Influence of solver to timesteps

Hello my name is Philipp,

i am currently writing my Master's thesis with MIKE-SHE. I implemented a small scale model (~50[km][/2]) coupled with a MIKE11 model for the hydraulics. My task is to accelerate the simulation, while the error is acceptable small. Therefor i compared for the OL (which needs the most time of the simulation) the explicit solver and the implicit SOR solver. The explicit simulation takes around 3-4 hours with an mean timestep of 36 min and the SOR needs around 18 hours with an maximum mean  timestep of 8 min, independent from the defined timesteps.
Has someone enough knowledge about the solver that he knows, why the implicit solver enables just smaller timesteps than the explicit solver? Because that feature is the extreme opposite of all i've ever learned, normally the implicit solver should use higher timesteps with more iteration steps and the explicit timestep should be much smaller...

Thanks a lot in advance!

Philipp

Regardless the running time of both methods, I am trying to stay away from the SOR method. This method may not converge, and you may have mass balance errors when this happens.

I hope this helps.