2016 October 26 - December 13
2017 January 25 - March 7
Integrated Simulation and x-ray Interrogation Tools and Training for μmechanics at CHESS
- CHESS (Cornell High Energy Synchrotron Source)
- 1 of 2 high energy storage rings in the U.S.
- Funded by The National Science Foundation
- Beams of x-rays transmit through metallic samples
- FEpX FEM code on parallel cluster
- InSitμ: a new center at CHESS on the Cornell Campus sponsored by the US Office of Naval Research
- Build understanding and model crystal-scale behavior of engineering alloys
- High Energy X-ray Diffraction (HEXD) experiments at A2 and F2 CHESS stations
- Finite Element-based polycrystal model developed at Cornell by P. Dawson (FEpX)
- Enhanced support for a new generation of industrial users working on the most important problems
- Augment user experience with training on both experiment and simulation - both available at CHESS
- People and applications are the Industrial “contribution”
- Create a new generation of industrial users
- Crystal-based mechanical design
Crystal-Based Mechanical Design
OMC copper EBSD map (left); FEpX Virtual OMC Polycrystal (right).
Direct comparison of HEXD and FEM Diffraction Results During Cyclic Loading
Grain Scale Heterogeneity from FEM
High Energy X-ray Diffraction (HEXD) Setup
Wong et al, Comp. Mat. Sci. 77, 456, 2013
Obstalecki et al, Acta. Mat. 75, 259, 2014
- The crystal scale is the source of properties and behavior of engineering alloys
- Anisotropy - significant crystal to crystal variation in strength/stiffness - heterogeneity
- Macroscale design, use average values and factors of safety
- Compare HEXD to FEpX data directly
- Diffracted intensity from each crystal
- Compare to model using “virtual diffractometer”
- FEpX estimation of grain heterogeneity
- Distill knowledge (of grain scale behavior) to develop
- Blend process models with key HEXD measurements
- “Informed” stress concentration factors
- Residual stress estimations
Deformation Behavior at the Crystal Scale
AA2099 (Al-Li) (left); Finite Element Model of AA2099 (right).
Measured (squares) and modeled (lines) lattice strains.
Beaudoin et al., MSMSE 1 8(6), 2012
- Anisotropy creates heterogeneity in mechanical response
- Each crystal is a boundary value problem
- Crystal scale models for elastoplastic deformation
- Explicit representation of crystal structure and constitutive behavior
- Impossible to validate or calibrate using macroscale mechanical testing data
- HEXD experiments with in situ loading
- Snapshots of internal grain structure
- Interrogate every grain within an aggregate
- Hybrid HEXD-FEpX methodology
- Build trust in model (or experiment) by matching grain scale data (thousands of crystals)
- Example: Crack turning in AA2099
- Significant issue for shuttle fuel tanks
- “Delamination” between pancake grains
- Potential source: Shear stress distributions in neighboring grains
Diffraction volumes (DV) shrink-fit sample (left); (200) SPFs from inner(a) to outer DV at α = 90° (right).
Shrink-fit sample (left); σxx from Hybrid Optimization Shrink-fit sample interpolated with FEM (right).
McNelis et al, JMPS, 61, 428, 2013
Demir et al, Comp. Meth. Appl. Mech.,265, 120, 2013
Park et all, Exp. Mech., DOI 10.1007/s11340-013-9771-0, 2013
- Diffraction has been used for decades on residual stress
- Crystal lattice as a strain gage
- Fidelity doesn’t match design standards
- Converting lattice (elastic) strains to stress
- Measuring residual stress “fields”
- Lattice Strain Pole Figures (SPFs)
- Thousands of lattice strain measurements
- FE mesh over the workpiece
- Enforce equilibrium and free surfaces
- Multiscale optimization
- Stress field that best matches all the lattice strain measurements and satisfies equilibrium and free surfaces
- Moving forward
- Developing methods for “blending” processing models with “a few” HE x-ray measurements
- Motivating, calibrating and validating lab source methods
Structure/Goals of InSitμ
CHESS IT Director Werner Sun, CHESS Computing Cluster
Left to Right: Matt Miller, InSitμ Director; Armand Beaudoin, Industrial Program Director; Joel Brock, CHESS Director; Ernie Fontes, CHESS Associate Director
- HEXD Experiments
- Enormous growth in the past decade
- New class of data for polycrystalline alloys
- Difficult experiments, lots of details
- CHESS has state-of-the-art HEXD capabilities
- FEpX Finite Element Simulations
- Developed over the past 20 years at Cornell
- Explicit representation of crystal-scale physics
- Demanding parallel implementation
- Available to CHESS users on CHESS Cluster
- Can be implemented at the beamline
- InSitμ Structure
- Complements CHESS by providing specific expertise on HEXD & FEpX
- InSitμ Goals
- Get HEXD and FEpX “into the hands” of engineers working on most important structural problems
- InSitμ personnel will enhance CHESS support of HEXD experiments and FEpX simulations
- Experiment/simulation design and implementation
- Interpretation / blending of data
- Training opportunities and workshops
- Overarching goal: create a set of committed industrial users
A New Center at CHESS for Characterization of Engineering Materials
A new center based at the NSF-Sponsored Cornell High Energy Synchrotron Source (CHESS) has been recently funded by the Office of Naval Research (ONR). The Integrated Simulation and x-ray Interrogation Tools and training for micromechanics (InSitμ) center will focus on the mechanical behavior of structural materials by providing enhanced user support for the high energy x-ray diffraction (HEXD) methods developed at the CHESS A2 and F2 experimental stations over the past decade, as well as the crystal-based finite element code, FEpX, developed by Professor Paul Dawson from Mechanical Engineering at Cornell. Used together, the data and model can help to shed light on life limiting processes such as plasticity and fatigue crack initiation in polycrystalline metallic alloys. InSitμ will focus on enabling the use of these new methods on the most important structural materials challenges by creating a new generation of industrial users.
It is safe to say that structural materials – the materials used to support loads in bridges, buildings, aircraft and automobiles – impact the lives of nearly every human every day. Of the various classes of structural materials, metals and metallic alloys are the most prevalent. Metals are used across the spectrum of structural applications from the steel in a high-rise tower to titanium screws used in facial reconstruction hardware. Mankind has been processing metals for centuries. Early alloy development was a purely trial and error process of combining elements to achieve a particular set of mechanical properties. Over the past century, microscopy methods employing visible light, electrons or x-rays have enabled researchers to observe the size, shape, orientation and distribution of the tiny (1-100 micrometers) individual crystals that make up a metallic alloy component. Now, a particular set of alloy properties or performance measures can be associated with the image of the material microstructure. Using these data, an enormous microstructure-property “atlas” has been constructed over the years and has greatly enhanced the material selection process. However, predicting properties for an arbitrary microstructure, has not been possible, in general. This makes the development of a new alloy or a new material processing operation an extremely long, tedious undertaking involving extensive amounts of microstructural characterization experiments and mechanical tests. An accurate mechanical model of the microstructure as it responds to loading would enable such predictions by creating understanding of the crystal-scale mechanical response that drives life limiting processes such as plasticity and fatigue crack initiation. With detailed knowledge of the stresses and strains at the scale of the individual crystal we will be able to design important engineering components such as jet engine turbine disks and aircraft wings starting with single crystals. Currently, the lack of this high fidelity knowledge of mechanical state (stresses and strains) at the crystal scale requires designers to use large “factors of safety” to assure that failure will not occur. The reduction in mass that could be realized by eliminating such design conservatism in a typical structure such as an aircraft would be enormous.
Using High Energy X-rays and Finite Elements to Understand Crystal-Scale Deformation
X-rays have been used for interrogation and characterization of structural materials for a century. The discoveries behind x-ray diffraction and the overarching principles that basically enabled the field of modern crystallography are all based on the fundamental relationships that exist between diffracted x-ray intensity and the underlying material structure. The tremendous utility of x-ray diffraction ex- periments is the ability to utilize a diffraction model to convert a collection of diffracted x-ray intensity into a quantitative, often geometric description of the material structure that produced it. A high energy synchrotron light source produces well-characterized, concentrated beams of monochromatic high energy (short wavelength) x-rays that are capable of penetrating through bulk metallic samples during high energy x-ray diffraction (HEXD) experiments. Capturing large patterns of diffracted x-rays is now possible using modern area detectors with increasingly fast (sub-second) detection times and ever smaller pixel size. By employing ancillary equipment that can precisely control external loads, temperature and other environmental conditions, diffraction patterns can be obtained in real time in situ from a test specimen subjected to conditions that closely approximate those experienced during processing or in service.
Accurate representation of the evolving geometry of a multi-dimensional domain and the accompanying approximation of spatially-varying fields over that domain are the hallmarks of the finite element method (FEM), as is the ability to specify the relevant conditions on the boundary of a domain and the manner in which those conditions vary with time. The basic premise of the FEM is the discretization of the domain into elements and the representation of each field variable (such as stress, velocity or temperature) within each element using a decomposition into shape or interpolation functions of the spatial variables (ie. x,y,z) and the nodal point values of the field variable. While FEM is typically thought of as an engineering tool for designing structures or representing fluid flows and temperature fields, the great promise of finite elements for fundamental material science is this ability to systematically approximate the time- varying topology of a deforming material and at several relevant size scales, the values of important mechanical field variables such as stress, lattice orientation and strain rate.
Researchers from the Sibley School of Mechanical and Aerospace Engineering at Cornell have played major roles in the development of both HEXD methods and crystal-based finite element modeling. Professor Matt Miller began building mechanical loading machines for use at the Cornell High Energy Synchrotron Source (CHESS) in 2001. Since that time, Miller and his group have built several generations of loadframes and used them together with the high energy, high flux synchrotron x-rays produced at CHESS and at the Advanced Photon Source (APS) to study a broad spectrum of metallic materials subjected to monotonic and cyclic loading at room and elevated temperatures. Miller’s Sibley School colleague, Professor Paul Dawson, pioneered the use of finite elements for the representation of crystal-scale elastic-plastic deformation over 20 years ago. Dawson and his students have developed the state-of-the-art crystal scale finite element package called FEpX, which can be used to discretize a polycrystalline aggregate into millions of finite elements. Together, Miller and Dawson have examined crystal-scale deformation behaviors in copper, steel, titanium, aluminum and nickel alloys. In addition to examining elastic-plastic deformation behaviors from the in situ loading experiments, they have constructed a hybrid HEXD/FEpX methodology for determining processing-induced residual stresses in polycrystalline alloys.
Mission of InSitμ
The HEXD experiments and FEpX finite element formulation created and developed by Miller and Dawson, together with the experimental and computational infrastructure and excellent legacy of support at CHESS form the basis of InSitμ. The overarching goal of InSitμ is to connect these emerging experimental and computational methods with engineers working on important classes of structural materials challenges. InSitμ will supply enhanced experimental and computational support to enable industrial users with very little experience with x-rays to come to CHESS and conduct HEXD experiments and to blend the resulting diffraction data with FEpX simulations running on CHESS computers to extract crystal-scale properties and performance attributes.
Structure of InSitμ
Miller is the PI of the ONR grant for InSitμ joined by CHESS director Joel Brock and Associate Director Ernie Fontes as Co-PIs. Armand Beaudoin, Professor Emeritus of Mechanical Engineering at the University of Illinois Urbana Champaign, has been recently named as the first InSitμ director. In addition to nearly 20 years as a UIUC faculty member, Professor Beaudoin brings years of experience in the aluminum metals industry. A full time research assistant and post-doctoral researcher will join Beaudoin on the InSitμ staff. Professor Dawson will be a charter member of the InSitμ advisory board.
InSitμ Operation - Experiments
CHESS has three specimen loading systems that range from a small loadframe that can be easily mounted within a standard six circle diffractometer to state of the art closed-loop controlled load frames with submicron positioning capabilities. Recently, the Materials Directorate within the Air Force Research Laboratory at Wright Patterson AFB built and commissioned a new loading and positioning system capable of conducting near- and far-field HEXD experiments (yielding both structural and micromechanical response data) simultaneously. Having the right equipment is just the start. Most metallurgists or mechanical design engineers lack the x-ray science prowess to arrive at a beamline ready to conduct a strain pole figure experiment with in situ loading, for instance. The fundamental role of the InSitμ research scientist or post-doc, therefore, is to meet the user beyond “half way”. The bulk of the x-ray expertise will come from the InSitμ scientist. This will involve some experimental design before the actual experiment to understand what conditions might be optimal for the measurement at hand. In addition to real time experimental assistance, InSitμ will provide experimental training for users - specifically for industrial users - so they can start to extrapolate to new problems and, perhaps, to new experimental configurations.
InSitμ Operation - Simulations
A key attribute of InSitμ is the fact that the simulations and experiments are conducted at the same location. This is a significant departure from the more traditional approach of combining experiments and simulations by “throwing information over the wall”. The HEXD experiments and FEpX simulation framework were developed in a research configuration where both experimentalist and modeler contributed equally, an environment of checks and balances - where neither experiment nor the model is “validating” the other. This somewhat non-traditional environment will be replicated at InSitμ. Hosting extensive computational activities, such as FEpX, at a light source, however, is completely unprecedented. One of the main activities in the first year of the project, therefore, is implementing FEpX along with pre- and post-processing programs. Some progress has already taken place. A two day computational InSitμ school is planned in conjunction with the CHESS users meeting in June. We will use the school to jumpstart the computational aspect of the project and to identify challenges and opportunities for the initiation of the computational component of the project. In addition to integrating the HEXD results with FEpX, data analysis software of various kinds will be supported and further developed. We will build upon open-source software that has been developed by the HEXD community and code for ”on-the-fly” analysis that has been developed at CHESS. Integration of the HEXD analysis tools with FePX will be undertaken. We will host a software framework that enables the user to take a snapshot of their experiment, preserving the state of analysis present during the beamtime. This will foster interaction after the beamtime; InSitμ scientists will collaborate with users and help to bring the analysis to completion.
Traditional access to synchrotrons is through scientific proposals seeking to accomplish a single measurement or task using a specific beamline or instrument. There has been precious little industrial use of synchrotron experimental stations for investigating structural materials at CHESS or at other light US light sources. Since the creation of new industrial users is the primary goal of InSitμ, we will pilot a new approach by forming Industrial Project Team (IPTs) composed of faculty and students from academia, post-docs and synchrotron experts from CHESS, and scientists and engineers from industry. The IPTs will originate from discussions that InSitμ management initiates with potential industrial partner organizations. InSitμ management will also identify faculty and graduate students who are interested in joining the IPTs based on common interests in the underlying materials research. Together with InSitμ staff at CHESS, these teams will identify and analyze industrial challenges, configure or design experimental data collection and modeling tools, and solve materials problems. As a location that hosts the x-ray data collection facilities and main computing cluster, CHESS will host IPT activities and serve as a central meeting place. Part of the management of InSitμ will be to form an InSitμ Advisory Board (IAB) comprised of at least one representative from each of the industrial partners.
Training personnel and involving students
The student pipeline is quite well established, wherein Masters and Ph.D. candidates spend between 1-5 years (typ.) training at academic institutions. Student projects utilizing synchrotron light sources lead to training on one or more techniques, established at one or more existing beamline facilities. It is very rare for students to learn beamline design, or carry out technical development. Rarer still is students having the opportunity to alter existing beamline or endstation equipment to suite their own project needs. Training of professional staff in industry happens in a similar way to student training in the sense that visiting a synchrotron is the only venue to learn about existing techniques and individual beamline capabilities. CHESS has a very different model for student involvement and a much more flexible, academic laboratory style for beamline and endstation development and utilization. InSitμ will build upon the CHESS training model for all participants in the IPTs. Visiting faculty and visiting graduate students will reside part time at CHESS and will be considered staff and have full access to the engineering, design and scientific staff. Industrial scientists and engineers on the IPT will be afforded the same opportunities, and expected to shoulder whatever tasks are needed in the course of their project(s). InSitμ also expects to organize annual workshops similar in nature to the FEpX school being held at the 2014 CHESS Users meeting. Some future workshops will focus on specific issues associated with the experiments and/or simulations but others will be basic introduction to HEXD and HEXD/FEM interface methods.