[Nek5000-users] param(2)

[nek5000-users at lists.mcs.anl.gov]

Thank you Paul!

p2 = \mu is indeed what we have used so far, and it is consistent with 
changes in p1 and p2 (same \nu = p2/p1 gives same solution).

The only "problem" then is the LES model from the turbChannel example, 
where
UDIFF = EDIFF in uservp, and
EDIFF = param(2) + dynamic Smagorinsky term in eddy_visc,
Here, param(2) is \mu and the dyn.Smag term has dimension as \nu.
If we multiply the dyn.Smag.-term by param(1) to obtain a "\mu_T", the 
simulation crashes after a couple of time-steps (with \rho=1000, for water).

But on closer inspection, for PN-PN, the dyn.Smag. term is split off as 
VDIFF_E (if nu_star = param(2)) and the correponding stress is added to 
BFX/BFY/BFZ in makevis, and then BFX (etc) is multiplied by \rho in 
makeabf, so it gets the correct dimension.

For PN-PN-2, it's not clear to me how this is handled, but the effect 
seems to be the same. EDIFF should be set to param(2) + \nu_T in 
eddy_visc (as it is turbChannel) to get the correct result for different 
\rho.

Carl Erik

On 01/18/2017 04:34 PM, nek5000-users at lists.mcs.anl.gov wrote:
>
> Thanks Carl Erik.
>
> It should be p2 = \mu.   (I hadn't noticed that error in the manual.)
>
> Examples are stand-alone entities and thus as long as they are self-consistent,
> that suffices.  Thus, in turbChannel (originally set up by me), I know that rho==1,
> and therefore ignore it always... such that \mu=p2==\nu, and I never bother to
> divide by rho.
>
> So, I think the answer to your question, if you want physical variables is:
>
> p1 = rho
> p2 = mu
> p7 = rhoCp
> p8 = conduct.
>
> You should also then set
>
> p21=0
> p22=0
>
> and
>
> p24=.01
> p25=.01
>
> These latter changes will turn off the non dimensional tolerances and tell
> the iterative solvers to set the tolerances in a relative sense according to the
> anticipated scale of your variables.
>
> Paul
>
>
> ________________________________________
> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov]
> Sent: Wednesday, January 18, 2017 8:07 AM
> To: nek5000-users at lists.mcs.anl.gov
> Subject: [Nek5000-users] param(2)
>
> Hi all,
>
> There seems to be some inconsistencies regarding dynamic (\mu)/kinematic
> (\nu) viscosity in nek:
>
> - The manual says that P002 is kinematic viscosity (\nu)
> - param(2) is stored in VDIFF, which is split by nu_star in split_vis
> - In the turbChannel example, param(2) is added to the Dynamic
> Smagorinsky term, which has dimension m^2/s, as has \nu
>
> On the other hand:
> - MAKEVIS is stated to calculate 2*DEL*[mue*(S ....)], using VDIFF_E
> - The manual says about P008: mavier5.f: param(8) = param(2) ! conduct =
> dyn.visc.
> - \mu is listed as a parameter in Chapter 3.3.1 of the manual, regarding
> the .rea file.
>
> In all the examples, param(1) = 1, so \nu = \mu, but for some physical
> cases it is convenient to keep the dimensional variables.
>
> So what is the correct use of param(2)?
>
> Thanks,
> Carl Erik,
> FFI, Norway
>
>
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>