Big Area Additive Manufacturing

Out of Bounds Additive ManufacturingChristopher Holshouser, Clint Newell, and Sid PalasLockheed MartinCorp.Fort Worth, Tex.

Chad Duty, Lonnie Love, Vlastimil Kunc,Randall Lind, Peter Lloyd, John Rowe, Ryan Dehoff,William Peter, and Craig Blue,FASM*Oak Ridge NationalLaboratoryOak Ridge, Tenn.

The Big AreaAdditive Manufacturingsystem has thepotential tomanufactureparts completely unbounded insize.

*Member of ASM International

L ockheed Martin and Oak Ridge NationalLaboratory (ORNL) are working on anadditive manufacturing (AM) system

(Big Area Additive Manufacturing, or BAAM)capable of manufacturing components meas-ured not in terms of inches or feet, but multipleyards in all dimensions (Fig. 1). It has the po-tential to manufacture parts completely un-bounded in size. Lockheed Martin pioneeredthe development of advanced materials, equip-ment, and processes for AM, and continues tobe an industry leader in the field, realizing a vi-sion that started in the late 1990s for manufac-turing advanced thermoplastic composites ofarbitrary size and shape. To further this goal,Lockheed Martin teamed with ORNL to lever-age the lab’s historic expertise in materials de-velopment and processing through the recentlyestablished U.S. Department of Energy’s (DOE)Manufacturing Demonstration Facility (MDF).The partnership will accelerate the maturationand eventual commercialization of the Big AreaAdditive Manufacturing (BAAM) system.

AM benefitsMost current AM systems are confined to

a limited build envelope defined by a powderbed or an enclosed deposition zone, such as anoven. By contrast, BAAM is not confined by acontrolled-deposition chamber or restricted toa single deposition head. Instead, the system isdesigned to accommodate a team of coordi-nated robots working in an open-air environ-ment to produce components and structuresunbounded in size. The system not only ex-pands the working volume of AM, but also en-ables the use of advanced polymer composites,multiple materials within a single component,multifunctional materials, and automated in-sertion of dedicated subcomponents, enablinga “CAD-to-system” approach rather than a“CAD-to-part” philosophy.

Most conventional manufacturing tech-niques (e.g., drilling, milling, and turning) areclassified as subtractive because they involveremoving material from a block of raw stock.Alternatively, AM refers to a suite of rapid, fullyautomated manufacturing technologies basedon additive principles. AM enables productionof complex structures directly from three-di-mensional (3-D) CAD models in a layer-by-layer process using metals, polymers, andcomposite materials. Selective laser sintering(SLS), fused deposition modeling (FDM; Strata-sys Inc., Eden Prairie, Minn.), and electronbeam melting (EBM; Arcam AB, Mölndal, Swe-

den) are just a few examples of AM technolo-gies. Unlike subtractive techniques that requiretime and energy to remove material, AM makesvery efficient use of energy and raw materials,depositing material only where it is needed.This leads to significant time, energy, and costsavings in the manufacture of complex compo-nents for the automotive, biomedical, aero-space, and robotic industries.

AM volume capabilitiesA number of AM systems can make com-

ponents in the range of inches to several feet.For example, commercial systems currentlyavailable from Stratasys (Fortus 900) andExOne (M-Print) produce polymer and metalcomponents, respectively, measuring two tothree feet along a side. The German companyVoxeljet recently unveiled a continuous powderbed system (VXC800) that allows building andunpacking processes to occur simultaneously,effectively producing unbounded parts along asingle dimension. In the Netherlands, an artistis using a robotically controlled, extruded poly-mer-deposition system to make freeform de-signs several feet long.

The BAAM system at ORNL’s MDF de-posits high-performance engineered thermo-plastics and customized thermoplasticcomposites via melt extrusion processing.BAAM leverages industry standard feedstockmaterials, such as polymer pellets, powders,

ADVANCED MATERIALS & PROCESSES • MARCH 2013 15

Fig. 1 — Big Area Additive Manufacturing (BAAM)system at ORNL’s Manufacturing DemonstrationFacility.

mar amp features_am&p master template new QX6.qxt 2/18/2013 10:23 AM Page 15

fiber reinforcements, and specialty additives ratherthan requiring pre-extruded filament feedstock com-monly used in industry standard extrusion-based sys-tems, such as FDM. A benefit of eliminating filamentas feedstock material is the ability to create high-per-formance composite materials, such as compoundscontaining significant amounts of carbon-fiber rein-forcement. Extruding filaments made of highly rein-

forced compounds flexible enoughto be wound on a spool is very chal-lenging, and is a major limitation toimplementation in functional end-use structures. Therefore, the depo-sition head of the BAAM system isdesigned to combine melting, com-pounding, and extruding functionsto deposit a high-performancepolymer compound at a controlledrate. Deposition-head positioningand movement are controlled usinga multiaxis robotic arm (Fig. 2),which can either be stationary ormounted to a large multiaxis gantrysystem, similar to automated fiber-placement technologies. The depo-sition head could also be mounted

directly to a conventional three-axis gantry (Fig. 3). Inprevious research at ORNL, this gantry system wasused to demonstrate freeform deposition of concretestructures. Figure 4 illustrates a composite structurewhere deposited concrete defines the internal and ex-ternal surfaces of a wall structure, and the internal vol-ume is filled with either insulating or structuralmaterials.

ADVANCED MATERIALS & PROCESSES • MARCH 201316

Fig. 2 — Robotically controlled extrusion-deposition head on theBAAM system.

Fig. 3 — ORNL gantry deposition system used for additivemanufacturing of concrete structures.


www.thermocalc.com

[email protected]

Part size not a limitationThe BAAM system is designed to produce parts that

are completely unbounded in size. It can deposit materialat any point within a build volume defined by a specificrange of motion, followed by the coordination of neighbor-ing build volumes to produce very large components. Inthis configuration, a single deposition head is primary re-sponsible for a specific region of the component, but coor-dinates with neighboring deposition systems to mergeoverlapping sections. In practice, multiple robotically con-trolled deposition systems are configured according to theoverall dimensions of the desired component, simulating a“swarm” of deposition robots, each responsible for a givenarea, but coordinating with its neighbor to produce theoverall component. Figure 5 illustrates the concept for themanufacture of an unmanned aerial vehicle.

Validation of the BAAM system will be initially demon-strated for the application of large, low-cost tooling. Sev-eral process-optimization tasks will be addressed inparallel throughout the advancement of BAAM, such asflow control, spatial resolution, geometric tolerances, ad-vanced materials, and thermal control. Follow-on work willinvolve expanding the concept beyond the manufacture oflarge scale polymer components. Because AM enables in-corporating multiple materials and multifunctional sys-tems directly into a component structure,further developments will involve place-ment of reinforcing materials for struc-tural applications, conductive materialsfor on-board electronics, and pick-and-place of specific subsystems (such as nav-igation, communication, and energystorage). The BAAM platform can also beextended to metallic-materials deposition.

BAAM potentially can revolutionizeAM for large scale, highly complex sys-tems, and radically impact tomorrow’sfleet of aerospace vehicles. Imagine a full-scale squadron of unmanned aerial vehi-cles (UAVs) being produced directly froma CAD file in a “lights out” facility by aswarm of BAAM robots—from “file-to-flight” in a matter of days. Lockheed Mar-

tin and ORNL are accelerating technology to realize thisvision through numerous innovations on the BAAM plat-form.Research sponsored by the U.S. Department of Energy, Officeof Energy Efficiency and Renewable Energy, Advanced Man-ufacturing Office, under contract DE-AC05-00OR22725 withUT-Battelle LLC. The authors acknowledge the technical con-tributions of Ken Cluck, Stephen Wood, and Blake Arthur ofLockheed Martin Corp.

For more information: Chad Duty, Ph.D., Group Leader &Senior Research Staff, Deposition Science & TechnologyGroup, ORNL, tel: 865/574-5059; email: [email protected];www.ornl.gov/manufacturing.


Fig. 4 — Composite concrete wall structuredeposited using AM.

Fig. 5 — Conceptual illustration of using coordinated multiple BAAM robotsto manufacture a large, complex system.


www.clemex.com


Case Study: Additive Manufacturing ofAerospace Brackets

Oak Ridge National Laboratory (ORNL)and Lockheed Martin recently demon-strated the capability of additive man-

ufacturing technologies to drastically reducethe cost and material scrap associated with theproduction of aerospace components. Over thepast decade, layered deposition techniques (i.e.,rapid prototyping) transitioned toward a truemanufacturing platform that can produce theform, fit, and function of a component. This ap-proach, referred to as additive manufacturing(AM), demonstrates the ability to fabricate fullyfunctional complex components at reducedproduction costs and significant material andenergy savings. AM also has the potential to de-crease component lead times and to fabricatecomplex designs not possible through conven-tional processing technologies. Metal AM tech-nologies have not been widely adopted by theindustrial sector partially because informationpertaining to the mechanical integrity of com-ponents and the associated business case arenot well understood. This article discusses acomponent that is well suited for productionusing metal AM techniques, demonstrates themechanical properties of Ti-6Al-4V samples asa function of orientation and layout within thebuild chamber, and examines the associatedbusiness case for production.

Component selectionApplication of metal AM has concentrated

around components made of high-value mate-rials such as Ti-6Al-4V or Ni-base superalloysbecause it increases material use and minimizescostly machining. A Bleed Air Leak Detect(BALD) bracket used in the hot side of the en-gine on Lockheed Martin’s Joint Strike Fighterplatform was selected for this study. The BALDbracket is classified as a tertiary structure underlight loading, which simplifies the certificationprocess and increases the potential to imple-ment AM technology into production. Tradi-tionally, the BALD bracket is machined fromwrought Ti-6Al-4V plate to the specified geom-etry. The bracket has a very thin cross section,resulting in a buy-to-fly ratio of 33:1; meaning33 lb of raw stock plate is purchased to producea 1-lb machined bracket. Because of these inef-ficiencies, AM was expected to be an attractiveproduction method if the appropriate mechan-ical properties and cost structure could beachieved.

Electron beam meltingElectron beam melting (EBM), developed

by Arcam AB (Mölndal, Sweden), is a powder-bed AM technology that fabricates fully densemetal parts by selectively melting specific re-gions within a 50-200 mm layer of powder. Suc-cessive layers are fused together to build up anear-net three-dimensional (3-D) structure.The process is conducted under vacuum toeliminate the potential for contamination/oxi-dation at elevated temperatures. Upon comple-tion of the component, excess powdersurrounding the solid structure is reclaimedand can be used in subsequent builds, which re-sults in very limited waste for each component.An animation of the process can be found athttp://www.ornl.gov/sci/manufacturing/video/direct_manufacturing.shtml.

The purpose of this work was to evaluatematerial properties as a function of location ororientation within the build chamber and to de-termine the economic feasibility of the EBMsystem for the BALD bracket. To go into pro-duction using EBM technology, it is essentialthat material properties are consistent and in-dependent of orientation or location within thebuild chamber. This stems from the limited sizeof current system, requiring large parts to beoriented non-orthogonally to a primary axiswithin the chamber. Economic viability also re-quires that the number of parts included in the

Ryan Dehoff,Chad Duty,William Peter*,Yukinori Yamamoto, Wei Chen, andCraig Blue, FASM*Oak Ridge NationalLaboratoryOak Ridge, Tenn.

Cory TallmanLockheed MartinAeronautics Co.Fort Worth, Tex.

Electron beammelting technology is asuitable additive manufacturingtechnology to producecomplex aerospace components.

*Member of ASMInternational

Fig. 1 — Ti-6Al-4V bleed air leak detect (BALD) bracket fabricatedusing additive manufacturing.


build chamber be maximized. Thus, flexibility in ori-entation maximizes the number and type of parts thatcan be built using the Arcam process at minimizedproduction costs.

The components for this study were fabricated atthe Manufacturing Demonstration Facility at ORNL(www.ornl.gov/sci/manufacturing/mdf.shtml) usingArcam A2 direct manufacturing technology. The partswere made from Ti-6Al-4V powder provided byArcam. Following EBM, components were hot isosta-tically pressed (HIPed) for two hours at a temperatureof 1650°F and pressure of 15,000 psi. The surface fin-ish associated with the EBM process is not suitable in

the as-processed condition, so all components were over-sized by 0.05 in. on all surfaces to allow for final machin-ing. Tensile specimens were post-machined to ASTM E8requirements while the BALD brackets and witnesscoupons were machined to a surface roughness of 125 rms.

Material propertiesFigure 2 shows the layout of a build designed to inves-

tigate variation of mechanical properties, microstructure,and porosity with respect to location within the buildchamber. Tensile samples produced as near-net-shape(NNS) and machined from large blocks were tested to de-termine the impact of the quantity of material on materialproperties. Most of as-deposited material was free ofporosity, but voids as large as 300 mm were detected. HIPeliminated all porosity. The resulting microstructure isshown in Fig. 3.

Tensile tests were conducted on 50 samples from thebuild shown in Fig. 2. A total of seven and nine tensile sam-ples were extracted from each vertical and horizontalblock, respectively, and NNS samples were finish ma-chined. Average tensile properties (Fig. 4) were 127-ksiyield stress, 137-ksi tensile strength, and 14.4% elongation


Fig. 2 — Arcam EBM build layout (left) and as-built samples (right) showinglarge blocks and near-net-shape tensile samples. Samples were fabricatedin both the horizontal and vertical direction at specified locations within thebuild volume.

Front

Fig. 3 — Microstructure ofTi-6Al-4V couponafter hot isostatic

pressing.

0.0010 in

Fig. 4 — Tensile-test results as a function of location and orientation withinthe build chamber. Each location and orientation contains numerous duplicate samples.

Fig. 5 — Layout of BALD brackets within the Arcam build chamber. Colored circles

with black dots inside represent tensile samples parallel to the z axis.

A4F A5F

A6F

A1T

A8FA7F

A2T

A3F A9F

A2FA1F

A3T

0 0.05 0.1 0.15 0.2 0.25Strain

160

140

120

100

80

60

40

20

0

Str

ess

, ksi

Horizontal block from corner 1

Horizontal block from corner 2

Horizontal block from center

Horizontal block from edge

Vertical block from edge

Vertical block from center/edge

Horizontal near net shape

Vertical near net shape


at fracture. Although there is some variation in results, alltensile properties exceed ASTM specification for wroughtTi-6Al-4V plate material. Elongation values at break forNNS tensile bars are higher than those for samples ex-tracted from vertical and horizontal blocks. Based on theresults, it was determined that the Arcam EBM process issuitable to fabricate components requiring wrought Ti-6Al-4V tensile properties.

Bracket fabrication and testingNine BALD brackets were produced in a single build;

the location and orientation of the brackets was variedwithin the build chamber (three groups at the center, cor-ner, and edge) to determine the impact on mechanicalproperties (Fig. 5). Each group consisted of three brackets,oriented with the bracket web aligned with the x, y, and z-axes of the chamber. The layout was used for two buildshaving different powder compositions as shown in Table1. Powder for the first build met AMS4928 chemical spec-ifications. Powder for second run had higher oxygen con-tent than allowed by specification.

Tensile test results are shown in Fig. 6. Yield and ul-timate tensile stress were greater in samples with higheroxygen content, but both sample sets had similar totalelongation. This result is only for simple tensile loading;it is expected that the higher oxygen content would neg-atively impact properties such as fracture toughness andfatigue. Because the design of the BALD bracket is lim-ited by strength, mechanical properties were deemed ad-equate for production. Results also indicate that oxygencontent below the ASTM specification for wrought ma-terial can be achieved using additive manufacturingtechniques.

After EBM manufacturing and HIP, BALD bracketswere machined (Fig. 7) and tested alongside unma-chined brackets for comparison. Testing was conductedin a customized loading device where both tension andtorsional forces were applied. Brackets were subjectedto loading at a ten times factor of safety over serviceconditions. Permanent bending deformation of £0.03 in.occurred along the vertical web, but no brackets failedduring testing despite repeated loading. The observeddeformation is not enough to prevent brackets fromserving their intended purpose and would not impact


TABLE 1 — CHEMICAL COMPOSITION OF SAMPLES FABRICATED USING ARCAM EBM TECHNOLOGY(a)

Chemical composition, wt%

Other, Other,Al V Fe O C N H each total

AMS4928 5.5-6.75 3.5-4.5 0.30 max 0.2 max 0.08 max 0.05 max 0.0125 max 0.1 max 0.4 max

Build in Fig. 2 6.06 4.08 0.07 0.17 0.01 0.02 0.0043 <0.1 <0.4

BALD bracket (build 1) 6.06 4.10 0.07 0.14 0.01 0.02 0.0016 <0.1 <0.4

BALD bracket (build 2) 6.34 4.13 0.15 0.26(b) <0.1 0.02 0.0019 <0.1 <0.4

(a) Chemical composition of the component is identical to that of the starting powder material. (b)Value is outside of ASTM chemical specification.

Fig. 6 — Tensile-test results for vertical samples included in the BALDbracket build and initial sample.

Fig. 7 — As-deposited (top) and post-machined BALD bracket (bottom).

Str

ess

, ksi

0 0.05 0.1 0.15 0.2 0.25Strain

180

160

140

120

100

80

60

40

20

0

Center of build with high oxygen content

Edge of build with high oxygen content

Corner of build with high oxygen contentCenter of build with on-specification oxygen contentEdge of build with on-specification oxygen contentCorner of build with on-specification oxygen contentVertical tensile center of orientation build

Vertical tensile edge of orientation build

Vertical tensile corner of orientation build


the system they support. There was no statistical differ-ence in performance due to differences in surface fin-ish, part orientation, or powder composition.

Cost analysis for productionThe cost to fabricate a BALD bracket using EBM

technology was directly compared to the current pro-duction method of machining the brackets fromwrought plate. Wrought Ti-6Al-4V plate is significantlyless expensive than spherical Ti-6Al-4V powder used in

the EBM process, but the waste associated with machin-ing complex geometries dominates manufacturingcosts. Once scrap rate is included for the manufacture ofa complex component, material costs can exceed$1000/lb of finished product. By comparison, the EMBprocess uses nearly 100% of the raw material and recov-ers any waste for recycling. This reduces the buy-to-flyratio for the BALD bracket from 33:1 to just over 1:1.This study showed that using EBM technology (includ-ing HIP and final machining) would reduce current

bracket cost by more than 50%.

ConclusionsEBM technology is a suitable addi-

tive manufacturing technology to pro-duce complex aerospace components,such as the BALD bracket. As-deposited and HIPed brackets haveconsistent mechanical properties re-gardless of location or orientationwithin the build chamber, meeting theASTM specification for wrought Ti-6Al-4V material for yield strength, ul-timate tensile strength, and elongation.Component testing also satisfied therequired mechanical performance fornonflight-critical hardware. Numerousbrackets can be placed into the buildchamber at various locations and ori-entations to maximize the productionrate without sacrificing componentquality. A simple cost analysis showsthat EBM technology provides a 50%cost reduction over the current pro-duction method. Additional certifica-tion will be required prior to fullproduction.

Arcam is a registered trademark of ArcamAB, Mölndal, Sweden.

Research was sponsored by the U.S. DOE,Office of Energy Efficiency and RenewableEnergy, Advanced Manufacturing Office,under contract DE-AC05-00OR22725 withUT-Battelle LLC.

For more information: Ryan R. Dehoff,Ph.D., Technical Lead for Metal Manufac-turing, Deposition Science and Technol-ogy Group, Oak Ridge NationalLaboratory, Oak Ridge, Tenn.; tel: 865/574-1094; email: [email protected];www.ornl.gov/manufacturing.



www.bruker.com/s1titan

[email protected]


Oak Ridge National Laboratory (ORNL)is leveraging decades of experience inneutron characterization of advanced

materials together with resources such as theSpallation Neutron Source (SNS) and the HighFlux Isotope Reactor (HFIR) shown in Fig. 1 tosolve challenging problems in additive manu-facturing (AM). Additive manufacturing, orthree-dimensional (3-D) printing, is a rapidlymaturing technology wherein components arebuilt by selectively adding feedstock material atlocations specified by a computer model. Themajority of these technologies use thermallydriven phase change mechanisms to convertthe feedstock into functioning material. As themolten material cools and solidifies, the com-ponent is subjected to significant thermal gra-dients, generating significant internal stressesthroughout the part (Fig. 2).

As layers are added, inherent residualstresses cause warping and distortions that leadto geometrical differences between the final partand the original computer generated design.This effect also limits geometries that can be fab-ricated using AM, such as thin-walled, high-as-pect-ratio, and overhanging structures.Distortion may be minimized by intelligent tool-path planning or strategic placement of supportstructures, but these approaches are not well un-derstood and often “Edisonian” in nature.

Residual stresses can also impact compo-nent performance during operation. For exam-ple, in a thermally cycled environment such asa high-pressure turbine engine, residualstresses can cause components to distort un-predictably. Different thermal treatments onas-fabricated AM components have been usedto minimize residual stress, but componentsstill retain a nonhomogeneous stress stateand/or demonstrate a relaxation-derived geo-metric distortion.

Industry, federal laboratory, and universitycollaboration is needed to address these chal-lenges and enable the U.S. to compete in theglobal market. Work is currently being con-ducted on AM technologies at the ORNL Man-ufacturing Demonstration Facility (MDF)sponsored by the DOE’s Advanced Manufactur-ing Office. The MDF is focusing on R&D of bothmetal and polymer AM pertaining to in-situprocess monitoring and closed-loop controls; im-plementation of advanced materials in AM tech-nologies; and demonstration, characterization,and optimization of next-generation technolo-gies. ORNL is working directly with industrypartners to leverage world-leading facilities infields such as high performance computing, ad-vanced materials characterization, and neutronsciences to solve fundamental challenges in ad-vanced manufacturing. Specifically, MDF isleveraging two of the world’s most advanced neu-tron facilities, the HFIR and SNS, to characterizeadditive manufactured components.

Why neutrons?Neutron techniques might be less familiar

than x-ray methods to materials scientists andengineers, but are similar in many ways. For ex-ample, x-ray and neutron diffraction both useBragg’s Law to describe scattering from singlecrystals, powders, and polycrystalline solids.However, one important difference is that neu-trons are much more penetrating than x-rays inmost materials. X-ray scattering occurs withina few microns (with lab sources) to a few mil-limeters (using high-energy synchrotron radi-ation) of the surface, whereas neutrons oftenpenetrate several centimeters. Therefore, neu-trons serve as a nondestructive probe of thebulk microstructure in engineering solids. An-other difference is that x-ray scattering inten-

Neutron Characterizationfor Additive Manufacturing

Thomas Watkins,Hassina Bilheux,Ke An, Andrew Payzant,Ryan Dehoff, Chad Duty, William Peter*,and Craig Blue,FASM*Oak Ridge NationalLaboratoryOak Ridge, Tenn.

Craig Brice,NASA Langley Research CenterHampton, Va.

Nondestructiveexamination of complex additive manufacturedcomponentsusing neutronsis a valuabletechnique forimaging andmeasuringresidualstress.

*Member of ASMInternational

Fig. 1 — (left) Spallation Neutron Source (SNS); (right) High Flux Isotope Reactor (HFIR) at Oak Ridge National Laboratory, Tenn.

Fig. 2 — Thermal image of a layered additive manufacturing process shows significant temperature gradients near the deposition zone[1].



sity increases proportionally to the electron density of amaterial (or approximately to the atomic number Z). Con-sequently, light elements such as hydrogen, boron, andlithium make very little contribution to the overall scatter-ing in the presence of heavier elements. In contrast, neu-tron scattering depends on the neutron scattering length,which is not a function of Z, and which may vary widelyeven for different isotopes of a particular element. Conse-quently, appreciable scattering signal can be obtained fromlight elements using neutrons.

While portable neutron sources exist, most materialscience and engineering work occurs at large, fixed neu-tron sources located at large national and international fa-cilities[2]. Two source types are reactor and spallation.Reactor-based sources (e.g., HFIR) use fission to continu-ously generate a flux of neutrons, while spallation sources(e.g., SNS) generate neutrons by accelerating protons tohigh energies, which hit a target such as tungsten or mer-cury, causing emission of neutron pulses. The spallationprocess is analogous to a cue ball hitting a rack of balls onthe break in billiards. Combining neutron source and de-tectors enables characterization of various aspects of ma-terials.

Neutron computed tomographyConventional two-dimensional (2-D) neutron radiog-

raphy and three-dimensional (3-D) computed tomographyare based on neutron beam attenuation through matter.Attenuation and contrast are directly related to total neu-tron cross section, which depends on neutron interactionswith the atomic nucleus (e.g., coherent and incoherentscattering and absorption of neutrons). Neutron cross sec-tion does not depend on Z in a simple way: in many cases,a low-Z material is readily imaged despite the presence ofsurrounding high-Z material; not easily done using x-rays.Neutron cross sections are known, tabulated, and ex-pressed in units of barn (1 barn = 10-24 cm2). Neutron crosssection can be thought of as the area of each nucleus asseen by the neutron[3] with larger cross sections being moreattenuating. Attenuation coefficient can be measured at acontinuous-source facility, such as a reactor-based neutronimaging beamline (HFIR).

Neutron computed tomography(nCT) reconstructs a 3-D image ofthe contrast due to attenuation ofneutrons inside an object by trans-forming multiple 2-D radiographs ofthe object as it is rotated around itsaxis. Neutron CT was performed atHFIR’s CG-1D cold neutron imagingprototype facility[4] on Inconel 718turbine blades fabricated by additivemanufacturing using direct lasermetal sintering (DLMS) at MorrisTechnologies (Cincinnati, Ohio).Neutron radiographs are generatedusing a LiF/ZnS scintillator thatconverts neutrons into light, which

is subsequently detected by a CCD camera. Spatial res-olution at the sample position is approximately 75 µm.The radiograph and nCT data in Fig. 3 indicate a rela-tively low neutron attenuation through the blade itself,and higher attenuation at the base of the object. Thedark/low transmission region is likely due to the in-creased thickness at the blade base. Some internalair/sample interfaces show an interesting texture thatcould affect airflow through the turbine. After recon-struction, the volume rendering provides a 3-D visuali-zation of the object (Fig. 3 center). Quantitative analysisis often based on transverse slices obtained from the re-constructed piece (Fig. 3 right).

ORNL is developing a world-class neutron imaging in-strument called VENUS (Versatile nEutron imagiNg in-strUment at Sns), which will use the SNS in a unique wayto measure and characterize large-scale and complex sys-tems. VENUS will offer the opportunity to advance scien-tific research in a broad range of areas such as energy,materials, additive manufacturing, transportation, engi-neering, plant physiology, biology, and archeology[5]. It willhave unprecedented spatial resolution (~10 µm), provid-ing 3-D visualization of surface texture and porosity. Theunique capabilities are derived from the intrinsic SNStime-of-flight (TOF) properties, which allow probing thewavelength-discrete neutron attenuation[6]. Energy-selec-tive and Bragg edge-imaging capabilities will provide en-hanced contrast mechanisms, material identification, andthe opportunity to map strain inside samples.

Neutron residual-strain measurementsIn addition to tomography, neutron diffraction can be

used to nondestructively measure residual strain in a ma-terial and determine residual stresses. Strain is determinedthrough Bragg’s Law by measuring distances between crys-tallographic planes of the strained sample and a reference.Figure 4 illustrates how the sample interior is mapped. Thenominal gauge volume is fixed in space on a goniometerand defined by the intersection of projections of incidentand receiving slits. The sample is moved (x, y, and z, androtation) relative to the gauge volume and mapped. Whilediffraction outside the gauge volume occurs along the

Fig. 3 — Neutron radiograph (left), volume rendered (center), and transverse slices at top, middle,and bottom (right) of turbine blade fabricated using additive manufacturing. Blade height is ~76mm.



beam path, it is not recorded because these neutrons areblocked by the receiving slits. The direction of the meas-ured strain component is given by the scattering vector,which bisects the incident and diffracted beams and is per-pendicular to the diffracting planes.

Characterization of EBF3 samplesNASA Langley Research Center (Hampton, Va.) has

been working on large-scale metal additive manufacturingfor the past ten years. The electron beam freeform fabrica-tion process (EBF3) uses a high-power electron beam withwire feedstock to fabricate large structural aerospace com-ponents. These components often exceed a meter in lengthand require many kilograms of deposited material. The in-cremental nature of the additive process combined withthe large scale of EBF3 fabricated components results inhigh levels of distortion and/or residual stress. Under-standing how these residual stresses evolve is critical in de-veloping effective mitigation strategies.

Two applications of primary interest for the EBF3 plat-form are terrestrial-based manufacturing of aerospacestructures and remote, space-based fabrication of hard-ware. For the former, knowledge of residual stress accumu-lation helps develop fixturing, path planning, andintermediate heat treatment strategies minimize through-put time and out-of-tolerance distortion. For the latter,knowledge of the residual stresses helps predict final bulkproperties, as heat treatment may not be available forstructures fabricated extraterrestrially.

NASA recently initiated neutron diffraction measure-ments on EBF3-deposited aluminum samples using theVulcan Engineering diffractometer at SNS. Initial meas-urements evaluated the impact of an intermediate thermalanneal treatment on residual stress. The stress relief heattreatment shifted the residual stresses from the substrateinto the deposited material. This is important for the EBF3

process where the substrate plate is usually incorporatedinto the final part and often accounts for the bulk of out-of-plane distortion. These data were used to develop basicmodels using commercially available finite element mod-eling software. Additional trials at SNS scheduled for early2013 will help validate the early models and provide a pre-dictive capability to minimize residual stress and distor-tion in large-scale additive structures.

Characterization of DLMS turbine bladesExperiments were also conducted on the VULCAN

beam line to determine if residual strains could be meas-ured in complex components with internal structure,which is also of interest elsewhere[8]. A series of turbineblades were fabricated from Inconel 718 using DLMS tech-nology. The blades have a wall thickness down to 1 mmand curved internal cooling passages (see Fig. 3). Threeblades were analyzed at VULCAN together with a strain-free reference powder (Fig. 5). Slits on both the incidentand diffracted beams give a gauge volume of 2 ́ 2 ´ 2 mm.Two detector banks (not shown) are located to the left andright of the samples allowing for simultaneous collection of

two orthogonal strain components.Time-of-flight measurement gives concurrent analysis

of a series of Bragg reflections. Given the material’s elasticanisotropy[9], only residual strains from the (311) reflectionare reported here as the modulus for the (311) reflection isclose to the bulk average. Figure 6 shows residual strainsand their position within the blade for two samples in theas-built and stress relieved condition. Residual strain mapsare superimposed on the representative “slice” of the bladefrom an auto-cad drawing. Slices A, B, C, D, and E are at15, 12, 10, 8, and 0 mm up from the origin on the dovetailbase, respectively. The arrow in slice A indicates the direc-tion of the strains (ppm) at each location. The green andblue values originate from the as-built and stress relieved

Incident slits

Direction of strain measurement

Diffracted beam

Sample 2q

Diffracted slits or radial collimators

Transmitted beam

Fig. 4 — Schematic of neutron diffraction set up viewed from above showsnominal gauge volume (dark square) as defined by the slits[7].

Fig. 5 — Precision alignment of the turbine blades and powdersample (top) in the VULCAN beam line (bottom).

Incid

ent

beam

Dete

cto

r(s)


conditions, respectively. Each square and diamond in Fig.6 is nominally 1 and 4 mm2, respectively.

In the as-built blade, strains parallel to the long axis ofthe blade are more compressive on the convex surface thanthe concave surface, which may be related to sequencing inthe DLMS build process. In general, residual strains in theas-built sample become more tensile (or less compressive)as the dovetail base is approached. At the base, residualstrains return to being more compressive. After a propri-etary stress relief process, residual strains become very

compressive and much more uniform throughout thelower portion of the blade, suggesting that the blade willbe less likely to distort during service. The combination ofneutron tomography and residual strain measurement is acritical approach for examining complicated componentsthat can be built using additive manufacturing.

HFIR residual stress mappingPreliminary experiments are underway at the HFIR’s

NRSF2 beamline (HB-2B, Fig. 7) to map stress distribu-tions in complex parts produced by additive manufac-turing. Whereas the VULCAN beamline allows forsimultaneous acquisition of strains from multiple crys-tallographic planes, HB-2B selects a single diffractionpeak as the strain sensor. Peak data are recorded for a se-ries of sample locations and orientations to produce aresidual stress map throughout the structure. Diffractionpeak shifts are a measure of changes in atom-to-atomspacing, and may be related to tensile or compressivestrains in the sample. Peak intensity changes may indi-cate preferred orientation, or changes in sample compo-sition, and peak width may be related to crystallite size,dislocation density, and other microstructural features.The facility is being used to measure residual stress insimple prismatic shapes fabricated using various AMtechniques to gain a fundamental understanding ofprocess parameters on stress evolution. This informationwill be used to verify process models and stress mitiga-tion strategies currently under development.

SummaryThe use of neutrons to characterize additive manu-

factured components is a unique, valuable technique forimaging and measuring residual stress. Neutron and x-ray techniques can be used in a complementary fashion,providing information about the bulk and surface of acomponent, respectively. Neutrons are useful to charac-terize complex components, such as those fabricatedusing additive manufacturing techniques. In this article,the penetrating power of neutrons facilitated uniquecharacterization of aircraft parts fabricated using addi-tive manufacturing, imaging internal structures and pas-sageways and mapping residual strains in a complexturbine blade. Future projects combining these tech-niques will be required to fully use the potential of addi-tive manufacturing.

Research sponsored by the U.S. DOE, Office of Energy Effi-ciency and Renewable Energy, Advanced Manufacturing Of-fice, under contract DE-AC05-00OR22725 with UT-BattelleLLC. Research conducted at ORNL’s High Flux Isotope Reac-tor and Spallation Neutron Source was sponsored by the Sci-entific User Facilities Div., Office of Basic Energy Sciences,Department of Energy.

References1. R.B. Dinwiddie, et al., Real-time Process Monitoring andTemperature Mapping of the 3D Polymer Printing Process,Thermosense XXXV, Proc. SPIE, to be published 2013.2. www.ncnr.nist.gov/nsources.html.


Fig. 6 — Residualstrain maps (in

microstrain) of aturbine blade

fabricated usingadditive

manufacturing inthe as-built (greenvalues) and stress

relieved (blue values) conditions.

A

B

C

D

E



3. M.T. Hutchings, et al., Introduction to the Characterizationof Residual Stress by Neutron Diffraction, CRC Press, BocaRaton, Fla., p 27-32, 2005.4. H. Bilheux, et al., Neutron imaging at the Oak Ridge Na-tional Laboratory: present and future capabilities, Physics Pro-cedia (in press).5. I.S. Anderson, et al., Neutron Imaging and Applications: AReference for the Imaging Community, Series: Neutron Scat-tering Applications and Techniques, Springer, 2009.6. A.S. Tremsin, et al., High-Resolution Strain MappingThrough Time-Of-Flight Neutron Transmission Diffractionwith a Microchannel Place Neutron Counting Detector, doi:10.1111/j.1475-1305.2011.00823, 2011.7. P.J. Webster, et al., Problems with Railway Rails, Measure-ment of Residual and Applied Stress Using Neutron Diffrac-tion, Proc. NATO Advanced Research Workshop, NATO ASISeries E: Applied Sciences, V 216, Kluwer Academic Publish-ers, Boston, p 517-524, 1992.8. S. Pierret, et al., Combining neutron diffraction and imag-ing for residual strain measurements in a single crystal tur-bine blade, NDT&E Intl., 45, p 39–45, 2012. 9. T.M. Holden, et al., Intergranular strains in Inconel-600 andthe impact on interpreting stress fields in bent steam-genera-tor tubing, Matls. Sci. and Engrg., A246, p 180–198, 1998.

For more information: Thomas Watkins, Ph.D., GroupLeader & Senior Research Staff, Scattering and Thermo-physics Group, MS&T Div., ORNL, tel: 865/387-6472; email:[email protected]; http://www.ornl.gov/sci/physical_sciences_directorate/mst/dtp/watkins.shtml.

Fig. 7 — HB-2B beamline at the High Flux Isotope Reactor.


www.TiniusOlsen.com

Documents

Big Area Additive Manufacturing