Multiscale Modeling and Simulation
Multiscale modeling or multiscale mathematics is the field of solving problems that have important features at multiple scales of time and/or space.
In physics and chemistry, we use multiscale modeling to calculate material properties or system behavior on one level. To do this, we might use information or models from different levels. On each level, we use particular approaches to describe the system correctly.
Most materials have some complexity of structure at the nano or microscale that influences their behavior at the continuum level. Our goal is to build a valid continuum model that captures this complexity of the micro and mesoscale. To do this, we need to bridge the gap between molecular scale models and the continuum.
Understanding Continuum-level Material Behavior
Materials scientists have for many years tried to understand continuum-level material behavior by looking closely at the material’s nanostructure. For example, we might want to study structure and properties at this scale. In this case, we can use molecular and mesoscale dynamics simulations based on classical equations of motion. However, using the results directly within finite element models is not entirely straightforward.
The purpose of this blog is to describe a suitable workflow to:
- Bridge the gap between very fine-grained work at the atomic/molecular level
- Work through a mesoscale simulation level
- Understand continuum-level material behavior for use in finite element (FE) simulations of real components.
Bridging the Gap
There are many materials and properties to calculate. Faced with this overabundance of opportunities, BIOVIA has made a consolidated effort to examine soft materials like polymers. Is it possible to take the material information from micro through meso to arrive at a Representative Volume Element (RVE) model? This approach is the simplest way to bridge the gap. Figure 1 represents the workflow from quantum mechanics, molecular mechanics and meso-scale. By understanding the atomistic scale, we can then connect to the continuum level to predict device-level properties. Machine Learning occurs at the initial stage. Here we want to find a new material based on experience or available information. We can move Machine Learning up or down in any of the methods based on data concentration.
A recent presentation at the SPE ACCE conference has adopted and tested this multiscale scenario over polyuria. The goal was to compare stress-strain curves at the micro and continuum level using Finite Element-RVE (FE-RVE). With an FE-RVE method, we can use FE tools to investigate and optimize the underlying constituent behaviors. Most importantly, FE-RVE captures the morphology of the microstructure with good accuracy. It also allows us to investigate “what-if” scenarios with the constituent behaviors.
Dr. Chatterjee has spent the last decade working on material design from catalysis to battery to semiconductor in support of R&D in the Asia Pacific region. Before joining Dassault Systemes he conducted materials science research as a research associate and post-doctoral follow in India, France and the USA before working as a visiting Associate Professor and senior research fellow in Japan. He has published some 140 peer reviewed papers and is involved in scientific consultant services and other national research projects in India and Japan. His current interest is in data science, material informatics and multi-scale multi-physics platform design. He holds a Ph.D. in Physico-analytical Chemistry from Burdwan University, INDIA.