from steady flow model results. I even tried reducing it to just the one inflow at the top, and a constant WL at the bottom.

But I still get the same error...

For what it's worth the model is 35000' long with elevations from 55' to -15', it is quite uneven with many bridges and obstructions and 2 split flows. However I dont think the geometry is too relevant at this stage as the model is falling over straight away.

Any assistance appreciated!

Thanks

[Zidacos]

UFFF

Well, you have inestabilities in the model, you must to interpolate cross sectios, reduce the time interval calculation, and study every parameters to make de geometry model as homogeneus as you can... and move the optimization and calculation tolerances.

sorry for my English, good luck.

Jorge

[Lih]

What do you mean by "make the geometry homogeneous"?

I am trying interpolating cross sections where required.

At the moment I'm trying the create a hotstart file, and ramping down the tailwater level. The model is going unstable just by creation of the hotstart!

[tsairi]

*try setting a minimum flow

*check the XS Htab Param.: use the max number of points, smallest interval possible for the expected WS. if the flow very low, use the channel minimum elev as the starting El (the default is a foot higher).

*as hotstart, use a stage hydrograph as lower boundary condidtion, starting from high stage and decreasing.

[AJP]

I had the same problem, and the only thing I can think of that you haven't already said you did is to set your minimum flow value (bottom left corner of unsteady flow data editor screen). I believe I was reaching the same error due to my cross-sections running out of water. When I set the minimum to a value roughly 5-10% of the peak value in my UPPER BOUNDARY - FLOW HYDROGRAPH, the model was able to execute without error. (also, I was not modeling baseflow conditions, which would have probably provided the constant minimum flow value that this fix works around.) I hope that is able to help you out

[MattZ]

Another important step is to run your model using steady flow with several different profiles that cover the range of flows you expect in the unsteady flow run.

It is easier to identify and fix problems using steady flow first.

[Lih]

Thanks everyone for the ideas. Model is up and running now. In the end I did the following:

* Took out all bridges to simplify the model. Then added them in one by one and addressing any instability issues as they came up.

* Refined pilot channel to be continuous with smooth transitions between reaches

* Due to irregularity of channel invert, a number of inline weirs wer added at locations displaying sharp drops

* hotstart file created by running a constant flow from top of model at a rate that approx half filled the channel. starting tailwater level filled the entire model domain and was ramped down to realistic levels.

* I did not change any of the calculation tolerances, except for maximising the number of iterations.