Process parameter optimization and characterization of friction stir welding of aluminum alloys.

Vijayan, S. ; Raju, R.

Introduction

Welding is one of the vital and widely used manufacturing processes in any manufacturing/production industries. The welding technology has grown substantially in aerospace and shipping industries with a common goal of achieving higher strength and weld/joint efficiency. Selection of optimal welding conditions is a key factor in achieving this condition. The manufacturer has to set the process related controllable variables at their operating conditions with minimum effect of uncontrollable or noise variables on the levels and variability in the outputs.

The study of Friction Stir Welding focuses on the features of process parameters like tool profile, input work materials, and machine parameter setting which influences process efficiency and output quality characteristics (or responses).A significant improvement in process efficiency may be obtained by process parameter optimization that identifies and determines the regions of critical process control factors leading to desired outputs or responses with acceptable variations ensuring a lower cost of manufacturing. [1]

In this paper an approach that is able to evaluate the influence of the input parameters on the strength of friction stir welding process is described. Based on this approach a setting of the input parameters, which ensures a maximal strength of the welding process with respect to uncontrollable input parameters can be identified. This paper is organized as

* General issues in fusion welding

* Aluminum welding

* Friction stir welding and its benefits

* FSW Process parameters and its effects

* Process parameter optimization and its tools

* Research gap

* Objectives of this research

* Conceptual research framework

* Scheme of investigation

* Conclusion

General problems in fusion welding

* Deterioration of mechanical properties due to melting & re-solidification.

* Hot cracking and solidification cracking.

* Inclusions of Oxygen, Hydrogen and Nitrogen from atmosphere.

* Porosity.

* Filler material.

* Flux and Shielding gas.

* Environmental problems due to flue gases.

* Consumption of high energy

Aluminum Welding

Aluminum is light in weight, yet some of its alloys have strengths exceeding mild steel. The main characteristics of aluminum, which influences welding, are hydrogen solubility, aluminum oxides, thermal conductivity, thermal expansion and solidification shrinkage, filler material and non -discoloration [2]

Hydrogen solubility

The moisture or hydro carbons on the surface might decompose beneath the arc, and create hydrogen. Hydrogen is extremely soluble in molten aluminum, so when the weld freezes, the hydrogen can get trapped in the weld itself. That is the main source of porosity in aluminum welds.

Aluminum oxides

Aluminum and its alloys rapidly develop a tenacious, refractory oxide film when exposed to air. Aluminum oxides melts at about 2066[degree] C or about three times the melting point of the aluminum alloy itself. It is obvious in welding that the base metal will be melted long before the oxide. Aluminum oxide is an insulator. It does all kinds of wild and crazy things to arc. It can cause preferential arcing or if the welding plates are thick enough, no arcing at all. It is a major cause of poor fusion in aluminum welds.

Filler material

The filler metal composition does not match with the base metal. This is true, the matching filler 2319, which contains slightly higher content of Ti and Zr. The loss of strength is due to melting and quick solidification, which renders all the strengthening precipitates to dissolve and the material is as good as a cast material solute segregation and large columnar grains [3].The cleanliness of the filler metal is another consideration. The importance of the storage and handling of the filler material is often underestimated.

High Thermal conductivity

The thermal conductivity of aluminum is about six times that of steel. If aluminum is welded too slowly, the heat travels ahead of the arc. Even thought most aluminum alloys melts in the range of 5650 to 6480 C ,the high thermal conductivity of aluminum necessities that a more intense heat be employed, than that used in welding steel. It can affect mechanical properties.

Thermal expansion

Thermal expansion is about twice that of steel. Of particular concern here is the greater expansion in thinner materials. Solidification shrinkage in aluminum weld metal is about 6% by volume and it can be the main cause for distortion, especially in thicker welds.

Non Discoloration

Unlike steel, aluminum does not change color when it is heated up. In welding in metal, the operator has to get in position where he can see the molten pool and the end of the electrode tip if he expects to make good welds. Aluminum surfaces are also highly reflective as such the welders require protection against radiation

Friction Stir Welding

Friction stir welding (FSW) was invented at The Welding Institute (TWI) of UK in 1991 as a Solid-state joining technique, and it was initially applied to aluminum alloys [4,5]. The basic concept of FSW is remarkably simple. A non-consumable rotating tool with a specially designed pin and shoulder is inserted into the abutting edges of sheets or plates to be joined and traversed along the line of joint (Fig. 1).

The tool serves two primary functions:

(a) Heating of work piece, and

(b) Movement of material to produce the joint

The heating is accomplished by friction between the tool and the work piece and plastic deformation of work piece. The localized heating softens the material around the pin and combination of tool rotation and translation leads to movement of material from the front of the pin to the back of the pin. As a result of this process a joint is produced in 'solid state'. Because of various geometrical features of the tool, the material movement around the pin can be quite complex [6]. During FSW process, the material undergoes intense plastic deformation at elevated temperature, resulting in generation of fine and equiaxed recrystallized grains [7-10]. The fine microstructure in friction stir welds produces good mechanical properties

[FIGURE 1 OMITTED]

FSW joints usually consist of four different regions as shown in Fig. 2. They are: (a) unaffected base metal; (b) heat affected zone (HAZ); (c) thermo-mechanically affected zone (TMAZ) and (d) friction stir processed (FSP) zone. The formation of above regions is affected by the material flow behavior under the action of rotating non-consumable tool. However, the material flow behavior is predominantly influenced by the FSW tool profiles, FSW tool dimensions and FSW process parameters [11,12].

[FIGURE 2 OMITTED]

Benefits of FSW

Defect free welds with good mechanical properties have been made in a variety of aluminum alloys, even those previously thought to be not weldable. [13] Due to the absence of parent metal melting, the new FSW process is observed to offer several advantages over fusion welding [4,14 -15].Key benefits of friction stir welding with respect to the Metallurgical, Environment and Energy are listed bellow [16]

Metallurgical benefits

1. Solid phase process

2. Low distortion of work piece

3. Good dimensional stability and repeatability

4. No loss of alloying elements

5. Excellent metallurgical properties in the joint area

6. Fine microstructure. Absence of cracking Replace multiple parts joined by fasteners

Environmental benefits

1. No shielding gas required

2. No surface cleaning required

3. Eliminate grinding wastes

4. Eliminate solvents required for degreasing

5. Consumable materials saving, such as rugs, wire or any other gases

Energy benefits

1. Improved materials use (e.g., joining different thickness) allows reduction in weight

2. Only 2.5% of the energy needed for a laser weld

3. Decreased fuel consumption in light weight aircraft, automotive and ship applications

Process Parameter

The tool geometry, welding parameters, joint designs are the significant parameters which significant effect on the material flow pattern and temperature distribution, thereby influencing the micro structural evolution of material. [16]. The detailed list of FSW process parameters are listed bellow. FSW process parameters

1. Rotational speed (rpm)

2. Welding speed (mm/s)

3. Axial force (KN)

4. Tool geometry

(i) D/d ratio of tool

(ii) Pin length (mm)

(iii) Tool shoulder diameter, D (mm)

(iv) Pin diameter, d (mm)

(v) Tool inclined angle ([degree])

Effect of tool rotational speed

Rotational speed appears to be the most significant process variable since it also tends to influence the translational velocity. Very high rotational speeds (>10,000 rpm) could raise strain rate, and there by influence the recrystallisation process; which in turn could influence the FSW process [17]. Higher tool rotational speed resulted in a higher temperature and slower cooling rate in the FSP zone after welding. A higher rotational speed causes excessive release of stirred materials to the upper surface, which resultantly left voids in the FSP zone. Lower heat input condition due to lower rotational speed resulted in lack of stirring. The area of the FSP zone decreases with and decreasing the tool rotation speed and affect the temperature distribution in the FSP zone. [18].

Effect of axial load

In friction stir welding the shoulder force was directly responsible for the plunge depth of the tool pin into the surface of the work piece. The degree of material mixing and inter diffusion ,the thickness of the deformed aluminum lamellae, and material flow patterns highly depend upon the geometry of the tool, welding temperature, and material flow stress. these in turn depends on the axial force. [19] The heat generation in the FSW process depends on the friction coefficient, friction area between the tool shoulder and the work piece surface as well as on the rotational speed of the welding head pin and the pressure applied to the welding head shoulder. [20]

Effect of tool pin profile

The primary function of the non-consumable rotating tool pin is to stir the plasticized metal and move the same behind it to have good joint [21]. Pin profile plays a crucial role in material flow and in turn regulates the welding speed of the FSW process. The pin generally has cylindrical plain, frustum tapered, threaded and flat surfaces. Pin profiles with flat faces (square and triangular) are associated with eccentricity. This eccentricity allows incompressible material to pass around the pin profile.

Eccentricity of the rotating object is related to dynamic orbit due to eccentricity [22]. This dynamic orbit is the part of the FSW process. In addition, the triangular and square pin profiles produce a pulsating stirring action in the flowing material due to flat faces. The square pin profile produces 80 pulses/s and triangular pin profile produces 60 pulses/s when the tool rotates at a speed of 1200 rpm. There is no such pulsating action in the case of cylindrical, tapered and threaded pin profiles The higher number of pulsating action experienced in the stir zone of square pin profile produces very fine microstructure and in turn yields higher strength and hardness. [13]

[ILLUSTRATION OMITTED]

FSW tool dimensions

[ILLUSTRATION OMITTED]

Effect on welding speed

The welding speed has a vital role in productivity of welded joints. The ultimate tensile strength of AA 5083 aluminum decreased significantly when the transverse speed was increased. The voids were formed due to poor consolidation of the welded interface when the tool traveled at higher transverse speeds, and hence lower heat inputs. The reduced plasticity and rates of diffusion in the material might have resulted in a week interface. Low transverse rates resulted a weld with a higher strength [23].When the welding speed was faster than the critical value, welding defects were produced in the joints. [24]

Process Parameter Optimization and its tools

In any process the input parameters play a vital role in determining the output characteristics of the process. How to control the input process parameters is the common problem in any process. In order to overcome this problem, various optimization methods can be applied to define the desired output variables through developing mathematical models to specify the input and output variables. In the last two decades, design of experiment (DoE) techniques has been carried out such optimization. Evolutionary algorithms and computational network have also grown rapidly and been applied in different areas. [25]

Process parameter optimization in these machining operations is required to be undertaken in two stages:

(i) Modeling of input-output and in-process parameter relationship,

(ii) Determination of optimal or near-optimal cutting conditions.

Modeling of input-output and in-process parameter relationship is considered as an abstract representation of a process linking causes and effects or transforming process inputs into outputs [26]. The resulting model provides the basic mathematical input required for formulation of the process objective function. An optimization technique provides optimal or near-optimal solution(s) to the overall optimization problem formulated, and subsequently implemented in actual metal cutting process. The classification of modeling (a) and optimization techniques (b) is listed bellow. [27]

[ILLUSTRATION OMITTED]

Taguchi method

Taguchi addresses quality in two main areas: off-line and on-line quality control. Both of these areas are very cost sensitive in the decisions that are made with respect to the activities in each. Offline quality control refers to the improvement in quality in the product and process development stages. On-line quality control refers to the monitoring of current manufacturing processes to verify the quality levels produced [28]. The most important difference between a classical experimental design and a Taguchi-method-based robust design technique is that the former tends to focus solely on the mean of the quality characteristic while the later considers the minimization of the variance of the characteristic of interest. Although, the Taguchi method has drawn much criticism due to several major limitations, it has been able to solve single response problems effectively. The Taguchi method attempts to optimize a process or product design and is based upon three stages, as follows:

1. Concept Design or System Design

2. Parameter Design

3. Tolerance Design.

The following are the steps to be followed for process parameter optimization. [29]

Step 1: Determine the quality characteristic to be optimized

Step 2: Identify the Noise factors and Test conditions

Step 3: Identify the control factors and their Alternative Levels

Step 4: Design the Matrix Experiment and Define the data analysis procedure.

Step 5: Conduct the matrix Experiment

Step 6: Analyze the Data and determine optimum levels for control factors

Step 7: Predict the performance at these levels.

Response Surface Design Methodology (RSM)

The RSM is a dynamic and foremost important tool of design of experiment (DoE) wherein the relationship between response(s) of a process with its input decision variables is mapped to achieve the objective of maximization or minimization of the response properties. It is a set of statistical DOE techniques, intrinsic regression modeling, and optimization methods useful for any field of engineering. [30] The first necessary step in RSM is to map response(s), Y as a function of independent decision variables (X1,.,Xn).If the model is adequate, hill climbing or descending technique for maximization or minimization problem is attempted and the same mapping technique is repeated. In the vicinity of optimal point, a second order regression model is generally found adequate. [31].

The response function representing any of the mechanical properties of the weld can be expressed as

Y =f (RS, AF, WS, T)

Where Y is the response, e.g. yield strength, joint efficiency etc., RS is the rotational speed of the tool in rpm, AF is the Axial force in KN, WS is the Transverse speed in mm/sec and T is the tool geometry.

The second-order polynomial (regression) equation used to represent the response surface for K factors is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Where b0 is the free term of the regression equation, the coefficients b1, b2, ... , bk the linear terms, the coefficients b11, b22, ... , bkk the quadratic terms and the coefficients b12, b13, ... ,bk-1k are the interaction terms [32 -36].

For three factors, the selected polynomial could be expressed as given below: Y=bo+b1RS+b2AF+b3WS+b4T+b12RS*AF+b13RS*WS+b14RS*T+b23 AF*WS+b24 AF*T+b34 WS*T+b11 RS2+b22 AF2+b33 WS2+b44 T2

Genetic Algorithms [GA]

The working of GA [37] generally preferred for large and complex process parameter optimization problems, is based on three basic operators, viz., reproduction, crossover, and mutation, in order to offer a population of solutions. The algorithm creates new population from an initial random population (obtained from different feasible combination of process decision variables) by reproduction, crossover, and mutation in an iterative process. The selection, crossover and mutation on initial population create a new generation, which is evaluated with pre-defined termination criteria. The procedure continues by considering current population as initial population till the termination criteria are reached. GA is very appealing for single and multi-objective optimization problems [38], and some of its advantages are as follows: (i) as it is not based on gradient-based information, it does not require the continuity or convexity of the design space, (ii) it can explore large search space and its search direction or transition rule is probabilistic, not deterministic, in nature, and hence, the chance of avoiding local optimality is more, (iii) it works with a population of solution points rather than a single solution point as in conventional techniques, and provides multiple near-optimal solutions, (iv) it has the ability to solve convex, and multi-modal function, multiple objectives and non-linear response function problems, and it may be applied to both discrete and continuous objective functions.

Research gap

From the literature reviewed, the following are the potentials areas to do research and they are listed bellow.

* FSW during FSW process is quite complicated and poorly understood process parameters have not been optimized in a systematic way

* Research is going on the tool profile

* Material flow.

* A very limited studies on the tool wear during FSW is reported

Objectives of this research

The objectives of this proposed research is listed bellow

* To optimize the process parameters of FSW process.

* To investigate the effect of parameters on mechanical properties of the weldment. (micro and macro structure)

* To evaluate the Metallurgical properties of AA 5083 & AA2219

Scheme of investigation

* Identifying the important FSW welding parameters

* Finding the upper and lower limits of the identified parameters

* Developing the experimental design matrix

* Conducting the experiments as per the design matrix

* Recording the responses

* Developing mathematical models

* Identifying the significant factors

* Checking the adequacy of the developed models

* Results obtained are tested for confirmation

Conceptual research framework

The research framework has three components. They are simulation, characterization and process parameter optimization. The simulation is carried out by means of finite element modeling. Based on the numerical studies the behaviors of the input process variables can be simulated to the desired output. And a thermo--mechanical model is developed to predict the material deformation and temperature histories in FSW process. A commercial finite element package--ABAQUS/EXPLICIT is used for simulation. The characterization is done with the help of optical microscope. The TEM and SEM pictures of FSW were captured and investigated to the base metal structures. In the case of process parameter optimization both convectional and non convectional methods are used to predict the optimized process parameters. The detailed flow chart of the research framework is given bellow.

[ILLUSTRATION OMITTED]

Conclusion

The above research framework for process parameter optimization in FSW attempts to provide a single, unified and systematic approach to determine optimal or near optimal welding conditions. Generally this frame work helps us to visualize the difference between the convectional and non convectional methods of process parameter optimization. And it lays the platform for comparison as well as combination of above said methods in process parameter optimization.

References

[1] Montgomery, D.C., (1990). Introduction to Statistical Quality Control (2 edition) New York: Willey.

[2] N.R Mandal (2005) Aluminium Welding (2 edition) Narosa Publishing Home Pvt Ltd.

[3] S.R. Koteswara rao, R. Madhusudhana Reddy, G. Srinivasa rao, K. Kamaraj,M. Prasad Rao K, Mater. Charact. 40 (2005) 236-248.

[4] W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Murch, P. Templesmith, C.J. Dawes, G.B. Patent Application No.9125978.8 (December 1991).

[5] C. Dawes, W. Thomas, TWI Bulletin 6, November/December 1995, p. 124.

[6] B. London, M. Mahoney, B. Bingel, M. Calabrese, D.Waldron, in: Proceedings of the Third International Symposium on Friction Stir Welding, Kobe, Japan, 27-28 September, 2001.

[7] C.G. Rhodes, M.W. Mahoney, W.H. Bingel, R.A. Spurling, C.C. Bampton, Scripta Mater. 36 (1997) 69.

[8] G. Liu, L.E. Murr, C.S. Niou, J.C. McClure, F.R. Vega, Scripta Mater. 37 (1997) 355.

[9] K.V. Jata, S.L. Semiatin, Scripta Mater. 43 (2000) 743.

[10] S. Benavides, Y. Li, L.E. Murr, D. Brown, J.C. McClure, Scripta Mater. 41 (1999) 809.

[11] Yingchun Chen, Huijie Liu, Feng Jicai, Mater. Sci. Eng. A 420 (2006) 21-25.

[12] H.J. Liu, Y.C. Chen, J.C. Feng, Scripta Mater. 55 (2006) 231-234.

[13] K. Elangovan, V. Balasubramanian / Materials and Design (2007)

[14] Threadgill PL. Friction stir welding--The state of the art, Bulletin 678, TWI, UK, 1999.

[15] Peel M, Steuwer A, Preuss M, Withers PJ. Microstructure, mechanical properties and residual stresses as a function of welding speed in AA5083 friction stir welds. Acta Mater 2003;51:4791-801.

[16] R.S. Mishra, Z.Y. Ma / Materials Science and Engineering R 50 (2005) 1-78

[17] V.F. Olga, Scripta Mater. 38 (5) (1998) 703-708.

[18] B.L.Won, Yun FY.M., J.B Seung, Mater. Trans. 45 (5) (2004) 1700-1705.

[19] Ouyang j.h and Kovacevic R(2002) Journal of Material Engineering and Performance,Vol. 11, No. 1,pp 51-63

[20] Buffa G,Hua J,Shivpuri R and Fratini L (2006) Material Science and Engineering, Vol.419,pp 389-396.

[21] Thomas WM, Dolby RE. Friction stirs welding developments. in: Proceedings of 6th international trends in welding research conference, 2002. p. 203-11.

[22] Thomas WM, Nicholas ED. Friction stir welding for the transportation industries. Mater Des 1997; 18:269-73.

[23] Peel M.J,Preuss M, Steuwer A, and Withers P.J (2003) Friction Stir Welds, Acta Materialia, Vol 51 ,PP 4791 -4801.

[24] Yung Chun Chen, Huijie Liu and Jicai Feng (2006) Material Science & Engineering A, Vol.420, pp 21 -25

[25] K.Y Benyounis , A.G Olabi Advances in Engineering Software 39 (2008) 483-496.

[26] Markos, S., Viharos, Zs.J., & Monostori, L. (1998). Quality-oriented, comprehensive modelling of machining processes. Sixth ISMQC IMEKO symposium on metrology for quality control in production (pp. 67-74).

[27] I. Mukherjee, P.K. Ray / Computers & Industrial Engineering 50 (2006) 15-34

[28] Ross PJ (1996) Taguchi Techniques for Quality Engineering. McGraw-Hill, Singapore.

[29] Saurav Datta , Asish Bandyaopadhyay, Pradip kumar Pal(2006) Int. J. Adv. Manufacturing Technology. Springer

[30] Montgomery, D. C. (2001). Design and analysis of experiment (5th ed.). New York: Wiley.

[31] Mayers, R. H., & Carter, W. H. (1973). Response surface technique for dual response system. Technometrics, 15(2), 301-317.

[32] D.C. Montgomery, G.C. Runger, Applied Statistics and Probability for Engineers, second ed., John Wiley & Sons Inc., New York,1999.

[33] Walpole, Myers, Probability and Statistics for Engineers and Scientists, sixth ed., Prentice Hall, New Jersey, 1998.

[34] P.N. Cheremisinoff, Practical Statistics for Engineers and Scientists, Technomic Publishing Co. Inc., USA, 1987.

[35] W.G. Cochran, G.M. Cox, Experimental Designs, second ed., John Wiley & Sons, Singapore, 1957.

[36] A.I. Khuri, J.A. Cornell, Response Surfaces, Designs and Analyses, Marcell Dekker Inc., New York, 1996

[37] Goldberg, D. E. (2002). Genetic algorithms in search, optimization, and machine learning. Boston, MA: Pearson Education Asia (5th print).

[38] Deb, K. (2002). Multi-objective optimization using evolutionary algorithm (1st ed.). Chichester: Wiley

S. Vijayan (a) and R. Raju (b)

(a) Department of Mechanical Engineering, SSN College of Engineering, Chennai, Tamilnadu, India.

(b) Department of Industrial Engineering, Anna University, Chennai, Tamilnadu, India