Supercomputer_2The U.S. Department of Energy (DOE) recently announced that it awarded $3 million to a portfolio of 10 projects as part of its High Performance Computing for Manufacturing (HPC4Mfg) program (see my previous blog post on the program). In reviewing the projects selected, I noticed a common theme that speaks to a critical aspect of using high performance computing (HPC) to advance manufacturing: nearly all of the projects involve integrated computational materials engineering (ICME).

ICME tools enable the multiscale data produced through HPC to be used to optimize manufacturing processes and products. By making critical connections between the properties and performance of materials, devices, and aspects of design and manufacturing processes, ICME enables manufacturers to improve efficiency, cut cost, and make materials and products with superior performance.

The basic theory behind ICME lies in the synthetic system view movement. According to Prof. Albert-László Barabási at Northeastern University, while reductionism in science has enabled physicists and materials scientists to develop theories and computational tools that allow us to understand various physical mechanisms governed at a range of specific time and length scales, it has also brought to light the difficulties of putting the dividend information back together to holistically understand the phenomena. Many scientists today agree that the system approach linking various chemical and physical processes occurring at different time and length scales is critical to our ability to advance manufacturing:

This system view that is the basic tenet of ICME is reflected in the projects DOE selected as part of its HPC4Mfg program. The majority of the selected projects aim to use or develop ICME tools to optimize materials properties and design processes, integrate materials structures and manufacturing processes, and/or conduct multiscale HPC modeling:

  • Optimization of materials properties and design processes: The Lightweight Innovations for Tomorrow (LIFT) consortium will partner with the Lawrence Livermore National Laboratory (LLNL) to optimize the strength of forged turbine blades for aircraft engines using ICME tools. Optimization of the design of a new drying method and transistor design will be performed by the team of LLNL and ZoomEssence, Inc., as well as another team of Global Foundries and Lawrence Berkeley National Laboratory (LBNL).
  • Integration of materials structures and manufacturing processes: United Technologies Research Center will partner with Oak Ridge National Laboratory (ORNL) and LLNL to develop integrated predictive tools that link microstructure with additive manufacturing processes, while General Electric (GE) and ORNL will try to connect microstructure with a laser powder bed fusion manufacturing process.
  • Multiscale HPC modeling: Procter & Gamble and LLNL will perform highly scalable multiscale finite element analysis (FEA) simulations of paper manufacturing, while GE will partner with ORNL and LLNL to improve the efficiency and component life of aircraft engines through design optimization using parallel multiscale simulations.

In 2013, my Nexight colleague Warren Hunt wrote a blog post that questioned when ICME would become a feasible tool for more widespread use, particularly for advancing materials manufacturing. Based on the focus of the newly awarded HPC4Mfg projects and current expert thinking, I can confidently say that the era of “ICME for all” is now.