HYPERFIT

Although not in each model, the following load protocols are currently available:

- Uniaxial (1) tension

- Uniaxial (2) tension (i.e. in direction 2 - for anisotropic materials)

- Uniaxial tension + Tr.Def. (i.e. including transversal deformation data)

- Planar tension (~Pure shear)

- Simple shear

- Equi-biaxial tension

- Biaxial tension

- Triaxial tension

- Confined tension

- Volumetric tension

- ... and others

Note: - Each load protocol supports also compressive load data.

The test data can weighted in several ways to distribute and control their individual relevancy and/or reliability. The whole data-sets (i.e. individual tests) or the individual data-points (e.g. in a specific strain range) can be controled and weighted by the following weight instruments:

- W1 -
Manual weight factor for individual data-set - user can change significance of any data-set

- W2 -
Automatic correction of different number of data-points in individual data-sets - user does not have to provide equivalently sampled data-sets

- W3 -
Manual weight factor for individual data-point - user can change significance of any data-point

- W4 -
Automatic correction of different data-point density along the data-set - user does not have to provide perfectly sampled data-sets

- W5 -
Manual weight factor for individual data-channel - user can change significance of any data-channel (e.g. specific axis in multi-axial loading)

- W6 -
Automatic correction of different quantities in multi-channel response data - different physical meaning of individual channel are correctly taken into account

Least square optimization (or other if chosen) is performed to find optimal set of material parameters. The following optimization methods are available:
LINEAR METHODS:

- Linear least squares

NONLINEAR METHODS:

- Nelder-Mead (simplex)

- Levenberg-Marquardt

- Genetic algorithm

- Gradient descent

- ... and others

Note: - Setting of each method (internal parameters) can be controled/adjusted by user.
Note: - Methods can also be combined in a sequence (e.g. "Genetic algorithm" to locate rough solution + "Nelder-Mead" to refine).

Various objective quantities used for optimization are available:

- SS - Sum of Squares of differences *

- SA - Sum of Absolute differences *

- NE - Normalized error

- NRMSE - Normalized Root Mean Square Error

- r - Correlation coefficient

- R^2 - Coefficient of determination

Note: * Normalized and non-normalized versions are available. Various types of (stress) normalization is available.

- Any parameter can be fixed by the user at any (known) value.
- Any parameter can be constrained by user-specified low-limit or high-limit value.
- Multiple data-sets (unlimited number of stress-strain curves) can be used for a single fit.
- Any type of stress (Cauchy, Eng., etc.) and strain (Stretch ratio, Eng., etc.) can be chosen.
- Real-time stress-strain plots are available during iteration process.
- Real-time residuum history plot is available during iteration process.
- Iteration history is available and can be exported into a file.
- Any data-set and/or any data-point can be easily activated/deactivated.
- Each response channel (e.g. stress in direction 2) can be easily activated/deactivated.
- Solution can be interrupted and continued with (e.g.) changed parameters.
- ... and others