DNS of passive scalar transport in canonical flows

Bibill · September 15, 2022, 05:31

Dear all,

I worked several years on low-fidelity CFD simulations and LES as well, but I have to run DNS of passive scalar transport for a new project.

As I am quite a beginner with DNS, do you have any nice codes with tutorials to advice ? Would be even better if some tutorials are made for scalar transport!

Any views on DNSfoam?

Thanks in advance for any kind of suggestions!

FMDenaro · September 15, 2022, 07:45

You can use any Navier-Stokes code, also a standard second order FD-based.
Your real problem is the computational cost for your simulation, have you estimate an order of magnitude of the number of grid node required for your flow problem?

sbaffini · September 15, 2022, 08:33

For what I can say, Nek5000 is the openfoam equivalent for DNS, meaning that it has a large user base (for DNS of course), is used at universities and research centers and is actively maintained. Also, it is not more difficult to use.

Still, differently from OF, it is seriously verified and validated.

LuckyTran · September 15, 2022, 10:47

If you know how to LES, you know how to DNS. LES is wayy more complicated. DNS requires 2 brain cells.

dnsFoam is not a general purpose dns solver, it's for turbulence in a box problems. dnsFoam has a little random source term to offset the dissipation that occurs when you do turbulence in a box. Otherwise, dnsFoam is just icoFoam, i.e. incompressible and piso. You can use just about any other openfoam solver and it would support more options, like non-piso solvers and compressible N-S instead of purely incompressible N-S. You could also just turn off the source term in dnsFoam, but then it now a worse version of icoFoam.

September 15, 2022, 05:31	DNS of passive scalar transport in canonical flows	#1
Bibill New Member Join Date: May 2020 Posts: 23 Rep Power: 6	Dear all, I worked several years on low-fidelity CFD simulations and LES as well, but I have to run DNS of passive scalar transport for a new project. As I am quite a beginner with DNS, do you have any nice codes with tutorials to advice ? Would be even better if some tutorials are made for scalar transport! Any views on DNSfoam? Thanks in advance for any kind of suggestions!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Implication Of modified Scalar transport equation in PimpleFoam	shouvik ghorui	OpenFOAM Pre-Processing	0	August 25, 2022 03:19
undetermined cells in mesh cause passive scalar transport equation to diverge	backscatter	OpenFOAM Running, Solving & CFD	3	February 20, 2022 09:42
Peng Robinson Equation of state	shock77	OpenFOAM Running, Solving & CFD	0	November 18, 2021 05:48
Validation of passive scalar transport	kostnermo	OpenFOAM Verification & Validation	3	December 18, 2016 16:25
dieselFoam problem!! trying to introduce a new heat transfer model	vivek070176	OpenFOAM Programming & Development	10	December 24, 2014 00:48

September 15, 2022, 07:45		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,882 Rep Power: 73	You can use any Navier-Stokes code, also a standard second order FD-based. Your real problem is the computational cost for your simulation, have you estimate an order of magnitude of the number of grid node required for your flow problem?

September 15, 2022, 08:33		#3
sbaffini Senior Member Paolo Lampitella Join Date: Mar 2009 Location: Italy Posts: 2,192 Blog Entries: 29 Rep Power: 39	For what I can say, Nek5000 is the openfoam equivalent for DNS, meaning that it has a large user base (for DNS of course), is used at universities and research centers and is actively maintained. Also, it is not more difficult to use. Still, differently from OF, it is seriously verified and validated.

September 15, 2022, 10:47		#4
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,754 Rep Power: 66	If you know how to LES, you know how to DNS. LES is wayy more complicated. DNS requires 2 brain cells. dnsFoam is not a general purpose dns solver, it's for turbulence in a box problems. dnsFoam has a little random source term to offset the dissipation that occurs when you do turbulence in a box. Otherwise, dnsFoam is just icoFoam, i.e. incompressible and piso. You can use just about any other openfoam solver and it would support more options, like non-piso solvers and compressible N-S instead of purely incompressible N-S. You could also just turn off the source term in dnsFoam, but then it now a worse version of icoFoam.