benefits of using boundary conditions

i am just wondering whats the use and benefit of applying boundary conditions when we solve any problem using full wave electromagnetic simulation software like cst or hfss?

i am just thinking what will happen if i dont apply these boundary conditions to my model. what difference it will have on my solution.
looking for any comments on this question


thanks

Hi, although I have never use hfss or the other simulation software you mentioned but from my experience with winfeko, xfdtd and supernec, i believe if you don't specify any boundary conditions there is already default values built into the software this default will be applied to your simulation.
Theoretically, when using software to simulate radiation of wave through unbounded space you have to specify boundary conditons because of the limited resources of the computer you are using(e.g speed of the processor and RAM available). Also, if you run simulation without BCs, there will be reflection at the boundary plane of your simulation space because no specifying the BC will make the boundary plane behave like a PEC (perfect electric conductor) plane, and these reflections will corrupt the result you will obtain from the simulation. You can check Allan taflove book on FDTD: computational electromagnetics, chapter seven.
Hope I have been of help. Regards.

With the boundary the radiations from the structures are observed and any back radiations are prevented..

HAbeeb

Computers can only store a finite amount of information due to memory limitations. Most numerical algorithms must then limit how big of a volume of space they operate to calculate fields. Fields at the boundary are heavily dependent on fields outside of the stored space. If these are completely ignored, fields outside the problem space usually default to zero making your problem look like it is surrounded by a perfect electric conductor. Many times you want waves incident on the boundaries to "keep going" and not reflect back into your problem space. This requires special assumptions about the fields outside of the problem space. These are called boundary conditions.

Dirichlet boundary conditions are the simplest and assume fields outside the problem space are all zero as discussed above. This is commonly used in waveguide models. Neuman boundary conditions assume the first derivative of the field is continuous across the interface. I rarely see this applied. There are transparent boundary conditions that try to estimate what the fields outside of the grid should be if waves are exiting the problem space. This is a great boundary condition when it can be used. Perhaps the most common now is the perfectly matched layer (PML). This is an absorbing boundary to absorb outgoing waves, but the impedance is matched to avoid reflections going into the boundary.

-Tip

In a simulation model you have to define everything. There is no such thing as "undefinded". rrumpf's explanation is very theoretical, so I want to add two easy examples:
1) When you simulate a hollow waveguide (hwg), you create the inside as a vacuum box. Then you define the boundaries as perfect condutor (and the waveports of course) and you have a perfect hwg.
2) To simulate an antenna you place it inside a box of air. The boudary of this airbox has to be of an absorbing kind. It must not reflect antenna radiation.

In HFSS the standard is perfect conductor (pec). Any boundary that is not defined is automatically set to pec. That is useful when you simulate hwg but can cause funny solutions of other problems when you forget to change.