[mnobrega]

Comments and questions are welcomed on the article Barestrand et al., Modeling Convective Heat Transfer of Air in a Data Center Using OpenFOAM: Evaluation of the Boussinesq Buoyancy, OpenFOAM Journal, https://doi.org/10.51560/ofj.v3.59.

[Marpole]

Thanks for the paper. Can you elaborate "An opening is therefore made in the geometry to act as a pressure reference to alleviate this, enabling the use of a specified flow rate for the CRAH-return areas."? Is this opening the supply area (grey)?

[broccolibadger]

Hi Marpole,

The pressure opening acts as a sort of reference, indirectly modelling that the data center is not airtight towards the atmosphere. Here it is done by making a small 10x10 cm hole with a simple pressure boundary condition. The effect on the stiffness of the CFD solution is substantial in how it improves convergence due to the other boundaries using enforced mass flows to preserve continuity across racks.

/Henrik

[milocheung]

Hi everyone, I am really appreciate this paper, and also provided the case files. Unfortunately, I've encountered some issues while attempting to run the "steady_incompressible" case, which has brought my progress to a halt. The detail error file also attached. It will be nice if anyone can help. Thank you very much.

Here's my setting and changes.

use finest grid no changes,"40mm, UUF" that one
system/meshParDict file. Updated "numberOfSubdomains" to 16
create logs folder

Copy and run the commands from "Allrun" file.

nmesh=16
rm -r constant/polyMesh/
blockMesh | tee logs/log1_blockMesh
surfaceFeatureExtract | tee logs/log1_features
rm -r processor*
decomposePar -decomposeParDict system/meshParDict | tee logs/log1_decompose
mpirun --oversubscribe -np $nmesh renumberMesh -decomposeParDict system/meshParDict -overwrite -parallel | tee logs/log1_rnM
mpirun --oversubscribe -np $nmesh snappyHexMesh -decomposeParDict system/meshParDict -overwrite -parallel | tee logs/log1_snappy
reconstructParMesh -constant -fullMatch | tee logs/log1_reconstruct
rm -r processor*
checkMesh | tee logs/log1_check
rm -r 0
cp -r 0.org 0
rm -r postprocessing
renumberMesh -overwrite | tee logs/log2e_rnM
buoyantBoussinesqSimpleFoam

Error when I run "renumberMesh".No problem when I pass "--oversubscribe" option in every mpirun.

command:
mpirun --oversubscribe -np $nmesh renumberMesh -decomposeParDict system/meshParDict -overwrite -parallel | tee logs/log1_rnM

error message:
There are not enough slots available in the system to satisfy the 16
slots that were requested by the application:

renumberMesh

Error when I run "buoyantBoussinesqSimpleFoam"

command:
buoyantBoussinesqSimpleFoam

error:
Time = 9

#0 Foam::error::printStack(Foam::Ostream&) at ??:?
#1 Foam::sigFpe::sigHandler(int) at ??:?
#2 ? in /lib/x86_64-linux-gnu/libc.so.6
#3 Foam::scalarProduct<double, double>::type Foam::sumProd<double>(Foam::UList<double> const&, Foam::UList<double> const&) at ??:?
#4 Foam::PBiCGStab::scalarSolve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const at ??:?
#5 Foam::PBiCGStab::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const at ??:?
#6 ? at ??:?
#7 Foam::fvMesh::solve(Foam::fvMatrix<Foam::Vector<do uble> >&, Foam::dictionary const&) const at ??:?
#8 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/buoyantBoussinesqSimpleFoam
#9 ? in /lib/x86_64-linux-gnu/libc.so.6
#10 __libc_start_main in /lib/x86_64-linux-gnu/libc.so.6
#11 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/buoyantBoussinesqSimpleFoam
Floating point exception (core dumped)

__________________
Milo

[Marpole]

As you can read from the error message "error message: There are not enough slots available in the system to satisfy the 16", the program asks for using 16 CPU and your computer might have less than 16 CPU.

[milocheung]

Hi Marpole,

Thank you for your reply. Apologies for the confusion due to the unclear typesetting; I have two distinct issues to address.

For the first issue regarding MPI, I believe I've found a potential solution. I'm considering using the "oversubscribe" option with the mpirun command, which seems appropriate given my computer's specs: an AMD Ryzen™ 9 7940HS Processor featuring 8 cores and 16 threads. This should allow me to utilize all available threads even if the number of processes exceeds the number of physical cores.

Regarding the second problem, I encountered a "Floating point exception" error while running buoyantBoussinesqSimpleFoam. Any idea for it?


__________________
Milo

[Marpole]

Hi Milo,

The second error is obviously a divergence error as you can see the residual of epsilon, k and the error of continuity. When reading the max/min U from the log, they are way large (e.g. 295,721m/s at time step 2).

This can be caused by a number of reasons. But I would review mesh and mesh quality first (file log1_check).

[milocheung]

Hi Marpole,

File uploaded. Thank you.

[Marpole]

The mesh quality is Okay. Can you display your decomposeParDict?

[milocheung]

Hi Marpole,

I pass "system/meshParDict" in "-decomposeParDict" option in following commands:

decomposePar -decomposeParDict system/meshParDict | tee logs/log1_decompose
mpirun -np $nmesh renumberMesh -decomposeParDict system/meshParDict -overwrite -parallel | tee logs/log1_rnM
mpirun -np $nmesh snappyHexMesh -decomposeParDict system/meshParDict -overwrite -parallel | tee logs/log1_snappy

system/meshParDict content

FoamFile
{
 class dictionary;
 format ascii;
 object decomposeParDict;
 version 2.0;
}

numberOfSubdomains  16;

method          scotch;

system/decomposeParDict

FoamFile
{
 class dictionary;
 format ascii;
 object decomposeParDict;
 version 2.0;
}

numberOfSubdomains  88;

method          metis;

coeffs
{
    n           (4 4 6);
}

The log1_decompose, log1_rnM, log1_snappy file also uploaded. Thank you.


__________________
Milo

[milocheung]

Hi,

I can execute the case files now. I downgrade the swak4Foam to v2021.05, and ubuntu from 22.04 to 20.04. Then I am able to run the cases. I think the ubuntu version is the main reason.

Thank you.
__________________
Milo