open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,
I am conducting a wave-current interaction simulation but found that the computation time is somewhat long.

- When it is telemac3d only, the model took about 15 minutes to simulate 1 day.
- When it is only tomawac, the model took about 7 minutes to simulate 1 day.
- when I coupled the two models together, the model took about 50 minutes to simulate 1 day.

Is there any suggestion on how I improve the computation time? Perhaps relax some accuracy constraints in TOMAWAC computation? I have tried to change SEUIl values in some subroutines but didn't improve the computation time.

Thanks.

Edwin

Can you post the timesteps for tomawac and telemac and the 'COUPLING PERIOD FOR TOMAWAC' variable?

Costas

Can you give some precisions about the SEUIL values you changed ?

Elodie

Hi,

The time steps for Telemac and Tomawac are 60 seconds and 600 seconds respectively. The coupling period for tomawac is 10. Based on my earlier runs, the TOMAWAC model generated spurious wave results when I use timestep greater than 600 sec. Thus, I didn't try to increase that timestep.

I note that the value for SEUIL is defined at various subroutines.
I have only modified the value from 1e-6 to 1e-4 in qbrek1.f.

I have also tried to increase the value for "MINIMUM WATER DEPTH". This resulted in reduction in "NUMBER OF SUB-ITERATIONS" but not in the overall computation time.

Edwin

Hello ewang,

I after all think it is totally normal that you get a much higher computation time when you couple Telemac with Tomawac. Indeed, the propagation step is solved in TOMAWAC by means of the method of characteristics. When there is no current and the depth is constant over time, the characteristics can be traced back only once, at the beginning of the simulation. But when the current is not stationary, as in your case, the characteristics should be traced back at each time step when the current is updated. This takes some of the computation time...

Regards,

Elodie