Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Available online at www.sciencedirect.

com

ScienceDirect
Materials Today: Proceedings 18 (2019) 3717–3725 www.materialstoday.com/proceedings

ICMPC-2019

Topology Optimization Of Industrial Manipulator-Link Considering


Dynamic Loading
G Lakshmi Srinivasa,*, Arshad Javedb
ab
Birla Institute of Technology and science-Pilani, Hyderabad Campus, Hyderabad, Telangana-500078, India.

Abstract

In the present work, mass of an industrial manipulator-link is minimized using topology optimization method. Topology
optimization is established substantial method for mass reduction of structural and machine components. A single link of
manipulator is considered for optimization. For optimizing the design region, minimum compliance is chosen as objective
function. While robotic arm rotates continuously about a fixed end, applied force acts dynamically on the other end. In such case
topology obtained at one angular position is not optimum for other angular positions. Hence, a single topology is not sufficient
for dynamic loading condition. Here, a weighted density parameter approach is proposed to consider the dynamic effect and
produce a single weighted-topology. To compare the performance of weighted-topology, performance values such as maximum
Von-Mises stress and deflection are computed at all discrete angular positions. A MATLAB based routine is developed to
generate the weighted-topology and to simulate the performance values. Further, the obtained results are validated using
simulation software ANSYS Workbench 18.1. The optimal design of weighted-topology is proven as better compare to any other
individual topologies at different angular positions.
© 2019 Elsevier Ltd. All rights reserved.
Selection and peer-review under responsibility of the 9th International Conference of Materials Processing and Characterization, ICMPC-2019

Keywords: Energy efficiency; Industrial manipulator; Topology optimization; Robotic arm; Solid isotropic material with penalization; Dynamic
loading

1. Introduction

In industrial automation, the necessities of industrial manipulators are rising on account of their advanced superiority
and productivity [1]. In this present scenario different attempts are made to reduce the energy consumption of the

* Corresponding author. Tel.: +91-9652456219.


E-mail address: g.l.srinivas7@gmail.com

2214-7853 © 2019 Elsevier Ltd. All rights reserved.


Selection and peer-review under responsibility of the 9th International Conference of Materials Processing and Characterization, ICMPC-2019
3718 G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725

Nomenclature

AP angular position
DOF degree of freedom
EX number of elements in x-direction
EY number of elements in y-direction
SIMP solid isotropic material with penalization

industrial manipulator. Topology optimization is the promising method to remove needless material from the design
space [2]. Substantial study on the application of topology optimization was started by Lohmeier et al. [3]. A few parts of
22-DOF humanoid were optimized. Albers et al. applied topology optimization method on hybrid multibody dynamic
system for reducing weight by 15% [4]. Huang & Zhang focused on static loading condition by considering worst case
of the obtained topologies [5]. Similarly for loading, the worst case condition was chosen Junk et al. Yunfei et al. Chu et
al. Briot & Goldsztejn [6-9]. In the recent attempts light weight serial robots were optimized by topology optimization
and parametric system optimization. Overall mass of the serial robot reduced by 12% [10].
2. Dynamic loading of topology optimization

2.1. Topology optimization method

Topology optimization is a mathematical method to help in synthesizing the optimum material distribution of a
design space [11]. The design space is discretized into bi-linear four node elements with 2 DOF each. Each element
(z) is allocated a density ‘ z ’ that estimates the Young’s modulus ‘  z ’ by SIMP method as shown in equation 1.

 z (  z )   min   zn (    min ),  z  [0,1] (1)


Where, ‘  0 ’ is stiffness of the design space,  min is a minimum stiffness value allocated to the void regions for
avoiding stiffness matrix from singularity; and n is a penalization number incorporated to ensure a binary solid
(black) and void (white) solution. The topology optimization problem is solved by optimality criteria algorithm by
selecting minimum compliance as the objective function shown in equation 2. Compliance represents is reverse of
stiffness, also it represents the amount of strain energy stored in robotic link.
N
min :  ( z )   T      z (  z ) zT    z
z
z 1
     Subjected to :  ( z ) /      0   (2)
    0
0  z 1

where  is the volume fraction,  ( z ) is the compliance,  ,  and  are the global displacement, force vectors, and
global stiffness matrix, respectively,  z is the element displacement vector, N is the total number of elements,
 ( z ) and   are the material volume and design domain volume, respectively, z is the design variable,  z is the
element stiffness matrix,  is the stiffness matrix.

2.2. Topology optimization of robotic arm

The robotic link of 1 DOF fixed at one end other end is subjected to applied force as shown in Fig. 1 [12]. The
length and height of the robotic link are 300 mm and 80 mm respectively. For making lightweight, the material
chosen as aluminum and thickness of the robotic link is 3 mm. While robotic link is rotating continuously at lower
speed, gravitational force and centrifugal forces can be neglected. Black color rib is excluded from topology
optimization problem for assembly purpose. The developed mathematical gradient algorithm is compatible with
MATLAB.
G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725 3719

Pre-defined
solid region

Design region

Fig. 1. Initial material domain for robotic link, specifying the predefine and design region

The design space discretizes into bilinear square finite elements. In each iteration process optimization algorithm
eliminates the needless densities. It updates the design variables upto convergence of the compliance value. A
MATLAB code is developed with sensitive analysis, finite element analysis and filtration of sensitivities. The code
is devolved to compute the deflection and Von-Mises stress on each finite element. Also it is parameterized to run at
any larger mesh value. In order to ensure the accuracy of results a mesh independency check is performed here. For
this analysis a simple horizontal position of link is considered, where the applied force of 5N is acting vertically
downward. The link is considered to be made of Aluminum (density 2770 kg/m3, Poisson’s ratio 0.33, Young’s
modulus 72 GPa) with a volume reduction of 50% i.e. volume fraction of 0.5. The convergence of EX × EY is
obtained at 2100×560 for deflection and von-Mises stress values as shown in Table 1.

Table 1. Mesh independency of Von-Mises stress and deflection at different grid sizes.
S. No. Grid Size Von-Mises stress (MPa) Deflection (mm) Time (min)
1 300 × 80 2.52 0.00767 0.43
2 600 × 160 2.32 0.00772 1.78
3 900 × 240 2.10 0.00774 4.58
4 1200 × 320 2.45 0.00776 8.51
5 1500 × 400 2.41 0.00777 13.82
6 1800 × 480 2.37 0.00780 21.07
7 2100 × 560 2.35 0.00780 26.18
8 2400 × 640 2.35 0.00780 39.51
9 2700 × 720 2.35 0.00780 52.25
10 3000 × 800 2.35 0.00780 56.57

2.3. Methodology for dynamically loaded robotic link

When the robotic arm rotates dynamically, loading or applied force also acts dynamically. It implies that the
topologies obtained for certain angular position will not be optimum for other angular values of applied forces,
because the optimal topology is generated in conjunction with applied forces and its direction. Further, when this
topology will be loaded at different angular positions, the stress and deflection values will not remain in the
minimum state. In Table 2, topologies are different angular position is shown with maximum stress and deflection
values. It is observed that the minimum Von-Mises stress obtained at angular position 00 and minimum deflection is
at 1700.
3720 G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725

Table 2. Maximum Von-Mises stress and deflection of topologies at different angular positions.
Max
Max Stress Max Def Max Stress
AP Topology AP Topology Def
(MPa) (mm) (MPa)
(mm)

0.007
0 3.07366 0.007973 180 3.320476
973

0.008
10 3.769653 0.007925 190 4.991987
122

0.008
20 5.540738 0.008503 200 12.0664
317

0.008
30 6.049187 0.009191 210 19.24732
499

0.008
40 18.38805 0.009787 220 18.38291
705

0.008
50 25.10066 0.018552 230 7.833945
985

0.016
60 15.98333 0.018214 240 30.96347
715

0.025
70 26.65435 0.027416 250 33.09061
514

0.030
80 19.52533 0.025668 260 20.72323
677

0.024
90 13.31952 0.019893 270 17.10901
61

0.030
100 14.45024 0.025643 280 20.70712
596

0.025
110 23.79384 0.027359 290 24.74518
489

0.016
120 20.03142 0.01828 300 22.73593
742

0.008
130 39.4623 0.018595 310 8.021129
984

0.008
140 10.49469 0.009786 320 17.39506
73

0.008
150 6.086783 0.009185 330 18.36728
513

0.008
160 5.778148 0.008506 340 5.775364
325

0.008
170 3.876248 0.007919 350 4.846139
129
G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725 3721

Hence, a single topology will not be sufficient for dynamic conditions. A weighted density parameter approach
will be proposed to consider the dynamic effect and produce a single topology which will be sufficient in dynamic
condition as shown in equation 3.
   z   S1  z1   S2  z2     Sn  zn  (3)
  S
T   (4)

Where, [z] is design variable and (Sn) is weighted coefficient are selected in such a way that it represents the
highly sensitive element at a higher priority and less sensitive elements can be neglected. Since the design variable
(or density value) is binary in nature, the weight coefficient also represented in binary number (i.e. 0 or 1). In order
to satisfy the same volume fraction, the number of weight coefficients having a value ‘1’ is decided based on
equation 4. In this equation T is the number of reference optimal topologies. The proposed weighted density
approach is applied to the synthesized optimum topologies of different angular position and a consolidated topology
named as weighted-topology developed, which is presented in the next section.
3. Results and discussions

3.1. Weighted-topology
As discussed earlier, to satisfy the loading condition and provide minimum stress ad deflection in all angular
position of the link weighted density method is proposed. Different weight values are applied to get a conclusive
topology. The constraint here was to not allow any fluctuation in the desired volume fraction value. In the present
example the number of reference optimal topology is 36. According to equation (3) and (4) to the weighted-topology
is obtained, as shown in Fig. 2. The weighted-topology is optimum for all angular positions.
3.2. Performance estimation of weighted-topology
In order to validate the enhanced performance of the weighted-topology, the stress value and deflection is
obtained. For this, force is applied to the weighted- topology in all set of angular positions and stress with deflection
values are recorded as shown in Table 3 and Table 4.

Table 3. Maximum Von-Mises stress and deflection of weighted-topology at different angular positions.
AP Max Stress (MPa) Max Deflection (mm) AP Max Stress (MPa) Max Deflection (mm)
0 1.976 0.006097 180 1.599 0.006093
10 1.570 0.005919 190 2.485 0.006088
20 1.494 0.007620 200 1.560 0.005898
30 1.377 0.005057 210 1.465 0.005509
40 1.221 0.004413 220 1.322 0.004938
50 1.033 0.003663 230 1.133 0.004188
60 0.819 0.002825 240 0.905 0.003288
70 0.583 0.001920 250 0.647 0.002269
80 0.332 0.000975 260 0.571 0.001164
90 0.234 0.000083 270 0.607 0.000136
100 0.325 0.000974 280 0.704 0.001165
110 0.567 0.001917 290 0.630 0.002272
120 0.796 0.002822 300 0.881 0.003292
130 1.005 0.003660 310 1.102 0.004191
140 1.188 0.004418 320 1.283 0.007428
150 1.339 0.005061 330 1.422 0.005506
160 1.453 0.005566 340 1.515 0.005895
170 1.546 0.005921 350 1.579 0.006086
3722 G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725

Table 4. Deflection and von-Mises Stress of angular position 00 for different Mesh tool based on ANSYS 18.1.

Mesh Tool Maximum von-Mises stress (MPa) Max deflection (mm)

Coarse

Stress
1.2629

Deflection
0.0078914

Fine

Stress
1.2629

Deflection
0.0078917

Finer

Stress
1.2630

Deflection
0.0078916

It is observed that among all angular positions, the maximum von-Mises stress and deflection for weighted-
topology is 2.485 MPa and 0.00762 mm. Which is lower that of best deflection (0.007917 mm) at angular position
1700 and best Von-Mises stress (3.07366 MPa) at angular position 00 as well as individual reference optimal
topologies as shown in Table 2.

Fig. 2. Weighted topology suitable for all angular positions

3.3. Validation of Performance simulation

The results obtained from MATLAB validated through ANSYS at different mesh tools. Out of all angular
positions two topologies are selected based on best performance with respective to von-Mises stress and deflection
at 00 and 1700 respectively. After the execution of algorithm in MATLAB, results are generated in two dimensional
.TIF format. Adobe illustrator is used for converting .TIF file into .DXF file. Two dimensional .DXF file imported
in modeling software SOLIDWORKS and extruded for thickness 3 mm. Generated three dimensional .X_T file
from SOLIDWORKS imported in simulation software ANSYS 18.1. Apply the boundary conditions and constraints
G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725 3723

at different mesh tool to compute the results. The outcome of deflection and von-Mises stress for angular position 00
and angular position 1700 are shown in Table 4 and Table 5 respectively.

Table 5. Deflection and von-Mises Stress of angular position 1700 for different Mesh tool based on ANSYS 18.1.

Mesh Tool Maximum von-Mises stress (MPa) Max deflection (mm)

Coarse

Stress
1.2929

Deflection
0.00785

Fine

Stress
1.2929

Deflection
0.0078501

Finer

Stress
1.293

Deflection
0.0078503

From Table 4 and 5 it is observed that the deflection values are in good agreement with that to MATLAB results.
However, the Von-Mises stress values are shown a variation. The reason for this variation is the different element
type used in MATLAB and ANSYS. The MATLAB uses Q4 bi-linear four node elements with 2 DOF each
element, while ANSYS employed tetrahedron element.

3.4. Performance simulation of weighted-topology using ANSYS 18.1

Based on weighted density parameter weighted-topology is created which is optimum for all angular
positions. The MATLAB results shows that, the created topology is better, because the values of von-Mises
stress and deflections are minimum compared to any other topologies. The validation of performance simulation
of weighted-topology based on ANSYS 18.1 shown in Table 5 and Table 6.
3724 G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725

Table 6. Deflection and von-Mises Stress of weighted-topology for different Mesh tool based on ANSYS 18.1.

Mesh Tool Maximum von-Mises stress (MPa) Max deflection (mm)

Coarse

Stress
1.0676

Deflection
0.0075193

Fine

Stress
1.0691

Deflection
0.0075196

Finer

Stress
1.0695

Deflection
0.0075197

The performance result obtained using MTALB code and ANSYS are compared. The convergence of deflection
is within acceptable limits. The MATLAB result of von-Mises stress are slightly deviates from ANSYS. The reason
for this change is already stated. However, results are within permissible limits of safe design [13]. In order to
prepare a concise manuscript, MATLAB 2017a routine is not presented in the document. The convergence values
are simulated in ANSYS, for illustration only 3 mesh tools coarse, fine and finer are presented because of space
constraint. The complete program and simulation executed in the Dell Precision Tower 5810-825W, 32GB RAM
(2400MHz), Intel Xeon Processor E5-1650 v3 (3.5GHz Turbo).

4. Conclusion

In this paper, a method of weighted density parameter approach is proposed to consolidate the individual
optimum topologies that can suffice dynamic loading condition. A volume fraction of 0.5 was considered for
optimization. The topology optimization was performed by optimality criteria method. A MATLAB based code was
developed to generate the optimum topology and produce a weighed- topology. From simulation of the
G L. Srinivas et al./ Materials Today: Proceedings 18 (2019) 3717–3725 3725

performances, Von-Mises stress and deflection were found as 2.485 MPa, 0.00762 mm, respectively. On the other
hand, the best topologies among the individual angular position topologies were showing Von-Mises stress of
3.07366 MPa and deflection of 0.007917 mm. Moreover these values were obtained for two different topologies.
Hence the proposed method is found to be suited for dynamic loading condition.

Acknowledgements

The work presented in this paper was funded by the “Department of Science and Technology (DST) – Science and
Engineering research board (SERB), India” (File Number: ECR/2017/000799), and Research Initiation Grant-BITS
Pilani.

References

[1] Statistics – IFR International Federation of Robotics Ifr.org. Retrieved 18 October 2018, from https://ifr.org/downloads/press2018/2018-10-
18_Press_Release_IFR_WR_2018_Industrial_Robots_ENG.pdf/ (2018)
[2] M. Bendsøe and O. :Sigmund, Topology optimization: Berlin: Springer (2003).
[3] Lohmeier, S., Buschmann, T., Schwienbacher, M., Ulbrich, H., & Pfeiffer, F. (2006, December). Leg design for a humanoid walking robot.
In Humanoid Robots, 2006 6th IEEE-RAS International Conference on (pp. 536-541). IEEE.
[4] Albers, A., & Ottnad, J. (2008, December). System based topology optimization as development tools for lightweight components in
humanoid robots. In Humanoid Robots, 2008. Humanoids 2008. 8th IEEE-RAS International Conference on(pp. 674-680). IEEE.
[5] Huang, H. B., & Zhang, G. (2012). The topology optimization for l-shape arm of Motorman-HP20 robot. In Applied Mechanics and
Materials (Vol. 201, pp. 871-874). Trans Tech Publications.
[6] Junk, S., Klerch, B., Nasdala, L., & Hochberg, U. (2018). Topology optimization for additive manufacturing using a component of a
humanoid robot. Procedia CIRP, 70, 102-107.
[7] Yunfei, B., Ming, C., & Yongyao, L. (2016). Structural Topology Optimization for a Robot Upper Arm Based on SIMP Method.
In Advances in Reconfigurable Mechanisms and Robots II (pp. 725-733). Springer, Cham.
[8] Chu, X. Y., Xu, H. H., Shao, G. F., & Zheng, W. F. (2016, August). Multi-objective topology optimization for industrial robot.
In Information and Automation (ICIA), 2016 IEEE International Conference on (pp. 1919-1923). IEEE.
[9] Briot, S., & Goldsztejn, A. (2018). Topology optimization of industrial robots: Application to a five-bar mechanism. Mechanism and
Machine Theory, 120, 30-56.
[10]Wang, X., Zhang, D., Zhao, C., Zhang, P., Zhang, Y., & Cai, Y. (2019). Optimal design of lightweight serial robots by integrating topology
optimization and parametric system optimization. Mechanism and Machine Theory, 132, 48-65.
[11]Andreassen, E., Clausen, A., Schevenels, M., Lazarov, B. S., & Sigmund, O. (2011). Efficient topology optimization in MATLAB using 88
lines of code. Structural and Multidisciplinary Optimization, 43(1), 1-16.
[12]Kulkarni A, & Javed A (2014) Performance investigation of topologically optimized manipulator link. In Proceedings of International
Conference on Advancements and Futuristic Trends in Mechanical and Materials Engineering, 342-345.
[13]Chalhoub, N. G., & Ulsoy, A. G. (1987). Control of a flexible robot arm: experimental and theoretical results. Journal of Dynamic Systems,
Measurement, and Control, 109(4), 299-309.

You might also like