Skip to content

mathias

Due to recent bot attacks we have changed the sign-up process. If you want to participate in our forum, first register on this website and then send a message via our contact form.

Forum Replies Created

Viewing 15 posts - 661 through 675 (of 705 total)
  • Author
    Posts
  • in reply to: Change the value of nx and ny for cavity2d examples? #2120
    mathias
    Keymaster

    Hi Thir,rnrn No. If you can change the source code in order to have any size you want. Have a look at our example “”cavity3d””.rnrnRegardsrnMathias

    in reply to: Change the value of nx and ny for cavity2d examples? #2118
    mathias
    Keymaster

    Hallo Thir,rnrn You can change the resultion in the xml-file by changing the parameter “”latticeL”” which is the size of a voxel in meter. In this example we have a square cavity of lenth 1 m. The function “”converter->numNodes(x)”” returns the number of voxels of size “”latticeL”” for x meter. In this example we set “”int N = converter->numNodes();”” which gives us the number of voxels of size “”latticeL”” for 1 meter. Have a look at the other examples e.g. “”cylinder2d”” and our user guid.rnrnMathias

    in reply to: compressible flow in LBM #2116
    mathias
    Keymaster

    Dear MAT,rnrnit seems not to be difficult to implement a model like “”http://www.math.ualberta.ca/ijnam/Volume-9-2012/No-2-12/2012-02-21.pdf”” into the OpenLB code. Basically, you would just have to cange the definition of the collision step in the classes “”dynamics.*”” starting by copying an existing implementation like “”BGK””. We could include your code in the next release if you are happy to share under GPL2.rnrnMathias

    in reply to: rho vs. rho +/- 1 (esp. for computePressureField()) #2113
    mathias
    Keymaster

    Hi,rnrnI understand. Sorry for confusing you. By doing it the way “”P = (RHO – 1) * (Cs*Cs)”” you obtain “”P=0″” for “”RHO=1″”. If you dont do that scaling, you will get different “”pressure levels”” for different discretisations. Then, with the help of the unit converter you can add any constant “”pressure level”” you want. rnrnMathias

    in reply to: natural convection #2111
    mathias
    Keymaster

    Hi Saida29,rnrn Have a look at the OpenLB examples “”thermal2d”” and “”thermal3d””.rnrnMathias

    in reply to: cylinder3d #2110
    mathias
    Keymaster

    Dear Atiya,rnrnyou can rn a/ change the stl-file another software like FreeCADrn b/ set the material numbers “”by hand””rn “”blockGeometry.setMaterial(iX, iY, iZ, 2);””rn c/ set the boundary conditions “”by hand”” like in “”cylinder2d””rnrnMathias

    in reply to: Data Extraction from 2D simulations #2109
    mathias
    Keymaster

    Dear Atul,rnrnIt is working like in the example “”cavity2d””:rn””vtkOut.writeData<double>(analysis.getVorticity(), “”vorticity””, (T)1/dt);””rnrnJust change “”Vorticity”” to “”Pressure”” and use the right scaling.rnrnMathias

    in reply to: rho vs. rho +/- 1 (esp. for computePressureField()) #2108
    mathias
    Keymaster

    Dear Randy,rnrnSorry for the late reply. Dividing by rho0 is just doing scaling. You need to do that also for the results you get in OpenLB for which you can use the class UnitConverter. The term -1 comes from the followed implemetation strategy in OpenLB (cf. answer question 1).rnrnMathias

    in reply to: rho vs. rho +/- 1 (esp. for computePressureField()) #2106
    mathias
    Keymaster

    Dear Randy,rnrn1) Why do you do this? Does this yield a more stable or accurate simulation?rnA: Yes indeed, we win one or more order of magnitude for the representation of the distribution function f which leads to more stability and accuracy. This is because f-f^eq is small and will be distributed around zero while f is distributes around f^eq which is not zero.rn2) Why is pressure calculated the way it is (shown below)? Isn’t the equation of staternp = rho/c^2,rnnotrnp = (rho-1)/c^2rnA: Have a look at some standard text books on LBM or “Lattice Boltzmann Model for the Incompressible Navier–Stokes“. In the later paper of He/Lou it is explained very nicely in my opinion. rnrnMathias

    in reply to: More than two components? #2103
    mathias
    Keymaster

    Hi,rnrnUnfortunately, we do not have an example code with a three- or more-component flow. However, it should be possible to extent the implementation. rnIf you do so, we will be happy to include your example in our next release.rnrnMathiasrn

    mathias
    Keymaster

    Hi,rnrn””mpiCC”” and “”mpi++”” should work since OpenLB is written in C++. “”mpicc”” is for C.rnrnMathias

    mathias
    Keymaster

    Dear lucasmr,rnrn Thanks for reporting the bug. If you cange line 359 in “”momentaOnBoundaries.hh”” in “”src/core”” from “”defineRhoU(cell, rho_, u_);”” to “”this->defineRhoU(cell, rho_, u_);”” it should work. rnrnWe will correct this in the next release.rnrnBest regardsrnMathias

    in reply to: velocity with LBM #2098
    mathias
    Keymaster

    Dear saida29,rnrnI am n ot really a Fortan expert. To speed up, can you please indicate the line where the velocity is finally computed, shorten the code and comment it.rnrnMathias

    mathias
    Keymaster

    Have a look at: http://dx.doi.org/10.1063/1.1399290 rnrnMathias

    in reply to: External flow in openLB – doubts #2092
    mathias
    Keymaster

    Dear Shreyas,rnrn The function “”setOffset”” of “”BlockGeometry”” does not change the matrix itself. You need to create a new “”BlockGeometry”” and copy the data by hand. Yet, it is much easier to use the read function of the “”STLReader”” with an appropriate offset. rnrnMathias

Viewing 15 posts - 661 through 675 (of 705 total)