Exact analytical solutions to EMC problems apply only to a limited and unrealistic range of geometries. To solve most real world problems full wave numerical electromagnetic solvers must be used. However typical problems can take from hours to weeks to solve and many problems require too much memory and computer time to solve at all. An EMC designer would usually like to solve a problem is seconds so that the effects of design changes can be seen quickly. Full wave numerical electromagnetic solvers can be divided into two broad areas, finite techniques, and integral methods which are described briefly below.
Finite Numerical modelling techniques solve Maxwell's equations in differential form. They include, the Finite Difference method, The Finite Element method, and The Transmission line Matrix method. Finite methods have the following attributes:
Showing how an enclosure with contents might be discretised into volume elements for a finite solver (cross-section)
Integral methods solve Maxwells equations in integral form. They are also known as "boundary element methods" or "method of moments" solvers. Integral methods consider only the surface or boundaries of objects represented. Integral methods have the following properties:
Showing how an enclosure with contents might be discretised into boundary elements for an integral solver (cross-section)
For most EMC problems a useful, approximate, solution can be obtained with an integral, or finite solver. However the problem geometry must be simplified. The skill of the modeller lies in knowing what parts of the problem can be simplified whilst still obtaining a valid solution. The long solution times and large computing resources required are also a significant barrier to the widespread use of numerical modelling tools, though with the constant improvements in speed and reduction in computing costs it is now possible to solve may problems on desktop computers in hours or days.
Intermediate level modelling tools aim to provide a faster solution (seconds) with less computing resources to allow designers to obtain fast approximations. They are however less flexibile than the finite and integral techniques outlined above.