High performance computing

Overview

The national laboratories of the U.S. Department of Energy (DOE) possess world-class expertise in high-performance computing (HPC) and operate some of the world’s most powerful computers. Complex computational studies can yield valuable insights and improve technology across scientific disciplines. Leveraging DOE’s HPC resources can expedite the development of energy-efficient manufacturing processes across U.S. industry—saving energy, reducing emissions, boosting competitiveness, and building global technology leadership.

Objectives

The DOE Advanced Manufacturing Office (AMO) launched the High-Performance Computing for Manufacturing (HPC4Mfg) program in 2015 to create targeted partnerships between U.S. manufacturers and the national laboratories. Under this program, competitively selected projects apply modeling, simulation, and data analysis to industrial processes and products to lower production costs and shorten the time to market. The aim is to advance future technology investments by optimizing designs, predicting performance, and reducing the number of testing cycles during development. In each case, the industry partner identifies the manufacturing challenge to ensure that the process or product will have direct commercial impacts.

This program is designed to foster an open exchange of best practices and know-how, breaking down communication barriers between world-class researchers, equipment suppliers, and manufacturing companies. Computer scientists at the national laboratories will benefit by expanding software codes for application to diverse problem sets. In the short term, industry benefits by reducing investment risk and improving energy efficiency. In the longer term, the U.S. manufacturing sector will increasingly turn to HPC modeling and simulation as trusted tools to accelerate innovation and achieve significant energy and cost savings.

For more information on the HPC4Mfg program, selected projects, and information on solicitations, please visit https://hpc4mfg.llnl.gov/.

For additional information, please contact:

Bob Gemmer
Technology Manager
U.S. Department of Energy

Advanced Manufacturing Office
Phone: (202) 586-5885
Email: bob.gemmer@ee.doe.gov

Recent projects

Study of Fluid Behavior Inside an Ammonothermal Gallium Nitride Reactor Using CFD

Gallium nitride (GaN) is a semiconductor material that could potentially lower the cost and improve the output of light-emitting diodes (LEDs). Researchers are looking for an inexpensive way to develop this promising material. The ammonothermal process is a relatively new way to synthesize single-crystal GaN, but the process is not well-understood and difficult to analyze experimentally.

Sorra Inc., which possesses specialized knowledge of GaN crystal growth, is working with LLNL to apply high-performance computing to conduct high-fidelity modeling of the complex physical-chemical processes involved in the growth of these crystals. The project could potentially save several years of costly trial-and-error experimentation and accelerate global market uptake of affordable, energy-saving LED products—possibly reducing U.S. electricity consumption by up to 20% in some sectors (EIA).

Soraa is using LLNL’s expertise in computational fluid dynamics (CFD) and HPC to simulate reactive flows affecting GaN growth. The team has shown that HPC capabilities are essential to capture key process physics at the required level of detail, including turbulent flow transition and temperature gradients near crystal growth sites on the substrate. The team has successfully used CFD to simulate Soraa’s crystal growth apparatus and is now comparing model results to a range of experimental configurations.

High-Fidelity Model for Coupling Flow and Mechanical Deformation of the Porous Paper Web

Maximizing water removal can significantly reduce the energy-intensive drying process in papermaking. The Agenda 2020 Technology Alliance, a consortium focused on the paper industry, is working with two national laboratories (LLNL and LBNL) to computationally simulate how water flows through paper pulp during and after the pressing process. According to Agenda 2020, new press designs based on this simulation could improve the solid content of the paper web entering dryers by 10%, reducing energy use in drying by up to 20% and saving an estimated 80 trillion Btu per year.

After water is initially pressed out of the paper pulp, some re-wetting typically occurs. Currently, neither the paper de-watering nor the re-wetting phenomena are well-understood, primarily due to the lack of sufficient data and reliable models describing these processes. The team is leveraging advanced simulation capabilities, experimental measurements, and paper machine data to develop an integrated, multi-physics modeling framework. LLNL is developing a continuum simulation framework to couple mechanical and two-phase flow models; LBNL is developing a pore-scale flow model (70% complete); and Agenda 2020 is providing the domain expertise and data to support model calibration, validation, and application. Project results will be made available to the scientific community through technical presentations and publications. The laboratories will work with industry to apply the developed model in guiding press section configuration and roll/felt design to minimize rewet. 

The Virtual Blast Furnace: An Integrated HPC Modeling, Simulation, and Visualization Capability for Steel Manufacturing

Integrated steelmaking is an energy-intensive process. The steel mills feed coke, a fuel derived from coal, into blast furnaces to produce the molten iron used to make high-tonnage carbon steels. Existing limits on modeling capabilities constrain efforts to optimize the energy efficiency of blast furnaces. Simulations are often run on desktop computers—and complex reactive flows or 3D simulations can take 30 days or more to complete.

The University of Purdue-Calumet is working with LLNL and an integrated steel mill to integrate and transform the mill’s existing codes to run on high-performance computer clusters—potentially improving simulation resolution/ times by a factor of 100. If optimized blast furnace processes were to reduce coke demand by 5%, the iron and steel industry might save $80 million annually.
The team has conducted parameter studies to better understand the effects of furnace inputs on coke consumption in a working furnace. These studies simulate internal furnace conditions under varying conditions, such as wind rate, percentage O2 enrichment, natural gas injection rate, and hot blast temperatures. LLNL has used CFD models on high-performance computers to examine hundreds of parameter combinations and study the impacts on blast furnace operations.

2015: Seed Projects

  • Study of Fluid Behavior Inside an Ammonothermal Gallium Nitride Reactor Using Computational Fluid Dynamics (Sorra Inc. and Lawrence Livermore National Laboratory [LLNL])
  • High-Fidelity Model for Coupling Flow and Mechanical Deformation of the Porous Paper Web (Agenda 2020 Technology Alliance, LLNL, and Lawrence Berkeley National Laboratory [LBNL])
  • The Virtual Blast Furnace: An Integrated High-Performance Computing (HPC) Modeling, Simulation and Visualization Capability for Steel Manufacturing (University of Purdue-Calumet and LLNL)
  • Utilizing HPC to Model the E-Iron Nugget Process (Carbontec Energy Corp., Purdue University and LLNL)
  • Microstructured Prediction in Additively Manufactured Metal Parts (Eaton Corp. and LLNL)

2016: Selected Projects

  • Numerical Simulation of Fiber Glass Drawing Process via Multiple Tip Bushing (PPG Industries and LLNL)
  • Integrated Predictive Tools for Customizing Microstructure and Material Properties of Additively Manufactured Aerospace Components (United Technologies Research Center, LLNL, and Oak Ridge National Laboratory [ORNL])
  • Massively Parallel Multi-Physics, Multi-Scale Large Eddy Simulations of a Fully Integrated Aircraft Engine Combustor and High-Pressure Vane (GE Aviation, LLNL, ORNL)
  • Process Maps for Tailoring Microstructure in Laser Powder Bed Fusion Additive Manufacturing Process (GE and ORNL)
  • Computational Design and Optimization of Ultra-Low Power Device Architectures (Global Foundries and LBNL)
  • Development of Reduced Glass Furnace Model to Optimize Process Operations (PPG Industries and LLNL)
  • Integrated Computational Materials Engineering Tools for Optimizing Strength of Forged Al-Li Turbine Blades for Aircraft Engines (Lightweight Innovations for Tomorrow Consortium and LLNL)
  • Highly Scalable Multi-Scale FEA Simulation for Efficient Paper Fiber Structure (Procter and Gamble and LLNL)
  • High-Performance Computing Analysis for Energy Reduction of Industry Spray Drying Technology (Zoom Essence Inc. and LLNL)
  • Weld Predictor App (Edison Welding Institute, Ohio Supercomputer Center, and ORNL)
  • Improving the Manufacturability, Performance, and Durability of Microporous Polymer Membrane Separators for Li–S Batteries using First Principles Computer Simulations (Sepion Technologies of California and LBNL)
  • Modeling High-Impulse Magnetron Sputtering (HiPIMS) Plasma Sources for Reactive Physical Vapor Deposition (PVD) Processes Used in Fabrication of High-Efficiency LEDs (Applied Materials Inc. and LLNL)
  • High Performance Computing Tools to Advance Materials Joining Technology (General Motors LLC of Michigan, EPRI of California, and ORNL)
  • Development and Validation of Simulation Capability for the High-Capacity Production of Carbon Fiber (Harper International Corp. and ORNL)
  • Level-set Modeling Simulations of Chemical Vapor Infiltration for Ceramic Matrix Composites Manufacturing (Rolls-Royce Corp. and ORNL)
  • Catalytic Pulping of Wood (Agenda 2020 TechnologyAlliance and ORNL)
  • Minimization of Spatter during Direct Metal Laser Melting Additive Manufacturing Process using ALE3D Coupled with Experiments (GE Global Research Center and LLNL)
  • Modeling Paint Behavior During Rotary Bell Atomization (PPG of Pennsylvania and LBNL)
  • HPC Modeling of Synthetic Jet Actuators for Increased Freight Efficiency in the Transportation Industry (Actasys Inc. and ORNL)
  • Multi-Physics Modeling of Continuous Liquid Interface Production (CLIP) for Additive Manufacturing (Carbon Inc. and LBNL)
  • Accelerating Industrial Application of Energy-Efficient Alternative Separations (Am. Chemical Society Green Chemistry Institute and LBNL)
  • Improving Gas Reactor Design With Complex Nonstandard Reaction Mechanisms in a Reactive Flow Model (Alzeta Corp. and LBNL)
  • Development of a Transformational Micro-Cooling Technology for High-Pressure Die Casting using High-Performance Computing (Shiloh Industries of Ohio and ORNL)