You are on page 1of 5

Analysis of Tensegrity Rotationally Repetitive Space

Structures Using the Substructuring Method


Aida Mottahedin 1; Jafar Keyvani 2; and Ali Kaveh 3
Downloaded from ascelibrary.org by Dr Vishwanath Karad MIT-World Peace University on 02/09/23. Copyright ASCE. For personal use only; all rights reserved.

Abstract: In this paper, a substructuring method is presented for calculating the displacement of cable domes. This method obtained matrices
as a canonical form in cyclic structures using graph theory. This method helps to avoid the generation of the entire matrices. Then, a solution
for the eigenproblem is presented and nonlinear equations are solved by the Newton–Raphson method. In this paper, the cable domes are
prestressed rotationally repetitive structures that are cyclically symmetric, having cable part and compression strut part. For clarifying the
benefits of the proposed method, both the usual method and the substructuring method are performed for some examples. Results show that
the time and memory are considerably (more than 80%) saved in the substructuring approach. DOI: 10.1061/(ASCE)SC.1943-
5576.0000677. © 2022 American Society of Civil Engineers.
Author keywords: Tensegrity structure-cable dome; Nonlinear analysis; Substructuring method.

Introduction for the eigensolution in symmetric structures can be found in the


work of Kaveh (2004), Kaveh and Salimbahrami (2004), and
A cable dome is a structure built by prestressed cables as tension Kaveh and Sayarinejad (2005). Kaveh and Rahami (2006) studied
elements, and struts as compression members. The prestress level canonical forms for adjacency and Laplacian matrices using graph
was such that all cables were in tension. Usually prestress load was models and suggested methods for decomposing regular structures.
between 5% and 10% of the cable yield stress. In this paper pre- A structure is called regular if its model can be considered as the
stressing is considered as 50% of the yield stress of the cables. product of two graphs. Other researchers (Aghayere 1983; Thomas
Geiger et al. (1986) designed a “cable dome” that was used in 1979; Hasan and Hasan 2002; Williams 1986a, b) also worked
the structures of the gymnastics and fencing arenas for the Seoul on the eigensolution methods for cyclically symmetric structures.
Olympic Games in 1986. One of the largest existing structures of Kaveh and Nemati (2010) presented the eigensolutions for buck-
this type is the Georgia Dome, designed for the Atlanta Olympics ling load and natural frequencies in vibrating systems considering
Games in 1996 (Yuan and Dong 2003). the canonical forms. Optimal analysis occurs when matrices in so-
However, this structure could not be considered as an exact ten- lutions are spares and well-structured to solve the eigenvalue prob-
segrity system. Levy (1989) built the Georgia Dome (completed in lem of the structure (Kaveh 2013).
1992) in quasi-elliptical shape for the Atlanta Olympic Games. In this paper the nonlinear static analysis of cable domes is in-
Some researchers like Rebielak (2000) investigated new shapes of vestigated. For obtaining the eigensolution of stiffness matrix a
cable domes. canonical form is used. With associating graph products and rota-
The elastic stability of a long-span dome is poor (Abedi and tion repeated property, a dome is divided into substructures, and the
Parke 1996; Han and Liu 2002), so nonlinear analysis should be solution is simplified by calculating the eigensolution of its repeat-
used in a cable dome. Kebiche et al. (1999) researched nonlinear ing substructures.
analysis of tensegrity systems structures. Actually, instead of calculating the inverse matrix directly, graph
There are different variations of the geometries in cable-built products are used to write the matrix as the sum of Kronecker prod-
domes, which researchers investigated. For engineers, it is interest- ucts, and many simpler matrices developed for solving the eigen-
ing to know the behavior of the structures with different shapes and problems. Thus, the internal forces and displacements are obtained
the way they affect the weight and cost of the structure. Therefore, without complicated calculations.
the design of the tensegrity structure is also important for its
performance.
Many researchers have discussed a solution to the equilibrium Analysis of Dome Structures
by considering symmetry and regularity. One of the best methods
Tensegrity structures consist of two kinds of structural members
1
Ph.D. Candidate, School of Civil Engineering, Kharazmi Univ., Tehran called cables and bars. They are self-equilibrium stable structures
15719-14911, Iran (corresponding author). ORCID: https://orcid.org/0000 established by interacting compression and tensile components
-0002-6057-8045. Email: aida.mottahedin.9@gmail.com (Pugh 1976; Motro 1992, 2003).
2
Associate Professor, School of Civil Engineering, Kharazmi Univ., Large displacements cause geometrical changes in the struc-
Tehran 15719-14911, Iran. Email: jkeyvani@khu.ac.ir tures, resulting in nonlinearity. Cable domes have large displace-
3
Professor, School of Civil Engineering, Iran Univ. of Science and ments, and the linear theory is not suitable for their analysis. In
Technology, Tehran 13114-16846, Iran. Email: alikaveh@iust.ac.ir
Note. This manuscript was submitted on July 14, 2021; approved on
these structures equations of equilibrium are written for the de-
December 2, 2021; published online on January 31, 2022. Discussion per- formed configuration. For analysis, first initial geometrical equilib-
iod open until June 30, 2022; separate discussions must be submitted for rium is obtained, then Newton–Raphson methods are used for static
individual papers. This paper is part of the Practice Periodical on Struc- analysis. After determination of the boundary condition, the pre-
tural Design and Construction, © ASCE, ISSN 1084-0680. stressed force is applied to the tensegrity structure. For extracting

© ASCE 04022004-1 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2022, 27(2): 04022004


the initial equilibrium, unbalancing of the nodes under prestress
force must be considered. Therefore, the coordinates and the pre-
stress forces must be modified step by step until the entire structure
is balanced.
The formula to determine the initial equilibrium is

K gT u ¼ Fext ð1Þ

where K gT and Fext are the stiffness matrix and external load vector,
Downloaded from ascelibrary.org by Dr Vishwanath Karad MIT-World Peace University on 02/09/23. Copyright ASCE. For personal use only; all rights reserved.

respectively. Stiffness matrix is a summation of elastic stiffness and


geometric stiffness from prestresses

K gT ¼ K gE þ K gG ð2Þ

where

2 3
  1 0 0
E i Ai I 0 −I 0 6 7
kE ¼ ; I0 ¼ 6
40 0 07
5 ð3Þ
I 0i −I 0 I0
0 0 0

and

2 3
  1 0 0
Fi I 3 −I 3 6 7
kG ¼ ; I3 ¼ 6 7
4 0 1 0 5; i ¼ 1; : : : ; b ð4Þ
I i −I 3 I3
0 0 1 Fig. 1. Flowchart of the nonlinear analysis.

In Eq. (4), Fi = pretension load.


After extracting the equilibrium equation using the Newton– 2 3
A B 0 0 ··· 0 0 B1
Raphson method for solving it. In each iteration, the “unbalanced” 6 T 7
section of the external load is estimated for the next increment. As 6B A B 0 ··· 0 0 0 7
6 7
a summary, Fig. 1 shows a flowchart of steps in the nonlinear 6 7
6 0 B1 A B ··· 0 0 0 7
analysis. 6 7
6 .. .. 7
6 0 0 B1 . . 0 0 0 7
6 7
K¼6 . .. .. .. 7 ð5Þ
6 . 7
Dome Analysis with a Canonical Form 6 . . . . A B 0 0 7
6 7
6 0 0 0 0 B1 A B 0 7
6 7
In this section, a canonical form is presented for circular repetitive 6 7
6 0 0 0 0 0 B1 A B7
structures, and the eigensolution via this form is obtained. 4 5
In the classic methods, the entire structures have been used to B 0 0 0 0 0 B1 A
find required matrices (Chen and Feng 2012; Chen et al. 2012). The
proposed method considers some properties of symmetric and
Generating A, B, and BT requires less effort in the substructur-
regular structures to improve eigensolution and analysis, so analy-
ing method.
sis needs only small parts of the structures, and matrices have
After numbering the nodes in substructure, the corresponding
much smaller dimensions to solve large dimension matrices of
elements should be defined. Eq. (6) shows the stiffness matrices
eigensolution.
of the extracted substructure
For derivation of a proper substructure, all boundaries of
segments may not have any joint and three arbitrarily selected seg- 2 3
ments choose. Nodal numbering in this method is important, and in C B 0
6 t 7
each segment the number difference of arbitrarily selected nodes is ½K ¼ 6
4B A B7
5 ð6Þ
constant in each corresponding node in the adjacent segment.
All substructures have the same conditions in analysis under a 0 Bt D
symmetric loading. However, rotation substructure that leads to dis-
placements will not be identical in Cartesian directions. Thus, for The blocks A, B, Bt , C, and D are the submatrices of the sub-
calculating a governing equation, the cylindrical coordinate system structure’s stiffness matrix. For finding the eigenvalues of the stiff-
is more suitable than the Cartesian coordinate system. ness matrix A, B, and Bt are needed.
With proper numbering of the nodes, the global stiffness matrix After generating the blocks A, B, and Bt , eigenproblems are
of the entire structure is formed as shown in Eq. (5) solve by the following solution.

© ASCE 04022004-2 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2022, 27(2): 04022004


Table 1. Roots matrix L
n Real roots Complex roots θ
If n is even 1, −1 cos θ  isinθ θ ¼ 2kπ=n, k ¼ 1; 2; 3; : : : ; ðn − 2Þ=2
If n is odd 1 cos θ  isinθ θ ¼ 2 kπ=n, k ¼ 1; 2; 3; : : : ; ðn − 1Þ=2
Downloaded from ascelibrary.org by Dr Vishwanath Karad MIT-World Peace University on 02/09/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. (a) Plane view of the dome of Example 1; (b) section (Z direction) of the dome of Example 1; and (c) section (Y direction) view of the dome of
Example 1.

Using Kronecker products to rewrite stiffness matrix in Eq. (7) For finding the eigenvector of K using the transformation in
yields the following: Eq. (13), as shown here
K nm×nm ¼ I n×n ⊗ Am×m þ Ln×n ⊗ Bm×m þ Ltn×n ⊗ Btm×m ð7Þ ϕi ¼ Uðej ⊗ Y i Þ ¼ ðX ⊗ IÞðej ⊗ Y i Þ ¼ Xej ⊗ IY i ð13Þ
where L = unsymmetric matrix shown in Eq. (8):
ϕi ¼ X j ⊗ Y i ð14Þ
2 3
0 1 ··· 0
6 .. 7 this method decreases computational time and memory
6. 0 1 7 significantly.
L¼6 .. 7 ð8Þ
40 ··· . 15
1 0 ··· 0 nn
An Illustrative Example
Thus, the eigenvalues of the matrix K can be obtained using the
union of the eigenvalues of n blocks as follows: The following assumptions are for simplifying the solution:
1. Cables are tension-only members;
n n 2. Cables have negligible buckling; and
eigðKÞ ¼ ∪ eigðBLj Þ ¼ ∪ eigðA þ λj ðLÞB þ λ̄j ðLÞBt Þ ð9Þ
j¼1 j¼1 3. Twisting and bending strength and transfer loads are found by
developing direct tension only (Krishnan 2015; Gurfinkel and
where λj = eigenvalue of L generated as follows: Krishnan 2017).
λn − 1 ¼ 0 ð10Þ

Eq. (10) has n real and complex roots shown in Table 1. Table 2. Node displacement with total analysis time
The matrix X in Eq. (11) is the eigenvectors of L: Classic method Present method
Displacement Time Displacement Time
X ¼ ½x1 ; x2 ; : : : ; xn  ð11Þ Node (mm) (s) Node (mm) (s)
the eigenvector Y i is for the jth block calculated as follows: 2 −0.15338136 2 −0.16460926
3 −0.228321712 16.07 3 −0.263196485 1.042
BLj Y j ¼ μi Y i ð12Þ 5 −0.635667521 5 −0.680987944

© ASCE 04022004-3 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2022, 27(2): 04022004


Downloaded from ascelibrary.org by Dr Vishwanath Karad MIT-World Peace University on 02/09/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Example 2 dome and substructure.

Table 3. Material properties used for structural analysis Table 5. Example 3 nodes displacement with total analysis time
Modulus of Ultimate tensile Classic method Present method
Member type elasticity (MPa) strength (MPa) Area (mm2 ) Member Force (kN) Time (s) Member Force (kN) Time (s)
Ridge and hoops 137,895 1,751 2,130 Diagonal1 391.654 Diagonal1 389.268
Diagonal cables 165,474 1,862 1,400 and 2,800 Struts1 −164.128 37.549 Struts1 −144.127 4.43
Hoop cables 165,474 1,862 6,135 Ridge1 181.781 Ridge1 182.814
Struts 200,000 290 9,613 and 11,680 Hoop1 297.14 Hoop1 295.489

Table 4. Example 2 nodes displacement with total analysis time Fig. 3 shows a cable dome with span = 120 m and its substruc-
Classic method Present method ture. This substructure is selected by considering a minimum num-
Member Force (kN) Time (s) Member Force (kN) Time (s)
ber of elements that covers the entire structure property with
rotation. A prestress between 45% and 50% of the yield stress
Diagonal1 187.3 Diagonal1 180.46 of the cable is assumed as an initial value. Members’ properties
Struts1 −559.95 17.98 Struts1 −556.32 2.842 are shown in Table 3. Displacement and time for both methods
Ridge1 356.96 Ridge1 352.68
are presented in Table 4.
Hoop1 297.14 Hoop1 295.48
Fig. 4 shows a cable dome with span = 120 m and its substruc-
ture. A prestress between 45% and 50% of the yield stress of the
cable is assumed as the initial value. Members’ properties are
shown in Table 3. Displacement and time are presented in Table 5.
Assumptions 1 and 2 are satisfied by suitable prestress load. The time ratio in example 3 is 0.158065 and in example 4 is
Fig. 2 shows the cable dome studied by Pellegrino (Kangwai 0.117979 that show how effective this method in time saving.
et al. 1999). The total span is 1,725 mm. The Young’s modulus
of cables and struts are E ¼ 185 × 106 kN=m2 and E ¼ 210 ×
106 kN=m2 , respectively. Conclusion
As shown in Table 2, the results of the present method are near
to those of the classic method, with remarkable time reduction for Prestressed symmetrical cable domes are investigated with the
their calculation. numerical decomposition method. By using suitable subsystems

Fig. 4. Example 3 dome and substructure.

© ASCE 04022004-4 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2022, 27(2): 04022004


from the structures, complicated eigenproblems turned into much Hasan, M. A., and J. A. Hasan. 2002. “Block eigenvalue decomposition
simpler eigenproblems. using nth roots of the identity matrix.” In Proc., 41st IEEE Conf. on
Eigenvalues of a structure with n rotationally repeating segment Decision and Control, 2002. New York: IEEE.
Kangwai, R., S. Guest, and S. Pellegrino. 1999. “An introduction to the
need an nm × nm matrix where m is the number of degrees of
analysis of symmetric structures.” Comput. Struct. 71 (6): 671–688.
freedom in a substructure, in the proposed method eigenvalues https://doi.org/10.1016/S0045-7949(98)00234-X.
of m × m matrix are calculated n times. This saves considerable Kaveh, A. 2004. Structural mechanics: Graph and matrix methods.
time and memory for eigensolution. Somerset, UK: Research Studies Press.
Prestressed levels and cable cross sections are the most effective Kaveh, A. 2013. Optimal structural analysis using the concept of symmetry
items. Prestrain must prevent cable slacking and help to avoid over- and regularity. Berlin: Springer.
Kaveh, A., and F. Nemati. 2010. “Eigensolution of rotationally repetitive
Downloaded from ascelibrary.org by Dr Vishwanath Karad MIT-World Peace University on 02/09/23. Copyright ASCE. For personal use only; all rights reserved.

stressing the struts. In this work, prestress keeps cables in tension.


The nonlinear equilibrium equation of structure and substruc- space structures using a canonical form.” Int. J. Numer. Methods
ture are developed and solved using the Newton–Raphson method. Biomed. Eng. 26 (12): 1781–1796. https://doi.org/10.1002/cnm.1265.
Kaveh, A., and H. Rahami. 2006. “Block diagonalization of adjacency and
The results of the illustrative example show that the analysis in the
Laplacian matrices for graph product; applications in structural mechan-
cable dome is effective, correct, and reliable. ics.” Int. J. Numer. Methods Eng. 68 (1): 33–63. https://doi.org/10.1002
/nme.1696.
Kaveh, A., and B. Salimbahrami. 2004. “Eigensolution of symmetric
Data Availability Statement frames using graph factorization.” Commun. Numer. Methods Eng.
20 (12): 889–910. https://doi.org/10.1002/cnm.711.
All data, models, and code generated or used during the study Kaveh, A., and M. Sayarinejad. 2005. “Eigenvalues of factorable matrices
appear in the published article. with form IV symmetry.” Commun. Numer. Methods Eng. 21 (6): 269–
287. https://doi.org/10.1002/cnm.744.
Kebiche, K., M. Kazi-Aoual, and R. Motro. 1999. “Geometrical non-linear
analysis of tensegrity systems.” Eng. Struct. 21 (9): 864–876. https://doi
References .org/10.1016/S0141-0296(98)00014-5.
Krishnan, S. 2015. Prestressed cable domes: Structural behavior and
Abedi, K., and G. Parke. 1996. “Progressive collapse of single-layer braced design. Champaign, IL: Univ. of Illinois at Urbana-Champaign.
domes.” Int. J. Space Struct. 11 (3): 291–306. https://doi.org/10.1177 Levy, M. 1989. “Hypar-tensegrity dome.” In Proc., Int. Symp. on Sports
/026635119601100302. Architecture, 157–162. London: Thomas Telford.
Aghayere, A. O. 1983. “Structural systems with polar symmetry: Solution Motro, R. 1992. “Tensegrity systems: The state of the art.” Int. J. Space
by quasicirculant matrices.” Doctoral dissertation, Dept. of Civil Engi- Struct. 7 (2): 75–83. https://doi.org/10.1177/026635119200700201.
neering, Massachusetts Institute of Technology. Motro, R. 2003. Tensegrity: Structural systems for the future. Amsterdam,
Chen, Y., and J. Feng. 2012. “Generalized eigenvalue analysis of symmetric Netherlands: Elsevier.
prestressed structures using group theory.” J. Comput. Civ. Eng. 26 (4): Pugh, A. 1976. An introduction to tensegrity. Berkeley, CA: University of
488–497. https://doi.org/10.1061/(ASCE)CP.1943-5487.0000151. California Press.
Chen, Y., J. Feng, L. Zhuang, and S. Xia. 2012. “Elastic stability of sym- Rebielak, J. 2000. “Structural system of cable dome shaped by means of
metric dome structures using group theory.” In Proc., Earth and Space simple form of spatial hoops.” In Lightweight structures in civil engi-
2012: Engineering, Science, Construction, and Operations in Chal- neering, edited by J. B. Obrebski Wydawnictwo Naukowe, 114–115.
Warsaw, Poland: Micro-Publisher.
lenging Environments, 655–663. Reston, VA: ASCE. https://doi.org/10
Thomas, D. L. 1979. “Dynamics of rotationally periodic structures.” Int. J.
.1061/9780784412190.071.
Numer. Methods Eng. 14 (1): 81–102. https://doi.org/10.1002/nme
Geiger, D. H., A. Stefaniuk, and D. Chen. 1986. “The design and construc-
.1620140107.
tion of two cable domes for the Korean Olympics.” In Proc., IASS Symp.
Williams, F. 1986a. “An algorithm for exact eigenvalue calculations for
on Shells, Membranes and Space Frames. Amsterdam, Netherlands: rotationally periodic structures.” Int. J. Numer. Methods Eng. 23 (4):
Elsevier Science Publishers BV. 609–622. https://doi.org/10.1002/nme.1620230407.
Gurfinkel, G., and S. Krishnan. 2017. “Analysis and design of cable-stayed Williams, F. 1986b. “Exact eigenvalue calculations for structures with rota-
steel columns using the stiffness probe method.” Eng. J. Am. Inst. Steel tionally periodic substructures.” Int. J. Numer. Methods Eng. 23 (4):
Constr. 54 (3): 195–210. 695–706. https://doi.org/10.1002/nme.1620230411.
Han, Q., and X. Liu. 2002. “Behavior of a single layer spherical dome with Yuan, X., and S. Dong. 2003. “Integral feasible prestress of cable domes.”
openings and large depth-to-span ratio.” Adv. Struct. Eng. 5 (3): 137– Comput. Struct. 81 (21): 2111–2119. https://doi.org/10.1016/S0045
142. https://doi.org/10.1260/136943302760228086. -7949(03)00254-2.

© ASCE 04022004-5 Pract. Period. Struct. Des. Constr.

Pract. Period. Struct. Des. Constr., 2022, 27(2): 04022004

You might also like