The purpose of this web site is to assist engineers wishing to perform EM
model/simulation validation. A set of problems is provided, with results
in some cases, to be used to help insure the simulation results are accurate. Measurement
data and modeling data is provided with enough detail so the user can repeat
the work, and compare his/her new results to the previous results.
This site is a joint effort between the IEEE Electromagnetic Compatibility (EMC)
Society TC-9 subcommittee and the Applied Computational Society (ACES). All
papers and data on this site have been reviewed by a special committee for completeness,
technical content, and value to the EMC modeling community.
All papers and data may be downloaded and used as desired, as long as proper
credit to the source is recognized in any new publication of results using this
data.
In addition to the test/model data results, a section is provided for ‘Standard’ EMC
modeling problems. These problems have been documented with sufficient
detail so that a user can easily create these models, and potentially analyze
commercial or new software codes for their ability to perform meaningful EMC
simulations.
The validation process for modeling and simulation results is dependent on exactly
what is being validated and what data is available to support the validation. The
preferred process is to validate the unknown case against a known reference. To
this end, standard problems and other means of obtaining a reference are described
in IEEE P1597.1™/D4.2 together with a procedure to compare the resulting
data sets. Where no suitable reference case can be obtained, validation
is possible by using self referencing schemes. While less encompassing,
these schemes still permit a high level of confidence to be achieved.
NOTE—The terms ‘modeling’ and ‘simulation’ can,
in most places, be used interchangeably.
The IEEE P1597.2 Recommended Practice provides a wide variety of reference problems
that are similar to the types of problems encountered at the work place, and
the results of these given problems can be used to compare to the user’s
problem and hence validate a modeling or simulation tool. The Feature Selective
Validation (FSV) method is a means to consistently compare the model being validated
to the reference data (as described in P1597.1).
There are three different levels to a complete model validation. When
deciding how to validate a model, it is important to consider which levels
of validation are needed. The levels are:
- Mathematical level: Computational technique validation
- Implementation level: Individual software code implementation validation
- Model level: Specific model validation
The first level determines if the underlying computational technique
is correct. The second expands this to include a particular code implementation,
and the third includes the specific details of a model. For a complete
model validation, all three of these levels are required to be correct. See
IEEE P1597.1™/D4.2 for more details on these three levels of validation
and how they are analyzed and compared using external and self references,
and the Feature Selective Validation (FSV) procedure.
NOTE—The term ‘numerical’ could be used equally in place
of ‘computational’. The authors of this document have chosen
to use ‘computational’.
|