Magneto-QuasiStatic: Interpolation error

[Sylvain]

For coupled LF simulations, where the MQS solver uses the Vector Potential from another simulation as a source (e.g. as in the WPT exposure tutorial), one can get the following error message (emphasis mine):

Error : Simulation 'LF 1' reports the following failure:
Error : Interpolation error: Component 2 seems to be outside the interpolation grid: -0.144094 < -0.106346.
Error : Simulation 'LF 1' failed on 2018-Jul-18 08:02:14
Error : iSolve framework failed (see previous error messages).
Error : The solver process returned an error code:

The interpolation error in the quasi-static simulation occurs when the
solver is unable to assign a value for the vector potential for every point
of the computational domain. In other words, the vector potential used as a
"source" is defined over a region that is too small compared to the size of
the computational grid (of the quasi-static solver). To assign values at
each grid node, the solver uses an interpolation method. This allows you in
principle to use a different grid for the Vector Potential simulation and
the MQS simulation. To address this issue, you have two solutions:

1. increase the size of the region where the vector potential is defined
   (e.g. by adding extra padding in the grid of the Vector Potential
   simulation)

2. decrease the size of the region where the vector potential is needed
   (e.g. by reducing the padding in the grid of the MQS simulation)

To know by how much you need to increase or decrease the grid size, and in
which direction, I would suggest comparing the values of the "Boundary x+",
"Boundary x-", "Boundary y+", etc... in the Option window of each simulation
when the Grid setting is selecting.
For the interpolation method to work, the grid of the MQS simulation has to
be "well inside" (i.e. one or two cells) the grid of the Vector Potential
one.

[gbgbha]

@Sylvain Hi Sylvain, thanks for the helpful hints. I followed your instructions and my simulation works. However, I did not calculate exactly how much to adjust the padding. I would like to ask if this affects the simulation result significantly? As a related question, when I work through the tutorials myself, should I expect to get similar numerical results as shown in the manuals? I get similar-looking graphics and am happy about that, but the numerical results do not agree.

[Sylvain]

hi,

1. regarding padding: as a general rule, if you are not sure whether a numerical parameter will affect your simulation results, the best course of action is to just "try and see". Change parameter -> see if it affects the results significantly. For the padding, you should try to increase it and check that your simulation results are not affected by spurious "boundary effects".
2. as for tutorials: yes, in principle you should obtain the same results as shown in the manual. If you do not, please check carefully that you followed the steps faithfully (pay attention, for example, to the Normalization settings in the Analysis/postprocessing part, after extracting the field sensors - forgetting to tick this box would often lead to orders of magnitude difference in the results).