Multiscale engine combines macro FEA analysis and the micro simulation solver in the Multiscale Designer. (All images courtesy of Altair.)
Traditionally, these models are time-consuming and computationally taxing to develop because they require a combination of an FEA solver and a micro simulation solver. However, their advantage is that they allow users to obtain stresses within the individual constituents of the material which make up the unit cell.
As a result, engineers can, for example, assess the carbon fiber and epoxy matrix in carbon fiber-reinforced plastics (CFRP) instead of the homogenized ply-level stresses readily available from FEA codes.
“When you extract a stress or strain out of an FEA code from a composite simulation, it represents the homogenized stress over some volume of the ply,” said Jeff Wollschlager, senior technical director at Altair. “You have limited ability to determine which portion of that stress or strain is in the fiber and which portion is in the matrix.”
“This has been the major limitation with composites; we are modeling at the homogenized ply-level where sometimes composite failure assessment can be averaged or cancelled out,” Wollschlager added. “Modeling at the constituent level usually determines composite failure assessments more accurately since there is no homogenization over the ply, but traditionally at a high computational cost.”
Altair reports that its Multiscale Designer implements a reduced order model (ROM) technique that can determine the stress or strain within the individual constituents of the composite with lower computational costs while retaining the accuracy of the full FEA unit cell.
“The reduced order model capability of Multiscale Designer allows for the equivalent accuracy of more computationally expensive methods but provides those answers at a fraction of the computational cost,” said Wollschlager. “With multiscale methods, you have to run a macro FEA Solver and a micro solver that is responsible for the additional computation time.”
“It is important because, with the invention of the ROM, we now have a method that is computationally acceptable for industry to apply to most problems,” Wollschlager continued
The Multiscale Designer is part of Altair’s Multiscale Engine, which combines the company’s microscale simulation solver with macroscale FEA solvers like Altair’s OptiStruct and RADIOSS.
How Altair’s Multiscale Engine Works
Composite unit cells for various material types: continuous composites, fiber composites and woven composites.
“Multiscale Designer simplifies the creation of a unit cell by providing a built-in library of parametric unit cells for many commonly used micro structures in the industry,” said Wollschlager. “The built-in parametric unit cell library gives engineers more time to work on optimizing their designs.”
For instance, let’s say an engineer is designing a complex geometry. The parts will need multiple unit cells to characterize the composite material. This is because the concentrations of the composite components may vary at different regions of the part due to its shape.
In this case the engineer can enter the parameters of the unit cell, typical volume fractions and a couple of dimensions. The unit cell model is then built automatically.
“This saves a significant amount of time,” said Wollschlager. “For those materials for which we do not provide a parametric unit cell model, you simply have to develop your own and hook that into Multiscale Designer. This allows engineers to expand their library of materials.”
After the unit cell model is developed, Multiscale Designer uses the ROM algorithm to numerically reduce the unit cells in order to streamline the characterization of the part’s material properties. The unit cell is a volume of the composite that contains hundreds or thousands of finite elements with repeating compositions—the ROM algorithm will transform the computation of these unit cells into a series of deformation modes and state variables.
“Multiscale Designer also allows you to de-homogenize the stresses on a composite to get the stresses in the fiber and matrix individually,” explained Wollschlager. “This way, you get a fiber stress or strain and a matrix stress or strain to assess a limit state within each component.”
Other features, such as Multiscale Designer’s forward homogenization or inverse optimization approach, allow engineers to characterize materials that would otherwise be difficult and-time consuming to describe.
“The challenge with forward homogenization processes is that you have constituent properties, of which the fiber properties are typically unknown or easily available,” said Wollschlager.
Multiscale Designer can obtain these constituent properties using the inverse optimization technique. In this case, the program will determine the constituent properties of a composite given the homogeneous properties of the material which are typically available and determined experimentally.
“Typically, customers will have composite ply (homogenized) properties and not the constituent properties,” explained Wollschlager. “The inverse optimization process determines the constituent properties of the fibers and matrix that reproduce the composite ply property test data. In this way, it is much easier to develop a multiscale material model that requires constituent properties hard to obtain via experimental methods.”
Once the multiscale material model is developed, engineers can link it to various macro FEA solvers using the appropriate Multiscale Designer plugins for that macro FEA solver. The macro FEA solvers supported by Multiscale Designer include:
- OptiStruct
- RADIOSS
- LS-DYNA
- Abaqus
What Engineers Can Simulate with a Multiscale Engine for Composite Materials
Fatigue results from Multiscale Designer.
- Unidirectional
- Woven
- Chopped fiber
- Concrete
- Organics (bones)
- Natural inorganics (soil and stones)
Once these materials are simulated, the engineer can then determine a number of characteristics of the material in various geometries using the Multiscale Designer macro FEA solver plugins.
Some of the analyses compatible with the Multiscale Designer include:
- Linear and nonlinear structural simulations
- Ultimate failure assessments
- Statistical-based material allowables
- Fatigue
- Fracture
- Impact
- Crash
- Environmental degradation
Altair reports that when performing FEA mechanical computations, the unit cell will perform at a computational cost that is comparable to macroscale materials. This is because the Multiscale Designer produces and stores the material property databases for the unit cell for which the ROM can access easily and quickly for any multiscale simulation.
Altair has tested and validated Multiscale Designer to over 50 benchmark materials. This was done using coupons and component-level analysis of various composite components and forms.
Stochastic results from Multiscale Designer.
Engineers can use the stochastic capabilities of Multiscale Designer to assess problems with material property variability.
Using stochastic calculations, Multiscale Designer will translate the geometric and material uncertainties into macro-level component uncertainties. This will allow users to better assess the failure criteria of their products.
To find out more about Altair’s Multiscale Designer, follow this link.