Professional Documents
Culture Documents
To cite this article: Iria Heitel (Engineer) & Feng Fu(Senior Lecturer) (2021) Form Finding and
Structural Optimisation of Tensile Cable Dome Using Parametric Modelling Tools, Structural
Engineering International, 31:2, 271-280, DOI: 10.1080/10168664.2020.1750937
Fig. 1: Code design diagram for using Grasshopper to generate the tensile structure
Fig. 5: Code for generating the second ring of cables and struts in tensile dome
Conventional Methods
For conventional methods, tracking
changes can be complex and maintain-
ing the quality of the design in complex
structures with a tight deadline can be
especially challenging. More time is
required, and usually several models
have to be built, to test the design
options.
Case Studies
Fig. 8: CENTRAL TRUSS definition generated by Grasshopper (green)
A parametric model of a tensile ellipti-
cal dome was created in Rhino 3D
Grasshopper, as shown in Fig. 9. This
The initial sizes of members can be cal- under such modifications can be replicates the existing tensile structure
culated to assess the initial design. checked. For example, the form- Georgia Dome (the dome was demol-
Changes are adjusted automatically, finding method could be thought of as ished). Different parameters were set
saving time along the design pipeline. a memory foam mattress that will out allowing variations in the size of
As a result, the quality of the design take the shape of your body when the elements and their connectivity.
is enhanced. In the case of tensile you lie on it. In the case of tensile struc- The resulting free-form geometry was
cable domes, this is a great advantage tures, the prestress force will define the then rationalised by changing par-
because it allows determination of a shape of the net, so understanding how ameters through the pipeline code in
set of parameters to assess feasibility the forces are distributed is key to the Grasshopper. The form finding and
and optimise the individual members design. Form finding with the means optimisation results can be changed
and the structural system, allowing of parametric design will allow this freely and quickly in each iteration by
changes at any time. Form finding opti- conversation between forces and varying the parameters. An initial
mises the right level of prestress force shape. The benefits of implementing form finding of different options was
by the way in which loads are applied parametric design in the early stages conducted with the plug-in Kangaroo
to the system. of construction are that creative think- using the dynamic relaxation method.
ing and rationalisation are involved The results were compared and then
Parametric modelling can simulate the early in the design process. It requires exported to GSA for further form
behaviour of the structure under thinking about and analysing the finding, which allowed an initial sizing
loading and make initial calculations system and its parts, and computer pro- of the elements with the force density
to study the feasibility of the design gramming offers a set of tools to create method. Further nonlinear analysis
options. This can be visualised and an algorithm that will assist with those was then carried out and the results
changes to the geometry can be per- changes. So, when coding the script, were compared. Therefore, an iterative
formed so that engineers can instantly one is addressing the body of the process of form finding was carried out
gain a sense of force and deflection design and the assumptions that are with the objective of minimising the
changes in the global structural being made, while identifying the vari- sizes of the cross-section areas of
system when the geometry is modified. ables. With the movement towards pre- cables by distributing the force
The efficiency of the whole system fabrication, offsite construction and 3D density ratios. The rigidity and stiffness
Model 1
It can be seen that, after loading, the
forces on the ridge cables decrease,
especially those on the top and inner
layers. The forces on the outer and
lower layer of diagonal cables increase,
while for the inner and upper layer the
Fig. 9: Model 1 in Rhinoceros 3D forces decrease. The forces on the
hoop cable increase more rapidly on
the bottom layer.
of the dome comes are provided specifications, if a ridge cable on the The results in Fig. 13 show that Model 1
through the equilibrium of internal outer layer has a cross-section area of presents considerable vertical displace-
forces between cables and struts. To 8540 mm2, the maximum stress ment and deformation. The distribution
avoid slackening of the cables, a high reached before failure will correspond of the stresses and dimensions of the
level of prestress is required. This will to 158,844 kN. The struts are S275 cir- cables and struts need to be further
determine the sizes of the cables, and cular tubes in compression. adjusted. The necessary changes are
thus the cost of the materials and the performed in the parametric model in
construction process. It was determined that an initial pres- Grasshopper. Longer struts are added
tress for averaging 30% of cable tensile and higher prestresses are considered.
Meanwhile, different forms of cable capacity was needed. Thus, a starting The central zone of the tensile dome
domes, Models 1, 2 and 3, are proposed value of 30% was taken to define the needs to be reinforced with vertical
and compared. The topology of the initial force density value and the itera- struts, and adding another concentric
different models created was associ- tive process of form finding.19 For a ring of ridge cables and diagonal cables
ated with different force density ridge cable with a cross-section area of seems to be a good idea since the hori-
ratios. A nonlinear analysis was per- 8540 mm2, the initial force density zontal distances between the struts are
formed and the results were compared. value q = fy * cross-section area * 0.3 = too large and the cables need to be
4279 kN was taken. highly prestressed to maintain the
For comparative study purposes, the
A live load of 0.6 kN/m2 is applied on stresses.
dimensions of all the models have the
same height 30 m, width 192 m, length the roof, with a projected wind load of
240 m and 20 subdivisions in each 1 kN/m2. The nonlinear analysis is Model 2
layer. As the dome has an elliptical defined for a combination of the The new Model 2 is an improvement
shape, the cables on the same layer internal load due to prestresses and on Model 1. An extra layer of struts
have different lengths. The dome has a the dead load of 1 kN/m2. A and ridges and diagonal cables is
triangulated geometry similar to Levy’s
Georgia Dome and has a span of 240 m.
Model 3
Model 3 is a further improvement on
Model 2 where more cables were
added. The central zone is again the
weakest part of the structure. The
design process has the main objective
of strengthening the central part of
the dome.
Figures 17–19 show the analysis results
of Model 3. The initial force density
result is in the range of 4278–1092 kN
for the ridge and diagonal cables, and
Fig. 15: Static analysis results (reactions) of Model 2 13,416–3 787 kN for the tension hoop
cables. After form finding in Oasys
GSA, it has increased to 60%, with a
maximum of 26,822 kN, compared to
the initial 30% of the prestress level
of a cable. The stresses are distributed
through a more homogeneous
network. The maximum vertical
displacements have decreased with
the new geometry and prestress
distribution. By comparison, under
the same load conditions, Model 3 per-
formed better than Models 1 and 2,
mainly owing to the higher
prestresses of the cables. The weakest
part of the dome is the central zone.
Strengthening the central section of
the dome, especially the ridge and
diagonal cables in the central section,
Fig. 16: Static analysis results (vertical deflection Uz) of Model 2 was necessary.
Conclusions
In this paper, a new parametric tool for
form finding and design optimisation of
tensile structures is presented. Using
the new tool, a prototype model
which replicates the Georgia Dome is
first designed and analysed. Based on
the parametric form finding and optim-
isation results, this dome is further
improved and two new elliptical
tensile domes with new geometric con-
figurations are designed.
Fig. 18: Static analysis results (displacements Uz) of Model 3
The following conclusions can be drawn:
. Computational design using para-
metric tools allows an initial design
evaluation of multiple form options,
facilitates an intuitive design
exploration of the topology, and
drives the structural optimisation
and exploration of the design in a
more efficient way.
. Any improvements in the design of the
tensile dome will focus on strengthen-
ing the ridge and diagonal cables of
the inner and upper parts of the
central section of the dome while mini-
mising the use of steel in the structure.
. The two new geometric configurations
of elliptical tensile domes were devel-
Fig. 19: Static analysis results (reactions) of tensile dome Model 3 ] oped, primarily to strengthen the
central part of the dome. The new
types of tensile domes exhibit better
Analysis of Results and Discussion The configuration of the cable net can load-bearing features.
be either in a wedge shape or in a tri-
In elliptical domes, the weakest part of
the structure is the central region angular shape. A combination of both
will improve the stability and perform- Disclosure statement
between the centre point and the first
hoop cable. Slackening of the cables ance of a tensile dome.
No potential conflict of interest was
may cause failure. This applies The benefits of implementing para- reported by the authors.
particularly to the ridge and diagonal metric design in the early stages of con-
cables. The failure occurs mainly struction are that creative thinking and
by breaking of the hoops and rationalisation are involved early in the ORCID
diagonal cables, buckling of the struts, design process. This requires thinking Feng Fu http://orcid.org/0000-0002-
or both. about and analysing the system, and 9176-8159