scientific paper summary

Engineering Science and Technology, an International Journal 22 (2019) 346–352
Contents lists available at ScienceDirect

Engineering Science and Technology,
an International Journal

journal homepage: www.elsevier.com/locate/jestch
Full Length Article
Design optimization of industrial robot arm to minimize redundant
weight
https://doi.org/10.1016/j.jestch.2018.11.009
2215-0986/� 2018 Karabuk University. Publishing services by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.
E-mail address: [email protected] (M. Karali).
URL: http://www.konya.edu.tr (M. Karali).

Peer review under responsibility of Karabuk University.
Mustafa Bugday, Mehmet Karali ⇑
Necmettin Erbakan University, Engineering and Architectural Faculty, Mechatronics Engineering Department, Konya, Turkey

a r t i c l e i n f o a b s t r a c t
Article history:
Received 6 January 2018
Revised 6 November 2018
Accepted 19 November 2018
Available online 8 December 2018

Keywords:
Robot arm optimization
Industrial robot
Finite element method (FEM)
In industrial robots, projected torque increases depending on the extending reach length and payload.
This requires selection of powerful motors particularly on the second axis. Since arm rigidity becomes
more important as the expected positioning precision increases, less flexible materials are used.
Therefore, during operation conditions, 70% of motor’s energy is used for redundant weight. In this study,
analyses conducted on five different robotic arms belong to different brands. Arm’s payload distribution
in terms of region and amount is examined, alternative designs are analysed and compared with each
other. Geometry and materials are changed in alternative designs. In this way, redundant weight is min-
imized (without increasing the share amount) at the same positioning precision. Results of this study
demonstrated a 10% decrease in inert payloads.
� 2018 Karabuk University. Publishing services by Elsevier B.V. This is an open access article under the CC

BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction deformations formed on the robotic arm. From those studies, Kur-
Servomotors are used in industrial robots because of their pre-
cise positioning control and stable structures and the cost of those
motors increases in proportion with their torque. A robot with a
lifting capacity between 5 and 7 kg has about 300 kg average
weight. Especially the second axis motor carries 70–75% of the
robot’s total weight. Considering the high acceleration of robot
arms, it will be better to use a more powerful motor. Therefore,
it is important to select a different material with same strength
or to change the geometric structure of the same material to get
a reduced motor load.

There are some other studies analyzed performance or effi-
ciency of robots by employing different criteria and methods
Pupaza et al. [1] reduced material by geometric changes on the sec-
ond axis and conducted strength analysis. Their analysis showed
that robot arm became lighter and no extra deformation occurred
under the same load. In a similar study, Supriya Sahu et al. [2] con-
ducted FEM analyses of loads on a robotic arm with six axis, and
calculated the deformation, stress and stress values. They identi-
fied the location with the maximum deformation and tried to sta-
bilize the robot by implementing proper minimization techniques
on those parts. In some other studies, alternative materials and
arm geometries are tested by examining the stress and elastic
nool ve K. Urmila [3] by using cylinder or square segments made of
steel, al-356 and aramid epoxy materials; Prasad et al. [4] by using
magnesium, aliminium, zink and steel materials; Chong et al. [5]
and Rueda [6] by changing material properties and geometric
design of the arm, all conducted stress and deformation analyses.
Results of those studies presented the proper motor and load quan-
tity. While calculating the load range that the robot can securely
lift, Singh et al. [7] employed analytic calculations and Chitte
et al. [8] employed vibration analyses. On the other hand, Zhou
et al. [9] focused on the design change by identifying the load range
that a robot can securely lift to increase robot’s lifting capacity.

When the articulated robots extend their arms in full and hor-
izontally as shown in Fig. 1, the torque caused by robot’s own
weight is extremely high. In addition to that, when the robot is
loaded at its maximum capacity, the stress on the second axis
arm is increasing and it is growing closer to the rotation axis. There
are other studies supporting this conclusion [10–12].

Since it would heterogeneously increase the amount of defor-
mation, concentration of stress on a narrow region is an undesired
situation. If those loads could be dispersed by a geometrical
change, particularly on the second axis arm, deformation at the
tip of the robot will be decreased. It is known that in structures
with a support on one side and load on the other, conversion of
rotating forces to the axial forces in the form of pull an push,
increases the strength and decreases the mass [13,14]. Fig. 2 shows
that when a proper channel is opened on a part, stress that concen-
trated on a specific region is dispersed to different regions.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jestch.2018.11.009&domain=pdf

http://creativecommons.org/licenses/by-nc-nd/4.0/

https://doi.org/10.1016/j.jestch.2018.11.009

http://creativecommons.org/licenses/by-nc-nd/4.0/

mailto:[email protected]

http://www.konya.edu.tr

https://doi.org/10.1016/j.jestch.2018.11.009

http://www.sciencedirect.com/science/journal/22150986

http://www.elsevier.com/locate/jestch

Fig. 2. The advantage of digging a hole.

Fig. 1. Articulated robot.

M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352 347
By opening a proper channel on the second axis, this study
attempts to decrease the weight through material reduction, to
increase maximum strength by reducing mass inertia. Following
is a list of some features distinguish this study from others;

1. Instead of designing a new robot arm, this study tries to opti-
mize existing arms belong to known brands.

2. An articulated robot is analyzed by considering the motor load.
3. The chosen method to dispose the redundant weights is to

reduce the material by opening channels. Location of channels
are determined with ANSYS 16.2 Shape Optimization package.

4. The force caused by the weight of disposed material is not
ignored and this weight is added to the tip of the robot. When
reducing material, a simulation that would not deteriorate
arm’s strength and deformation values and would not increase
the motor load was applied

2. Preliminary works

In to select the proper robot model for this study, 3D
models of five most preferred robots in industry were obtained
Fig. 3. Appearance of six axis robot.

Table 1
Total amount of deformation d and maximum stress ru on the robots found at FEM analy

ABB44 FANUC
(45 kg) (200 kg

Steel d (base) 0,30105 2,1368
d (hole) 0,28803 2,1105
ru (base) 40,696 162,3
ru (hole) 40,32 160,7

Aluminium d (base) 0,52679 1,9242
d (hole) 0,52277 1,9316
ru (base) 32,553 141,73
ru (hole) 33,194 141,74

Epoxy d (base) 1,5239 3,2338
d (hole) 1,4439 3,2766
ru (base) 31,945 200
ru (hole) 31,828 180,91
through Solidworks. The maximum load that they can carry
was applied on the sixth axis. To facilitate the job, a P force
which equals to the distributed load caused by the weights of
the third, the fourth, the fifth and the sixth axis and motor’s
weight was calculated. As shown in Fig. 3, that P force was
reduced as the F force to the second axis of the robot arm after
static calculations.

The total amount of deformation was calculated for 5 different
robots. For each robot, by holding the assumed maximum load
conditions constant, some holes were dug on the second axis
arm. Simulation was repeated after the forces that caused by
weights of the excavated pieces was added to the F force. The
results are presented in Table 1.

Table 1 shows comparison of robot arms before and after dig-
ging the holes (maximum weights robot arms can lift are shown).
In Abb44, Fanuc and Kuka robot arms holes were drilled on second
axes where stress values are at minimum, in Staubli ve Abb120
robot arms holes were drilled on second axes where stress values
are at maximum. Assessment of results in terms of material and
holes showed that maximum stress and displacement on the sec-
ond axes of Abb44 and Fanuc robot arms has decreased but maxi-
mum stress and displacement on the second axes of Staubli and
Abb120 robot arms has increased. For Kuka robot arm, a fluctua-
tion was observed in both maximum stress and displacement val-
ues. In to identify accurate place and hole that would lower
the fluctuations to the minimum, Shape Optimization module of
Ansys applied to Kuka’s 3D model. Simulation results were pre-
sented in Table 1.
3. Finite elements method

3.1. Mesh model

In to measure more precise values during the mesh anal-
ysis, body mesh method was employed. Three nodes triangle ele-
ment is chosen as element type and element size was set as
10 mm. In Fig. 4, triangle element with three nodes was presented
and formulized. The number of node was 1,056,426 and the num-
ber of element was 746,142 on the robot arm. The mesh analysis is
shown in Fig. 5.

u

v

� �
¼ N1 0 N2 0 N3 0

0 N1 0 N2 0 N3

� �
u1
v1
u2
v2
u3
v3

8>>>>>>>>< >>>>>>>>:

9>>>>>>>>=
>>>>>>>>;

ð1Þ

In this equation ‘‘u” and ‘‘v” values are displacement functions.
sis.

KUKA STAUBL ABB120
) (6 kg) (5 kg) (4 kg)

0,028138 0,012314 0,0092191
0,0284 0,012407 0,0087651
5,1494 2,4966 2,442
4,0097 2,4247 1,8396

0,035918 0,014729 0,014337
0,039875 0,01574 0,014727
3,4193 2,0609 3,0381
2,8065 2,071 3,0816

0,097524 0,070882 0,017839
0,16775 0,078369 0,019391
3,2955 3,6083 2,4243
3,8603 3,7975 2,5067

Fig. 4. Three nodes triangle element type.

348 M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352
u ¼ b1 þ b2x þ b3y v ¼ b4 þ b5x þ b6y ð2Þ
‘‘u” and ‘‘v” values are expressed in this form (bi (i = 1,2. . .6) con-
stant coefficient).

Shape functions at ‘‘x” and ‘‘y” points:

N1 ¼
1
2A

½ðx2y3 � x3y2Þ þ ðy2 � y3Þx þ ðx3 � x2Þy� ð3Þ

N2 ¼
1
2A

½ðx3y1 � x1y3Þ þ ðy3 � y1Þx þ ðx1 � x3Þy� ð4Þ

N3 ¼
1
2A

½ðx1y2 � x2y1Þ þ ðy1 � y2Þx þ ðx2 � x1Þy� ð5Þ

Here ‘‘A” represents region.

A ¼ 1
2
det

1 x1 y1
1 x2 y2
1 x3 y3

2
64

3
75 ð6Þ
3.2. Static analysis

In Fig. 6 point A represents joint of first and second axes, point B
represents center of gravity of second axis, point C represents joint
of second and third axes, distance between C and E represents part
where third, fourth, fifth and sixth axes are located. ‘‘k” represents
distance between [AB], ‘‘l” represents distance between [BC], ‘‘m”
and ‘‘n” represent distance between [CD] and [DE] of robot arm.
‘‘WB” shows weight of second axis and ‘‘WD” shows total weight
of third, fourth, fifth and sixth axes of robot arm. ‘‘P” force repre-
sents maximum weight that robot arm can lift. Since the optimiza-
tion will be applied to the second axes, ‘‘P” and ‘‘WD” weights
turned into an ‘‘F” force at point C and the moment resulted from
force transmission applied to point C.

Formulization of conversion of ‘‘P” and ‘‘WD” weights into ‘‘F” is;X
F ¼ P þ WD ð7Þ
Fig. 5. Mesh analys
F ¼ 6 � 9; 81 þ 107 � 9; 81

F ¼ 1108N

X
MC ¼ P � ðm þ nÞ þ WD � m ð8Þ

M ¼ 58; 86 � 0; 7 þ 107 � 0; 3

Fig. 6. Analytic calculation.
C

MC ¼ 347; 3 N � m

The maximum weight that Kuka Kr210 industrial robot can
carry is 6 kg. After static calculations, the maximum weight that
the robot can lift was reduced from the sixth axis to the second axis
and converted to a 1108N force. Moment as a result of force
transmission in the second axis became 347,3 N�m. Fig. 7a shows
the limiting values of the robot. In Fig. 7a, point A represents
acceleration due to gravitation. 1108N to point B, 40N to points
E, D and 347,3 N�m as a result of force transmission to point F were
applied. That force (1108N) was applied to the second axis
of the robot arm in ANSYS Static Structural software. The deforma-
tion and stress values were calculated as 0.03444 mm and
6.4506 MPa in simulation. Maximum stress and deformation val-
ues are shown in Fig. 7b and c.

3.3. Modal analysis

This analysis is conducted to identify the causes of vibrations
occurred at the tip of the robot arm and the dynamic characteris-
tics of the structure. Natural frequency and the mod types were
identified by conducting modal analysis on the robot. Natural fre-
quency is important to spot the places where resonance may hap-
pen. Generally the resonance is not a desired situation for a
structure. Resonance can be removed by either changing outer
force frequency or by changing structure’s frequency. Solution of
equation of motion belongs to a multiple degree of freedom system
yields natural frequencies and shape mods. Equation of motion of
an undamped system with multiple degree of freedom is;

½M� €xðtÞf g þ ½K� xðtÞf g ¼ fðtÞf g ð9Þ
[M] and [K] represent mass and rigidity matrices, €xðtÞ; xðtÞ ve

fðtÞ represent momentum, displacement and force vectors.
Analysis results showed that mode 1 and mode 2 were calcu-

lated as f 1 ¼ 121:75 Hz and f 2 ¼ 150:28 Hz, and the structure
is of the robot.

Fig. 8. Robot modes.

Fig. 7. Static analysis. a: Limiting values ve b: Maximum stress c: Maximum Deformation.

M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352 349
twisted in mode 1 and deflection in mode 2. The third and the
fourth modes were calculated as f 3 ¼ 323:48 Hz and f 4 ¼
369:85 Hz, and the structure twisted in mod3 and deflection in
mode4. Finally, the fifth and the sixth modes were calculated as
f 5 ¼ 626:61 Hz and f 6 ¼ 774:73 Hz, and the structure twisted in
both modes. These modes are depicted in Fig. 8.
4. Optimization works

Although the optimization algorithm does not provide more
certain and clear answers than the traditional methods, it gives
an approximate result. This method has two advantages over tradi-
tional methods: cost and time reductions [1]. The main objective of

350 M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352
this study is to identify suitable locations to reduce material from
robot arm by using ANSYS Shape Optimisation analysis. In addi-
tion, different holes were dug and analyzed to identify the best
geometric shape when reducing material. Fig. 9 shows best places
for the holes.
Fig. 9. Shape op

Fig. 10. Sampling points

Fig. 11. Precisio
Main objective of optimization work is to reduce the loads of
motors by decreasing the mass of the arm and getting smaller
deformation and stress values than calculated in static analysis.
Hole’s height L and radius r were inputs to ANSYS Response Surface
Optimization module. The weight of robot arm, the amount of
timization.

on the design space.

n analysis.

Fig. 12. Optimizasyon sonucu a: Maximum Deformation and b: Maximum stress.

Fig. 14. Distribution of stress amounts on the curve.

M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352 351
deformation and stress value were defined as outputs. The soft-
ware searched for the best results by changing L and r within a pre-
defined range ð�3 mmÞ.

The software made some improvements on design space by
using Kriging Response Surface algorithm to accurately identify
the best design domain. It solved the structure through 45 different
samplings. Fig. 10 shows sampling points on the design space.

According to the precision analysis method, the hole placed on
the upper side of the robot arm has greater impact on the output
parameters, namely stress, deformation and robot arm mass.
Fig. 11 shows the precision analysis.

Finally, Fig. 12a and b shows optimized robot arm and mini-
mum amount of deformation achieved.

After the optimization the weight of the robot was reduced by
10% percent. The stress value was reduced about 23% from
6.4506 MPa to 4.9832 MPa. Finally, deformation value is reduced
from 0.034444 mm to 0.012214 mm with a 65% decrease. Results
showed that neither deformation nor stress values was not close
to the anticipated limiting values.

5. Simulation results and discussion

A busy industrial robot can make about 10,000 cycle daily.
Therefore, energy efficiency is the major concern when designing
robot components. Since the diameter of robot arm decreases
toward tip of the robot, the load of servomotor lightens and its
energy consumption is reduced. However since the motors of the
fourth, the fifth and the sixth axises are placed on the shoulder
of the robot, the load of the second axis servomotor increases
and bigger motors are needed. Therefore, this study analyzed stress
and deformation on the second axis arm. In addition, an optimiza-
tion by reducing material to increase efficiency of the robot was
conducted.
Fig. 13. Maximum stress reg
Fig. 13a shows the initial distribution of stress on the robot arm.
Fig. 13b shows the distribution of stress on the robot arm after the
optimization.

Examination of Fig. 14 shows that the region close to the motor
on the second axis has greater amount of stress. Curvilinear accu-
mulation of stress would increase deformation on that region.
Increase in the deformation would eventually affect the robot’s
precision. Industrial robot vendors are trying to solve this problem
by using thicker materials. This study achieved to reduce the max-
imum stress value and motor’s load by distributing the stress on
the second axis to a wider region. Fig. 14 shows base and optimized
stress values.

Fig. 14 shows that the stress values decreases on regions where
maximum stress values are accumulated. There are two reasons for
this decrease:
ions on the second axis.

Fig. 15. Distribution of deformation amounts on the curved line.

352 M. Bugday, M. Karali / Engineering Science and Technology, an International Journal 22 (2019) 346–352
1. By opening channel, the stress in the regions with maximum
stress values are turned into pull and push stress. As a result
of that, the stress is distributed to a wider region.

2. As a result of geometric change weight is reduced.

Normally, when the weight of an arm decreased, the amount of
deformation due to the load increases. However, this situation
reverses depending on the location and geometry of the channel.
In addition to that there is an optimum point between the wide
of the channel and its contribution. Fig. 15 shows that this study
improved the distribution of deformation by using Shape Optimi-
sation module.

Examination of Fig. 15 shows that there is not any increase in
the region with maximum deformation.

6. Conclusion

When mechanical systems working on different loads and fre-
quencies are optimised in terms of energy efficiency and cost, a
partial reduction in operating costs can be achieved. Considering
the limiting factors like elasticity, rigidity, and precision that are
expected from mechanical systems, some costs can be ignored.
However, the optimization of long-life robots not only improves
the cost and efficiency but also the increases the motor’s lifetime.

This study found that in spite of the material reduction by 10%
on the second axis, there was not any increase in the elastic defor-
mation values; moreover, an improvement on the rigidity values of
the material was observed.

References

[1] C. Pupăză, G. Constantin, T. Negrilă, Computer aided engineering of industrial
robots, Proc. Manuf. Syst. 9 (2) (2014) 87–92.

[2] S. Sahu, B. Choudhury, Static analysis of a 6-axis industrial robot using finite
element analysis, Int. J. Mech. Eng. Technol. (IJMET) 8 (3) (2017).

[3] N. Francis, K. Urmila, Design and analysis of circular and square arm robot,
Anveshana’s Int. J. Res. Eng. Appl. Sci. 1 (11) (2016).

[4] C.S.V. Uma Sai Vara Prasad, M. Balaji, Geometric parameters optimization &
analysis of packing robot arm using ansys, Int. J. Adv. Res. Basic Eng. Sci.
Technol. (IJARBEST) 3 (30) (2017) 1–6.

[5] W. Choong, K. Yeo, Structural design for a 3dof robot lower-arm via computer
aided engineering, Centre Mater. Miner., Universiti Malaysia Sabah 88999
(2007) 8–18.

[6] J. Rueda, L. Ángel, Structural analysis of a delta-type parallel industrial robot
using flexible dynamic of ansys 11.0, Industrial Electronics, 2009. IECON’09.
35th Annual Conference of IEEE, IEEE, 2009, pp. 2247–2252.

[7] M. Singh, P. Yadav, S. Pandey, S. Patil, Stress analysis of pick and place robot
using ansys, Int. J. Res. Sci. Eng. (IJRISE) (7) (2017).

[8] P.G. Chitte, S.S. Bansode, S.S. Rathod, N.S. Motgi, Structural and vibrational
anlysis of six axix aristo robot using ansys, IJIRT 3 (1) (2016) 291–296.

[9] J. Zhou, Z. Yang, S. Chen, Analysis of the harvesting robot arm modal based on
cae, J. Chem. Pharm. Res 6 (11) (2014) 669–673.

[10] M. Akdağ, Design and analysis of robot manipulators by integrated cae
procedures, Ph.D. thesis, DEÜ Fen Bilimleri Enstitüsü (2008).

[11] B. Reddy, R.G. Brioso, Automated and generic finite element analysis for
industrial robot design.

[12] M. Makaracı, S. Demir, O. Bahçacı, Direksiyon aksonlarının statik hasar analizi,
Academic Platform J. Eng. Sci. 1 (2) (2013) 21–27.

[13] K. Abdel-Malek, B. Paul, Criteria for the design of manipulator arms for a high
stiffness-to-weight ratio, J. Manuf. Syst. 17 (3) (1998) 209–220.

[14] A.A.H. Ali, D.O. Ramadhan, Comparative analysis for link cross-section of
manipulator arms, J. Eng. 16 (2) (2010).

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0005

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0005

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0010

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0010

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0015

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0015

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0020

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0020

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0020

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0025

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0025

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0025

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0030

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0030

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0030

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0030

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0035

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0035

http://refhub.elsevier.com/S2215-0986(18)30025-9/h9000

http://refhub.elsevier.com/S2215-0986(18)30025-9/h9000

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0045

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0045

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0050

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0050

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0060

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0065

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0065

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0070

http://refhub.elsevier.com/S2215-0986(18)30025-9/h0070

Design optimization of industrial robot arm to minimize redundant weight
1 Introduction
2 Preliminary works
3 Finite elements method
3.1 Mesh model
3.2 Static analysis
3.3 Modal analysis

4 Optimization works
5 Simulation results and discussion
6 Conclusion
References