Welding Journal : January 2018 - DOI

Introduction andMotivation

In 2000, the American Welding Soci-ety developed a road map (Ref. 1) withthe following goals: 1) Increase the usesof welding by 25%, decrease the cost,and increase the productivity; 2) en-hance the process technology that al-lows for the use of welding across allmanufacturing sectors; 3) develop newwelding technology along with new ma-terials so that it can be used for all ap-plications; 4) ensure that welding can bepart of the six-sigma quality environ-ment; 5) increase the knowledge base ofpeople employed at all levels of thewelding industry; and 6) reduce energy

use by 50% through productivity im-provements. To meet these goals, a re-search and development pathway (Ref.2), spanning from 2000 to 2040, thatintegrates the process and computa-tional modeling, including process-based quality was articulated — Fig. 1.Interestingly, rapid qualification ofgoods and components is critical to ex-isting applications that are based on tra-ditional supply chain with slow and seri-al flow of information, materials andperformance expectations, as well asemerging trends such as smart manu-facturing that will rely on an agile andintegrative approach facilitated by closecoupling of technologies (Refs. 3, 4) rel-evant to the whole manufacturing lifecycle. The manufacturing life cycle (Ref.

5) involves design, selection of materi-als, selection of process and environ-ment, deployment of robotics and au-tomation, certification through destruc-tive testing of simulative samples andnondestructive evaluation (NDE), fol-lowed by recycling and rejuvenation atthe end of useful life. Since fusion weld-ing, solid-state joining, brazing, and sol-dering are crucial to the above manufac-turing life cycle (Ref. 6), this paper ex-plores the feasibility of ensuring thequality of welded components throughmodification of process parametersbased on a fundamental understandingof microstructural evolution. Note thatthis paper is not a comprehensive re-view, and readers are referred to variouspublications throughout the documentfor an in-depth review.

Current Methodologiesand Process-BasedQuality

The AWS Welding Handbook classi-fies the technologies relevant to thejoining of advanced materials intoeight categories including arc welding,solid-state joining, resistance welding,oxyfuel gas welding, brazing, solder-ing, and other allied processes. A quickaccounting of these categories andsubclassifications shows there aremore than 70 processes (Ref. 7). At thesame time, there are hundreds ofstructural metals and alloys based oniron, aluminum, copper, magnesium,

WELDING RESEARCH

JANUARY 2018 / WELDING JOURNAL 1-s

SUPPLEMENT TO THE WELDING JOURNAL, JANUARY 2018Sponsored by the American Welding Society

Toward Process-Based Quality through aFundamental Understanding of Weld

Microstructural Evolution

Qualifying welded components by integrating software and hardware tools that arecapable of describing physical processes that occur during welding was explored

BY SUDARSANAM SURESH BABU

ABSTRACT In this overview paper, the possibility of qualifying welded components by inte-grating software and hardware tools that are capable of describing physicalprocesses that occur during welding is explored. The existing and emerging toolsto define geometrical boundary conditions, process parameters, heat and masstransfer, solidification, solid-state transformation, plastic deformation, distortion,residual stress, and performance of welded components are reviewed. Existingchallenges relevant to the fundamental understanding of complex alloying,processing environment, and transients in energy deposition are discussed withreference to microstructure evolution. Approaches to address these challengeswere articulated with published case studies relevant to welding and additive manufacturing.

KEYWORDS • Process-Based Quality • Microstructural Evolution • Energy Deposition • Additive Manufacturing • Residual Stress

https://doi.org/10.29391/2018.97.001

Babu (Adams Lecture) January 2018.qxp_Layout 1 1/24/18 8:47 AM

titanium, zinc, tin, lead, nickel, cobalt,zirconium, hafnium, and precious andrefractory metals (Ref. 8). As a result of the large number ofpermutations that is possible based onthe above options, the industry hastaken a cautious and pragmatic ap-proach of deploying joining processesthrough simulative tests, qualification,and certification defined by the stan-dards (Ref. 9). These are usually docu-mented in terms of Welding ProcedureSpecifications (WPSs) (Ref. 10) andProcedure Qualification Records(PQRs) (Ref. 11), and often used astransactional documents between theend user of welds and the provider offabrication services. These founda-tional standards published by AWS(Refs. 12, 13) and other organizationshave gained the confidence and trustof industry allowing for welding to beused as a common tool for critical ap-plications. For example, the AmericanPetroleum Institute (API) has pub-lished standards for welding ofpipelines (API 1104) that specify vari-ous aspects including equipment, ma-terials, manual and automatic proce-dures, welding with and without weld-ing wires, welders, design and prepara-tion of joint geometry, inspection andtesting of welds, repair and removal ofdefects, procedures for nondestructivetesting, and acceptance tests based onNDE (Ref. 14). With emerging criticalapplications, these standards may be-come prescriptive and will not allowflexibility to minimize the variabilityof expected performance in service. Asa result, the engineering advance will

be a single point solution specific to agiven geometry and application. In addition, because WPSs andPQRs are developed with an extensiveamount of resources, they are notopenly disseminated to the public inpeer-reviewed papers. As a result,many innovative and engineering so-lutions may remain proprietary andpart of transactional documents. Withevery new application, every entityhas to repeat the above procedures,again and again, thereby limiting theagility of the welding and joining in-dustry to adopt new processes and ma-terials while maintaining the trust ofcustomers. This leads to the funda-mental question: is there a possibilityof qualifying welded components withvarying geometries by characterizingthe processes within the context of ex-isting standards with minimal trialand error experimentation? Process-based quality is defined asmaterial and information flow whereall aspects of welding life cycle, i.e., de-sign, materials, process, and estima-tion of performance are consideredand evaluated simultaneously. A typi-cal process-based quality flow will in-volve concurrent activities rangingfrom modeling to making and measur-ing. In the modeling step, the geome-try of the component (e.g., fillet weldgeometry) for a given application isdesigned with knowledge from materi-al properties (e.g., yield strength andresidual stress) predicted by integratedprocess models for a given process(e.g., laser welding) based on bound-ary conditions (Ref. 15). In the making

step, the boundary conditions (e.g.,welding current, weaving, speed, re-straints, etc.) are recorded and provid-ed as input to the modeling activity. Inthe measuring step, responses of thematerial to process (e.g., temperaturedistribution, geometric displacements,and cracking) are measured using in-situ sensors. With the informationfrom all these activities, a welding en-gineer may be able to qualify a compo-nent for final service with minimal de-structive testing and NDE.

Existing Tools forProcess-Based Quality

Many aspects of the process-basedquality approach have already been ad-dressed by welding researchersthrough development of empirical,phenomenological, physics-basedmodeling tools for computational weldmechanics (Refs. 16, 17), as well as,deployment of in-situ sensors (Ref.18), controls (Ref. 19), and data log-ging devices (Ref. 20). These tools fo-cus on comprehensive descriptions ofboundary conditions for welding (Ref.21), heat and mass transfer (Refs. 22,23), solidification (Ref. 24), solid-statephase transformation (Refs. 25, 26),transient thermal stresses (Ref. 27),elastic and plastic deformation (Ref.28), incipient crack formation (Ref.29), residual stress, and distortionevolution (Ref. 30). These aspects arebriefly reviewed below.

Description of Geometry,Material, and ProcessingBoundary Conditions

It is well known from engineeringfundamentals that boundary condi-tions have a great influence on theoutcome of any model predictions. Forexample, in welding and joining, theseboundary conditions include the ini-tial geometry of the joints, restraintsused for holding the joints, thermalconductivity of the substrate, locked-in residual stress within the incomingmaterials, variations in compositionsof the base material and filler materi-als, and transients in the heat sourcemovement. Pahkamaa et al. (Ref. 31) havedemonstrated the complex interrela-tionship between variations in geome-

WELDING RESEARCH

WELDING JOURNAL / JANUARY 2018, VOL. 972-s

Fig. 1 — Schematic of research and development roadmap spanning four decadesdeveloped by AWS in the year 2000.

Babu (Adams Lecture) January 2018.qxp_Layout 1 12/12/17 3:42 PM

try welded assembly and evolvingresidual stress and distortion. In an-other example, microstructural refine-ment in aluminum alloys was achievedby just changing the backing plate ma-terials from stainless steel to copper(Ref. 32) for the same processing con-ditions. This approach had also beenextended to stainless (Ref. 33) to mod-ify the microstructure. In another ex-ample, microstructure and propertiesin steel welds were found to vary sig-nificantly on moving from stringer toweaving mode of welding, while keep-ing the same average heat input condi-tions (Ref. 34). This is attributed tothe onset of local brittle zones due totransient variations in the thermal sig-nature at localized regions. In anotherexample, the role of initial microstruc-ture (e.g., distribution of carbides andchemical segregation) in advancedhigh-strength steels (AHSS) and its re-sponse to welding was attributed toimproved ballistic properties in com-parison to welds made with homoge-

neous materials (Refs. 35, 36). Similar-ly, during spot welding a small changein the electrode curvature led to largechanges in weld nugget development— Fig. 2. Using incrementally coupledelectrical, thermal, and mechanicalmodels, the above changes were attrib-uted to the changes in the spatial andtemporal variation of current densityacross the electrode and steel sheet in-terfaces. Although limited, the aboveexamples clearly stress the need for in-situ monitoring of the geometry, ma-terial, and processing boundary condi-tions. With the advent of commercialdata logging devices, rapid and mobilethermal, optical (Ref. 37), and phonon(Ref. 38) tools, it is indeed possible todefine some aspects of the geometry,material, and boundary conditions.

Prediction of Heat and MassTransfer

As early as the 1940s, it was clearthat for qualification of welds, it is

necessary to predict temperature dis-tribution around arc welds. Rosenthalpublished a classic paper that solvedthe heat flow in a quasi-steady statecondition in infinite or semiinfiniteplates (Refs. 39, 40). The above solu-tions have been extended to otherprocesses and conditions by variousauthors (Ref. 41). With the advent ofcomputational hardware and finite ele-ment methodologies, the heat transfermodels above were to consider specificgeometries and boundary conditions,as well as transient conditions. For ex-ample, Yang and Babu (Ref. 27) usedthe finite element models to capturethe transients (Fig. 3) in heat-transferconditions during laser cladding ofnickel-based alloys. Although, the above models do notconsider the fluid flow effects, theyhave been extensively used in industryThe role of fluid flow on description ofthe melt pool shape was pioneered byresearchers including Kou (Ref. 42),Zacharia (Ref. 43), and DebRoy (Ref.

WELDING RESEARCH


Fig. 2 — Predicted temperature distribution (blue — room temperature; red — melting point of steel) during spot welding of a steelsheet at a different number of current cycles is shown in a quarter symmetry. The results show that flat electrodes (A and B) lead to atoroidal-shaped weld pool after 7 7⁄8 cycles, while the curved electrodes lead to stable hemispherical spot weld growth immediatelyafter 2 7⁄8 cycles for similar current levels.

A

C

B

D


44). All of these models share a similarflow of information, starting with pro-cessing conditions (i.e., mode of energydelivery); material addition (e.g., weld-ing wire feed rate); joint geometry; anddefining the boundary conditions forconductive, convective, and radiativeheat transfer. Then, for these boundaryconditions, the melting, fluid flow, andcooling to ambient temperatures arecalculated by solving the conservationof mass, momentum, and energy. Theoutputs of the model include tempera-ture variations at different locations,melt pool shape, thermal gradients (G),and liquid-solid interface velocity (R).The above data can be used to predict

the solidification, sol-id-state transforma-

tion, and deformation characteristics inthe weld metal (WM) and heat-affectedzones (HAZ).

Prediction of Phase Stability

The thermodynamic and kineticframework for describing liquid-to-solid (Ref. 45) and solid-to-solid phasetransformations (Ref. 46) are well es-tablished in physical metallurgy text-books (Ref. 47). Based on many decadesof research and development across theworld, commercial solutions exist forcalculating phase diagrams for multi-component metals and ceramic sys-tems. In the above theoretical con-

struct, for calculating the phase dia-grams and interface or interphase sta-bilities, the following equality (Equation1) is assumed, i.e., the chemical poten-tial of ith element is the same at eitherside of the interface between the twophases, P1 and P2 (Refs. 48, 49).

This equality is also known as thetie-line construction or common tan-gent between molar Gibbs energy vs.composition curves for two phases, ina graphical representation of the phasediagram. The chemical potential of theith element in a multicomponent sys-tem can be calculated using the follow-

μiP1 =μi

P2 (1)

WELDING RESEARCH


Fig. 4 — Calculated stability diagram for Ti(CN) carbo-nitridein steel melts as a function of titanium and carbon concen-tration with different levels of dissolved nitrogen showed thesensitivity to small additions of nitrogen to the system.

Fig. 5 — Schematic of integrated computational weld mechan-ics approach proposed by Kirkaldy (Ref. 72).

Fig. 3 — Predicted temperature distribution during laser cladding for repairing of nickel alloys at the following: A — Early stages; B — middle of the build. The calculations show the transients in temperature distribution due to changes in the heat sink broughtabout the change in the build geometry.

BA


ing expression (Equation 2) that re-lates to the molar Gibbs free energyand partial derivative of the Gibbs freeenergy at a given temperature.

The molar Gibbs free energy of thephase P1 is given (Equation 3) by thecontribution from the pure element(Gi

0), ideal mixing, and excess energy ofmixing defined by the parameter Ωij

v.

In the above equation, R is the gasconstant, T is the temperature, xi corre-sponds to the mole fraction of ith ele-ment in the solid solution, and v is apower factor that takes the value fromzero to two. Similar equations also existfor the description of the molar free en-ergy of compound. For example, it ispossible to calculate the equilibrium be-tween liquid steel and Ti(CN) carbo-nitride as a function of dissolved car-bon, titanium, and nitrogen concentra-tion — Fig. 4. These calculations mayallow us to estimate the role of dis-solved nitrogen on inducing a widerange of inclusions. Such diagrams canbe used to develop an inoculation strat-egy to arrive at the equiaxed grainstructures in welds (Ref. 50). With the

above formulations, it is indeed possibleto predict the multicomponent phasediagrams relevant to weld metal con-sumables, as a function of compositionand temperature.

Prediction of Liquid-SolidTransformation

The solidification conditions duringwelding are related to the liquid-solid(l/s) interface instability. Phenomeno-logical theories describing l/s interfacestability under unidirectional thermalgradients are well known (Ref. 51) inthe casting (Ref. 52) and welding litera-ture (Ref. 24). It is indeed possible topredict columnar to equiaxed solidifica-tion transition (CET) as a function ofthermal gradient (Gl/s) and l/s interfacevelocities (Vl/s)) based on dendrite tiptemperature (Td), with interface re-sponse function theories and by solvingthe coupled equations (Ref. 53) that de-scribe the (kvi) kinetic solute partition-ing coefficient, (mv

i) nonequilibriumslope of the liquidus, dendrite tip radius(R), and (cil/s) concentration at the l/s in-terface, as a function of Vl/s and Gl/s.

In Equations 4–9, koi is the equilibri-um partitioning coefficient between theliquid and solid, Di is the interphase dif-fusivity, mo

i is the equilibrium liquidusslope, ao is the characteristic diffusiondistance, is the Gibbs-Thompson coef-ficient, Pei is the Peclet number given byVl/s/(2Di), is the interface kinetic coef-ficient, and Iv{} is the Ivantsov functionthat depends on the Peclet number. Us-ing thermodynamic information for amulticomponent system, Equations 4–9can be iteratively solved to predict thedendrite tip temperature and therebyconstitutional supercooling, as well asCET in a wide range of alloy systems.These equations can be extended tononequilibrium eutectic growth as well.This phenomenological model has beensuccessfully applied to a wide range ofmulticomponent alloys (Ref. 54). Fundamentals of these theories andextension to welding are reviewed inthe paper by David and Vitek (Ref. 24).With the above formulations, oneshould be able to estimate the solidifi-cation morphologies (planar, colum-nar, equiaxed), dendrite arm spacing,texture, and alloying element segrega-tion as a function of thermal gradients(G) and liquid-solid interface velocity(R), predicted by heat and mass trans-fer models.

Prediction of Solid-StatePhase Transformation

The microstructural evolutions dur-ing welding of steels, titanium, andnickel alloys are also related to solid-

kvi = ko

i +ao Vl/s / Di( )/ 1+ao Vl/s / Di( ) , 4( )

mvi =mo

i 1= kvi 1= ln kv

i / koi{ }( )

/ 1= koi( ) 5( )

4+ 2 1/R2( )+(2 i mv

i Pei 1–kvi( )cl /si* c

l)(1/R)+G s/i = 0, (6)

where ci =1=2kv

i /

2kvi 1+ 1+ 2 / Pei( )( )0.5 7( )

cl/si* = co

i 1= 1 kvi Iv Pei{ }( ) 8( )

Td =TL + i cl/si* mv

i coimo

i( )=2 / R Vl/s /μ Gl/sDi / Vl/s( ) 9( )

GmolarP1 = i xiGi

0 +RT i xi log xi

+ i xij>1 x j v ijv xi x j( )v 3( )

μ iP1 =Gmolar

P1 + j=2r

ij x j( ) GmolarP1

xj;

where ij =0 for i≠j and

ij =1 for i=j 2( )

WELDING RESEARCH


Fig. 6 — Scanning electron microscopy images from the following: A — High-Al weldexhibits AlN inclusion; B — low-Al weld shows both Al2O3 and Ti(CN) inclusions.

A B C

Table 1 — Composition of Self-Shielded Weld Metal (wt-%)

ID C Si Mn Si Al Ni Ti O N Balance

high-Al 0.234 0.28 0.50 0.28 1.70 0.02 0.003 0.006 0.064 Fe low-Al 0.149 0.30 0.64 0.30 0.53 0.01 0.058 0.030 0.033 Fe


solid (s/s) interphase motion eitherthrough a reconstructive or displacivemechanism (Ref. 55). Let us revisit ex-isting theories for interphase stabili-ties for s/s phase transformationsoriginating from the classic papers ofHultgren published in 1947 (Ref. 56)and Coates (Refs. 57, 58). Underisothermal conditions, the velocity ofthe interface vint between the two sol-id-state phases, P1 and P2, can be de-scribed by Equation 10.

In the above equation, ciP1P2 is theinterface concentration of solute “i” inthe P1 phase, and ciP2P1 is the interfaceconcentration of solute “i” in the P2

phase. Similarly, DiP1 and Di

P2 are thediffusivities of the solute in the P1 andP2 phases, respectively. In a multicom-ponent system, the above equationshave to be solved simultaneously forall solute elements. Due to rapid cool-ing and/or large differences in the dif-fusivity, the equality represented byEquation 10 may break down and mayinduce constrained equilibrium. Forexample, paraequilibrium (Ref. 59)growth of ferrite into austenite insteel is controlled only by carbon dif-fusion, and the substitutional ele-ments (Ref. 60) are configurationallyfrozen. In the case of displacive trans-

formation, one can describe the sameby coupling thermodynamics and the-ories of nucleation of martensite onlattice defects (Ref. 61). Bhadeshia etal. (Ref. 62) have extended to predictthe microstructural evolution in low-carbon steel weld metals. In the early1980s, due to limited access to ther-modynamic data, a semi-empirical ap-proach was based on carbon equiva-lence, austenite grain size, and thetime taken to cool from 800° to 500°C(Refs. 63, 64). However, these phe-nomenological models are not capableof describing the 3D microstructuralevolution. In this regard, the phasefield models are very beneficial (Ref.65) and have been used for describingboth solidification and solid-statetransformation in a unified way. Withthe above tools, it is possible to predictthe microstructural features such asphase fractions, grain size, and mor-phology as a function of weld compo-sition for a given thermal signature.

Prediction of ThermalStresses and PlasticDeformation

One of the critical issues with weld-ing is the evolution of residual stressdistribution due to nonuniform accu-mulated plastic strains within the builds(Ref. 66). Under certain cases, thermalstresses can also lead to prematurecracking during weld cooling conditions(Ref. 67). The plastic strain evolution

during repeated thermal cycling is calcu-lated based on thermal stresses broughtabout by macro-scale thermal gradients,local changes in thermal expansion co-efficients, crystallographic misfit be-tween phases, and constitutive stress-strain properties measured underisothermal conditions (Ref. 68). Theconstitutive properties for thermal sim-ulation can be described either by ex-perimental measurements or by cali-brating constitutive relationships devel-oped by Zener-Holloman (Ref. 69) orJohnson-Cook (Ref. 70) models. Re-cently, the Zener-Holloman model wasused to describe the flow properties ofTi-6Al-4V alloys under torsional defor-mation conditions as a function of tem-perature and strain rate (Ref. 71). In ad-dition, there are ongoing worldwide ef-forts to move away from the phenome-nological models to describe the consti-tutive properties through crystal plas-ticity models (Ref. 72). With the abovetools, it is indeed possible to predict theconstitutive properties of metal defor-mation during typical thermomechani-cal conditions induced by spatial varia-tion of thermal gradient and thermalsignature typical to that of welding.

Integrated Process andPerformance Models

Although each and every predictivetool is important in its own sense, forarriving at process-based quality, all ofthe above tools need to be integrated

ciP1P2 � ci

P2P1( )vint= �Di

P1 dcidx int

P1

�DiP2 dci

dx int

P2

10( )

WELDING RESEARCH


Fig. 7 — Optical microscopy images from A — High-Al weld shows skeletal columnar -ferrite interspersed with residual austenitethat has transformed to bainitic microstructure. B — In contrast, the low-Al weld undergoes classic phase transformation to grainboundary ferrite and bainitic microstructure that is expected from 100% austenitic parent phase.

A B


within a single framework. This needwas articulated by Kirkaldy in the1990s (Ref. 73) with a classic diagramshown in Fig. 5 for predicting the steelweld properties. In his proposed inte-grated framework, hardness and prop-erty distributions, as well as residualstress and distortion in steel welds canbe predicted as a function of boundaryconditions (thermal, mechanical),thermo-physical-chemical propertiesby exchanging the information be-tween submodels for heat and masstransfer, solidification, solid-statetransformation, and mechanical re-sponse. As soon as one understandsthe spatial variations, the above datacan also be mapped into a perform-ance model to predict the static anddynamic properties of welded compo-nents (Ref. 74). Interestingly, the above modelingframework has become the founda-tional tenet for many of the existingcommercial solutions for integratedweld modeling. In 2007, the above in-tegrated process modeling frameworkwas deployed within a high perform-ance computational framework capa-ble of on-demand modeling over theinternet (Refs. 75, 76). At this junc-ture, it needs to be stressed that appli-cability of these welding simulationtools always require some amount ofcalibration due to variability in ther-mo-physical-chemical properties (Ref.

77). In order to develop confidence incomputational modeling, there is aneed to develop verification and vali-dation (V&V) standards. Recently, theabove V&V document (Ref. 78) wasdeveloped and published by AWS,which enables the adoption of thesetools by industries. Furthermore, there is ongoing workon validation of residual stress predic-tion (Ref. 79) by the Nuclear ResearchCouncil (NRC) and Electric Power Re-search Institute (EPRI). These studieshave shown the importance of materi-al constitutive properties and the or-der of weld bead placement affectingthe annealing on the variability ofresidual prediction. For example, thevariations in predicted residual stress-es may exceed more than ± 200 MPa(Ref. 80). In addition, extensive workis ongoing to understand the residualstress evolutions based on phasetransformation kinetics and its effecton constitutive properties (Refs. 81,82), especially for low-temperature-transformation (LTT) wires.

Unresolved ScientificChallenges Based on the brief literature reviewon integrated process modeling, onemay be tempted to conclude there areno scientific or technical challenges for

process-based quality. However, thereare two major challenges that need tobe addressed with reference to mi-crostructural evolution in welds, as de-scribed below.

Complex Alloying and Shielding Environments

In most of the published literaturerelated to integrated process modeling,the compositions within the base metalor weld metal region are often consid-ered to be uniform. Nevertheless, it iswell known that small changes in theliquid-slag (Ref. 83) or liquid-gas (Ref.84) reactions may lead to variations indeoxidizing elements (e.g., Ti and Al),which in turn may lead to large changesin inclusions, solidification, and solid-state transformation microstructure(Ref. 85), as well as associated proper-ties (Refs. 86, 87). In addition, by changing from non-reactive to reactive gas shielding, it ispossible to induce in-situ alloying andarrive at hard particles to induce wearresistance (Ref. 88). Although therehas been pioneering work by weldingresearchers to understand the liquid-slag reactions (Ref. 89), dissolution ofgases (Refs. 90, 91), evaporation (Ref.92), and condensation (Ref. 93), thereare no integrated models to predictthese reactions a-priori, without cali-

WELDING RESEARCH


Fig. 8 — A — Calculated stability diagram for high-Al weld supports the preferential formation of AlN inclusions within liquid steel;B — similar calculations with low-Al weld composition shows the preferential formation of Al2O3 and Ti(CN) and not AlN.

BA


bration. Therefore, there is a need toimprove the capability of computa-tional thermodynamic tools to addressthe above phenomena.

Transients in WeldingConditions

In most of the engineering solutionsrelevant to welding, steady-state condi-tions are valid to predict thermome-chanical changes. However, in certaincases involving manual welding for ver-tical and overhead configurations, auto-matic welding with weaving and varyinggeometrical cross sections, the above as-sumption may not be valid. These tran-sients may lead to local changes in mi-crostructures that may lead to scatter inproperties (Ref. 94). Furthermore, dur-ing welding with large energy density,thermal gradients may exceed 106 K/m.Most of the phase stability and kineticmodels for solid-state transformationswere developed under isothermal condi-tions. Currently, the validity of the localequilibrium assumed by Equation 1 forthese conditions is not known. Further-more, in multipass welds, thermal sig-natures may induce repeated dissolu-tion and growth under severe thermalgradients (Ref. 95). Therefore, there is aneed to understand the effect of thesetransients and development of method-ologies to track the same under in-situconditions (Refs. 96, 97) and map thesame to integrated process models toarrive at process-based quality.

Scientific Approaches toAddress Challenges

In this paper, four case studies arepresented with increasing complexitiesto demonstrate the approaches thatcan be used to address the fundamen-tal challenges relevant to weld mi-crostructural evolution. In order toprovide continuity of the discussions,the first three case studies were re-stricted to self-shielded flux cored arcweld metal with an Fe-C-Al-Mn alloysystem. The results from each casestudy in the Fe-C-Al-Mn alloy systembuild on the results from the previousones. Finally, extension of these con-cepts to emerging metal additive man-ufacturing is also presented.

Case Study 1: Role of Thermodynamicand Kinetic Modeling

Problem Statement: Kotecki andMoll (Refs. 86, 87) found that inclu-sion formation and microstructuralevolution in self-shielded flux coredarc (FCAW-S) welds were quite differ-ent from that of low-alloy steel weldsmade by submerged arc (SAW), shield-ed metal arc (SMAW), and gas metalarc (GMAW) welding. Since theFCAW-S processes do not use anyform of gas shielding, the dissolvedoxygen and nitrogen from atmospherein the liquid steel is removed by largeadditions of aluminum. In this casestudy, the possibility of predicting

these complex microstructural evolu-tions was explored using ThermoCalc®and DicTra® software (Refs. 98, 99). Approach: Microstructural evolu-tion in weld metal compositions withtwo different aluminum concentra-tions (Table 1) were characterized.Then the observed microstructuralevolution was compared with the pre-dictions from computational thermo-dynamic and kinetic models. Results: Scanning electron mi-croscopy showed welds with high alu-minum concentration exhibited onlyAlN-type inclusions — Fig. 6A. In con-trast, the weld with low aluminumconcentration contained both Al2O3

and Ti(CN)-type inclusions — Fig. 6B,C. Interestingly, the high-Al welds alsoexhibited columnar -ferrite (BCCcrystal structure) network inter-spersed with bainitic ferrite () mi-crostructure that formed from stableaustenite () — Fig. 7A. This resultsuggests that high-Al welds do not gothrough 100% austenite phase field. Incontrast, low-aluminum welds did notshow any columnar ferrite network(Fig. 7B) and exhibited grain boundaryferrite and bainitic microstructurethat forms directly from the 100%austenite phase field. Physics-Based Modeling of In-clusion Formation: First, the abilityto predict inclusion formations wasanalyzed with stability maps that de-pict the equilibrium between AlN,Al2O3 or Ti(CN), and liquid steel at1527°C — Fig. 8A, B. The calculationswere performed with ThermoCalc®

WELDING RESEARCH


Fig. 9 — Calculated Fe-Al quasi-binary diagram shows a largechange in the phase evolution as a function of temperaturewith small changes in dissolved aluminum concentrations.

Fig. 10 — Schematic of the camera setup developed by Halland Robino for the in-situ measurement of liquid-solid in-terface stability during welding (reproduced with permis-sion from authors (Ref. 102)).


software that relies on foundationalEquations 1 to 3. For the high-Alwelds, the concentration of dissolvedAl and N lies on the right side of theAlN stability line, which conforms tothe tendency for the formation of AlNinclusions. In contrast, the concentra-tions of Al and O lie outside the stabil-ity loop for Al2O3. Similarly, the con-centration of Ti and N fall outside thestability loop for the formation ofTi(CN) inclusions. The stability dia-gram also confirmed that with the re-duction of aluminum concentration,the formation of Al2O3 and Ti(CN) isfavored, while the formation of AlN isstifled in low-Al welds. It is indeed in-triguing to notice that even with thepresence of high aluminum concentra-tions, the Al2O3 formation is not possi-ble. Careful analyses of thermodynam-ic description show that this effect isdue to the 2nd order interactions (Ref.100) between dissolved Al and O in

the liquid steel, which stabilizes theliquid phase with reference to Al2O3. Physics Based Modeling of Mi-crostructural Evolution: Applicationof Fe-C phase diagram to high-Al weldswould not have supported the retentionof -ferrite at room temperature. How-ever, the Fe-Al quasi-binary diagram(Fig. 9) calculated for the compositionsshown in Table 1 shows interesting fea-tures. For high-Al welds, on coolingfrom high temperature, the liquid steelwill transform first to -ferrite. Withsubsequent cooling to low temperature,the austenite will form at the liquid -ferrite dendrite boundaries. Subsequentcooling to low-temperature high-Alweld never enters the 100% austenitephase field. As a result, there is a highprobability for the retention of -ferriteon cooling further. In contrast, the low-Al welds do enter into 100% austenitephase field and thereby promotes theformation of grain boundary and

bainitic microstructure seen in otherlow-alloy steel welds. The above ther-modynamic tendencies were also con-firmed by calculating the growth ofaustenite into -ferrite as a function ofweld cooling rate through kinetic mod-els capable of describing the diffusion ofC, Si, Mn, and Al between these twophases (Ref. 99). Relevance to Process-BasedQuality: The above examples show theability of thermodynamic models topredict inclusion formation and mi-crostructural evolution. At the sametime, the results stress the need for agood thermodynamic description of liq-uid steel. For example, the thermody-namic data relevant to liquid steel, in-cluding the 2nd order interaction, wasmeasured in the 1960s to 1970s by ladlemetallurgy researchers (Ref. 101). With-out this data, the models would havenot been able to predict the absence ofAl2O3 formation in high-Al welds. The

WELDING RESEARCH


Fig. 11 — Overview of the liquid-solid interface instabilities during the pulsed GTAW process measured using in-situ optical method:A, B — Two successive frames showing the motion of large inclusions (orange color) and small inclusions (green color) within theliquid steel (blue regions). In addition, the same images show the motion of the band of the liquid-solid interface. C — Calculatedtemporal variation of the liquid-solid interface from the images show large scatter; D — the data shown in C is converted into fre-quency domain using DFFT analyses.

BA

DC


above example demonstrates the needfor continued experimental measure-ments of thermodynamic properties ofvarious phases in alloys relevant towelding and joining.

Case Study 2: Effect of Liquid-SolidInterface Transients

Problem Statement: The thermo-dynamic and kinetic models allowed usto predict the inclusion formation andmicrostructural evolution in weldsmade with normal steady-state condi-tions. However, the applicability of themodel to transient conditions was notevaluated. For example, under pulsedwelding conditions, one may experiencelocal changes in thermal gradients andliquid-solid interface velocity. This casestudy focuses on the measurement ofthe transients in weld solidification con-ditions in an Fe-C-Al-Mn alloy system,specifically within high-Al welds. Approach: In this research, autoge-nous gas tungsten arc (GTA) weldswere made with the following condi-tions: 18.5 V, pulse peak current 130A, pulse background current 90 A,pulse frequency of 300 Hz, weldingspeed 0.6 mm/s, and helium shielding.A high-speed in-situ monitoring sys-tem (Fig. 10) developed by Hall andRobino (Ref. 102) was used to trackthe instabilities of the liquid-solid in-terface. In this setup, the camera wasmounted on a fixture attached to the

weld arc and recorded at a time inter-val of 0.005 s. Therefore, the relativevelocity between the camera and theweld pool was set at zero. Results: The in-situ images wereevaluated through standard imageanalysis tools (Ref. 103). Typical imagesfrom two sequential time frames areshown in Fig. 11A, B. Original black andwhite images were converted to color toillustrate various observed phenomena.First, the images show the motion ofcoarse and fine inclusions in the liquidregion. Second, the snapshots also showalternating motion of the liquid-solidinterface with reference to the cameralocation. The locations of these inter-

face positions were extracted from a se-ries of image frames. Since the relativevelocity between camera and arc is zero,the motions of the interface are inter-preted as growth and dissolution (Fig.11C) reaching the rates of ±0.015 m/s.The above data were further analyzedusing discrete fast Fourier transforma-tion (FFT) algorithm. The analyses (Fig.11D) show the pulsed interface motionoccurs at different frequencies, i.e., 25,75, 100, and 125 Hz. Currently, the rea-sons for such specific frequency evolu-tions are not understood. The effect of such l/s oscillations onmicrostructural evolution was evaluatedby optical microscopy — Fig. 12. The

WELDING RESEARCH


Fig. 12 — A — Cross section (Y-Z) of the pulsed GTA welds relevant tothe data shown in Fig. 11. The images in B–D show the presence of analternating band of d-ferrite network and bainitic microstructure thatforms from 100% austenite.

Fig. 14 — A — Fe-Al quasi-binary diagram shows the concentration of aluminum inthe welds used for the in-situ TRXRD experiments; B — illustration of different spotwelding experiments performed in conjunction with TRXRD experiments.

Fig. 13 — Schematic of the time-resolved X-ray dif-fraction experiments performed within a synchro-tron beamline. The image was reproduced withpermission from the authors of Ref. 97.

B

B

C D

E

F

A

A


micrographs showed interesting phe-nomena of alternating arcs of columnard-ferrite and bainitic microstructure.This microstructure was interpreted aspossible changes in the phase selectiondue to large changes in liquid-solid in-terface velocity (Ref. 104). At slow ve-locities, the equilibrium d-ferrite is ex-pected and with each and every acceler-ation of the interface nonequilibrium gaustenite might have manifested itself,similar to the classic work of Fukumotoand Kurz (Ref. 105). Relevance to Process-BasedQuality: The above data shows the so-

lidification mi-crostructural evo-lution is sensitiveto the local ther-mal transients. In-ability to describethe above tran-sients in modelsmay not allow usto predict the scat-ter in propertiesassociated withthe banded mi-crostructure

shown in Fig. 12. Therefore, it is impor-tant to track these changes using in-situsensors and correlate to the scatter inmicrostructure and properties.

Case Study 3: In-Situ Analyses ofLiquid-Solid Interface Instabilities

Problem Statement: Although wepostulated that the changes in mi-crostructure shown in Figs. 11 and 12are due to local changes in liquid-solidinterface velocity, there was no directproof. To validate our hypothesis, in-

situ time-resolved X-ray diffraction(TRXRD) experiments (Fig. 13) wereperformed using the setup developed byElmer (Ref. 97). Initial experimentswith high-Al welds confirmed that alarge change in liquid-solid interface ve-locity will indeed induce transition fromequilibrium d-ferrite to nonequilibriumaustenite (Ref. 106). At this juncture,the generality of the above phenomenafor a wide range of alloy compositionswas not understood and is the focus ofthe current case study. Approach: To validate the abovephenomena, a special Fe-C-Al-Mn alloyweld with higher aluminum concentra-tion was selected for the TRXRD experi-ments. The new alloy composition is Fe-0.28C-0.45Mn-3.7Al (wt-%). The quasi-binary Fe-Al diagram (Fig. 14A) showsthat in this alloy, the stability of d-fer-rite is greater than g-austenite. There-fore, we expect that for a wide range ofliquid-solid interface velocities, weshould only see d-ferrite solidificationand phase selection to nonequilibriumg-austenite should be highly improba-ble. The phase evolutions at the HAZ

WELDING RESEARCH


Fig. 15 — A — Image representation of the in-situ TRXRD datashows the presence of BCC peaks before and after the arcstrike; B — optical micrograph confirms the presence of epi-taxial columnar d ferrite.

Fig. 16 — A — Image representation of the in-situ TRXRD datashows the lack of any diffraction peak before 20 s, confirmingthe presence of liquid before the arc-extinguishing event.After 20 s, when the arc extinguishes, a small speck of dif-fraction peak corresponding to d-ferrite appears and disap-pears, while the austenite peak forms immediately andappears to be stable during cooling until 45 s. The coolingprocess is confirmed by the shift in the peak position. Atabout 45 s, the austenite diffraction is replaced by broad bccpeaks corresponding to martensite. B–D — Optical micro-graphs confirm the transition from columnar d-ferrite to planaraustenite, which transforms to martensite.

B

B C D

AA

Fig. 17 — Comparison of calculated dendrite-tip and planar in-terface temperature for both d-ferrite and g-austenite usinginterface response function theories (Equations 4–9) and cali-brated parameters given in reference shows scenarios thatmay lead to phase selection phenomenon in 3.7 wt-% Al self-shielded flux cored arc welds. The higher the interface or den-drite tip temperature, the higher the probability for formationof that phase with a given morphology.


and WM were tracked for various spotwelding experiments (Fig. 14B). The ex-periments considered 35 transient con-ditions relevant to spot welding includ-ing arc strike, arc ramp up, and arc rampdown (Ref. 106). In this paper, resultsfrom two conditions are discussed —Figs. 15, 16. Results from Arc-Strike Experi-ment: In-situ diffraction data from arc-strike experiments are shown in Fig.15A, B. Under this condition, the arcwas switched on for 1 s and then extin-guished. The diffraction data show thepresence of BCC phase at room temper-ature. As soon as the arc strike was in-duced, the diffraction signals from theprobed region were lost, indicating theformation of liquid. As soon as the arc

wasswitchedoff, the dif-fractionpeaks corre-sponding to(110) planereemerged.The resultsconfirmedthat evenunder thenonequilib-rium condi-tions thatmay inducelarge liquid-solid inter-face veloci-ties, only d-ferrite will

form. Optical microscopy of the samespot weld shows epitaxial growth of fer-rite from the HAZ — Fig. 15B. Howev-er, a quick analysis of the optical micro-graph indicated that due to the smallsize of the weld metal region, the maxi-mum velocity we could achieve would bein the order of only 3 × 10–4 m/s. As aresult, one can postulate that higher liq-uid-solid interface velocities might nothave been reached to induce nonequilib-rium austenite phase selection. Results from Current Ramp-Down Experiment: Previous research(Ref. 106) has shown that by forminga large weld pool and then using cur-rent ramp down, rapid liquid-solid in-terface velocity can be accessed. Theseconditions were imposed for new alloys.

The diffraction data from the ramp-down conditions are shown in Fig. 16A.At the start of the ramp down of cur-rent, the data show no evidence for anycrystal structure. However, at around20 s, the data showed a brief formationof ferrite. At this time, the arc was ex-tinguished, which leads to accelerationof the liquid-solid interface. Interesting-ly, the data show formation of nonequi-librium g-austenite and transformationof the austenite to martensite on reach-ing low temperature at about 45 s. Simi-lar experiments were performed on an-other build and the microstructural het-erogeneity was evaluated — Fig. 16B–D. The micrographs show the forma-tion of equilibrium d-ferrite close to theweld interface — Fig. 16B, C. However,with an increase in the liquid-solid in-terface velocity, instability in d-ferritedendrites was observed — Fig. 16D. Fi-nally, most of the weld shows the for-mation of martensite laths that arelarger than 100 μm, which could haveonly formed from large austenitegrains. The above results suggest thatwe have accessed nonequilibrium g-austenite with planar solidification. Theabove phase selection phenomena wereanalyzed using the interface responsefunction theories outlined in the previ-ous section (Ref. 104). The dendrite tiptemperature for d-ferrite and g-austen-ite was calculated using calibrated pa-rameters relevant to Equations 4–9.The results are shown in Fig. 17. Thecalculations show that for a given ther-mal gradient, it is possible to access theplanar g-austenite solidification, before

WELDING RESEARCH


Fig. 18 — Calculated (A) G and (B) R for Cube 1, which uses a spot meltpattern leading to (C), the columnar grain region along the build direc-tion. In contrast, (C) and (D) show the calculated G and R for Cube 2,which uses a spot melt pattern with shallow G and R, thereby inducing(E) an equiaxed microstructure (images were used with permission).

Fig. 19 — Overlay of calculated G and R for Cubes 1 and2 on the columnar to equiaxed (CET) maps show thateven with the scatter in the values, Cube 1 will predom-inantly lead to columnar microstructure and Cube 2 toequiaxed microstructure.

DA

EB

FC


giving away to planar d-ferrite. Relevance to Process-BasedQuality: Although the models haveshown the feasibility of predicting thesetransient behaviors, the parametersused in Equations 4–9 have been cali-brated. Therefore, extension of thesemodels to other alloy systems must beaccompanied by simple calibration ex-periments and heat transfer simulationwith ramp-down and arc-strike experi-ments. Nevertheless, results from theabove experiments showed it is indeedpossible to control microstructure byspatially and temporally controllingthermal gradients and liquid-solid inter-face velocities.

Case Study 4: Extension to PowderBed Additive Manufacturing

Problem Statement: It has been adream of metallurgists to arrive at site-specific control of microstructures innear net shape components. The discus-sions in the previous section showed itis indeed possible to arrive at process-based quality by controlling microstruc-tural heterogeneity. This microstructur-al control is possible in complex compo-nents if one can modify the spatial andtemporal thermal signatures includingthermal gradients and liquid-solid inter-face velocity. The emergence of powderbed additive manufacturing provides anopportunity to test out this hypothesis(Ref. 107). Approach: Electron beam powderbed fusion (E-PBF) of Alloy 718 wasselected for this case study. To inducedifferent melt pool shapes, thermalgradients (G), and liquid-solid inter-face velocities (R), the beam spot-ontime and location of the beam shouldbe varied. A heat transfer model (Ref.108) was used to design these spotmelting strategies, a priori before theexperiment. The predicted G and R areshown in Fig. 18. The first cube sam-ple was designed (Fig. 18A, B) to in-duce columnar microstructure and thesecond cube sample was designed (Fig. 18D, E) to induce the equiaxedmicrostructure. Results: These cube samples wereproduced by E-PBF melting of eachlayer with the spot melt patterns de-signed by models. The cube sampleswere then sectioned and analyzed us-ing electron backscattered diffractionimaging. As expected, the first cube(Fig. 18C) showed a columnar mi-

crostructure and the second cube (Fig.18F) showed an equiaxed microstruc-ture. The above experiment was fur-ther verified by calculating the solidifi-cation maps using interface responsefunction theories (Equations 4–9) andoverlaying the calculated G and R val-ues for Cubes 1 and 2 — Fig. 19. Evenwith a large amount of scatter, most ofthe G and R data from Cube 1 layabove the CET line confirming the ten-dency for the formation of columnargrains. Similarly, most of the datapoints from Cube 2 lay below the CETline cinforming the tendency for theformation of equiaxed grains. Theabove transitions have also been veri-fied with in-situ infrared thermogra-phy and in complex geometries (Ref.109). Relevance to Process-BasedQuality: The above example demon-strates that process-based quality isindeed achievable even under tran-sient conditions, provided that we canreproduce these transients repeatedlyby careful control of energy depositionand heat transfer. Recently, the aboveprocess-based quality framework, inconjunction with in-situ near-infraredimaging, was extended to qualificationof topology optimized Ti6Al4V partsmade by the E-PBF process (Ref. 110).

Future Directions

This review, results, and discus-sions demonstrate it is indeed possibleto arrive at process-based quality ofcomponents much earlier than in 2040as proposed in the AWS roadmap. Fur-thermore, the process-based qualityframework must be extended to fusionwelding, solid-state joining, brazing,soldering, and additive manufacturingby coupling integrated process models,in-situ process sensors, and the collec-tion of all parameters relevant toprocess and other boundary condi-tions. The above data can be analyzed,tracked, and archived using emerginghigh-performance computing, cloudcomputing (Ref. 76), and data analyt-ics (Ref. 111). Once we have developedand deployed proof-of-principle indus-trial standards based on process-basedquality, the above approach can be ex-tended to hybrid materials that joinmetals, polymers, ceramics, glass, andelastomers in different forms andfunctions (Ref. 112).

Summary andConclusions

In this paper, the feasibility ofprocess-based qualification of weldedcomponents was explored. A process-based qualification framework will in-volve concurrent activities rangingfrom modeling, making, and measur-ing with state-of-the-art computation-al and sensing tools. To succeed in thisalternative approach to qualification,it is important to define all the bound-ary conditions including geometry, re-straints, and the processing environ-ment. Then, for given process condi-tions, an integrated modeling andsensing tool should be capable of pre-dicting or describing the spatial andtemporal variation of heat and masstransfer, solidification, solid-statetransformation, and plastic deforma-tion. With the above data, it is possi-ble to predict the performance ofwelded components with minimal tri-al-and-error experimentation. Few ex-amples were presented to bolster theargument for process-based quality ofwelded components. Two major scien-tific challenges were identified, i.e.,variability in describing the alloying el-ement concentration in weld metaland transients in thermal signatures,thermal gradient, and liquid-solid in-terface velocity. Approaches to addressthe above challenges are illustratedwith four case studies. In the first case study, the need forreliable multi-component thermody-namic data of liquid and solid phasesare emphasized to describe inclusionformation and microstructural evolu-tion in Fe-C-Al-Mn self-shielded fluxcored arc welds. In the second casestudy, the role of in-situ monitoring byhigh-speed optical imaging to track theliquid-solid interface velocities and as-sociated microstructural heterogeneitywas presented. In the third case study,the role of in-situ time-resolved X-raydiffraction to probe the phase selectionduring transient welding was demon-strated. The data and analyses con-firmed the importance ofspatial andtemporal variations of the thermal gra-dient and liquid-solid interface velocityin forcing the equilibrium d-ferrite ornonequilibrium -austenite solidifica-tion. Finally, with the above fundamen-tal knowledge of weld solidification,site-specific control of microstructurewas achieved in Alloy 718 builds madeby electron beam powder bed fusion

WELDING RESEARCH



technique. The results presented in thispaper support the notion that process-based qualification of componentsmade by welding and additive manufac-turing is indeed achievable in the nearfuture.

Research was sponsored by the U.S.Department of Energy, Office of EnergyEfficiency and Renewable Energy, Ad-vanced Manufacturing Office, undercontract DE-AC05-00OR22725 withUT-Battelle LLC. This manuscript hasbeen authored by UT-Battelle LLC un-der Contract No. DE-AC05-00OR22725with the U.S. Department of Energy.The United States Government retainsand the publisher, by accepting the arti-cle for publication, acknowledges thatthe United States Government retains anonexclusive, paid-up, irrevocable,world-wide license to publish or repro-duce the published form of this manu-script, or allow others to do so, for Unit-ed States Government purposes. TheDepartment of Energy will provide pub-lic access to these results of federallysponsored research in accordance withthe DOE Public Access Plan (http://ener-gy.gov/downloads/doe-public-access-plan).Author also thanks Dr. A. C. Hall (San-dia National Laboratories, Albuquerque,N.Mex.) and Dr. John Elmer (LawrenceLivermore National Laboratory, Liver-more, Calif.) to use the data during thecollaborative research. Author acknowl-edges the support of UT/ORNL Gover-nor’s Chair program of University ofTennessee, Knoxville.

1. Welding Technology Roadmap, FinalDraft, September 2000, prepared by Ener-getics, The American Welding Society andEdison Welding Institute. The article canbe accessed from the following internet lo-cation: https://app.aws.org/researchroadmap.pdf 2. Babu, S. S. 2010. Introduction to in-tegrated weld modeling. ASM Handbook,Vol. 22B, Metals Process Simulation, Eds.D. U. Furrer and S. L. Semiatin. ASM International. 3. Davis, J., Eagar, T., Porter, J.,Bernaden, J., and Sarli, M. 2012. Smartmanufacturing, manufacturing intelligenceand demand-dynamic performance. Com-puters and Chemical Engineering 47:145–156.

4. Integrated Computational MaterialsEngineering: A Transformational Disciplinefor Improved Competitiveness and NationalSecurity. 2008. The National AcademiesPress: Washington, DC. 5. Kalpakjian, S., and Schmid, S. 2013.Manufacturing Engineering and Technology,7th Edition, Pergamon. 6. Saaksvuori, A., and Immonen, A.2002. Product Lifecycle Management. Berlin,Germany: Springer-Verlag. 7. Welding Handbook, Vol. 1, WeldingTechnology. 1987. Miami, Fla.: AmericanWelding Society. 8. Metals Handbook, Desk Edition, 2nd

Edition. 1998. Materials Park, Ohio: ASMInternational. 9. AWS D1.1/D1.1M: 2015, StructuralWelding Code — Steel. Miami, Fla.: Ameri-can Welding Society. 10. AWS B2.1-1/8-010: 2015, Standardfor Welding Procedure Specification (SWPS).Miami, Fla.: American Welding Society. 11. Zhan, X., Ou, W., Wei, Y., and Jiang,J. 2016. The feasibility of intelligent weld-ing procedure qualification system forQ345R SMAW. Int. J. Adv. Manuf. Technol.83: 765–777. 12. AWS A5.5/A5.5M: 2014, Specifica-tion for Low-Alloy Steel Electrodes for Shield-ed Metal Arc Welding. Miami, Fla.: Ameri-can Welding Society. 13. AWS C7.3:2016, AWS A5.5/A5.5M:2014, Process Specification for Electron BeamWelding. Miami, Fla.: American WeldingSociety. 14. API Standard 1104, Welding ofpipelines and related facilities, 21st edition,2013. American Petroleum Institute 15. Weld Quality: The Role of Computers.1988. Proceedings of an international con-ference on improvement of weldment con-trol with reference to computer technology.Vienna, Austria: Pergamon Press. 16. Goldak, J. A., and Akhlaghi, M.2005. Computational Welding Mechanics.Springer, USA. 17. Radaj, D. 1992. Heat Effects of welding— Temperature Field, Residual Stress and Dis-tortion. Berlin, Germany: Springer-Verlag. 18. Wickle III, H. C., Zee, R. H., and Chin,B. A. 1999. A sensing system for weldprocess control. J. Materials Processing Tech-nology 89-90: 254–259. 19. Chen, S. J., Huang, N., Liu, Y. K., andZhang, Y. M. 2017. Machine assisted manualtorch operation: System design, responsemodeling and speed control. J. Intell. Manuf.28: 1249–1258. 20. Nilsiam, Y., Haselhuhn, A., Wijnen,B., Sanders, P., and Pearce, J. M. 2015. Inte-grated voltage-current monitoring and con-trol of gas metal arc weld magnetic ball-jointed open source 3D printer. Machines 3:339–351. 21. Goldak, J. A., Chakravarti, A., andBibby, M. 1984. A new finite element model

for welding heat sources. Metall. Trans. B 15:299–305. 22. Kou, S., and Wang, Y. H., 1986. Com-puter simulation of convention in movingarc weld pools. Metall. Trans. A 17:2271–2277. 23. DebRoy, T., and David, S. A., 1989.Physical processing in fusion welding. Rev.Mod. Phys. 67: 85–112. 24. David, S. A., and Vitek, J. M., 1989.Correlation between solidification parame-ters and weld microstructures. Int. Mater.Reviews 34: 213–245. 25. Ashby, M. F., and Easterling, K. E.1982. A first report on diagrams for graingrowth in welds. Acta Metall. 30:1969–1978. 26. Bhadeshia, H. K. D. H., Svensson, L.–E., and Gretoft, B. 1985. A model for the de-velopment of microstructures in low-alloysteel (Fe-Mn-Si-C) weld deposits. Acta Met-all. 33: 1271–1283. 27. Yang, Y., and Babu, S. S. 2010. An in-tegrated model to simulate laser claddingmanufacturing process for engine repair ap-plications. Welding in the World 54:R298–R307. 28. Babu, S. S., Livingston, J., and Lip-pold, J. C. 2013. Physical simulation and mi-crostructure evolution during friction stirprocessing of Ti6Al4V alloy. Metallurgicaland Materials Transactions A, 44: 3577–3591. 29. Yang, Y., Babu, S. S., Vaze, S., Kikel, J.and Dewees, D. 2008. Cracking mitigationduring buttering and cladding of a low alloysteel pipe. Proceedings of Trends in WeldingResearch Conf., Pine Mountain, Ga. ASM In-ternational. 30. Withers, P. J., and Bhadeshia, H. K. D.H. 2001, Overview — Residual stress: Part 2— Nature and origins. Materials Science andTechnology 17: 366–375. 31. Pahkamma, A., Warmefjord, K.,Karlsson, L., Soderberg, R., and Goldak, J.2012. Combining variation simulation withwelding simulation for prediction of defor-mation and variation of a final assembly. J.Computing and Information Science and Engi-neering 12; #021002. 32. Babu, S. S., Rajasekhar, M. V., andAchar, D. R. G. 1988. Studies on improvingaluminum zinc magnesium alloy weld metalproperties, presented at the internationalconference on Welding technology in Devel-oping countries — Present status and futureneeds. Organized by University of Roorkee,India. 33. Rao, K. P. 1990. Effect of weld cool-ing rate on delta-ferrite content ofaustenitic weld metals. Journal of MaterialsScience Letters 90: 675–677. 34. Quintana, M. A., Babu, S. S., Major,J., and Dallam, C. 2010. Weld metal tough-ness — Sources of variation. Proceedings ofthe 8th International Pipeline Conference, IPC2010, September 27–October 1, Calgary,Alberta, Canada.

WELDING RESEARCH


Acknowledgments

References


35. Hanhold, B., Babu, S. S., and Cola,G. 2013. Investigation of heat affectedzone softening in armor steels — Part 1:Phase transformation kinetics. Science andTechnology of Welding and Joining 18:247–252. 36. Hanhold, B., Babu, S. S., and Cola,G. 2013. Investigation of heat affectedzone softening in armor steels — Part 2:Mechanical and microstructural hetero-geneity. Science and Technology of Weldingand Joining 18: 253–260. 37. Chen, J., Yu, X., Miller, R. G., andFeng, Z. 2015. In situ strain and tempera-ture measurement and modeling duringarc welding. Sci. Technol. Weld. Join. 20:181–188. 38. Joseph A., Palanichamy, P., andJayakumar, T. 2015. Evaluation of residualstresses in carbon steel weld joints by ul-trasonic Lcr wave technique. J. Nondestruc-tive Evaluation 34: #266. 39. Rosenthal, D. 1941. Mathematicaltheory of heat distribution during weldingand cutting. Welding Journal 20(5): 220-sto 234-s. 40. Rosenthal, D. 1946. The theory ofmoving sources of heat and its applicationto metal treatments. Trans. ASME 68:849–866. 41. Grong, O. 1997. Metallurgical Model-ing of Welding, 2nd Edition. London, UK: In-stitute of Materials. 42. Tsai, M. C., and Kou, S. 1990. Elec-tromagnetic-force-induced convection inweld pools with a free surface. WeldingJournal 69(6): 241-s to 246-s. 43. Zacharia, T., David, S. A., Vitek, J.M., and DebRoy, T. 1989. Weld pool devel-opment during GTA and laser beam weld-ing of Type 304 stainless steel, Part I —Theoretical analysis. Welding Journal68(12): 499-s to 509-s. 44. Mundra, K., DebRoy, T., Babu, S. S.,and David, S. A. 1997. Weld metal mi-crostructure calculations from fundamen-tals of transport phenomena in the arcwelding of low alloy steels. Welding Journal76(4): 163-s to 171-s. 45. Kurz, W., and Fisher, D. J. 1998.Fundamentals of Solidification, 4th Edition,CRC Press. 46. Cahn, J. W. 1956. The kinetics ofgrain boundary nucleated reactions. ActaMetallurgica 4: 449–459. 47. Laughlin, D. E., and Hono, K. 2014.Physical Metallurgy, 5th Edition. Elsevier. 48. Saunders, N., and Miodownik, A. P.1998. Calculation of Phase Diagrams: A Com-prehensive Guide. Pergamon. 49. Hillert, M. 1998. Phase Equilibria,Phase Diagrams and Phase Transformations—Their Thermodynamic Basis. CambridgeUniversity Press. 50. Villafuerte, J. C., Kerr, H. W., andDavid, S. A. 1995. Mechanisms of equiaxedgrain formation in ferritic stainless steel

gas tungsten arc welds. Materials Scienceand Engineering A 194: 187–191. 51. Trivedi, R., and Kurz, W. 1994. Den-drite growth. International Materials Re-views 39: 49–74. 52. Dantzig, J., and Rappaz, M. 2017.Solidification, 2nd Ed., EPFL Press. 53. Babu, S. S., Elmer, J. W., David, S.A., and Vitek, J. M. 2002. Time-resolved X-ray diffraction investigation of primaryweld solidification in Fe-C-Al-Mn steelwelds. Acta Materialia 50: 4763–4781. 54. Fukumoto, S., and Kurz, W. 1999.Solidification phase and microstructure se-lection maps for Fe-Cr-Ni alloys. ISIJ Inter-national 39: 1270-1279. 55. Babu, S. S. 2009. Thermodynamicand kinetics models for describing mi-crostructure evolution during joining ofadvanced materials. International MaterialsReviews 54: 333–367. 56. Hultgren, R. 1947. Isothermaltransformation of austenite. Trans. AIME39: 915–1005. 57. Coates, D. E. 1972. Diffusion-con-trolled precipitate growth in Ternary-sys-tems — 1. Metallurgical Transactions 3:1203–1212. 58. Coates, D. E. 1972. Diffusion-con-trolled precipitate growth in Ternary-sys-tems — 2. Metallurgical Transactions 4:1077–1086. 59. Babu, S. S., Hono, K., and Sakurai,T. 1994. APFIM study of the partitioningof substitutional elements during temper-ing of a low alloy steel martensite. Metall.Trans. A. 25A: 499–508. 60. Aaronson, H. I., Domain, H. A., andPound, G. M. 1966. Thermodynamics ofaustenite — Proeutectoid ferrite transfor-mation — 2: Fe-C-X alloys. Trans. of theMet. Soc. of AIME, Vol. 236, p. 768. 61. Ghosh, G., and Olson, G. B. 2001.Computational thermodynamics and ki-netics of martensitic transformations.Journal of Phase Equilibria 22: 199–207. 62. Bhadeshia, H. K. D. H., Svensson, L.E., and Gretoft, B. 1985. The influence ofalloying elements on the formation of al-lotriomorphic ferrite in low-alloy steelweld deposits. J. Materials Science Letters 4:305–308. 63. Ion, J. C., Easterling, K. E., and Ash-by, M. F. 1984. A 2nd report on diagrams ofmicrostructure and hardness for heat-af-fected-zone in welds. Acta Metallurgica 32:1949–1962. 64. Loberg, B., Nordgren, A., Strid, J.,and Easterling, K. E. 1984, The role of al-loy composition on the stability of nitridesin Ti-microalloyed steels during weld ther-mal cycles. Metall. Trans. A. 15A: 33–41. 65. Ohno, M., Yamaguchi, T., Sato, D.,and Matsuura, K. 2013. Existence ornonexistence of thermal pinning effect ingrain growth under temperature gradient.Computational Materials Science 69: 7–13.

66. Feng, Z., Wang, X. L., David, S. A.,and Sklad, P. S. 2007. Modeling of residualstresses and property distributions in fric-tion stir welds of aluminum alloy 6061-T6.Sci. Technol. Weld. Joining 12: 348–356. 67. Feng, Z., David, S. A., Zacharia, T.,and Tsai, C. L. 1997. Quantification ofthermomechanical conditions for weld so-lidification cracking. Sci. Technol. Weld.Joining 2: 11–19. 68. Heinze, C., Pittner, A., Rethmeier,M., and Babu, S. S. 2013. Dependency ofmartensite start temperature on prioraustenite grain size and its influence onwelding-induced residual stresses. Compu-tational Materials Science 69: 251–260. 69. Schmidt, H., and Hattel, J. 2005. Alocal model for the thermomechanical con-ditions in friction stir welding. ModellingSimul. Mater. Sci. Eng. 13: 77–93. 70. Johnson, G. R., and Cook, W. H.1985. Fracture characteristics of threemetals subjected to various strains, strainrates, temperatures and pressures. Engi-neering Fracture Mechanics 21: 31–48. 71. Semiatin, S. L., and Bieler, T. R.2001. The effect of alpha platelet thicknesson plastic flow during hot working of Ti-6Al-4V with a transformed microstructure.Acta Mater. 49: 3565–3573. 72. Roters, F., Eisenlohr, P., Kords, C.,Tjahjanto, D. D., Diehl, M., and Raabe, D.2012. DAMASK: the Dusseldorf advancedMaterial Simulation Kit for studying crys-tal plasticity using an FE baser or a spectralnumerical solver. Procedia IUTAM 3: 3–10. 73. Kirkaldy, J. S. 1991. Diffusion-con-trolled phase transformations in steels —Theory and applications. Scand. J. Metall.20: 50–61. 74. Yang, Y. P., Babu, S. S., Orth, F., andPeterson, W. 2008. An integrated computa-tional model to predict the mechanical be-havior of spot welds. Science and Technologyof Welding and Joining 13: 232–239. 75. Yang, Y., Zhang, W., Gan, W., Khu-rana, S. P., Xu, J. and Babu, S. S. 2008. On-line software tool for predicting weld resid-ual stress and distortion. ASME 2008 Pres-sure Vessels and piping Conference, Vol. 6,Materials and Fabrication, Parts A and B,PVP2008-61123, pp. 279–288. 76. Babu, S. S., Conrardy, C., Khurana, S.P., Zhang, W., and Gan, W. 2012. Remotehigh performance computing materials join-ing and material forming modeling systemand method. U.S. Patent, #8,301,286 B2. 77. Pittner, A., Weib, D., Schwenk, C.,and Rethmeier, M. 2008. Methodology toimprove applicability of welding simula-tion. Science and Technology of Welding andJoining 13: 496–508. 78. AWS A9.5:2013, Guide for Verificationand Validation in Computation Weld Mechan-ics. Miami, Fla.: American Welding Society. 79. Fredette, L. F., Broussard, J. E.,Kerr, M., and Rathbun, H. J. 2011.

WELDING RESEARCH



NRC/EPRI Welding Residual Stress Valida-tion Program — Phase III details and find-ings. Proceedings of PVP2011, 2011 ASMEPressure Vessels and Piping Division Confer-ence, July 17–21, Baltimore, Md. 80. Rathbun, H. J., Fredette, L. F., andRudland, D. L. 2011. NRC welding residualstress validation program internationalround robin program and findings. Pro-ceedings of PVP2011, 2011 ASME PressureVessels and piping Division Conference, July17-21, Baltimore, Md. 81. Caron, J., Heinz, C., Schwenk, C.,Rethmeier, M., Babu, S. S., and J. C. Lippold.2010. Effect of continuous cooling transfor-mation variations on numerical calculationof welding induced residual stresses. WeldingJournal 89(7): 151-s to 160-s. 82. Heinze, C., Pittner, A., Rethmeier,M., and Babu, S. S. 2013. Dependency ofmartensite start temperature on prioraustenite grain size and its influence onwelding-induced residual stresses. Compu-tational Materials Science 69: 251–260. 83. Babu, S. S., David, S. A., Vitek, J. M.,Mundra, K., and DebRoy, T. 1995. Devel-opment of macro- and microstructures ofC–Mn low alloy steel welds — Inclusionformation. Materials Science and Technology11: 186–199. 84. Babu, S. S., Reidenbach, F., David, S.A., Böllinghaus, Th., and Hoffmeister, H.1999. Effect of high-energy-density weld-ing processes on inclusion and microstruc-ture formation in steel welds. Science andTechnology of Welding and Joining 4: 63–73. 85. Song, H. Y., Evans, G. M., and Babu,S. S. 2014. Effect of microstructural het-erogeneities on scatter of toughness inmulti-pass weld metal of C-Mn steels. Sci-ence and Technology of Welding and Joining19: 376–384. 86. Kotecki, D. J., and Moll, R. A. 1970,A toughness study of steel weld metal fromself shielded flux-cored electrodes — Part1. Welding Journal 49(4): 157-s to 165-s. 87. Kotecki, D. J., and Moll, R. A. 1972.A toughness study of steel weld metal fromself shielded flux-cored electrodes — PartII. Welding Journal 51(3): 138-s to 155-s. 88. Babu, S. S., Kelly, S. M., Muru-gananth, M., and Martukanitz, R. P. 2006.Reactive gas shielding during laser surfacealloying for production of hard coatings.Surface Coating and Technology 200:2663–2671. 89. Brandi, S., Taniguchi, C., and Liu, S.1991. Analysis of metal transfer in shield-ed metal arc welding. Welding Journal70(10): 261-s to 270-s.

90. Palmer, T. A., and DebRoy, T. 2000.Numerical modeling of enhanced nitrogendissolution during gas tungsten arc weld-ing. Metallurgical and Materials Transac-tions B 31: 1371–1385. 91. Elmer J. W., Vaja, V., Carlton, H. D.,and Pong, R. 2015. The effect of Ar and N2

shielding gas on laser weld porosity in steelstainless steel and nickel. Welding Journal94(10): 313-s to 325-s. 92. He, X., Norris, J. T., Fuerschbach, P.W., and DebRoy, T. 2006. Liquid metal ex-pulsion during laser spot welding of 304stainless steel. J. Phys. D: Appl. Phys. 39:525–534. 93. Khan, P. A. A., and DebRoy, T. 1984.Alloying element vaporization and weldpool temperature during laser welding ofAISI 202 stainless steels. Metall. Trans. B15B: 641–644. 94. Quintana, M. A., Babu, S. S., Major,J., and Dallam, C. 2010. Weld metal tough-ness — Sources of variation. Proceedings ofthe 8th International Pipeline Conference, IPC2010, Sept. 27–Oct. 1. Calgary, Albera,Canada. 95. Hong, T., DebRoy, T., Babu, S. S.,and David, S. A. 2000. Modeling of inclu-sion growth and dissolution in the weldpool. Metallurgical Transactions B 31:161–169. 96. Smith, R. J., Li, W., Coulson, J., Clark,M., Somekh, M. G., and Sharples, S. D. 2014.Spatially resolved acoustic spectroscopy forrapid imaging of material microstructureand grain orientation. Measurement Scienceand Technology Vol. 25, #055902. 97. Elmer, J. W. 2008. A new path for-ward for understanding microstructuralevolution during welding. Welding Journal87(6): 149-s to 168-s. 98. Quintana, M. A., McLane, J., Babu,S. S., and David, S. A. 2001. Inclusion for-mation in self shielded flux-cored arcwelds. Welding Journal 80(4): 98-s to 105-s. 99. Babu, S. S., David, S. A., and Quin-tana, M. A. 2001, Modeling microstructureevolution in self-shielded flux cored arcwelds. Welding Journal 80(4): 91-s to 97-s. 100. Babu, S. S., David, S. A., Vitek, J.M., Mundra, K., and DebRoy, T. 1999. Amodel for inclusion formation in low alloysteel welds. Science and Technology of Weld-ing and Joining 4: 276-284. 101. The Making, Shaping and Treatingof Steel, 10th Edition, United States Steel.1985. Published by the Association of Ironand Steel Engineers. 102. Hall, A. C., and Robino, C. V. 2004.Association of microstructural features

and rippling phenomenon in 304 stainlesssteel gas tungsten arc welds. Science andTechnology of Welding and Joining 9:103–108. 103. Babu, S. S., Specht, E. D., David, S.A., Karapetrova, J., Zschak, P., Peet, M.,and Bhadeshia, H. K. D. H. 2005. In-situobservations of lattice parameter fluctua-tions in austenite and transformation tobainite. Metal. Mater. Trans. A 36A:3281–3289. 104. Babu, S. S., Elmer, J. W., David, S.A., and Quintana, M. 2002. In-situ obser-vations of nonequilibrium austenite for-mation during weld solidification of Fe-C-Al-Mn low alloy steel. Journal of Proceed-ings of Royal Society (Mathematical andPhysical Sciences) A 458: 811–821. 105. Fukumoto S., and Kurz, W. 1999.Solidification phase and microstructure se-lection maps for Fe-Cr-Ni alloys. ISIJ Inter-national 39: 1270–1279. 106. Babu, S. S., Elmer, J. W., Vitek, J.M., and David, S. A. 2002. Time-resolvedX-ray diffraction investigation of primaryweld solidification in Fe-C-Al-Mn steelwelds. Acta Materialia 50: 4763–4781. 107. Sames, W., List, F. A., Pannala, S.,Dehoff, R. R., and Babu, S. S. 2016. Themetallurgy and processing science of metaladditive manufacturing. International Ma-terials Reviews 61: 315–360. 108. Plotkowski, A., Kirka, M. M., andBabu, S. S. 2017. Verification and valida-tion of a rapid heat transfer calculationmethodology for transient melt pool solid-ification conditions in powder bed metaladditive manufacturing. Additive Manufac-turing 18: 256–268. 109. Raplee, J., Plotkowski, A., Kirka, M.M., Dinwiddie, R., Okello, A., Dehoff, R. R.,and Babu, S. S. 2017. Thermographic mi-crostructure monitoring in electron beamadditive manufacturing. Sci. Rep. 7: #43554. 110. Yoder, S., Morgan, S., Barnes, E.,Kinzy, C., Nandwana, P., Kirka, M. M.,Babu, S. S., Paquit, V., and Dehoff, R. R.2017. Qualification of topology optimizedTi-6Al-4V components made by electronbeam powder melting process. Submittedfor publication in Additive Manufacturing. 111. Powers, S. A., Dehoff, R. R., Paquit,V. C., Steed, C. A., and Kistler, D. E. 2016.Application of data analytics to additivemanufacturing. Conference: 11th INFORMSWorkshop on Data Mining and Decision Ana-lytics (DM-DA 2016). Nashville, Tenn. 112. Ashby, M. F. 2005. Hybrids to fillholes in material property space. Philosoph-ical Magazine 85: 3235–3257.

WELDING RESEARCH


SUDARSANAM SURESH BABU is with Department of Mechanical, Aerospace and Biomedical Engineering, Tickle College of Engineer-ing, The University of Tennessee, Knoxville, Tenn. He is also with Manufacturing Demonstration Facility, Energy and TransportationSciences Division, Oak Ridge National Laboratory, Oak Ridge, Tenn.

Paper based on 2017 Comfort A. Adams Lecture, which was delivered on November 6, 2017, at the AWS Annual Convention, duringFABTECH in Chicago, Ill.


Welding Journal : January 2018 - DOI

Documents

Transcript of Welding Journal : January 2018 - DOI