[MITgcm-support] Fwd: model time step 'deltaTmom'

[Martin Losch]

Madhu,

as you can probably read in any textbook about numerical modelling and also many times on this email-list and in the documentation (there's for examle this section in one of the verification experiments: <http://mitgcm.org/public/r2_manual/latest/online_documents/node131.html#SECTION004122100000000000000>),
the maximum possible time step is limited by things like the CFL criterion: velocity * deltatT/ deltaX < .5 needs to be satisfied for the model to be stable. Or the viscosity must not be too high, something like Ah * deltaT/deltaX^2 < .5, etc.

Sometimes the initial conditions are chosen so that large adjustment processes take place which require (because of large velocities) a very small time step. Once these adjustments are over, one can often increase the time step.

Martin

On Jul 12, 2012, at 7:44 AM, Madhu wrote:

> 
> Hello MITGCM Users
> I am using curvilinear grid to rum the MITGCM model. Total number of grid points are 175x200x15. The horizontal grid resolution  dx: 3- 5km and  dy: 4 -6 km. The model is running with time step 1sec, which is too small   'deltaTmom = 1.'  (for more details following is the data file used for model run).  
> 
> The issue is, I want to increase the  model time step so that the simulations can be done faster.
> 
> 
> #**********data***********
> # ====================
> # | Model parameters |
> # ====================
> #
> # Continuous equation parameters
> &PARM01
> 
> # viscAhGrid = 0.0
> viscAh=5.E2,
> viscAz=5.E-3,
> diffKzT=1.E-4,
> diffKzS=1.E-4,
> # diffKhT=0,
> # diffKhS=0,
> 
> no_slip_sides=.TRUE.,
> no_slip_bottom=.TRUE.,
> tempAdvScheme=33, 
> saltAdvScheme=33, 
> staggerTimeStep=.TRUE.,
> eosType='LINEAR',
> 
> tAlpha=2.E-4,
> sBeta =2.E-6,
> gravity=9.81,
> f0=0.000001365,
> beta=1.E-11,
> hFacMin=1.,
> readBinaryPrec=64,
> writeBinaryPrec=64,
> 
> nonHydrostatic=.FALSE.,
> rigidLid=.FALSE.,
> implicitFreeSurface=.TRUE.,
> exactConserv=.TRUE.,
> useSingleCpuIO=.TRUE.
> globalFiles=.FALSE.,
> &
> 
> # Elliptic solver parameters
> &PARM02
> cg2dMaxIters=200,
> cg2dTargetResidual=1.E-13,
> # cg3dMaxIters=100,
> # cg3dTargetResidual=1.E-13,
> &
> 
> # Time stepping parameters
> &PARM03
> startTime=0.,
>  endTime=43200.,
> deltaTmom = 1.
> pChkptFreq=0.,
> dumpFreq=21600.,
> monitorFreq=21600.,
> 
> &
> 
> # Gridding parameters
> &PARM04
> usingCurvilinearGrid=.TRUE.,
> delZ= 5.0 , 10.0 , 20.0, 30.0, 40.0 ,
>       50.0 , 60.0 , 70.0, 80.0, 90.0,
>       100.0 ,200.0 ,300.0 ,400.0,500.0,
> rSphere=6370.E3,
> &
> 
> # Input datasets
> &PARM05
> bathyFile='TOPO.bin',
> zonalWindFile='Ustr.bin',
> meridWindFile='Vstr.bin',
> hydrogThetaFile='T15lev.bin',
> hydrogSaltFile='S15lev.bin',
> checkIniTemp=.false.,
> checkIniSalt=.false.,
> &
> 
> The issue is, I want to increase the  model time step so that the simulations can be done faster.
> 
> How to increase the model time step?
> If I am increasing  'deltaTmom' >1 model outputs.....  NaN.  
> the files 'packeges.conf' , 'CPP_OPTIONS.h' and 'data.exf'  along with the screan output dump file 'exp1' are atteched.      
> 
> 
> 
> Waiting for reply.
> 
> Madhu
> 
> 
> 
> -- 
> 
> Madhu
> 
> 
> 
> -- 
> 
> Madhu
> <packages.conf><CPP_OPTIONS.h><data.exf><exp1>_______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support