International Journal of Solids and Structures 40 (2003) 63476367

www.elsevier.com/locate/ijsolstr

Equilibrium conditions of a tensegrity structure

Darrell Williamson a, Robert E. Skelton

b,*

, Jeongheon Han

b,*

a
b

The Faculty of Informatics, University of Wollongong, Wollongong, NSW 2522, Australia

Department of Mechanical and Aerospace Engineering, University of California, San Diego,
9500 Gilman Dr., La Jolla, CA 92093-0411, USA
Received 13 March 2003; received in revised form 7 July 2003

Abstract
This paper characterizes the necessary and sucient conditions for tensegrity equilibria. Static models of tensegrity
structures are reduced to linear algebra problems, after rst characterizing the problem in a vector space where direction
cosines are not needed. This is possible by describing the components of all member vectors. While our approach
enlarges (by a factor of 3) the vector space required to describe the problem, the advantage of enlarging the vector space
makes the mathematical structure of the problem amenable to linear algebra treatment. Using the linear algebraic
techniques, many variables are eliminated from the nal existence equations.

2003 Elsevier Ltd. All rights reserved.
Keywords: Tensegrity; Structure; Statics; Equilibrium conditions; Linear algebra

1. Introduction
The tensegrity structures introduced by Snelson (1996) pose a wonderful blend of geometry and mechanics. In addition, they have engineering appeal in problems requiring large changes is structural shape.
Tensegrity structures exist as a prestressed stable connection of bars and strings. Most existing smart
structure methods are limited to small displacements, but the control of tensegrity structures allows very
large shape changes to occur (Skelton and Sultan, 1997; Skelton et al., 2001a; Motro, 1992). Therefore, an
ecient set of analytical tools could be the enabler to a host of new engineering concepts for deployable and
shape controllable structures. This paper characterizes the static equilibria of tensegrity structures in terms
of vectors which describe the elements (bars and strings), thereby eliminating the need to use direction
cosines and the subsequent transcendental functions that follow their use. For a comparison of previous
methods of form-nding in tensegrity structures, see Tibert and Pellegrino (2001), Vassart and Motro
(1999), Motro et al. (1994), Linkwitz (1999), Barnes (1998), and Schek (1974).

Corresponding authors. Tel.: +1-858-822-1054/534-6146; fax: +1-858-822-3107.

E-mail addresses: darrell_williamson@uow.edu.au (D. Williamson), bobskelton@ucsd.edu (R.E. Skelton), jehan@ucsd.edu
(J. Han).
0020-7683/$ - see front matter
2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0020-7683(03)00400-1

6348

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

It is well known in a variety of mathematical problems that enlarging the domain in which the problem
is posed can often simplify the mathematical treatment. In fact, many nonlinear problems admit solutions by
linear techniques by enlarging the domain of the problem. For example, nonlinear Riccati equations are
known to be solvable by linear algebra in a space that is twice the size of the original problem statement. The
purpose of this paper is to show that the mathematical structure of the equations admits some treatments by
linear algebra methods by enlarging the vector space in which the tensegrity statics problem is characterized.
Our results characterize the equilibria conditions of tensegrity structures in terms of a very small number
of variables since the necessary and sucient conditions of the linear algebra treatment allow the elimination of several of the original variables. These results can be used for ecient algorithms to design and
simulate a large class of tensegrity structures. Tensegrity concepts have been around for 50 years without
ecient design procedures (Kenner, 1976; Pugh, 1976; Connelly, 1982, 1993, 1999; Ingber, 1993, 1997,
1998; Williamson and Skelton, 1998a,b; Motro, 1984, 1990, 2001; Skelton et al., 2001b).
The paper is organized as follows. After the review of mathematical preliminaries in Section 2, Section 3
introduces the network representations of tensegrity structures as an oriented graph in real three dimensional
space. Geometric connectivity, equilibrium, and a coordinate transformation will be introduced. Section 4
introduces the algebraic equilibrium conditions. After we derive necessary and sucient conditions for the
existence of an unloaded tensegrity structure in equilibrium, we write the necessary and sucient conditions
for the externally loaded structure in equilibrium. A couple of examples will illustrate the results.
2. Algebraic preliminaries
We let In dene the n
n identity matrix, and 0 dene an n
m matrix of zeros. (The dimensions of 0 will
be clear from the context.) We also let qA dene the rank of the matrix A. Let A 2 Rm
n and B 2 Rp
q ,
then the Kronecker product (Horn and Johnson, 1985) of A and B is dened as
2
3
A1; 1B A1; 2B    A1; nB
6 A2; 1B A2; 2B    A2; nB 7
6
7
mp
nq
AB6
72R
..
..
..
4
5
.
.
.
Am; 1B

Am; 2B



Am; nB

where Ai; j is the i; j element of a matrix A. Then we have the following result.
Lemma 1. The following statements are true.
i(i) Suppose A 2 Rn
r , B 2 Rn
r , C 2 Rr
p . Then
A  Im T AT  Im
A  Im B  Im A B  Im
A  Im C  Im AC  Im
rankA  Im m
rankA:
(ii) Suppose A 2 Rn
n has eigenvalues fk1 ; k2 ; . . . ; kn g. Then A  Im also has eigenvalues fk1 ; k2 ; . . . ; kn g
where each eigenvalue is repeated m times.
The derivations in this paper rely heavily on the singular value decomposition svdA fUA ; RA ; VA g of a
matrix A as expressed in the following result (Horn and Johnson, 1985).

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6349

Lemma 2
ii(i) Suppose an n
m matrix A has rank rA , then there exists an n
n unitary matrix UA , an m
m unitary
matrix VA and a positive denite rA
rA diagonal matrix R1A such that


R1A 0
T
A UA RA VA ; RA
:
1
0 0
i(ii) If fUA ; VA g are partitioned such that
UA U1A ; U2A ;

V V1A ; V2A 

with
U2A 2 Rn
n rA ;

U1A 2 Rn
rA ;

V1A 2 Rm
rA ;

V2A 2 Rm
m rA

then
UT1A  U1A IrA ;

UT1A  U2A 0;

UT2A  U2A In rA

VT1A  V1A IrA ;

VT1A  V2A 0;

VT2A  V2A Im rA

UT2A A

0;

AV2A 0:

(iii) The algebraic equation

Ax y
has a solution if and only if UT2A y 0. When this condition is satised, then all solutions x are of the form
T
x V1A R 1
1A U1A y V2A zx

where zx 2 Rn rA is arbitrary.
(iv) Suppose A 2 Rn
m and
~ A  Ip ;
A

svdA fUA ; RA ; VA g:

~ 2 Rpn
pm and
Then A
~ fUA  Ip ; RA  Ip ; VA  Ip g:
svdA
i(v) Suppose A 2 Rn
m and
T
~
b ~
bT1 ; ~
bT2 ; . . . ; ~
bTn  ;

~
bk bk1

bk2

...

bkp  2 Rp :

Then the algebraic equation

A  Ip ~
x~
b

has a solution if and only if the equations

Ax b ;

166p

have solutions fx 2 Rm ; 1 6 6 pg where

b b1

b2

...

bn T :

x2

...

xm  ;

If
x x1

166p

6350

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

are solutions of (7), then

h
iT
~T2 ; . . . ; x
~Tm ; x
~j xj1
~ x
~T1 ; x
x

xj2

...

xjp T 2 Rp

10

is a solution of (6).
In particular, (6) has a solution if and only if UT2A  Ip ~b 0. When this condition is satised, then all so~ are of the form
lutions x
T
~
~ V1A R 1
zx
x
1A U1A  Ip b V2A  Ip ~

where ~zx 2 Rpn rA is arbitrary.

3. Network representation of structures
A tensegrity structure consists of a connection of tensile components (or strings) and compressive
components (or bars) in which the tension and compressive forces are directed along the strings and bars.
Consequently, in this paper, the equilibrium conditions are completely specied in terms of translational
forces.
We represent a tensegrity structure as an oriented graph (Desoer and Kuh, 1969) in real three dimensional space R3 dened in terms of nodes and directed branches which are all represented as vectors in R3 .
A loop is any closed path in the graph. As we shall see, the advantage of this approach is that both the
magnitude and the direction of the forces are contained in vectors which can be solved using linear algebra.
Thus linear algebra plays a larger role in this approach compared to the usual approach in mechanics and
nite element methods using direction cosines.
In particular, suppose there are nb bars and ns strings for which bar contacts can only occur at the bar
ends. Then in the oriented graph of a tensegrity structure, the np nodes consist of the ends of the bars as
represented by the np vectors fpk g, and the nb ns directed branches consist of the ns string branches (or
vectors) fsn g and the nb bar branches (or vectors) fbm g. If there are nb0 6 nb bars which are not in contact
with any other bar, then
np 2nb0 n~p ;

n~p 6 2nb nb0 :

For example, if no two bars are in contact, then n~p 0 and np 2nb . Or, if nb nb0 bars all contact at a
single bar end, then n~p nb nb0 1.
Thus given a tensegrity structure consisting of np nodes, nb bars and ns strings, the positions of the nodes
are described by the np vectors fp1 ; p2 ; . . . ; pnp g, the positions of the bars are described by the nb vectors
fb1 ; b2 ; . . . ; bnb g, and the positions of the strings are described by the ns vectors fs1 ; s2 ; . . . ; sns g.
Denition 3. The geometry of the tensegrity structure is dened by the tensegrity node vector p 2 R3np , the
tensegrity bar vector b 2 R3nb , and the tensegrity string vector s 2 R3ns where
pT pT1 ; pT2 ; . . . ; pTnp ;

bT bT1 ; bT2 ; . . . ; bTnb ;

sT sT1 ; sT2 ; . . . ; sTns :

11

Denition 4. A class k tensegrity structure connects only k compressive members to a node.

Geometric connectivity. Each directed branch can undergo a displacement in reaching its equilibrium
state. String and bar vectors can change both their length and orientation. Node vectors can change both
their length and orientation but subject to a Law of Geometric Connectivity which we state as follows:
The vector sum of all branch vectors in any loop is zero:
These equations are in the form of a set of linear algebraic equations in the branch vectors.

12

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6351

Force equilibrium. In our study of tensegrity structures, we are concerned with structures in which bars
sustain compressive forces and strings do not. We therefore choose to distinguish between the string (or
tensile) forces ftn g and the bar (or compressive) forces ff m g which are dened in terms of the string and bar
vectors respectively as follows.
Denition 5. Given the tensile force tn in the string characterized by the string vector sn and the compressive
force f n in the bar characterized by the bar vector bn , the tensile force coecient cn > 0 and the compressive
force coecient km > 0 are dened by
tn cn sn ;

f m km bm :

13

The forces of the tensegrity structure are dened by the external force vector w 2 R
vector f 2 R3nb , and the tension vector t 2 R3ns where
h
i




wT wT1 ; wT2 ; . . . ; wTnw ; f T f T1 ; f T2 ; . . . ; f Tnb ; tT tT1 ; tT2 ; . . . ; tTns :

3nw

, the compression
14

It follows from (14) that (13) can be expressed in the form

t C  I3 s;

f K  I3 b

15

where
C diagfc1 ; c2 ; . . . ; cns g;

K diagfk1 ; k2 ; . . . ; knb g:

16

The diagonal matrices fC; Kg shall be referred to as the tensile force coecient matrix and compressive
coecient force matrix, respectively.
Force convention. Suppose each node pk is subjected to compressive vector forces ff mk g, tensile vector
forces ftnk g and external force wk . Then the Law for Static Equilibrium may be stated as follows:
X
X
tnk
f mk wk 0
17
n

where a positive sign is assigned to a (tensile, compressive and external) force vector leaving a node, and a
negative sign is assigned to a (tensile, compressive and external) force vector entering a node. The negative
sign in (17) is a consequence of the fact that we choose to dene positive force coecients kn and cn .
From the network, it follows that components of the string vector s and the bar vector b in (11) can be
written as a linear combination of components of the node vector p. Also, if branch k is a bar which leaves
node i and enters node j, then bk pj pi , whereas if branch k is a string which leaves node i and enters
e T p sT ; bT T where the matrix C
e consists only of block matrices
node j, then sk pj pi . Hence we have C
of the form f0; I3 g. In particular, if we consecutively number the ns nb branches fs1 ; s2 ; . . . ;
e C
e ij  is dened
sns ; b1 ; b2 ; . . . ; bnb g as f1; 2; . . . ; ns ; ns 1; . . . ; ns nb g, then the 3np
3ns 3nb matrix C
by
8
if force vector j enters node i
< I3
e ij I3 if force vector j leaves node i
18
C
:
0
if force vector j is not incident with node i:
e has exactly one block I3 and one block I3 with all other column blocks 0, and
Also (i) each column of C
e ij I3 . Specically, the bar connectivity matrix B
e
(ii) for any row i there exists a column j such that C
e form the matrix C
e as follows:
and the string connectivity matrix S
   T
e
s
eT
19
S
eT p C p
b
B

6352

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

where
e 2 R3np
3ns ;
S

e 2 R3np
3nb :
B

20

e D
e ij  to be the external force incidence matrix dened by
Similarly, we dene D

e ij I3 if external force vector j enters node i
D
0 if external force vector j is not incident with node i:

21

From network analysis, the law for static equilibrium and linear algebra, we have the following result.
Lemma 6. Consider a tensegrity structure as described by the geometric conditions given by (19). Then the
equilibrium force equations for a tensegrity structure under the external load w are
2
3
t


~ 4 f 5 0; A
~, S
e B
e D
e
A
22
w
or equivalently
et B
ef D
ew
S

23

where
e S  I3 ;
S

e B  I3 ;
B

e D  I3
D

24

for some S 2 Rnp
ns , B 2 Rnp
nb , and D 2 Rnp
nw .

We shall refer to fS; B; Dg as the string connectivity matrix, the bar connectivity matrix, and load incidence matrix respectively. These incidence matrices are binary matrices whose components are f 1; 0; 1g.
The computational signicance of this fact is that roundo errors in digital computers are avoided, dramatically increasing the size of problems that can be solved accurately on a digital computer.
~ is known as the incidence matrix. This matrix is not the reduced inIn network analysis, the matrix A
cidence matrix since we have included the datum node which means that one block row of equations in (22)
is dependent on the other rows. This fact does not cause any diculties in subsequent developments. On the
contrary, some symmetry is preserved in the algebraic equations.
Lemma 7. Consider a tensegrity structure consisting of nb P 1 bars and ns P 1 strings as dened by the
connectivity matrices fS; Bg in (24). Then
qB , rankB 6 nb 6 np 1:
In particular, if there are exactly r independent loops of only bar vectors, then qB nb r.
Proof. Each loop formed from bars connected end to end results in a linear relationship between the
corresponding bar vectors. There are at most np 1 connections between any np nodes in order that there
be no loops. Hence, we have the required result. h
Example 8. Consider the 2-bar 4-string class 1 planar tensegrity structure illustrated in Fig. 1 with tensile
force vectors ft1 ; t2 ; t3 ; t4 g and compressive force vectors ff 1 ; f 2 g. This structure has nb 2 bars, np 4
nodes and ns 4 strings. The Geometric Connectivity conditions (19) can be written in the form

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

s3

s1

6353

w1

b2
s2

s4

b2

b1

s4

s3

s2

b1
s2

s4

s1

w2

s1

Fig. 1. 2-Bar 4-string class 1 tensegrity.

p1 p3 s 1 ;
p2 p4 s 3 ;

p2 p3 s 2
p1 p4 s 4

p1 p2 b1 ;

p3 p4 b2 :

In (19) and (24), the connectivity matrices are given by

2
3
2
3
1 0
0
1
1 0
6
6 0
1 1 0 7
0 7
7; B 6 1
7
S6
4 1 1 0
4 0 1 5:
0 5
0
1
0
0
1 1

25

26

Here nb rankB 2 < np 4. Also, in terms of the stated force convention (17), the conditions for static
equilibrium at nodes 14 are
9
t1 t4 f 1 0
>
>
=
t2 t3 f 1 w1 0
:
27
t1 t2 f 2 w2 0 >
>
;
t3 t4 f 2 0
These static equilibrium conditions and the geometric conditions can be written in the form (19)(24), where


3
2
0
1 0
0
1  1 0  0
h   i
6
0  1 0 7
1 1 0  1
~ S
e  B
e  D
e  I3 6 0
7  I3 :
A

4 1 1 0
0  0 1  0 1 5
1  0
0
0
0
1 1  0
Example 9. Consider the 4-bar 8-string planar class 2 tensegrity structure illustrated in Fig. 2. Now the bar
connectivity B and the string connectivity S in (24) are given by

6354

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

Fig. 2. 4-Bar 8-string class 2 tensegrity.

1 0
6 0
1
6
6 1 1
S6
6 0
0
6
4 0
0
0
0

0
1
1 0
0
0
1 1
0
0
0
0

0
0
0
0
1 0
0
0
0
1
1 1

3
0
1 7
7
0 7
7;
0 7
7
1 5
0

1
6 1
6
6 0
B6
6 0
6
4 0
0

0
0
1
1
0
0

0
0
1
0
1
0

3
0
1 7
7
0 7
7:
0 7
7
0 5
1

Here
rankB 4 nb < np 6:
Example 10
(a) A 3-dimensional class 1 tensegrity consisting of
strings is illustrated in Fig. 3. The corresponding
2
1 0
0
0
1
0 1 0
6 0 1 0
0
0
1
0
1
6
6 1
0 1 0
0
0
0 1
6
S6
1
0 1 0
0
0
0
6 0
4 0
0
1
0 1 0
0
0
0
0
0
1
0 1 1
0

one stage of nb 3 bars, np 6 nodes and ns 9

connectivity matrices fS; Bg are given by
3
2
3
1 0
0
0
6 1
0
0 7
0 7
7
6
7
6 0 1 0 7
0 7
7
7; B 6
6 0
1
0 7
1 7
7
6
7
4 0
0 1 5
1 5
0
0
1
0

where rankB 3 nb , np 2nb .

(b) A class 2 tensegrity structure may be derived from the structure in part (a) by replacing the strings
fs1 ; s3 ; s5 g by bars fb4 ; b5 ; b6 g where the nodes fp1 ; p3 ; p5 g now become ball joints. In this new structure,
T
T
np nb ns 6. With s sT2 sT4 sT6 sT7 sT8 sT9  , b bT1 bT2 bT3 bT4 bT5 bT6  , we now have that
3
2
3
2
1 0
0 1 0
1
0
0
0 1 0
0
6 1
6 1 0
0
0
0
0
0 7
1
0
1
0 7
7
6
7
6
6
7
6 0
1 1 0 7
0
0
0 1 0 7
7:
6 0 1 0
;
B

S6
6 0
6 1 1 0
1
0
0
0
0 7
0
0
1 7
7
6
7
6
4 0
4 0
0 1 0
1 1 5
0
0
0
0 1 5
0
0
1
0
0
0
0
1 1 1
0
0

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6355

Fig. 3. 3-Bar 9-string class 1 tensegrity.

Here rankB 5 nb 1 < np 6. Note that since there is exactly one loop fb4 ; b5 ; b6 g of bars, it
follows form Lemma 7, rankB nb 1.
(c) Another class 2 tensegrity structure may be derived from the structure in part (b) by replacing the string
s9 by a bar b7 . In this new structure, nb 7; np 6 and ns 5. Since there are now two independent
loops fb4 ; b5 ; b6 g and fb2 ; b7 ; b5 g of bars, we have from Lemma 7 that rankB nb 2 5 < np 6.
We have the following result.
Theorem 11. Suppose the np
nb connectivity matrix B of rank qB has the singular value decomposition



RB1 0 VTB1
B UB1 UB2 
28
; q B < nb
0 0 VTB2
where UB1 2 Rnp
qB , VB1 2 Rnb
qB or


R
B UB1 UB2  B1 VTB ; qB nb
0

29

where UB1 2 Rnp
nb , VB 2 Rnb
nb . Also, given the tensegrity node vector p, dene the coordinate transformation
e Tq
pP
30
where
e P  I3 ;
P

PT PT1 ; PT2 

qB
np
T
;
P1 R 1
B1 UB1 2 R

P2 UTB2 2 Rnp qB
np :

31

Then
i(i) In terms of the transformed tensegrity node vector q, the tensegrity geometry is given by

T
e Tq S
e T q ; q VT  I3 b
q qTd qTe ; s S
d
1 d
2 e
B1

32

6356

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

where qd 2 R3qB , qe 2 R3np qB and

e 2 S2  I3
S

e 1 S1  I3 ;
S
where

S1 P1 S 2 RqB
ns ;

S2 P2 S 2 Rnp qB
ns :

33

(ii) The tensegrity force equilibrium is given by

S1  I3 t VTB1  I3 f D1  I3 w
S2  I3 t D2  I3 w

34

D1 P1 D;

35

D2 P2 D:

Proof. By Lemma 7, qB < np which means fUB1 ; UB2 g are well dened. Part (i) follows directly from the
e T q s, B
e with S
e T q b with
denition of P
q
q
e T; S
e T ;
e T S
S
q
1
2

e T VB1  I3 ; 0
B
q

36

e1 P
e2 P
e; S
e2 P
e g and f P
e1S
e2S
e2B
e VT  I3 ; S
e 0g. Part (ii) follows from the expansion of
e1B
with f S
B1
e
e
e
the transformed equilibrium force equation S q t B q f D q w, where
h
i


e S
eq B
eq D
eq P
e B
e D
e :
S
For notational simplicity, we assume in subsequent expressions that VB1 VB whenever qB nb . h
3.1. Class 1 structures
In a class 1 tensegrity structure, no two bars are connected, and so np 2nb . Without any loss of
generality, we can then label the nodes of the bar bm to be p2m and p2m 1 . Hence for class 1 tensegrity
structures, we have
bm p2m 1 p2m ;

m 1; 2; . . . ; nb :

37

Lemma 12. Given the tensegrity node vector p with (37), bar vector b, and string vector s, the geometry of the
class 1 tensegrity structure can be described by the algebraic equations
e T p b;
B

e Tp s
S

38

for some reduced np
ns connectivity matrix S. The reduced np
nb connectivity matrix B is given by
B Ie Io

39

where odd and even node selection matrix Io , Ie 2 R3np
3np are dened by
ITo , blockdiagf1; 0; 1; 0; . . . ; 1; 0; 1; 0g
ITe , blockdiagf0; 1; 0; 1 . . . ; 0; 1; 0; 1g:

40

Then in (29)
np 2nb ;

qB np ;

VB I:

41

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6357

The following two corollaries give two special choices for transformations.
Corollary 13. From Theorem 11
ii(i) Let P1 , P2 2 Rnb
np in (31) be given by


PT PT1 PT2 12 B J 

42

and the inverse transformation is

P 1 2PT B

43
nb
np

where B is given by (39) and J 2 R

are dened by

J , Ie Io :

44

i(ii) The transformed coordinate q is given by

qT qTd ; qTe ;

qd b

45

where b is the bar vector and qe 2 R

qTe

3nb

is the vector of the mass center of each of the bars given by

pTc1 ; pTc2 ; . . . ; pTcn 

b

where pcj ,

1
p2j
2

46

p2j 1 .

Corollary 14. From Theorem 11

i(i) Let P1 , P2 2 Rnb
np in (31) be given by
PT Io

47

and the inverse transformation is

P 1 B

Ie 

48

where B and J are given by (39) and (44) respectively, and odd and even selection matrix Io , Ie 2 R
are given by (40).
(ii) The transformed coordinate q is given by
qT qTd ; qTe ;

qd b

where b is the bar vector and qe 2 R

qTe

3np
3np

49
3nb

is a vector of the even nodes given by

pT2 ; pT4 ; . . . ; pTnp :

50

Example 15. Eqs. (25), (27) of the 2-bar 4-string tensegrity introduced in Example 8 can be written in the
form (22) and (19) where b in (11) is already in the form (37). Then from Corollary 14, we have




1 0 0 1
1 1 1 1
S1
; S2
1 1 0 0
1 1 1 1
51




1 0 0 0
1 1 0 0
P1
; P2
0 0 1 0
0 0 1 1
or from Corollary

1 1
P1
2 0

1 1
S1
2 1

13, we have



1 1 1 0 0
1 0 0
; P2
0 1 1
2 0 0 1 1



1 1 1 1 1
1 1 1
; S2
:
1 1 1
2 1 1 1 1

6358

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

4. Analysis of the transformed equilibrium conditions for a tensegrity structure

Denition 16. A tensegrity structure with tensile force coecients fcm > 0g, compressive force coecients
fkn > 0g, node vector p, string vector s and bar vector b is said to be in equilibrium if the element relationships (13), the force equations (23) and the geometric equations (19) are all satised.
For the remainder of this paper, we choose to use the coordinate transformation derived in Theorem 11.
Requirements for equilibrium. Given an external force vector w, the problem of determining the geometric
and force conguration of a tensegrity structure consisting of ns strings and nb bars in equilibrium is
therefore equivalent to nding a solution qd 2 R3qB , qe 2 R3np qB of the equations
s ST1  I3 qd ST2  I3 qe

52

t C  I3 s;

53

C , diagfc1 ; c2 ; . . . ; cns g

S2  I3 t D2  I3 w

54

VTB1  I3 f S1  I3 t D1  I3 w;
f K  I3 b;

VTB1 VB1 I

K , diagfk1 ; k2 ; . . . ; knb g

55
56

for given matrices

S1 2 RqB
ns ; S2 2 Rnp qB
ns ;

D1 2 RqB
nw ; D2 2 Rnp qB
nw :

57

Beyond equilibrium requirements, one might require shape constraints by requiring p to take on a specic
e T q. However, in this paper, our focus is only to characterize possible
set of values p 
p, where p P
equilibria, and so the freedom in choosing the nodal vector p will appear as free variables in the vector qe , as
the sequel shows.
As a result of Lemma 2, conditions for the existence of solutions fqd 2 R3qB ; qe 2 R3np qB g of (52)(56)
are equivalent to conditions for the existence of solutions fq1 2 RqB ; q2 2 Rnp qB g of the equations:
sr ST1 q1 ST2 q2

58

tr Csr ;

59

C , diagfc1 ; c2 ; . . . ; cns g

S2 tr D2 wr
VTB1 f r S1 tr D1 wr ;
f r Kbr ;

60
VTB1 VB1 I

K , diagfk1 ; k2 ; . . . ; knb g:

61
62

4.1. Prestressed equilibrium structure

We now proceed to derive necessary and sucient conditions for the existence of a structure in equilibrium that is prestressed in the absence of any external load; that is, w 0 in (52)(56), or equivalently,
wr 0 in (58)(62). Our strategy for the examination of the conditions (58)(62) is as follows. The solution
of the linear algebra problem (60) yields nonunique tr which lies in the right null space of S2 . The existence
condition of q2 for linear algebra problem (58) yields a condition on the left null space of S2 . Hence (58) and

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6359

(60) can be combined to obtain a unique expression for tr in terms of q1 . This is key to the main results of
this paper.
We now establish necessary and sucient conditions for a solution of equations (58)(62) in the absence
of external forces (i.e. wr 0) by examining each of these equations in turn beginning with the solution of
(58) and (60). The next result follows from Lemma 2.
Lemma 17. Suppose
q2 , rankS2 6 minfnp qB ; ns g

63

and let S2 have the singular value decomposition fU; R; Vg given by

S2 URVT 2 Rnp qB
ns
where

U U1 ; U2 ;

R11
R
0

U1 2 Rnp qB
q2 ;

64

0
;
0

V V1 ; V2 

U2 2 Rnp qB
nb q2 ;

V1 2 Rns
q2 ;

V2 2 Rns
ns q2 :

Then a necessary and sucient condition for (58) to have a solution q2 2 Rnp qB is given by
VT2 sr ST1 q1 0:

65

Furthermore, when (65) is satised, all solutions q2 are of the form

T
T
q2 U1 R 1
11 V1 sr S1 q1 U2 z2

where z2 2 R

nb q2

66

is arbitrary.

We now consider the solution of (60) when wr 0.

Lemma 18. When wr 0, all solutions tr of (60) which guarantee (65) are of the form
tr V2 M 1 VT2 ST1 q1 ;

M , VT2 C 1 V2

67

in which case
T
T
1 T
1
q2 U1 R 1
11 V1 C V2 M V2 I3 S1 q1 U2 z2

where z2 2 R

nb q2

68

is arbitrary.

Proof. From (64), (65) and Lemma 2, we have tr V2 zt where zt is the free solution of (60). Then from (59)
VT2 sr ST1 q1 VT2 C 1 V2 zt ST1 q1 :
Since V2 has full column rank, the matrix M VT2 C 1 V2 is invertible if it exists (that is, if cn > 0,
n 1; . . . ; ns ). Hence (65) is satised when zt M 1 VT2 ST1 q1 , and this gives (67). h
We now consider the solution of (61) and (62) when wr 0.
Lemma 19. When wr 0, a necessary and sucient condition for (61) and (62) to have a solution q1 2 RqB is
given by
X VTB1 KVB1 q1 0

69

6360

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

where
X , S1 V2 M 1 S1 V2 T ;

M , VT2 C 1 V2

70

In particular, dene
qY , rankY;

Y , X VTB1 KVB1 :

71

Then
ii(i) when qY qB , q1 0 is the only solution of (69),
i(ii) when qY 0, any q1 2 RqB is a solution of (69), and
(iii) when 0 < qY < qB , all solutions q1 are given by
q1 VY 2 z1

72
qB qY

where z1 2 R
is free, and where fUY ; RY ; VY g is the singular value decomposition of the matrix
Y 2 RqB
qB ; that is


RY 1 0
UY UY 1 ; UY 2 ; RY
; VY VY 1 ; VY 2 ; qY rankY
73
0 0
with UY 1 2 RqB
qY , VY 2 2 RqB
qB qY .
Proof. From (67), (62) and (61)
S1 tr VTB1 f r S1 V2 M 1 VT2 ST1 q1 VTB1 Kbr X VTB1 KVB1 q1
where X is given in (70). The result then follows from the singular value decomposition of X after writing
(69) in the form Xq1 VTB1 f r , VTB1 f r VTB1 KVB1 q1 . h
There are many choices of fcj ; kk g which guarantee a solution q1 6 0 of (69). One choice is provided in
the following result.
Corollary 20. If C1 cI, and K1 kI, then all solutions of (69) are characterized by the modal data of the
T
(symmetric) matrix S1 V2 S1 V2 . That is, all admissible values of k=c and q1 are eigenvalues and eigenvectors
T
of S1 V2 S1 V2 .
 kIq 0 or X qIq 0,
Proof. From (69) and C1 cI, K1 kI, X cS1 V2 VT2 ST1 cX. Then cX
1
1
where q k=c. Hence q and q1 are the eigenvalues and eigenvectors of X. h
We now make reference to Lemma 2, part (iv) to relate solutions of (58)(62) as provided in Lemma 19
to solutions of (52)(56).
Theorem 21. Consider a tensegrity structure as dened by the geometry and force equations in the absence of
external load as described by the geometric conditions
BT  I3 p b;

ST  I3 p s;

p 2 R3np ; b 2 R3nb ; s 2 R3ns

and the equilibrium force equations

S  I3 t B  I3 f;

t C  I3 s;

f K  I3 b

where for fcm > 0, kn > 0g,

C diagfc1 ; c2 ; . . . ; cns g;

K diagfk1 ; k2 ; . . . ; knb g:

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6361

Then given any tensile force coecients fcm > 0; 1 6 m 6 ns g, there exist compressive force coecients
fkn > 0; 1 6 n 6 nb g which dene an equilibrium structure, if for some q1 6 0, fK; Cg satisfy the condition:
X VTB1 KVB1 q1 0
T

X , S1 V2 M 1 S1 V2 ;

M , VT2 C 1 V2 ;

74

where VB1 is given by (28) and (29), S1 is given by (33), and V2 is given by (65).
Moreover, for any qd 6 0 which satises
X VTB1 KVB1  I3 qd 0;
the nodal vector p is of the form
p PQ  I3 zTd ; zTe T

75

where fzd 2 R3qB ; ze 2 R3np qB g may be arbitrarily chosen, and

P UB1 R 1
UB2 
B1 ;


I3
0
T
T
1 T
1
Q,
; L , U1 R 1
11 V1 C V2 M V2 I3 S1
LVY 2 U2

76

where fUB1 ; RB1 g are given by (28) and (29).

The corresponding tensegrity string vector s, tension vector t, bar vector b and compression vector f are
given in terms of fqd ; pg by
s C 1  I3 t;

t V2 M 1 VT2 ST1  I3 qd

b BT  I3 p;

f K  I3 b:

77

4.2. Externally loaded structures

Under the action of an external force vector w with component vectors fwj g given by


wT wT1 ; wT2 ; . . . ; wTnw :

78

Suppose that the new equilibrium structure is assumed to be given by node vector p, bar vector b, string
vector s, compressive force vector f, tensile vector t, compressive force coecient matrix K and tensile force
matrix C as described by (52)(56). As a result of Lemma 2, conditions for the existence of solutions
fqd 2 R3qB ; qe 2 R3np qB g of (52)(56) are equivalent to conditions (58)(62) for the existence of solutions
fq1 2 RqB ; q2 2 Rnp qB g. Note that all force coecients together with all node geometry will normally
change. We now seek necessary and sucient conditions for the externally loaded structure to be in geometric and force equilibrium. An extension of Lemmas 18 and 19 gives us the following result.
Theorem 22
i(i) All solutions tr of (60) which guarantee (65) are of the form
tr V2 M 1 VT2 ST1 q1 G1 wr ;

M , VT2 C 1 V2

T
G1 Ins V2 M 1 VT2 C 1 V1 R 1
11 U1 D2 :

79

(ii) A necessary and sucient condition for (61) and (62) to have a solution b 2 Rnb is given by
X VTB1 KVB1 q1 Gwr
UT2 P2 wr 0

80

6362

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

where X is given by (70), and

G , D1 S1 G1 :

81

Proof. Since (58) is not directly dependent on wr , Lemma 17 applies for wr 6 0. Now consider the solution
of (60) for wr 6 0. A necessary condition for the existence of a solution t is UT2 P2 wr 0, and in this case, all
solutions tr are of the form
T
tr V2 zt V1 R 1
11 U1 D2 wr

for any zt 2 Rnb r where as in (63), r is the rank of S2 . Now in order that condition (65) is satised, zt must
be selected such that
T
T T
VT2 C 1 V2 zt V1 R 1
11 U1 D2 wr V2 S1 q1 0:

That is
T
zt M 1 VT2 ST1 q1 M 1 VT2 C 1 V1 R 1
11 U1 D2 wr

which gives (79). From (60), (62) and (79)

S1 tr D1 wr VTB1 f r S1 V2 M 1 VT2 ST1 q1 S1 G1 wr D1 wr VTB1 Kbr
which gives (81). h
The rst condition in (80) is a nonhomogeneous equivalent of condition (69). However it is unlikely
(although not impossible) that VTB1 KVB1 X for wr 6 0. Instead, is more likely that rY , qVTB1 KVB1 X
satises 0 < rY 6 nb , where fUY ; RY ; VY g is the singular value decomposition of the matrix Y 2 RqB
qB given
1
by (73). If rY nb , then q1 K X Dwr is unique.
When UT2 P2 wr 0, the solution q1 is of the form
T
q1 VY 2 zY VY 1 R 1
b11 Ub1 Dwr ;

UT2 P2 wr 0

where zY 2 Rnb rb is unknown. The possibility of multiple solutions is interesting; either only one solution is
possible and more information is required to determine zY , or many solutions are possible. In the latter
case, the particular equilibrium obtained will then depend on way in which the external load wr is introduced. The structural implications of the null space condition UT2 P2 wr 0 on the external load wr would
then also require a physical interpretation.
The existence of an equilibrium solution however requires the second condition in (80) on the external
force wr to be satised. In this regard, we have the following result.
Lemma 23. For all structures fS; Bg, the nb r
ns matrix product UT2 P2 is of the form
UT2 P2 e1; 1; . . . ; 1

82

for some vector e. Hence UT2 P2 wr 0 if and only if

nw
X

wk 0:

83

k1

Proof. It follows from Lemma 2 and svdS2 in Lemma 17 that UT2 S2 0 which from (33) implies
UT2 P2 S 0. Now from Lemma 6, each column of S has exactly 1 and exactly )1 with all other column
elements 0. Furthermore, for every ith row of S, there exists a column j such that the ijth component of S is
1. These properties of S then imply that UT2 P2 is of the form (82) for some vector e, and so

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

UT2 P2 wr e

np
X

6363

wk :

k1

e T Z1 ; Z2 ; . . . ; Zn  it follows from
e 2 is partitioned into the block form U
Now if the full row rank matrix U
2
b
(42) that
e TP
e
U
2 2 Z1 ; Z1 ; Z2 ; Z2 . . . ; Znb ; Znb 
e TP
e
which then guarantees that U
e is invert2 2 also has full row rank, and consequently that the matrix ~
ible. h
Condition (83) expresses the requirement that for an externally loaded tensegrity structure to be in force
and geometric equilibrium, it is necessary (but not sucient) that the sum of the external forces is zero.
5. Computational algorithm for equilibria
One procedure for construction of a tensegrity structure in equilibrium is provided as follows.
Step 1. Given the connectivity matrices S and B from the network topology, nd a nonsingular matrix
P PT1 ; PT2  such that BTq BT P VTB1 ; 0nb , and calculate fS1 P1 S; S2 P2 Sg.
Step 2. Choose fcm > 0g and fkn > 0g such that detX VTB1 KVB1 0.
Step 3a. Select suitable zd and compute qd by
qd VY 2  I3 zd :
Step 3b. When the bar connectivity matrix B has full rank qB nb (i.e. no loops of bar vectors), then one
can select suitable zd and compute b by
b VB1 VY 2  I3 zd :
Step 4. Select ze and compute the node vector p from (75).
Step 5. Compute ft; s; f; bg from
s C 1  I3 t;

b BT  I3 p;

f K  I3 b

t V2 M VT2 ST1  I3 qd D1  I3 w
T
D1 Ins V2 M 1 VT2 C 1 V1 R 1
11 U1 P2 :

84

6. Illustrated examples
We now illustrate the construction procedure for a simple tensegrity structure.
Example 24. A general force conguration for the class 1 tensegrity structure in Example 8 with
w1 w2 0 will be investigated in this section. Suppose the force coecient matrices are given by
K diagfk1 ; k2 g and C diagfc1 ; c2 ; c3 ; c4 g.
Step 1. The connectivity matrices S, B and the coordinate transformation P PT1 ; PT2  are given in
Example 8. Since B is of full column rank matrix and V2 spans the null space of S2 , we compute VB1 I2
and

6364

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

1
61
V2 6
40
0

1
0
1
0

3
1
07
7:
05
1

Step 2. Choose fcm > 0g and fkn > 0g such that detX VTB1 KVB1 0, where


1
c2 c3 c4 c1
c3 c1 c2 c4
1
X S1 V2 VT2 C 1 V2 VT2 ST1
:
c3 c1 c2 c4
c3 c4 c2 c1
c1 c2 c3 c4
In all choices for fkn g that led to the rank of X K having rank 1, the 4 2 matrix VY 2 is of the form
VTX 2 AT1 ; AT1  in (72); that is the two bar vectors fb1 ; b2 g are always parallel, so the equilibrium structure
is one dimensional with fp1 p3 ; p2 p4 g. Hence for a two-dimensional structure, X K must have rank
zero. This requires
c3 c1
c2
c2 c3 c4 c1
k1
c2 c3 c4 c1
c3 c4 c2 c1
k2
c2 c3 c4 c1
c4

85

where c1 , c2 , and c3 are free positive constants. We choose fck 1; k 1; 2; 3g and then C I4 . It follows
that X I2 and K I2 satisfy condition (74) in Theorem 21.
Step 3b. When the bar connectivity matrix B has full rank qB nb (i.e. no loops of bar vectors), then one
can select suitable zd and compute b by b VB1 VY 2  I3 zd . Since we choose c and k such that X K 0,
T
T
VY 2 I; that is the bar vector is arbitrary. Let us choose b1 2; 0 , b2 0; 2 .
Step 4. Select ze and compute the node vector p from (75).
The nodes fp1 1; 0T ; p2 1; 0T ; p3 0; 1T ; p4 0; 1T g dene an equilibrium solution from (75)
setting ze is zero. When we set ze 1; 1T , we obtain the nodal vector








0:6464
1:3536
0:3536
0:3536
p1
; p2
; p3
; p4
:
0:3536
0:3536
0:6464
1:3536
T

This choices of ze only translate the geometric center of the structure from 0; 0 to 0:35361; 1 , since all
force coecients and b have been specied.
Step 5. Compute ft; s; fg from (84).
Example 25. Consider the 3; 9; 3 class 1 tensegrity structure dened in Example 10. A symmetrical force
conguration will be investigated with equal bar force coecients fk1 k2 k3 kg, equal base string
force coecients fc1 c2 c3 , cb g, equal top string coecients fc4 c5 c6 , ct g, and equal vertical
string coecients fc7 c8 c9 , cv g. Then
1

X S1 V2 VT2 C 1 V2 VT2 ST1

1
X1
2c2b 8cb ct 6cb cv 6cv ct 2c2t 3c2v c3t c3b c3v

where
2

3
ct cv cb c2v 3cv ct 3cb cv 4cb ct c2b c2t
X1 4 cb cv ct cb c2t 4c2b ct 3c2b cv c3b cv c2t 2ct c2v c3v 5
4cb c2t cb cv ct c2b ct 2cb c2v c2b cv 3cv c2t c3t c3v

X2

X3 ;

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6365

cb cv ct cb c2t 4c2b ct 3c2b cv c3b cv c2t 2ct c2v c3v

4
X2 10cb cv ct 4cb c2t 7c2b ct 5cb c2v 6c2b cv c3b 2cv c2t 3ct c2v c3v 5
3cb c2t cb cv ct cv c2t ct c2v 3c2b ct cb c2v c2b cv c3v
2

3
4cb c2t cb cv ct c2b ct 2cb c2v c2b cv 3cv c2t c3t c3v
5:
X3 4
3cb c2t cb cv ct cv c2t ct c2v 3c2b ct cb c2v c2b cv c3v
2
2
2
2
3
3
2
2
6cv ct 5ct cv 10cb cv ct 7cb ct 4cb ct ct cv 3cb cv 2cb cv
Now we need to choose the force coecients such that
detX VTB1 KVB1 0:
Since bar connectivity matrix is of full column rank, VB1 I and
detX K

2c2b

k cv
ksecond 0
8cb ct 6cb cv 6cv ct 2c2t 3c2v

where
ksecond , 2c3t k 2c2v c2t 2cv c3t 6c3b ct 15c2b c2t 2c2b c2v 6cb c3t 2c3b cv 6c2v ct k 10cv c2t k 16c2b ct k
2c2b k2 2c3b k 10c2b cv k 6cb k2 cv 6cb kc2v 8cb k2 ct 16cb kc2t 22cb kcv ct
2k2 c2t 3k2 c2v 6k2 cv ct 16cb c2t cv 8cb ct c2v 16c2b cv ct :
Since smaller rank of X K yields more freedom for the choice of b, we choose cv k. Next evaluating
the second term when cv k, we have
ksecond jcv k 6k2 c2t 6c2b k2 6cb c3t 6c3b ct 15c2b c2t 6cb k2 ct 3k4
k2 ct cb 2c2b k2 ct cb 2c2t :
p
p
We conclude k ct cbp

2c2b or k ct cb 2c2t , since k should be positive.

When we apply k ct cb 2c2t ,
cb ct cb ct
X:
2c2b 6cb cbt 11cb ct 8c2t 6ct cbt
p
where cbt , ct cb 2ct and
2
3
ct cbt
cb 2cbt 3ct cb 4ct 3cbt
X 4 cb 4ct 3cbt cb 4cbt 6ct 2ct cbt 5:
ct cbt
2ct cbt
ct
X K

case is when ct cb . In this case, an equilibrium

Note that the rank of the matrix X is 1. Anpinteresting

solution with K X is provided by k cv 3ct forpall

choices
of fct g. Hence the bar vector can be freely

chosen in this case. When ct cb 1 and k cv 3, the structural shape is the prism given in Fig. 3.

7. Conclusion
This paper characterizes the static equilibria of tensegrity structures. Analytical expressions are derived
for the equilibrium condition of a tensegrity structure in terms of member force coecients and string and
bar connectivity information. We use vectors to describe each element (bars and tendons), eliminating the
need to use direction cosines and the subsequent transcendental functions that follow their use. By enlarging the vector space in which we characterize the problem, the mathematical structure of the equations

6366

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

admit treatment by linear algebra methods, for the most part. This reduces the study of a signicant portion
of the tensegrity equilibria to a series of linear algebra problems. Our results characterize the equilibria
conditions of tensegrity structures in terms of a very small number of variables since the necessary and
sucient conditions of the linear algebra treatment has eliminated several of the original variables. This
formulation oers insight and identies the free parameters that may be used to achieve desired structural
shapes. Since all conditions are necessary and sucient, these results can be used in the design of any
tensegrity structure. Special insightful properties are available in the special case when one designs a
tensegrity structure so that all strings have the same force per unit length (c), and all bars have the same
force per unit length (k). In this case, all admissible values of k=c are the discrete set of eigenvalues of a
matrix given in terms of only the string connectivity matrix. Furthermore the only bar vectors which can be
assigned are eigenvectors of the same matrix. Future papers will integrate these algorithms into software to
make these designs more ecient.

Acknowledgements
Part of this research was conducted by Dynamic Systems Research Inc. under the sponsorship of the Jet
Propulsion Laboratory, technical monitors Dr. Marco Quadrelli and Fred Hadaegh.

References
Barnes, M.R., 1998. Form-nding and analysis of prestressed nets and membranes. Computers & Structures 30 (3), 685695.
Connelly, R., 1982. Rigidity and Energy. Inventiones Mathematicae 66 (1), 1133.
Connelly, R., 1993. Rigidity. In: Gruber, P.M., Wills, J.M. (Eds.), Handbook of Convex Geometry. Elsevier Publishers Ltd,
Amsterdam, pp. 223271.
Connelly, R., 1999. Tensegrity Structures: Why are they stable. In: Thorpe, M.F., Duxbury, P.M. (Eds.), Rigity Theory and
Applications. Plenum Press, New York, pp. 4754.
Desoer, C.A., Kuh, E.S., 1969. Basic Circuit Theory. McGraw-Hill, New York.
Horn, R.A., Johnson, C.R., 1985. Matrix Analysis. Cambridge University Press.
Ingber, D.E., 1993. Cellular tensegrity: dening new rules of biological design that govern the cytoskeleton. Journal of Cell Science 104,
613627.
Ingber, D.E., 1997. Tensegrity: the architectural basis of cellular mechanotransduction. Annual Review of Physiology 59, 575599.
Ingber, D.E., 1998. Architecture of Life. Scientic American. pp. 4857.
Kenner, H., 1976. Geodesic Math and How to Use It. University of California Press, Berkeley, California.
Linkwitz, K., 1999. Form nding by the direct approach and pertinent strategies for the conceptual design of prestressed and hanging
structures. International Journal of Space Structures 14 (2), 7387.
Motro, R., 1984. Forms and forces in tensegrity systems. In: Nooshin, H. (Ed.), Proceedings of Third International Conference on
Space Structures. Elsevier, Amsterdam. pp. 180185.
Motro, R., 1990. Tensegrity systems and geodesic domes. International Journal of Space Structures 5 (34), 341351.
Motro, R., 1992. Tensegrity systems: the state of the art. International Journal Space Structures 7 (2), 7583.
Motro, R., 2001. Foldable Tensegrities. In: Pellegrino, S. (Ed.), Deployable Structures. Springer Verlag, Wien-New York.
Motro, R., Belkacem, S., Vassart, N., 1994. Form nding numerical methods for tensegrity systems. In: Abel, J.F., Leonard, J.W.,
Penalba, C.U. (Eds.), Proceedings of IASS-ASCE International Symposium on Spatial, Lattice and Tension Structures, Atlanta,
USA. ASCE, New York. pp. 707713.
Pugh, A., 1976. An Introduction to Tensegrity. University of California Press, Berkeley, California.
Schek, H.J., 1974. The force density method for form nding and computation of general networks. Computer Methods in Applied
Mechanics and Engineering 3, 115134.
Snelson, K., 1996. Snelson on the tensegrity invention. International Journal of Space Structures 11 (1 and 2), 4348.
Skelton, R.E., Sultan, C. 1997. Controllable tensegrity. In: Proceedings of the SPIE 4th Annual Symposium of Smart Structures and
Materials, San Diego, March.
Skelton, R.E., Helton, W.J., Adhikari, R., Pinaud, J.P., Chan, W., 2001a. An Introduction to the Mechanics of Tensegrity Structures,
Handbook of Mechanical Systems Design. CRC Press, Boca Raton, FL.

D. Williamson et al. / International Journal of Solids and Structures 40 (2003) 63476367

6367

Skelton, R.E., Pinaud, J.P., Mingori, D.L., 2001b. Dynamics of the shell class of tensegrity structures. Journal of the Franklin Institute
338 (23), 255320.
Tibert, G., Pellegrino, S., 2001. Review of form-nding methods for tensegrity structures. In: Tensegrity Workshop, Rome, May.
Vassart, N., Motro, R., 1999. Multiparametered form.nding method: application to tensegrity systems. International Journal of Space
Structures 14 (2), 147154.
Williamson, D., Skelton, R.E., 1998a. A general class of tensegrity systems: geometric denition, engineering mechanics for the 21st
century. In: ASCE Conference, La Jolla, California, May.
Williamson, D., Skelton, R.E., 1998b. A general class of tensegrity systems: equilibrium analysis, engineering mechanics for the 21st
century. In: ASCE Conference, La Jolla, California, May.