
[Sponsors] 
March 18, 2014, 10:38 
Adding Boussinesq Approximation to multiphaseEulerFoam?

#1 
New Member
Dominik Schmidt
Join Date: Mar 2014
Posts: 11
Rep Power: 4 
Hello,
I'm quite new to openFoam, but I'm trying to implement the Boussinesq approximation into the multiphaseEulerFoam Solver (OF 2.2.x). My first step was to add a simple temperatureequation for the hole system, including the calculation of rhok, as in other Boussinesq solvers. TEqn.H: Code:
fvScalarMatrix TEqn ( fvm::ddt(T) + fvm::div(phi, T)  fvm::laplacian(DT, T) ); TEqn.relax(); TEqn.solve(); rhok = 1.0  beta*(T  TRef); pEqn.H  lines 8589 Code:
phiHbyAs[phasei] += rAlphaAUfs[phasei] *( fluid.surfaceTension(phase)*mesh.magSf()/phase.rho() + /*(g & mesh.Sf())*/ fvc::interpolate(rhok)*(g & mesh.Sf())/phase.rho() ); pEqn.H  lines 235241 Code:
phase.U() = HbyAs[phasei] + fvc::reconstruct ( rAlphaAUfs[phasei]*/*(g & mesh.Sf())*/ fvc::interpolate(rhok)*(g & mesh.Sf())/phase.rho() + rAlphaAUfs[phasei]*mSfGradp/phase.rho() ); So my first question is, if this implementation makes sense or am I missing something important? I also made some modifications to TEqn based on Cryogenics, compressible multiphase and heat transfer solver Code:
{ volScalarField rhoCp = fluid.rho()*fluid.Cp(); surfaceScalarField kByCpf = fvc::interpolate(fluid.kappa()/rhoCp + sgsModel>nut()/fluid.Prt()); fvScalarMatrix TEqn ( fvm::ddt(T) + fvm::div(phi, T)  fvm::laplacian(kByCpf, T) ); TEqn.relax(); TEqn.solve(); rhok = 1.0  beta*(T  TRef); } left: buoyantBoussinesqPimpleFoam  right: multiphaseEulerFoamBoussinesq boussinesqSmall.jpg Setup: Fluid: Air (rho: 1.205; beta: 3.43e03; Pr/Prt 0.71, nu: 15.11e6; Cp: 1.005, kappa(multiphase): 0.0257) T_left: 303 K T_right: 283 K TRef: 293 K I believe/hope the reason can be found in the different TEqnimplementations, rather then in the PEqn, but would be glad if someone could support this assumption, and suggests an easy way to validate the multiphaseBoussinesqversion. My first idea would be to edit the TEqn of PimpleBoussinesq Solver so it matches the TEqn with fixed diffusion coeff. (DT) and compare the results with the modified multiphaseSolver. \\EDIT: After implementing the same TEqn in both solvers I couldn't see much difference. Then I remembered that multiphaseEulerFoam always has a residual phase fraction for overcoming the problems with nonphase intensive handling of the momentum equations. So I edited the transport properties for the second phase (alpha uniform 0) to match the ones of the "alpha uniform 1" phase (air). After that the temperature distribution looks more like the one with buoyantBoussinesqPimpleFoam, but the velocities are still lower with the modified multiphaseSolver. So the differences seem to be related to the fact, that multiphaseEulerFoam is not designed for dealing with a single phase system ? Thanks ! Last edited by dschmidt; March 19, 2014 at 05:36. 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Boussinesq approximation, help please!  engahmed  FLUENT  0  May 20, 2010 11:25 
Question on boussinesq approximation  panos_metal  FLUENT  1  January 11, 2010 12:18 
boussinesq approximation  jamal  FLUENT  2  March 25, 2008 09:57 
InterFoam with boussinesq approximation  sinusmontis  OpenFOAM Running, Solving & CFD  0  March 28, 2007 09:35 
Boussinesq approximation again  Gabriel  Main CFD Forum  3  May 11, 2000 09:24 