Introduction

Friction stir welding involves the joining of metals without fusion or filler materials. It is a solid–state joining process that takes place below the melting point of the materials to be joined. The maximum temperature reached was generally of the order of 0.8 of the liquidus temperature of the material being welded. This welding process was invented at The Welding Institute, UK in the early 90’s [1]. The welds were created by the combined action of frictional heating and mechanical deformation due to a rotating tool. Welds produced by FSW generally exhibits a higher strength and formability due to the low welding temperatures compared to fusion welding processes [2, 3]. FSW can potentially replace the riveting and resistance spot welding of aluminum and steel sheets in the aircraft and automotive industries, respectively [4]. Friction stir welding has been widely researched in marine and aerospace application because it is thought to be the appropriate technique without defects [5], which may occur in arc welding of aluminum alloys. The disadvantageous effects of arc welding like very high localized heating beyond the melting temperature, residual stresses and deformations are minimized in FSW, as the heat generated is less than the melting point of base plate. In marine application, marine grade steel is generally used for haul and superstructures. But aluminum has become a very good choice as it is lightweight, and allows for minimal rusting of aluminum to take place as compared to steel. The geometrical effect of FSW tool on welding quality [6–10] is a crucial part of this welding process. Some authors analyzed the influence of the tool rotation speed [11, 12], welding speed [13–16] and both parameters simultaneously [17–22] of 7XXX, 6XXX and 5XXX welds by considering the same tool geometry. They observed that different rotational speeds did not result in significant differences in the hardness profile in FSW welds, except for the width of the softened region in the weld of 6063-T5 Al. They suggested that frictional heat and plastic flow of plasticized material [23] during friction stir welding created fine and equiaxed [16] grains in the stir zone, and elongated and recovered grains in the Thermo-mechanically affected zone (TMAZ).The aim of the present study is to investigate the effect of process parameters [24, 25] viz. tool geometry, plunging force and rotational speed of the tool on mechanical properties [26–29] of similar friction stir butt welds AA1100.

Model Methodology

Let us consider a cylindrical tool whose shoulder radius is R S and radius of its probe is R Pt , whereas, considering height of the probe as H p as shown in Fig. 1. If the angular speed of the tool is considered as ω and it is given a plunging force P n due to which a pressure P is generated by the tool on the shoulder surface, due to this pressure the generation of total heat is calculated as follows.

Fig. 1
figure 1

Schematic diagram of flat cylindrical shoulder and cylindrical probe

Full size image

An elemental ring of thickness δr is taken which is at r distance away from the centre O shown in Fig. 2. Then the area of the elemental ring will be,

$$\delta A = 2\pi r\delta r$$
(1)

Normal force acting on this elemental ring,

$$\delta N = P \cdot \delta A = P \cdot 2\pi r\delta r$$
(2)

If one considers the frictional co-efficient between the tool and workpiece to be μ, the frictional force will be,

$$\delta F = \mu \cdot \delta N = 2\mu P\pi r\delta r$$
(3)

Heat generation due to elemental ring,

$$\delta Q_{Sp} = \delta F \times V_{t} = 2\mu P\pi r\delta r \cdot \omega r$$
(4)

where \(V_{t}\) is the tangential velocity. Integrating both sides of Eq. (4) one gets,

$$Q_{Sp} = 2\pi \mu \omega P\mathop \int \limits_{{R_{Pt} }}^{{R_{S} }} r^{2} \delta r$$

or

$$Q_{Sp} = \frac{2}{3}\pi \mu \omega P \left( {R_{S}^{3} - R_{Pt}^{3} } \right)$$
(5)

Let us consider the depth by which the tool shoulder penetrate the work plate due to plunging force in welding area, is h 1 and the tool moves with a velocity of V. Now the yield strength of the material at 80 % of the melting point temperature is considered as \(S_{ys}^{*}\). Since the tool is moving forward with a traverse velocity then one force will act upon the front half curved surface area of the shoulder. The area responsible for heat generation,

$$A = \pi R_{S} \cdot h_{1}$$
(6)

Frictional force which is responsible for heat generation is,

$$F = \mu S_{ys}^{*} \cdot A = \mu \pi S_{ys}^{*} R_{S} h_{1}$$
(7)
Fig. 2
figure 2

Shoulder small elementary area

Full size image

Heat Generation

$$Q_{Sm} = F \times V = \mu \pi S_{ys}^{*} R_{S} h_{1} V$$
(8)

So, the total heat generation,

$$Q_{s} = Q_{Sp} + Q_{Sm}$$
$$Q_{s} = \frac{2}{3}\pi \mu \omega P \left( {R_{S}^{3} - R_{Pt}^{3} } \right) + \mu \pi S_{ys}^{*} R_{S} h_{1} V$$
(9)

Heat Generation in Cylindrical Probe

There are two important surfaces of probe. These are responsible for heat generation i.e. (i) probe tip surface and (ii) probe side surface.

The details of mathematical models of heat generation [30, 31] both on probe tip and side surface are described below.

Heat Generation Due to Vertical Pressure at Probe Tip

Similar to Eq. (5),

$$Q_{1} = \frac{2}{3}\pi \mu \omega P \cdot R_{Pt}^{3}$$
(10)

Heat Generation due to the Rotational Movement of the Probe Side Surface

If one considers τ to be the shear strength of the workpiece and \(\tau^{*}\) be the shear strength of the material at 0.8 of its melting point temperature, then the force acting on the probe on its surface will be,

$$F_{p} = \mu \tau^{*} \cdot 2\pi R_{Pt} H_{P}$$
(11)

Heat generation by the curved surface area,

$$Q_{2} = 2\mu \pi \tau^{*} \omega R_{Pt}^{2} \cdot H_{P}$$
(12)

Heat Generation Due Traverse Speed of the Tool by the Probe Side Surface

Let us assume that probe is moving with a velocity of V and the yield strength of the workpiece material at 0.8 of its melting temperature be \(S_{ys}^{*}\). Here only half of the curved surface area is responsible for heat generation during the travelling of probe i.e.

$$A = \pi rH_{P}$$

Therefore, force acting on the probe due to travelling on workpiece is,

$$F_{p} = \mu S_{ys}^{*} \pi R_{Pt} H_{P}$$
(13)

Therefore heat generation due to travelling of probe on workpiece is,

$$Q_{3} = \mu S_{ys}^{*} \pi R_{Pt} \cdot H_{P} \cdot V$$
(14)

So the total heat generation by the probe is calculated by adding Eq. (10), (12) and (14),

$$Q_{p} = Q_{1} + Q_{2} + Q_{3}$$

or,

$$Q_{p} = \frac{2}{3}\pi \mu \omega P \cdot R_{Pt}^{3} + 2\mu \pi \tau^{*} \omega R_{Pt}^{2} \cdot H_{P} + \mu S_{ys}^{*} \pi R_{Pt} \cdot H_{P}$$
(15)

Therefore, total heat generation by flat cylindrical shoulder with cylindrical probe is calculated by,

$$Q = Q_{s} + Q_{p}$$

or

$$Q = \frac{2}{3}\pi \mu \omega P \left( {R_{S}^{3} - R_{Pt}^{3} } \right) + \mu \pi S_{ys}^{*} R_{S} h_{1} V + \frac{2}{3}\pi \mu \omega P \cdot R_{Pt}^{3} + 2\mu \pi \tau^{*} \omega R_{Pt}^{2} \cdot H_{P} + \mu S_{ys}^{*} \pi R_{Pt} \cdot H_{P} \cdot V$$
(16)

Similarly, for flat cylindrical tool with conical pin the total heat generated can be calculated by the below mentioned equation,

$$Q = \frac{2}{3}\pi \mu \omega P \left( {R_{S}^{3} - R_{Pt}^{3} } \right) + \mu \pi S_{ys}^{*} R_{S} h_{1} V + \frac{2}{3}\pi \mu \omega P \left( {R_{S}^{3} - R_{Pt}^{3} } \right) + \left\{ {\frac{2}{3}\pi \mu \omega \tau^{*} \left( {R_{Pb}^{2} + R_{Pb} R_{Pt} + R_{Pt}^{2} } \right) + \frac{\mu }{2}\pi S_{ys}^{*} V\left( {R_{Pb} + R_{Pt} } \right)} \right\}\sqrt {\left( {R_{Pb} - R_{Pt} } \right)^{2} + H_{P}^{2} }$$
(17)

Heat Flux Calculation

Heat Flux in Cylindrical Shoulder

Heat flux (q) can be obtained by dividing Eq. (4) by Eq. (1)

$${\text{i}} . {\text{e}} .\quad \frac{{\delta Q_{Sp} }}{\delta A} = \frac{{2\mu \pi \omega Pr^{2} \delta r}}{2\pi r\delta r}$$
(18)

or

$$q = \mu \omega \Pr$$
(19)

Multiplying and dividing Eq. (19) by \(\frac{2}{3}\pi \left( {R_{S}^{3} - r_{p}^{3} } \right)\), one gets,

$$q = \frac{{\frac{2}{3}\mu \pi \omega P\left( {R_{S}^{3} - R_{pb}^{3} } \right)r}}{{\frac{2}{3}\pi \left( {R_{S}^{3} - R_{pb}^{3} } \right)}}$$
(20)

Using Eq. (5),

$$q = \frac{{3Q_{Sp} r}}{{2\pi \left( {R_{S}^{3} - R_{pb}^{3} } \right) }}$$
(21)

Heat flux due to travelling of shoulder

As the radius of shoulder is constant at the curved surface area therefore heat flux is taken as uniform. Heat generation is taking place at half of the depressing curved surface area which is given in Eq. (6).

Therefore heat flux can be calculated by using Eq. (8) as,

$$q = \frac{{Q_{Sm} }}{A}$$
(22)

Heat Flux in Probe

There are two important areas where flux is needed to be calculated (a) Curved surface area of the probe and (b) Tip of the probe.

For Curved Surface Area

Since radius of the probe is not changing the heat flux through curved surface area is taken as uniform. Curved surface area of the probe

$$A = 2\pi R_{Pt} H_{P}$$
(23)

Therefore heat flux is calculated as,

$$q = \frac{{Q_{2} + Q_{3} }}{A}$$
(24)

where Q 2 and Q 3 are taken from Eq. (12) and (14).

For Tip of the Probe

Similar to Eq. (21),

$$q = \frac{{3Q_{1} r}}{{2\pi R_{Pb}^{3} }}$$
(25)

Similarly one can calculate heat fFlux for conical shoulder and conical pin.

Experimental Procedure

Experimental Details

Two similar aluminium alloy of AA1100 grade test samples of size 200 mm long, 100 mm wide and 6 mm thick plates were welded with the below referred FSW tools. Specifications of milling machine used for FSW experimentation are vertical type milling machine, motor capacity: 7.5 HP, rotational Speeds: 50–1500 rpm, traversing speeds: 22–555 mm/min. The tool was rigidly fixed on an arbor. The workpiece edges are clamped to the machine bed without any root gap to restrict the movement in all degrees of freedom to with stand mainly plunging and translational forces of the tool. The butting line of two workpieces was seen to match exactly with the center of the tool pin. After allowing the tool to plunge on the butted plate and visually ensuring full contact of the tool shoulder with the plate surface, the bed horizontal movement was switched on and continued up to the end. A typical FSW setup is shown in Fig. 3. For measurement of tool plunging force a 5000 kg capacity load pad was used. Agilent data logger 34,970 A was used to record temperature data. The FSW tools were fabricated in house using SS310 shown in Fig. 4.

Fig. 3
figure 3

Friction stir welding setup

Full size image
Fig. 4
figure 4

Conical tools (top) and cylindrical tools (bottom)

Full size image

Four flat shouldered but different pin dimensioned FSW tools have been used in this work. Two are of flat shouldered straight cylindrical pin tool and other two are flat shouldered tapered cylindrical pin shown in Fig. 4. The FSW tools were fabricated in workshop using SS-310. With the constant traverse speed of 112 mm/min and zero tool inclination angle the plunging forces are taken as 5100 and 6500 N in case of 1400 and 1000 rpm tool rotational speed respectively, for FSW of AA1100 grade.

Vickers Micro Hardness Test

Procedure to measure material’s resistance to localized plastic deformation is done through hardness test. Required hardness value can vary depending upon the application and purpose of use. So to calculate the actual hardness value of welding zone compared to other zone of the welded plate, the test samples were made to carried micro hardness test which have been carried out on a Vickers micro hardness testing machine (Make BUEHLER, model no: 1600-6036, loading range: 10–2000 gf). Hardness measurements were taken on cross-sections perpendicular to the welding direction. Welding was carried out using four types of FSW tools with varying process parameters as shown in Table 1.

Table 1 Different process parameters used for similar FSW of AA1100
Full size table

The hardness at the different zones of the test samples was measured. At a particular location on the weld zone ten number of hardness reading have been noted and the average of these readings was taken. The indentation load for the hardness measurement was kept at 100 gf. The variation of hardness value for all the eight test samples is shown in Fig. 5.

Fig. 5
figure 5

Vickers micro hardness for eight different FSW samples

Full size image

Tensile Test

As strength of any structure or a body is determined by its load carrying capacity so to quantify the strength or toughness or yield strength of the welded joint, tensile test was carried out and specimens were prepared after welding as per ASME standard and tensile tests have been carried out for the job specimens with 1400 rpm and 5100 N plunging force. Instron-8801 tensile testing machine was used for tensile test. The specifications of this machine are closed loop servo hydraulic dynamic testing machine, maximum load capacity is 100 kN, and actuator displacement is about 75 mm and in full scale 150 ± 5 mm.

The tensile test specimens were taken perpendicular to the welding direction from the welded test samples. All tensile tests were performed at a constant crosshead displacement rate of 4 mm/min using an Instron tensile testing machine. Tests were performed to study the load versus elongation characteristics for the specimens welded under different parameters. Load versus elongation curves were plotted for the all the four samples welded under 1400 rpm and 5100 N plunging force have shown in Fig. 6. In this result, specimens 1 and 2 were welded by the cylindrical pin tool and the specimens 3 and 4 were welded with the tool of tapered pin geometry. It is observed that the load versus elongation plots of specimen 3 and 4 have enhanced nature compared to all others. It can be shown that the ultimate tensile strength for the specimens 3 and 4 were 30 % higher than the specimens 1 and 2.

Fig. 6
figure 6

Load versus elongation plot for four samples welded by different tool geometry for same process parameters

Full size image

Numerical Thermal Analysis

Three-Dimensional Finite Element Modeling [32, 33]

A three dimensional finite element transient thermal model was developed in the present work to analyze the heat transfer and temperature distribution in a FSW process. In the thermal model the actual conditions were accommodated as far as possible. The following assumptions were made in developing the present thermal model of a FSW process.

  1. 1.

    All the thermal properties were considered as function of temperature.

  2. 2.

    Linear Newtonian convection cooling was considered on all the surfaces.

  3. 3.

    Heat generation was considered as a load.

The governing differential equation is:

$$\frac{\partial }{\partial x}\left[ {K \frac{\partial T}{\partial x}} \right] + \frac{\partial }{\partial y}\left[ {K \frac{\partial T}{\partial y}} \right] + \frac{\partial }{\partial z}\left[ {K \frac{\partial T}{\partial z}} \right] + q^{\prime\prime\prime} = \rho C\frac{\partial T}{\partial t}$$
(26)

where \(\frac{\partial }{\partial x}\left[ {K \frac{\partial T}{\partial x}} \right]\) = variation of heat flux in X direction, \(\frac{\partial }{\partial y}\left[ {K \frac{\partial T}{\partial y}} \right]\) = variation of heat flux in Y direction, \(\frac{\partial }{\partial z}\left[ {K \frac{\partial T}{\partial z}} \right] =\) variation of heat flux in Z direction, \(q^{\prime\prime\prime}\) = heat generation, ρ = density of plate material, C = specific heat and K = thermal conductivity.

The following boundary conditions were applied in the present FE model.

Equation (26) can be written as:

$$\rho c\frac{\partial T}{\partial t} = - \left\{ L \right\}^{T} \left\{ q \right\}$$
(27)

where {L} = \(\left\{ {\begin{array}{*{20}c} {\frac{\partial }{\partial x}} \\ {\frac{\partial }{\partial y}} \\ {\frac{\partial }{\partial z}} \\ \end{array} } \right\}\) = vector operator, {q} = heat flux vector.

$$\left\{ L \right\}^{T} \left\{ q \right\} = \nabla \cdot \, \left\{ q \right\}{\text{ and}}\left\{ L \right\}^{T} = \nabla T$$

where \(\nabla .\) represents divergence operator and \(\nabla\) represents grad operator.

Fourier’s law is used to relate the heat flux vector to the thermal gradient

$$\left\{ q \right\} = - \left[ D \right]\left\{ L \right\}^{T}$$

where \(\left[ D \right] = \left[ {\begin{array}{*{20}c} K & 0 & 0 \\ 0 & K & 0 \\ 0 & 0 & K \\ \end{array} } \right]\) = conductivity matrix.

Equation (27) can be written as:

$$\rho c\frac{\partial T}{\partial t} = \left\{ L \right\}^{T} \left( {\left[ D \right]\left\{ L \right\}^{T} } \right)$$
(28)

Boundary Conditions

Initial Condition

A specified initial temperature for the line heating that covers all the elements of the specimen = T α for t = 0, where T α  = the ambient temperature

To develop second and third boundary conditions one should consider energy balance at the work surface as:

$${\text{Heat}}\,{\text{supply}} = {\text{Heat}}\,{\text{loss}}$$

First Boundary Condition

A specific heat flow acting over surface S 1

$$q_{n} = - q_{sup}$$

or

$$- K \frac{\partial T}{\partial n} = h_{f} \left( {T - T_{\alpha } } \right)\quad {\text{for}}\,t > 0$$
(29)

The quantity \(q_{n}\) represents the component of the conduction heat flux vector normal to the work surface. The quantity \(q_{sup}\) represents the heat flux supplied to the work surface in \(\frac{\text{W}}{{{\text{m}}^{2} }}\), from an external welding arc.

$$q_{n} = \left\{ q \right\}^{T} \left\{ n \right\}$$
(30)

on the surface S 1 for t > 0, where \(\left\{ n \right\}\) = unit outward normal vector.

Second Boundary Condition

Considering heat loss (\(q_{conv}\)) due to convection over surface S 2 (Newton’s law of cooling):

$$q_{n} = q_{conv}$$

or

$$\left\{ q \right\}^{T} \left\{ n \right\} = h_{f} \left( {T - T_{\alpha } } \right)$$
(31)

on surface S 2 for t > 0.

Pre-multiplying Eq. (28) by a virtual change in temperature, integrating over the volume of the element, combining with Eqs. (30) and (31), and with some algebraic manipulation one gets:

$$\mathop \int \limits_{vol}^{{}} \left( {\rho c\delta T\left( {\frac{\partial T}{\partial t}} \right) + \left\{ L \right\}^{T} \left( {\delta T} \right)\left( {\left[ D \right]\left\{ L \right\}T} \right)} \right)d\left( {vol} \right) = \mathop \int \limits_{{S_{1} }}^{{}} \delta Tq_{sup} d\left( {S_{1} } \right) + \mathop \int \limits_{{S_{2} }}^{{}} \delta Th_{f} \left( {T_{\alpha } - T} \right)d\left( {S_{2} } \right)$$
(32)

where vol = volume of the element, \(\delta T\) = an allowable virtual temperature \(\left( { = \delta T\left( {x,y,z,t} \right)} \right)\).

As stated before, the variable T is allowed to vary both in space and time. This dependency is expressed as:

$$T = \left\{ N \right\}^{T} \left\{ {T_{e} } \right\}$$
(33)

where \(T = T\left( {x,y,z,t} \right)\) = Temperature, \(\left\{ N \right\} = \left\{ {N\left( {x,y,z} \right)} \right\} = {\text{element shape function}}\), \(\left\{ {T_{e} } \right\} = \left\{ {T_{e} \left( t \right)} \right\}\) = nodal temperature vector.

The time derivatives of Eq. (33) may be written as:

$${\dot{\text{T}}} = \frac{\partial T}{\partial t} = \left\{ N \right\}^{T} \left\{ {T_{e} } \right\}$$
(34)

\(\delta T\) has the same form as T:

$$\delta T = \left\{ {\delta T_{e} } \right\}^{T} \left\{ N \right\}$$
(35)

The combination of \(\left\{ L \right\}T\) is written as:

$$\left\{ L \right\}T = \left[ B \right]\left\{ {T_{e} } \right\}$$

where

$$\left[ B \right] = \left\{ L \right\}\left\{ N \right\}^{T}$$
(36)

The variational statement of Eq. (32) can be combined with Eqs. (33), (34), (35) and (36) to yield:

$$\mathop \int \limits_{vol}^{{}} \rho c\left\{ {\delta T_{e} } \right\}^{T} \left\{ N \right\}\left\{ N \right\}^{T} \left\{ {{\dot{\text{T}}}_{e} } \right\}d\left( {vol} \right) + \mathop \int \limits_{vol}^{{}} \left\{ {\delta T_{e} } \right\}^{T} \left\{ B \right\}^{T} \left[ D \right]\left[ B \right]\left\{ {T_{e} } \right\}d\left( {vol} \right) = \mathop \int \limits_{{S_{1} }}^{{}} \left\{ {\delta T_{e} } \right\}^{T} \left\{ N \right\}q_{sup} d\left( {S_{1} } \right) + \mathop \int \limits_{{S_{2} }}^{{}} \left\{ {\delta T_{e} } \right\}^{T} \left\{ N \right\}h_{f} \left( {T_{\alpha } - \left\{ N \right\}^{T} \left\{ {T_{e} } \right\}} \right)d\left( {S_{2} } \right)$$
(37)

The density ρ is assumed to remain constant and specific heat C may vary over the element. Finally \(\left\{ {T_{e} } \right\}\), \(\left\{ {{\dot{\text{T}}}_{\text{e}} } \right\}\) and \(\left\{ {\delta T_{e} } \right\}\) are nodal quantities and do not vary over the element, so that they also may be taken out from the integrals. Now, since all quantities are pre-multiplied by \(\left\{ {\delta T_{e} } \right\}\) this term may also be dropped from the resulting equation. Thus Eq. (37) may be reduced to:

$$\left[ {C_{e}^{t} } \right]\left\{ {{\dot{\text{T}}}_{e} } \right\} + \left( {\left[ {K_{e}^{tb} } \right] + \left[ {K_{e}^{tc} } \right]} \right)\left\{ {T_{e} } \right\} = \left\{ {Q_{e}^{f} } \right\} + \left\{ {Q_{e}^{c} } \right\}$$

which is used for solving thermal analysis in ANSYS software.

Where \(\left[ {C_{e}^{t} } \right] = \rho \mathop \int \limits_{vol}^{{}} c\left\{ N \right\}\left\{ N \right\}^{T} d\left( {vol} \right) = {\text{element specific heat matrix}}\), \(\left[ {K_{e}^{tb} } \right] = \left[ B \right]^{T} \left[ D \right]\left[ B \right]d\left( {vol} \right) = {\text{element diffusion conductivity matrix}}\), \(\left[ {K_{e}^{tc} } \right] = \mathop \int \limits_{{S_{2} }}^{{}} h_{f} \left[ N \right]\left[ N \right]^{T} d\left( {S_{2} } \right) = {\text{element convection surface conductivity matrix}}\), \(\left\{ {Q_{e}^{f} } \right\} = \mathop \int \limits_{{S_{1} }}^{{}} \left\{ N \right\}q_{sup} d\left( {S_{1} } \right) = {\text{element heat flow vector for surface S}}1\), \(\left\{ {Q_{e}^{c} } \right\} = \mathop \int \limits_{{S_{2} }}^{{}} T_{\alpha } \left\{ N \right\}h_{f} d\left( {S_{2} } \right) = {\text{element convection surface heat flow vector}}.\)

This can be used for easily calculating temperature {\(T_{e}\)}.

Heat Source Modeling [34–40]

The input of heat is assumed to be linearly proportional to the radial distance from the center of the tool which is derived from the supposition (a) due to downward force on the workpiece by the tool creates a uniform pressure between the shoulder and the workpiece, and (b) from the work done by the frictional effect between tool-workpiece interface the heat is generated. The distribution of the rate of heat flux to work piece can then be represented as shown in Eq. (38).

$$q\left( r \right) = \frac{3Q r}{{2\pi r_{0}^{3} }}\quad {\text{for}}\,r \le r_{0}$$
(38)

where r 0 = outside radius of the shoulder of the tool, Q is the rate of heat input to the work piece.

The distribution of heat flux over the pin-plate interface due to tool pin side surface is given by,

$$q_{p} = \frac{{Q_{p} }}{{\left( A \right)_{ps} }}$$
(39)

where (A) ps  = The area of tool pin side surface.

Table 2 shows the temperature dependent material properties of aluminum alloy [41, 42] used for the transient heat transfer analysis. Constant convection co-efficient (i.e. 30 W/m2 °C) was used [43] in the analysis. The melting point of commercial grade aluminum alloy is about 660 °C. Table 3 shows the temperature dependent friction coefficient of aluminum and steel [41, 42] used for the transient heat transfer analysis. The recrystallization temperature range of commercial grade aluminum alloy is generally 325–290 °C.

Table 2 Temperature dependent material properties of Aluminum used in FE analysis
Full size table
Table 3 Temperature dependent friction coefficient of aluminum and steel
Full size table

FE Results and Discussions

For calculation of weld induced frictional heat the vertical force first considered was obtained from published literature. It was seen the vertical force varied with welding speed. The vertical force obtained from published results is given in Table 4. In this present FE model the vertical plunging force considered with different welding parameters is shown in Table 4.

Table 4 Vertical plunging force on FSW tool and welding speed
Full size table

Analysis of the results (Results and Discussion)

Vickermicro hardness test

From Vicker Micro Hardness Test results (Fig. 5) it is shown that the hardness value of the weld nugget is lower than the base materials because of the stirring effect of the pin with huge mechanical deformation, grain refinement [44] happens around the pin. So ductility improves of that zone due to which hardness decreases. It is seen that there is not much difference in the hardness values of weld zone and heat affected zone (HAZ).

Tensile Test

The tensile testing results shows (Fig. 6) that toughness for the welding sample welded by tapered cylindrical tool pin is more compared to straight cylindrical tool pin because area under the curve is greater here (Fig. 6), and the displacement before failure is more for tapered cylindrical tool compared to straight cylindrical tool of same base diameter (Table 5).

Table 5 Process parameters upon which the experimental and numerical results are comprised
Full size table

Verification of Numerical Analysis with Experimental Result

From the numerical and the experimental study it can be seen the both the result has come more or less same and the error between them is very less.

Figures 7, 8, 9, 10, 11 and 12 are showing mesh generation model and different numerical results based on FE model. It is understood from Figs. 10 and 11 that the heat generation is maximum at the weld nugget zone due to huge mechanical deformation arising from shear deformation as a result of traverse speed and frictional heating and decreases gradually with increase of the shoulder radius of the tool.

Fig. 7
figure 7

FE model and meshing view

Full size image
Fig. 8
figure 8

Isothermal pattern for a particular time step during welding for 6 mm tapered pin diameter

Full size image
Fig. 9
figure 9

Isothermal pattern for a particular time step during welding for 5 mm pin diameter

Full size image
Fig. 10
figure 10

Peak temperature distributions along plate breadth perpendicular to the weld line using conical pin

Full size image
Fig. 11
figure 11

Comparison of peak temperature distribution with distance perpendicular to the weld line for varying welding parameters

Full size image
Fig. 12
figure 12

Temperature distribution from center of weld line to away from weld line for 6 mm cylindrical probe diameter and flat circular shoulder frictional surface where the tool rotational and transverse speed were 1400 rpm and 112 mm/min respectively

Full size image

By numerical model the isotherm pattern can be seen from Figs. 8 and 9 at the time of welding. Like if one considers the peak temperature error calculation, then it can be seen from Fig. 13 that in experimental result it is approximately 494 °C and numerically it is 480 °C which leads to the maximum error of 2.91 %.

Fig. 13
figure 13

Numerical and experimental comparison of temperature distribution at 5 mm away from center of weld line

Full size image

Conclusion

In this work 6 mm thick commercial grade aluminum plates have been welded by using 4 different types of tool(two straight cylindrical and two tapered cylindrical). From the above designed FSW tool geometry and the above process parameters, a very good quality weld has been achieved by which the following conclusions can be made:

  1. a)

    The hardness value is more in case of straight cylindrical tool pin than tapered cylindrical tool pin for the same process parameters.

  2. b)

    The average hardness value of the weld nugget zone reduces near about 13.28 % compared to base metal or parent metal zone.

  3. c)

    It is observed that the tool with tapered pin profile generate less temperature than the tool with straight cylindrical pin profile.

  4. d)

    For a particular welding parameter the increase in peak temperature for increment of 1 mm diameter in case of cylindrical pin was around 10 % only. Since increase in diameter does not significantly contribute to heat generation. It is preferable to keep the diameter as low as possible to avoid occurrence of wormhole defect.

  5. e)

    Toughness was higher for welded sample done by tapered cylindrical tools with same process parameter because area under the curve is greater here as seen in tensile test.

  6. f)

    The hardness values of the weld zone and HAZ are lower than the base material which indicates the improved ductility of the weld.