Note that the error messages that you reported are coming from the FORTRAN floating point library when the math co-processor signals an exception with an attempted operation. These error message are generally not very meaningful, since they are coming from such a low level in the computations.
Here is what they are:
POW - Power Function
SQRT - Square Root Function
For example, calling the SQRT function with a negative number will generate an error. This is coming from the original NWS DAMBRK code when the model input is not correctly setup. BOSS DAMBRK traps on these floating point errors, resets the match co-processor and attempts to continue with the flow computations. However, this is not always successful in getting a model to run.
I have listed below some suggestions that you can try in an attempt to get your model to run. These items are only meant to help clear up any possible problem that you are having, and can be removed once you have found out what has caused the problem:
1. Try modeling steady-state conditions at first. A flat hydrograph and no dam-break failure. See if it will run steady-state.
2. Make certain that there is enough flow contained in the river. DAMBRK is a "wet" model, meaning that it must have some sort of base flow to "start-up". Therefore, to get the model up and running, I suggest that you artificially raise the amount of base flow that exists at start-up. For example, use a minimum flat hydrograph of 1000 cfs. Have a turbine flow of 1000 cfs as well, to make certain that the base flow is getting through to the river.
3. If the model runs steady-state, try changing the hydrograph to a more "realistic" hydrograph. Again, make certain to make small adjustments. If this causes the model to fail, the cross-section geometry may need to be smoothed.
4. A "general rule" is to allow the cross-section top-width to contract down to 0.5 and expand up to 2.0 from one cross-section to another (i.e., a cross-section top-width of 100 m can contract to 50 m or expand to 200 m). Any greater contraction or expansion should be "absorbed" using storage top width (i.e., a 100 m top-width could expand to 700 m by defining a channel top-width of 200 m and 500 m of storage).
5. Make certain that there is no large flat bottom inverts. For example, a channel bottom width of 100 m with a small base flow will prevent the model from having a sufficient depth of flow to start-up.
6. If subcritical and supercritcal profile conditions exist, make certain to use the "mixed flow option."
7. A specified dam-failure cannot occur too rapidly. The specified inflow hydrograph cannot change too quickly.
8. Try to simplify the model, and then add complexity to it. Try to reduce the number of items that could be causing your model to fail to run. For example, if a bridge is specified downstream of a dam, temporarily undefine it (changing the "Flow Structure at Cross-Section" entry to "NONE"). Then see if the model runs.
9. A small enough time-step value must be specified. For example, this generally ranges from 0.05 to 0.005 hours.
10. If a bridge is causing problems in routing your model, is there another way to model it? Could you model it as a dam with a spillway rating curve? Could you assume that the bridge gets washed out (conservative estimate)? Could you model it as simply a cross-section geometry constriction?
If you continue to have problems, you can contact me directly and I can troubleshoot your engineering model.
Sincerely,
Chris E. Maeder, M.S., P.E.
Senior Technical Engineer
chris.maeder@bossintl.com
Linear Mode
