STATIC ANALYSIS OF TENSEGRITY STRUCTURES

Carl D. Crane III Joseph Duffy

University of Florida University of Florida
Department of Mechanical and Department of Mechanical and
Aeronautical Engineering Aeronautical Engineering
P.O. Box 116300 P.O. Box 116300
Gainesville, Florida, 32611 Gainesville, Florida, 32611
USA USA
telephone: 352-392-9461, fax: 352-392-1071
ccrane@ufl.edu

Julio C. Correa
Universidad Pontificia Bolivariana
Department of Mechanical
Engineering
P.O. Box 56006
Medellín, Colombia
telephone: 574-415-9020
fax: 574-411-8779
jccorrea@logos.upb.edu.co

ABSTRACT The development of tensegrity structures is relatively new

In this paper the mathematical model to perform the static
analysis of an antiprism tensegrity structure subjected to a wide
variety of external loads is addressed. The virtual work
approach is used to deduce the equilibrium equations and a
method based on the Newton’s Third Law is used to verify the
numerical results. Two numerical examples are provided to
demonstrate the use of the mathematical model, as well as the
verification method.
Keywords: tensegrity, compliance, self-deployment.
INTRODUCTION
Tensegrity structures are spatial structures formed by a
combination of rigid elements (the struts) and elastic elements
(the ties). No pair of struts touch and the end of each strut is
connected to three non-coplanar ties [1]. The struts are always
in compression and the ties in tension. The entire configuration
stands by itself and maintains its form solely because of the
internal arrangement of the ties and the struts [2]. Tensegrity is
an abbreviation of tension and integrity.
and the works related have only existed for the 25 years.
Kenner [3] established the relation between the rotation of the
top and bottom ties. Tobie [2] presented procedures for the
generation of tensile structures by physical and graphical
means. Yin [1] obtained Kenner’s results using energy
considerations and found the equilibrium position for the
unloaded tensegrity prisms. Stern [4] developed generic design
equations to find the lengths of the struts and elastic ties needed
to create a desired geometry. Since no external forces are
considered his results are referred to the unloaded position of
the structure. Knight [5] addressed the problem of stability of
tensegrity structures for the design of a deployable antenna.
In this paper the problem of the determination of the
equilibrium position of a tensegrity structure when external
forces and external moments act on the structure is addressed.
NOMENCLATURE
Figure 1a shows a tensegrity structure formed by n struts
each one of length LS . In every structure it is possible to

1
identify the top ties, the bottom ties and the lateral or
connecting ties which are denoted as T, B and L respectively.
Figure 1b shows the same structure. The bottom end of each
strut is labeled consecutively as E1 , E 2 , L , E j ,L , E n where 1
identifies the first strut and n stands for the last strut. Similarly
the top ends of the struts are labeled as A1 , A2 , L , A j , L , An .
The selection of the first strut is arbitrary.
For each strut the generalized coordinates are the
horizontal displacements aj, bj of the lower end of the strut
together with two rotations about the axes of the universal joint,
εj and βj. εj corresponds to the rotation of the strut about B x
axis and βj corresponds to the rotation about C y axis, see Fig.
2b.
D
z

Top tie

Connecting tie P LS
•
l

Strut
A
z

A
y
Bottom tie D
A
x
(a)
(a)
An D
z
A1
Aj
A2
C
z

βj B
z

E2
E1 C D
y, y

Ej En A
z
εj B
(b) y
A
y
t
Figure 1. Nomenclature for tensegrity structures. a) B
x ,C x
Components; b) Strut ends. a j
A
x
(b)
GENERALIZED COORDINATES AND bj
TRANSFORMATIONS MATRICES
Figure 2. Degrees of freedom associated with one of the struts
Figure 2a shows an arbitrary point P located on a strut of
of a tensegrity structure. a) Arbitrary point on the strut; b) Strut
length LS . In a reference system D whose z axis is along the modeled as a universal joint.
axis of the strut and with its origin located at the lower end of
D
the strut, the coordinates of P are simply (0,0,l). However The alignment of the z axis on the fixed system with the
frequently is more convenient for purposes of analysis to axis of the rod can be accomplished using the following three
express the location of P in the global reference system A. consecutive transformations, [7]: translation, t=(aj, bj, 0),
If the lower end of one strut is constrained to move on the rotation ε about the current x axis ( B x) and rotation β about the
horizontal plane and also the rotation about its longitudinal axis current y axis ( C y).
is constrained, the strut can be modeled by a universal joint. In The coordinates of P expressed in the global reference
this way the joint provides the 4 degrees of freedom associated system are
with the strut. The total system has 4*n degrees of freedom
which means there are 4*n generalized coordinates.

2
A
P = ABTa ,b, 0 B
C ε T C
T
D β
D
P (1) a j 
E j =  b j  .
A
(7)
where  0 
1 0 0 a Similarly the coordinates of the upper end of the struts A j
0 1 0 b 
A
Ta ,b,0 =  , (2) are evaluated by replacing l by the length of the struts LS in (6)
B
0 0 1 0
   Ls sin β j + a j 
0 0 0 1  
= − Ls sin ε j cos β j + b j  .
A
Aj (8)
1 0 0 0  Ls cos ε j cos β j 
0  
 cos ε − sin ε 0  Equations (7) and (8) permit one to obtain expressions for
B
Tε = , (3)
C
0 sin ε cos ε 0 the lengths of the top, bottom and lateral ties in terms of the
  generalized coordinates as follows
0 0 0 1
Tj = ((Aj +1, x − Aj,x ) + (A
2
j +1, y − Aj, y ) + (A
2
j +1, z − Aj,z ))
2 1/ 2
(9)
 cos β sin β
 0
0
1 0
0
0 
Bj = ((E j +1, x − E j,x ) + (E
2
j +1, y − E j, y ) + (E
2
j +1, z − E j,z ))
2 1/ 2
(10)

=
((A ))
C
Tβ , (4)
D
 − sin β 0 cos β 0 Lj = j,x − E j +1, x ) + (A
2
j, y − E j +1, y ) + (A
2
j, z − E j +1, z
2 1/ 2
(11)
  where if j = n then j + 1 = 1
 0 0 0 1
D
P = [0 0 l 1]T . (5) THE PRINCIPLE OF VIRTUAL WORK FOR
TENSEGRITY STRUCTURES
Substituting the above three expressions into (1) yields The virtual work for systems able to store potential energy
can be stated from [6] by
 x  l sin β + a 
 y − l sin ε cos β + b δW = δWnc + δWc (12)
P =   =  
A
(6) where δW is the total virtual work, δWnc is the virtual work
z  l cos ε cos β 
    performed for non-conservative forces and moments and δWc
1   1  is the virtual work performed by conservative forces. δWnc can
be represented as
In addition to the constraint imposed that the lower ends
are to remain in the horizontal plane and for each strut to avoid δWnc = δWF + δWM (13)
the rotation about its longitudinal axis the following where δWF is the total virtual work performed by non-
assumptions are made without loss of generality: conservative forces and δWM is the total virtual work
The external moments are applied along the axes of the performed by non-conservative moments.
universal joints. The virtual work performed by the conservative force j,
• The struts are massless. δWcj is δWcj = −δV j where δV j is the potential energy
• All the struts have the same length. associated with the conservative force j , therefore the total
• Only one external force is applied per strut.
contribution of the conservatives forces δWc is
• There are no dissipative forces acting on the system.
• All the ties are in tension at the equilibrium position, δWc = − δV (14)
i.e., the initial lengths of the ties are greater than their where δV is the summation over all the δV j present in the
respective free lengths. structure.
• The free lengths of the top ties are equal. Substituting (13) and (14) into (12) yields
• The free lengths of the bottom ties are equal. δ W = δW F + δW M − δV . (15)
• The free lengths of the connecting ties are equal. In equilibrium the virtual work described by (15) must be
• The stiffness of all the top ties is the same. zero [6], then the equilibrium conditions can be deduced from
• The stiffness of all the bottom ties is the same.
δ W F + δW M − δV = 0 . (16)
• The stiffness of all the connecting ties is the same.

COORDINATES OF THE ENDS OF THE STRUTS THE VIRTUAL WORK DUE TO THE EXTERNAL
j
FORCES
The Cartesian coordinates of the lower ends E , expressed As it is assumed that there is only one external force acting
in the global reference system A, are obtained in terms of the
on each strut, the virtual work δWF performed by all the
generalized coordinates substituting l in (6) by 0, and replacing
external forces is given by
a and b by aj, bj

3
 n THE VIRTUAL WORK DUE TO THE EXTERNAL
δWF = ∑ j =1
F j ⋅ δrj (17) MOMENTS
The virtual work performed by the external moments is
where F j is the external force acting on the strut j and δr j is given by
n
the virtual displacement of r j which is the vector that goes
from the origin of the global reference system to the point of
δW M = ∑
j =1
Mε j ⋅ δε j + M β j ⋅ δβ j
. (21)
application of the external force. If the distance between the
Provided that in this model of the tensegrity structure the
point of application of the force and the lower end of the strut is
external moments can be exerted only along the axis of the
LFj , see Fig. 3, then an expression for r j in the global system
universal joint, M ε j , which is collinear with δε j , and M β j ,
can be obtained from (6) replacing l by LFj and its rectangular
which is collinear with δβ j (see Fig. 4), and therefore Eq.(21)
coordinates are
is simplified to
 A rj x   LFj sin β j + a j  n
A
rj

=  A rj
  
y  =  − LFj sin ε j cos β j + b j  . (18) δW M = ∑
j =1
M ε j δε j + M β j δβ j . (22)
 A r j z  LFj cos ε j cos β j 

C
z

βj

Mβ
C
A L Fj y
A Fj A
rj z
εj
B
A y
z A
y
B Mε
A x ,C x
A
y x
A
x
Figure 4. External moments applied to one of the struts of a
Figure 3. Location of the external force acting on the strut j. tensegrity structure.

The virtual displacements can be deduced from (18) where

the derivatives are taken with respect to the generalized THE POTENTIAL ENERGY
coordinates εj, βj, aj and bj as Since the struts are considered massless the term related to
the potential energy in the principle of virtual work is the
resultant of the elastic potential energy contributions given by
 Aδr j x
  the ties. The potential elastic energy for a general tie j is given
δ rjA =  Aδr j y by, [6]
 Aδr j z
 1
Vj = k (w j − w j 0 ) 2 (23)
 δa j + LFj cos β j δβ j  2
  where Vj is the elastic potential energy for tie j, k the tie
= δb j − LFj cos ε j cos β j δε j + LFj sin ε j sin β j δβ j  (19)
stiffness, wj the current length of the tie j and wjo the free length
 − LFj sin ε j cos β j δε j − LFj cos ε j sin β j δβ j  of the tie j. Therefore the differential of the potential energy for
tie j is
Substituting (19) into (17), the general expression for the δ V j = k ( w j − w j 0 ) δw j . (24)
virtual work performed by external forces is given by
The differential of the potential energy for all the
n
tensegrity structure, δV, is the resultant of the contributions of
δW F = ∑ j =1
( LFj [− AF j y cos ε j cos β j − AF j z sin ε j cos β j ]δε j the top ties, the bottom ties and the lateral ties and can be
expressed as
+ LFj [ AFj x cosβ j + AFj y sinε j sinβ j −AFj z cosε j sinβ j ]δβj
+ A F j x δa j + A F j y δ b j ) . (20)

4
 n n Equation (26) must be satisfied for all the values of the
δV = ∑ (
j =1
kT T j − To δT j + ) ∑ (j =1
k B B j − Bo δB j ) virtual displacements which in general are different from zero,
then
n f1 = 0
+ ∑ k (L
j =1
L j − Lo δL j ) (25)
f2 = 0
(31)
where kT , k B , k L are the stiffness of the top, bottom and M
connecting ties respectively, T0 , B0 and L0 are the free f 4n = 0
lengths of the top, bottom and connecting ties respectively and where f i is given by Eq. (27) to (30). Equations (31) represent
T j , B j and L j are given by (9), (10) and (11) and are a strongly coupled system of 4*n equations depending only on
functions of some of the generalized coordinates. the 4*n generalized coordinates. The equilibrium position for a
general tensegrity structure is obtained by solving numerically
the set (31) for a1 , b1 , ε 1 , β 1 , ........ , a n , bn , ε n , β n .
THE GENERAL EQUATIONS After that Eq. (7) and (8) yield explicitly expressions for the
Now that each one of the terms contributing to the virtual coordinates of the ends of the struts in the global coordinate
work has been evaluated, the equilibrium condition for the system.
general tensegrity structure can be established. Substituting
(20), (22) and (25) into (16) and re-grouping yields INITIAL CONDITIONS
f1δa1 + f 2δa2 + ..... f nδan + f n +1δb1 + f n + 2δb2
To be able to solve (31) iteratively it is necessary to first
find a proper set of values for the generalized coordinates in the
unloaded position. This is accomplished using Yin’s method
+..... f 2 nδbn + f 2 n +1δε 1 + f 2 n + 2δε 2 + ..... f 3nδε n [1], which is presented here without proof. Yin derived the
following expressions:
+ f 3n +1δβ1 + f 3n + 2δβ 2 + ..... f 4 nδβ n = 0 (26)
 L  γ
where
k L 1 − o  R B − 2 k T (RT − RTo ) sin = 0, (32)
 L 2
n
∂T j
fi = A
Fi x − ∑ kT (T j − To ) ∂ai
 L 
k L 1 − o  RT − 2k B (R B − RBo ) sin
γ
= 0, (33)
j =1  L 2
n
∂B j n
∂L j [cos(α + γ ) − cosα ]
− ∑ kB (B j − Bo ) ∂ai
− ∑ kL (L j − Lo ) ∂ai
(27) L − L2s + 2 RB RT =0 (34)
j =1 j =1 where
To Bo
RT 0 = and RB 0 = . (35)
n
∂T j γ γ
f n+i = A
Fi y − ∑ kT (T j − To ) ∂bi
2 sin
2
2 sin
2
j =1
The angles γ and α are given by
n
∂B j n
∂L j
− ∑ kB (B j − Bo ) ∂bi
− ∑ kL (L j − Lo ) ∂bi
(28)
γ =
2π
and α =
π
−
π
(36)
j =1 j =1 n 2 n
where n is the number of struts.
The three unknowns RB , RT and the length of the
[
f 2 n + i = LFi − AFi y cos ε i cos β i − A
Fi z sin ε i cos β i ] connecting ties L are solved using Eq. (32) through (34). These
values are used to solve the following set of generalized
n
∂T j n
∂B j
+ Mε i − ∑ kT (T j − To ) ∂ε i
− ∑ kB (B j − Bo ) ∂ε i
coordinates for the unloaded position.
j =1 j =1 a j ,0 = R B cos( ( j − 1) γ ) , j = 1, 2, .... , n , (37)
n
∂L j
− ∑ kL (L j − Lo ) ∂ε i
(29) b j ,0 = R B sin ( ( j − 1) γ ) , j = 1, 2, .... , n , (38)
j =1
b j ,0 − RT sin( ( j − 1) γ + α )
tan ε j ,0 = , (39)
f 3n+i = LFi [ A Fi x cos βi + Fi y sin ε i sin β i − Fi z cosε i sin β i ]
A A
H
RT cos( ( j − 1) γ + α ) − a j , 0
n
∂T j n
∂B j tan β j , 0 = , (40)
+ Mβ i − ∑ kT (T j − To )
∂β i
− ∑ kB (B j − Bo )
∂β i
 b j ,0 − RT sin( ( j − 1) γ + α ) 




j =1 j =1
 sin ε j ,0 
n
∂L j
− ∑ j =1
kL (L j − Lo )
∂β i
(30)
where H = L2s − RB2 − RT2 − 2 RB RT sin
γ
(41)
2
i = 1, 2, . . . , n . and if j = 1 then j − 1 = n .

5
 1 0 0 
VERIFICATION OF THE NUMERICAL RESULTS  
Because of the complexity of the equilibrium equations it C
A
R = 0 cos ε j − sin ε j  (50)
is essential to verify the answers obtained. An independent 0 sin ε cos ε j 
 j
validation of the results can be accomplished using Newton’s
Third Law. If there are no external moments acting on an and 03 is a 3 by 3 zeroes matrix.
isolated strut, it is sufficient to perform the summation of Figure 5 shows the free body diagram for an arbitrary strut
moments with respect to the lower end of the strut. If there are modeled with a universal joint. In addition to the forces in the
external moments the verification process involves additional ties there is an external force Fj which is known, a reaction
steps. force R passing through the lower end and a reaction moment
The unitized Plücker coordinates of a line joining two RM at the lower end. Newton’s Third Law expressed in
finite points (x1, y1, z1) and (x2, y2, z2), as is the case of the Plücker coordinates in the C system is
forces in the ties, can be written in the global reference system FAj Aj +1 C $̂ Aj Aj +1 + FAj Aj −1 C $̂ Aj Aj −1 + FAj Ej +1 C $̂ Aj Ej +1
as, [8]
1 + FEj Ej +1 C $̂ Ej Ej +1 + FEj Ej −1 C $̂ Ej Ej −1 + FEj Aj −1 C $̂ Ej Aj −1
A
$̂ = [L M N P Q R]
T
(42)
L2 + M 2 + N 2 C
+ $ F + M ε C $̂ Mε + M β C $̂ Mβ + $ R + $ RM = 0
C C
(51)
where
1 x1 1 y1 1 z1
L = , M = , N = (43)
1 x2 1 y2 1 z2
A j −1
and
F AjAj −1
y1 z1 z1 x1 x1 y1 F AjAj +1 A j +1
P = , Q = ,R = (44) Aj
y2 z2 z2 x2 x2 y2
Fj
The subindex 1 in (43) and (44) identifies the end of the tie F AjEj +1
attached to the current strut and the subindex 2 is for the
remaining end of the tie. Further the coordinates of the ends of Mβ F EjAj −1
the ties can be evaluated using (7) and (8). C
y C
Mε R
x
If L, M and N are simultaneously equal to zero (42) must
be modified to A
z RM
1
A
$̂ = [0 0 0 P Q R]
T
(45) E j +1 F EjEj +1 A
y
A
x Ej
P2 + Q2 + R2
F EjEj −1
When an external force A Fj and its point of application are
known, the Plücker coordinates are obtained by E j −1

A
 A
Fj 
$F =  A A  (46) Figure 5. Free Body diagram for an arbitrary strut modeled with
 r j × F j  a universal joint.
where A Fj corresponds to the external force expressed in the
global reference system and A r j is given by (18). The coefficients FAj Aj +1 , FAj Aj −1 , L represent the
The Plücker coordinates can be expressed in a new system magnitudes of the forces in the ties A j Aj +1 , A j Aj −1 L and are
that is translated and rotated with respect to the global reference
system. If the new system is the C system, this is the system given by k*(w-wo) where k is the stiffness, w the current length
defined for the axes of the universal joint, the expression that and wo the free length of the tie. The current lengths are given
relates the Plücker coordinates in the A system and the C by (9) and (11) for the top ties and connecting ties respectively.
system is, [8] It should be noted that the magnitude does not depend of the
reference system which is used.
$ = e −1 $
C A
(47) The unitized Plücker coordinates C $̂ for each one of the ties
where can be calculated in the A system using (42) thorough (44) and
 CA R T O3  then converted to the C system using (47) through (50).
−1   The Plücker coordinates of the external force acting on the
e =   (48)
current strut $ F , can be evaluated in the A system using (46)
 CA R T A3T A T
C R  and then converted to the C system with the aid of (47) through
 0 0 bj  (50).
 
A3 =  0 0 −aj (49)
− b j aj 0 


6
 Mε and M β are the magnitudes of the external moments
and their unitized Plücker coordinates in the C system are given
Table 3. External forces and their application points
by C $̂ Mε = [0 0 0 1 0 0] T and C $̂ Mβ = [0 0 0 0 1 0] T . acting on the structure of the example 1.
Since the reaction force C $ R expressed in the C system is Strut 1 Strut 2 Strut 3
a pure force and the reaction moment C $ RM expressed in the C Fx (N) 0 0 0
system is a pure moment they have the form [CRx CRy CRz 0 0 Fy (N) 0 0 0
0] T and [0 0 0 CRMx CRMy CRMz] T respectively. Further they
are the only unknowns in (51). Fz (N) -10 -10 -10
After expanding (51), rows four and five represent the LF (mm) 100 100 100
components in the x and y directions of the summation of
moments about the lower end of the strut. A universal joint
cannot exert moment along its own axes. If after a numerical Since the system has 3 struts and there are no constraints
evaluation, RMx and RMy are both zero, then the equilibrium of then there are 12 degrees of freedom and therefore 12 equations
moments is maintained solely due to the forces in the ties and are required, one per each generalized coordinate. Equation
to the external loads (if any) and therefore the current position (27) yields f1 , f 2 and f 3 , Eq. (28) yields f 4 , f 5 and f 6 , Eq.
is an equilibrium position.
(29) yields f 7 , f 8 and f 9 , and Eq. (30) yields f10 , f11 and
f12 . Each f i is equated to zero and then the system is solved
EXAMPLE 1: ANALYSIS OF A TENSEGRITY numerically using the software developed and the initial values
STRUCTURE WITH 3 STRUTS. listed in Table 2. The values of the generalized coordinates for
ANALYSIS FOR THE UNLOADED POSITION the structure when the external forces are applied are shown in
It is required to evaluate the unloaded equilibrium position Table 4.
of a tensegrity structure with 3 struts and with the stiffness and
free lengths shown in Table 1. Each of the struts has a length Table 4. Generalized coordinates for the final
Ls = 100mm . position for the structure of the example 1.
Strut 1 Strut 2 Strut 3
a (mm) 40.8573 -20.4241 -20.4332
Table 1. Stiffness and free lengths for example 1.
Stiffness (N/mm) Free lengths (mm) b (mm) -0.0053 35.3861 -35.3808
Top ties kT = 0.5 T0 = 35 ε (rad) -0.0269 0.6808 -0.6643
Bottom ties k B = 0.3 B0 = 52 β (rad) -0.7434 0.3271 0.3635
Connecting ties kL = 1 L0 = 80
Using the values of Tables 2 and 4, Eq. (7) and (8) yield
the coordinates of the ends of the struts for the initial and final
position. The results are summarized in Table 5. Figure 6
For this example n = 3 then (35) and (36) yield γ = 120° ,
shows the structure in its initial and final equilibrium position.
α = 15° , RT 0 = 20.207mm and RB 0 = 32.02mm .
The solution of (32), (33) and (34) yields Table 5. Lower and upper coordinates for the unloaded and
RB = 33.0568mm , RT = 22.8422mm and from (41), final position for the structure of the example 1 (mm).
H = 84.1287mm . The initial values of the generalized Strut 1 Strut 2 Strut 3
Initial Final Initial Final Initial Final
coordinates are obtained from (37) through (40) and are listed Pos. Pos. Pos. Pos. Pos. Pos.
in Table 2.
Ex 33.056 40.857
-16.528 -20.424 -16.528 -20.433
8 3
Table 2. Initial values of the generalized coordinates Ey 28.628 35.386
for the structure of the example 1. 0 -0.0053 -28.628 -35.380
0 1
Strut 1 Strut 2 Strut 3
a (mm)
Ez 0 0 0 0 0 0
33.0568 -16.5284 -16.5284
11.701 19.781 15.122
b (mm) 0 28.6280 -28.6280 Ax -19.781 -26.824 0
6 9 4
ε (rad) -0.1349 0.5491 -0.4443 Ay 11.421 11.421 22.242
β (rad) -0.5567 0.1660 0.3716 1.9750 -22.842 -24.217
1 1 9
84.128 73.588 84.128 73.588 84.128 73.588
Az 7 8 7 8 7 8
ANALYSIS FOR THE LOADED POSITION
It is required to evaluate the final equilibrium position of
the structure when the external forces listed in Table 3 are
applied vertically at the upper end of the struts and there are no
constraints acting on the struts.

7
 the equilibrium of moments the result is 0 what means the
current position is indeed an equilibrium position.
A3
A1 A2 A3
F A 2 A3
F A2 A1
A2
A1

F
r F A2 E 3

A
z
F E 2 A1
A
A
z E1
y A E1 F E 2 E1
E2 x
E2 A
x
E3 A
y

R
A3
E3
A2
A1
Figure 7. Free body diagram for the second strut of the
structure of example 1 in the last position.

EXAMPLE 2: ANALYSIS OF A TENSEGRITY

STRUCTURE WITH 4 STRUTS.
ANALYSIS FOR THE UNLOADED POSITION
A
z
It is required to evaluate the unloaded equilibrium position
of a tensegrity structure with 4 struts with the stiffness and free
A
y A
x E1 lengths shown in Table 6. Each of its struts has a length
E2 Ls = 100mm .

E3 Table 6. Stiffness and free lengths for example 2.

Stiffness (N/mm) Free lengths (mm)
Figure 6. Unloaded and final equilibrium position for the
structure of the example 1. Top ties kT = 0.5 T0 = 40
Bottom ties k B = 0.5 B0 = 40
Figure 7 shows the free body diagram of strut 2 for the Connecting ties k L = 0.5 L0 = 40
final position. The summation of moments with respect to the
lower end E2 is given by the following equation
For this example n = 4 then (35) and (36) yield γ = 90° ,
r × F + r × F A 2 A1 + r × F A 2 A3 + r × F A 2 E 3 = 0
where α = 22.55° , RT 0 = 28.2843mm and RB 0 = 28.2843mm .
The solution of (32), (33) and (34) yields
r = A2 − E 2 ,
RB = 41.2528mm , RT = 41.2528mm and from (41),
A − A2
F A2 A1 = kT ( A1 − A 2 − T0 ) 1 , H = 64.7280mm . The parameters that define the location of
A1 − A 2 the struts at the initial position are obtained from (37) through
A3 − A 2 (40) and are listed in Table 7.
F A2 A3 = kT ( A3 − A 2 − T0 ) ,
A3 − A 2 Table 7. Parameters for the location of the struts for the
E − A2 structure of example 2 in the unloaded position.
F A2 E 3 = k L ( E 3 − A 2 − L0 ) 3 . Strut 1 Strut 2 Strut 3 Strut 4
E 3 − A2
a (mm) 41.2528 0 -41.2528 0
The numerical values are obtained from Tables 1 and 5. In b (mm) 0 41.2528 0 -41.2528
addition from Table 2, F=[0 0 –10] T . It can be verified that ε (rad) -0.4234 0.8275 0.4234 -0.8275
when the numerical values are replaced in the expression for β (rad) -0.7813 -0.2960 0.7813 0.2960

8
ANALYSIS FOR THE LOADED POSITION Table 9. Generalized coordinates for the final
It is required to evaluate the final equilibrium position of position for the structure of example 2.
the structure when the external moments listed in Table 8 are Strut 1 Strut 2 Strut 3 Strut 4
applied along the axes of the universal joints that model the ε (rad) -0.468 0.6311 0.8140 -1.1716
structure, see Fig. 8, and the lower ends of the struts are β (rad) -0.291 0.1598 1.2136 0.5948
constrained in such a way that they cannot move in the
horizontal plane.
A2

A3
A1
A4
M β2

C
y
2

M β3 C
M ε1 A
x2 C
x1 E4 z

M ε2 A
E3
x A
C
y M β1 y
C 1 E1
y
3

C
E2
x3
M ε3 C
x4
A2
M ε4
C
y A1
4
M β4

Figure 8. Directions of the external moments for the structure

of example 2.

A3
Table 8. External moments acting on the A
z
structure of example 2.
Strut 1 Strut 2 Strut 3 Strut 4
A4 E4
M ε (N.mm) 450 -900 450 -900 E3
A
M β (N.mm) 450 450 450 450 x A
y
E1
E2
Because there are 2 constraints per strut there are 8 degrees Figure 9. Unloaded and final equilibrium position for the
of freedom for this system, and they are associated with the structure of example 2.
rotations of the struts. The generalized coordinates are
ε 1 , β 1 , ε 2 , β 2 , ε 3 , β 3 and ε 4 , β 4 , where the subscript Figure 10 shows the free body diagram for the fourth strut
indicates the number of the strut. in its final position modeled with a universal joint. In addition
to the forces in the ties and the external moments, all known,
The required 8 equations are obtained as follows: Eq. (29) there are a reaction force R and a reaction moment RM, these
yields f 9 , f10 , f11 and f12 and Eq. (30) yields f13 , f14 , f15 both of which are unknowns. The equilibrium equation
expressed in Plücker coordinates in the C system, this is the
and f16 . The solution of this system yields the generalized system defined by the axes of the universal joint, see Eq.(51),
coordinates for the final position listed in Table 9. is
kT ( A 4 − A1 − T0 ) C $̂ A 4 A1 + kT ( A 4 − A3 − T0 ) C $̂ A 4 A3
Using the values of Tables 7 and 9, Eq. (7) and (8) yield
the coordinates of the ends of the struts for the initial and final + k L ( A 4 − E 1 − L0 ) C $̂ A 4 E1 + k B ( E 4 − E 1 − B0 ) C $̂ E 4 E1
position. The results are summarized in Table 10. Figure 9
+ k B ( E 4 − E 3 − B0 ) C $̂ E 4 E 3 + k L ( E 4 − A3 − L0 ) C $̂ E 4 A3
shows the structure in its initial and final equilibrium position.
It should be noted that the numerical values of Ex,Ey and Ez are + M ε C $̂ Mε + M β C $̂ Mβ +
C
$R +
C
$ RM = 0
the same for the unloaded and final position due to the
constraints imposed to the lower ends.

9
 Table 10. Lower and upper coordinates for the unloaded and final position for the example 2 (mm).
Strut 1 Strut 2 Strut 3 Strut 4
Init. Pos. Fin. Pos. Init.Pos. Fin. Pos. Init.Pos. Fin. Pos. Init.Pos. Fin. Pos.
Ex 41.2528 41.2528 0 0 -41.2528 -41.2528 0 0
Ey 0 0 41.2528 41.2528 0 0 -41.2528 -41.2528
Ez 0 0 0 0 0 0 0 0
Ax -29.1701 12.5079 -29.1701 15.9151 29.1701 52.4370 29.1701 56.0322
Ay 29.1701 43.2859 -29.1701 -16.9992 -29.1701 -25.4176 29.1701 35.0631
Az 64.7280 85.4404 64.7280 79.7083 64.7280 24.0035 64.7280 32.1911

C
$̂ A4 A1 , C $̂ A4 A3 L are the unitized Plücker coordinates of
the lines joining the points A4 A1 , A4 A3 L They are calculated
in the A system with the aid of (42) thorough (44) and using the
data of Table 10 for the structure in its final position. Then
they are converted to the C system using (47) through (50).
A1
Similarly Mβ C $ Mβ are the Plücker coordinates of the
moment acting along axis C y 4 and is given by 450* [0 0 0 0 1
0] T , see Table 4.
C
$ R is the reaction force through E 4 which components
are [Rx Ry Rz 0 0 0] T and C $ RM is the reaction moment along
the axes of the C system which components are C [0 0 0 RMx
RMy RMz] T .
The substitution of numerical values in the last expression
yields

A2  - 0.6284   - 0.0588   - 0.2965 

 - 0.6622   - 0.2611   0.3217 
F A4 A1   
A4  0.4082   - 0.9635   - 0.8992 
14.6319   + 10.5691  + 4.9205 
F A 4 A3  54.8493   21.6248   - 26.6442
 - 74.9186  49.1177   25.8241 
     
F A 4 E1  - 37.1052   - 14.6290  18.0246 

A3
 0.7071  - 0.7071  0.8768 
E2  0.2748  0.2748   - 0.2669
E1     
F E 4 E1  0.6515  0.6515   0.4 
+ 9.1701  + 9.1701  + 9.9022  
F E 4 A3 R  0   0   0 
E3 F E 4E3  0   0   0 
     
 0     
C 0 0
x
E4
Mε RM
0  0  C Rx   0 
0  0   C   0 
   Ry  
C
y   
0  0  C Rz   0 
Mβ − 900   + 450   +   + C  = 0
1 0  0   RM x 
0  1  0   C RM y 
Figure 10. Free body diagram for the fourth strut of example 2       C 
in the last position. 0 0  0   RM z 

It should be noted that aj and bj in (49) correspond to the
coordinates x and y for the lower end of the strut 4. They do
not change for this example and their numerical values are
listed in Table 7. The angle εj in (50) is one of the generalized
coordinates associated with the rotations of strut 4 in the last
position and is listed in Table 9.
Mε C $ Mε are the Plücker coordinates of the moment acting
along axis C x 4 and according to the initial data is given by
900* [0 0 0 -1 0 0] T , see Table 8.
The expansion of rows four and five yield RMx=0 and
RMy=0, this is the component of the reaction moment along the
axes of the universal joint is zero. Since the universal joint
cannot exert any reaction moment along its axes, the foregoing
results confirm that the current position is an equilibrium
position. The same procedure is executed for the software for
all the struts and all the positions of the structure.

10

CONCLUSIONS
The model allows one to analyze a general anti-prism
tensegrity structure subjected to a wide variety of external loads
and the software developed is able to solve the system of
equations generated for the model. The results are presented
both numerically and in a three dimension graphical
representation, which permits one to visualize the behavior of
the structure.
The model is developed using the virtual work approach
and all the results are checked using Newton’s Third Law. This
verification assures that the answers produced by the numerical
method accurately correspond to equilibrium positions.
Mathematical models for variations of the basic
configuration of tensegrity structures such as the reinforced
tensegrity prisms might be developed following the same
procedure presented in this research.
The mathematical model always assumes that the ties are
in tension. If under the action of the external load the distance
between two strut ends which are connected by a tie is less than
the free length for that tie, the model is no longer a valid
representation of the structure and as result no convergence is
found and the software cannot yield a solution for that
particular situation.
Also when two struts or a tie and a strut intersect, the
Jacobian for the structure vanishes, Lee et al. [9], and
corresponds to a singular configuration that the software cannot
solve.
In certain situations even though there are no singularities
and all the ties are in tension, a small increase in the external
load can make it impossible for the software to converge to a
solution, i.e., it is not possible to find a new equilibrium
position. The system suffers a sudden change and it jumps from
one equilibrium position to another for a smooth transition
force. This is known as a catastrophe, Hines [10], and Arnold’s
[11]. Catastrophe Theory is a well developed classical method.
It describes sudden changes caused by a gradually changing
input. It offers a better understanding of the phenomena
reported here which is beyond the scope of this work.
[2] Tobie, R.S., 1976, “A Report on an Inquiry into The
Existence, Formation and Representation of Tensile
Structures,” Master of Industrial Design thesis, Pratt Institute,
New York.
[3] Kenner, H., 1976, Geodesic Math and How to Use It,
University of California Press, Berkeley and Los Angeles, CA.
[4] Stern, I.P., 1999, “Development of Design Equations for
Self-Deployable N-Strut Tensegrity Systems,” Master of
Science thesis, University of Florida, Gainesville, FL.
[5] Knight, B.F., 2000, “Deployable Antenna Kinematics using
Tensegrity Structure Design,” Ph.D. thesis, University of
Florida, Gainesville, FL.
[6] Doughty, S., 1998, Mechanics of Machines, John Wiley &
Sons, New York.
[7] Crane, C. and Duffy, J., 1998, Kinematic Analysis of Robot
Manipulators, Cambridge University Press, New York.
[8] Kunt, K.H., 1978, Kinematic Geometry of Mechanisms,
Oxford, New York.
[9] Lee, J., Duffy, J., and Hunt, K., 1998, “A Practical Quality
Index Based on the Octahedral Manipulator,” The International
Journal of Robotics Research, 17, pp. 1081-1090.
[10] Hines, R., 1995, “Catastrophe Analysis of Compliant
Mechanisms,” Master of Science thesis, University of Florida,
Gainesville, FL.
[11] Arnold’s, V.I., 1992 Catastrophe Theory, Springer-Verlag,
Berlin.

ACKNOWLEGEMENTS
The authors would like to gratefully acknowledge the
support of the Air Force Office of Scientific Research, Grant
Number F49620-00-1-0021.

DEDICATION
This paper is dedicated to the memory of Dr. Joseph Duffy
whom the other two authors had the pleasure of working with
in this endeavor. Dr. Duffy was a mentor and colleague and an
inspiration to a generation of kinematicians. His contributions
to the field are truly significant and he will be dearly missed
both professionally and personally.

REFERENCES

[1] Yin, J.P., 2000, “An Analysis for the Design of Self-
Deployable Tensegrity and Reinforced Tensegrity Prisms with
Elastic Ties,” Report for the Center for Intelligent Machines
and Robotics, University of Florida, Gainesville, FL.

11