Projects

The HPC4Manufacturing program has funded several rounds of solicitations. Details about each of the projects are in the list bar.

Round 12: Fall 2020 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

Raytheon Technologies Research Center will partner with Argonne National Laboratory to develop a physics-informed machine learning technique to desensitize film cooling effectiveness to manufacturing variability and to inform design practitioners of the impact of manufacturing uncertainties on the lifecycle energy efficiency of gas turbine engines in project a titled “Robust Film Cooling Under Manufacturing Uncertainty For Improved Jet Engine LifeCycle Energy Efficiency (P.E00.0623)”.

Polyceed Inc (dba Glass Dyenamics) and Oak Ridge National Laboratory will utilize HPC- and Machine-Learning-Based Modelling to develop new electrochromic dyes for smart glass building windows with improved roll to roll manufacturability and low-cost in a project titled “HPC- and Machine-Learning-Based Modelling of Electrochromic Dyes for High Performance and Reduced-Cost Manufacturability of Electrochromic (EC) Devices”.

3M Company in partnership with Argonne National Laboratory will use a combination of HPC based CFD simulations and machine learning to minimize energy consumption of melt blown (MB) fiber manufacturing processes. Such processes are widely used for 3M products including filters, fabrics and insulation materials. Project title is “Next Generation nonwovens Manufacturing Based on Model-driven Simulation Machine Learning Approach”.

With the computing expetise of National Renewable Energy Laboratory, Element 16 Technologies, Inc., will improve Element 16’s molten sulfur TES product design with a high-fidelity HPC model validated by experimental data in a project titled “High-Fidelity and High-Performance Computational Simulations for Rapid Design Optimization of Sulfur Thermal Energy Storage”.

The Procter & Gamble Co will partner with Sandia National Laboratories to create an eco-system of HPC-enabled fiber manufacturing models to allow for defect-free production of solvent-free detergents with an accelerated timescale and reduced waste streams compared to traditional approaches such as build-test cycles in a project titled “Reinventing the Green Consumer Products Landscape with Material and Process Design using High Performance Computing”.

Electric Power Research Institute, Inc. will leverage Agronne National Laboratory's HPC expertise to apply state-of-the-art modeling and simulation tools to induction pipe bending nickel-based alloys for energy applications in a project titled “Modeling Dynamic Stress-strain-Temperature Profiles in Induction Pipe Bending to Improve Productivity and Avoid Cracking in Energy Intensive Applications”.

Generon IGS and Oak Ridge National Laboratory will use HPC to model the flow patterns in a shell-side fed gas separation module to maximize counter current flow patterns which could lead to a 50% reduction in the methane lost through the CO2 removal process in a project titled “Modeling of Shell-Side Gas Membrane Modules to Optimize Counter-Currency and Improve Selective Gas Permeation”.

General Motors LLC and Oak Ridge National Laboratory will utilize ICME tools to develop a high-performance lightweight additive manufacturing (AM) engine piston through material, shape and process optimization in a project titled “Improving Additive Manufactured Component Performance through Multi-Scale Microstructure Simulation and Process Optimization”.

Commonwealth Center for Advanced Manufacturing and Oak Ridge National Laboratory will establish foundational knowledge for developing and implementing technologies that enable the use of directed energy deposition (DED) for additively producing large gas turbine components using refractory metals in a project titled “Integrated Process and Materials Modeling for Development of Additive Manufacturing of Refractory Materials for Critical Applications”.

In partnership with Ames Laboratory, Praxair Surface Technologies will use HPC improve quality and yield of metal powder for additive manufacturing produced via close-coupled gas atomization in a project titled “Optimization of Processing Parameters for Metal Powder Production by Gas Atomization Utilizing CFD Simulations”.

Improving Jet Engine LifeCycle Energy Efficiency

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Dr. Michael Joly
National Lab Partner: Dr. Pinaki Pal, Dr. Muhsin Ameen, Dr. Opeoluwa Owoyele - Argonne National Laboratory

Summary: This proposal aims to quantify the impact of manufacturing uncertainties in gas turbine engines and to better assimilate lifecycle sensitivities in the development of next-generation energy-efficient technologies. Reliable film cooling drives durability and thermal efficiency of turbine stages in gas turbine engines, but is greatly sensitive to variations in the shape of cooling holes (such as, machining offset, blockage from thermal barrier coating, and surface roughness) induced by manufacturing processes. The primary objective of this proposal is to develop a machine learning technique to desensitize film cooling effectiveness to manufacturing variability. The novelty of this proposal is in the development and application of composite neural network as a surrogate of multi-fidelity computational fluid dynamics (CFD) simulations towards the development of a reduced-order model to inform gas turbine engine design practitioners of the impact of manufacturing uncertainties on the energy efficiency and durability of gas turbine engine components.

Cross-section of CMOS chip

Figure. Contour plot of velocity magnitude (normalized by the crossflow inlet velocity) on a vertical cut plane from a Nek 3D wall-resolved LES of an inclined jet-in-crossflow configuration with a shaped cooling hole (40 millio grid points). The simulation was run on 1800 processors.

High Performance and Reduced-Cost Manufacturability of Electrochromic (EC) Devices

Industry Partner: Polyceed Inc (dba Glass Dyenamics)
Principle Investigator: Dr. Anoop Agrawal, Dr. John Cronin, Dr. Sahila Perananthan
National Lab Partner: Dr. Stephan Irle, Dr. Debsindhu Bhowmich, Dr. Dmitry Ganyushin - Oak Ridge National Laboratory

Summary: The use of electrochromic (EC) dyes in Glass Dyenamic’s devices has shown to significantly reduce assembly cost for smart glass building windows with improved energy efficiency. Low manufacturing cost and aesthetic consideration has the potential of significantly increasing the adaptation of this technology in commercial and residential glass markets. However, the experimental design of suitable EC dyes with desired photophysical properties is highly resource intensive. We therefore propose a combined high-performance computing (HPC)- and machine learning (ML)-driven inverse structural design of anodic EC dyes based on high-level electronic structure theory to predict their photophysical properties in neutral and oxidized states. Highly accurate ab initio multireference wavefunction methods will be employed on OLCF’s Summit supercomputer to compute UV/Vis absorption spectra for a large training set of dye molecules. This data will drive a novel ML approach to predict novel EC dyes with superior properties.

Cross-section of CMOS chip

Figure: The goal of this proposal is to develop and employ a high-performance computing (HPC)- and machine learning (ML)-driven electrochromic dye design methodology. The results from this project will be used to produce energy efficient building windows that can be manufactured at an attractive cost using roll-to-roll processing methods.

Reducing Consumption of Melt Blown Fiber Manufacturing Processes

Industry Partner: 3M Company
Principle Investigator: Dr. Bill Klinzing
National Lab Partner: Dr. Ian Foster, Dr. Sibendu Som, Dr. Debolina Dasgupta - Argonne National Laboratory

Summary: This proposal aims at minimizing energy consumption of melt blown (MB) fiber manufacturing processes. Such processes are widely used for 3M products including filters, fabrics and insulation materials. The most impactful recent example is the base material for the making N95 mask during the COVID-19 pandemic. The process is extremely energy intensive since it relies heavily on compressed air and electrical heating. This proposal seeks methods to minimize energy consumption through a combination of High-Performance Computing, Computational Fluid Dynamics, and Machine Learning. It is estimated that the optimization will lead to a 20% reduction in energy consumption. Approximately 300 tons of MB nonwovens are produced worldwide each year by 3M and other manufacturers, consuming approximately 245 GW hour/year. A 20% (49 GW hour/year) reduction in energy consumption would have a global impact as 3M is a major player in the nonwoven manufacturing market and other manufacturers would likely follow suit.

Cross-section of magnetic field

Figure F1: (Left) 3M schematic of the MB process on a single fiber. (Right) An example of the MB process at a pilot scale. Fiber is extruded and blown from the die on the left to the rotating drum collector on the right, thereby forming the nonwoven fabric.

Optimization of Sulfur Thermal Energy Storage

Industry Partner: Element 16 Technologies, Inc.
Principle Investigator: Dr. Karthik Nithyanandam
National Lab Partner: Dr. Zhiwen Ma, Dr. Michael Martin – National Energy Renewable Laboratory

Summary: Industrial process heating (IPH) accounts for ~70% of US manufacturing energy use and is primarily produced by fossil fuel combustion. Approximately, 1500 TWht (~60%) of IPH demand is in the temperature range of 100-300℃. Industrial applications in this temperature range include drying, hydrothermal processing, thermal enhanced oil recovery, food and beverage, bioethanol production, etc. Cost-effective thermal energy storage (TES) that increases the utilization of waste and renewable heat (solar, geothermal, etc.) could provide significant energy savings and reliable heat sources, decrease emissions, and increase US manufacturing competitiveness through reductions in fuel consumption. This HPC4EI project will facilitate Element 16’s development and commercialization of low-cost and high-impact molten sulfur TES for dispatchable IPH and support its broad applications and deployment. The project will accelerate Element 16’s molten sulfur TES product design with a high-fidelity HPC model validated by experimental data.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

HPC Enabled Design Optimization of Sulfur Thermal Energy Storage for Industrial Process Heat Applications

Defect-free Production OF Solvent-free Detergents

Industry Partner: The Procter & Gamble Co
Principle Investigator: Dr. William Hartt IV
National Lab Partner: Dr. Rehka Rao - Sandia National Laboratory

Summary: A new-to-the-world product form may revolutionize consumer cleaning products such as laundry detergent, shampoo, dentifrice, and lotions, by lowering energy usage for manufacturing and transportation and significantly reducing the carbon footprint compared to traditional products. These novel products will eliminate the need for water transportation, yet still perform effectively and receive excellent consumers reviews. However, formulation and processing of these game-changing materials are challenging. P&G desires modeling and simulation technology to theoretically determine process windows and formulate new products optimized for reduced defects. The rheology of the precursor solution, containing surfactants, polymers, and other actives, must be optimized such that fibers may be formed and solidified without instabilities such as droplet formation, fiber folding, or breakage. Process development and optimization require significant insight into these physical process for predictivity and control; process breakdowns and low yield can be costly. In this proposal, P&G will utilize expertise in rheology and process development complemented by HPC transient 3D multiphase viscoelastic flow models developed at Sandia to provide a modeling and simulation approach, using machine learning, to capture the complex rheology and advance process design. This “digital manufacturing” approach will allow for defect-free production of solvent-free detergents with an accelerated timescale and reduced waste streams compared to traditional approaches such as build-test cycles.

Cross-section of CMOS chip Cross-section of CMOS chip

Improving Modeling and Simulation Tools to Induction Pipe Bending

Industry Partner: Electric Power Research Institute, Inc.
Principle Investigator: Dr. John Shingledecker, Mr. Kavarana Firdosh
National Lab Partner: Dr. Noah Paulson – Argonne National Laboratory

Summary: The proposed project seeks to leverage high performance computing (HPC) and active machine learning to apply state-of-the-art modeling and simulation tools to induction pipe bending of nickel-based alloys for energy applications. Induction bending offers significant improvements to the production of energy application piping systems, which enable highly efficient power cycles. However, the process has largely been ignored by the modeling community, and therefore the introduction of new piping alloys designed for high-temperature service require a trial-and-error experience-based approach. HPC offers the possibility of developing an accurate model of the non-symmetric 3-D temperature and strain profiles during heating, bending, cooling, and heat-treatment. Active machine learning can then efficiently construct an optimal surrogate model for the high-fidelity simulations, and therefore enable a large multi-variable assessment of the wide range of potential pipe sizes and process controls to develop scientifically sound approaches to enhance product quality and reduce overall energy intensity.

Cross-section of CMOS chip

Induction Bending Process (courtesy of Shaw)

Cross-section of CMOS chip

Integral Induction Bends for an Inconel 740H Advanced Energy Piping System

Optimizing Counter Currency and Improve Selective Gas Permeation

Industry Partner: Generon IGS
Principle Investigator: Mr. John Jensvold
National Lab Partner: Dr. Ramanan Sankaran – Oak Ridge National Laboratory

Summary: Several gas separation applications such as the removal of CO2 from natural gas require a highly efficient gas separation membrane device to purify the feed stream with minimal loss of methane or other light hydrocarbons and a minimal loss of natural gas pressure. In both ways, energy loss as well as green-house gas emissions are minimized. This is often carried out with shell-side fed hollow fiber membrane modules equipped with membranes that can selectively permeate CO2 from the feed gas by means of a partial pressure driving force across the membrane. To maximize this driving force, the preferred module design is counter-current in which the permeate gas runs counter to the feed gas that is being processed. We propose to develop a CFD model that enables optimizing the counter-current flow patterns in the module while minimizing pressure drop. The CFD model will use effective media models informed by fiber resolved direct numerical simulations and will be validated against experimental flow measurements.

Cross-section of magnetic field

High pressure shell-side fed membrane module for separating the feed gas to a preferential permeate product stream and a separate retentate gas. The inset image shows a typical hollow fiber membrane bundle packaged within a module.

Improving Additive Manufactured Component Performance

Industry Partner: General Motors LLC
Principle Investigator: Dr. Qigui Wang, Dr. Andy Wang
National Lab Partner: Dr. Alex Plothowski – Oak Ridge National Laboratory

Summary: The additive manufacturing (AM) process is increasingly used to manufacture critical automotive components because it provides significant advantages for production of unique and complex geometries. As an emerging and revolutionizing process, AM also faces a lot of challenges and uncertainty in product quality and performance. This HPC4EI project will utilize state-of-the-art ICME (Integrated Computational Materials Engineering) tools (e.g. ECP ExaAM) to develop a high-performance internal combustion engine component such as a piston through material design, microstructure control, and process optimization. The alloy composition will be tailored for AM for high temperature material property requirements based on previous research at ORNL and large-scale simulations. The unique microstructure and defect population in the AM part will be simulated and controlled using scalable melt-pool process modeling, and phase field modeling of the solidification microstructure, coupled with CALPHAD predictions of relevant free-energy relationships and defect predictions adopted from casting simulations that have been successfully developed at GM. Model validation will be carried out against experimental data available via existing ORNL programs. The validated model will be used to optimize the AM process for fabricating an aluminum piston, which will be engine tested within GM’s medium-duty truck engine light-weighting program.

Vector of open cell

Coupling scalable process and microstructure models to optimize fatigue performance of new high-performance aluminum alloys for additive manufacturing to increase performance and efficiency of automotive engines.

Development of Additive Manufacturing of Refractory Materials for Critical Applications

Industry Partner: Commonwealth Center for Advanced Manufacturing, University of Virginia, Siemens Corporation, Virginia State University
Principle Investigator: Dr. Yuri Plotnikov, Dr. Kaushik Joshi, Dr. Rich Martukanitz, Dr. Nasser Ghariban, Dr. Gaurav Ameta
National Lab Partner: Dr. Yousub Lee – Oak Ridge National Laboratory

Summary: The proposed program is focused on establishing computational framework, foundational knowledge, and additive manufacturing (AM) capabilities for accelerating the use of refractory metals for gas turbine generators, which is considered a significant enabling technology for increasing operating temperatures and improving efficiency of these systems. The program will develop and apply high-fidelity process and material models for simulation of potential defects, deposition geometry, and resultant microstructure of refractory alloys produced using directed energy deposition (DED) AM. Upon validation of the developed model, virtual and physical experiments will be designed and conducted to create process and material maps. The maps will assist in establishing quantitative relationships to define the influence of primary processing parameters on attributes used to delineate process consistency and product quality for meeting the stringent requirements for this industry. The developed models will be used to conduct significant virtual experimentation at the supercomputing facilities within Oak Ridge National Laboratory.

Schematic of a solid-state battery

Optimization of Processing Parameters for Metal Powder Production

Industry Partner: Praxair Surface Technologies
Principle Investigator: Dr. Andrew Heidloff
National Lab Partner: Dr. Iver Anderson – Ames Laboratory

Summary: Additive Manufacturing (AM) technologies are redefining next generation, energy critical component/system designs and manufacturing (e.g., stationary gas turbines, heat exchangers for extreme environments, etc.). The ability to produce complex geometries coupled with rapid development of new materials capable of harsh environments allows for unprecedented energy efficiencies through AM. Gas atomization (GA) is one of the most promising methods of producing feedstock powders used in AM processes, but suffers from inefficient powder yields and poor powder quality characteristics. This follow-on project aims to further develop the current understanding of breakup mechanisms during GA by using 2D and 3D computational fluid dynamics (CFD) to study the key variables leading to enhanced efficiency/ precision and powder quality. The results will aid US powder manufacturers in optimizing GA technologies to improve powder yield/quality, reduce material and energy production costs, and expedite the availability of novel and fully developed alloy powders for the AM marketplace.

Photo of VAST combustor test

Example of a 3D VOF simulation of the atomization process in a close-coupled gas die with several breakup mechanisms identified.

Round 11: Spring 2020 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

Ford Motor Company will partner with ORNL to improve part-scale modeling of laser powder bed fusion to improve car part quality and reduce scrap rate in a project titled "Extend an innovative HPC-Compatible Multiple Temporal-spatial Resolution Concurrent Finite Element Modeling Approach to Guide Laser Powder Bed Fusion Additive Manufacturing".

Raytheon Technologies Research Center (RTRC) will collaborate with ORNL to develop multi-physics and machine learning optimization algorithms to upscale MAP technology to an industrial level in a project titled "Multiphysics Models and Machine-learning Algorithms for Energy Efficient Carbon Fiber Production Using Microwave-assisted Plasma".

In collaboration with NREL, Futamura Group will accelerate development of next generation recyclable cellulose-based packaging materials in a project titled "In-Silico Design of Next Generation Cellulose-Derived Packaging Materials".

Raytheon Technologies Research Center (RTRC) and ORNL will address the need to optimize microwave-enhanced manufacturing of ceramic matrix composites in a project titled "Modeling Driven Manufacturing Process Intensification".

Machina Labs in collaboration with LLNL will perform informed aluminum sheet metal processing for bending and reducing springback for aerospace and automotive applications in a project titled "Advanced Machine Learning for Real-time Performance-informed Thermo-mechanical Processing of Sheet Metal Parts".

In a multi-lab partnership with ORNL and LLNL, Rolls-Royce Corporation will use HPC to study a key modeling component, heat transfer coefficients between the quench oil and solid-state components in the quench heat-treatment processes for gas turbine parts in a project titled "Nucleate Boiling of Quench Oils Used in the Heat Treatment of Critical Aerospace Components".

General Electric, GE Research will partner with ORNL to improve ceramic matrix composites for aviation by using advanced computational fluid dynamics and modern data analytics on HPC to rapidly develop a high-fidelity CVI kinetics model in a project titled "Data-driven Kinetics Modeling of Chemical Vapor Infiltration for Ceramic Matrix Composites Manufacturing".

The Procter & Gamble Company and SNL will collaborate to identify process parameters to efficiently and effectively utilize raw materials and for reducing energy consumption in the dewatering/drying of random foam & structured papers in a project titled "Highly-Scalable Multi-Physics Simulation for an Efficient Absorbent Structure".

Toyota Motor Engineering & Manufacturing North America will partner with LLNL to improve understanding of relationship between properties in specific solid electrolytes in a project titled "Multiscale Simulations of Novel Lithium Electrolytes for Improved Processability and Performance of Solid-state Batteries".

VAST Power Systems, Inc. ANL, and LLNL will continue their partnership in a Phase 2 project titled "Ultra-Clean Transient Turbine Combustor", which seeks to increase the number of simulations to improve the efficiency of VAST’s combustors for rapid transients in energy production.

Laser Powder Bed Fusion to Improve Car Part Quality

Industry Partner: Ford Motor Company
Principle Investigator: Dr. Mei Li and Dr. Yang Huo
National Lab Partner: Dr. Xiaohua Hu - Oak Ridge National Laboratory

Summary: Laser power bed fusing (L-PBF) additive manufacturing is a key enabling technology to manufacture highly complex and integrated automotive structures. L-PBF processes usually produce excessive and nonuniform residual stresses, which increase quality uncertainties and manufacture issues, leading to increases in cost and energy consumption in the form of rejected parts. We propose to extend an HPC-compatible in-house ORNL finite element (FE) code, which was demonstrated on pseudo-3D fully-coupled thermomechanical L-PBF simulations, to part scale and use it to predict temperature evolution and residual stress during L-PBF with experimental validation. The innovative multi-resolution and concurrent modeling approach adopted in this code ensures accuracy and computational efficiency, which will enable energy-efficient and high-yield, low-cost manufacturing of optimized, qualifiable automotive structures. The successful completion of this project will contribute towards reaching technical targets outlined in AMO’s Program Plan to develop additive manufacturing systems that deliver consistently reliable parts with predictable properties.

Cross-section of CMOS chip

Figure. (a) Graphical representation of a conventional HPDC Al structural node to connect longitude rails and cross beams; (b) Graphical representation of an AM structural nodes designed at Ford using DfAM digital tools, which has a weight of 24.6kg and 46% weight saving compared to conventional HPDC Al structural nodes, and stress analysis results, showing that AM technology can significantly advance automotive products.

Use of Machine Learning to Upscale MAP Technology

Industry Partner: Raytheon Technologies Research Center (RTRC)
Principle Investigator: Dr. Yehia F. Khalil and Dr. Vadim Yakovlev
National Lab Partner: Dr. Srdjan Simunovic, Dr. Merlin Theodore, Dr. Max L. Pasini - Oak Ridge National Laboratory

Summary: U.S. carbon fiber (CF) annual demand reached ≈73.1 million lbs/yr in 2020 and the primary energy intensity of PAN carbonization-step is ≈13.4 TBtu/yr. Using microwave-assisted plasma (MAP), ORNL demonstrated ≈45% energy savings, ≈67% reduction in residence-time, ≈40% reduction in CF production cost at small-scale, which can lead to energy savings of ≈1.7 TBtu/yr, based on 2010 current typical technologies. Building on ORNL work, this project aims to: (i) develop robust multi-physics model and machine-learning (ML) optimization algorithms to upscale MAP-carbonization to industrial levels and (ii) further optimize and validate techno-economic viability of MAP-based PAN carbonization. Structural light-weighting will benefit from advancing MAP-based technology in U.S.-manufacturing for energy-efficiency and it will positively impact the commercial aircraft manufacturing-supply-chain (which includes Raytheon Technologies), and the reduction of CF manufacturing-energy consumption. ORNL HPC capabilities and expertise are crucial to overcome key challenges in the computationally intensive optimization, testing, and validation of ML-driven MAP systems.

Cross-section of CMOS chip

Model development of a ML-driven microwave-assisted plasma (MAP).

Next Generation Recyclable Cellulose-based Packaging Materials

Industry Partner: Futamura Group
Principle Investigator: Kelly Williams
National Lab Partner: Dr. Peter Ciesielski, National Renewable Energy Laboratory

Summary: Fossil plastics in single-use packaging is one of the top existential problems in the world, and post-consumer collection of discarded materials continues to be elusive. Compostable packaging offers substantial energy savings relative to plastic packaging that requires recycling or upcycling in circular economy scenarios, and brands across the globe are seeking compostable options for flexible packaging. Cellulose, particularly dissolvable pulp that can be converted into high barrier packaging films, is currently in very high demand. We will leverage high-performance computing to accelerate evolution of art and science related to cellulose-derived films to meet societal demands and displace environmentally detrimental incumbent products. Specifically, molecular variations of cellulose dissolving pulp will be designed in-silico and their performance metrics, including mechanical, thermal, and barrier properties, will be predicted by large-scale simulation of polymer assemblies. The results will be used to identify production targets for next generation cellulose-based packaging materials to meet industry needs.

Cross-section of magnetic field

Snapshot from molecular dynamics simulations used to predict performance metrics for novel bio-based polymeric materials. Top: Simulation of O2 diffusion through a polymer matrix is used to estimate barrier properties. Bottom: Simulation of a water droplet on a polymer surface provides an estimate of the contact angle, a measure of hydrophobicity.

Microwave-enhanced Manufacturing of Ceramic Matrix Composites

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Dr. Ying She
National Lab Partner: Dr. Vimal Ramanuj, Dr. Wenjun Ge, Dr. Ramanan Sankaran, Oak Ridge National Laboratory

Summary: This project addresses the use of microwaves to intensify the manufacturing process of Ceramic Matrix Composites (CMCs) that enable light-weighting and energy efficiency improvements of gas turbines when deployed in the hot section. A conservative estimate is that CMCs can reduce thrust specific fuel consumption (TSFC) in commercial aerospace by ~2.5%, resulting in US annual nationwide energy savings of 113 TBTU. High Performance Computing (HPC) will be used to develop pore- and geometry-resolved modeling capabilities of an advanced Chemical Vapor Infiltration (CVI) process and corresponding reactor design. This will address the technical challenge of more uniform heating and temperature control, as required for manufacturing high-quality CMCs in a shorter manufacturing time. It will accelerate the development of CMCs for commercial aerospace and showcase HPC capabilities at Oak Ridge National Laboratory (ORNL).

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Microwave-assisted chemical vapor infiltration of a stationary article. Ribbons visualize the flow of reactive gases around article from left to right. The temperature distribution of the article (yellow=low and white=high) affects the resulting densification quality. The concentration of gaseous reaction products that get transported out of the article is illustrated in the blue background (blue=low and red=high).

Reducing Springback for Aerospace and Automotive Applications

Industry Partner: Machina Labs
Principle Investigator: Dr. Babak Raeisinia
National Lab Partner: Dr. Victor Castillo - Lawrence Livermore National Laboratory

Summary: The total onsite energy use for the Fabricated Metals (NAICS 332) sector in the U.S. is about 344 TBTU (with 11 MMT CO2-equiv of emissions) [1]. It is possible to reduce this energy and emissions footprint by ensuring that energy is only used when and where dictated by product performance needs. Uptake of such performance-informed processing strategies has been limited due to challenges in connecting product performance to processing parameters in real-time for control purposes. With current advancements in artificial intelligence and simulation capabilities, coupled with advanced sensors, it is now possible to overcome this challenge. This HPC effort is aimed at developing a lean, reduced-order model based on process simulation and sensor data to enable performance-informed thermo-mechanical processing of sheet metal parts. Broad adoption of such strategy across the industry would reduce the process energy of sheet metal parts, lead to development of novel products, while improving manufacturing yields.

Cross-section of CMOS chip

Scan of an incrementally formed aluminum alloy part, approximately 300 mm x 300 mm x 150 mm in size, depicting the redistribution of material across the part.

Quench Heat-treatment Processes for Gas Turbine Parts

Industry Partner: Rolls-Royce Corporation
Principle Investigator: Dr. Michael Glavicic and Dr. Chong Cha
National Lab Partner: Dr. Ramanan Sankaran (ORNL) and Dr. Ik Jang (LLNL)

Summary: To manufacture light-weight, advanced metal alloy components for gas turbine engines, quench heat-treatment processes are typically used. By quenching the component from elevated temperatures, the alloy sometimes undergoes a solid-state phase transformation which produces special microstructures with the required, enhanced mechanical properties. However, the quenching can also lead to cracks forming in the component. Addressing the quench cracking problems adds a significant burden to the cost, schedule, and energy demand of manufacture. Currently, optimizing the quench process to mitigate or avoid the cracking is performed largely by trial-and-error, relying heavily on costly experimental (thermocouple) trials to understand the local thermal gradients which cause the cracks to form. In this work, high-performance computing is employed to establish the ability of modern CFD (computational fluid dynamics) to alleviate or wholly replace the experimental quenching trials by virtual testing.

Cross-section of CMOS chip

Improvement of Ceramic Matrix Composites for Aviation

Industry Partner: General Electric, GE Research
Principle Investigator: Dr. Joseph Shiang
National Lab Partner: Dr. Dongwon, Shin, Oak Ridge National Laboratory

Summary: Current chemical vapor infiltration ceramic matrix composite (CVI-CMC) technology does not yet meet all requirements for commercialization in aircraft engines, in part due to the difficulty of optimizing CVI processes for batch scales and the significant capital expenses required. GE and ORNL will team to enhance ORNL’s recently demonstrated data-driven CVI simulation workflow (CVISim) by explicitly incorporating the complex chemical kinetics of the CVI process. This project will exploit high-throughput computational fluid dynamics (CFD) and modern data analytics on HPC to rapidly develop a high-fidelity CVI kinetics model. Project success will enable accurate physics, data-based forecasting of advanced processing costs, and description of the operational performance of the CVI process prior to capital equipment acquisition, simultaneously reducing scale-up risk and accelerating commercialization. Enabling the introduction of a CVI-CMC material system to both aircraft engines and land-based turbines is expected to result in significant fuel consumption reductions.

Cross-section of magnetic field

Multi-Physics Simulation for an Efficient Absorbent Structure

Industry Partner: The Procter & Gamble Company
Principle Investigator: Dr. Mel Allende and Dr. Ken Comer
National Lab Partner: Dr. Scott Roberts, Sandia National Laboratories

Summary: Open Cell Foams (Random Foams) manufacturing, as well as papermaking, is a highly energy intensive manufacturing process. Tremendous amounts of energy can be saved if the microstructures can be designed and optimized for dewatering/drying while maintaining a desirable consumer experience.

The objectives are: To utilize a model-based approach to predict the process parameters required to efficiently and effectively utilize raw materials while also reducing energy consumption in the dewatering/drying of random foam & structured papers while generating a final product which is consumer preferred. Meeting these objectives requires optimizing a truly multi-physics problem.

In this project, The Procter & Gamble Company (P&G) will use codes developed by Sandia National Laboratories to represent the needed multi-physics with high HPC scalability. These codes will then enable P&G to design and optimize foam/fiber structures that meet the consumer needs and require much less energy and cost to manufacture.

Vector of open cell

Feminine hygiene product with open cell foam as absorbent material.

zoom of open cell Vector of open cell foam structure

Micro-CT of open cell foam structure (absorbent material).

New class of Li-ion solid-state electrolytes

Industry Partner: Toyota Motor Engineering & Manufacturing North America
Principle Investigator: Dr. Rana Mohtadi
National Lab Partner:

Summary: Electrochemical energy storage technologies that are durable, efficient, energy dense, cheap, safe, and industrially scalable are highly demanded by a wide range of applications. Solid-state battery technologies are promising in this regard, but they remain challenged by difficulties in simultaneously achieving energy-efficient processability, mechanical durability, and efficient performance of manufactured electrolyte components. Toyota Research Institute of North America has developed a new class of Li-ion solid-state electrolytes that promise highly efficient performance and easier processability and therefore are expected to enable practical production of solid-state batteries. However, optimizing processing requires understanding the critical connection between mechanical robustness, ionic transport, and thermodynamic properties, which is very challenging utilizing available experimental tools due to the high levels of structural complexity. This project integrates experiments with a multiscale modeling approach that can offer the necessary insights to advance this area and accelerate the deployment of practical and easily processible solid-state batteries.

Schematic of a solid-state battery

Schematic of a solid-state battery electrolyte, which typically requires energy-intensive processing to achieve sufficient ionic conductivity while maintaining mechanical integrity and suitable operating temperature.

Ultra-Clean Transient Turbine Combustor

Industry Partner: VAST Power Systems, Inc.
Principle Investigator: Dr. David L. Hagen, Dr. Gary Ginter, Dr. John N. O’Halloran
National Lab Partner: Dr. A. Cody Nunno (ANL), Dr. Prithwish Kundu (ANL), and Dr. Victor Castillo (LLNL)

Summary: This research furthers VAST® TriFluid™ combustor and VAST Power Cycle™ design optimization for ~70% higher net power through a single expander, and ~24% better single turbine efficiency, with NOx and CO emissions below mandates, without catalysts, or ammonia.

To prevent state-wide blackouts from large wind/solar dropouts, California requires rapid 10-minute and 5-minute dispatch Peaker turbines, and 1-minute emergency dispatch. Frequent ramping severely harms turbines, increasing replacement costs. Clean air emission mandates cause high catalyst expenses. Emission control is difficult during rapid turbine startups, for pilot flames, and hydrogen combustion.

VAST® FastRamp™ turbines offer higher profitability with faster dispatch over >5% to <50% capacity use with renewable energy constraints. VAST’s patented independent temperature control minimizes cyclic fatigue, improving relative operating life. Accurate temperature control extends blade life. FastRamp turbines enable >40% US renewable grid penetration and international deployment. They create a profitable new niche between peakers and constrained combined cycle turbines.

Photo of VAST combustor test

VAST combustor test with sub ppm NOx and CO.

Round 10: Winter 2020 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

CHZ Technologies, LLC will partner with NREL to use HPC to deepen understanding of material transport, heat transfer, phase-change, and chemistry in the Thermolyzer™ technology that converts waste hydrocarbon materials into fuel gas and saleable byproducts in a project titled "Simulation of Complex Reacting Media in Multidimensional Reaction Chamber.

Raytheon Technologies Research Center (RTRC) will partner with ORNL to use HPC based phase-field simulations along with experimental validation to design novel Ti alloy compositions for AM to potentially replace currently-used wrought Ti alloys in a project titled Development of HPC Based Phase Field Simulation Tool for Modification of Alloy Morphology to Enhance Material Properties During Additive Manufacturing (AM) Process

Materials Sciences LLC will partner with LLNL to combine recent advances in topology optimization-based design, high performance computing (HPC), and additive manufacturing (AM) technology to develop high pressure and temperature heat exchangers in a project titled "HPC-Enabled Optimization of High Temperature Heat Exchangers.

ESI North America, Inc will partner with PNNL to use HPC resources to develop a data driven approach to link features of the material and manufacturing processes to the mechanical properties of thermoplastic composite parts in a project titled ""Development of Efficient Process for Manufacturing of Thermoplastic Composites with Tailored Properties.

Efficient Larger-scale Thermolyzer Systems

Industry Partner: CHZ Technologies, LLC
Principle Investigator: Dr. Henry W. Brandhorst, Jr. - CHZ Technologies, LLC
National Lab Partner: Dr. Hariswaran Sitaraman, Dr. Shashank Yellapantula, Dr. Vivek Bharadwaj, Dr. Marc Henry de Frahan - National Renewable Energy Laboratory

Summary: Thermolyzer™ is the only technology that can convert all waste hydrocarbon materials cleanly and safely into a fuel gas and salable byproducts. This means that tons of plastics now in storehouses can be converted into energy, thereby conserving non-renewable fossil fuels. The impact on the U.S. economy can be huge. However, pyrolysis of plastics is a complex process. The feedstock material that is of high variability is continuously gasified creating multiple species as it gets converted to a complex synthesis gas and carbon. The geometry and temperature gradients within the reactor are also complex. Thus, computational modeling of the reactor using high performance computing is essential in order to understand the physico-chemical interactions and to derive the best operating conditions for maximum efficiency. This project will provide the capability to achieve efficient larger-scale Thermolyzer systems (~200 ton/day capacity) that can significantly reduce the backlog of scrap plastics in the US.

Photo of chip

Image of 7 ton/day R&D Thermolyzer™

Titanium Alloy Development for Additive Manufacturing

Industry Partner: Raytheon Technologies Research Center (RTRC)
Principle Investigator: Dr. Tahany El-Wardany, Dr. Ranadip Acharya - Raytheon Technologies Research Center (RTRC)
National Lab Partner: Radhakrishnan Balasubramaniam, Ph.D. - Oak Ridge National Laboratory

Summary: Raytheon Technologies Research Center (RTRC) in collaboration with ORNL proposes use of model-based tools to design alloys for additive manufacturing (AM) in order to obtain as-desired microstructure for performance improvement in aerospace and automotive applications. The performance and cost of AM products still controls the business value of deploying AM to replace conventional manufacturing processes. The digital benefit of digitally designing a component and rapidly manufacturing it through AM is often lost due to extensive experimental iterations to remedy poor performance of fabricated components. The lack of performance is often attributed to intrinsic defects formation and undesirable microstructural features since the alloy composition and microstructure are not designed optimally for the given application. RTRC and ORNL will use HPC based phase-field simulations along with experimental validation to design novel Ti alloy compositions based on forming fine equiaxed grains during AM to potentially replace currently used wrought Ti alloys.

Cross-section of CMOS chip Graph of CMOS chip

Faster Heat Conduction Using Advanced Heat Exchanger (HEX) Designs

Industry Partner: Materials Sciences LLC
Principle Investigator: Mr. Devlin Hayduke - Materials Sciences LLC
National Lab Partner: Dr. Boyan Lazarov - Lawrence Livermore National Laboratory

Summary: The Project Team proposes to combine recent advances in topology optimization-based design, high performance computing (HPC), and additive manufacturing (AM) technology to develop high pressure and temperature heat exchangers (HEX) concepts with greater than 85% effectiveness and a 50% reduction in volume in order to overcome the current design and economic limitations of conventional manufacturing methods. If realized, this technology could provide significant energy savings for power generation, aviation, and space industries.

Cross-section of magnetic field

Figure F1: Example of 3D printed topology optimized multi-scale beam [Materials Sciences LLC]

Cross-section of magnetic field

Figure F2: Example of 3D printed jet impingement cooling solution [University of Pittsburgh, United Technologies Corp.]

Link Processes to Properties in Thermoplastic Composite Manufacturing via Machine Learning (ML)

Industry Partner: ESI North America, Inc.
Principle Investigator: Dr. Ravi Raveendra - ESI North America, Inc.
National Lab Partner: Dr. Ram Devanathan - Pacific Northwest National Laboratory

Summary: This HPC4EI proposal seeks to develop a data driven approach to link features of the material and manufacturing processes to the mechanical properties of thermoplastic composite parts. This work will leverage data from physics-based commercial codes for manufacturing simulation and micromechanical analysis. There is a need to develop and manufacture lightweight materials with enhanced performance to improve the energy efficiency of automobiles. With outstanding strength to weight ratio, good fatigue resistance and good corrosion/fire resistance, composite materials are well positioned to meet the lightweight challenge. However, computational tools are needed to develop composites with enhanced performance given the large number of parameters that can be tuned to improve the performance. The proposed work will use high performance computing (HPC) and data analytics to optimize the design, shorten the time to market and generate reduced order models that are ultimately usable by U.S. industry without the need for HPC resources.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Round 9: Fall 2019 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

OxEon Energy, LLC will partner with LLNL to reduce the number of reactor tubes in Fischer Tropsch (FT) reactors in order to lower cost and increase performance in a project titled, "Topology Optimization of Fischer Tropsch Reactor Design for Synthetic Fuel Production".

Flawless Photonics and LLNL will simulate both the heated glass flow and nucleation and growth of crystal nuclei to find the drawing conditions that suppress the growth of light scattering crystalline defects in ZBLAN in a project titled, "Modeling and Simulation of the Manufacture of a Superior Fiber Optic Glass".

3M Company will continue a partnership with SNL for a follow-on project to enhance non-equilibrium thermal radiation computation capability in a multi-physics framework of a passive cooling installation on a project titled, "Passive Cooling Film Optimization".

NLMK USA will partner with ORNL to use computational fluid dynamics (CFD) methodology to optimize scrap melting using the electric arc in electric arc furnaces in a project titled, "Optimization of Scrap Melting Using an Electric Arc in Steel Manufacturing".

PPG Industries will partner with LBNL to use high performance computing in the modeling of the paint drying process to enable energy savings through co-curing in a project titled, "Modeling Coating Flow and Dynamics During Drying".

Raytheon Technologies Research Center will partner with ANL to predict flow and heat transfer characteristics of cooling air in gas turbine hot section combustion liners in order to increase operating efficiency and reduce fuel consumption in a project titled, "Pseudo-Spectral Method for Conjugate Heat Transfer Prediction of Impinging Flows Over Rough Surfaces". The Raytheon Technologies Research Center project was selected by both the HPC4Mfg and HPC4Mtls programs and will be co-funded by the Advanced Manufacturing Office in the Department of Energy's Energy Efficiency and Renewable Energy Office and the Office of Fossil Energy.

ArcelorMittal USA and ORNL will collaborate to reduce the yield loss caused by inclusions forming in the refining ladle process in a project titled, "Reduced Order Modeling and Performance Prediction for Steel Refining Ladle Processing via HPC".

ArcelorMittal, LLNL, and ANL will work together to develop next generation lightweight advanced high strength steels with the help of HPC and artificial intelligence to positively impact the U.S. energy landscape during both production and use in a project titled, "Ab-initio Guided Design and Materials Informatics for Accelerated Product Development of Next Generation Advanced High Strength Steels (AHSS)".

Guardian Glass, LLC and LLNL will collaborate to reduce energy consumption in glass making by using CFD simulations and Machine Learning in a project titled, "Rapid CFD Using Machine Learning Algorithms".

Optimization of Fisher Tropsch Reactor Design

Industry Partner: OxEon Energy, LLC
Principle Investigator: Joseph Hartvigsen, Dr. S. Elangovan, and Michele Hollist – OxEon Energy, LLC
National Lab Partner: Victor Beck, Lawrence Livermore National Laboratory

Summary: The proposed work will use coupled thermal-fluid analysis to drive the topology optimization design of an internal thermal management structure, enabling FT reactors to be designed with fewer, larger reactor tubes. Utilizing computational topology optimization, the interior structure of reactor tubes and catalyst supports will be optimized to enable effective heat removal, achieve lower cost, improve volumetric efficiency, and increase catalyst lifetime. The envisioned design will allow the use of reactor tubes that are more than five times the diameter of those in present commercial reactors, enabling significant energy and cost savings.

By lowering the manufacturing cost and improving economics of small, transportable, distributed FT plants, valuable liquid hydrocarbons can be produced from gas resources that would otherwise be flared. Energy lost to flaring in the US amounted to 0.48 quads in 2018, equivalent to 9.3% of residential gas use and 1.5% of total gas use.

Cross-section of CMOS chip

OxEon 2 barrel per day gas to liquid (GTL) pilot plant.

Cross-section of CMOS chip

OxEon 2nd generation thermal management structure designed using model driven geometric optimization as an illustration of the approach being proposed for a 3rd generation design.

Superior Fiber Optic Glass

Industry Partner: Flawless Photonics
Principle Investigator: Michael Vestal, Flawless Photonics
National Lab Partner: Chris Walton and Tomorr Haxhimali, Lawrence Livermore National Laboratory

Summary: Heavy metal fluoride glasses (HMFG) can replace silica fiber as the telecommunications and photonics material of choice because of superior (near "theoretical absolute") optical transparency. ZBLAN, a well-characterized and studied HMFG, has proven superior mid-IR transparency and lower intrinsic loss with the potential for 20 times less attenuation than industry-standard silica. These performance improvements can result in longer telecom transmission before amplification with the potential to save ~20 TBTUs annually. The challenge is in manufacturing – creating long fiber without crystalline optical defects. Flawless Photonics has demonstrated making clear fiber onboard the ISS using our microgravity drawing platform, but to fully and rapidly exploit future results requires an accurate and multiscale model. This project will build such a model, simulating both the softened glass flow and nucleation and growth of crystal nuclei, to find the drawing conditions that suppress the growth of light scattering crystalline defects.

Passive Cooling Film Optimization (Phase II)

Industry Partner: 3M Company
Principle Investigator: Dr. Robert Secor and Dr. Chris Pommer, 3M Company
National Lab Partner: Dr. Rekha Rao, Sandia National Laboratory

Summary: Electrical power for cooling systems account for about 3% of the 33 gigatons of global energy-related CO2 emissions. In addition, these systems use about 32 trillion liters of evaporated water annually. Passive cooling, which requires no power, is a new way to dramatically reduce cooling energy and freshwater usage. Passive cooling films combine solar heat rejection with thermal radiation to outer space and the upper atmosphere to produce a sub-ambient heat sink. The challenge for passive cooling commercialization is to improve the cost-competitiveness by optimizing the emissive film structures for functionality as well as low-cost, large-scale manufacturing. Team will use high performance computing to maximize functional thermal emission and reduce manufacturing costs of this composite metamaterial thin film.

Cross-section of magnetic field

Figure F1: Example of 3D printed topology optimized multi-scale beam [Materials Sciences LLC]

Optimization of Scrap Melting Using an Electric Arc

Industry Partner: NLMK USA
Principle Investigator: Harold Kincaid, NLMK USA and Dr. Chenn Zhou, Purdue University
National Lab Partner: Dr. Hong Wang, Oak Ridge National Laboratory

Summary: Steel industry is crucial to the national economy and security. Around 68% of crude steel in the U.S is produced in electric arc furnaces (EAF), which is energy intensive. Around 140 EAFs operate in the U.S., consuming about 9.2x10^7 MMBtu/year of electricity. One of major challenges for EAFs includes maximizing the efficiency of the electrical energy provided in the form of electric arcs to melt various scrap mixes. To address this issue, a computational fluid dynamics (CFD) methodology is chosen to optimize scrap melting using the electric arc. Due to complex furnace phenomena and the wide variety of potential scenarios, high performance computing (HPC) is essential to yield comprehensive and detailed CFD analyses and systematic parametric studies for optimized EAF operation. The objectives are to 1) simulate scrap melting using electric arc, and 2) evaluate electrode/arc position for optimum scrap melting. This project has the potential to save $65 million/year on energy usage with improved EAF productivity.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

EAF at NLMK Indiana in Portage, IN.

Modeling Coating Flow and Dynamics During Paint Drying Process

Industry Partner: PPG Industries, Inc
Principle Investigator: Dr. Xinyu Lu and Dr. Reza Rock, PPG Industries, Inc.
National Lab Partner: Dr. Robert Saye and Dr. James Sethian, Lawrence Berkeley National Laboratory

Summary: Achieving a smooth, defect-free film is often the biggest technical hurdle to commercializing energy-efficient coating systems. We propose to use high performance computing to model the complex physics driving flow and leveling in a two-layer coating system as solvents evaporate, during film formation, and ultimately as the film cures. This model will provide new insights to develop coating formulas that can be co-cured with a single, lower temperature bake. Beyond the energy savings achieved from fewer cure steps, less energy consumption, and faster process times, the proposed work provides a foundation for future models for water-based coatings and lightweight substrates such as carbon-fiber reinforced composites. The U.S. consumes 100-200 TBtus annually applying and curing paint.1 Energy-efficient paint lines that co-cure layers have been demonstrated to reduce energy consumption by up to 30%.2.

3D VOF simulation

Film smoothness and metallic color are critical to manufacturers adopting energy-efficient coating systems. It is so finely controlled that the crispness of the left side of the image below is acceptable while the right side is not. In this project, we model how two-layer paint systems flow and level as they dry and cure. The new model will help formulators and manufacturers understand and quickly control the fundamental processes driving film smoothness and pigment orientation.

Heat Transfer Prediction of Cooling Air in Gas Turbine

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Dr. Miad Yazdani and Dr. Peter Cocks, Raytheon Technologies Research Center
National Lab Partner: Dr. Mushin Ameen, Argonne National Laboratory

Summary: Gas turbine hot section components leverage cooling air for thermal management. Methods for thermal management can introduce complex flow structures (e.g., impinging flow) and can be substantially impacted by surface roughness. Surface roughness is aggravated by modern manufacturing techniques and materials, such as additive and ceramic matrix composites. This proposal aims to develop a first-principles based simulation framework for predicting flow and heat transfer characteristics of cooling air in gas turbine hot section components, including surface roughness effects and with specific focus on combustor liners. Accurate prediction is crucial in order to successfully develop the advanced cooling concepts required to operate gas turbine engines more efficiently at higher pressures and temperatures. A multiscale approach for capturing the impact of surface roughness on flow and thermal characteristics at relevant operating and flow conditions will be developed. This proposal would leverage world-class simulation tools and expertise at Raytheon Technologies Research Center and Argonne National Lab.

3D VOF simulation

Impinging jet cooling architecture, along with example surface roughness and notional expectation of surface heat transfer characteristics (and its distinct features compared to smooth surface counterpart)

Improving Steel Refining Ladle Processing

Industry Partner: ArcelorMittal USA
Principle Investigator: David White, ArcelorMittal, and Dr. Chenn Zhou, Purdue University
National Lab Partner: Dr. Hong Wang, Oak Ridge National Laboratory

Summary: Representing over $520 billion in economic output and 7% of the total energy used in the United States manufacturing sector, the U.S. steel industry is an energy-intensive yet critical component of the domestic and global economy. With over 5% of produced steel ending up rejected and reprocessed as scrap, inefficiencies and production faults represent an enormous energy and financial burden. A critical stage to prevent product rejection is the refining ladle process. However, the steel refining in a ladle is a nonlinear dynamic and stochastic process. To address this issue, numerical simulation is used to model ladle operation, with HPC being necessary due to the computational requirements to simulate details accurately. Furthermore, reduced order modeling using simulation and industry data allows for real-time process control improvements. Reduction of product rejection due to improper refining and improvement of processing time can save an estimated $50.7 million annually.

3D VOF simulation

Steel refining ladle seen from above

Development of Next Generation Advanced High Strength Steels (AHSS)

Industry Partner: ArcelorMittal USA, LLC
Principle Investigator: Brian Lin, ArcelorMittal, Dr. Aditi Datta and Dr. Chen Zhou, Purdue University
National Lab Partner: Dr Sylvie Aubry and Dr. Amit Samanta, Lawrence Livermore National Laboratory, and Dr. Prasanna Balaprakash, Argonne National Laboratory

Summary: Transportation accounted for approximately 28% of the total energy consumed in the U.S. in 2010 (JOM, Vol. 64, No. 9, 2012, p. 1032). The iron and steel industry is the fourth largest energy-consuming industry in the U.S. About 80% of the total energy consumed in the steel industry is used to produce liquid steel, therefore, producing fewer tons of steel by developing lightweight high strength steel would lead to a huge benefit in terms of energy savings and reduction of CO2 emissions during production phase of the steel. Therefore, accelerated development of next generation lightweight advanced high strength steels (AHSS) with the help of HPC and artificial intelligence (AI) would positively impact the U.S. energy landscape during both production and use phase.

3D VOF simulation

Choudhury et al. J. Engg. Mat. Tech., Vol. 140, p 020801 April 2018

3D VOF simulation

LLNL

3D VOF simulation

LLNL

Reducing Energy Consumption in Glass Making

Industry Partner: Guardian Glass, LLC
Principle Investigator: Dr. Yousef Mohassab, Guardian Glass, LLC
National Lab Partner: Dr. Vic Castillo, Lawrence Livermore National Laboratory

Summary: Flat glass “melting” is an energy intensive continuous process involving several processes that occur in two chambers prior to forming of a glass ribbon. CFD simulations of production glass furnaces currently take two or more weeks to process with a supercomputer. With the support of Lawrence Livermore National Laboratory, Guardian Glass proposes that CFD simulations to train a machine learning algorithm that can achieve comparable accuracy to the CFD simulation in seconds on a standard desktop computer in a production facility. In doing so, a reduction of 5% or more in U.S. float glass energy consumption can be possible through inline optimization and real-time control of the glass making process. This would result in savings of roughly 2.5 million GJ per year from natural gas. The associated cost savings are crucial for maintaining global competitiveness of U.S. glass manufacturing while also reducing the environmental impact of evolved CO2.

3D VOF simulation

Chargers feed raw materials to furnace where glass reaches temperature above 1500 °C.

3D VOF simulation

Temperature distribution and batch concentration profile inside the furnace

Round 8: Spring 2019 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

Raytheon Technologies Research Center will partner with ANL to develop innovative and affordable machine learning enabled high-fidelity flow-physics models to be used in the design cycle of a gas turbine engine in a project titled “Deep Learning-Augmented Flow Solver to Improve the Design of Gas-Turbine Engines”.

General Motors LLC (GM) will partner with ORNL to develop residual stress model for laser-welded dissimilar joints (HSLA/CE steel) for car light-weighting in a project titled “Simulation Tools for Characterizing Stress Distribution in Laser Welded Dissimilar Joints”.

UTRC in collaboration with SNL will aim to develop a first-principles based simulation framework for predicting deposition of dirt, sand, volcanic ash and other particulates on aero-engine components operating in polluted urban environments in a project titled “Fully-Resolved DNS Simulation of Particulate Deposition for Aeroengine Combustor Applications”.

Praxair, Inc. will partner with ORNL to develop a multi-physics 3D CFD model of an Oxygen Transport Membrane (OTM) reactor module in a project titled “High Performance Computing for Improvement of Syngas Production Efficiency with OTM Technology."

Dow Chemical Company will partner with NREL to model how flow in plastic impacts polymers at a molecular level in a project titled “Non-equilibrium Molecular Simulations of Polymers under Flow: Saving Energy through Process Optimization”.

Owens Corning & Saint Gobain Ceramics & Plastics., DBA SEFPRO and LLNL will collaborate to optimize the operating conditions in the glass manufacturing process in a project titled “Spectral Radiative Modeling of Glass Furnaces”.

Raytheon Technologies Research Center and LANL will collaborate on developing a multiscale model to predict the mechanical behavior of additively manufactured components, particularly for creep applications in a project titled “Integrated Predictive Tools for Property Prediction in Additive Manufacturing”.

Reduce Fuel Usage in Gas Turbine Engines

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Michael Joy, Raytheon Technologies Research Center
National Lab Partner: Dr. Pinaki Pal, Dr. Prithwish Kundu, Dr. Opeoluwa Owoyele, Argonne National Laboratory

Summary: Project aims to develop innovative and affordable machine learning enabled higher fidelity flow-physics models that can be used in the design cycle of a gas turbine engine to develop next generation energy-efficient technologies. Specifically, models for near-wall cooling flow physics are needed to optimize combustor and turbine cooling designs for modern engines that operate at very high pressures and temperatures. Reductions in cooling air flow would improve the thermal efficiency and hence the overall efficiency of the engine. The primary objective is to leverage recent advances in deep learning techniques and supercomputing to develop predictive but affordable models for near-wall mixing and heat transfer. The novelty of our proposal is in the development and application of machine learning-based spatial emulators that can predict a spatially varying flow field quantity in an engineering fidelity simulation of a combustor or turbine blade at a reduced cost.

Cross-section of CMOS chip

Contour plot of velocity magnitude (normalized by the inflow jet velocity) on a vertical cut plane from a Nek5000 3D wall resolved LES of a jet-in-crossflow configuration with 40 million cells. The simulation was run on 1900 processors.

Cross-section of CMOS chip

Contour plot of static temperature on a vertical cut plane from a Fluent 3D hybrid RANS-LES of effusion cooling flow with wall conduction (ASME-GT2019-91423).

Optimizing Laser Welding Process

Industry Partner: General Motors LLC (GM)
Principle Investigator: Dr. Liang (Andy) Wang
National Lab Partner: Dr. Zhili Feng, Oak Ridge National Laboratory

The automotive industry has increasing needs of joining dissimilar materials by high-power (> 4 kW) laser welding to achieve narrow-gap deep penetration (> 4mm) for the rotating parts in the powertrain components. The benefits of dissimilar material welding include weight reduction such as using new joint design by replacing bolt joint with a laser weld. Currently, welding residual stresses in dissimilar welded materials are rarely studied in both experiment and modeling aspects. The project aims to take advantage of massively parallel thermo-metallurgical-mechanical simulation to accurately predict the microstructural evolution and residual stress in laser welding between dissimilar metals, such as high strength low alloy (HSLA) steel and high carbon equivalent (CE) gear steel. The simulation tool will help to optimize the laser welding process to minimize the weld residual stress, resulting in weld quality improvement and production increase, as well as energy savings by reducing experimental trial-and-errors.

typical secondary lead reverberatory furnace

Figure. Schematic and FEM modeling results of a large-diameter dissimilar material weld.

Prediction of Particulates on Aero-engine Components

Industry Partner: UTRC
Principle Investigator: Miad Yazdani, UTRC
National Lab Partner: Jackie Chen, Sandia National Laboratories

Summary: Project aims to develop a first principles based simulation framework for predicting deposition of dirt, sand, volcanic ash and other particulates on aero-engine components operating in polluted urban environments. Such deposition increases the cooling air needs for thermal management which reduces overall efficiency and increasing fuel consumption. Models developed in this proposal will provide physical insights needed to develop mitigation concepts for engines to operate at higher pressures and temperatures with commensurate improvement in energy efficiency. This is a first-of-its-kind attempt to develop a multiscale approach for capturing the physics of deposition process from sub-microscopic scales to the engine component scale including physics of particle-surface interactions, particle-surface morphology, composition effects on sticking and surface reactions. This project would leverage UTRC and Sandia National Lab’s world-class simulation tools-expertise to develop a unique capability for predicting particulate ingestion, deposition and the resultant thermal penalty that leads to reduced energy efficiency.

Cross-section of magnetic field

3D CFD Model of an Oxygen Transport Membrane (OTM)

Industry Partner: Praxair, Inc., a member of the Linde group
Principle Investigator: Jamie Wilson, Praxair, Inc.
National Lab Partner: Prashant K. Jain, Oak Ridge National Laboratory

Summary: Chemical renaissance in the Gulf coast is driven by the combination of readily available, inexpensive natural gas in the US and the high demand for its chemical derivatives in the global marketplace. More energy efficient processes will improve profitability and sustainability of natural gas conversion technologies. Praxair’s high temperature, ceramic based combined reformer technology, referred to as Oxygen Transport Membrane (OTM) combined reformer, offers a significant advantage over conventional technologies, such as, steam methane and autothermal reforming for production of methanol, gas-to-liquids, carbon monoxide and hydrogen. For example, 20% reduction in energy consumption and 80% reduction in carbon dioxide emissions can be achieved for methanol production. Although, the OTM technology has been successfully demonstrated at the pilot scale, robustness during transient operation needs improvement. Transient simulations of the physics underlying the OTM reactor are expected to assist in optimization of thermal integration and management of the associated thermal stresses.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Fig. D1. Praxair’s combined reforming with OTM panel array technology.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Fig. D2. Combined reforming in a single integrated efficient package.

Nonequilibrium MD of Polymers Under Flow

Industry Partner: Dow Chemical Company
Principle Investigator: Kurt Koppi, Ph.D.
National Lab Partner: Michael Crowley, Ph. D, National Renewable Energy Laboratory

Summary: Critically important manufacturing of parts from plastics include the operations of extrusion, film blowing, curtain coating, injection molding, fiber spinning, and 3D-printing. When these processes are performed industrially, deformation rates are very large - the polymer molecules flow under extreme conditions. This project combines DOE HPC capabilities with established and emerging molecular simulation techniques to address industrially relevant issues associated with polymers under flow conditions of practical interest to US Manufacturing competitiveness.

3D VOF simulation

Visual representation of a single polymer chain embedded in other chains in the molten state. (A.) A single chain under quiescent conditions (B.) A single chain subject to flow shown stretching out.

Optimization of Operating Conditions for Glass Manufacturing Processes

Industry Partner: Owens Corning & Saint Gobain Ceramics & Plastics., DBA SEFPRO
Principle Investigator: Bruno A. Purnode, Owens Corning; Kristen Pappacena, Saint Gobain Ceramics & Plastics., DBA SEFPRO
National Lab Partner: Hai P. Le, Lawrence Livermore National Laboratory

Summary: Refractory materials are used in E-glass melting furnaces thanks to their ability to withstand extremely high-temperatures (>1600°C). In the presence of natural gas combustion, batch materials chemically react, melt and become glass. Modern computational fluid dynamics (CFD) simulations are used to simulate combustion, and while all three modes of heat transfer are important in melting furnaces, improving radiation heat transfer calculations is the primary focus of this project. Radiation is currently considered using gray body approximations, in order to limit computational cost. However, non-gray (i.e., spectral) characteristics are important, especially for radiative exchange occurring at refractory surfaces. Using HPC, the team hopes to demonstrate a way to increase the fidelity of our simulation models by running multi-band radiation models, in massively parallel CFD calculations. It is expected that this project will lead to more rapid optimization of operating conditions for glass manufacturing processes, as well other energy intensive industries in the U.S.

3D VOF simulation 3D VOF simulation

Combustion flames in a glass furnace along with CFD mode.

Develop Multiscale Model to Predict Mechanical Behavior of Additive Manufacturing

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Dr. Ranadip Acharya, UTRC
National Lab Partner: Laurent Capolungo and M. Arul Kumar, Los Alamos National Laboratory

Summary: Sparsity in mechanical characterization and the limited understanding of microstructure effects on mechanical response significantly hinder the certification of Additively Manufactured (AM) components, particularly for creep applications. This can be remedied with use of highly predictive microstructure sensitive models. It is proposed to derive a computationally efficient and predictive engineering model that can capture the role of point defect diffusion on plasticity of AM processed IN718. The strategy is to develop a high-fidelity model predicting the material’s mechanical response and microstructure changes as function of stress and temperature and to use this model to modify/verify an engineering model used for design applications. The high-fidelity model will rely on LANL’s existing crystal plasticity Fast Fourier Transform framework while the engineering model will be based on UTRC’s existing framework. Information transfer between the high fidelity and the engineering model will be performed with data analytics, thereby requiring the use of high-performance computers.

3D VOF simulation

Schematic of proposed program combining effect of variable AM microstructures in crystal plasticity framework and surrogate model development through machine learning.

Round 7: Fall 2018 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded federal funding for public/private projects aimed at solving key manufacturing challenges.

The HPC4Mfg Program expands its impact with new awards:

Ferric, Inc. will partner with LBNL to develop analytical tools that combine traditional electromagnetic finite-element analysis with micromagnetic simulation in a project titled “Combined Micromagnetic and Finite-element simulation of Integrated Magnetic Inductors for Improve DC-DC Voltage Regulator Energy Efficiency and Manufacturing Yield”.

Gopher will partner with Gas Technology Institute (GTI) and ORNL to develop a high-fidelity computational fluid dynamics (CFD) model of a directly fired reverberatory style secondary lead furnace in a project titled “High-Performance Computing to Increase Productivity of Secondary Lead Furnaces”.

Applied Materials will partner with LLNL for a Phase II project to continue development of predictive modeling capabilities for the advanced film deposition technique, High Power Impulse Magnetron Sputtering (HiPIMS), in a project titled “Modeling High Impulse Magnetron Sputtering (HiPIMS) plasma sources at reactor scale for reactive Physical Vapor Deposition (PVD) processes used in fabrication of high efficiency LEDs and solid-state non-volatile energy efficient storage class memory devices”.

Eastman will partner with ANL to optimize the non-Newtonian slurry atomizers for large-scale energy production facilities in a project titled “Simulation-driven optimization of non-Newtonian slurry atomizers for large-scale energy production”.

Praxair Surface Technologies, Inc. will partner with Ames Laboratory to enhance the efficiency/precision and the quality of metal powder production in project titled “CFD Simulations of Metal Powder Production by Gas Atomization”.

Integrated Voltage Regulators

Industry Partner: Ferric, Inc.
Principle Investigator: Michael Lekas, Ferric, Inc.
National Lab Partner: Peter Nugent, Lawrence Berkeley National Laboratory

Summary: In the U.S., data centers already represent 1.8% of total energy consumption, and this figure continues to grow at an annualized rate of approximately 4%. To combat this trend, electronics manufacturers are adopting novel DC-DC power converter technology, such as the integrated voltage regulators (IVRs) developed by Ferric. Ferric’s IVRs employ thin-film ferromagnetic inductors to improve energy efficiency and reduce size relative to traditional solutions. Accurate modeling and optimization of these magnetic devices in conjunction with integrated circuits is paramount for achieving high conversion efficiency (η) in these products. However, thin-film magnetics simulation remains computationally expensive, and is a primary obstacle to reducing design-cycle time and further optimizing converter products. This proposal outlines the development of analytical tools that combine traditional electromagnetic finite-element analysis with micromagnetic simulation on HPC resources to improve the η of IVRs, and the energy efficiency of the data center processors they supply..

Cross-section of CMOS chip

Cross-section of CMOS chip with Ferric’s thin-film magnetic inductor technology integrated into the back-end-of-line process layers.

Lead Furnace

Industry Partner: Gopher Resource, Inc. (Gopher)/ Gas Technology Institute (GTI)
Principle Investigator: Alexandra Anderson, Ph.D., Gopher Resource, Inc. (Gopher)/ Gas Technology Institute (GTI)
National Lab Partner: Prashant K. Jain, Ph.D., Oak Ridge National Laboratory

Summary: In 2018, about 1.3 million tons1 of refined lead was produced using secondary sources in the United States (US). This was accomplished primarily by recycling lead batteries through a combination of physical concentration, hydrometallurgical, and pyrometallurgical processes. Secondary lead production in the US provided for more than 70% of the total lead demand1 and was estimated to have a total US market value of $3 billion1 in 2018. With such a prominent role of secondary lead, it is important that vital design and manufacturing process improvements are pursued to realize significant energy, cost and environmental benefits. In this project, Gopher is partnering with GTI and ORNL to propose development of a high-fidelity computational fluid dynamics (CFD) model of a directly fired reverberatory style secondary lead furnace. The proposed simulations on the Department of Energy (DOE’s) high-performance computing (HPC) clusters will enable a detailed scientific understanding of the underlying multiphysics interactions in the high-temperature environment of a lead furnace. The model will simulate the combustion and melting processes from the first principles while accounting for complex interphase interactions between the gas, solid charge material, slag, and metal phases. The proposed CFD model, which will be validated against the data obtained from Gopher facilities and burner characterization work performed at GTI, will enable significant improvements in design, operational parameters, and energy efficiency, hence improving productivity and refractory lifetime of secondary lead melting furnaces. The resulting improvements can provide energy savings of at least 1 trillion BTUs, reductions in greenhouse gas emissions of at least 1 million ton/year, and a total cost savings of more than $50 million/year to the US lead industry.

typical secondary lead reverberatory furnace

A typical secondary lead reverberatory furnace for Gopher’s recycling operations.

High Power Impulse Magnetron Sputtering (HiPIMS)

Industry Partner: Applied Materials, Inc.
Principle Investigator: Adolph Miller Allen, Ph.D., Applied Materials, Inc.
National Lab Partner: Andrea Schmidt, Ph.D., Lawrence Livermore National Laboratory

Summary: We propose the continued development of predictive modeling capabilities for the advanced film deposition technique, High Power Impulse Magnetron Sputtering (HiPIMS), which can be used to deposit materials for energy-efficient lighting such as high efficiency LEDs and in energy-efficient non-volatile memory storage devices. In 2018, Applied Materials and LLNL completed a 2016 HPC4Mfg collaboration culminating in an experimentally validated, small-scale, 3-Dimensional HiPIMS plasma model of a carbon/krypton material system. This model pushed the boundaries on high-fidelity magnetron simulations, essentially demonstrating for the first time a particle-in-cell magnetron model with realistic densities and magnetic fields. Despite this model being state-of-the-art, more computationally intensive models of larger systems and longer time scales are required to address full-scale reactor design. Thin carbon films are used to fabricate non-volatile memory storage that can significantly reduce national energy consumption when used in large scale data centers. In this project, we propose to model larger systems and to extend the model to reactive processes such as aluminum nitride to improve the quality of the AlN layer in LEDs, making them higher in efficiency and cheaper. This Phase II HPC4Mfg project is meant to address the additional model development that is needed to overcome challenges in HiPIMS based deposition.

Cross-section of magnetic field

Cross-section of magnetic field in magnetron device for first 3D magnetron simulation. The longest 3D simulation has been run out about 15 µs, which is roughly the length of the entire discharge.

Optimizing Non-Newtonian Slurry Atomizers

Industry Partner: Eastman
Principle Investigator: Wayne Strasser, Eastman
National Lab Partner: Haomin Yuan, Argonne National Laboratory

Summary: Eastman seeks to optimize by means of high-performance computing (HPC) non-Newtonian slurry atomizers for large-scale energy production facilities at our Kingsport, TN, site. The atomizers disintegrate fuel feeds for a water cracking operation that is critical for meeting our global business obligations. To improve atomization, we will use computational fluid dynamics (CFD) models to evaluate various geometry and feed condition permutations. The laboratory partner will work on developing open-source-software-based (OpenFOAM and Nek5000) models of the atomizer starting from ongoing work being performed at Eastman. This project will involve large-scale calculations on laboratory clusters and leadership-class computing facilities. The models generated and the simulations performed will be used by Eastman to gain insight into the physics of the phenomena involved and to optimize the process. Success is defined as discovering a mechanism by which we achieve at least a 2% improvement in energy production. That would result in a net present value of $10 million for Eastman.

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Vector plot colored by Mach number of a normal shock wave moving from left to right in air just having passed over a single droplet of water

Computational Fluid Dynamics Simulations

Industry Partner: Praxair Surface Technologies, Inc.
Principle Investigator: Andrew Heidloff, Praxair Surface Technologies, Inc.
National Lab Partner: Iver E. Anderson and Bo Kong, Ames Laboratory

Summary: Many crucial components in energy production, e.g., stationary gas turbines and A-USCS powder plants, are desirable to manufacture by additive manufacturing (AM) with the full development of the materials and processes. Typical gas atomized (GA) feedstock for metal AM is produced in spherical powder form, but often inefficiently with a wide size distribution and with reduced quality, e.g., having internal porosity, heavily oxidized surfaces, and “satellites” that degrade printing and build quality. The proposed project aims to significantly enhance the efficiency/precision and, moreover, the quality of metal powder production by computational fluid dynamics (CFD) simulations, probing the details of the physical processes occurring in supersonic gas atomizers. The goals are improving the desired powder size range yield and quality of GA powder, reducing material and energy production costs, and helping fulfill the immense the potential of AM, thereby enhancing the competitiveness of US manufacturing and of energy and transport sectors.

3D VOF simulation

Example of a 3D VOF simulation of the atomization process in a close-coupled nozzle.

Round 6: Special Call Summer 2018 Selectees

The High Performance Computing for Manufacturing Program (HPC4Mfg) awarded $1.2 million in federal funding for public/private projects aimed at solving key manufacturing challenges in steelmaking and aluminum production through supercomputing.

The special call for proposals, the sixth overall for the program, focused on applying the unparalleled high-performance computing capabilities of the Department of Energy’s (DOE) national laboratories to steelmaking and aluminum production processes. Under the program, each selected industry partner will have access to the national labs’ HPC machines and expertise to help these industries become more competitive, boost productivity and support American manufacturing jobs.

The HPC4Mfg Program expands its impact with new awards:

United States Steel Corporation (USS) will partner with LLNL in developing the expansion of the thermomechanical profile across a hot strip mill simulation model that provides predictions of through-thickness temperature, deformation behavior, and associated microstructure in a project titled “Hot Strip Mill Simulation Model”.

AK Steel Corporation will partner with LLNL to improve real-time modeling of hot strip milling in a project titled “Application of HPC for Hot Rolling of Next Generation Steels”.

ArcelorMittal USA, LLC will partner with LLNL to enable production of defeat free steel slabs with minimal trial and error in a project titled “Energy Efficient Manufacturing of Steel Slabs with the Application of High Performance Computing (HPC) and Machine Learning”.

Alcoa USA Corporation will collaborate with ORNL to advance the performance of Alcoa’s new advanced smelting cell in a project titled “Optimization of Alumina and Aluminum Fluoride Feeding in Advanced Aluminum Smelting Cells Using Large Eddy Simulation”.

Hot Strip Mill Simulation Model

Industry Partner: Purdue Northwest, Carbontec Energy Corporation
Principle Investigator: Evgueni Nikitenko, United States Steel Corporation (USS)
National Lab Partner: Aaron Fisher, Lawrence Livermore National Laboratory

Summary: To meet the Corporate Average Fuel Economy (CAFE) standards mandated by Department of Transportation’s National Highway Traffic Safety Administration, automakers are incorporating increasing volumes of light-weight Advanced High Strength Steels (AHSS) into their designs. Demands for thinner and wider AHSS presents a significant challenge to steelmakers as they need to manufacture them within the constraints of existing rolling mills in a cost-effective approach. USS has developed a hot strip mill simulation model that provides predictions of through-thickness temperature, deformation behavior, and associated microstructure at a selected single location along the slab length throughout the hot rolling process. Expansion of the thermomechanical profile across the strip width is desired to improve the model’s predictive capability. Potential impacts include cost reductions by optimizing rolling operations and reductions in unforeseen metallurgical changes by decreasing the number of required trials to develop robust AHSS rolling practices.

Steel Mill

Hot Rolling Steels

Industry Partner: Purdue Northwest, Carbontec Energy Corporation
Principle Investigator: Ronald H. Radzilowski, Ph.D., AK Steel Corporation (USS)
National Lab Partner: Victor Castillo, Lawrence Livermore National Laboratory

Summary: To use High Performance Computer (HPC) capabilities for constructing a reduced order model (ROM) usable in the real-life application of a hot rolling mill based on an existing, vetted physics-based off-line hot rolling model. The steel industry needs a near real-time model to be applicable for actual production and calculate the results along the entire length of a coil. The ROM is expected to predict with a reasonable accuracy the mechanical properties through the entire length of the hot rolled coil. The benefits of such a tool for the steel industry are: 1) Produce products with consistent properties i.e. reduce variation and non-conformance, 2) Develop new products with new required properties, 3) Save time, money, and energy by reducing the number of expensive industrial trials, 4) Reduce both thermal and electrical energy using an optimized, fast hot rolling model in real time.

Steel Roll

Improvement of Steel Production

Industry Partner: Purdue Northwest, Carbontec Energy Corporation
Principle Investigator: Dr. Tathagata Bhattacharya, ArcelorMittal USA, LLC.
National Lab Partner: Victor Castillo, Lawrence Livermore National Laboratory

Summary: The iron and steel industry is the fourth largest energy-consuming industry in the U.S. The iron and steel industry consumed an estimated 6% (~1470 PJ) of the total energy consumed in the whole U.S. manufacturing sector (www.eia.gov, 2006). About 80% of the total energy consumed in the steel industry is used to produce steel slabs via the continuous casting route (96% of steel in the U.S. today is produced via this route). Therefore, being able to produce defect free slabs (by making it right, first time, every time) would lead to a huge benefit in terms of energy savings and reduction of CO2 emissions during the steelmaking process by minimizing wastes and increasing quality and yield. Therefore, as the most recycled material on earth (more than all other materials combined), a slight improvement to steel production would have a lasting positive impact on the environment and on the U.S. energy landscape.

Diagram of Steel Production

Large Eddy Simulation to Improve Cell Design

Industry Partner: Purdue Northwest, Carbontec Energy Corporation
Principle Investigator: Rajneesh Chaudhary, Alcoa USA Corp
National Lab Partner: Prashant K. Jain, Oak Ridge National Laboratory

Summary: Alcoa is developing a novel hybrid advanced smelting process to increase productivity and cell performance while minimizing emission of greenhouse gases. The design utilizes novel materials and a unique anode-cathode combination. Computational tools and pilot prototypes are being developed to address materials, process, and design for this technology. Alumina and the bath ratio in the process both must be maintained at the desired levels for the new smelting process, which is quite challenging due to the unique flow field and limited area of the narrow space between the anode and cathode pair. Sophisticated computational tools are required to understand the alumina dissolution and bath ratio distribution while optimizing the number of feed, feed rates, and feeder locations. This work proposes application of large eddy simulations (LESs) to understand alumina dissolution and bath ratio distribution in the proposed cell design. Optimized feeder operation is necessary to achieve the target 15% energy efficiency improvement over the conventional Hall-Héroult process.

Mechanical Sketch

Winter 2018

The HPC4Mfg Program expands its impact with new awards:

  • Alliance for Pulp & Paper Technology Innovation (APPTI) (formerly Agenda 2020 Technology Alliance) will partner with NREL and ORNL to improve pulp yield during kraft pulping process in a project titled "Molecular Modeling to Increase Kraft Pulp Yield".
  • Dow Chemical Company will partner with SNL to reduce the thermal conductivity of insulating foam polyurethane products while using less polymer in its products in a project titled "Predictive Modeling of Polyurethane Foam Processes to Optimize Thermal Performance and Reduce Waste".
  • Seurat Technologies will partner with LLNL to optimize their innovative laser powder bed fusion additive manufacturing printer in a project titled "Fluid and Particle Dynamics in Metal Area Printing".
  • SFP Works, LLC will partner with ORNL to understand phase transformations that occur during the flash processing of steels in a project titled "High Performance Computing to Quantify the Evolution of Microscopic Concentration Gradients During Flash Processing".
  • 3M will partner with SNL to enhance the design of emissive films on building windows for cooling in a project titled "Passive Cooling Film Optimization". Project will be co-funded by the Building Technologies Office and the Advanced Manufacturing Office.
  • VAST Power Systems, Inc. will partner with LLNL and ANL to increase the efficiency and reduce start-up times of gas turbine combustors in a project titled "Ultra-Clean Transient Turbine Combustor". Project will be co-funded by the Office of Fossil Energy and the Advanced Manufacturing Office.
  • Arconic Inc. will partner with ORNL to model rolling processes to observe the evolution of porosity in a project titled "Computational Modeling of Industrial Rolling Processes Incorporating Microstructure Evolution to Minimize Rework Energy Losses".
  • GE Global Research Center will partner with LANL to improve the Truchas code for single crystal casting in a project titled "Highly Parallel Modeling Tool to Drive Casting Development for Aerospace and Industry Gas Turbines (IGT) Industries".
  • 3M will partner with ANL to optimize its fiber spinning manufacturing process in a project titled "Next Generation Nonwovens Manufacturing based on Model-driven Simulation Machine Learning Approach".
  • Raytheon Technologies Research Center will partner with ORNL to understand microstructure evolution during heat treatment of additively manufactured parts in a project titled "Predictive Tools for Customizing Heat Treatment of Additively Manufactured Aerospace Components".
  • Steel Manufacturing Simulation and Visualization Consortium (SMSVC) and ArcelorMittal USA will partner with ANL to improve the efficiency of the reheat furnace process in steel manufacturing in a project titled "Application of High-Performance Computing (HPC) to Optimize Reheat Furnace Efficiency in Steel Manufacturing".
  • Transient Plasma Systems Inc. will partner with ANL to develop more efficient dilute-burn engines in a project titled "Modeling of Non-equilibrium Plasma for Automotive Applications". This project will be co-funded by the Vehicle Technologies Office and the Advanced Manufacturing Office.

Increase Kraft Pulp Yield

Industry Partner: Alliance for Pulp & Paper Technology Innovation
Principle Investigator: David Turpin, Alliance for Pulp & Paper Technology Innovation (APPTI)
National Lab Partner: Brandon Knott, National Renewable Energy Laboratory, Jerry Parks Oak Ridge National Laboratory

Summary: The pulp and paper industry is an essential segment of the American economy, making products necessary for everyday life from renewable resources. Kraft pulping, the predominant technology is energy-intensive, capital-intensive, and provides a less efficient use of wood resources than desired. An approach to improve energy efficiency, increase mill profitability, and preserve jobs (this industry employs 378,000 American workers) is to increase pulp yield. By increasing pulp yield from 45 to 50% the energy intensity decreases by ~10% (13T BTU/year). A 5% yield increase has never been achieved because of alkaline lability of wood carbohydrates. Softwood pretreatment with a confidential chemistry (CC) increases yield by 3% primarily by stabilizing cellulose against degradation by alkaline peeling. Unfortunately, glucomannan stabilization by CC is minimal. A fundamental understanding of CC and alkali reactivity with wood components using molecular modeling may lead to identification of operating conditions to reach a 5% kraft pulp increase.

Green Glob

Molecular model of individual chains of hemicellulose (blue and red ‘sticks’) interacting with a cellulose microfibril (green surface).

Optimize Thermal Performance during Polyurethane Foam Processes

Industry Partner: Dow Chemical Company
Principle Investigator: Dr. Irfan Khan, The Dow Chemical Company
National Lab Partner: Dr. Rekha Rao, Sandia National Laboratory

Summary: Enhancing appliance insulation efficiency is an optimization problem requiring both improvements in the chemical formulation and the manufacturing process. Efficiency is a complex function of local solid/gas thermal conductivity, mixture density, and foam cell size distribution. A HPC-based polyurethane (PU) foaming model to predict the impact of formulation changes on bubble-scale thermal properties, concurrent to foam filling and curing, will expedite the development timeline for more energy efficient foams and appliances. Dow Chemical, the leading manufacturer of PU formulations, has state-of-the-art research to develop the next-generation PU formulations that both minimize material usage and maximize the energy efficiency of insulation. A new formulation/process-design paradigm, informed by predictive modeling, promises to allow Dow to achieve these goals with a significant reduction in slow and costly build-test cycles. This effort focuses on partnering with Sandia to enhance their HPC foam models via Dow’s expertise in kinetics and bubble-scale physics.

Wood Blocks

PMDI has a short pot-life: models can help reduce defects and improve filling process. The picture shows foam slabs made in various densities: 6 lb/ft3, 12 lb/ft3, and 20 lb/ft3 (from left to right). Defects can be seen in the slabs from coarse microstructure, exotherms, and poor surface finish.

Cake Mold Diagrams

Simulation of a “Cake Mold” geometry using SNL foaming model.

Reduce Spatter during Laser Powder Bed Fusion

Industry Partner: Seurat Technologies
Principle Investigator: James DeMuth, Seurat Technologies
National Lab Partner: Manyalibo Matthews, Lawrence Livermore National Laboratory

Summary: Metal additive manufacturing is growing rapidly and is of immense interest worldwide to decrease weight, increase functionality and improve manufacturing efficiency. Laser powder bed fusion additive manufacturing (PBFAM) printers from today’s market leaders routinely make high-value parts. However, today’s machines also suffer from low productivity and have major quality problems largely due to spatter, defects, and residual stress. Seurat’s novel printers operate by area printing enabling a high degree of spatial-temporal laser intensity control. Careful tuning of thermal history can lead to minimal spatter, ability to control microstructure, and decreased residual stresses. We propose to use high-speed video, material analysis and multiphysics modeling to characterize the effects of the interaction between the lasers and powder and to elucidate the underlying physics. Simulation guided advancements to Seurat’s area AM printing can impact energy-related technologies such as light-weighted vehicles, heat exchangers, inventory reduction and novel high-performance parts.

Green Flame

Improve Flash Process

Industry Partner: SFP Works, LLC
Principle Investigator: Gary M. Cola, Jr., SFP Works, LLC
National Lab Partner: Balasubramaniam Radhakrishnan, Oak Ridge National Laboratory

Summary: Flash processing employs rapid thermal cycling to strengthen commercial, off the shelf steel sheet, plate, or tubing into advanced high strength steel. The proposed HPC4Mfg project will utilize massively parallel Monte Carlo and phase field simulations to quantify the microstructural evolution that occurs during Flash processing including (1) the kinetics of ferrite grain growth in the presence dissolving carbide particles, and (2) the kinetics of austenite nucleation and growth. The objective of the simulations is to quantify the evolution of chemical heterogeneity in the Flash process that is responsible for the presence of through-thickness heterogeneity in hardness due to bainitic and martensitic transformations occurring in austenite grains with composition gradients. The simulations will help in designing the incoming material structure and the Flash processing cycle to obtain target microstructural gradients required for a desired combination of strength and ductility in the Flash processed material.

Diagram of FLash Process

Flash Processing Setup

Passive Cooling Film

Industry Partner: 3M
Principle Investigator: Dr. Robert Secor, 3M
National Lab Partner: Dr. Rekha Rao, Sandia National Laboratory

Summary: Over $372 billion is spent on electricity for cooling annually. Passive cooling, which requires no electricity, is a new way to dramatically reduce energy usage for refrigeration and other cooling methods.

Several candidate film structures have been proposed as prototypes in academic settings which are based on complex metamaterial fabrication processes. These combine a visible mirror to reflect sunlight with infrared radiation emission to the cold upper Earth atmosphere for cooling. The challenge for commercialization of this technology is to optimize the emissive film structures for functionality as well as low-cost, large-scale manufacturing. We propose using high performance computing to first develop an optimized design of emissive structures for functionality, and subsequently ensure the manufacturability of the design by detailed process modeling of this multistep manufacturing process for this composite metamaterial thin film.

Mirrors on grass

Ultra-Clean Transient Turbine Combustor

Industry Partner: VAST Power Systems, Inc.
Principle Investigator: David L. Hagen, VAST Power Systems, Inc.
National Lab Partner: Victor Castillo (LLNL), Lawrence Livermore National Laboratory and Prithwish Kundu, Argonne National Laboratory

Summary: California’s 50% renewable mandate is creating a 30,000 MW 3 hour “Duck Curve” backup power ramp by 2030. To prevent high wind/solar transients from causing South Australia-like grid blackouts requires very rapid <10-minute response gas turbine backup for 5-minute dispatch, 1-minute emergency dispatch. Clean air legislation mandates 95% cleaner gas turbine combustion, incurring high catalyst and increased operating costs. Emissions control during rapid turbine startup is difficult. This research develops VAST® combustor design optimization enabling 24% better single turbine efficiency, with 60% higher net power, and emissions below California mandates without catalysts or ammonia. It demonstrates ultra-accurate peak temperature control, potentially doubling gas turbine blade life. Fast ramp VAST® turbines enable major US renewable grid penetration, displacing peaking turbines, and competing with combined cycle turbines by higher profitability with lower capital and cleanup costs on shorter use. They offer rapid distributed power systems deployment for Africa, Asia, and Latin America.

Thermogenrator Laser

VAST® Thermogenerator™ Lab Test

Computational Modeling of Industrial Rolling Processes

Industry Partner: Arconic Inc.
Principle Investigator: Patrick Ulysse, Arconic Inc.
National Lab Partner: Sarma Gorti, Oak Ridge National Laboratory

Summary: Because light-weight materials are fuel efficient, aluminum, already well-established in the aerospace industry, is gaining significantly in the automotive industry and other markets. Currently, greater amounts of aluminum must be produced than are needed because of manufacturing waste and scrap inefficiencies that increase energy consumption and lead to higher costs to end consumers. This project addresses rolled plate and sheet recovery issues in aerospace, industrial and automotive products. Arconic will comprehensively model the commercial rolling process using high fidelity three-dimensional models to virtually observe the evolution of porosity in greater detail, with the ultimate objective of optimizing rolled plate properties. Arconic’s rolling and research facilities and over 100 years of rolling experience provide specific knowledge of the process, and facilities for defining and validating processing conditions. DOE’s national labs will provide scientific support, computational resources and expertise, and data storage to close the current industrial computation gap.

3D model

Highly Parallel Modeling Tool to Drive Casting Development for Aerospace and Industry Gas Turbines (IGT) Industries

Industry Partner: GE Global Research Center
Principle Investigator: Dr. Huijuan Dai, GE Global Research Company
National Lab Partner: Dr. Neil Carlson, Los Alamos National Laboratory

Summary: Casting is one of the oldest and most energy-intensive industries. US casting shipments in 20151 totaled $28.5 billion and 10.8 million tons and consumed 100 trillion BTUs of energy in 20102. The average defect rate is 5% leading to over $1.4 billion in lost production each year. For single crystal /directional solidification investment casting, development yield is less than 75%. Modeling could be used to improve this but is not adequately utilized due to long simulation run time.

GE GRC will collaborate with LANL to demonstrate the use of a scalable open-source modeling tool to design the optimal manufacturing process window for an industrial gas turbine (IGT) blade. This aims to reduce casting trials by 50%, improve first time yield and lower cost by 60%.

Schematic of gas turbine

Next Generation Nonwovens Manufacturing based on Model-driven Simulation Machine Learning Approach

Industry Partner: 3M
Principle Investigator: Alejandro Londono Hurtado, 3M
National Lab Partner: Ian Foster, Argonne National Laboratory

Summary: This project is aimed at minimizing energy consumption of the fiber spinning manufacturing process. This process is used in a wide variety of 3M products including filters, fabrics and insulation materials. The process is extremely energy intensive since it relies mostly on compressed air and electrical heating. This proposal seeks the minimization of energy consumption by using a combination of HPC-based Computational Fluid Dynamics (CFD) simulations and a machine learning approach. It is estimated that optimization will lead to a 20% reduction in energy consumption.

Image of spinning fibers

Predictive Tools for Customizing Heat Treatment of Additively Manufactured Aerospace Components

Industry Partner: Raytheon Technologies Research Center
Principle Investigator: Dr. Ranadip Acharya, Raytheon Technologies Research Center
National Lab Partner: Radhakrishnan Balasubramaniam, Oak Ridge National Laboratory

Summary: This project will utilize the phase-field based predictive tool MEUMAPPS developed by ORNL to simulate the effect of heat treatment on microstructure evolution of additively manufactured parts. Experimental measurement of as-deposited morphology and segregation patterns during additive manufacturing (AM) of the Ni-base alloy IN718 will be used as input for modeling of solid-state transformations in MEUMAPPS. MEUMAPPS will be integrated with a computational thermodynamics model of a Ni-Al-Nb ternary alloy serving as a surrogate alloy for IN718. The modeling results will be calibrated and validated against experiments conducted at Raytheon Technologies Research Center. The goal of the solid-state transformation modeling is to develop a novel heat treatment scheme that focuses on meeting property targets by eliminating deleterious Laves and delta phases from the microstructure while significantly reducing the annealing time compared to the existing practice, resulting in potential saving of 30-40% of the current energy consumption of the process.

Heat Graph

Optimizing Reheat Furnace Efficiency in Steel Manufacturing

Industry Partner: Steel Manufacturing Simulation and Visualization Consortium (SMSVC) and ArcelorMittal USA
Principle Investigator: David White, Steel Manufacturing Simulation and Visualization Consortium (SMSVC) and ArcelorMittal USA
National Lab Partner: May Wu, Argonne National Laboratory

Summary: Reheat furnace operation is essential to the steel industry and is thus crucial to national security and the economy. It is also very energy-intensive. Around 140 reheat furnaces operate in the U.S. and altogether consume approximately 1.6x1014 Btu/year of natural gas alone. Major technical challenges for reheat furnaces include low energy efficiency and inconsistent productivity and reheating quality. To address these issues, a computational fluid dynamics (CFD) methodology is chosen to optimize furnace operations. Due to complex furnace geometries, phenomena, and numerous scenarios, high-performance computing is essential to yield comprehensive and detailed CFD analyses and systematic parametric studies. Project objectives are to use CFD to 1) simulate furnace and process phenomena to build databases for optimization, and 2) evaluate hot/mixed charging, recuperator use, and other applications that have high energy-recovery capability. This project has the potential to save $350+ million /year through reduced energy consumption, improved energy efficiency, and increased productivity.

Dilute-burn Internal Combustion Engines

Industry Partner: Transient Plasma Systems Inc.
Principle Investigator: Dan Singleton, Transient Plasma Systems Inc.
National Lab Partner: Riccardo Scarcelli, Argonne National Laboratory

Summary: The development of enabling technologies for dilute-burn internal combustion engines is considered by automotive engine manufacturers as the most practical solution to meet stringent emissions regulations. Dilute-burn engines are key to developing more efficient and environmentally friendly vehicles, nevertheless ignition instabilities associated with dilute mixtures prevent wide-spread application to the transportation sector. TPS technology is currently the only solution that meets the requirements from these manufacturers, with minimal implementation cost and reduced component wear with respect to the conventional spark-plug approach. A key to market acceptance is the ability to properly characterize and optimize the plasma properties in order to develop solid control strategies. Relying on multi-dimensional modeling and high-performance computing (HPC) appears as the only viable solution to efficiently investigate and optimize transient plasma characteristics. High-fidelity modeling can accelerate TPS’s path to market and result in a reduction of emissions over 50% and fuel savings over 20%.

Image of VizGlow calculations

VizGlow calculations showing the evolution of thermal (Temperature) and chemical (O atom) plasma properties in a nano-pulsed discharge between two pin electrodes

Spring 2017

The HPC4Mfg Program expands its impact with new awards:

  • PPG Industries, Inc. will partner with LBNL to continue the modeling of an electrostatic rotary bell atomizer used to paint automobiles in a follow-on project titled “Optimizing Rotary Bell Atomization.”
  • Vitro Flat Glass LLC. will partner with LLNL to develop real-time glass furnace control using a neural net-based reduced order model of a CFD simulation of molten glass flow in a follow-on project titled “Advanced Machine Learning for Glass Furnace Model Enhancement.”
  • Caterpillar Inc. will partner with ANL to increase efficiency and reduce emissions on optimizing heat transfer in diesel engines through simulations of piston and spray geometry in a project titled “Heavy-duty Diesel Engine Combustion Optimization for Reduced Emissions, Reduced Heat Transfer, and Improved Fuel Economy.” Project will be co-funded by the Vehicle Technology Office and the Advanced Manufacturing Office.
  • Eaton Corporation will partner with ORNL to develop waste heat recovery (WHR) technology that can be applied to industrial manufacturing processes and vehicle operations in a project titled “High Performance Computing to Enable Next-generation Low- temperature Waste Heat Recovery.”
  • General Motors LLC will partner with LLNL to reduce cycle time in composite manufacturing in a project titled “Computational Modeling of High Pressure Resin Transfer Molding (HP-RTM) for Automotive Structural Carbon Fiber (CF) Composites.” Project will be co-funded by the Vehicle Technology Office and the Advanced Manufacturing Office.
  • Arconic Inc. will partner with LLNL and ORNL to develop advanced understanding of the non-equilibrium metallic phases established during metal additive manufacturing (AM) processes in a project titled “Multiscale Modeling of Microstructure Evolution During Rapid Solidification for Additive Manufacturing.” Project will be co-funded by the Office of Fossil Energy as an HPC4Materials for Severe Environments seedling project and by the Advanced Manufacturing Office as part of the HPC4Mfg portfolio.
  • Vader Systems, LLC will partner with SNL to understand the physics needed to apply transition MagnetoJet (MJ) 3D printing technology to a higher melting point metals and higher ejection rates in a project titled “Computational Modeling of MHD Liquid Metal 3D Printing.”

Optimizing Rotary Bell Atomization

Industry Partner: PPG Industries, Inc.
Principle Investigator: Reza Rock, Sr., PPG Industries, Inc.
National Lab Partner: James Sethian, Lawrence Berkeley National Laboratory

Summary: The U.S. automotive industry spray-applied over 60 million gallons of paint in 2014. Much of this paint is applied by an electrostatic rotary bell atomizer. Commonly used bells are capable of applying paint at 20% higher throughput, but desirable atomization, which controls paint appearance and transfer efficiency, i.e., the volume of sprayed paint that ends up on the part, suffers when fluid delivery is increased. In our pilot HPC4Mfg grant, we have made significant progress toward modeling paint behavior during rotary bell atomization as a function of paint fluid properties. In this proposed extension, we will capitalize on this model to compute the complex fluid dynamics and extend the model to capture even more rheologically-complex behaviors. With this information, new coatings that atomize well at higher flow rates can be developed to increase productivity and reduce booth size. These improvements can deliver significant energy savings and enhance manufacturing competitiveness.

Photo of water droplets

This 2D model developed in our pilot program illustrates droplet shedding at industrially-relevant conditions.

Graphic simulation of water droplets

Training of a Neural Network Reduced-order Model for Glass Furnace Operations

Industry Partner: Vitro Flat Glass LLC.
Principle Investigator: Rajiv Tiwary, Vitro Flat Glass LLC.
National Lab Partner: Vic Castillo, Lawrence Livermore National Laboratory

Summary: Vitro proposes to collaborate with LLNL under the HPC4Mfg program to extend the recently developed reduced glass furnace model by enhancing it with real world production data. A machine learning approach will then be used to identify the boundary in operating space between good and poor-quality product. Finally, a control framework will be identified which facilitates control of the furnace. Glass manufacturing is an energy intensive process. The furnace performs multiple functions in the same space and small variations in input parameters can shift where in the furnace and how efficiently those functions occur.

This enhanced model will enable fast and accurate control of furnace operations. Similar models, if deployed across the glass industry, could improve operational efficiencies and reduce overall costs and energy usage by 3.5 trillion BTUs per year. These reductions will help maintain U.S. global competitiveness in this industry.

Flowstreams of molten glass

The image above shows a top view of flowstreams within the molten glass within the glass furnaces as modeled with CFD and visualized with LLNL Visit software. The glass flow is from left to right.

Output Chart

Output from the visualization tool of predicted critical response variables (middle) and visualization of flow streams within the glass melt (bottom) for a given set of input variable (top sliders) the glass furnace model developed in phase 1. The actual proprietary data and parameters is not shown. This is for illustration only.

Heavy-duty Diesel Engine Combustion Optimization for Reduced Emissions, Reduced Heat Transfer, and Improved Fuel Economy

Industry Partner: Caterpillar Inc.
Principle Investigator: Jon Anders, Caterpillar Inc.
National Lab Partner: Sibendu Som, Argonne National Laboratory

Summary: Heavy-duty diesel engines continue to be a primary power source for construction and mining equipment, transportation, and power generation. Improving the efficiency with which diesel fuel energy is converted into useful work while minimizing pollution is critical to reducing energy consumption and enhancing the clean diesel technology that is integral to the sustainability of Caterpillar’s Construction and Resource industries. The manner in which the combusting fuel jet and in-cylinder air interact with the piston has a significant impact on combustion and emissions characteristics, as well as heat transfer to the piston. Heat loss through the piston is important from an overall engine efficiency perspective and component life. This project will optimize piston and fuel spray geometry for performance, emissions, and heat transfer objectives using state-of-the-art CFD inclusive of simultaneous gas- and solid-phase simulations on Leadership-class HPC infrastructure and will validate the optimized solution via single-cylinder engine testing leveraging additive manufacturing (AM) technologies.

Model of combustion engine parts

(Image used with permission from Convergent Science)

HPC to Enable Next-generation Low-temperature Waste Heat Recovery

Industry Partner: Eaton Corporation
Principle Investigator: Swami Subramanian, PhD., Eaton Corporation
National Lab Partner: Prashant K. Jain, PhD., Oak Ridge National Laboratory

Summary: The US manufacturing industry fails to recover an estimated 900 trillion BTUs of low-temperature waste heat from its processes each year. A grand research challenge has been to develop waste heat recovery (WHR) technology that can be applied to industrial manufacturing processes and vehicle operations. Despite considerable investment, cost-effective solutions remain elusive. The research team proposes to develop an innovative direct-contact heat exchanger technology to deliver low-cost, compact, longer-lifetime, high-efficiency waste heat recovery that is optimized for a low-temperature organic Rankine cycle. National laboratory expertise in high performance computing (HPC) and multiphase flows will be needed to realize this goal while advancing the fundamental understanding of two-phase, two-immiscible-fluid turbulent flow heat transfer and immiscible-fluid membrane separation for heat exchangers and membrane filtration systems. The team will leverage Eaton Corporation’s state-of-the-art testing facilities to acquire process performance data with which to validate these fundamental models. These simulations will unlock the development of smaller, higher-efficiency thermodynamic components for use in a novel low-temperature heat recovery process.

Sketch of Model part Sketch of Model part, side angle

Computational Modeling of High Pressure Resin Transfer Molding (HP-RTM) for Automotive Structural Carbon Fiber (CF) Composites

Industry Partner: General Motors LLC
Principle Investigator: Venkat Aitharaju, General Motors LLC
National Lab Partner: Michael Homel, Lawrence Livermore National Laboratory

Summary: High Pressure Resin Transfer Molding (HP-RTM) is a potentially game-changing manufacturing technology for allowing advanced composite materials to meet high volume automotive requirements. The proposed project addresses the challenges involved in predicting the outcome of HP-RTM manufacturing. This objective requires solving coupled multi-physics problems of fluid flow in a porous media, including fiber deformation, curing of the resin, and heat transfer between the resin and mold. This highly nonlinear problem will require simulating large finite element models across various scales by utilizing a large number of high performance computers. Improving predictability of the HP-RTM process will accelerate the large scale introduction of carbon fiber composites to achieve lightweighting and thus reduce both fuel consumption and emissions. A 10% implementation of carbon fiber composites in 10% of new vehicles can reduce fuel usage by 1.9 billion gallons of gasoline per year and correspondingly reduce 2.0 million tons of CO2.

Macro-scale simulation chart

Multiscale Modeling of Microstructure Evolution During Rapid Solidification for Additive Manufacturing

Industry Partner: Arconic Inc.
Principle Investigator: Tyler E. Borchers, Arconic Inc.
National Lab Partner: Tomorr Haxhimali, Lawrence Livermore National Laboratory and Jean-Luc Fattebert, Oak Ridge National Laboratory

Summary: The project aims to developing the advanced understanding and data necessary to establish the processing-microstructure relationship for metal additive manufacturing (AM). Using high-performance computing and multiscale modeling capabilities, the team proposes to simulate the highly non-equilibrium kinetic behaviors of solute chemistry at an overdriven liquid/solid interface and their impact on rapidly solidified microstructures during AM. In particular, the team will (1) employ large-scale classical molecular dynamics (MD) simulations to investigate the kinetic behaviors of interfacial atoms at or near a rapidly migrating liquid/solid interface; and (2) apply the MD-derived non-equilibrium interfacial parameters to a mesoscopic phase-field model to analyze the effects of the non-equilibrium interfacial chemistry on the solidification microstructure during AM. Research will provide the essential kinetic information for tailoring mechanical performances of AM alloys by controlling their solidification microstructures.

Photo of microscopic structure

Computational Modeling of MHD Liquid Metal 3D Printing

Industry Partner: Vader Systems, LLC
Principle Investigator: Scott Vader, Vader Systems
National Lab Partner: Joseph Bishop, Sandia National Laboratories

Summary: Vader Systems’ “MagnetoJet” (MJ) 3D printing technology uses a magnetohydrodynamic, drop-on-demand method of creating aluminum 3D parts more quickly, safely and efficiently than current powder-based 3D printing technologies. The MJ does this by melting aluminum wire in a ceramic nozzle and then ejecting liquid aluminum droplets at frequencies of 500-1000Hz, instead of melting specially prepared metal powders which are difficult to reuse between builds. The technology is not restricted to printing aluminum or to these ejection rates. However, in order to transition the technology to higher melting point metals and higher ejection rates, achieving an understanding of the physics and identifying and optimizing critical printing parameters is essential. By replacing high power sintering lasers and energy intensive powder preparation processes, the MJ printing technology is uniquely positioned to provide energy, time, and cost savings, while addressing critical powder safety issues (fire and explosion potential).

3D printed shapes Image of 3d Printer

Fall 2016

The HPC4Mfg Program expands its impact with new awards:

  • Ford Motor Company will partner with ANL to study the effect of dimensional tolerances on the cylinder to cylinder variation in engine performance in a project titled “CFD Study of Impact of Part-to-Part Variations on Spark-Ignition Engine Charge Formation.”
  • Arconic, Inc. will partner with ORNL to develop high-melting-point, lightweight alloys in a project titled “High Performance Computing for Phase Predictions for Multi-Component Alloy Systems.”
  • GE Global Research will partner with ORNL to reduce process development time and accelerate process certification of additively manufactured parts in a project titled “Powder Spreading Process Maps for Metal Additive Manufacturing.”
  • LLNL will partner with Raytheon Technologies Research Center to reduce defects in additively manufactured parts in a project titled “High Fidelity Physics-based Model Driven Process Parameter Selection for LPBF Additive Manufactured Metallic Aerospace Components.”
  • General Electric Research Corporation will partner with ORNL to reduce manufacturing costs in a project titled “Surface Roughness Effects from Additive Manufacturing in High Efficiency Gas Turbine Combustion Systems.”
  • Applied Materials, Inc., will partner with LLNL to improve powder bed formation in additively manufacturing processes in a project titled “Simulating Properties of Metal Powder Beds Used for Additive Manufacturing of Parts in Semiconductor, Solar and Display Equipment.”
  • LBNL will partner with Samsung Semiconductor, Inc. (USA) to optimize the performance of semiconductor device interconnects in a project titled “Making semiconductor devices cool through HPC ab initio simulations."
  • Arconic, Inc. will partner with LLNL and ORNL to develop new lightweight alloys in a project titled “Computational Modeling of Multi-Strand Aluminum DC Vertical Casting Processes Incorporating Cast Structure and Thermal Treatment Effects Contributing to Rework Energy Losses."
  • 7AC Technologies will partner with NREL to improve air conditioning technologies in a project titled “Modeling water vapor transport at liquid/membrane interfaces for applications in liquid desiccant air conditioners.”
  • Sierra Energy will partner with LLNL to enable gasification technologies to reduce landfill waste and create renewable energy in a project titled “High Performance Computing of Sierra Energy’s FastOx® Gasification Polisher to Optimize Waste-to-Syngas Conversion."
  • 8 Rivers Capital will partner with LLNL to develop a robust oxy-fuel sCO2 combustion CFD model to evaluate the performance of the Allam Cycle combustor in a project titled “Advancement of combustion design and modelling techniques through the application of high performance computing to sCO2 combustor development.”
  • The Timken Company will partner with ORNL to improve reliability and lifetime of wind turbines in a project titled “Crystal Plasticity Finite Element Model to Study the Effect of Microstructural Constituents on White Etch Area formation in Bearing Steels.”

Computational fluid dynamics simulations of spark ignition in engines as built to investigate the importance of manufacturing tolerances on performance and efficiency.

Industry Partner: Ford Motor Company
Principle Investigator: James Yi, Sibendu Som
National Lab Partner: Argonne National Laboratory

Summary: Spark-ignition engines are the backbone behind people transportation around the world. The efficiency of spark-ignition engines is limited in practice by variations between engine cycles and cylinders within an engine that result from the manufacturing processes/tolerances. These variations impact knock limits and dilution tolerance, which results in more conservative settings for design and calibration settings, such as compression ratio, valve timing, and exhaust gas recirculation rates. Engine variations also have a significant impact on emissions generation, which can have a secondary impact on efficiency. A deeper understanding of the relative importance of these variations and their interactions on the charge preparation process can guide future decisions on machining tolerances and control strategies. This project will develop simulation tools and methodology to include the effects of some key manufacturing tolerances and their impact on engine performance and emissions.

Image of a motor Graphic of heat transfer

Using first principles calculations to develop novel alloys for the automotive industry that can replace existing aluminum alloys in high temperature environments to further reduce the weight of automobiles and aircraft and thereby increase fuel efficiency.

Industry Partner: Arconic Inc.
Principle Investigator: Andreas Kulovits, James Robert Morris
National Lab Partner: Oak Ridge National Laboratory

Summary: This project will utilize high-throughput computing to generate knowledge and data necessary for accelerating new lightweight alloy discovery, important for the automotive industry to meet the increasingly stringent gas mileage regulations. The focus will be on systems that have not been explored due to experimental and manufacturing challenges, where limited data exists. Recent advances in fabrication might make the implementation of these systems viable, if justifiable. The advent of high performance computing may now be brought to provide reliable information that makes research time-, cost-effective and the expense defendable. Specifically, we propose to (1) utilize high-throughput calculations on important lightweight systems with limited data; (2) use and refine recently developed phase identification approaches including metastable states; (3) apply Monte Carlo (MC) approaches, including recently established direct first-principles based MC utilizing high-performance computing, for further refinement of finite- temperature behavior, and for development of databases of validated empirical models for faster exploration.

Development of powder spreading process maps to enable significant reduction of process development time and acceleration of process certification and new product introduction.

Industry Partner: General Electric Global Research Center
Principle Investigator: Meisam Salahshoor, Srdjan Simunovic
National Lab Partner: Oak Ridge National Laboratory

Summary: The powder spreading process in powder-bed based additive manufacturing (AM) often leads to layers with varying heights and areas of high porosity (resulting in defective parts). However, the bulk of the research in powder-bed based AM is mostly concerned with powder fusion or the binding step, effectively presuming even powder spreading. To bridge that gap, this work will develop a computational modeling framework to generate powder spreading process maps. Those maps will assist with optimal process design to ensure even layers. GE Global Research (GEGR), with extensive expertise in AM using metal powders and characterization of their rheological and tribological properties, will collaborate with ORNL to utilize lab’s high performance computing resources and expertise in multi-scale, multi-physics modeling to develop and validate the computational framework. The resulting process maps will reduce the cost and time associated with process development resulting in a faster and less costly realization of the energy and emission savings potential of AM.

Diagram of roller process

Demonstrate the use of high fidelity physics-based simulation models can facilitate the process parameter selection in additive manufacturing (AM), with focus on process defects and part surface roughness, leading to better AM processes and approaches.

Industry Partner: United Technologies Research Corporation
Principle Investigator: Vijay Jagdale, Andrew Anderson
National Lab Partner: Lawrence Livermore National Laboratory

Summary: Current Laser Powder Bed Fusion (LPBF) process development approaches are predominantly experimental requiring multiple iterations to obtain the required part form, fit and function. The process chain tends to have a large number of post-processing steps to achieve high density components with desired surface finish and properties. Experimental approaches are costly, inherently energy intensive, require long lead times and are not easily transferrable to other material systems and part geometries. Aerospace safety critical components demand aggressive performance metrics often requiring multi-objective process parameter selection. The proposed work focuses on utilizing LLNL’s detailed powder spreading and powder consolidation models and HPC capabilities for developing detailed physics based understanding of layer and part level defect and surface roughness generation. The simulation model outputs will help generate process maps and guide process parameter selection for reduced defects and improved surface finish, while enabling ~30-40% energy savings through reduced experimentation and post-processing operations.

Chart of melt pool model

As-built additively manufactured parts exhibit surface roughness which may impact turbine performance. The project will assess the need for post-processing of additively manufactured parts to reduce surface roughness.

Industry Partner: General Electric Company
Principle Investigator: Yuxin Zhang, Srikanth Allu
National Lab Partner: Oak Ridge National Laboratory

Summary: Additive manufacturing has been increasingly adopted by gas turbine manufacturers for rapid prototyping and mass production of complex, light-weight parts for combustion systems. However, the additive manufacturing process typically results in higher surface roughness compared to conventional manufacturing. Surface roughness effects, which are difficult to efficiently model with high accuracy, result in a performance difference between drawing board concepts and as-built parts, which poses challenges on combustion system efficiency improvement and manufacturing cost reduction.

Today, engine flow simulation boundary conditions are not well quantified in terms of surface roughness, which is either ignored or empirically estimated based on manufacturing material and surface treatment processes. We will use efficient integral wall modeled LES and surface reconstruction with X-ray computed tomography to model surface roughness effects in high efficiency combustion systems. This project is expected to fundamentally advance the design methodology of additive manufacturing products with significant energy savings.

Image of turbine flame

Scale resolved simulation of model single cup combustor rig at GE Global Research (not to scale)

Developing a comprehensive understanding powder bed formation in additive manufacturing processes will reduce defects in adaptively manufactured parts leading to reduced energy costs, minimized material waste and new design flexibility.

Industry Partner: Applied Materials, Inc.
Principle Investigator: Ajey M Joshi, Eric Herbold
National Lab Partner: Lawrence Livermore National Laboratory

Summary: dditive manufacturing (AM) of parts in semiconductor manufacturing equipment (SME) can reduce material needs and costs by up to 90% and enable returning end-of-life products to as new condition using only 2-25% of the energy required to make new parts. The powder bed laser fusion process is the most widely accepted process in AM that selectively solidifies powdered material in successive layers to form 3D parts. Powder bed properties significantly influences melt pool dynamics and may cause defects in the fabricated parts. Comprehensive understanding of metal powder bed layer properties is extremely important to utilize the full potential of AM for SME industry. Lawrence Livermore National Laboratory (LLNL) modeling expertise in Discrete Element Method (DEM) algorithms coupled to Lagrangian Finite Element (FE) solvers will be utilized to investigate powder-bed formation for AM. This work could reduce energy use and minimize materials waste while providing new design flexibility and shortening product time to market.

Sketch of Powder simulation

Powder bed spreading simulation over an existing melt track using geodyn-l

Optimize materials and reduce interconnect resistance on semiconductor devices using ab initio simulations with 100,000 atoms, allowing the continued down-scaling of semi-conductor devices.

Industry Partner: Samsung Semiconductor, Inc. (USA)
Principle Investigator: Byounghak Lee, Lin-Wang Wang
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: For decades, the semiconductor technology has followed the Moore’s law, making newer devices more powerful and energy efficient. Recently, however, it has reached a point where the performance and the energy efficiency of the device do not improve with the shrinking device size. One of the fundamental reasons of this deviation from the past trend is the interconnect resistance, which becomes larger with the shrinking size. It causes the delay of signals and heating of the circuits. The devices size is so small that the quantum mechanical effects can no longer be ignored and the traditional continuum simulation tools such as TCAD become inadequate. In this project, Samsung Semiconductor Inc.(USA) and Lawrence Berkeley National Laboratory will collaborate to perform first of kind device-scale ab initio simulations to optimize materials and interconnect morphology to minimize interconnect resistance. Mitigating this problem will allow the continued down scaling of the devices.

Image of electron microscope

Left: Scanning electron microscope image of Samsung’s 14nm logic device (Courtesy of Chipworks). M1, M2, M3, M4 are metal. The red circled areas indicate the interconnect contacts. Right: Schematic diagram of interconnect contact area. The dashed lines denote the simulation region. The barrier layer and the liner are also made of metal.

Reducing the number of aluminum alloy ignots rejected in direct chill casting through modeling and simulation of the casting and solidification processes will decrease production costs.

Industry Partner: Arconic, Inc.
Principle Investigator: Patrick Ulysse, Vic Castillo
National Lab Partner: Lawrence Livermore National Laboratory

Summary: Direct Chill (DC) casting facilities produce about 70% of aluminum ingots, with Arconic casting several billion pounds of metal each year. A significant amount is reprocessed because the cast material does not meet required standards. Many casting rejections, and even some aborts, can be traced to irregularities in the casting/solidification process. Arconic’s program will include physical modeling of the DC casting process using existing commercial software and a new sampling strategy, to efficiently explore the high-dimensional space of input parameters. Arconic’s casting and research facilities, along with more than 100 years of casting experience, will provide specific knowledge of the process; facilities for defining and validating processing conditions; and the requirements necessary for this type of modeling. DOE’s national labs will provide scientific support, modeling simulations, computational speed, data storage, and data structures needed to perform and post-process large amounts data to close the current industrial computation gap.

Photos of cracked metal

Several images of cracks exhibited in alloy 7050

Optimizing the hydrophobic membrane of a desiccant air dehydrator will improve the efficiency of air conditioning systems in under extreme humidity conditions.

Industry Partner: 7AC Technologies
Principle Investigator: Peter Luttik, Jason Woods
National Lab Partner: National Renewable Energy Laboratory

Summary: Conventional compressor-based cooling has undergone incremental changes over the past 100 years. Further improvements in efficiency require evermore complex systems, especially under humid conditions. Meso-porous membranes offer unique opportunities for efficient humidity control in buildings using an absorbent desiccant solution, but membranes are not designed or optimized for this purpose. This project will use molecular dynamic simulations to determine optimal membrane properties for these air conditioning applications, focusing on the membrane properties at the membrane/liquid/air interface. Optimal membrane designs will enable smaller, more durable, and less expensive designs of these membrane air conditioning systems.

Sketch of Microporous Membrane

This project will aid the scale up the FastOx gasification process to convert waste to syngas by optimizing process parameters using detailed 3D models and experimental data.

Industry Partner: Sierra Energy
Principle Investigator: Daniel Dodd, Nick Killingsworth
National Lab Partner: Lawrence Livermore National Laboratory

Summary: Sierra Energy has successfully tested its FastOx gasification, waste to renewable energy, system at bench scale and is building a first-of-a-kind plant at Fort Hunter Liggett in California. The team has conservatively designed FastOx using a zero-dimensional model based on heat and mass balances, but this model does not capture process dynamics or spatial variations. More sophisticated modeling of system components would enable greater understanding of waste to syngas conversion. Sierra Energy will partner with Lawrence Livermore National Laboratory to conduct high performance computing modeling of the FastOx gasification system’s polisher to advance understanding of its complex physical and chemical interactions, optimize polisher efficiencies, and reduce polisher equipment and operational costs. Advancing the energy efficiency and economics of gasification systems, such as FastOx, could ultimately enable the industry to convert 100 million tons of waste, annually landfilled in the US, into 120,000,000 MWh of renewable electricity.

Photo of pressure tank

FastOx polisher undergoing fabrication and pressure testing at Andy J. Egan Co. in Grand Rapids, MI (May 2016)

Advancement of combustion design and modelling techniques through the application of high performance computing to sCO2 combustor development.

Industry Partner: 8 Rivers Capital, LLC
Principle Investigator: Dr. Xijia Lu, Dr. Greg Burton
National Lab Partner: Lawrence Livermore National Laboratory and National Energy Technology Laboratory

This project is co-funded by the DOE Office of Fossil Energy.

Summary: A novel supercritical CO₂ power cycle, known as the Allam Cycle, holds significant promise to provide low-cost, emissions-free power from fossil fuels. The Allam Cycle employs oxy-combustion and a sCO₂ working fluid to generate highly-efficient, low-cost power while inherently capturing CO2. This cycle is currently being developed to utilize natural gas at a demonstration facility in Texas. The Allam Cycle is also capable of utilizing coal as a fuel, but this requires design of a novel combustor capable of utilizing low-BTU fuels, such as coal-derived syngas. This program proposes to use experimentally derived kinetic data and advanced computer modeling to optimize the design and produce a robust analysis of the combustor’s ability to operate on a wide range of fuels.

Diagram of Combustion model

Crystal Plasticity Finite Element Model to Study the Effect of Microstructural Constituents on White Etch Area formation in Bearing Steels

Industry Partner: The Timken Companyn
Principle Investigator: Rohit Voothaluru, Sarma B. Gorti
National Lab Partner: Oak Ridge National Laboratory

Summary: Wind turbines are subjected to a wide variety of operating conditions which stretch the bearings beyond their design limits and result in premature damage of the bearings due to a condition widely referred to as white etch cracking (WEC). Today, the actual service life of wind turbines is often much less than the designed 20 years. Although the fundamental root cause of WEC bearing damage remains unknown, much of the industry suggests that retained austenite plays a vital role in preventing WEC from forming in bearings. The ability to understand and computationally model the transformation plasticity and the transformation kinetics of retained austenite in mechanical power transmission components would result in the development of engineered materials and microstructures that can overcome the white etch cracking problem. This would lead to an improved performance of the bearings and significantly increase the reliability and life of wind turbine gearboxes (WTG), which currently see an average loss of 650,000 MWh due to downtime because of bearing WEC failures.

Image diagram of turbine

Bearings in a Wind Turbine Gearbox

Spring 2016

The HPC4Mfg Program expands its impact with new awards:

  • Shiloh Industries of Ohio will partner with ORNL to study phase change cooling of tooling to speed up casting processes in a project titled "Development of a Transformational Micro-Cooling Technology for High-Pressure Die Casting using High-Performance Computing."
  • Rolls-Royce Corporation of Indiana will partner with ORNL to improve silicon carbide composites in a project titled "Level-set Modeling Simulations of Chemical Vapor Infiltration for Ceramic Matrix Composites Manufacturing."
  • ORNL will partner with Agenda 2020 Technology Alliance, a consortium focused on the paper industry to design better catalysts for lignin breakdown in a project titled “Catalytic Pulping of Wood."
  • LLNL will partner with GE Global Research Center in New York to study how to mitigate defects caused by direct metal laser melting in a project titled “Minimization of Spatter during Direct Metal Laser Melting (DMLM) Additive Manufacturing Process using ALE3D Coupled with Experiments."
  • PPG of Pennsylvania will partner with LBNL to decrease the time needed to paint automobiles in a project titled "Modeling Paint Behavior During Rotary Bell Atomization."
  • Actasys, Incorporated of New York, will partner with ORNL to decrease the fuel consumption of trucks by actively modifying the flow around the trucks in a project titled "High Performance Computational Modeling of Synthetic Jet Actuators for Increased Freight Efficiency in the Transportation Industry."
  • Carbon, Incorporated of California will partner with LBNL to increase the speed of polymer additively manufactured components in a project titled "Multi-physics Modeling of Continuous Liquid Interface Production (CLIP) for Additive Manufacturing."
  • The American Chemical Society Green Chemistry Institute will partner with LBNL to systematically explore lower energy mechanisms of chemical separation using adsorbents and membranes in a project titled "Accelerating Industrial Application of Energy-Efficient Alternative Separations."
  • The Alzeta Corporation of California will partner with LBNL to destroy effluents from semiconductor processing that could potentially harm the ozone layer in a project titled "Improving Gas Reactor Design With Complex Non-Standard Reaction Mechanisms in a Reactive Flow Model."
  • Sepion Technologies of California will partner with LBNL to make new membranes to increase the lifetime of Li-S batteries in a project entitled “Improving the Manufacturability, Performance, and Durability of Microporous Polymer Membrane Separators for Li–S Batteries using First Principles Computer Simulations.”
  • Applied Materials, Incorporated will partner with LLNL to enable the manufacture of higher quality, more efficient LEDs for lighting in a project titled "Modeling High Impulse Magnetron Sputtering (HiPIMS) plasma sources for reactive Physical Vapor Deposition (PVD) processes used in fabrication of high efficiency LEDs."
  • General Motors LLC of Michigan and EPRI of California will partner with ORNL to improve welding techniques for automobile manufacturing and power plant builds in a project titled "High Performance Computing Tools to Advance Materials Joining Technology."
  • Harper International Corp. of New York will partner with ORNL to reduce the cost of carbon fibers in a project titled "Development and Validation of Simulation Capability for the High Capacity Production of Carbon Fiber."

Development of a Transformational Micro-Cooling Technology for High-Pressure Die Casting using High-Performance Computing

Industry Partner: Shiloh Industries of Ohio
Principle Investigator: Adrian Sabau
National Lab Partner: Oak Ridge National Laboratory

Summary: In high-pressure die casting (HPDC), latent heat is extracted from the molten metal, and the casting solidifies. Heat is extracted through the steel die by a system of cooling channels.

Challenge: Owing to the low steel thermal diffusivity, the heat concentrates near liquid metal-die steel interface, increasing the solidification time and hence limiting the productivity.

Solution/Concept: To overcome this challenge, Shiloh Industries is developing an innovative two-phase micro-channel cooling system and a comprehensive numerical model. This will allow the prediction of heat transfer coefficients (HTC) while obtaining optimum flow parameters necessary to achieve a uniform temperature distribution, controlling casting quality and mechanical properties, and eliminating common heat transfer related defects (soldering, hot cracking, and shrinkage porosity). Basis for proposed technology: A micro-channel cooling system has been demonstrated at lab scale. Several flow regimes (which influence the heat transfer between fluid and channel walls) were experimentally identified in micro-channels: bubbly, slug, churn, wispy-annular, and annular. Implementation challenge: These experimental results cannot include the wide range of designs, heat flux rates, shapes of the channel, wall boundary conditions, and most importantly, specific high-heat fluxes experienced in HPDC.

Level of scientific computing: Currently, a general purpose Computational Fluid Dynamics (CFD) commercial software is used for the simulation of HPDC but with limited application to micro-channel cooling. Wider use of the micro-channel system is limited by a lack of numerical models available to simulate the combined fluid flow and vapor-liquid phase changes in the complex 3D shapes utilized. Gap addressed by national labs: ORNL has access to several CFD source codes that can be used to test various boiling models in the micro-channels. The ORNL computer codes are installed on high-performance computing machines at ORNL.

Level-set Modeling Simulations of Chemical Vapor Infiltration for Ceramic Matrix Composites Manufacturing

Industry Partner: Rolls-Royce Corporation of Indiana
Principle Investigator: Ramanan Sankaran
National Lab Partner: Oak Ridge National Laboratory

Summary: Silicon-carbide (SiC) reinforced ceramic matrix composites (CMCs) are a key enabling technology to reduce fuel consumption and emissions of gas turbine engines. In one manufacturing approach, chemical vapor infiltration (CVI) is limited to only coating SiC fibers. The preform is then fabricated using a “lay-up” of basic plies or 2-D sheets composed of the precoated fibers. At the other extreme, CVI is used to completely densify a 3-D preform shaped in large part like the gas turbine component itself. The latter approach is more suitable for highly engineered components which sit directly in the gas path of the engine, for example, a high pressure turbine blade. In this case, the geometry is necessarily complex for aerodynamic, stress, and lifing (multi-physics) requirements. Presently, optimizing the CVI-dominated manufacturing approach is largely by trial-and-error. In this work, a first-principles modeling of CVI is performed to realize optimization of SiC/SiC CMC manufacturing.

Catalytic Pulping of Wood

Industry Partner: Agenda 2020 Technology Alliance
Principle Investigator: Jerry Parks
National Lab Partner: Oak Ridge National Laboratory

Summary: The pulp and paper industry is an essential segment of the national economy, making products necessary for everyday life from renewable resources. Kraft pulping, the predominant technology, is capital-intensive and energy-intensive, and it provides a less efficient use of wood resources than desired. Catalytic methods may offer a viable alternative to increase fiber yield through improved selectivity, reduce energy use and eliminate the formation of odor-causing mercaptans. This project proposes a coupled experimental/computational approach to develop catalysts. The work will focus on cobalt-based catalysts shown to be capable of delignifying wood. Correlations established between experimental and computational results can be used to design and predict the performance of prospective catalysts prior to synthesis, accelerating the development process. Catalytic methods, if successful, could improve yield, reduce fiber waste, reduce energy use, lower the cost of bleaching and lessen the cost of air emission controls by eliminating sulfur-based pulping chemicals.

Minimization of Spatter during Direct Metal Laser Melting (DMLM) Additive Manufacturing Process using ALE3D Coupled with Experiments

Industry Partner: GE Global Research Center
Principle Investigator: Wayne King
National Lab Partner: Lawrence Livermore National Laboratory

Summary: During a laser melting additive manufacturing process, several defects such as spatter, porosity, and precipitate alignment form that lead to poor mechanical properties of the built part. We propose to study the laser-particle interaction through modeling using ALE3D from LLNL coupled with experiments to be performed at GE GRC on different materials. This will provide valuable insights to the nascent additive industry at large and lead to a better understanding of the role of material properties and process parameters as well as their interactions on defect formation. It has been shown that additive manufacturing could be used to enable lightweight structures and engines to replace conventionally manufactured one. In the transportation industry such as aviation and automobile, a 10% weight reduction would translate into millions of gallons of fuel savings per year and lead to a reduction in the total greenhouse gas emissions from the US.

Modeling Paint Behavior During Rotary Bell Atomization

Industry Partner: PPG of Pennsylvania
Principle Investigator: James Sethian
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: The U.S. automotive industry spray-applied over 60 million gallons of paint in 2014.1 Much of this paint is applied by an electrostatic rotary bell atomizer. Commonly used bells are capable of applying paint at 20% higher throughput, but desirable atomization, which controls paint appearance and transfer efficiency (i.e., the volume of sprayed paint that ends up on the part), suffers when fluid delivery is increased. We propose to model paint behavior during rotary bell atomization as a function of paint fluid properties. With this information, new coatings that atomize well at higher flow rates can be developed to increase productivity and reduce booth size. These improvements can deliver significant energy savings and enhance manufacturing competitiveness.

High Performance Computational Modeling of Synthetic Jet Actuators for Increased Freight Efficiency in the Transportation Industry

Industry Partner: Actasys, Incorporated of New York
Principle Investigator: David Pointer
National Lab Partner: Oak Ridge National Laboratory

Summary: Actasys is developing a cutting edge technology that will dramatically increase fuel efficiency in the transportation sector. The technology is based on Active Flow Control (AFC) which uses synthetic jet actuators to improve vehicle aerodynamics by changing the way air flows around the vehicle but without requiring a change to the vehicle's shape.

Actasys, in cooperation with Rensselaer Polytechnic Institute (RPI) and Price Chopper, is initially targeting heavy trucking. As the US trucking industry struggles to meet the goals of reducing fuel consumption and CO2 emissions, Actasys has shown that AFC can provide a step change in aerodynamic performance leading to significant improvement to freight efficiency.

Actasys has already demonstrated the potential for its technology in wind tunnel tests using 1:14 scale tractor-trailer model and in road tests of full-scale prototypes. But unlocking the full potential requires a multivariable parametric study that is only practical by high performance computing (HPC).

Multi-physics Modeling of Continuous Liquid Interface Production (CLIP) for Additive Manufacturing

Industry Partner: Carbon, Incorporated of California
Principle Investigator: Dan Martin
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: Continuous Liquid Interface Production (CLIP) is poised to bring additive manufacturing to multiple American manufacturing sectors owing to its unique combination of rapid print speeds and material options that resemble injection molding thermoplastics. In spite of these benefits, CLIP is far from a mature process. Here, we propose the development of a multi-physics computational model that encompasses the coupled chemical-physical processes of photopolymerization, heat transport, fluid flow, and structural deformation to predict part outcomes. A physically relevant model would enable rapid optimization of CLIP and significant gains in performance. Ultimately, this effort would enable the key benefits of additive manufacturing, namely mass customization, unlimited design space, and a cost- and energy-effective path to mainstream manufacturing.

Accelerating Industrial Application of Energy-Efficient Alternative Separations

Industry Partner: American Chemical Society Green Chemistry Institute
Principle Investigator: Debbie Bard
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: Distillation in the chemical industry accounts for roughly 10% of energy use in the U.S. While porous mass separating agents (MSAs) appear capable of achieving the same separations for a fraction of the energy, the fundamental lack of a well-understood relationship between the behavior of fluid mixtures confined in MSA pores and the selectivity of MSA-based processes presents a major barrier to their widespread industrial application. This work will be the first study to systematically and self-consistently explore the range of parameters describing the various molecular-material interactions in MSA-based separations. Through high performance computing (HPC), this study will yield a fundamental understanding of the influence of confined fluid behavior on selectivity. The results will also serve as a knowledge base for subsequent investigations needed to advance the framework for the rational design of MSA-based separation processes to enable significant reductions in the energy required for separations central to chemical manufacturing.

Improving Gas Reactor Design With Complex Non-Standard Reaction Mechanisms in a Reactive Flow Model

Industry Alzeta Corporation of California
Principle Investigator: Marcus Day
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: A large number of critical industrial processes require an understanding of chemically reacting flows and most of these same processes would benefit from improved computational modeling capability. Computational fluid dynamics (CFD) is now a widely used tool for understanding non-reacting 3-dimensional (3-D) flow fields. Reactive flow modeling (for example methane-air combustion using ORI-Mech 3.0) has also proven to be a useful design tool, but in even moderately complicated flow geometries computing time can quickly exceed the capabilities of available computer hardware. Access to high performance computing (HPC) would allow us to model 2-D and relatively simple 3-D flow fields while using relatively complicated and non-standard reaction mechanisms. These non-standard mechanisms are necessary for us to model the gases encountered in our industry, but our computational needs currently exceed the computing capability available to us. Advances made in this project will provide benefits applicable to a broad range of other industries.

Improving the Manufacturability, Performance, and Durability of Microporous Polymer Membrane Separators for Li–S Batteries using First Principles Computer Simulations

Industry Partner: Sepion Technologies of California
Principle Investigator: David Prendergast
National Lab Partner: Lawrence Berkeley National Laboratory

Summary: Energy systems on board aircraft are rapidly being electrified by the aviation industry. To meet industry targets for these systems, next-generation batteries with high specific energy density are essential. Efforts to commercialize light-weight, energy-dense lithium-sulfur secondary batteries (2510 Wh kg–1) have been stalled by ongoing problems with the battery’s membrane, which limits cycle-life. Sepion’s polymer membrane technology provides a counterpoint, yielding long-lasting lithium-sulfur cells. Advancing to 10 Ah battery prototypes, Sepion faces challenges in membrane manufacturing related to polymer processing and the molecular basis for membrane performance and durability. High-performance computing offers critical new insight into these phenomena, which in turn will accelerate our product’s entry into the market. Our successes will catalyze a transformation in aviation, where fuel-burning aircraft are replaced with hybrid-electric planes featuring 30–50% reductions in fuel costs and emissions.

Modeling High Impulse Magnetron Sputtering (HiPIMS) plasma sources for reactive Physical Vapor Deposition (PVD) processes used in fabrication of high efficiency LEDs

Industry Partner: Applied Materials, Incorporated
Principle Investigator: Andrea Schmidt
National Lab Partner: Lawrence Livermore National Laboratory

Summary: Light emitting diodes (LEDs) are becoming a strong option for an efficient lighting source in the world. The manufacture of LEDs can benefit from an improved AlN buffer layer. A requirement of the buffer layer is a high crystalline orientation alignment with substrate. A low energy ion flux of AlN ions as the deposition source can be used to modulate the crystallinity and bond structure of the deposited material. One potential deposition method is using a High Power Impulse Magnetron Sputtering (HIPIMS) plasma source. The high ion fraction and energy tunability of the HIPIMS source makes it an attractive option for AlN deposition. However, challenges in HIPIMS need to be overcome including deposition rate, target poisoning, and particle generation. Partnering with Lawrence Livermore National Lab’s (LLNL’s) modeling expertise in molecular dynamics (MD), implicit particle in cell (PIC), and kinetic Monte Carlo (kMC) modeling could help overcome HIPIMS source challenges. A viable HIPIMS source for material deposition could impact LEDs pricing and efficiency.

High Performance Computing Tools to Advance Materials Joining Technology

Industry Partner: General Motors LLC of Michigan and EPRI of California
Principle Investigator: Zhili Feng
National Lab Partner: Oak Ridge National Laboratory

Summary: GM and EPRI, representing two major US manufacturing industry sectors (automotive and nuclear energy) and both having welding research & development capabilities, will work with DOE national labs (having critical computational expertise in the welding field) for the purpose of advancing HPC weld modeling tools for broad industrial applications. This potential will be demonstrated with two representative welded structural components – the prediction of welding-induced dimensional changes and stresses during laser welding assembly of a complex roof panel made of high-strength lightweighting materials and the prediction of weld residual stresses in the dissimilar welds of nuclear piping systems. Our goal is to reduce the computational time of the two above examples from days or months to several hours while providing adequate solution accuracy so that the HPC weld modeling tools could effectively optimize welding technology in order to minimize dimensional distortion and proactively mitigate the detrimental impact of weld-induced residual stresses.

Development and Validation of Simulation Capability for the High Capacity Production of Carbon Flier

Industry Partner: Harper International Corporation
Principle Investigator: Srdjan Simonovic
National Lab Partner: Oak Ridge National Laboratory

Summary: There is a big incentive to incorporate Carbon Fiber Reinforced Polymer (CFRP) materials in automotive application for the purpose of light weighting and fuel savings. This requires very large quantities of fiber [~1M tons/yr], at a low price, and consistent high quality. The scale-up required to achieve this production and reduce cost is very challenging.

The conversion of PAN to carbon fiber is a thermochemical process where the precursor loses 50% of its mass, undergoes softening, shrinkage, recrystallization and reorientation. It is imperative that the temperature, gas flow, and chemical reactions, which critically affect properties, be simulated to ensure that the design provides the necessary uniformity. The proposed work will start with purchased software packages and then adapt them to carbon fiber production. The simulations will be validated at the pilot scale equipment available at ORNL followed by iteration and refinement and tested in a commercial production facility.

Fall 2015

The HPC4Mfg Program expands its impact with new awards:

Management Team of HPC4MFG

HPC4Mfg project awards were announced at NERSC, a partner in the HPC4Mfg Program. From left to right are Robin Miles, LLNL; Horst Simon, LBNL; Peter Nugent, LBNL; Trish Damkroger, LLNL; Dona Crawford, LLNL; Mark Johnson, DOE EERE; Kathy Yelick, LBNL; Jeff Roberts, LLNL; Peg Folta, LLNL; and John Turner, ORNL. Participating remotely in the announcement were David Danielson, Assistant Secretary for Energy Efficiency and Renewable Energy, DOE; and Mark D’Evelyn, Vice President, Bulk Technology, Soraa Corp. The group is shown with Cori, NERSC's newest supercomputer.

Fall 2015 Solicitation Selectees

The Awards from the first HPC4Mfg solicitation were announced by David Danielson, Assistant Secretary, Office of Energy Efficiency and Renewable Energy (EERE), Department of Energy, during an online media presentation on February 17, 2016.

The selected projects entering into CRADA agreements include:

  • United Technologies Research Center, located in East Hartford, Connecticut, will partner with ORNL and LLNL to develop and deploy simulation tools that predict the material microstructure during the additive manufacturing process to ensure that critical aircraft parts meet design specifications for strength and fatigue resistance, under a project entitled: "Integrated Predictive Tools for Customizing Microstructure and Material Properties of Additively Manufactured Aerospace Components."
  • Procter & Gamble of Ohio will partner with LLNL to reduce paper pulp in products by 20 percent, which could result in significant cost and energy savings in one of the most energy intensive industries, under a project entitled: "Highly-Scalable Multi-Scale FEA Simulation for Efficient Paper Fiber Structure."
  • General Electric (GE), New York, will partner with ORNL to assist in the local control of melt pool and microstructure in additive manufactured parts, under a project entitled: "Process Map for Tailoring Microstructure in Laser Powder Bed Fusion Manufacturing (LPBFAM) Process."
  • In a separate project, GE will partner with ORNL and LLNL to improve the efficiency and component life of aircraft engines through design optimization, under a project entitled: "Massively Parallel Multi-Physics Multi-Scale Large Eddy Simulations of a Fully Integrated Aircraft Engine Combustor and High Pressure Vane."
  • The AweSim program at the Ohio Supercomputer Center (OSC) and the Edison Welding Institute (EWI) will partner with ORNL to deploy cloud-based advanced welding simulation tool for broad industry use, under a project entitled: "Weld Predictor App."
  • PPG Industries, Inc. of North Carolina will partner with LLNL to model thermo-mechanical stresses involved in forming and solidifying glass fibers to understand fracture-failures mechanisms to significantly reduce waste, under a project entitled: "Numerical Simulation of Fiber Glass Drawing Process via a Multiple-Tip Bushing."
  • In a separate project, PPG of Pennsylvania will partner with LLNL to develop a reduced computational fluid dynamics (CFD) model of a glass furnace to make informed line adjustments in hours in near real-time, under the title: "Development of Reduced Glass Furnace Model to Optimize Process Operation."
  • The Lightweight Innovations for Tomorrow Consortium in Michigan will partner with LLNL to develop, implement and validate a defect physics-based model to predict mechanical properties of Al-Li forged alloy, under a project entitled: "Integrated Computational Materials Engineering Tools for Optimizing Strength of Forged Al-Li Turbine Blades for Aircraft Engines."
  • ZoomEssence, Inc. of Kentucky will partner with LLNL to optimize the design of a new drying method using HPC simulations of dryer physics, under a project entitled: "High Performance Computing Analysis for Energy Reduction of Industry Spray Drying Technology."

Integrated Predictive Tools for Customizing Microstructure and Material Properties of Additively Manufactured Aerospace Components

Industry Partner: United Technologies Research Center
Principle Investigator: Balasubramaniam Radhakrishnan and Jean-luc Fattebert
National Lab Partner: Lawrence Livermore National Laboratory, Oak Ridge National Laboratory

Summary:

  • Better prediction of additively manufactured materials will help engineers qualify parts for lightweighting and other applications. ORNL and LLNL researchers will partner with UTRC to work towards predicting material strength by modeling microstructure evolution to predict grain size and segregations (secondary phases) as function of process-parameters.
  • Coupled CFD-phase field multi-scale models will be coupled with select experimental data to predict the corresponding microstructure and mechanical properties.
  • A 30%-50% time and energy savings can be achieved in the process development cycle using these methods.

Graph of dendrite

Process Maps for Tailoring Microstructure in Laser Powder Bed Fusion Additive Manufacturing (LPBFAM)

Industry Partner: GE
Principle Investigator: Dr. Adrian Sabau
National Lab Partner: Oak Ridge National Laboratory

Summary:

  • The cost and time associated with LPBFAM process development is very high due to a lack of fundamental process understanding that link process parameters to microstructure.
  • A transformational local-global multi-physics model will be developed with the goal of providing experimentally validated process maps for tailoring microstructure to achieve desired performance for LPBFAM.
  • AM can reduce materials needs and cost by up to 90% and use only 2-25% of energy to remanufacturing parts to return end-of-life products to as-new condition. Energy saving potential of AM for aerospace industry showed that with lightweight and cost-effective designs for aircraft components, fleet-wide life-cycle primary energy savings were estimated to reach 70-173 million GJ/year in 2050, with cumulative savings of 1.2–2.8 billion GJ.

Chart of Temperature v. Velocity Microscopic images

Massively Parallel Multi-Physics Multi-Scale Large Eddy Simulations of a Fully Integrated Aircraft Engine Combustor and High Pressure Vane

Industry Partner: GE
Principle Investigator: Ramanan Sankaran and Greg Burton
National Lab Partner: Lawrence Livermore National Laboratory, Oak Ridge National Laboratory

Summary:

  • Design optimization of combustor and High Pressure Turbine (HPT) interaction will lead to significant improvements in durability and fuel efficiency.
  • ORNL and LLNL are performing Large Eddy Simulations (LES) to understand the physical interaction between the combustor and HPT components, thus guiding design optimizations.
  • Design optimization of gas turbine components can be worth up to 2% specific fuel consumption, 1.5% reduction in weight, 3% reduction in cost and 20% improved component life.

Schematic of turbine engine

Weld Predictor App

Industry Partner: Ohio Supercomputer Center (OSC) and the Edison Welding Institute (EWI)
Principle Investigator: Jay Jay Billings
National Lab Partner: Oak Ridge National Laboratory

Summary:

  • A cloud-based advanced simulation tool for welding will be deployed which will be especially useful for small to medium-sized companies with limited in-house capability.
  • ORNL researchers will partner with OSC to extend and improve an existing, proven, successful tool which uses an integrated 3D thermo-mechanical-material model for aluminum, titanium and steels welds.
  • This will greatly reduce prototyping time and costs by a factor of 10-to-50 fold, decrease energy use, and provide access to HPC resources for medium-sized manufacturers.

Graph of Thermo-Mechanical Model

Numerical Simulation of Fiber Glass Drawing Process via a Multiple-Tip Bushing

Industry Partner: PPG Industries, Inc. of North Carolina
Principle Investigator: Chris Walton
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Manufacturing of glass fiber loses significant time and energy from fiber breakage, when an entire 4000-fiber station must stop and restart. Breakage is linked to thermal engineering of the drawing environment.
  • LLNL’s computing capability will allow a full-scale model of the complex thermal environment and of fluid behavior of the solidifying fiber.
  • Energy costs of 1.7 trillion BTUs worth $8.5M, plus uptime and wasted glass, could be saved annually, industry-wide if up-time can be improved.

Man working machine Close-up of manufacting

Development of Reduced Glass Furnace Model to Optimize Process Operation

Industry Partner: PPG of Pennsylvania
Principle Investigator: Vic Castillo
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Production of glass is not only energy intensive but difficult to monitor in production due to the extreme environment. Computational models provide some insight into the molten flows but can often take too long to resolve the necessary details to make timely adjustments to the production line.
  • LLNL is using VisIT, the premier visualization and analysis tool for the HPC environment, along with machine learning methods to develop a reduced-order glass furnace model to enable plant engineers to make informed, real-time process adjustments.
  • This fast-running prediction tool can save roughly two weeks of production per year per furnace and increase productivity by 2%. Extrapolating that improvement to the entire U.S. glass manufacturing industry suggests 2.5 TBTUs of energy and 130,000 metric tons of carbon dioxide emissions could be saved.

Graphic of production model

Integrated Computational Materials Engineering Tools for Optimizing Strength of Forged Al-Li Turbine Blades for Aircraft Engines.

Industry Partner: Lightweight Innovations for Tomorrow Consortium
Principle Investigator: Tom Arsenlis and Sylvie Aubry
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Light-weighting of aircraft parts using new materials such as aluminum-lithium which can substitute for heavier titanium alloys can lead to significant savings in fuel industry-wide.
  • LLNL is using a unique materials modeling code, ParaDis, to predict the strength of Al-Li alloys produced under different process conditions.
  • Over 13 million gallons ($26M) could be saved per year industry-wide using the new material for turbine blades in aircraft engines.

Schematic of Turbine

High Performance Computing Analysis for Energy Reduction of Industry Spray Drying Technology

Industry Partner: ZoomEssence, Inc. of Kentucky
Principle Investigator: Greg Burton and Victor Beck
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Low temperature drying of such industrial products as food, pharmaceuticals, chemicals, polymers and ceramics has been shown to significantly reduce energy consumption and improve product quality
  • LLNL will be using massively parallel turbulent mixing and transport computational fluid dynamics (CFD) research codes in massively parallel configurations to fully resolve and better understand the relevant physics of the food-drying process.
  • The project will result in fundamental first-principles understanding of how the low temperature drying process works, and will suggest design improvements to the technology, further enhancing energy efficiency and product quality.

Graph of suspended emulsion

Seedlings

HPC4Mfg seedlings established the program infrastructure with a focus on challenges in a broad range of energy intensive industries.

  • Agenda2020, a paper making consortium, is partnering with LLNL and LBNL to reduce energy in paper-making, saving up to 80 trillion BTU’s per year.
  • Purdue Calumet and LLNL are advancing steel blast furnace modeling to reduce coke usage in steel-making that could save up to $80 million per year.
  • Soraa is partnering with LLNL to scale up their new GaN process to yield cheaper LED lighting and new power electronics.
  • Eaton and LLNL are developing predicting microstructure in additively manufactured metal parts to aid in qualification.
  • Carbontec Energy Corporation, Purdue Northwest, and LLNL are enabling the scale up of the E-Nugget production process which replaces coal with bio-mass in pig iron production.

Reducing energy in paper-making could save 80 trillion BTUs per year

Industry Partner: Agenda 2020
Principle Investigator: Yue Hao, David Trebotich
National Lab Partner: Lawrence Livermore National Laboratory, Lawrence Berkeley National Laboratory

Summary:

  • Rewetting of paper pulp following pressing-is widely considered to be a leading contributor to the energy intensity in paper making (3rd largest).
  • rainbow glob
  • LLNL and LBNL researchers are developing coupled-physics simulations to determine how water flows through porous paper pulp during and after the pressing process.
  • New press designs could reduce energy consumption by up to 20% (80 trillion BTU, in $240M - $400M annually).

Technical Topic and Abstract

Paper rollers Chart of deformation

Simulation of pressure in felt image data using Chombo-Crunch (photo courtesy of David Trebotich, LBNL)

Reducing coke usage in steel-making could save $80 million per year

Industry Partner: The Steel Manufacturing Simulation and Visualization Consortium (SMSVC) and the Center for Innovation through Visualization and Simulation (CIVS) at Purdue Northwest
Principle Investigator: Aaron Fisher
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Carbon – rich natural gas and coke are used in large quantities in blast furnaces to produce molten iron. Process optimization can allow for lower coke usage and will reduce carbon loads to the environment and process costs.
  • LLNL and CIVS researchers work to improve Purdue’s existing blast furnace models and enable them to run on HPC clusters.
  • Improved software will be used to simulate complex reactive flows through particles of coke and iron ore and identify furnace conditions with reduced coke utilization.
  • Optimized blast furnace processes could save $80 million/year industry-wide by reducing coke consumption.

Steel Manufacturing Simulation

Scaling up a new GaN process will yield cheaper LED lighting and new power electronics

Industry Partner: SORAA
Principle Investigator: Matt McNenly
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Scale-up of GaN crystal growth technology will result in reduced production costs of highly efficient, high brightness LED lighting. Next generation power electronics for renewables will also be enabled.
  • LLNL researchers are modeling the chemical processes in the chemical vapor deposition equipment used to grow crystals to assist in process scale-up.
  • A new high-fidelity model will save the years of trial-and error experimentation typically needed to facilitate large-scale commercial production.

LED lightbulbs LED diagram

Microstructure prediction in additively manufactured metal parts can help qualify parts

Industry Partner: Eaton
Principle Investigator: Wayne King/Jean-Luc Fattebert
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Additively manufactured parts can be used to reduce weight in aircraft and automobiles parts saving transportation fuel.
  • LLNL researchers are modeling the material microstructure of parts made using the laser-melt process of powder-bed additive manufacture (3D printing). Material properties can be predicted from the microstructure helping to ensure that the parts will meet engineering requirements such as strength and fatigue life.
  • Qualification of as-built parts is critical for meeting product reliability standards and hasten adoption of the new technology.

Laser-melt process

The E-Nugget bio-mass replacement for coal in pig-iron results in a carbon-neutral process

Industry Partner: Purdue Northwest, Carbontec Energy Corporation
Principle Investigator: Evgueni Nikitenko, United States Steel Corporation (USS)
National Lab Partner: Lawrence Livermore National Laboratory

Summary:

  • Large amounts of coal are currently used to convert iron-ore to pig-iron as a first step in steel-making. Replacing coal with bio-mass in the recently developed E-Nugget process results in a net-zero carbon emission iron smelting process, and could save $500M/yr industry-wide if adopted.
  • LLNL and Purdue Northwest researchers are utilizing HPC facilities to model a pilot E-Nugget furnace and a series of potential scaled-up furnaces. This process will guide the design process for a large scale E-Nugget furnace.
  • The resulting plant will be able to process 100,000 tonnes/yr of pig-iron at costs lower than conventional blast furnaces.

Technical Topic and Abstract

E-NUGGET BIO-MASS REPLACEMENT

HPC4Mfg is sponsored by the Advanced Manufacturing Office of the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy