[MITgcm-support] EXACTCONSERV ?

[Nicolas Grisouard]

Hi everybody,

I'm trying to simulate a laboratory experiment in a tank with motionless 
walls (energy is brought by a vertical oscillation of the tank).

I'm having heavy problems of stability. I noticed that removing the 
rigid lid was a great improvement, but the velocity field was still odd 
: the simulation became stable but I got sort of long, vertical 
"stripes" as wide as dx (it is a 2D simulation), but whose magnitudes 
were not strong enough to make the simulation explode. I then tried to 
set the EXACTCONSERV condition to FALSE and these "stripes" vanished 
(although other problems persist, but I won't tell about them for the 
moment).

The main question I have is about the consequences of setting this last 
condition to false : it's a tank experiment and I don't want water to 
vanish, or the fluid to be compressible, but I may have misunderstood 
what this condition really meant. If somebody could explain me what is 
this condition for, it would be great. It is quite frustrating to 
improve a simulation by making wild guesses...

The complementary question is : why does it improve the stability of my 
simulation ? I guess this is harder to tell, but it may help me to 
understand a lot of other things.

Many thanks,

Nicolas