Analysis Parameters
During adaptive analysis, the system iteratively refines the starting mesh in order to reduce the size of individual elements in areas of high error — thus improving the accuracy of the solution.
When an adaptive analysis is performed, the following steps are executed:
- Maxwell generates a field solution using the specified mesh.
- It then analyzes the accuracy of the solution by calculating an energy value based on the error in the solution. The exact mechanism for evaluating the error varies by solution type. For example, in magnetostatic it can use Curl H to find the current density and then subtract all input currents and other sources. For a perfect solution the result would be zero, for a real, finite mesh the result is some amount of residual current density. An energy value calculated from this residual current density is called the error energy. The “Energy Error %” is the error energy as a percentage of the total energy (calculated with the original sources).
- If more than 1 pass has been completed, the software also calculates the change in total energy from the previous pass. The percentage difference is the “Delta Energy (%)”. Adaptive refinement continues until both the “Energy Error %” and the “Delta Energy (%)” are below the target Percent Error specified by the user (or until it reaches the Maximum Number of Passes requested).
- When the error targets are not satisfied the mesh is refined. This is generally done by subdividing the elements with the highest error energy into smaller elements.
- The user can optionally request to “Use Output Variable Convergence”. This is an additional stopping criterion. The Energy Error % and the Delta Energy must still be below the target Percent Error, but the software will also compute the specified Output Variable for each adaptive pass and will calculate the percentage change in that value for each pass after the first (this is the “Output Var. Delta (%)”). The solution will continue until the energy error criteria are met and the Output Var. Delta is below the target “Max. Delta Per Pass” specified by the user for output variable convergence (or until Maximum Number of Passes).