Experimental Analysis, Defect Evaluation and Computational Developments of FSW

by Pedro Vilaça*, Telmo Santos**, Luísa Quintino***

pedro.vilaca@ist.utl.ptlquintino@ist.utl.pt

*Assistant Professor, **PhD Student, ***Associate Professor at the Instituto Superior Técnico, Technical University of Lisbon; Portugal

Abstract

For more than one century welding technology plays an important role in most of the technical developments of our society. Friction Stir Welding (FSW) invented in 1991 is unanimously considered one milestone in the development of the welding technology and it is foreseen it will bring important contributions in the development of present and future design and construction of metallic structures. This fact has already being confirmed by the growing number of industrial applications in many and different sectors mainly in the manufacturing of light alloys.

The reliability of using the process in industry is nevertheless dependent on a clear understanding of the phenomena involved and the capability of controlling the end result by defining the most appropriate welding procedure for each case. To achieve this goal there is a need to perform both extensive experimental and modeling of FSW. The present paper addresses experimental results obtained in different aluminium alloys focusing metallurgical, mechanical and corrosion aspects. It also addresses two models, an analytical and a numerical witch enable a better understanding of the process features and are used to predict final properties of the joint. Keywords: FSW, Experimental Fundaments, Computational Modelling, IST.

Introduction

Friction Stir Welding (FSW) is a solid state welding process, which proceeds below the melting point of the weld material. A rotating non-consumable tool, with a somewhat complex shoulder and pin profile, is plunged into the weld joint and forced to traverse along the joint line, heating the abutting components by interfacial and internal friction, thus producing a weld joint by extruding, forging and stirring the materials from the two workpieces in the vicinity of the tool [1, 2, 3].

The principles behind the technology have been patented [1], and the development of FSW has been taken up primarily by large companies including aerospace, aircraft, naval and automotive sectors [2, 3]. Moreover, research into FSW should not only be oriented to provide solutions for major industrial applications but also to achieve enough application flexibility for its transition into small and medium enterprises [4].

FSW encompasses complex phenomena related with plastic flow deformation resulting from the stirring of the workpieces materials. The development of process models contributes towards a better understanding of joint formation and resultant mechanical properties [5].

The presentation of some of the experimental fundaments of the FSW developed at the IST start with a description of the concepts for the tools developed and follows addressing the different patterns of the plastic flow of the material in FSW. The typical hardness fields for aluminium alloys are presented prior to the mechanical and metallurgical techniques for evaluation of the FSW joints. Corrosion studies on AA5083-H111 and an assessment of the tailor-blanks construction with FSW, GMAW and GTAW are described.

Finally a Quantitave Non-Destructive Characterization (QNDC) System which integrates three Non DestructiveTecniques (NDT) is presented. This approach using the advanced algorithms of data fusion based in the Fuzzy Logic to integrate information from ultrasound, ToFD and Eddy Currents, in order to increase the capabilities of each method individually by complementar benefits.

Before the conclusions the implementation of both thermal analytical model and coupled fluid dynamics and solid mechanics numerical approaches are reviewed.

2

Experimentations on FSW

The experimental work was developed at IST-Instituto Superior Técnico, Secção de Tecnologia Mecanica using a milling machine adapted for FSW. Tests were conducted with different welding parameter to achieve hot and cold welding conditions. Several types of tools were developed, with both mono-bulk tools and modular tools as it is represented in Figure 1. The modular iSTIRtool_v1 allow the combination between different pins and shoulder geometries and the continuously variation of the length of pin coming out of shoulder. Also clamping devices were developed.

Figure 1 – Modular FSW tool: iSTIRtool_v1 [2, 6].

The mechanical and metallurgical characteristics of all the main families of wrought

aluminium alloys have been investigated from 1.0 to about 10mm in different joint geometries.

One important result from the sensitivity analysis of the influence of FSW process parameters in the properties of the joint was the establishment of the main patterns of material flow around the tool (Figure 2). In the hot conditions the visco-plastic material flow is more concentrated around the pin and the heat affected zone is wider. In opposite under cold conditions the thermo-mechanically heat affected zone is wider and the heat affected zone is smaller.

Figure 2 – Classification of the typical material flow patterns [2].

In Figure 3 typical hardness profile for the non heat treatable aluminium

AA5083-H111 is represented.

3

Figure 3 shows the behavior of the non heat treatable alloys it is possible to conclude about the increased hardness of weld bead and heat affected zones when compared with base material. Because this alloy is very sensitive to strain hardening the increase is most significant in dynamically recovered zone and thermo-mechanically heat affected zone.

Figure 3 – Typical hardness profile for the non heat treatable aluminium alloys

(AA5083-H111 ; thickness: 4mm) [6].

Metallurgical features were also investigated in detail, such as, the diffusion of the initial precipitates, and changes of grain size in the heat affected zone and thermal-mechanically affected zone, resulting from the thermal-mechanical cycle of the process [2, 7, 8]. The results allow to understand the mechanisms that determine the metallurgical changes in the weld bead. The metallographic analysis is compared with the results of the hardness tests developed in all zones of the welded joint.

Figure 4 – Metallographic characteristics of the FSW of AA5083-H111 with a

thickness of 4mm [6, 7]. Results of metallurgy analysis can be interpreted analysing the grain morphology and

precipitates density and location, e.g., in Figure 4 the material analysed was the non-heat

4

treatable aluminium alloy AA5083–H111, where is possible to emphasise the small coalescence of grain and the reduction of precipitates in the grain boundaries of the HAZ and TMAZ which it is know to play an important role in the prevention of corrosion development .Also the TMAZ/Nugget interface enhances the significant difference between the structure of initial grain and the grain resultant of the dynamical recrystallisation process of the nugget zone.

The transfer of the FSW into industrial applications, such as shipbuilding, demands for a detailed investigation about the influence of corrosion in the performance of both parent materials and welded joints. In cooperation with the Mondego Shipyard, Figueira da Foz, Portugal, IST evaluated for the AA5083-H111 the main mechanisms of corrosion, such as: intergranular corrosion and exfoliation. In Figure 5 it is possible to observe some resultant corrosion in FSW specimens of AA5083-H111 after 7 days of exposition to real conditions.

From the final results it should be emphasised that the loss mass in the base material samples is much higher than in the friction stir welded samples because intergranular corrosion mechanism is most susceptible at the intergranular precipitates and these are more abundant in base material [6, 7].

Figure 5 – Corrosion developed in base material of AA5083-H111 after 7 days Exposition in Tagus River (20g/l NaCl) [7]. a) Intergranular corrosion along the βAl-Mg precipitates;

b) Pitting formation mechanism under the ship of the weld bead. Quantitave Non-Destructive Characterization System for FSW Non Destructive Testing on FSW is an indispensable tool to evaluate the welding quality. Depending on the process parameters, several types of defects might occur, such as root defects, inclusions, porosity, lack of penetration, etc. Fig. 6 shows a transversal macrograph of a welding where some of those defects are identified.

5

Figure 6 – Typical defects in SFL [2].

The different nature and formation mechanisms of the defects in FSW makes their

detection very difficult, as there is no NDT capable of ensure a safe and full inspection. In order to respond to this problem an advanced system of non destructive

characterization is under development for FSW based on data fusion algorithms with fuzzy logic. The main goal is to obtain a synergetic system that has the ability of aggregate different NDT in order to gain from their complementary and redundancy. This method allows a quicker, safer and more economical inspection. All the stages of this method are described in Figure 7.

characteristic variables

Probes

Side 1

Side 2

Low freq.

high freq.

Commom systems

NDT selection and

development

Numerical data acquisition

Characteristic variables

Data fusion architecture

Final result

Characteristic variables

fuzzification

Figure 7– Distinct techniques and the stages of the process. The system incorporates three distinct techniques: Ultrasound, ToFD and Eddy Currents. Several probes are incorporated in a device that enables to analyse the entire length of the weld bead in one go. The data gathered from each probe is numerically treated in order to extract the characteristic variables of each one of the inspection curves, as represented in Figure 7. These variables are then fuzzyfied using a membership fuzzy function that is dinamically parameterized according to the extreme conditions (Figure 8): Base material and total defects. These two conditions are used to calibrate the system.

6

( )nuu vZZ σ−′=

( )nll vZZ σ+′=

μ

X

1

nu vZ =′( )ivσ

nl vZ =′

( )ivσ

)( i

( ) nn

in

u

l

u

l

vofdeviationdardSvaverage

viablesticCharacterivcorrectionBeforeZcorrectionBeforeZ

parameterflawTotalZparameterflawNoneZ

tan

var

−

−−′−′−−

σ

Figure 8 – Dinamically parameterized membership function.

The output is the defect index for internal and external defects in each one of the weld bead cross sections. This defect index varies in the range [0; 1] and indicates the general state of the weld quality. The system allows the automatization of the defects detection process, not needing the intervention of an expert on NDT. A graphical user interface enables a simple and intuitive access to all results produced by the system. In this way, the evolution of the defect index is graphically represented for the acquired sections (Figure 9). It is also possible, to access each one of the NDT technical curves of each section. A three-dimensional representation of NDT techniques’ results along the entire bead can be plotted (Figure 10).

Figure 9 – Evolution of the flaw index.

Base material Real inspection Total defect

iμ v

iv

7

Figure 10 – Three-dimensional representation of some NDT techniques. Computational Modeling

The development of computational models can greatly contribute to better understanding any industrial process, particularly FSW. A validated model has the potential to produce reliable information about the deformation and mixing patterns that are important when designing FSW tools and thus should be capable of producing welds free of defects and voids. Further, a model can measure process characteristics that are difficult to observe experimentally such as local strains, strains rates and stresses [9, 10]. These strain and stress fields, together with temperature histories are seen as critical in predicting microstructure evolution. A detailed understanding of micro structural evolution can guide FSW designs by further improving mechanical properties, fatigue strength and corrosion resistance.

While considerable experimental work has been done to improve the knowledge on FSW, there’s yet a lot of work needed to create a satisfying global model that can produce consistent results. The main difficulties in modelling FSW are [2, 5]: • Extensive material deformation in the region containing fully-plasticized material; • The viscous-plastic flow imposed by the tool rigid surface, into the materials constrained

by the interaction with the cold base material, with an essentially elastic behaviour, and the rigid plate (anvil) supporting the joint;

• Heat generated due to the sliding between the surface of the tool and the materials in the joint, depends on an unknown the friction coefficient;

• The correct prediction of the viscous-plastic flow imposed by the tool rigid surface into the materials being welded is also important because the viscous dissipation contributes significantly to the heat development during the performance of the weld bead;

• The materials thermo-mechanical properties vary throughout the FSW process; • The thermal flow into the tool and support plate, needs to be considered in the models; • FSW process modelling does not allow geometric simplification because it deals with a

complex 3D material flow around the pin; • The highly rotating tool pin has, typically, a complex geometric profile (e.g. threaded),

which is rather difficult to consider for most of the numerical methods available.

8

The challenge is then to create a model able to fully describe the complex FSW process as illustrated in Figure 11. Analytical Modeling

iSTIR is the acronym of a thermal analytical model for 2D and 3D cases when FSW similar and dissimilar materials. The development of the iSTIR code came about, due to the perceived need to establish relationships between FSW parameters and easy to measure geometric features and/or mechanical properties via an assessment of the Heat Input developed during the steady-state FSW regime (Figure 12).

The thermal analytical model, iSTIR, is based on evolutions from thermal flow equations, established by D. Rosenthal [11], for point heat sources with uniform velocity. This model also takes in account the specificities of the heat generated by viscous energy dissipation during plastic flow deformation and interfacial friction, e.g., the asymmetric heat generation mostly at the periphery of the shoulder of the FSW tool, as result of the composition of both linear and rotational velocities and finally the differences between hot and cold FSW conditions [2].

Heat dissipation byintern friction (viscous)

Metallurgic structurechanges (dynamic

recrystallization of theNuget)

Movement of the tool

Materials plastic flowand deformation

Superficial appearanceDefective joints

Mechanical propertiesof the different regionsInterfacial Friction

(material / tool)

Heat Generation

Figure 11 – Ccoupled mechanical/thermal/metallurgical character of FSW process [2, 9, 10].

9

Figure 12 – Basic principles of the development of iSTIR [2, 12, 13].

Due to perceived difficulties in modeling the FSW process an iterative method was

employed where the heat power source is incremented so that the final value will minimize the difference between the iSTIR thermal field results and the ones resulting from the experimental temperature measurements.

Therefore, iSTIR code is intended to be used in a “reverse engineering approach”, i.e., based on a thermal field previously measured, the results of the thermal field from iSTIR may be iteratively superimposed on results obtained from the experimental thermal measurements (under identical thermo-physical conditions) and the value of the point power source, which produces such thermal field is then determined.

Figure 13 – Concept of the “reverse engineering approach”, used in the iSTIR code [13].

The FSW efficiency coefficient, is then systematically related with the FSW parameters

and characteristics of weld macrographs made in sections transverse to the welding direction, e.g., the hardness field and the ratio between the extension of the thermo-mechanically heat affected zone and the extension of the heat affected zone.

The classification to establish hot and cold weld conditions is based on the value of the ratio between rotational and transactional, velocities of the tool relatively to the parts being welded.

10

Numerical Modelling

The numerical modelling approach of the FSW access the material flow in the vicinity of the tool and the residual stresses and deformation field resulting from the process in the real geometry of the parts being welded.

Because there is no commercial software able to perform a complete analysis of the process, the strategy adopted was to use a coupled process based on a specially developed code, Integra3D, which bridges two commercial software’s, namely a fluid dynamics (Fluent), used for the visco-plastic regime steady-state analysis of the material flow in the vicinity of the tool and structural mechanics (Abaqus), used for compute in a transient analysis the residual deformation and stress fields based on the stress and thermal history resulting from the FLUENT analysis.

This integration is represented in the Figure 16. The integration of the two approaches is made through a FORTRAN code routine named Integra3D. The concept of integration of the two approaches used in this work is based on the fact that the FSW process has two regions with different behaviours. A first zone, near the tool, where the material has viscous-plastic behaviour, and a second zone, that includes the rest of the domain, which has predominantly elastic-plastic behaviour.

Elastic - Plastic Analysis of the plates remote from tool

HAZ and Base Material

Viscous - Plastic Analysis near the tool

Nugget and TMHAZ

Same nodesat interfaces

Updated at interface :- Temperature- Pressure

AbaqusStructural Mechanics

Approach

FluentFluid Dynamics

Approach

Results:• Residual Stress Field• Residual Deformation• Thermal History of the HAZ

Results:• Material Flow for different tool geometries• Thermal History at the vicinity of the tool

Integra 3DFE Mesh

- Pressure- Temperature

Elastic - Plastic Analysis of the plates remote from tool

HAZ and Base Material

Viscous - Plastic Analysis near the tool

Nugget and TMHAZ

Same nodesat interfaces

Updated at interface :- Temperature- Pressure

AbaqusStructural Mechanics

Approach

FluentFluid Dynamics

Approach

Results:• Residual Stress Field• Residual Deformation• Thermal History of the HAZ

Results:• Material Flow for different tool geometries• Thermal History at the vicinity of the tool

Integra 3DFE Mesh

- Pressure- Temperature

Figure 14 – Integration approach scheme for the numerical model of FSW process [9, 10].

The material behaviour used by the fluid dynamics analysis follows the Zener-Hollomon [14, 15] material model which is based on a viscosity parameter, depending of the temperature and strain rate. The material behaviour used by the structural analysis is the typical elastic-plastic behaviour depending on the temperature.

11

Conclusions Friction Stir welding has generated a strong enthusiasm in the international welding community namely due to the easiness of its practical application for aluminium welding. Other light alloys and steels can be welded using this process but to master friction stir welding for any given application there is a need to study in detail the characteristics of this process. In fact the achievement of a “good weld” depends on the correct occurrence of many different phenomena that involve:

- process parameters and tool geometries - heat flow and plastic deformation - metallurgical and mechanical characteristics

The development of analytical and numerical models facilitates the definition of adequate process parameters for any given application, and in FSW these models have to take in account both the elastic-plastic and viscous-plastic behavior of the material. Another relevant aspect that needs to be addressed by the industry using FSW is the control techniques which need to have different capabilities than the ones usually used for fusion welding. The integration of data from different non-destructive techniques and its on-line analysis will represent a step ahead towards achieving quality assurance in Friction Stir Welding Acknowledgements The authors would like to acknowledge the financial support from the Fundação para Ciência e Tecnologia (FCT) via the project: POCTI/CTM/41152/01 (acronym: iSTIR).

References [1] Thomas, W. M., Nicholas, E. D., Needham, M. G., Templesmith, Dawes, C. J., Friction

stir butt welding, International patent Application PCT/GB92/02203, GB Patent Application 9125978.8, US Patent 5.460.317, 6 December 1991.

[2] Vilaça P., “Fundamentos do processo de soldadura por fricção linear, análise experimental e modelação analítica”, Ph.D. Thesis, IST-Universidade Tecnica de Lisboa, September 2003.

[3] Luísa Quintino, Pedro Vilaça, “Soldadura por Fricção Linear – O Que a Indústria Portuguesa Precisa e Precisa de Saber”. Tecnologia e Qualidade, Revista do Instituto de Soldadura e Qualidade, pp.46-57, nº 38, Abril/Junho de 2000.

[4] P. Vilaça, J. P. Santos, A. Góis, L. Quintino, “Joining Aluminium Alloys Dissimilar in Thickness by Friction Stir Welding and Fusion Processes”, Welding in the World, Journal of the International Institute of Welding (IIW), pp.56-62, March/April 2005, Volume 49, Nº3/4-2005.

12

[5] Vilaça P., Fraga E., Sá J., Quintino L., “State of the Art in the Modelling of Friction Stir Welding”, FSW Flow Visualisation and Modelling Seminar at GKSS-Forschungszentrum, Germany, February 24-25, 2003.

[6] Pepe N., Vilaça P., Quintino L., “Metallurgical and Corrosion Features of Friction Stir Welding of AA5083-H111”, IIW Doc. IX-2148-05, Annual Assembly of the International Institute of Welding (IIW), Commission IX, Prague, July 10-15, 2005.

[7] Pepe, N. “Características Metalúrgicas, Comportamento à Fadiga e Resistência à Corrosão de Juntas Soldadas por Fricção Linear”, MSc Thesis, IST-Universidade Técnica de Lisboa, September 2005.

[8] David T., Vilaça P., Quintino L., “Caracterização Metalúrgica do Processo de Soldadura por Fricção Linear e Comparação com Processos de Soldadura por Fusão”, 3as Jornadas Politécnicas de Engenharia – Produção e Tecnologia Automóvel, Documento: PTA_6, Instituto Superior de Engenharia de Coimbra, 19 e 20 de Novembro de 2003.

[9] Fraga, E. “Modelação Numérica e Validação do Processo de Soldadura por Fricção Linear”, MSc Thesis, IST-Universidade Técnica de Lisboa, September 2004.

[10] Fraga E., Vilaça P., Quintino L., “Analytical and Numerical Modelling of FSW”, 2nd FSW Modelling and Flow Visualisation Seminar at GKSS-Forschungszentrum, Germany, 31st January - 1st February, 2005.

[11] Rosenthal, D., The Theory of Moving Sources of Heat and Its Application to Metal Treatments. Transactions of the ASME, 1946, pp.849-866.

[12] Vilaça P., Quintino L., Dos Santos J., “iSTIR – Analytical thermal model for friction stir welding”. Journal of Materials Processing Technology, accepted for publication in 15 December 2004.

[13] Vilaça P., Quintino L., Sheiki S., dos Santos J., “Determination of the Thermal Efficiency in FSW via Analytical Modelling Formulation”, 56th Annual Assembly of the International Institute of Welding, Commission XII, Bucharest, July 06-11, 2003.

[14] Zener, C. and Hollomon, J.H., “Effect of Strain Rate Upon Plastic Flow of Steel”, Journal of Applied Physics, vol. 15, pp. 22-32, 1944.

[15] Shepard, T. and Wright, D.S., ”Determination of flow stress: Part 1 constitutive equation for aluminum alloys at elevated temperatures”, Metals Technology, pp. 215-223, June 1979.

13