About Execution Modes and Simulation Models

The three distinct computational modes available for computing a material's response curve are the analytical, numerical, and FE execution modes. Analytical execution mode is the fastest option, but it supports calibration for fewer material models and is subject to other limitations. Numerical execution mode is slower but provides support for more material models. The FE mode is generally the slowest of the three modes, but it offers the greatest range of applications for model and material response complexity, and it supports the same set of material options as analytical mode.

Only one execution mode can be active during a given a calibration job or range response computation, but you can switch between the modes during a session.

Analytical Execution Mode

The primary advantage of the analytical execution mode is speed. It uses a highly performant computational kernel and typically completes calibrations in seconds. However, it supports hyperelastic and hyperfoam material models only.

The analytical execution mode has a number of limitations:

Analytical execution mode ignores the volumetric response for deviatoric tests of hyperelastic materials. Because the numerical execution mode can account for volumetric response, you will see variations in the material models produced using the analytical and numerical execution modes. Numerical execution mode uses no incompressible assumption about the deviatoric response of hyperelastic materials.
Viscoelasticity is not considered in the volumetric response.
The Marlow model is purely deviatoric.
Only uniaxial, biaxial, and planar tension deformation modes are supported for hyperelasticity with linear viscoelasticity, including temperature-time shift, and/or Mullins effect. For more information, see Time Domain Deformation Modes and Test Data.

Numerical Execution Mode

The primary advantage of the numerical execution mode is the breadth of material behaviors that can be calibrated. To cover a wide range of material behaviors, the numerical execution mode uses the material subroutines for Abaqus/Standard, which are a more computationally expensive kernel than the kernel used for the analytical execution mode. Calibrations using the numerical mode typically take on the order of seconds to tens of minutes to complete. The calibration time depends on the complexity of the material model, the amount of test data, and the quality of the initial guesses for the material constants.

The numerical execution mode can cut back automatically if it encounters convergence issues during a material response calculation. If no cutbacks are required, the app computes the material response only at the primary strain values included in the test data set. If one or more cutbacks are required, the app computes the material response at a set of intermediate strain values as required, but it always includes the response for all the strains in the test data set. During a calibration the error measure calculation (see Best-Fit Error Measures) includes only stress values associated with test data points.

The numerical execution mode uses a number of tolerances and solver settings to control the accuracy and performance of the material response calculation. The default values provide good general performance for most material models and test data, but you can modify the values as required.

FE Mode

The primary advantage of the FE mode is the arbitrary range of model complexity that it supports. Calibrations using the FE mode typically take on the order of minutes to hours to complete. The calibration time depends on the size of the models (DOF count), the number of finite element models that are active, the complexity of the material model, the amount of test data, and the quality of the initial guesses for the material constants.

The FE execution mode uses the automatic time stepping available to the procedure types that are included in the FE models. The app will automatically ensure that response data is computed at every time point that has been imported with the test data. During a calibration the error measure calculation (see Best-Fit Error Measures) includes only matched output requests and test data sets selected by you (see Performing Data Matching for Imported Finite Element Model Material Data).

The FE execution mode uses the convergence tolerances and control settings as defined in the imported FE models to control the accuracy and performance of the material response calculation. The convergence controls for the FE models cannot be modified from the app.

Execution Modes and Simulation Models

The Material Calibration app supports two classes of simulation models, built-in models and imported finite element (FE) models. The built-in models are very easy to use and cover the most common types of material testing procedures. However, they are only available in the analytical and numerical execution modes. Imported FE models require more effort to use since they must be authored by you, but they do offer a wider range of calibration scenarios, as compared to the built-in models. Imported FE models are only available for use within the FE execution mode.