Anda di halaman 1dari 11

A Study on an Interface Tracking Algorithm in Friction Stir Welding on

Computational Fluid Dynamics Analysis

Abstract
Friction stir welding (FSW) was studied using commercial tool, FLOW-3D. Computational
fluid dynamics (CFD) analysis is a computational fluid dynamics (CFD) analysis. Cylindrical
tool shape was used, and the interface between the tool surface and the workpiece were
tracked by its geometrical relations in order to consider the frictional heat in FSW. After
tracking the interface cells, the average area concept was used to calculate the frictional
heat, which is related to interface area. Also three-dimensional heat source and visco-plastic
flow were modeled. The frictional heat generation rate was calculated numerically from the
analytical solution. The numerical solution was well matched with the analytical solution,
and the maximum percentage of error was around 3%.
Key words : Friction stir welding , Computational fluid dynamics , Frictional
heat , Interface tracking
Go to:

1. Introduction

Friction Stir Welding (FSW) is then developed in 1991. The Welding Institute (TWI) 1 ) were
welding defects is less the quality of welds, because the welding by conventional methods
himdeuldeon can be applied to materials such as aluminum foil and solid welding also an
excellent new welding technology is being conducted studies 2 ) .
In FSW, weldability changes according to various conditions such as tool shape and rotation
speed. Therefore, analytical research must be preceded in order to find more optimal
experimental conditions. Among them, Computational Fluid Dynamics (CFD) analysis is
able to describe the flow field which is difficult to grasp experimentally and many
researchers choose it as an interpretation method that reflects the physical phenomena of
actual FSW. However, since it is difficult to express both the heat source due to the friction
generated during the process and the heat source due to the plastic deformation, the
existing researchers modeled the heat source as follows. Kim et al. 3 ) performed CFD
analysis using commercial code STAR-CCM +. The temperature history obtained from the
analysis was in good agreement with the measured values, but the heat source due to the
friction between the tool and the base material was not considered. Considering only the
amount of heat generated by plastic deformation as a heat source, the ratio of plastic
deformation to heat transfer was calculated as 1. Chen et al. [ 4 ] analyzed the commercial
code, ANSYS Fluent, and formulated the relationship of the total calorific value according to
the rotation speed when the moving speed of the tool is constant. In his paper, only the
caloric value due to plastic deformation was considered without considering the heat source
due to surface friction, and the ratio was calculated as 0.8. Ji et al. 5 )Investigated the effect
of tool geometry on the flow field using ANSYS Fluent. To account for the heat source, the
surface temperature of the tool was set at 80% of the melting point known as the
conventional concept. In this way, the conventional CFD approach solves the problem by
using only the heat source by the plastic deformation without considering the heat source by
the surface friction in many cases, and it is difficult to describe the rotation and movement
of the tool even in the article considering the friction heat source. Were used.
Therefore, in this study, we proposed an algorithm to track the surface in real time in order
to consider the rotation and movement of the tool simultaneously and apply the friction
heat source accordingly. The analytical results were verified by comparing with the
theoretical frictional heat source size.
Go to:

2. Surface tracking and surface area calculation

2.1 Surface Tracking


Since the tool moves at the same time as it rotates, this study approaches the tool surface
as shown in Table 1 . In Table 1 , r p and r s are the radius of the shoulder and pin
respectively, Δcell is the size of the cell and x is the distance from the center of the tool to the
center of any cell. The surface cell tracking principle of each region is as follows. Since the
cylindrical tool is symmetrical, the distance from the center of the tool to an arbitrary
surface is always known. For example, the distance from the tool center to the pin side is
always equal to the radius of the pin, and the side of the shoulder is always equal to the
radius of the shoulder. Therefore, when the distance from the center of the tool to the
arbitrary cell is calculated every time and the distance is calculated within the area shown
in Table 1 , the corresponding cells are recognized as the surface cells. The reason for adding
or subtracting half of the cell size at each radius r p and r sis to include the case where the
surface of the tool moves in a three-dimensional cell because the cell is three-dimensional
and the surface of the tool is two-dimensional. The divided shoulder, pin side and pin
bottom area for calculation are shown in Fig. 1 .
Fig. One
Area partition

Table 1
Interface tracking algorithm

Part Algorithm
rp-Δcell2<x≤rs+Δcell2
Shoulder

rp-Δcell2<x≤rp+Δcell2
Pin side

x≤rp+Δcell2
Pin type

2.2 Surface area calculation


To provide a friction heat source using surface cells obtained by surface tracing, the surface
area at which the tool and the substrate contact each other should be calculated in all the
surface cells. However, it is difficult to calculate the exact surface area considering both the
rotation and the movement of the tool. To do this, we first calculate the area where the tool
and the base material meet, and divide the area by the number of surface cells obtained
from the surface tracking method described earlier. We can then calculate the average
contact area per surface cell, assuming that this average area works in all surface cells. The
data obtained by this method was applied to the heat source modeling described in Section
3.4 and the frictional heat source was analytically obtained.

Go to:

3. Modeling

3.1 Home
In order to compare the heat source by the surface friction with the theoretical value, the
analysis proceeded from the beginning with the tool inserted into the base material. During
the analysis, the tool was assumed to be a rigid body without any deformation, and the
rotational speed and the moving speed of the tool were kept constant. The base material was
assumed to be a viscous fluid, and the condition of welding in the center part of one base
material was used.

3.2 Tool shape modeling


Fig. 2 , a relatively simple form of the tool has been considered, since theoretically the
amount of heat from friction can be obtained. In addition, the proposed surface tracking
algorithm is applied to the simplest type of tool, and after the verification, it is intended to
develop a model that can be applied to a more complicated tool shape.

Fig. 2
Tool shape

3.3 Modeling of viscosity coefficient


In FSW the viscosity is represented as a function of temperature and strain rate, Sheppard
et al., First to calculate the viscosity . 6 ) is calculated for the proposed flow stress (flow
stress) of the following:
(One)
σe=Onealphasinh-One[(ZA)Onen]

Where A , α, and n are material constants and are measured through experiments. In this
study, AA5052-O was used as the base material and the material constants of the base
material are shown in Table 2 . Z is the Zener-Hollomon parameter and is expressed as:
Table 2
Data used in the simulation

Property / Weld parameter Value


Workpiece length 10mm
Workpiece width 8mm
workpiece thickness 4mm
Welding speed 1m / min
Property / Weld parameter Value
Rotational speed 1000rpm
A 12 ) exp (24.47)
α 12 ) 0.016 MPa -1
n 12 ) 5.24
Q 12 ) 155167J

(2)
Z=∈·exp(QRT)

Where Q is the activation energy, R is the gas constant, T is the temperature, and ∈ is the
effective strain rate, expressed as:
(3)
∈·=(23∈ij∈ij)One2

Where ε ij is defined by the strain rate tensor as follows.


(4)
∈ij=One2(∂ui∂xj+∂uj∂xi)

The viscosity coefficient is expressed as Perzyna's visco-plasticity model 7 ) , expressed as a


function of the flow stress and the actual strain rate .
(5)
μ=σee∈·

As the expression above, the viscosity coefficient is dominated by the temperature and
strain rate. Therefore, it can be seen that proper heat source modeling affects not only
temperature distribution, but also viscosity and flow field.

3.4 Heat Source Modeling


In the FSW process, heat source by friction and plastic deformation occurs. The heat
source S fric due to friction is expressed as follows.
(6)
Sfric=mk(ωr-Usinθ)ArV
Where m is the friction factor and maximum shear stress at yielding. This value is calculated
according to the Von-Mises theory.
k=σyield3

Ω is the angular velocity of the tool, r is the distance from the tool center, θ is the angle
between the direction of movement and rotation of the tool, A r is the average contact area
obtained at 2.2 and V is the volume of the cell containing A r . In this study, 0.4 was used as
the friction factor. 8 ) To obtain the theoretical solution, we assumed that sigma yield is a
temperature-independent constant.
The heat source S pd by plastic deformation is expressed as follows [ 9 , 10 ] .
(7)
Spd=fmσe∈·

Where f m was 0.04 applies to the ratio that the heat transmission 11 ) .
In addition to the surface tracking algorithm described above, the analysis was performed
using the user subroutine of the Flow-3D analysis program to perform the analysis using the
heat source and viscosity relations.

3.5 Boundary Condition Modeling


Due to the nature of the FSW, the tool will hold the backing plate on the bottom of the base
material as the tool presses the base material firmly as the weld proceeds. Therefore, the
bottom surface of the base material only and the heat transfer by conduction occurs between
the base material for the backing plate, and applying the convection takes place and home,
and the boundary condition equivalent because of the difficulty to establish a conduction
coefficient 13 ) . In other areas, convection and radiation conditions are applied because the
base material and the atmosphere are in contact.

Go to:

4. Results

4.1 Theoretical Theory


To verify the model, the size of the frictional heat source was determined by the analytical
model presented in Schmidt et al. [ 14 ] . In this paper, based on the geometric relationship of
the cylindrical tool like that used in this study, the micro-frictional force acting on each
surface (shoulder and pin) was obtained while integrating the whole area. The method of
expressing the frictional force is similar to the expression used in the modeling part of the
paper. Table 3 shows the result of calculating the friction heat source by dividing the tool by
the shoulder, the pin side, and the pin bottom by this method . The largest heat was
generated in the shoulder, and the size of the heat source was large in the order of the pin
side and the pin bottom. The reason for this is that it has the largest contact area on the
shoulder. Also, because the radius of the shoulder is larger than the pin, the influence of the
proportional velocity along the radius is considered.
Table 3
Analytical solution

Part Heat generation rate


Shoulder 22.11 [W]
Pin side 16.11 [W]
Pin type 3.16 [W]

4.2 Comparison with theoretical solutions


The analysis results are shown in Fig. 3 . Since the frictional heat source is proportional to
the rotational speed of the tool, the larger the heat source distance from the center of the
tool, the larger the size of the heat source. Also, it can be seen that the amount of heat
generated in the center of the tool is very small because the rotational speed is close to
zero. In order to select the analysis time, we considered the time to move by the size of the
tool in the initial position. In other words, based on the diameter of the shoulder, the
analysis was carried out for a time of 4 mm, and the analysis time at that time was 0.24 s.

Fig. 3
Frictional heat generation rate

Because the tool moves at the same time as it rotates, the number of surface cells per cycle
varies. Therefore, after obtaining a sufficient number of data, the values were moved for 4
mm for analysis. By dividing the total analysis time of 0.24s into 1 / 5000s, a total of 1200
data are obtained. The data are averaged and the result is given as a numerical solution. The
results are shown in Table 4. The analysis results show that the size of the heat source
appears in the order of shoulder, pin side, and pin bottom. Table 5 shows the comparison
between the analytical results and the theoretical values . An error of about 10% occurred at
the bottom of the pin because the absolute amount of heat generated at the bottom of the
pin is small. On the other hand, errors of 0.7 ~ 2.7% were observed on the shoulder and pin
side, and about 1% error was found in comparison with total calories.
Table 4
Simulation results

Part Heat generation rate


Shoulder 22.71 [W]
Pin side 15.99 [W]
Pin type 3.47 [W]

Table 5
Percentage error between the analytical solutions and simulation results using average
values

Part Error
Shoulder 2.71 [%]
Pin side 0.74 [%]
Pin type 9.81 [%]
Total 1.01 [%]

During the analysis, the number of surface cells changes each time. That is, the heat
generated from the surface also changes every time. In this study, the average value of 1200
data is determined by numerical solution. In order to check the validity of the mean value,
the results of the minimum and maximum values of 1200 data are compared with the
theoretical values. As shown in Table 6, when the minimum value is used rather than the
average value, it can be confirmed that it is close to the theoretical value. However, the
difference is about 1%, and even when using the maximum value accompanied by the largest
error, the error is less than 3%. Therefore, it can be concluded that the maximum error is
less than 3% regardless of any variables. Therefore, it can be concluded that the approach
using the mean value used in this study is valid.
Table 6
Percentage error between the analytical solutions and the simulation results

Part Error
Shoulder (Minimum) 1.94 [%]
Shoulder (Maximum) 3.21 [%]
Pin side (Minimum) 2.48 [%]
Pin side (Maximum) 0.74 [%]
Pin type (Minimum) 5.38 [%]
Pin tip (Maximum) 11.71 [%]
Total (Minimum) 0.48 [%]
Total (Maximum) 2.89 [%]

Go to:

5. Conclusion

Unlike the analysis using computational fluid dynamics of the existing friction stir welding,
a new algorithm is proposed to express the rotation and movement of the tool and
simultaneously apply the friction heat source. By tracing the surface cells and applying the
average surface area per cycle, the frictional heat source could be applied at any time and
position. The analytical results showed error of about 3% with the theoretical value, and
about 1% when using the mean value.
Therefore, it was confirmed that it was a reasonable assumption to apply the concept of the
mean area to find the surface cell according to the shape of the tool and to find its
area. Using this algorithm, it is expected that it will be able to describe frictional heat source
more precisely in the future for arbitrary tool shape.
Go to:

References

1. Thomas WM, Nicholas ED, Needham JD, Murch MG, Templesmith P, Dawes CGB. 9,125,
978.8, Dec. 1991, U.S. Pat. 5,460, 317, Oct. (1995)
2. Chang IS, Cho YJ, Park HS, So DY Importance of Fundamental Manufacturing
Technology in the Automotive Industry and the State of the Art Welding and Joining
Technology. J. Welding and Joining . 34 (1) (2016), 21-25

3. Numerical simulation of friction stir welding process for AA5083-H18 sheets, Kim, DG,
Badarinarayan H, Kim JH, Kim CM, Okamoto K, Wagoner RH, Chung KS. European
Journal of Mechanics-A / Solids . 29 (2) (2013), 204-215

4. Chen GQ, Shi QY, Li YJ, Sun YJ, Dai QL, Jia JY, Zhu YC, Wu JJ Computational fluid
dynamics studies on heat generation during friction stir welding of aluminum
alloy. Computational Materials Science . 79 (2013), 540-546

5. Ji SD, Shi QY, Zhang LG, Zou AL, Gao SS, Zan LV Numerical simulation of material flow
behavior of friction stir welding influenced by rotational tool geometry. Computational
Materials Science . 63 (2012), 218-226

6. Sheppard T, Wright DS Determination of flow stress: Part 1 constitutive equation for


aluminum alloys at elevated temperatures. Metals Technology . 6 (1) (1979), 215-223

7. Zienkiewicz OC, Cormeau IC Visco-plasticity-plasticity and creep in elastic solids-a


unified numerical solution approach. International Journal of Numerical Methods in
Engineering . 8 (4) (1974), 821-845

8. Buffa G, Hua J, Shivpuri R, Fratini L. A continuum based fem model for friction stir
welding-model development. Science and Engineering Materials: amperes . 419-1 (2005),
389-396

9. Ayer R, Jin HW, Mueller RR, Ling S, Ford S. Interface structure in a Fe-Ni friction stir
welded joint. Scripta Materialia . 53 (12) (2005), 1383-1387

10. Carslaw HS, Jaeger JC Conduction of heat in solids . Oxford: 2nd Edition. Clarendon
Press; (1959)
11. Nandan R, Roy GG, Lienert TJ, and Deb-Roy T. Three-dimensional heat and material
flow during friction stir welding of mild steel. Acta Materialia . 55 (3) (2007)

12. Cho JH, Boyce DE, Dawson PR Mod-eling strain hardening and texture evolution of
friction stir welding of stainless steel. Science and Engineering Materials, amperes . 398 (1)
(2005), 146-163

13. Sheppard T, Jackson A. Constitutive equations for use in prediction of flow stress during
extrusion of aluminum alloys. Materials science and Technology . 13 (3) (1997), 203-209

14. Schmidt H, Hattel J, Wert J. An analytical model for the friction stir welding. Modeling
and Simulation in Materials Science and Engineering . 12 (1) (2003), 143-157

Anda mungkin juga menyukai