Motivation for this new calibration process.
There is a capability in Abaqus/CAE to calibrate a hyperelastic material model and to calibrate a linear viscoelastic, Prony series, model. But these calibrations are separate and independent. In our classes on rubber: "Modeling Rubber and Viscoelasticity with Abaqus" and "Testing and Analysis of Elastomers" , we teach you how to use the calibration tools in A/CAE, but the key element is that the hyperelastic calibration uses one kind of test data and the Prony series calibration uses a different kind of test data. The test data used in the A/CAE Prony series calibration is often taken from a portion of the stress relaxation test. This means
- the user has to chop apart and modify the actual test data prior to its use in Prony series calibration.
- the viscoelastic calibration process assumes one can determine either the "instantaneous", or "long term" hyperelastic material model.
The tools in A/CAE have been used successfully for two decades by people in sealing/gasket applications. This is because their typical time frame of interest for the viscoelastic behavior is quite long (hours, days, weeks) compared to the time frame of the hyperelastic test (seconds, minutes). For that class of problems the fact that the A/CAE calibration process "bakes" some viscoelastic behavior into the hyperelastic definition of the material model was acceptable.
About a decade ago, we started to struggle with helping people interested in other applications, especially where the time frame of interest was much shorter. At short time frames in drop testing and crash applications, there were no tools available to us for calibrating a Hyper+Prony material model in a simultaneous and integrated fashion from test data such as a family of rate curves (shown in the image above, and posted in Butyl Rubber Test Data). Our struggles led us to try to accomplish testing that would fit into our available calibration tools, see the post: Hyper-visco rubber material models in Abaqus. As the Isight optimization software became part of the SIMULIA portfolio, another solution approach was to run reverse engineering optimizations using an Abaqus unit-cube single element model with the hyperelastic+Prony series material model. The downside to this approach is that it can be quite slow owing to the hundreds and hundreds of simulation runs involved.
General Overview of this new calibration process.
A better solution is to create newer calibration tools around a set of math that integrates the Hyper+Prony response and can be applied to any time-varying test data. In recent years, the calibration tool, MCalibration by Veryst LLC, has done just that. For the past several years a few of us at SIMULIA have also gradually developed internal tools based on a set of math equations for the integrated Hyper+Prony response. This post, by Li Zhang (April, 2015) :
Calibration for Hyper-Viscoelastic Material using Isight + Fortran
makes some of those ideas and tools available to you. The solution process that Li describes in his post has a drawback that we are trying to address – there is no graphic depiction of the stress-strain curves evolving as the optimization process unfolds. Li produced a script to get the stress-strain plots at the end of the process. It would be better to see the stress-strain curves evolving as the optimization process proceeds. We are working to address that issue, and hopefully will have a new post soon to share that with you.
For some prototyping of useful stress-strain graphics we developed an Excel version:
Integrated Hyper+Prony Calibration using Isight + Excel
And in July, 2016 we gave a one-hour customer eSeminar describing the reasons for these new tools, and spoke about both the Fortran tool and Excel tool driven by Isight:
A newer solution has been developed by Miguel Oliveira (August, 2016). Building on Li's work, Miguel has extended the scope of the calibration and switched to using a C++ program. Miguel provides a great deal of detail of the C++ program, the math behnid it, and videos that show the creation of an Isight process to drive the C++ program. Here is Miguel's post:
Integrated Hyper+Prony Calibration using Isight + C++
Back to: Material Modeling and Calibration - An Overview and Curriculum
