Sie sind auf Seite 1von 8

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

Porto, Portugal, 30 June - 2 July 2014


A. Cunha, E. Caetano, P. Ribeiro, G. Mller (eds.)
ISSN: 2311-9020; ISBN: 978-972-752-165-4

Analysis of cable membrane


structures using the Dynamic Relaxation Method
M. Httner1, J. Mca2, P. Fajman3
Dep. of Mechanics, Faculty of Civil Eng., Czech Tech. University in Prague, Ing. Milo Httner, Prague, Czech Rep.
2
Dep. of Mechanics, Faculty of Civil Eng., Czech Tech. University in Prague, Prof. Ji Mca, CSc., Prague, Czech Rep.
3
Dep. of Mechanics, Faculty of Civil Eng., Czech Tech. University in Prague, Doc. Petr Fajman, CSc., Prague, Czech Rep.
1

email: milos.huttner@fsv.cvut.cz, maca@ fsv.cvut.cz, fajman@ fsv.cvut.cz

ABSTRACT: This paper compares the effectiveness of different schemes of dynamic relaxation method (DRM) for the
analysis of cable and membrane structures. DRM is an iterative process that is used to find static equilibrium. DRM is not used
for the dynamic analysis of structures; a dynamic solution is used for a fictitious damped structure to achieve a static solution.
The stability of the method depends on the fictitious variables (i.e. mass a damping) and time step. The effect of mass
distribution along the structure is also studied in the paper. Eight different schemes DRM will be used in this paper. Schemes A
and B are based on the theory of viscous damping. Schemes C, D and E are based on the theory of kinetic damping (KD) with a
peak in the middle of the time step and schemes F, G and H are based on the theory of KD with parabolic approximation.
A cable is approximated as a tension bar, a catenary (several tension bars) and a perfectly flexible element. For membrane
structures a triangular element is considered. The chosen methods are applied to six constructions. The cable structures are
analyzed in Examples 1 to 3, the membrane structures are analyzed in Examples 4 to 6.
The results imply that that it is impossible to determine the best scheme. In this context, it may be noticed that the methods
based on kinetic damping appear more stable and faster. For bar element, catenary and cable elements the results confirm that it
is beneficial to divide the same amount of mass into all nodes of the structure proportionally to the stiffest node of the solved
structure (schemes C and F). For membrane element it is preferred to use the kinetic damping method with the approximation of
the kinetic energy peak in the middle of the time step t .
KEY WORDS: Dynamic relaxation, Cable structures, Membrane structures, Kinetic damping.
1

INTRODUCTION

The load analysis of cable-membrane structures is a


geometrical nonlinear problem. For numerical modelling of
cable structures can be used idealization of the structure into
the elements and nodes. The surface of membrane structure is
discretized into a system of joints and triangle membrane
elements. The edges of the triangle form the connection
between the joints and they are called links.
The joints can be divided into two groups supported or
unsupported ones. The equilibrium of position unsupported
nodes loaded by the nodal loads can be searched iteratively.
Large displacements of the structure and small deformation of
elements are considered.
Several methods exist to solve these structures. The
dynamic relaxation method (DRM) will be examined in this
paper. The stability of the method depends on the fictitious
variables (i.e. mass a damping) and time step. This paper
compares the effectiveness of different schemes of DRM for
the analysis of cable and membrane structures. The effect of
mass distribution along the structure is also studied in the
paper. This paper develops papers [1] and [2].
2

2.1

Tension bar

The bar connects the endpoints and carries only positive


normal force. The internal force T (normal force) in one bar
element can be calculated according to the well-known
Equation (1).
T=

EA
(r s0 ) ,
s0

(1)

where:
is the Youngs modulus of elasticity,
is the cross-sectional area,
is the distance between two end joints in the chord
direction (current length),
s 0 is the un-elongated length of element (slack length).
If the force T is negative, then it is equal to zero. The
deadweight of strut has been assumed to be concentrated
equally at its two end joints. The used bar element can be seen
in Figure 1.
E
A
r

ELEMENTS

A cable can be approximated as a tension bar, a catenary


(several tension bars) and a perfectly flexible element (where
bending moments to zero). Homogeneous material with a
constant cross-section throughout its length is assumed in all
cases. For membrane structures a triangular element is
considered.

Figure 1. Bar element.

1919

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

2.2

2.4

Catenary

The basic assumption of this theory is that the behaviour of a


cable can be approximated by a few bars. These bars are
interconnected by joints and sustain only positive normal
force. The behaviour of individual bars is described in
Chapter 2.1. As found in [3], five bars are well enough to
describe correctly the characteristics of the cable.
2.3

Cable element

The basic assumption of the analysis of a flexible elastic cable


is that the cable is regarded to be perfectly flexible and is
devoid of any flexural rigidity. Load on a cable, which must
include at least self-weight, is distributed uniformly along the
curve of the cable which is assumed to be a parabola. The
detailed analysis can be found in [4] and [5].
It is necessary for the aim of the study to use an internal
force T , which is always positive and the importance of
which is shown in Figure 2.

Membrane element

For a membrane structures the natural stiffness element can be


used for calculation of internal forces. The original
formulation of the natural stiffness element is credited to
Argyris [6] but the formulation here follows the work of
Barnes [7] and Topping [8]. For the formulation of the natural
stiffness element a triangular element is considered. This
element has only in-plane stiffness so the element formulation
is with respect to displacements in the local coordinate
directions. Using equations of equilibrium, it is possible to
convert the surface stress within the element into forces along
the sides of the triangle. General application of this element is
described e.g. in [8].
In this case the idealization of a typical element is as shown
in Figure 3 where the local coordinate system is conveniently
chosen such a way that the axis coincides with the first side.
The stresses in the element with respect to x and y
directions, with z equal to zero, are the standard plane
stress formulation for an isotropic material [9].

Figure 2. Cable element.


Force T can be calculated iteratively from Equation (2).
g (T , r , l , c, s 0 , Q) =
2c rQ
l 2T 2c rQ
b
a
ln

+
+
+
ln
+
l
2rQ l
lT lT
lT
lT

(2)
1
c
+
a b +
a + b s0
4rQ
8T

Figure 3. Membrane element.

T l 2 c2 Q2r
+
+
EA r
r 12T 2

= 0

where:
E is the Youngs modulus of elasticity,
A is the cross-sectional area,
r
is the distance between two end joints in the chord
direction (current length),
s 0 is the un-elongated length of element (slack length).
l
is the horizontal distance between the two end joints,
c
is the vertical separation between joint j and joint i
(can be negative),
Q is the resultant of the vertical uniform load q acting
vertically the entire length of parabolic curved cable,
while Q = qs 0 .
For reasons of clarity, Equation (2) introduces two more
substitutions:

1920

a = Q 2 r 2 + 4c 2T 2 + 4l 2T 2 + 4crQT ,

(3)

b = Q 2 r 2 + 4c 2T 2 + 4l 2T 2 4crQT .

(4)

In Figure 3 is:

l1 , l2 , l3
is the length of the edge 1, 2, 3
is the inclination of the edge 2 to the local x axis,
2
3
is the inclination of the edge 3 to the local x axis.
The initial forces T1 , T2 and T3 of sides 1, 2 and 3 are
defined for membrane element as:
1/ l
T1

T2 = Ad 0
T
0
3

0
1

a3c2 a2c3
1
Q
0
c3

0 0

Q
1/ l30
c
0
2
Q

0
1/ l20
0

a2b3 a3b2

Q
x
b3

y (5)
Q

xy
b2

and
E

2
x 1

E
y =

2
1

xy
0

where:

E
(1 2 )

) (1 )

0
0
0
l1
l1

a3c2 a2c3 c3
c

2 l2 (6)
0

Ql2
Ql2
Ql2

l3
E a2b3 a3b2
b3
b
2

2(1+ ) Ql3
Ql3 Ql3

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

A
d
E

l10

is the area of the membrane element,


is the thickness of the membrane,
is the Youngs modulus of elasticity,
is the Poissons ratio,
is the length of the edge 1 of the unloaded element,

l20

is the length of the edge 2 of the unloaded element,

is the fictitious damping factor for the nodal point


i and in the direction m ,
( t + t / 2 )
vim
is the velocity at the nodal point i in the direction
m and at the time t .
The current coordinates of the nodal point i at the time
instant ( t + t ) may then be expressed as follows:
Cim

0
3

l
is the length of the edge 3 of the unloaded element,
l1 , l 2 , l3 is the elongation of the edge 1, 2 and 3
Furthermore, they are used substitutions (for h = 1, 2, 3):
ah = cos 2 h ,

(7)

bh = sin 2 h ,

(8)

ch = sin h cos h ,

(9)

xi(t + t ) = t vix(t + t / 2 ) .

Similarly, equations may be written for the y and z


coordinate directions. From the imbalance (between external
and internal forces) in the node i , we may calculate the
t
residual forces Rim
( m = x , y , or z ) for the corresponding
node at the time t .

and
(10)

The direct stiffness S h of side h ( h =1, 2, 3) is defined for


membrane element as:
Sh =

Rixt = Pix Tkt

xit x tj

Riyt = Piy Tkt

yit y tj

Q = b2c3 b3c2 .

3 EAd
.

2 0 2
lh

( )

The dynamic relaxation method (DRM) is an iterative process


that is used for the static analysis of structures. DRM is not
used for the dynamic analysis of structures; a dynamic
solution is used for a fictitious damped structure to achieve a
static solution.
The theory of this method was first described by Day [10].
During several years, the DRM have been improved
progressively. The kinetic damping technique was suggested
by Cundall [11]. Topping [8] and Lewis [12] also contributed
to the kinetic damping method. Practical examples of the
application can be seen in [1,2,3,12,13].
3.1

Principle

The basic unknowns are nodal velocities, which are calculated


from nodal displacements. The discretization from timeline
with time step t will be performed. During the step t a
linear change of velocity is assumed. The acceleration during
the step t is thus considered to be constant. By substituting
the above assumptions the velocity for joint i in direction m
( x , y and z ) can be expressed in a new time point ( t + t /2)
thus:
( t + t / 2 )
( t t / 2 )
vim
= vim

where:
t
Rim
M im

Rizt = Piz +

(11)

DYNAMIC RELAXATION

t
M im / t C im / 2
Rim
,(12)
+
M im / t + C im / 2 M im / t + Cim / 2

is the residual force at the nodal point i , in the


direction m and at the time t ,
is the fictitious mass at the nodal point i and in
the direction m ,

(13)

where:
k
j
Pix
Piy

rkt
rkt

zit z tj
Qk
,
Tkt
2
rkt
k

(14)

is the index of the link (element or edge)


entering the nodal point i .
is the second endpoint on the link k .
is the external load at the nodal point i in the
direction x ,
is the external load at the nodal point i in the

direction y ,
Piz
is the external load at the nodal point i in the
direction z ,
Qk
is the resultant of the vertical uniform load for
each link k ,
t
t
t
xi , yi , zi are the current coordinates of the nodal point i ,
rkt

is the distance between two end joints for each


link k .
The internal force Tkt for each link k can be calculated
from Equation (2) for the cable element; from Equation (1)
for the bar element; and from Equations (5) for the edge of
membrane element.
If the sum of the forces Tkt in the link k is less than zero so
it must be set equal to zero.
In order to start calculations, the velocity at the time point
t /2 must be calculated. Using the initial conditions for the
0
time t = 0 where vim
= 0 , we obtain:
( t / 2 )
vim
=

t
0
,
Rim
2 M im

(15)

where Rim0 are residual forces at the time t = 0.

1921

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

At each time instant, it is also possible to calculate the


kinetic energy U kin throughout the structure:
( t + t / 2 )
U kin
=

M (v
n

),

( t + t / 2 ) 2
im

im

(16)

3.2.2

Scheme B

In this scheme, the fictitious values M i and Ci are calculated


for each joint i separately.
The discretized mass M i for each joint i is calculated from
Equation (21):

where n is the number of joints and p is the number of


dimensions (3D or 2D).

3.2

Schemes

Eight different schemes DRM will be used in this paper.


Schemes A and B are based on the theory of viscous damping
[10]. Schemes C, D and E are based on the theory of kinetic
damping (KD) with a peak in the middle of the time step [8]
and schemes F, G and H are based on the theory of KD with
parabolic approximation [12].

3.2.1

where

t 2
Si ,
2

(17)

where S im is the largest direct stiffness of the i -th joint in the


m direction.
The stiffness S k of each link k (entering into joint i ) is for
the element (cable or bar) represented by two components
namely, the geometric stiffness S kG and the elastic stiffness
S kE .

S k = S kE + S kG =

Ek Ak Tk
.
+
s0 k
rk

S im =

k ,m

(19)

where S k ,m is the stiffness S k - from Equation (18),


respectively from Equation (11) - distributed in the m
direction.
Viscous damping coefficient C for the whole structure is
calculated using the coefficient of critical damping [12]:
The iterative algorithm converges fastest when using the
critically damped mode. In an undamped mode, the structure
will oscillate around its position of equilibrium, and the
viscous damping coefficient, known as critical damping, may
be found from Equation (20):
C = 2 S im M im =

4M im
,
Nt

3.2.3

(20)

where N denotes the number of iterations required to


complete one cycle of oscillation. It may now be seen that in
order to obtain the value of the viscous damping coefficient,
an additional computer run is necessary, with C set to zero.

Mi
t

(23)

Scheme C

This scheme is based on the theory of KD with a peak in the


middle of the time step [8].
When the technique of kinetic damping is employed, the
viscous damping coefficient is taken as zero. The system is
brought to rest by following a process stopping the iterations,
whenever a peak in the kinetic energy of the entire system is
detected, and then restarting the computation from the current
configuration, but with zero initial velocity [12].
The coordinates are set to xi(t t / 2 ) when the peak is
assumed to have occurred. Then
xi(t t / 2 ) = xit

(18)

Stiffness S k 2D membrane element should be determined


for each edge of the equation (11).
Hence, the following Equation (17) applies to the stiffness
of the node Sim :

(22)

The viscous damping coefficient Ci for the nodal point i is


calculated as follows:
Ci = 2 S i M i = 8

t 2
M =
(max Sim ) ,
2

(21)

S i = max S ix , S iy , S iz .

Scheme A

Discretized mass M is chosen in this scheme the same for


each node and all directions from follows Equation (17):

1922

Mi =

t ( t t / 2 )
.
vix
2

(24)

Similarly, equations may be written for the y and z


coordinate directions.
The mass for whole structure is calculated from
Equation (17).

3.2.4

Scheme D

This scheme is similar to the Scheme C, but the discretized


mass M i for each joint i is calculated separately from
Equation (21).

3.2.5

Scheme E

This scheme is similar to Scheme D but masses M i are


recalculated after each restart of the kinetic energy.

3.2.6

Schemes F, G and H

These schemes are similar to Scheme C (respectively D


and E) but there is used the theory of KD with parabolic
approximation [12].
The trace of kinetic energy near the peak can be
approximated by a parabolic curve. The coordinates are set to
xi(t t ) when the peak is assumed to have occurred. Then
xi(t t ) = xit t vix(t t / 2 ) ,

where

(25)

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

KE3 KE 2
,
KE3 2 KE 2 + KE1

(26)

(t + t / 2 )
and where KE3 = U kin
is the kinetic energy at the time

acts on the structure. An uniform load q = 1.510510-1 kN/m


acts on each cable. The coordinates z of the unsupported
nodes were always set to zero.
The limit value of a residual force is Rlim = 0.01 kN. The
accuracy of the calculation is approximately 1 cm.

(t t / 2 )
(t 3t / 2 )
, KE1 = U kin
.
point ( t + t / 2 ), KE 2 = U kin

EXAMPLES

The chosen methods are applied to six structures. The cable


structures are analyzed in Examples 1 to 3, the membrane
structures are analyzed in Examples 4 to 6. The initial
geometry (no internal stress) is evident from individual
figures.
The calculations were terminated when the kinetic energy of
the structure was less than 1.10-6 kJ and while the residual
forces of all the degrees of freedom were less than the limit
value Rlim (defined individually for each example). The
maximum number of iterations is chosen according to the
elements max. 500000 iterations for bar elements max. 5000
iterations for cable and membrane elements. The time step is
chosen t = 1 s in all calculations.
Self-created scripts in MATLAB 7.14.0.739 (2012a) was
used for all calculations. The calculations were carried out on
the computer ASUS processor AMD E-450 APU 1.65 GHz,
memory 4GB RAM.
The results of the calculations (the number of iterations and
the CPU time) are presented in Tables 1 to 9.

4.1

Example 1

A suspended cable ring shown in Figure 4, which has been


discussed in [3], is analysed here to show the accuracy and
speed of the computations developed in this paper.

Table 1. The number of iterations Example 1.


scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

bar
1420
128759
278
484
546
270
385
535

catenary
39894
14874
114559
16680

cable
1266
799
913
1591
777
904
1554

Table 2. Time of solution (CPU time in seconds) Example 1.


scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

4.2

bar
0.72
52.80
0.28
0.30
0.36
0.28
0.30
0.38

catenary
63.60
28.80
184.80
36.20

cable
306.20
249.20
278.90
527.30
247.50
275.20
539.50

Example 2 Hypar net

The example is taken from Lewis [12]: Two sets of straight


line cables generate a model of a hyperbolic paraboloid
surface, shown in Figure 5.

Figure 5. Topology of Hypar net.

Figure 4. A plan view of Example 1.


The structure consists of 16 cables connected to 16 joints
(1-8 are free, A-H are fixed) with an inner radius of 35 m and
outer radius of 75 m. The structures have 8 radial cables and 8
tangential cables. All cables have the same cross-sectional
area A = 1.9634410-3 m2 and the same Young's modulus
E = 170 GPa. The slack length s0 of all radial cables is 40 m,
and that of all ring cables is 32 m. Any external node loads

The structure has 36 degrees of freedom. The load Piz of


0.0157 kN is applied at all internal nodes, except for nodes 17,
21, and 22. The cross-sectional area of the cables is 0.785
mm2 and Youngs modulus is 124.8 kN/mm2. The pre-tension
force in all cables is 0.2 kN. The previous analysis and
experimental measurements of the same structure are reported
in [14].
The limit value of a residual force is Rlim = 0.01 kN. The
accuracy of the calculation is approximately 1 mm.

1923

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

Table 3. The number of iterations Example 2.


scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

bar
177
8889
214
197
376
215
209
252

catenary
1042
1176
3078
1161
2894

cable
135
5920
149
139
237
145
165
234

The parameters of cables are as follows:


cables 1 7: E = 140 GPa, A = 230 mm2, q = 0.021 kN/m
cables 8 30: E = 140 GPa, A = 180 mm2, q = 0.017 kN/m.
The following cable lengths (selected cables) were
considered:
cable 1 11.221 m, cable 2 13.618 m, cable 3 14.340 m, cable
4 13.020 m, cable 5 14.530 m, cable 6 13.820 m.
The load is applied at joints 18 and 19 where the forces
exerted by the canvas act. These forces amount to:
P18 z = 12 kN and P19 = 16 kN.
The limit value of a residual force is Rlim = 0.1 kN. The
accuracy of the calculation is approximately 5 cm.

Table 4. Time of solution (CPU time in seconds) Example 2.


scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

4.3

bar
0.12
6.90
0.15
0.14
0.32
0.15
0.15
0.32

catenary
3.40
4.10
12.10
3.73
11.39

cable
52.00
3246.70
55.80
52.80
88.40
55.00
66.90
86.90

Example 3 Tram stop

Furthermore, one real structure was tested, see Figure 6.

Figure 7. The perspective view of Barrandov tram stop.

Figure 6. Barrandov tram stop.


It is a cable-membrane structure with a central symmetry,
only one fourth of the structure was subjected to modelling.
The canvas was represented by forces exerted by the
structures own weight and prestressi in the model. The
background for the creation of the model was the geodetic
survey of the existing structure made in November 2013, see
Figure 7. The measurement of forces in cables has been
discussed in [15].
The anchoring of the cables into anchor blocks was
modelled as a fixed support. The point where the cables were
anchored into anchor blocks was modelled as a sliding joint
allowing the joints motion only in the direction of the x and y
axes. The structural scheme of the structure considered is
evident from Figure 8.

1924

Figure 8. A plan view of a tram stop.


Table 5. The number of iterations Example 3.
scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

bar
488
1353
136
115
309
125
130
332

catenary
11461
2767
5320
2880
4446

cable
448
1290
120
121
301
118
135
313

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

Table 6. Time of solution (CPU time in seconds) Example 3.


scheme/element
viscous
A
damping
B
kinetic
C
damping
D
t/2
E
kinetic
F
damping
G
t
H

4.4

bar
0.36
1.11
0.12
0.11
0.45
0.10
0.12
0.33

catenary
38.60
9.72
23.56
10.00
19.81

cable
175.40
475.10
59.60
60.40
130.20
58.20
64.20
132.30

It is a membrane structure with 15 nodes (of which 11 are


unsupported) interconnected with 16 membrane elements.
This structure is shown in Figure 10. The load P = 0.75 kN.
The parameters of membranes are always E = 500 MPa,
d = 1 mm and = 0.3.
The limit value of a residual force is Rlim = 0.1 kN. The
accuracy of the calculation is approximately 1 mm.
Table 8. The number of iterations and time of solution (CPU
time in seconds) Example 5.
scheme
viscous
damping
kinetic
damping

Example 4

It is a membrane structure with 6 nodes (of which 2 are


unsupported) interconnected with 4 membrane elements. This
structure is shown in Figure 9. The load Piz = 3 kN acts on
both unsupported joints. The parameters of membranes are
always E = 500 MPa, d = 1 mm and = 0.3.
The limit value of a residual force is Rlim = 0.1 kN. The
accuracy of the calculation is approximately 1 mm.

t/2
kinetic
damping
t

4.6

A
B
C
D
E
F
G
H

number of
iterations
224
3285
231
113
189
249
114
199

time of
solution
2.82
34.20
2.35
1.14
1.92
2.81
1.30
2.34

Example 6

It is a membrane structure with 65 nodes (of which 57 are


unsupported) interconnected with 96 membrane elements. The
topology and initial geometry of this structure is shown in
Figure 11.

Figure 9. Topology and initial geometry of Example 4.


Table 7. The number of iterations and time of solution (CPU
time in seconds) Example 4.
scheme
viscous
damping
kinetic
damping

t/2
kinetic
damping
t

4.5

A
B
C
D
E
F
G
H

number of
iterations
66
981
42
42
43
24
24
26

time of
solution
0.32
2.45
0.27
0.27
0.28
0.21
0.20
0.18

Example 5

Figure 11. Topology and initial geometry of Example 6.


The load Piz = 10 kN for internal joints and Piz = 5 kN for
all external joints. The parameters of membranes are always
E = 500 MPa, d = 1 mm and = 0.3.
The limit value of a residual force is Rlim = 1 kN. The
accuracy of the calculation is approximately 2 cm.
Table 9. The number of iterations and time of solution (CPU
time in seconds) Example 6.
scheme
viscous
damping
kinetic
damping

t/2
kinetic
damping
t

A
B
C
D
E
F
G
H

number of
iterations
617
7065
449
781
1310
1427
1641
1222

time of
solution
33.90
345.60
20.70
39.00
62.80
66.20
76.10
59.00

Figure 10. Topology and initial geometry of Example 5.

1925

Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014

Table 13. The summary of results membrane element.

FINAL COMMENTS

The overall ranking of methods sorted by the number of errors


(sum of all examples), the total number of iterations and the
total CPU time are shown in Tables 10 to13.
The results imply that that it is impossible to determine
clearly the best scheme. In this context, it may be noticed that
the methods based on kinetic damping appear more stable and
faster, which confirms the conclusions presented in [1,2,13].
For bar element, catenary and cable elements the results
confirm that it is beneficial to divide the same amount of mass
into all nodes of the structure proportionally to the stiffest
node of the solved structure (schemes C and F). For
membrane element it is preferred to use the kinetic damping
method with the approximation of the kinetic energy peak in
the middle of the time step t .
6

CONCLUSIONS

It may be concluded that the Scheme C based on kinetic


damping with a peak in the middle of the time step and the
equal mass divided into all nodes proportionally to the stiffest
node has proved the most comprehensive results.

scheme
viscous
damping
kinetic
damping

t/2
kinetic
damping
t

A
B
C
D
E
F
G
H

number of
iterations
907
11331
722
936
1542
1700
1779
1447

time of
solution
37.04
382.25
23.32
40.41
65.00
69.22
77.60
61.52

errors

rank

0
0
0
0
0
0
0
0

2
8
1
3
5
6
7
4

ACKNOWLEDGMENTS
The results presented in this paper are outputs of the research
project P105/11/1529 - Cable - membrane structures analyses
supported by Czech Science Foundation and project
SGS14/029/OHK1/1T/11 - Advanced algorithms for
numerical modelling in mechanics of structures and materials
supported by the Czech Technical University in Prague.
REFERENCES

Table 10. The summary of results bar element.


scheme
viscous
damping
kinetic
damping

t/2
kinetic
damping
t

A
B
C
D
E
F
G
H

number of
iterations
2085
139001
628
796
1231
610
724
1119

time of
solution
1.20
60.81
0.55
0.54
1.13
0.53
0.57
1.03

[1]

errors

rank

0
0
0
0
0
0
0
0

7
8
2
4
6
1
3
5

viscous
damping
kinetic
damping

t/2
kinetic
damping
t

A
B
C
D
E
F
G
H

number of
iterations
12503
0
43837
0
23272
118600
0
24020

time of
solution
41.95
0
77.42
0
64.46
198.55
0
67.40

viscous
damping
kinetic
damping

t/2
kinetic
damping
t

1926

A
B
C
D
E
F
G
H

number of
iterations
1849
7210
1068
1173
2129
1040
1204
2101

time of
solution
533.65
3721.75
364.60
392.12
745.88
360.81
406.23
758.71

[4]

errors

rank

[6]

1
3
0
3
0
0
3
0

5
8
3
8
1
4
8
2

[7]
[8]
[9]
[10]
[11]

Table 12. The summary of results cable element.


scheme

[3]

[5]

Table 11. The summary of results catenary.


scheme

[2]

errors

rank

0
1
0
0
0
0
0
0

5
8
2
3
7
1
4
6

[12]
[13]

[14]
[15]

M. Httner, J. Mca and P. Fajman, Analysis of Cable Structures using


the Dynamic Relaxation Method, in B.H.V. Topping, P. Ivnyi,
(Editors), Proceedings of the Fourteenth International Conference on
Civil, Structural and Environmental Engineering Computing, CivilComp
Press,
Stirlingshire,
UK,
Paper
145,
2013.
doi:10.4203/ccp.102.145
M. Httner, J. Mca, Membrane structures - dynamic relaxation,
Proceedings of the 4th Conference Nano & Macro Mechanics. Prague,
VUT, 2013, pp75-82. ISBN 978-80-01-05332-4.
M. Httner, J. Mca, Cable structures - numerical analysis. Proceedings
of the 3rd conference: NMM 2012 Nano & Macro mechanics. Prague,
VUT, 2012, pp 81-88. ISBN 978-80-01-05097-2.
H. Deng, Q.F. Jiang, A.S.K. Kwan, Shape finding of incomplete cablestrut assemblies containing slack and prestressed elements, Computers
and Structures, (83), 2005, 21-22, pp 1767-1779, ISSN 0045-7949.
J. Kadlk, Statika nosnch lan visutch stech. Academia,
Prague, 1990, 160 p. ISBN 80-200-0251-0 (in Czech).
J.H. Argyris, Recent Advances in Matrix Methods of Structural
Analysis, Progress in Aeronautical Sciences. Pergamon Press, London,
1964, 4. ISSN 0376-0421.
M.R. Barnes, Form and stress engineering of tension structures,
Structural Engineering Review, 1994, 6 (3-4), pp 175-202. ISSN 09525807.
B.H.V. Topping, P. Ivnyi, Computer Aided Design of Cable Membrane
Structures. Saxe-Coburg Publications, Kippen, Stirlingshire, Scotland,
2007, 233p, ISBN 978-1-874672-11-1.
O. Zienkiewicz, R. Taylor, The finite element method volume 1 the
basis. Butterworth-Heinemann, Oxford, England 2000. 125p. ISBN 07506-5049-4.
A.S. Day, An introduction to dynamic relaxation. The Engineer. January
1965, pp218-221, ISSN 0013-8029.
P.A. Cundall, Explicit finite-difference methods in geomechanics,
Proceedings E.F. conf. numerical methods in geomechanics.
Blacksburg, 1976, pp132-150.
W.J. Lewis, Tension structures: Form and behaviour, Thomas Telford,
London, 2003, 201p, ISBN 0-7277-3236-6.
M. Rezaiee-Pajand, S.R. Sarafrazi, H. Rezaiee, Efficiency of dynamic
relaxation methods in nonlinear analysis of truss and frame structures,
Computers and Structures. 2012, (112-113), s 295-310, ISSN 00457949.
W.J. Lewis, The Efficiency of Numerical Methods for the Analysis of
Prestressed Nets and Pin-jointed Frame Structures, Computers and
Structures, 33, pp791-800, 1989.
P. Fajman, M. Polk, Dopnn lan nosn konstrukce zastvky
Barrandov. Konstrukce, 1, 2014, ISSN 1803-8433 (in Czech).

Das könnte Ihnen auch gefallen