Designing and Calibrating Materials for Additive Manufacturing – Part 1

By

Designing Materials for Purpose

Additive Manufacturing is used to build models, prototypes, tooling, and can produce parts in a variety of materials including plastic, metal, ceramic, and composites. Its advantages come from the additional design freedoms that allow complex geometries to be realized with little additional manufacturing costs.

Although the industry is well known for its advantage in prototyping, the use of Additive Manufacturing for serious engineering applications is still evolving. With designs no longer restricted by subtractive manufacturing constraints, and with processes no longer guaranteed by subtractive manufacturing standards, there are many questions that will need to be answered:

  • How do we effectively design lattice structures and take advantage of the additional design freedoms?
  • How do we validate intuitive lattice designs?
  • How do we make sure the in-service performance of printed lattices?
  • Can we print parts that are functional and reliable?
  • Do we dare use printed parts in critical load bearing sites?

The success of using additive manufacturing requires a thorough understanding of all the physics of the materials and the printing processes. Temperature-dependent material properties, process-induced material anisotropy, thermal conduction, convection and radiation, and material phase transformation are all important design considerations.

Multiscale Simulations 

In this post we will address the material design challenge with multiscale simulations. During the Science in the Age of Experience conference held in Boston in May 2016, Bernard Charlès – CEO of Dassault Systemes – said “Material for purpose, not purpose for a material.” Improved product design requires pushing materials to their limit and using advanced heterogeneous materials like polymers and composites to optimize stiffness and strength. Multiscale simulation plays a large role in understanding and bridging the scales across atomic, meso and continuum length scales, which spans from nanometer to millimeter.

Watch a workflow that starts with a single molecule and extends to a microstructure of the multi-phase material:

For example, let’s take a look at the material Polyurea. Polyurea can exhibit hyperelastic behavior with hysteresis but can also include plastic permanent set and damage. For many applications, using a phenomenological model fitted with coupon test may work. However, what if there was a new design intent, such as creating an active spring like member for a jumping shoe. This active member is subject to extreme compression as the runner hits the ground. In addition, it should “spring back” and perform well for the next landing, and all landings thereafter.

Polyurea is a very complex material belonging to a family of microphase segregated and thermalplastically crosslinked elastomeric copolymers. It also includes hard inclusions and a soft matrix. For us to be able to design the active member, we need to accurately characterize its mechanical behavior, which is highly dependent on its microstructure.

Single Molecule Design to Part Level

The design of polyurea materials can be done in BIOVIA Material Studio – an environment for virtual property and microstructure prediction at molecular level and at mesoscales. The above video shows the workflow in BIOVIA beginning with constructing a single molecule to the microstructure of the multi-phase material. The constituents, like hard and soft segments, can be modeled with continuum finite elements in a Representative Volume Element (RVE), or their response can be approximated by analytical mean field homogenization methods.

Figure 1: Conversion of mesoscale morphology into finite element model

Based on the atomic scale molecular simulation and coarse-grain mesoscale phase separation, simulation results we construct with a continuum mechanics Representative Volume Element (RVE) model with Abaqus at the same mesoscale are shown in Figure 1. We can import our RVE model into the FE-RVE plugin in Abaqus, which automatically defines the periodic boundary conditions on all external surfaces.

Subsequent constituent material calibration can be done starting with local mechanical properties of each phase/segment given by atomic scale simulations. Alternatively, one can characterize the multi-phase microstructure and use mean field homogenization to approximate and calibrate the aggregate material behavior.

Figure 2: RVE calibrated, Mori-Tanaka calibrated and PRF model calibrated aggregate behavior vs experimental test data

As another alternative, one can calibrate the aggregate material using phenomenological models such as the Phenomenological Rheological Framework (PRF) model. Figure 2 shows the calibrated aggregate material behaviors versus experimental test data with 100% compressive strain loading and unloading.

Finally, lets come back to where we started, i.e. with defining the purpose for the material. At the part level, in-service performance, and in this case, active loads from a runner or jumper can be applied in Abaqus, while using calibrated phenomenological models such as the PRF model. The results can be used to drive the RVE to analyze resultant behavior at the microstructure level.

Figure 3: Multiscale in-service performance simulation of a jumping shoe made of polyurea using mean field homogenization

In order to obtain a concurrent multiscale solution, one may also use mean field homogenization. Figure 3 shows the resultant Mises stresses on a jumping shoe under live dynamic loads.

The entire workflow is illustrated in Figure 4. We started with atomistic to obtain a first guess at estimating the mechanical properties of the two phases in the copolymer by using classical atomistic potential simulations. Then we used the dissipative particle dynamics method at the coarse grain to assess the morphology of phase segregation. The potentials were calibrated from the lower atomistic scale and the mechanical properties were delivered at the mesoscale.

Figure 4: Multiscale simulation workflow understanding and bridging the scales across atomic, meso and continuum length scales

Next we used a finite element RVE at the coarse grain scale to calibrate constituent properties of the hard reinforcement filler and soft polymer base. And finally, we created a few concurrent multiscale material definitions, mean field homogenization, and PRF, which are ready to be used for part level simulations.

Up to this point we’ve explored the procedure to design materials for purpose—from the single molecule design to part level, in-service performance prediction.

Part 2 of this blog post will discuss how to design and optimize materials for lattice structures. Stay tuned!

What to learn more about Dassault Systèmes’ simulation solutions for additive manufacturing? Visit: go.3ds.com/Print2Perform

0

Jing Bi

SIMULIA Additive Manufacturing Technical Consultant at Dassault Systemes Simulia Corp.
Jing Bi is a technical consultant at Dassault Systèmes SIMULIA focused on solutions that help realize the potential of Additive Manufacturing technologies. She received her MS and PhD degree in Mechanical Engineering from the University of North Carolina at Charlotte in 2010 and 2012 respectively. She joined Dassault Systèmes SIMULIA in 2012. She has worked in a variety of technical roles at SIMULIA and engagement with key customers and partners.