Time step in EE

I used an existing model to run it under EE8. I find out have to use much smaller time step in EE than the EPA version that I compiled. I have to use 5 times smaller time step to get a stable run in EE8. I appreciate any insight and help ,

Thank you for your question, Fele. There could be a number of reasons for the time step difference. We would need to see the model to know for sure. One reason is that the EPA version can run even when it shouldn’t. It has very little error trapping and the arrays are not dynamic. So you could be running a model that in no way reflects your input files but wouldn’t know it. The DSI version has a lot of error trapping and debugging so if it runs it should mean it it is running correctly. Some things to check:

1. Are you sure EPA model is stable to begin with ? When a executable runs it doesn’t mean it’s stable.
2. Who compiled the source code ? Are all the arrays properly set in EPA code ? Usually a single EPA executable cannot be used on multiple projects as you need to change several parameters. With EFDC+ you don’t have to do that.
3. Is CFL condition valid in EPA code ? No negative depth checks in EPA version
4. Memory overwrite issues in EPA code.
5. Did you compare the results between EPA and EFDC+ ?

I hope this is helpful for you.

The model was forced with M2 tide at the ocean boundary- with known and expected tidal propagation (e.g., tidal range) at different locations within the water body. The model was compiled with proper arrays dimensions that fits this particular model setup. The results are comparable only if I used very small time step in E+.

Can you please send us the model that ran slowly in EE8? We have tested EFDCPlus on many old models and not had a similar result.