Analysis of contact conditions and its influence on strain rate and temperature in friction stir welding

Highlights

• Evolution of the contact conditions with welding parameters.

• Analysis of strain rate sensitivity to process parameters.

• Prediction of grain size evolution with process parameters.

• Influence of contact conditions on heat generation.

• Analysis of the mechanisms governing material flow.

Abstract

In friction stir welding (FSW), the real contact conditions between the tool and the workpiece and the range of strain rates experienced remain quite unclear. In this work, a coupled 3D thermo-mechanical numerical model was used to simulate the FSW process. A Parametric finite element analysis of the evolution of the contact conditions, strain rates and temperatures with the processing parameters, tool dimensions and base material plastic properties was conducted. The numerical model was able to capture the evolution of the mixed slipping/sticking contact conditions with the welding time and welding parameters. The temperature and strain rate gradients obtained in the numerical simulations were validated with experimental data, by calculating the grain size distribution, in the stirred volume, using the Zener-Hollomon parameter. Full sticking, full slipping and mixed slipping-sticking contact domains were identified in a process parameters chart. It was found that, meanwhile the temperature and the sticking fraction evolve in the same way with the processing parameters, the strain rate is mainly determined by the tool rotation speed, varying from an average of 68 to 324 s⁻¹, when the tool rotation speed is increased from 300 to 1200 rpm. The contact conditions and the base material plastic properties were also found to mutually influence the material flow. In full sticking contact, high strength materials, with high strain rate sensitivity, may display a similar flow pattern to that of low strength materials. However, coarser and more uniform grain structures may result from the welding of high strength materials, as a result of the narrower range of strain rates experienced during welding combined with high heat input.

Introduction

During Friction Stir Welding (FSW), the tool rotation and translation movements promote not only the heating by friction of the materials to be joined, but also its plastic deformation under complex loading conditions and variable strain rates. The complex loading conditions, at high temperatures, are responsible for the material flow and for the microstructural phenomena taking place during welding. So the understanding of the mechanisms that govern the plastic deformation during welding, which are conditioned by the contact conditions at the tool/workpiece interface, temperature and strain rates inside the stirred volume, are very important to predict the final microstructure of the welded materials as well as the possibility of defect formation.

Several works attempted to analyse and measure the plastic deformation and the strain rates during FSW, by using different techniques, such as microstructural analysis, tracing materials, analytical models and numerical simulation. Table 1 summarises the strain rate values reported in the literature, determined using the above described techniques, for different base materials and process conditions. Frigaard et al. [1], Gerlich et al. [2] and Gerlich et al. [3] measured the grain size in the stirring zone to compute the strain rate values by using the Zener–Hollomon parameter, in FSW and Friction Stir Spot Welding (FSSW) of Aluminium alloys. Frigaard et al. [1] calculated strain rate values, between 1 and 20 s⁻¹ in FSW of AA6082 and AA7018. According to the authors, these results indicated the occurrence of slipping contact conditions between the tool and the workpiece, since the calculated strain rate values were very low when compared to the angular velocity of the tool. On the other hand, Gerlich et al. [2] and Gerlich et al. [3] in FSSW of AA7075 and AA2024, respectively, reported a decrease in the strain rates, from 650 to 20 s⁻¹ and 1600 to 0.6 s⁻¹, by increasing the rotation speed from 1000 to 3000 rpm and 750 to 3000 rpm, respectively. They attributed these results to the local melting of second phase particles. Masaki et al. [4] determined the strain rate values during FSW of AA1050, by comparing the grain size in the welded zone, with the grain size of specimens loaded in plane-strain compression, under various temperatures and strain rates. Using this technique, the authors found that by increasing the rotation speed from 600 to 1200 rpm the strain rates varied between 1.7 to 2.7 s⁻¹.

Chen and Cui [5] and Liu et al. [6] determined the strain rates in FSW of A356 alloy and C1100P copper, respectively, by measuring the distortion of tracer materials in the post-weld microstructure. Chen and Cui [5] calculated strain rates between 3.5 to 85 s⁻¹ in the leading side of the tool, while Liu et al. [6] calculated an average strain rate of 20.8 s⁻¹ in the band formation zone. The use of tracers has also been used to determine the strain rates and the material flow velocity during FSW by Morisada et al. [7], Morisada et al. [8] and Kumar et al. [9]. Morisada et al. [7] calculated a maximum strain rate value of almost 15 s⁻¹, for the FSW of A1050 at 1000 rpm. Also, in the FSW of A1050 Morisada et al. [8], observed that the tracing particles rotated around the tool several times, when the rotation speed was higher than 400 rpm, although the angular velocity of the tracer was always lower than the angular velocity of the tool. For rotation speeds lower than 300 rpm the tracing particles stopped rotating around the tool and defects were observed in the weld. Kumar et al. [9] analysed the influence of the rotation and traverse speeds on the strain rate, in the FSW of a viscoplastic fluid. According to the authors, the tracing particles also rotated several times around the tool pin, up to a maximum velocity of 60% of the pin angular speed. The tool rotational speed was found to be the main factor governing the strain rates. Increasing the rotation speed from 75 to 425 rpm lead to an increase in the strain rates between 8 and 44 s⁻¹.

Chang et al. [10] and Long et al. [11] proposed analytical models to estimate the strain rates during welding. Chang et al. [10] proposed that the strain rates are proportional to the size of the dynamically recrystallised zone and to a fraction of the tool rotational speed, due to the sticking/slipping contact condition at the tool/workpiece interface. Using the previous model, Chang et al. [10] calculated an increase in the strain rates between 5 and 50 s⁻¹ by increasing the rotation speed from 180 to 1800 rpm. The Long et al. [11] model estimated the strain rates by considering the distance that the tool advance in one rotation, as the initial length of the undeformed material. Then, during the tool rotation, this portion of material is stretched in the front side of the pin and is finally compressed in the trailing side of the pin, where it is deposited. Considering the previous model, Long et al. [11] calculated an increase in the average strain rates from 20 to 350 s⁻¹ by increasing the rotation speed from 544 to 844 rpm, respectively.

Due to the difficulty in calculating the strain rates experimentally during welding, numerical simulation has been used as a tool to determine the strain rates experienced during FSW. Nandan et al. [12], Nandan et al. [13] and Nandan et al. [14] used a three-dimensional viscoplastic model to simulate the FSW of 304 stainless steel, AA6061 aluminium and AISI 1018 steel, respectively. The authors determined maximum strain rate values of 130 s⁻¹, 150 s⁻¹ and 40 s⁻¹, for rotation speeds equal to 344, 300 and 450 rpm, respectively. Du et al. [15] used numerical simulation to model the FSW of AA2017, AA5083 and AA6082 aluminium alloys and computed strain rate values between 23.16 to 434.25 s⁻¹ by varying the rotation speed from 100 to 1100 rpm. Mukherjee and Ghosh [16] used two-dimensional finite-element simulation using ABAQUS, to model the FSW of AA5083 aluminium alloy. The authors concluded that a 0.1 ratio between the base material velocity matrix and the tool velocity, best characterised the material flow. For these conditions, a maximum strain rate of 87 s⁻¹ was determined. Ammouri et al. [17] used a 3D thermo-mechanically coupled FE model to simulate the FSW of AZ31B, under different rotation and traverse speeds. The authors observed that the strain rates increased with the rotation and traverse speeds. Although, the rotation speed presented higher influence on the strain rate values than the traverse speed. Sharghi and Farzadi [18] used a three-dimensional model based on the computational fluid dynamics to simulate dissimilar welding of AA6061/Al-Mg₂Si aluminium alloys. The authors computed a maximum strain rate of 975 s⁻¹ near the top surface of the workpiece at the outer edge of the tool shoulder.

The contact conditions at the tool/workpiece interface are also critical to understand the welding mechanisms occurring during FSW, although they are difficult to study experimentally. In general, the contact conditions are considered to be fully sticking [19], [20], [21], [22], [23], [24], [25], [26], [27] or fully slipping [19,24,[28], [29], [30]]. However, this assumption may be restrictive in order to simulate the welding process accurately. Some works have also considered the partial slipping/sticking phenomena during the welding process by prescribing imposed velocity profiles at the tool/workpiece interface [19,24,[31], [32], [33]].

Considering all the works analysed, it is possible to conclude that the calculated strain rate values widely vary, in accordance with the different measurement techniques, process parameters, contact conditions and base materials used. In current work, a coupled three-dimensional thermo-mechanical model was used to simulate the evolution of the mixed slipping/sticking contact conditions and to compute the strain rates and temperatures during FSW of different base materials under a wide range of parametrically varied welding conditions. The range of temperatures and strain rate obtained in the numerical simulations were validated with experimental results and extrapolated to predict the evolution of the weld microstructure for the different welding conditions tested.