Isight has numerous applications. One of its popular application is Material Calibration.

What is Material Calibration?

Material calibration is the process of deriving material behaviors like elasticity and plasticity from sets of material test data. The material behavior is characterized by the value of the material constants and the underlying mathematical equations. In a simple workflow below, material calibration would mean finding out the values of the constants A, m and n that would match the curves made from the experimental data.

How Isight helps in Material Calibration?

Isight, with its ability to perform simulations, data comparison and a seamless encapsulation of this workflow  to perform Optimization, is indeed well suited to perform material calibration.

A typical Isight workflow for material calibration will look like :

Yes, it is really that simple, containing only three components, namely Abaqus Component, DataMatching Component and Optimization Component.

Abaqus component  :

Using the Abaqus component, which I would like to call as The Simulator, one can simulate the testing scenario from the material labs. The abaqus model usually will be  simple tension, compression or relaxation test, often done on single elements.  The material parameter values for the abaqus simulation are given by the Optimization component.  The abaqus component can read and edit either  .inp file or  .cae file.  It can read the results from either .odb file or .dat file. The results, usually the stress and strain outputs, are read as arrays and passed on to the DataMatching component. It is also possible to perform additional post processing on these results and also store them into a file, before passing them on to the DataMatching Component.

DataMatching Component :

The Data Matching Component, which I call as The Comparator, allows target data to be visualized and compared with any simulation results. Within a few mouse clicks one can read in the experimental data and simulation data and readily plot them. The data can be either read in from text files or  from array parameters which is usually the case for simulation data, as the simulation results are supplied by the preceding  Abaqus component.  DataMatching Component also includes the easy to use data selection functionality from DataExchanger component, which enable users  to quickly select the specific data  from text file  that needs to be used for comparison. Once the experiment and simulation data are available in this component,  one can use a rich variety of error metrics either of the data separately or of the comparison between the two data sets.

Optimization Component :

The Optimization Component, which I call The Brain, is where all the logic for choosing the values of  the material parameters.  The goal of the optimization component will be to ( usually) minimize the value of the metrics, or combination of metrics  supplied by DataComparison Component. The mapping between the design variables ( material parameters ) and the output variables ( DataComparison metrics) is termed as the Design Space. Depending on the how the design space is,  one can choose from a variety of optimization techniques available in the Optimization Component.  For example,  if the design space has a single minimum like shown the below left picture, one can use Gradient based techniques like Sequential Quadratic Programming ( NLPQL ),  Large Scale Generalized Reduced Gradient ( LSGRG)  etc. or Direct Methods like Hookes-Jeeves Pattern Search and Downhill Simplex. On the other hand, if the design space is complex with multiple minima, like below right, one can use Exploratory techniques like Particle Swarm, Adaptive Simulated Annealing, Evol and Genetic Algorithms.

The above mentioned methods work very well if used with the right design space and with an educated guess about the starting points. For that one either needs to be familiar with the optimization technique or know the design space or both.  But most of the times one will have no idea of the design space and might not be familiar with optimization techniques. For precisely situations like that, there exists a technique called Pointer-2 in the Exploration Component. Pointer-2 is like an automatic transmission for optimizers. It is an optimization strategy where the algorithm will decide which of the various optimization techniques to use for the given problem. And depending on how the optimization process proceeds, it will automatically change to a different, more suitable technique.

If you are still not convinced that Isight is an easy tool to do material calibration, look at the following video to check out yourself, how easy it is to create a material calibration workflow.