Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 97 (2014) 1991 – 2000

12th GLOBAL CONGRESS ON MANUFACTURING AND MANAGEMENT, GCMM 2014

Parameter optimization for surface roughness and wall thickness on

AA5052 Aluminium alloy by incremental forming using response
surface methodology
V.Mugendirana, A.Gnanavelbabub*, R.Ramadossc
a
Associate Professor, Department of Mechanical Engineering, S.K.P Institute of Technology, Tiruvannamalai-606 603, TamilNadu, India
b
Associate Professor, Department of Industrial Engineering, CEG campus, Anna University, Chennai-600 025, TamilNadu, India
c
Professor, Department of Mechanical Engineering, Easwari Engineering College, Chennai-600 089, TamilNadu, India

Abstract

Surface quality and wall thickness mainly depends on the input parameters during forming process. This study aims to optimize
surface roughness and wall thickness through incremental forming on AA5052 Aluminium alloy at room temperature by
controlling the effects of forming parameters. Design of experiments has been used to study the effects of forming parameters.
The influence of three input parameters, (spindle speed, tool feed, and steps size) along with surface roughness and wall
thickness as output parameters were analyzed. Obtained experimental results from incremental forming were used for analysis.
The optimal results were predicted based on Response Surface Methodology and the analysis of variance. The obtained results
predict a predominant interaction between the forming parameters which can be effectively and efficiently identified to produce
minimum surface roughness and maximum wall thickness.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2014 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selectionand
Selection and peer-review
peer-review underunder responsibility
responsibility of the Organizing
of the Organizing Committee Committee
of GCMM 2014of GCMM 2014.

Keywords:Incremental forming; Optimization; Surface roughness; Wall thickness; Response surface methodology

1. Introduction

Incremental forming (IF) is one of the most promising techniques due to its various applications. Single Point
Incremental Forming (SPIF) is a process for producing complex external shapes and profiles in a sheet metal using a

* Corresponding author; Tel.: +91- 9551133779 ;

E-mail: [email protected]

1877-7058 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and peer-review under responsibility of the Organizing Committee of GCMM 2014
doi:10.1016/j.proeng.2014.12.442
1992 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000

hemispherical shaped tool controlled by means of a CNC milling machine. Since it does not require dies and punch
to form a complex shape, it is very appropriate for rapid prototyping. The tool travels in the programmed path and
deforms the sheet into desired shape. Some of the outstanding features, such as flexibility, low cost tooling, makes it
suitable for various applications. It is capable to manufacture various irregular complex components and highly
customized medical components [1-3].
Many researchers studied the metal forming parameters like spindle speed, tool feed and step size. In this study,
an attempt has been made to optimize the metal forming parameters such as surface roughness and sheet thickness
after forming [4-7]. The aim in this study is to obtain minimum surface roughness and maximum wall thickness in
incremental forming. Response surface methodology has been used to develop mathematical relations between the
forming parameters (spindle speed (V), tool feed (F) and step size(S)) and response parameters (surface roughness
(Ra) and wall thickness (t)) by using the experimental data obtained through experimentation [8-10]. A five level
full central composite factorial design was chosen with quadratic model to optimize the forming parameters.
Analysis of variance test has been done to test the adequacy of the developed mathematical model.

Nomenclature

Ra surface roughness in micro m

t wall thickness in mm
A first factor or input variable investigated
B second factor or input variable investigated
C third factor or input variable investigated
V spindle speed in mm/min
F tool feed in mm/rev
S step size in mm
DF degree of freedom
Prob>F portion of time or probability on would expect to get the stated F-value

2. Material and Methodology

AA5052 Aluminium alloy sheet metal of thickness 1 mm in cold rolled condition was used for experimentation.
Tensile test specimens were prepared according to ASTM E8 standard. The chemical composition of AA5052 was
given in table 1 [11]. Tensile tests were carried out to determine the mechanical properties [12-14]. The yield
strength is 243.4 MPa, ultimate tensile strength is 272.5 MPa, percentage of elongation is 13% and average hardness
is 96.63 (HV 0.5). To carry out the experiments on numerically controlled milling machine HAAS V2 was used
(figure 1). The blank with size 150 mm x 150 mm was held in a fixture shown in figure 2(a). The fixture along with
the sheet metal is mounted on the table of the CNC machine is shown in figure 2(b). A frustum of a cone with 100
mm as maximum diameter, 50 mm as minimum diameter and 50 mm depth was formed incrementally in AA5052
Aluminum sheet (figure 3).

Table.1 Chemical composition of Al 5052 alloy both nominal and actual (wt %)

Composition Mg Cr Si Fe Cu Mn Zn Al
Nominal 2.23 0.18 0.14 0.31 0.01 0.05 0.001
Remaining
Actual 2.24 0.15 0.25 0.40 0.10 0.10 0.10
 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000 1993

Fig. 1 CNC Machine for metal forming process

Fig. 2(a) Fixture to hold sheet metal and (b) fixture with sheet metal mounted on CNC machine

Fig. 3 Cone formed by increment forming process

A high speed hemispherical end tool of 12 mm diameter is used to perform the experiments. The experiments
were carried out at room temperature. The Ra and t at the formed area of each workpiece was measured. Reading
was repeated three times for better results. Values are taken at different regions and the average was calculated and
tabulated.
1994 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000

3. Experimental Design and Response Surface Modeling

Design of Experiments (DOE) is a method, used to reduce the number of experiments to obtain the maximum
optimum conditions. Response Surface Methodology (RSM) explores the relationships between the primary
variables and one or more output response variables. Central Composite Design (CCD) tool is used to determine the
number of experiments required to study the responses [15-16]. The purpose of the analysis of variance is to
investigate which forming parameters significantly affect the Ra and t. The Fisher’s ratio is used to determine
whether the parameter has a significant effect on the output characteristics by comparing the F test value with T
table value (F0.05) at 5% significance level. If the F test value is greater than F0.05, the forming parameter is
considered significant [17].
The forming parameters and response parameters were modeled using response surface method. The aim is to
obtain the optimal response of the inputs to the output through a quadratic model. This design consists of the
following three portions: a) a complete 2k factorial design, where k is the number of variables whose factors level
are coded as -1 and 1, b) axial portion of 2k points arranged in a manner such that two points are chosen at a
distance of α from the design center and c) no center points. Thus the total number of design points in a CCD is n =
2k+2k+no. The minimum possible number of experiments (N) can be determined from the following equations.

N = nf + na (1)

Where nf = 2k and na = 2k, nf defines the number of factorial points and na defines the number of axial points or
star points [18]. The factors and levels used in the factorial design were given in table 2.

Table 2 Factors and levels used in factorial design

Medium High
Forming parameter Factor Unit Low Level
Level Level
Spindle speed V mm/min 1500 2000 2500
Feed F mm/rev 500 650 800
Step size S mm 0.25 0.5 0.75

A five level central composite experimental design with categorical factor was employed to optimize the surface
roughness and t in sheet metal during incremental forming. The design was composed of five levels and a total of 20
experiments were carried out to optimize the input variables. In this study, three parameters (8 factorial points and 6
axial points) and their output (Ra and sheet thickness) were studied.
Design Expert provides prediction equations in terms of actual units and coded units. The coded equations are
determined first, and the actual equations are derived from the coded. To get the actual equation each term in the
coded equation is replaced with its coding formula.

௑ೌ೎೟ೠೌ೗ ି௑ത
ܺ௖௢ௗ௘ௗ ൌ  (2)
ሺ௑ಹ೔ ି௑ಽ೚ೢ ሻȀଶ

The experimental results from the forming trials performed according to the matrix by central composite full
factorial design are tabulated in Table.3. These results are given as input in Design Expert software for further
analysis. The most commonly used quadratic equation to fit the experimental data and to determine the output
response is given by,

ܻ ൌ  ߚ଴ ൅ ߚଵ ‫ ܣ‬൅ ߚଶ ‫ ܤ‬൅ ߚଷ ‫ ܥ‬൅ ߚସ ‫ ܤܣ‬൅ ߚହ ‫ ܥܣ‬൅ ߚ଺ ‫ ܥܤ‬൅ ߚ଻ ‫ܣ‬ଶ ൅ ߚ଼ ‫ܤ‬ଶ ൅ ߚଽ ‫ ܥ‬ଶ (3)
 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000 1995

Table 3 Study of experimental variables in coded units

Factor 1 Factor 2 Factor 3 Response 1 Response 2

Std Run A:V B:F C:S Ra t
mm/min mm/rev mm micro m mm
1 15 -1 -1 -1 4.38 0.726
2 19 1 -1 -1 3.78 0.671
3 1 -1 1 -1 3.47 0.739
4 17 1 1 -1 3.85 0.697
5 7 -1 -1 1 3.37 0.732
6 16 1 -1 1 3.89 0.691
7 14 -1 1 1 3.29 0.726
8 12 1 1 1 3.94 0.701
9 2 -1.682 0 0 5.06 0.716
10 5 1.682 0 0 5.28 0.648
11 9 0 -1.682 0 3.54 0.711
12 11 0 1.682 0 3.32 0.738
13 6 0 0 -1.682 2.41 0.741
14 18 0 0 1.682 1.92 0.745
15 10 0 0 0 2.81 0.754
16 4 0 0 0 2.81 0.754
17 13 0 0 0 2.81 0.754
18 8 0 0 0 2.81 0.754
19 3 0 0 0 2.81 0.754
20 20 0 0 0 2.81 0.754

4. Result and Discussion

4.1. Statistical Analysis

The optimal conditions for Ra and t of AA5052 aluminium alloy sheet metal formed by incremental forming
were determined by means of central composite design using response surface methodology. The obtained ANOVA
for response surface quadratic models are tabulated in the tables 4 and 5. The quality of the fitted model was given
by the coefficient of determination, R2. This gives the proportion of the total deviation in the predicted response and
a high R2 is desirable (close to 1). Considering the determination coefficient R2(adj) = 98.91% for Ra and R2(adj) =
99.58% for t, the equation demonstrates that the model is well fitted. Model terms were evaluated by the F
probability value with 95% confidence level. The P values were used to check the significance of each coefficient.
The P values less than 0.05 indicates that the model and model terms were statistically significant [19]. By dividing
the difference between the maximum predicted response and the minimum predicted response by the average
standard deviation of all predicted responses adequate precision measures signal to noise ratio was computed. Ratios
greater than 4 are desirable. In case of Ra the value was 38.593 and in case of t the value was 56.094 which were
well above 4, which indicated adequate signals to use this model to navigate the design space.
1996 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000

Table 4 ANOVA for Response Surface Quadratic Model (response: Ra in μ m)

Sum of Mean p-value

Source df F Value
Squares Square Prob> F
Model 13.27882 9 1.475 100.574 < 0.0001 Significant
A-V 0.127583 1 0.127 8.696 0.0146
B-F 0.112587 1 0.112 7.674 0.0198
C-S 0.240969 1 0.240 16.425 0.0023
AB 0.154013 1 0.154 10.498 0.0089
AC 0.241513 1 0.241 16.463 0.0023
BC 0.082012 1 0.082 5.590 0.0397
A2 10.60282 1 10.602 722.754 < 0.0001
2
B 0.847922 1 0.847 57.799 < 0.0001
2
C 0.603751 1 0.603 41.155 < 0.0001
Residual 0.1467 10 0.014
Lack of Fit 0.1467 5 0.029
StdDev 0.12 R2 0.9891
2
Mean 3.42 Adjusted R 0.9792
2
C.V% 3.54 Predicted R 0.9066
PRESS 1.25 Adequate Precision 39.593

Table 5 ANOVA for Response Surface Quadratic Model (response: t in mm)

Sum of Mean p-value

Source df F Value
Squares Square Prob> F
Model 0.017679 9 0.001964 262.7464 < 0.0001 Significant
A-V 0.005633 1 0.005633 753.449 < 0.0001
B-F 0.000572 1 0.000572 76.55038 < 0.0001
C-S 4.12E-05 1 4.12E-05 5.513814 0.0408
AB 0.000105 1 0.000105 14.06102 0.0038
AC 0.00012 1 0.00012 16.06735 0.0025
BC 0.000153 1 0.000153 20.48127 0.0011
A2 0.009944 1 0.009944 1330.084 < 0.0001
2
B 0.001821 1 0.001821 243.6271 < 0.0001
2
C 0.000319 1 0.000319 42.60903 < 0.0001
Residual 7.48E-05 10 7.48E-06
Lack of Fit 7.48E-05 5 1.5E-05
StdDev 2.734E-3 R2 0.9958
2
Mean 0.73 Adjusted R 0.9920
2
C.V% 0.38 Predicted R 0.9680
PRESS 5.674E-4 Adequate Precision 56.094
 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000 1997

PRESS stands for ‘Prediction Error Sum of Squares’ and it is a measure indicates how well the model for the
experiment is likely to predict the responses in new experiments. Small values of PRESS are desirable. In case of Ra
the value was 1.25 and in case of t the value was 5.674E-4. Model fitting with the help of Design-Expert software
suggested that a quadratic model provided the best fit, and the model was found to have insignificant lack of fit.
Based on the fig. 4(a) and (b) a correlation response equation for Ra and t with respect to the input parameters in
terms of coded factors are given by the following equation (4) and (5).

ܴܽ ൌ ʹǤͺͳ ൅ ͲǤͲͻ͹‫ ܣ‬െ ͲǤͲͻͳ‫ ܤ‬െ ͲǤͳ͵‫ ܥ‬൅ ͲǤͳͶ‫ ܤܣ‬൅ ͲǤͳ͹‫ ܥܣ‬൅ ͲǤͳͲ‫ ܥܤ‬൅ ͲǤͺ͸‫ܣ‬ଶ ൅ ͲǤʹ‫ܤ‬ଶ െ ͲǤʹͲ‫ ܥ‬ଶ (4)

‫ ݐ‬ൌ ͲǤ͹ͷ െ ͲǤͲʹ‫ ܣ‬൅ ͸ǤͶ͹Ͷ‫ିͲͳݔ‬ଷ ‫ ܤ‬൅ ͳǤ͹͵͹‫ିͲͳݔ‬ଷ ‫ ܥ‬൅ ͵Ǥ͸ʹͷ‫ିͲͳݔ‬ଷ ‫ ܤܣ‬൅ ͵Ǥͺ͹ͷ‫ିͲͳݔ‬ଷ ‫ܥܣ‬

െͶǤ͵͹ͷ‫ିͲͳݔ‬ଷ ‫ ܥܤ‬െ ͲǤͲʹ͸‫ܣ‬ଶ െ ͲǤͲͳͳ‫ܤ‬ଶ െ ͶǤ͹Ͳʹ‫ିͲͳݔ‬ଷ ‫ ܥ‬ଶ (5)

Figure 4(a) and (b) shows the graph for Ra and t plotted against experimental and predicted values. The points
are well distributed and closer to the straight line (R2 = 98.91 and R2 = 99.58) which gives an excellent relationship
between the experimental and predicted Ra values and t.

Fig. 4 Plot for (a) Ra and (b) t by experimental and predicted responses.

4.2. 3D Response Surface Plot

The 3D response surface plots which are the graphical representation of the regression equation, are useful to
understand both interaction properties between the input and output parameters [20-21]. The ultimate aim of the plot
is to predict the optimum values of the variables such the responses is maximized or minimized. Each contour
represents an infinite number of combinations of two input variables with the response maintained at zero level.
Elliptical contour is considered as a measure of perfect interactions among independent variables. The response
surface models for Ra and t are given in the figure 5 (a) to (f). The figure shows the estimated Ra and t as a function
of input variables.
1998 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000

Fig. 5 (a-f) Response surface plot for surface roughness Ra and wall thickness t.

From the analysis of 3D graphs, the major parameters that influence Ra are spindle speed and feed. As far as t is
considered, all the three parameters have significant interactions between them.
 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000 1999

4.3. Multi Response Optimization

Numerical optimization will optimize any combination of one or more goals. The goals may apply to either
factors or responses. The ramp plot was used to analysis the results obtained. The obtained data was optimized
numerical for minimum Ra and maximum t. Desirability is an objective function that ranges from zero outside of the
limits to one at the goal. The numerical optimization finds a point that maximizes the desirability function. For
several responses and factors, all goals get combined into one desirability function. Ramps view shows the
desirability for each factor and response, as well as the combined desirability [22]. A highlighted point shows both
exact value of the factor or response and how well that goal was satisfied. A ramp plot for desirability of 0.916 was
shown in figure 5.

Fig. 6 Ramp function plot for optimized parameters

When spindle speed, feed and step size were 1931.94 rpm, 654 mm/rev and 0.65 mm a minimum Ra of 2.45151
μ m and a maximum t of 0.753 mm can be obtained.

5. Conclusion

Optimization of incremental forming of AA5052 Aluminium alloy sheet was achieved by five factorial full
central composite design using response surface methodology in 20 runs. A second-order quadratic model has been
obtained to predict the surface roughness (Ra) and wall thickness (t) as function of spindle speed, tool feed and step
size variables. A minimum Ra of 2.45 μ m and maximum t of 0.753 mm were obtained at a spindle speed of 1931
rpm with feed 654 mm/rev and step size 0.65 mm. The study also has a higher R2 value above 0.93 and a lower
PRESS value indicating their usefulness in incremental forming. This study will be helpful in characterizing the
input variable during incremental forming.

References

[1] Y.H. Kim, J.J. Park, Effect of process parameters on formability in incremental forming of sheet metal, Journal of Material Processing
Technology, 130-131 (2002), 42-46.
[2] J.J. Park, Y.H. Kim, Fundamental studies on the incremental sheet metal forming technique, Journal of Materials Processing Technology,
140 (2003), 447– 453.
[3] G.Ambrogio, L. De Napoli, L.Filice, F.Gagliardi, M.Muzzupappa, Application of incremental forming process for high customized
medical product manufacturing, Journal of Materials Processing Technology, 204 (2005), 290–303.
2000 V. Mugendiran et al. / Procedia Engineering 97 (2014) 1991 – 2000

[4] M.Bambach, M.Cannamela, M.Azaouzi, G.Hirt, J.l.Batoz, Computer-aided tool path optimization for single point incremental sheet
forming, Advanced Methods in Material Forming, (2006), 234–250.
[5] A.Attanasio, E.Ceretti, C.Giardini, L.Mazzoni, Asymmetric two points incremental forming: improving surface quality and geometric
accuracy by tool path optimization, Journal of Material Processing Technology, 197 (2008), 59–67.
[6] J.L. Yang, J.C. Chen, A systematic approach for identifying optimum surface roughness performance in end milling operations, Journal of
Industrial Technology, 17 (2001), 1–8.
[7] Mohamed Azaouzi, NadhirLebaal, Tool path optimization for single point incremental sheet forming using response surface method,
Simulation Modelling Practice and Theory, 24 (2012), 49-58.
[8] J. AshvinMakadia, J.I. Nanavati, Optimisation of machining parameters for turning operations based on response surface methodology,
Measurement,46 (2013), 1521–1529.
[9] M.Subramanian, M.Sakthivel, K.Sooryaprakash, R.Sudhakaran, Optimization of end mill tool geometry parameters for Al7075-T6
machining operations based on vibration amplitude by response surface methodology, Measurement, 46 (2013), 4005–4022.
[10] Dun Liu,ChuanzhenHuangn, JunWang,HongtaoZhu,PengYao,ZengWenLiu, Modeling and optimization of operating parameters for
abrasive waterjet turning alumina ceramics using response surface methodology combined with Box–Behnken design Ceramics
International, 40 (2014), 7899–7908.
[11] V.Mugendiran, A.GnanavelBabu, R.Ramadoss, Tensile Behaviour of Al5052 Alloy Sheets Annealed at Different Temperatures, Advanced
Materials Research, 845 (2014), 431-435
[12] Gilbert Kaufman, Properties of Aluminium Alloys – Tensile, Creep, and Fatigue Data at High and Low Temperatures, ASM International,
Material Park Ohio, 2006, p. 431-439.
[13] George Dieter, Mechanical Metallurgy, McGraw Hill Book Company, 1998, p. 275-324.
[14] Joseph R Davis, Tensile Testing, ASM International, Material Park Ohio, 2004, p.12-31.
[15] MuhammetDemirell and BerkantKayan, Application of response surface methodology and central composite design for the optimization
of textile dye degradation by wet air oxidation, International, Journal of Industrial Chemistry, (2012).
[16] AmitKohli, Hari Singh, Optimization of processing parameters in induction hardening using response surface methodology, Indian
Academy of Sciences, 36, Part 2, (2011), 141–152.
[17] D.Lazarevic, M.Madic, P.Jankovi, A.Lazarevic, Cutting parameters optimization for surface roughness in turning operation of
polyethylene (PE) using Taguchi method, Tribology in Industry, 34(2) (2012), 68-73
[18] Sameh S. Habib, Study of the parameters in electrical discharge machining through response surface methodology approach, Applied
Mathematical Modeling, 33 (2009), 4397-4407.
[19] M. Villeta, E.M. Rubio, J.M. Sáenz De Pipaón, M.A. Sebastián, Surface Finish Optimization of Magnesium Pieces Obtained by Dry
Turning Based on Taguchi Techniques and Statistical Tests.MaterialsandManufacturing Processes, 26 (2011), 1503-1510.
[20] Murat Sarikaya,AbdulkadirGullu, Taguchi design and response surface methodology based analysis of machining parameters in CNC
turning under MQL. Journal of Cleaner Production, 65 (2014), 604-616.
[21] G. Quintana, J.De. Ciurana, J.Ribatallada, Surface roughness generation and material removal rate in ball end milling operations.
Materials and Manufacturing Processes, 25 (2010), 386-398.
[22] Design Expert®, Software for Experiment Design, Version 7, Stat-Ease, Inc. Minneapolis, MN, 2005.