BCs for multiple natures

Razi Abdul
2010-01-11
2013-05-27
  • Razi Abdul
    Razi Abdul
    2010-01-11

    Hi all,

    I'm not sure under which topic (help/general) to ask about this matter. It seems to me currently netgen (the mesher) only supports a single nature of BC for a segment  (edge/face) where you specify the BC for the segment with -bc=n (integer).

    As an example, suppose I'm solving a thermo-electric problem where the BC could be for temperature T and voltage potential U. It would nice to be able to do -bc=m,n where m = BC(T) and n=BC(U).

    My planned workaround is to use  -bc=mn where I have to code in my solver to parse the digits into two types of information (BCs). I'm just wondering if there's already an established method in netgen to handle/specify a BCs for multiple natures/physics modes for a single boundary segment.

    Thanks much

    Razi

     
  • Hello Razi,

    A Good evening to you, and sorry for the delay in replying….

    Netgen supports only one boundary patch identifier per face. I do not see the need or the requirement to extend Netgen to be able to handle multiple boundary patch identifiers per face, because usually you can specify multiple boundary conditions for each boundary identifier within the solver setup.

    For example, if I was doing a CFD simulation, for the inlet patch or face, I would have to specify different boundary conditions for the pressure and the velocity…. what is usually done, is to use the same boundary identifier (which is basically equivalent to calling my patch for example…. "inlet" or, in netgen "1") when specifying the different boundary conditions for the different physical variables.

    Each set of boundary conditions are then applied on the same physical face, the same way you need to apply Temperature and Voltage.

    I strongly suggest that you modify your solver to use the concept of using the same boundary identifier across multiple boundary conditions.

    Have a nice day!

    Philippose

     
  • Thanks Philoppose,

    Thanks much. Yup, it retrospect I think the current approach is already optimal.

    We're working on our solver, and it's not that much of a hassle to deal with the Netgen's BC identifiers to adapt to our own data structures..

    Best,

    Razi