Heat flow prediction in friction stir welded aluminium alloy 1100

1. International INTERNATIONAL Journal of Mechanical JOURNAL Engineering OF and MECHANICAL Technology (IJMET), ISSN ENGINEERING
0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
AND TECHNOLOGY (IJMET)
ISSN 0976 – 6340 (Print)
ISSN 0976 – 6359 (Online)
Volume 5, Issue 9, September (2014), pp. 336-346
© IAEME:

2. Journal Impact Factor (2014): 7.5377 (Calculated by GISI)
336
IJMET
© I A E M E
HEAT FLOW PREDICTION IN FRICTION STIR WELDED ALUMINIUM
ALLOY 1100
S. Deivanai1, Dr. Reeta Wattal2, Mrs. Sushila Rani2, Ms. Surabhi lata3
1Lecturer Mech.Engg .Deptt, Pusa Polytechnic, New Delhi, India
2 Professor, Mech. Engg. Deptt, Delhi Technological University, New Delhi, India
2Asst.Professor, Mech.Engg. Deptt. Delhi Technological University, New Delhi, India
3Asst.Professor, Mech. Engg. Deptt. Maharaja Agresan Institute of Engg Technology New Delhi,
India
ABSTRACT
Aluminium alloy 1100 is used in manufacturing of Aircraft electrical conduits, Rivets, hose
reel, sewage pumps, pressure regulator, level indicator ,control valves etc,. Friction stir welding
(FSW) is an innovative solid state joining technique, employed for joining aluminium, magnesium,
zinc and copper alloys. The FSW process parameters play a major role in deciding the weld quality.
Effect of the tool rotational speed, weld speed, shoulder pin diameter, on the temperature at weld
nugget and temperature at HAZ were found out. The mathematical model developed. The model
checked for adequacy using ANOVA technique. Main and interaction effects of the process variables
are presented in graphical form using MINITAB 17.0. The developed model used for prediction of
Heat flow. Comparison of the performance of the experimental values, the mathematical Model and
ANN model was done. ANN using feed forward algorithm used for modelling of the FSW process
which provided satisfactory outputs. Heat flow calculations made and micro structural analysis done.
Keywords: Full Factorial Design, Design of Experiment, Mathematical Model, ANN, Heat Flow,
Microstructure.
1. INTRODUCTION
Friction stir welding provides Good mechanical properties than fusion welding.
Consumables, filler rod, gas shield is not required. The process is easily automated on simple milling
machines. Hence lower setup costs and less training is required. Good weld appearance and minimal
thickness under/over-matching, thus reducing the need for expensive machining after welding. It is
called as Green process because Low energy input and Absence of toxic fumes, No spatter of molten
material, makes FSW friendly to our environment. Tool Rotational Speed, Weld Speed, Tool

3. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
shoulder diameter, Tool pin diameter, Axial force, Tool tilt angle and Tool design are important
process variables in Friction stir welding.
The research work was planned to carry out in the following steps:
337
2. PLAN OF INVESTIGATION
1. Identification of important process control variables.
2. Deciding the working range of the process control variables, viz. Tool rotational speed, weld
speed, tool pin diameter.
3. Developing the design matrix.
4. Conducting the experiments as per the design matrix.
5. Recording the responses viz. Temperature at weld nugget and Temperature at heat affected
zone.
6. Development of mathematical model.
7. Checking the adequacy of the model.
8. Development of an ANN architecture to model and to predict the mechanical properties.
9. Comparison of performances of mathematical model, experimental value, ANN model.
10. Metallurgical analysis to study the micro structure of the welded material with the process
variables.
11. Plotting of graphs and drawing conclusions.
12. Discussion of the results.
3. IDENTIFICATION OF IMPORTANT PROCESS CONTROL VARIABLES
Based on literature review the important process parameters of friction stir welding are
identified as tool rotational speed. Weld speed, axial force, tool shoulder diameter, tool pin diameter,
tool pin shape etc.
4. DECIDING THE WORKING RANGE OF THE PROCESS CONTROL VARIABLES
Trial runs are conducted to find the upper and lower limits of the process parameters by
varying one of the parameter and keeping the rest of parameters at constant values. Feasible limits of
the parameters are chosen in such a way that the joint should be free from visible defects. Upper
limit of parameter is coded as HIGH and lower limit as LOW. The selected process parameters and
their upper and lower limits together with notations and units are given in
Table.1.
Table.1: Process Control Parameters And Their Limits
Sl.No parameters Units Notation -1 0 +1
1 Tool Rotational speed RPM N 460 690 1130
2 Weld speed Mm/min V 24 40 65
3 Tool pin diameter mm D 6 7 8

4. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
338
5. DEVELOPING THE DESIGN MATRIX
When three factors N, V and D, each at two levels, are of interest, the design is called a 23
factorial design and the eight treatment combinations can be as follows.
Table .2: Design matrix
Weld no. Trial no. Input parameters labels
N V D
1 4 -1 -1 -1 (1)
2 7 +1 -1 -1 a
3 1 -1 +1 -1 b
4 6 +1 +1 -1 ab
5 2 -1 -1 +1 c
6 8 +1 -1 +1 ac
7 5 -1 +1 +1 bc
8 3 +1 +1 +1 abc
There are seven degrees of freedom between the eight treatment combinations in the 23
design. Three degrees of freedom are associated with the main effects of N, V, and D. Four degrees
of freedom are associated with interactions; one each with NV, ND, and VD and one with NVD.
6. CONDUCTING THE EXPERIMENTS AS PER THE DESIGN MATRIX
The experiments were conducted at the central workshop of Delhi Technological University.
A conventional vertical milling machine was used for FSW of aluminium alloy 1100. The feed speed
and tool rotation (rpm) value of the milling machine play major roles in ensuring sound FSW joints.
During the investigation, basic limitations of the milling machine also became apparent, for example
rpm and feed speed setting were as per the machining operations. Furthermore, the load value is also
unknown. The experience gained with the milling machine for FSW of aluminium is that it is the tool
that is capable of demonstrating the FSW process for aluminium at a low feed speed.
6.1 The Base Metal
For carrying out the research work, test specimens were prepared from 5 mm thickness
Aluminium Alloy 1100 plate. Dimension of each plate was 100x50x5 mm. Composition of the base
material is identified by spectro analysis using Hilger Polyvac 2000 Optical Emission Spectrometer.
Optical emission spectrometry involves applying electrical energy in the form of spark generated
between an electrode and a metal sample, whereby the vaporized atoms are brought to a high energy
state within a so-called “discharge plasma”. These excited atoms and ions in the discharge plasma
create a unique emission spectrum specific to each element.
The chemical composition of AA1100 is tabulated as follows
Table.3: chemical composition of AA 1100
Al Cu Mg Si Fe Mn Ni Zn Pb Sn Ti Cr
99.16 % 0.123% 0.036 0.185 0.322 0.023 0.025 0.052 0.023 0.020 0.010 0.005

5. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
6.2 Tool Material Used
EN 8: It is a very popular grade of through-hardening medium carbon steel, which is readily
machinable in any condition. EN8 is suitable for the manufacture of parts such as general-purpose
axles and shafts, gears, bolts and studs.
339
Eight sets of plates were welded as per the design matrix by selecting trials at random and
welding was carried out. Figure.2 shows few samples in welded condition.
Figure.1: welded samples
7. RECORDING THE RESPONSE
The Temperature at work piece during Friction stir welding is measured using Laboratory
mercury thermometer at weld centre and at heat affected zone.
Table.4: Temperature and heat flow
Run
no.
Tool
speed
(RPM)
Weld
speed
(mm/min)
Toolpin
Diameter
(mm)
Temp
At
nugget
(0C)
Heat flow [Q]
in w/cm/0c
In weld nugget
Temp
At
HAZ
(0C)
Heat flow [Q]
in w/cm/0c
In HAZ of
weld
1 460 24 6 55 297.48 43 178.48
2 1130 24 6 70 446.22 55 297.48
3 460 65 6 85 594.96 61 356.97
4 1130 65 6 68 426.39 60 347.06
5 460 24 8 50 247.90 42 168.57
6 1130 24 8 55 297.48 40 148.74
7 460 65 8 65 396.64 58 327.23
8 1130 65 8 108 823.03 92 664.38
7.1 Heat flow calculations
The majority of the heat generated from the friction, i.e., about 95%, is transferred into the
work piece and only 5% flows into the tool. Heat flow during welding, strongly affect phase
transformations during welding and thus the resultant microstructure and properties of the weld. It is
also responsible for weld residual stresses and distortion. The analytical solution derived by

6. International Journal of Mechanical Engineering and Tec
ISSN 0976 – 6359(Online), Volume 5, Issue
hnology – 6340(Print),
Rosenthal.D for three-dimensional heat flow in a semi infinite work piece during welding was used
for calculation of heat input.
0 C; - initial temperature in 0 C ; k- thermal conductivity of
Where T – measured temperature in
aluminium alloy in J/ms K = 229 for aluminium ;R
+ y^2 + z^2)^1/2 ; x – desired distance from
- workpiece thermal diffusivity = 8.5 X 10
8. DEVELOPMENT OF MATHEMATICAL MODEL
The Response function can be expressed as: Y = ƒ (N,
Where, Y = response; N = tool rotational speed;
A linear regression model with three predictor variables can be expressed with the following
equation:
Y = b0+b1N+b2 V+b3D+b12N
Where b0 is constant and b1, b2, b3, b
b0, the Y-intercept, can be interpreted as the value to predict for Y if N=
Since N is a continuous variable,
each one-unit difference in N, if other independent variables remains constant. This means that if N
is differed by one unit, and V and D did not differ, Y will differ by
8.1 Evaluation of the co-efficient of model
The values of the co efficient of the response function were calculated using regression
analysis. The calculations were carried out using MINITAB17
Table.5: Evaluation
Sl.no
Co-efficient
1 b0
2 b1
3 b2
4 b3
5 b12
6 b13
7 b23
8 b123
Technology (IJMET), ISSN 0976
9, September (2014), pp. 336-346 © IAEME
340
- radial distance from the origin, namely, (x^2
origin ; Q – heat input (W)V – travel speed (mm/s);
-5 m2 / s for aluminium
V, D)
V = weld speed; D= tool pin diameter
NV+b13ND+b23 VD+b123NVD
, b12, b13, b23, b123 are co- efficients of model
V=D = 0
b1 represents the difference in the predicted value of Y for
b1 units, on average.
and values listed in Table 5
of the co-efficient of model
Temperature(0C)
At weld At HAZ
0 -102.1 -69.43
1 + 0.2786 + 0.2204
2 + 5.882 + 3.265
3 + 19.39 + 14.74
12 - 0.008810 - 0.005825
13 - 0.03804 - 0.03185
23 - 0.7690 - 0.4347
123 + 0.001274 + 0.000892
5.

7. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
Temperature at weld nugget (TN) = b0+b1N+b2V+b3D+b12NV+b13ND+b23VD+b123NVD
TN = -102.1 + 0.2786 N + 5.882V + 19.39 D – 0.008810 NV – 0.03804 ND – 0.7690 VD
+ 0.001274 NVD
Temperature at HAZ (T HAZ) = b0+b1N+b2V+b3D+b12NV+b13ND+b23VD+b123NVD
THAZ = -69.43 + 0.2204 N + 3.265 V + 14.74 D – 0.005825 NV – 0.03185 ND – 0.4347 VD
+ 0.000892 NVD
9. CHECKING ADEQUACY OF THE MODEL
341
8.2 Regression Equations
The analysis of variance (ANOVA) was used to check the adequacy of the developed models.
As per this technique:
a. The F-ratio of the developed model is calculated and is compared with the standard tabulated
value of F- ratio for a specific level of confidence.
b. If calculated value of F- ratio does not exceed the tabulated value, then with the
corresponding confidence probability the model may be considered adequate. For analysis ,
a confidence interval of 95% was taken
Table.6: ANOVA or Temperature at Weld Nugget
source Sum of
squares
DF Mean square F
value
P value
model 2546 7 363.7143 5.980913 0.04439346
N 264.5 1 264.5 4.349435 0.28463858
W 1152 1 1152 18.94347 0.14377340
D 0 1 0 0 0
NW 4.5 1 4.5 0.073997 0.83092317
ND 312.5 1 312.5 5.138726 0.26448893
WD 200 1 200 3.288798 0.32081320
NWD 612.5 1 612.5 10.07194 0.19432760
Pure
error
486.5 8 60.8125
Total 3032.5

8. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
342
Table.7: ANOVA for Temp at HAZ
10. PREDICTION OF RESPONSES BY REGRESSION MODEL
Table.8: Predicted values
source Sum of
squares
DF Mean
square
F
value
P value
model 1961.875 7 280.2679 1.430623 0.44330748
N 231.125 1 231.125 1.179774 0.47371849
W 1035.125 1 1035.125 5.283779 0.26123236
D 21.125 1 21.125 0.1078322 0.79801249
NW 66.125 1 66.125 0.337534 0.66493950
ND 55.125 1 55.125 0.2813847 0.68951297
WD 253.125 1 253.125 1.297245 0.45869807
NWD 300.125 1 300.125 1.531983 0.43261963
Pure
error
1567.25 8 195.9062
Total 3529.125
Weld no Tool
speed
rpm
Weld speed
Mm/min
Pin dia
mm
Temp TN Temp
THAZ
0C
1 460 24 6 51.75 38.00
2 1130 24 6 63.25 48.75
3 460 65 6 75.75 60.75
4 1130 65 6 87.25 71.50
5 460 24 8 51.75 41.25
6 1130 24 8 63.25 52.00
7 460 65 8 75.75 64.00
8 1130 65 8 87.25 74.75

9. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
11. MODELLING USING ARTIFICIAL NEURAL NETWORK
Figure 2: FFNN to predict temperature
A typical feed forward network is shown in fig 2. This network is used in the present research
for the modelling of the process. To model the process 3 input parameters viz , tool rotational
speed , weld speed , tool pin diameter were considered and their effect on temperature developed
during friction stir welding were studied , so for our modelling the input and the output layer of net
work had 3,and 7 neurons respectively. The goal was kept as 0.01.
Figure 3: Performance plot Figure 4: Fit plot
343
Tool
rotational
speed
Weld
speed
Tool pin
diameter
Temp at
nugget
Temp at
HAZ

10. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
12. COMPARISON OF ANN, MATH. MODEL AND EXPERIMENT VALUES
344
Table 9: Comparison of predicted temperature values
Weld
Run
ANN %ERROR MATH.
MODEL
%ERROR EXPERIMENT
VALUE
Temp
TN
7 69.43 +6.82% 75.75 +16.53% 65
8 99.12 -8.22% 87.25 -19.21% 108
Temp
THAZ
0C
7 61.00 +5.17% 64.00 +10.34% 58
8 85.61 -6.94% 74.75 -18.75% 92
13. MICROSTRUCTURAL ANALYSIS OF AA1100
Microscopic Examination using Radical 40X-800X Metallurgical Microscope With 5mp
Camera was carried out before and after welding. ASTM E- 3 standard testing procedure followed.
Etchant 25% Nitric Acid used for surface preparation. The image obtained at Magnification - 250 X
was given below
Figure 5: Microstructure of Base metal Figure 6: Microstructure of weld nugget

11. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
345
14. RESULTS AND DISCUSSIONS
Direct effect of process variables
C1
460 1130
85
80
75
70
65
60
Main Effects Plot for C7
Data Means
C2 C3
24 65 6 8
Me an
C1
460 1130
70
65
60
55
50
45
Main Effects Plot for C8
Data Means
C2 C3
24 65 6 8
M e a n
Figure 7: Effect of TRS, WS D on Figure 8: Effect of TRS, WS D on
Temp at weld Temp at HAZ
Interaction Plot for C8
Data Means
24 65 6 8
80
60
40
80
60
40
C1
C2
C3
C1
460
1130
C2
24
65
Interaction Plot for C7
Data Means
24 65 6 8
90
75
60
90
75
60
C1
C2
C3
C1
460
1130
C2
24
65
Figure 9: Interaction effect at HAZ Figure 10: Interaction effect at weld
15. CONCLUSIONS
1. The tool rotational speed has positive effect (11.5%) on temperature at weld nugget. The weld
speed has 24% positive effect on temperature. When weld speed is increased temperature at
weld nugget is also increased. Tool pin diameter has no effect on temperature at weld nugget.
Interaction effect of tool speed and weld speed is less (1.5%). Interaction effect of tool speed
and pin diameter has 12.5 % positive effect on temperature. Interaction of weld speed and pin
diameter has 10% positive effect on temperature. Interaction of tool speed, weld speed and pin
diameter increases weld nugget temperature.
2. When tool rotational speed and weld speed increases , the temperature at heat affected zone is
increased. When these parameters decreased the temperature also decreased. Interaction values
also prove that input parameters were directly proportional to temperature at heat affected zone.

12. International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 – 6340(Print),
ISSN 0976 – 6359(Online), Volume 5, Issue 9, September (2014), pp. 336-346 © IAEME
346
3. Increase in tool rotation speed causes more heat input which, in turn, enlarges the TMAZ and
HAZ. Increasing the weld speed reduces the heat input resulting in smaller TMAZ and HAZ
which leads to greater tensile strength.
REFERENCE
[1] The welding of aluminium by N.R. Mandal
[2] The Design and Analysis of Experiments by Monte Gomery
[3] Investigation on Interaction Effects of Tool Geometry and Welding Speed on Tensile
Strength of Friction Stir Welded AA1080 Joints 1Vinod Kumar, 2Sunil Kumar IJRMET
Vol. 3, Issue 2, May - Oct 2013
[4] Modeling the Effects of Tool Probe Geometries and Process Parameters on Friction Stirred
Aluminium Welds H. K. Mohanty1, D.Venkateswarlu1, M. M. Mahapatra1,*, Pradeep
Kumar1, N. R. Mandal2 Journal of Mechanical Engineering and Automation 2012, 2(4): 74-
79
[5] http://www.globalspec.com/industrial-directory/1100_h14_aluminum
[6] Microstructure and tensile strength of friction stir welded joints between interstitial free steel
and commercially pure aluminiumS. Kundua,c,, D. Roy b, R. Bholac, D. Bhattacharjee b B.
Mishrac, S Chatterjeea.
[7] Heat Transfer in Friction Stir Welding—Experimental and Numerical Studies YuJ.Chao
Mem. ASME X. Qi W. Tang.
[8] Comparison of RSM with ANN in predicting tensile strength of friction stir welded AA7039
aluminium alloy joints A. K. Lakshminarayanan, V. Balasubramanian, Centre for Materials
Joining Research (CEMAJOR), Department of Manufacturing Engineering, Annamalai
University, Annamalai Nagar-608 002, Tamil Nadu, India.
[9] Artificial neural network application to thefriction stir welding of modelling plates Hasan
Okuyucu a, Adem Kurt a,*, Erol Arcaklioglu b.
[10] D.Muruganandam and Dr.Sushil lal Das, “Friction Stir Welding Process Parameters For
Joining Dissimilar Aluminum Alloys”, International Journal of Mechanical Engineering
Technology (IJMET), Volume 2, Issue 2, 2011, pp. 25 - 38, ISSN Print: 0976 – 6340,
ISSN Online: 0976 – 6359.
[11] D. Kanakaraja, P. Hema and K. Ravindranath, “Comparative Study on Different Pin
Geometries of Tool Profile in Friction Stir Welding using Artificial Neural Networks”,
International Journal of Mechanical Engineering Technology (IJMET), Volume 4, Issue 2,
2013, pp. 245 - 253, ISSN Print: 0976 – 6340, ISSN Online: 0976 – 6359.
[12] Dalip Kumar, Antariksha Verma, Sankalp Kulshrestha and Prithvi Singh, “Microstructure
and Mechanical Properties of Mild Steel-Copper Joined by Friction Welding”, International
Journal of Mechanical Engineering Technology (IJMET), Volume 4, Issue 5, 2013,
pp. 295 - 300, ISSN Print: 0976 – 6340, ISSN Online: 0976 – 6359.
[13] C.Devanathan, A.Murugan and A.Suresh Babu, “Optimization of Process Parameters in
Friction Stir Welding of AL 6063”, International Journal of Design and Manufacturing
Technology (IJDMT), Volume 4, Issue 2, 2013, pp. 42 - 48, ISSN Print: 0976 – 6995,
ISSN Online: 0976 – 7002.