You are on page 1of 8

Proceedings of the IASS Annual Symposium 2017

“Interfaces: architecture.engineering.science”
25 - 28th September, 2017, Hamburg, Germany
Annette Bögle, Manfred Grohmann (eds.)

Study of a novel design for self-standing tensegrity domes


Andrés GONZÁLEZ*, Ani LUOa, Heping LIUa

*Harbin Engineering University


Nantong Street 145, Harbin, 150001 China
andres@hrbeu.edu.cn
a
Harbin Engineering University

Abstract
This paper proposes a novel design for a self-standing tensegrity dome, meaning that it doesn’t require
the use of rigid compression rings or fixed nodes. The structure is designed as the assembly of three
tensegrity units composed of six bars each one, with an initial hemispherical shape, and its equilibrium
matrix A is calculated. By using the force density method, the singular value decomposition SVD of
the equilibrium and force density matrices are executed iteratively until the nodal coordinates and the
force densities of the members correspond to a stable configuration. A numerical example and a real
model are illustrated to demonstrate the efficiency of this method.
Keywords: tensegrity, self-standing dome, lightweight, stability, force density.

1. Introduction
In a simple way, Guest [1] defines tensegrities as structures consisting of cables, struts and bars where
the compression members do not touch each other. This configuration has the characteristic that all the
forces are distributed along the axes of the members, avoiding bending or torques. The properties of
these novel lightweight structures, like being self-standing, the ability of adapting the shape through
control techniques, and a relative low price, make them ideal for the construction of big span
structures.
The use of tensegrity principles for the design of high span domes has been considered for a long time,
since Fuller [2] developed the concept of geodesic tensegrity for the construction of domes and roofs.
His proposals, however, are not tensegrities per the definition, but inspired later designs like the
Olympic Gymnastics and Fencing Arenas in Seoul, South Korea, and the Georgia Dome stadium, in
the USA.
A literature review on the topic brings that most of the dome proposals consist of a cable-strut network
fixed to an outer compression ring, in contradiction with the self-standing capacity of tensegrities.
Pellegrino [3] investigated the characteristics of these domes. Fu [4] also studied this kind of domes
and proposed new different configurations, but all of them considering fixed nodes. Burkhardt [5]
developed a methodology for the design of tensegrity domes by truncating double-layer tensegrity
spheres, but this also requires to consider the base nodes as fixed. One of the few exceptions is the
work of Peña et al. [6, 7] and their design of a tensegrity building. Yet, this design too can be divided
into a compression ring and a roof, both being tensegrity systems though.
In this paper, we propose the design of a multi stage class 1 (Skelton and de Oliveira [8]) tensegrity
dome. By using the force density form finding method, the geometrical parameters to modify the
shape of the 3-stage tensegrity tower into a quasi-spherical figure are determined, and its stability is
analyzed. Finally, a model built in the laboratory is presented to confirm the results.

Copyright © 2017 by <Andrés GONZÁLEZ, Ani LUO, Heping LIU>


Published by the International Association for Shell and Spatial Structures (IASS) with permission.
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

2. Initial formulations
The tensegrity dome problem is divided into two steps: first, the definition of a prototype topology,
nodal coordinates and connectivity matrix, which will be modified by a form finding algorithm until a
stable arrangement is found, and the stability equations that determine this self-standing configuration.

2.1. Geometry
By definition, a dome is a rounded structure. The geometry can vary from a perfect hemisphere to a
fraction of the sphere. To achieve this, it is necessary to create a dependence relation between the
height h of a single unit (or level) and the radius R of the dome. The radius ri of each level must also
be dependent on the radius R of the dome. This condition will assure that the topology will have a
spherical shape. Fig. 1 presents a graphic explanation of this correlation between parameters. The
initial prototype geometry for the tensegrity dome is a half-sphere, although the final geometry will
vary after the optimization procedure.

δ
h2

x
Figure 1: Geometric parameters for half a sphere. Correlation between the radius R of the dome and the height hi
and radius ri of the different levels of the tensegrity dome.
A single unit, composed of b bars, is considered as two b-sided polygons located on parallel planes,
and rotated an angle α, whose vertex are connected by bars and strings. If the angle between nodes in
2
the same plane is   , the bars connect a bottom node with a top node rotated an angle θ+α.
b
y z
β horizontal

diagonal

x
δ saddle
vertical

(a) (b)
Figure 2: Geometric parameters for the assembly of single units. Thick lines represent bars and thin lines,
strings. (a) top view; (b) side view.

2
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

Two kinds of strings are considered: horizontal strings, that connect adjacent strings located on the
same plane, and vertical strings that connect bottom nodes with the corresponding top node rotated an
angle α. For the assembly of units, two more kinds of strings are added. The first one, the saddle
strings, connect the top nodes of a unit with the adjacent bottom strings of the immediate superior unit.
The second type, the diagonal strings, connects a top node of a lower unit with a top node of an upper
unit rotated an angle β.
Two other parameters are considered: the rotation angle β, and the overlap between units δ. These two
parameters, as shown in Figure 1 and Figure 2, represent the interaction between two units. β measures
the rotation between the top nodes of a lower unit and the bottom nodes of the immediately upper unit,
2 
with a value of 0    . The overlap δ has a value of 0    , with ρ as the angular height of
b 2
the units measured from the center of the dome; if the value of δ exceeds this range, will cause the
overlapping of three units. With these parameters, the nodal coordinates [x y z]   nd
 can be
calculated.

2.2. Self-equilibrium equations


The stability problem for tensegrity structures has been extensively studied, using several approaches.
The method used in this paper is the well-known force density; it converts the non-linear system of
equilibrium equations into a linear system by using a force-to-member-length’s ratio. An extensive
explanation can be found in Motro [9].
The topology of a tensegrity structure of d-dimension with m members, can be expressed by the
 
coordinates of the nodes n and its connectivity matrix C  mn . If a member mk connects nodes i
and j  i  j  , the entries i and j of the kth row of C will be -1 or 1, respectively if

1 for p  i

C k , p    1 for p  j (1)
 0 other nodes


If q  m1
 is defined as the vector containing the force densities of the members of the structure,
with the first and external loads and self-weight are ignored, the equilibrium system of a tensegrity
structure can be expressed as
Aq  0 (2)

where A  dnm
 is known as the equilibrium matrix, and is defined by
 CT diag  Cx  
 
A   CT diag  Cy   (3)
 CT diag  Cz  
 
The linear homogeneous system of equilibrium can also be rewritten in terms of the force densities
and nodal coordinates as
Dx y z   0 (4)

3
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

where D  nn
 is known as the force density matrix, and is given by
D  CT diag (q)C (5)

3. Form-finding process
The form-finding procedure is based on the method first proposed by Estrada et al. [10], and then
extended by Tran and Lee [11] by using the singular value decomposition for the analysis of the
equilibrium and force density matrices. In this paper, the method is adapted by introducing a prototype
set of nodal coordinates as an initial value.

3.1. Rank deficiency requirements


Tensegrity structures in a state of self-stress must satisfy two necessary rank conditions (Motro [9]).
The first condition corresponds to the solution of the homogenous system of Eq.(2), and ensures the
existence of at least one state of self-stress if
nA  m  rA  1 (6)

where n A is the dimension of the null space of the equilibrium matrix A, and rA  rank ( A) . The
second rank condition refers to the semi-definite force density matrix D. In the same way as above, the
solutions to the Eq.(4) lay in the null space of D, nD , as in

nD  n  rD  d  1 (7)

where rD  rank (D) and d is the dimension of the structure. This condition brings d-particular
solutions for a vector space basis of a d-dimensional configuration. In this specific case, a nullity of
four is needed for a 3-dimensional tensegrity structure.

3.2. Iterative algorithm


The prototype topology, determined in Section 2, is used as an initial value for calculating A using
Eq.(3). The equilibrium matrix A is then factorized using the SVD as follows
A  UΣWT (8)

where U  dndn
 and W  mm
 are orthogonal matrices, and ∑  dnm
 is a diagonal matrix
containing the non-negative singular values of A. When the null space of the matrix A is equal to zero,
there exists no self-stress state. In this case, all the columns wi of W are checked until a column with
all its entries matching the sign of the entries of q0 is found, i.e. sign(wi) ≡ sign(q0). This column is
taken as the approximated q for the next iterative cycle. q0 is a prototype q with a value of 1 for the
first m-b entries, and a value of -1 for the final b entries.
When Eq.(6) is valid, the nA vectors of W matching in sign with q0 correspond to the states of self-
stress and any of these vectors can be taken as the final force density vector q. This is, in fact, the final
step of the form-finding process, as can be seen in Tran & Lee [11]. The existence of at least one state
of self-stress, if the rank deficiency condition for D is also valid, stops the iterative process and the
corresponding values of [x y z] and q are taken as the final equilibrium conditions.
The value of the force density vector q is next input into Eq.(5) to calculate D. Then, the force density
matrix D is factorized using the SVD in the following way
D  GΗJT (9)

4
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

where G  nn
 
and J  nn
 are orthogonal matrices, and H   nn
 is a diagonal matrix
containing the non-negative singular values of D. The last nD columns of J are taken as potential nodal
coordinates and arranged in a matrix J

x y z   J   jn-n*D +1 jn-n*D +2 jn  (10)

Next, the projected lengths of J along the n directions are calculated and arranged in a projected
length matrix L,

L  CJ   Cjn-n*D +1
   Cj n-n*D +2   Cjn  (11)

in order to identify and remove the vector ji from J that gives members of length zero
Cji  0 (12)

Finally, the remaining d vectors of J are taken as the [x y z] nodal coordinates and input in Eq. (3) to
calculate A for the next iterative cycle. These vectors, along with the force density vector q, will be
gradually modified until the rank deficiency conditions of both the force density matrix and the
equilibrium matrix are satisfied.

4. Numerical example
We demonstrate the feasibility of the procedure by presenting a numerical example of a tensegrity
dome. Table 1 contains the initial values for the computation of the prototype topology; the radius is
for reference only, as the dimension of the dome will be shrunk and should be scaled up again. The
rest of the dimensions correspond to arbitrary values (Zolesi et al.[12]).

Table 1: Initial values for the prototype topology of the dome

Radius R (mm) 3000


Unit height h (deg) 30º
Rotation angle ∝ (deg) 13º
Angle between units β (deg) 30º
Overlap between units δ (deg) 0,45×h

The form finding process carries out iterations until the rank deficiencies of the force density matrix D
and the equilibrium matrix A are met. It can be seen from Figure 3 that the initial semi-spherical
geometry varies through the iterations, but the overall shape is maintained. The final topology is
obtained after 14 iterations.0.3
0.3

0.2
0.2

0.1
0.1

0 0

-0.1 -0.1

-0.2 -0.2

-0.3 -0.3

(a) (b) (c)


-0.4 -0.4
-0.4 -0.3 -0.2 -0.1 0 0.1 -0.4 0.2 -0.3 0.3 -0.2 -0.1 0 0.1 0.2 0.3

Figure 3: Side view of the dome after: (a) zero iterations; (b) 9 iterations; (c) 14 iterations.

5
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

After introducing a scaling factor to adjust the base radius to 3000 mm, and displacing the Z-
coordinates to the positive axis, the final nodal coordinates are obtained, as shown in Table 3. It can be
noticed that the structure is still symmetrical around the Z-axis, but the final height is 1.45 times
higher than its radius, making it not hemispherical anymore.

Table 3: Nodal coordinates of the tensegrity structure

Node X Y Z Node X Y Z
1 -1749.0 -2437.5 0.0 19 -1904.7 -778.4 4357.8
2 1236.5 -2733.4 0.0 20 -278.3 -2038.7 4357.8
3 2985.5 -295.9 0.0 21 1626.4 -1260.3 4357.8
4 1749.0 2437.5 0.0 22 1904.7 778.4 4357.8
5 -1236.4 2733.4 0.0 23 278.2 2038.7 4357.8
6 -2985.4 295.9 0.0 24 -1626.4 1260.3 4357.8
7 1572.4 -2057.5 3003.8 25 -2493.1 251.9 3265.8
8 2568.1 333.0 3003.8 26 -1464.6 -2033.1 3265.8
9 995.6 2390.5 3003.8 27 1028.4 -2285.0 3265.8
10 -1572.4 2057.5 3003.8 28 2493.1 -251.9 3265.8
11 -2568.1 -333.0 3003.8 29 1464.6 2033.1 3265.8
12 -995.6 -2390.5 3003.8 30 -1028.4 2285.0 3265.8
13 266.6 -2857.1 1704.0 31 -525.8 -1251.8 5360.1
14 2607.7 -1197.7 1704.0 32 821.2 -1081.3 5360.1
15 2341.0 1659.5 1704.0 33 1347.0 170.5 5360.1
16 -266.6 2857.1 1703.9 34 525.8 1251.8 5360.1
17 -2607.7 1197.7 1703.9 35 -821.2 1081.3 5360.1
18 -2341.0 -1659.5 1703.9 36 -1347.0 -170.5 5360.1

The force density vector q is obtained from W, and its values are checked to confirm that the forces
corresponding to each member have the correct sign. After grouping the values per the symmetry,
Table 4 shows that all the strings are in tension and only the bars are in compression, as defined in q0.

Table 4: Force densities

0.02724
0.02328
0.09232
0.12494
0.09889
0.04232
0.06705
0.09870
0.14854
0.11530
0.03074
0.18298
0.09209
-0.07564
-0.13213
-0.11831

Finally, a scale model was built in the laboratory to confirm the results. The structure is made of
wooden bars and nylon strings. Figure 4 shows that the model obtained in this procedure is self-
standing and stable.

6
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

Figure 4: Scale model of the proposed tensegrity dome.

5. Conclusion
A proposal for a self-standing tensegrity dome has been presented. The geometrical parameters for a
prototype set of nodal coordinates have been determined and the form finding process for obtaining a
self-standing structure has been introduced. It is proved, through the analysis of the rank deficiency of
the force density and equilibrium matrices, that the structure is stable.
Despite its versatility, the force density method presents limitations for determining complex
tensegrity systems. However, in this paper it is shown that it is possible to obtain new structures by
modifying configurations whose stability is already known.
In this paper, only the topological characteristics and the form finding procedure for obtaining a
structure in equilibrium have been presented; further work on the topic should include vibration
analysis and response to external forces.

Acknowledgements
This paper is funded by the International Exchange Program of Harbin Engineering University for
Innovation-Oriented Talents Cultivation and National Natural Science Foundation of China (NSFC)
51605111, 51675114.

References
[1] S. D. Guest, "The stiffness of tensegrity structures," 2011, pp. 57-66.

7
Proceedings of the IASS Annual Symposium 2017
Interfaces: architecture.engineering.science

[2] B. Fuller, "Tensegrity," Portfolio and Art News Annual, vol. 4, pp. 112-127, 1961.
[3] S. Pellegrino, "A class of tensegrity domes," International Journal of Space Structures, vol. 7,
pp. 127-142, 1992.
[4] F. Fu, "Structural behavior and design methods of Tensegrity domes," Journal of
Constructional Steel Research, vol. 61, pp. 23-35, 1// 2005.
[5] R. W. Burkhardt, "A practical guide to tensegrity design," C. U. Press, Ed., ed, 2005.
[6] D. Pena, J. Ignasi Llorens, R. Sastre, D. Crespo, and J. Martinez, "TENSEGRITY RING FOR
A SPORTS ARENA FORMFINDING & TESTING," Textile Composites and Inflatable
Structures V (Structural Membranes 2011), pp. 244-251, 2011 2011.
[7] D. Pena, I. Llorens, and R. Sastre, "Formfinding and Structural Analysis of a Tensegrity
Dome," in Tensinet Symposium 2010 - Tensile Architecture: Connecting Past and Future,
2010, pp. 147-148.
[8] R. E. Skelton and M. C. de Oliveira, Tensegrity Systems. New York: Springer-Verlag, 2009.
[9] R. Motro, Tensegrity: Structural Systems for the Future, 1 ed. Oxford: Butterworth-
Heinemann, 2003.
[10] G. G. Estrada, H. J. Bungartz, and C. Mohrdieck, "Numerical form-finding of tensegrity
structures," International Journal of Solids and Structures, vol. 43, pp. 6855-6868, 11// 2006.
[11] H. C. Tran and J. Lee, "Form-finding of tensegrity structures using double singular value
decomposition," Engineering with Computers, vol. 29, pp. 71-86, 2013.
[12] V. S. Zolesi, P. L. Ganga, L. Scolamiero, A. Micheletti, P. Podio-Guidugli, G. Tibert, et al.,
"On an innovative deployment concept for large space structures," in 42nd International
Conference on Environmental Systems 2012, ICES 2012, July 15, 2012 - July 19, 2012,
United states, 2012, p. American Institute for Aeronautics and Astronautics (AIAA).

You might also like