Experimental Techniques (2022) 46:823–834

https://doi.org/10.1007/s40799-021-00509-6

RESEARCH PAPER

Experimental and Numerical Study of a Geodesic Dome Under Static

and Dynamic Loads and the Influence of Nodal Connections
D. Rossot1 · R. D. Machado2 · N. Barbieri1 · K. F. de Lima1

Received: 25 August 2020 / Accepted: 2 September 2021 / Published online: 28 October 2021
© The Society for Experimental Mechanics, Inc 2021

Abstract
Geodesic domes are 3D reticulated structures used to cover large areas. As lightweight structures, they are sensible for
dynamic effects. The structural behaviour of domes is affected by the kind of connections between their members. Semi -rigid
joints produce loss of stiffness compared with rigid joints. Dynamic tests may capture the effect of semi-rigid joints. The
objective of this work is to conduct both static and dynamic tests on a Geodesic Dome at the laboratory scale and compare
the results with a linear elastic FE model. The research method involved designing, building and testing a prototype con-
structed of aluminum wit h overall dimensions of 2000 mm in diameter and 800 mm in height. A 3D finite element model was
developed using beam and shell elements. Experimental and numerical results were compared in terms of displacements and
frequencies. Load—displacement curves are presented for the static loading. The dynamic behaviour is evaluated by hitting
an impact hammer on members and nodes to allow measurement of accelerations. An FFT technique is used to determine
the frequency spectrum. The results show that the laboratory prototype presented about 20% less stiffness than predicted by
the linear elastic numerical model. The difference is attributed to the transmissibility condition between nodal connectors
and emphasizes the influence of semi-rigid joints. Despite being largely used to assess dome structures, standard FE models
are shown to be insufficient in capturing local non-linearities.

Keywords Geodesic dome · single layer lattice dome · Transmissibility condition · semi-rigid joints

Introduction Some examples of geodesic domes are the Nagoia Dome,

the Astrodome, and the Cinesphere.
Geodesic domes are suitable solutions for large structures The nodal points of a geodesic dome are placed on geo-
such as gymnasiums, stadiums, tank covertures and those desic curves over a spherical surface. Therefore, the curves
which do not need any intermediate supports. As they are connecting two nodal points have the shortest possible
lightweight structures, dynamic and instability effects must length on the surface of sphere and the straight members
be carefully considered. Geodesic domes are constructed connecting the same two points also have the shortest length,
with straight members mutually connected by nodes forming contributing for making the structure lighter [3].
triangular panels. The resistance/weight ratio is greater than Currently, one of the most important applications of geo-
other structures, which allows to cover spans larger than 300 desic domes is the coverage of oil reservoirs, as indicated in
m without intermediate columns, as commented by [1, 2]. Fig 1. Due to the volatility of the fuel, most reservoirs have
a movable coverage whose height is function of the level of
fluid stored. Since the coverage is movable, there is a risk of
* R. D. Machado infiltration of water inside the tank affecting the fuel quality.
rdm@ufpr.br Therefore, an additional coverage such as a geodesic dome
1
Department of Mechanical Engineering, PolytechnicSchool, is recommended to avoid this problem.
Pontifical Catholic University of Paraná, Av. Domes are 3D structures that have been studied in the
Imaculada Conceição, 1155, Prado Velho, Curitba, literature through different aspects. The progressive col-
Paraná CEP 80‑125‑901, Brazil lapse of domes produced by static loads is one of the main
2
CESEC – Civil Engineering Study Center, Centro Politécnico topics addressed. Abatan and Holzer [4] studied the insta-
‑ Bairro Jardim das Américas, Federal University of Paraná, bility of geodesic domes considering independent loading
Curitiba, PR, Brazil

Vol.:(0123456789)
824 Experimental Techniques (2022) 46:823–834

It is important to emphasize that the structural behav-

iour is affected not only by the mechanical properties and
material parameters but also by the rigidity of joints. This
subject is addressed in many recent works as shown in Chen
[24], López et al [25], Alvarenga [26], Ma et al [6], Ma et
al [27], Han et al [28], Silva et al [29], Ye and Lu [18]. One
is observed that the angle between members located along
the same meridian line has a relevant effect on the load-
carrying capacity of domes [25]. Most research deal with
the semi-rigidity of joints for single layer lattice domes for
instability analyses due to progressive loads [27, 28]. The
Fig. 1  A geodesic dome covering a cylindrical tank of oil
structural behaviour is affected by a non-linear moment cur-
vature relationship [29]. The rigidity of member connections
parameters. Considering the initial curvature of the mem- is a local effect, and it is not expected to be captured by a
bers, the elastic-plastic instability for single layer domes was standard finite element model. In fact, initial imperfections
investigated by Fan et al [5]. Ma et al [6] developed a stabil- of semi-rigidly jointed connections can also lead to struc-
ity analysis for single-layer dome addressing the effect of tural collapse, as discussed by Kato et al [30]. In a similar
semi-rigid socket joints. Yan et al [7] analysed how occur the way, Morris [31] discussed the effect of imperfections on the
instability of a single layer reticulated dome. Wei et al [8] stability of lattice shells. These problems make the domes
investigated the progressive collapse of single layer domes more sensible for buckling, increasing them structural vul-
over long spans. Xu et al [9] also studied the progressive nerability, as reveal the studies from Yamada et al [32] and
collapse of single layer Lattice Domes developing experi- Nanhai and Jihong [33]. Ma et al [6, 27] developed tests with
mental and numerical models. The progressive collapse is socket joints and used the experimental results to consider
also tackled by Tian et al [10] in a special kind of domes, the semi-rigid connections in single layer domes. They used
named Kiewitt. Spherical coverages, including tanks, roofs, the results for developing a parametric study of single layer
and marine structures, were examined by Grünitz [11], who domes, taking into account the semi-rigid joints on stability
analysed the effect of welding on the buckling behaviour of analyses.
these structures. Chen et al [12] tested a suspendome failure However, the nodal semi-rigidity of 3D domes is not an
due to a cable rupture. Piotrowski and Szychowski [13] stud- easy task to evaluate. It depends on the kind of each member
ied lateral torsional instability of elastic beams restrained at connection and its relative position over the dome surface
the support nodes, which is a subjected to be considered in [28]. It means that each node will have a different behaviour
the dome structural analysis. in terms of mom ent- curvature relationship. This theme has
Many studies aim to find the minimum weight for a dome been considered for many studies for instability structural
applying topological optimization techniques. Saka [14] analysis. However, in the case of dynamic analysis, this is
used a genetic algorithm to get the optimal solution of sin- an important challenge to be attacked and the issue is not
gle layer domes considering nonlinear geometrical effects. investigated as desired yet.
Pokusinski and Kaminski [15, 16] focused the reliability of Although most of the commercial structural software can
Lattice domes via analytical and semi-analytical methods. analyse any kind of dome structures, the question is how
The reliability of steel structures subjected to fire are dis- the computational model can represent their real behaviour.
cussed by Kubicka et al [17], taking into account solutions How the connections between the members are represented
for connecting 3D elements. Ye and Lu [18] analysed a set by standard finite element models? Are the standard FE
of semi-rigid joints and applied optimization techniques to models able to capture the local effects for all connections
design domes against instability. The influence of the mesh in a geodesic dome? The dynamic performance of geodesic
refinement in terms of the division of spherical triangles domes needs to be better investigated. This is the main moti-
for geodesic domes affects the critical load capacity of the vation for developing this work.
structure, as discussed by Zabojszcza and Radon [19]. In this paper, static and dynamic tests of a geodesic dome
Regarding dynamic behaviour of domes, Szaniec and are considered. A physical prototype was built in laboratory
Biernacka [20] developed a modal analysis for selected bar scale, which is related to a real coverture over an oil tank like
domes. However, a relevant aspect to be considered is the the one shown in Fig 1. Experimental tests and numerical
type of joints. Kubik [21] investigated the static behaviour of simulations are carried out and the results are compared and
geodesic dome joints while Barbieri et al [22] investigated analysed. The numerical model is based on the Finite Ele-
their dynamic behaviour. Yan et al [23] suggested a new type ment Method, and 3D beam elements are adopted for each
of truss joint for prevention of progressive collapse. member of the structure. Shell elements are used to model
Experimental Techniques (2022) 46:823–834 825

the ring support cylindrical. Three load cases are consid- present research shows how mitigate this limitation and the
ered here for static analysis. Vibration tests are carried out reached results validate the global loss of stiffness observed
by hitting an impact hammer, measuring the accelerations, in the experimental tests.
and evaluating the frequency spectrum. A large range of
natural frequencies is identified, and they are compared with The Laboratorial Prototype of Geodesic Dome
numerical results. The influence of semi-rigid joints in the
global dynamic behaviour is emphasized. An aluminium geodesic dome for laboratorial prototype was
The main interest of dynamically analysing a geodesic conceived and built-in small scale, Fig 2(a)-(d). From now
dome is to give better understanding about its structural on, the real prototype will be addressed as “physical model”
behaviour. As geodesic domes are lightweight structures, to distinguish it from the “numerical model”. The dimen-
they are more sensible to vibrations and dynamic loads sions of the physical model are: 2000 mm of diameter and
than heavier structures [44]. The modal analysis gives us a 245 mm of height. The bulging radius, as suggested by [34]
diagnosis about the actual properties of the structure. There- for oil tanks, should be between
fore, natural vibration modes and frequencies can be used 0.8 to 1.2 times the tank’s diameter. For this prototype, it
as reference for the real state of the structure. The lower is adopted a bulging radius 1.08, that is, 2161.67 mm. The
the frequencies, the more flexible the structure will be and, design of the laboratorial dome also follows the scale 1:12
consequently, the more susceptible to dynamic behaviour. when compared with a real dome as depicted in Fig 1. All
Another related issue is that very flexible structures are also members are L-shape cross section with 40.132 m ­ m2 of area,
susceptible to elastic instabilities [45]. Therefore, geodesic which corresponds to angles equal-leg of 12.7 mm length
domes must be designed carefully since they are supposed as and 1.58 mm thickness. Figure 2(d) shows the nodal con-
light as possible and can be submitted to dynamic and elastic nection in the prototype model. The dome was built over a
instability effects. Due to these characteristics, predictive circular ring with the aim to represent the side of the oil tank
maintenance is recommended for this type of structures. One as can be seen in Fig 2(a). The height of the ring is 300 mm
of the most popular techniques for structural health control and its thickness is 1.6 mm.
are the vibration tests, which can determine the actual state The geometry of the physical model is defined consider-
of the structure by identification of vibration parameters. ing the math procedure for generation of geodesic domes,
Any chang e in the measured frequencies, for example, can taking into consideration the kind of Platonic solid, the
mean that the overall state of the structure is changed, and frequency and number of sides of the solid. The software
some preventive actions should be taken [46]. Therefore, the CADREGEO [43] was used to define the geometry and
knowledge about vibration modes is something essential for topology of the dome under the parameters defined above.
structural safety of geodesic domes. It is adopted an icosahedron, class 1 and frequency 9. Details
It must be highlighted that advanced numerical models of this procedure can be found on the works from Kenner
based on 3D nonlinear finite elements could be applied to [35], Clinton [36], Shirley [37], Diniz [38] and Rossot [39].
quantify the transmissibility in each connection, taking into The cross section of the members was designed to ensure
account the friction among the members and connectors the local and global elastic equilibrium of the prototype, for
[22]. However, the transmissibility condition is affected by the lo ad cases applied, and for avoiding plastic deformations
the node orientation, since each node of dome has a differ- on each member. A linear instability analysis showed that
ent spatial orientation, with normal vector that points in a the global buckling limit load would be equal to 561.6 N,
different direction from the others. Therefore, to including which is much greater than the experimental applied loads.
the semi -rigidity of each node may become an unfeasible That is way, neither local n or global collapse was expected.
task when it should be considered for all nodes for 3D dome. Instead, the structural behaviour of the prototype should be
Instead, combining static and dynamic tests, as in the present elastic and linear. Circular disks, screws and nuts were used
work, it is possible to check the loss of global stiffness due as connectors between the L-shape bars. The diameter of
to semi-rigid connections, which gives a real magnitude of the circular disks connections is 33 mm and their thickness
the general loss of rigidity produced by the total number of is 1.5 mm.
nodes.
The present work uses a commercial finite element soft- Numerical Model
ware, Solid Works, to carry on static and dynamic analy-
ses of the dome. Non-linear effects or large displacements The finite element model was developed using the commer-
approaches are not considered, as well it is not possible to cial CAD-CAE SolidWorks Premium software for static and
model local effects for se mi- rigid connections for nodes dynamic analyses. 3D frame elements are used for straight
of the dome. Despite this limitation, the linear approach is members and triangular shell elements are used for the ring
adopted widely in the daily routine of structural design. The support, each o ne with 42 mm of length. The mesh size was
826 Experimental Techniques (2022) 46:823–834

Fig. 2  The experimental model

(a) Top view (b) Bottom view
(c) Top view of a joint (d) Bot-
tom view of a joint

(a) (b)

(c) (d)

defined after some convergence tests. For static analysis, it ⎧ u(x, y, z;t) ⎫
is enough to model the dome coverture. However, for the ⎪ ⎪ � �
vibration analysis, it is essential to model the ring support
u(x, y, z;t) = ⎨ v(x, y, z;t) ⎬ = N(x, y, z) {U(t)} (1)
⎪ w(x, y, z;t) ⎪
together with the geodesic coverture. ⎩ ⎭
The area and the moment of inertia of the mem-
bers are related to L-shape sections with dimensions of where [N] is the matrix of shape functions. Defining {𝜀}
12.7 × 12.7 × 1.58 mm. as the vector of small strain; {σ} the vector of stress; [B] the
The material is aluminium alloy 1060. The constitutive strain-desplacement matrix and [D] the constitutive matrix,
relationship is elastic linear. The Young’s Modulus is 70 then:
GPa. All sup port nodes are restricted to displacements in {𝜀} = [B]{U} (2)
three directions as indicated in Fig. 3.
For a linear elastic finite element model, taking u =
u(x,y,z;t) as the displacement vector of a specific point and
{𝜎} = [D]{𝜀} (3)
U(t) the vector of nodal displacements, it is possible to write The classical equation of movement is given, in indicial
acording [40–42]: notation, as:

Fig. 3  The numerical models

and the boundary conditions (a)
without ring and (b) with ring

(a) (b)
Experimental Techniques (2022) 46:823–834 827

[ ]
∭
𝜎ji,j + bi = 𝜌ü i i, j = 1, 2, 3 (4) Ke = [B]T [D][B]dV (8)
Ve
In Equation (4), σ is the Cauchy’s stress tensor, b is the
body’s force vector, ρ is the mass density of the material, and The solution of Eq. (6) is obtained solving an generalized
u =u(x,y,z,t) is the displacement vector, with twin upper dots eigenvalue problem, that means,
meaning the ­2nd time derivative or acceleration vector. The [ ]
weak form of Equation (4) may be derived for a master ele-
[K] − 𝜔2 [M] {𝜆} = 0 (9)
ment in matrix form reaching: where, for free vibration problems, the pair (𝜔 , {𝜆}) are
( ) the natural frequencies and the vibration modes vector. As
∫
{𝛿𝜀}T {𝜎} + {𝛿u}T 𝜌{u}
̈ dV = 0 (5) the pair (𝜔, {𝜆}) depends on the mass and the stiffness matri-
Ve
ces, then they just depend on the mechanical and geometri-
The terms on the left-hand side of Eq. (5) are the inter- cal properties.
nal force and the inertial force, δ means a variation, and the
volume integration is carried out over the element volume
Ve. Combining (4) and (5) and named {u} and {ü} as the Materials and Method
nodal displacement and acceleration vectors, in the case of
undamped effects and supposing no external loads, yields: Test devices
[M]{u}
̈ + [K]{u} = 0 (6) For static tests, small circular weights are used for applying
where [M] and [K] are the global mass and stiffness matri- forces on the physical model. The weights are gently put on
ces developed from the assembling of the element mass and thin plates for applying forces in specific nodes through steel
stiffness matrices defined as: wires and pulley system. The nodes that are subjected to
loads are described ahead. The plates weigh 0.64 N each. For
[ ] measuring the nodal displacements, analogical deflectom-
∭
Me = 𝜌[N]T [N]dV (7)
Ve eters are used. Figures 4(a) and (b) illustrate both situations.
For dynamic tests, an impact hammer is used. Acceler-
ometers are fixed in some members to capture the dynamic

Fig. 4  Applying loads and

measuring their effects (a)
Nodal loads; (b) Displacement
Deflectometer (c) Impact ham-
mer (d) Accelerometers

(a) (b)

(c) (d)
828 Experimental Techniques (2022) 46:823–834

effects. The technical characteristics of these devices, such Load Case 2, the force is applied in the upward direction
as the type of model and their sensibilities, are presented in through steel wires and pulleys. In these cases, the displace-
Table 1. Figures 4(c) and 4(d) show them. ments are measured at nodes 26, 28 and 52 as depicted in
orange square marks. Considering the initial loading due to
Static Tests the weight of the plate, additional loads are applied succes-
sively. The first additional load step is equal to 2.0 N for the
The static tests are developed by hanging weights on small Load Case 1, and 7.0 N for Load Case 2. The successive load
plates on any desired node, as can be seen in Fig 4(a). Three steps are of 5 N. The Load Case 3 corresponds to mainte-
different kinds of loads are applied named Load Cases 1, 2, nance service loads and they are applied on nodes 21, 27 and
and 3, respectively, as indicated in Fig 5. Only one force is 31 (green circle marks) while de displacements are measured
applied in the upper point of dome for Load Cases 1 and 2. in nodes 16, 27 and 30 (orange square marks), as indicated
Load Case 3 corresponds to a practical situation which simu- in Fig 5(b). For each load step, the nodal displacements are
late a maintenance task over the dome. In this case, three measured using deflectometers, as illustrated in Fig 4(b).
points are loaded simultaneously. For Load Cases 1 and 2, a
vertical load is applied gradually on the central node, num- Dynamic Tests
ber 52, as indicated by the green circle marks - Fig 5(a). For
Load Case 1 the action is in the downward direction and for Non-destructive dynamic tests are useful to determine some
structural properties as the resonant frequencies [40, 41].
Preliminary, some tests were made for calibrating the acqui-
Table 1  Technical characteristics of the accelerometers and the sition of experimental data system, including the verifica-
impact hammer tion of symmetry of some mode shapes. The arrangement
Description Model Sensibility of accelerometers and the impact hammer are shown in Fig
4(d). The set of accelerometers is placed around the node
Impact hummer 086C03-PCB 2.13 mV/N
number 52. The impact hammer is applied in some distinct
Accelerometer 333A-PCB 84 mV/g
points (such as nodes 6, 8 and 30). After impacted, the dome
Accelerometer 3333B-PCB 109.9 mV/g
is excited by different kind of vibrations, and the structural
Accelerometer 3333B-PCB 94.5 mV/g
responses are recorded by the signal acquisition system.
Accelerometer 3333B-PCB 109.2 mV/g
After that, by one FFT analysis, it is possible to identify
Accelerometer 3333B-PCB 101.1 mV/g
which natural frequencies are excited. This i s enough to
Accelerometer 3333B-PCB 107.9 mV/g
characterize the dynamic behaviour in terms of natural fre-
Accelerometer 3333B-PCB 111.8 mV/g
quencies and modes.
Accelerometer 3333B-PCB 111.5 mV/g

Measured Nodes
Applied loads

(a) (b)

Fig. 5  Static tests: (a) Load Cases 1 and 2; (b) Load Case 3
Experimental Techniques (2022) 46:823–834 829

Table 2  Displacement values (mm) for the numerical and physical Results
models

(a) Numerical Static Results

Displacement (mm)
Load Node 52 Node 52 Node 28 Table 2 presents the vertical displacements of nodes 52,
0 0 0 0 26 and 28 for the numerical and physical models for Load
2.64 -0.05 -0.002 -0.0009 Case 1. These values are depicted in Fig 6. It is possible to
7.64 -0.156 -0.002 -0.001 observe that both models have close results up to 12.64 N
12.64 -0.256 -0.0024 -0.0039 and present the same stiffness. Above this value, the physi-
17.64 -0.356 -0.0022 0.006 cal model changes its behaviour, diminishes the stiffness
22.64 -0.4568 -0.0019 0.008 and presents much bigger displacements than the numeri-
(b) Physical cal model. Since the dome is supposed to be within the
Load Displacement (mm) elastic limit, the suddenly change of stiffness would not
Node 52 Node 26 Node 28 be expected. This behaviour alerts us to a critical problem,
0 0 0 0 which will be addressed further on.
2.64 -0.05 0 0 Similar results are observed for Load Cases 2 and 3 which
7.64 -0.18 0 0 the load displacement curves are presented, respectively, in
12.64 -0.3 0.11 0.01 Figures 7 and 8.
17.64 -0.75 0.05 0.01
22.64 -2.28 0.6 0.01 Dynamic Results

Three dynamic tests are applied on the physical model.

Nodes 6, 8 and 30 are hit separately by the impact

Fig. 6  Load Case 1—Verti-

cal load applied on node 52
(downwards): (a) numerical; (b)
measured
830 Experimental Techniques (2022) 46:823–834

Fig. 7  Load Case 2—Verti-

cal load applied on node 52
(upwards): (a) numerical (b)
measured

hammer which provides the vibration to the structure. For structure, both models should have the same displacements.
the dynamic numerical analyses, only results from the model But, as can be seen in Fig. 6 for Load Case 1, the numerical
with the ring support were considered. It was verified that, model is stiffer than the experimental one. The numerical
due to its elevated position and the support conditions in and physical models are close and present linear behaviour
laboratory, the simplest model, without the ring support, is until 12.64 N. Above this value, the experimental model
not able to capture the frequencies measured well. changes the load displacement curve, presents lower stiffness
Table 3 shows the first fifteen natural frequencies of the than the numerical one with much bigger displacements for
geodesic dome determined by FEM analysis. As can be the same load level. One is observed that the physical model
observed in Table 3, there are five successive modes with presents a new linear behaviour above the 12.64 N load. It
closer frequencies. This happens because the physical model may suggest an accommodation of the physical model, but
was developed from an icosahedron pattern with five buds this fact was not confirmed in laboratory, even after succes-
which are similar between themselves. This explain the mul- sive cycles of load - unload.
tiplicity of frequencies, as can be seen in Table 3. Some unpractical aspects may have contributed to this
After recovering the accelerometers data, a Fast Fourier issue. The connections between members are made by cir-
Transform (FFT) is applied to determine the frequency spec- cular metallic disks whose holes have a diameter of 1.5 mm.
trum. Fig 9 shows the frequency spectrums for accelerom- The screws, however, have 1.39 mm in diameter. This small
eter C1. The spectrum shows a first peak near 108 Hz (cor- difference can produce accommodation of the structure as
responding to frequencies 1 to 5 on the numerical model) the loads arise. Another probable aspect is the tightening
and a second peak close to 142 Hz (which can be related to of the screws. Due to their smal l dimensions, the control
the natural frequencies 6 to 11 in Table 2). of the screws’ tightening was not as precise as desired. A
third aspect is the kind of connections and bars adopted in
the physical model. As the members’ sections are L-shaped,
Discussion some torsional rotation may have occurred. To fix each bar
in the metallic disk, it was necessary to make connections
The initial discussion is about the unexpected difference like those shown in Fig. 2(d) cutting the end of each bar. All
between the numerical and experimental results for static these conditions may have affected the rigidity of the con-
tests. Si nce the dome is supposed to be an elastic linear nections. In this case, the static linear elastic computational
Experimental Techniques (2022) 46:823–834 831

Fig. 8  Load Case 3 – Mainte-

nance loads: (a) numerical; (b)
measured

happens a loss of energy showing peaks of low frequencies

Table 3  The first 15 natural frequencies by FEM that are not captured by the FE model. According to Barbieri
et al [22], there are some reasons for this difference. Among
Mode Frequency (Hertz) Mode number Frequency (Hertz)
number them, the way the pieces are fixed, th e torque applied on the
screws, the member adopted (L shape), and the cut necessary
1 118,93 9 142,95 to join the members on the disk connectors. The relationship
2 119,23 10 143,31 between the loss of dynamic transmissibility and the loss of
3 119,33 11 143,52 stiffness due to semi-rigid joints is out of the scope of this
4 119,57 12 151,26 work and will be addressed in a future study.
5 119,66 13 151,46 A practical solution is to reduce the global stiffness of the
6 142,29 14 152,65 numerical model. This approach is hereafter investigated. In
7 142,34 15 152,79 this work, a stiffness reduction is adopted by adjusting the
8 142,86 elasticity modulus of the material. With the aim of investi-
gating the stiffness reduction effect, four reductions levels
are adopted (5, 10, 15 and 20% of the elasticity modulus).
model is not able to consider all these small but important The ad hoc process was adopted to approximate the struc-
contributions. The results suggest that there is a loss of tural behaviour observed in static and dynamic tests and
global stiffness of the structure due to the nodal accommo- quantify the global loss of stiffness for whole structure.
dations in comparison with the theoretical model. To illustrate the consequences of changing the value of the
In fact, the conventional FE model adopted is not capable elasticity modulus, Fig 10 show, for Load Case 1 the load-
of simulating the tightening of screws and the semi-rigid displacement curve with 0%, 5%, 15% and 20% of reduction
joints. It explains why the computational model presents until 12.64 N for static analysis. Better results were obtained
greater stiffness than the laboratorial prototype. Barbieri et with 20% of reduction. A complete set of tests results can
al (2016) analysed the transmissibility of the nodal joints for be seen in the work of [39]. It is appropriate to comment
this dome. Applying the impact and impedance hammer on that, as can be seen in Fig. 10, there is a slight increase in
selected nodes, it was possible to measure the vibrations on the displacement measured in the physical model for node
nodes close to those impacted. Using an inverse FRF tech- 52. The change in slope of the load × displacement curve
nic and wavelet transf orms, the study concluded that there is due to nodal nonlinearity and suggest an accommodation
832 Experimental Techniques (2022) 46:823–834

Amplitude (g/N)

Frequency (Hz)

Fig. 9  Spectral frequencies for accelerometer C1 when node 30 is hit

of the structure. This characteristic is not determined by the static and dynamic load cases. Static loads were progres-
linear elastic model. sive and slowly applied. Dynamic loads were applied by
For dynamic tests, a reduction of the elasticity modulus impact hammer. Displacements and accelerations were
also provokes a reduction of the frequency’s magnitudes. measured by deflectometers and accelerometers. The fre-
Adopting a reduction of 20% of elasticity modulus, both quency spectrum was also determined. The experimental
physical and computational models brought about almost results were compared with those obtained by a linear
the same results. In the physical model, the first resonant elastic finite element model with 3D beam elements for
frequency is 108.25 Hz and, for the computational model, the geodesic coverture and shell elements for the ring
the frequency is 107.93 Hz. Consequently, based on the support.
static and dynamic results, in practical terms, it is advisable As commented earlier, dynamic parameters are very
to work within the range of 20% reduction in the modulus important for lightweight structures, such geodesic
of elasticity. However, the reduction depends on the type domes, and can be used as reference data for structural
of joint. health monitoring, and for flexibility and elastic sta-
Further studies will be done in order to develop local FE bility control. Therefore, it is supposed that numerical
models and investigate the real degree of trans missibility and measured frequencies are close to each other. This
for each joint, making it possible to evaluate the effective ideal situation was not observed in this study. The physi-
stiffness of the structure. cal model presented a structural and dynamic behavior
slightly different from the numerical one. In the design
phase, a linear elastic behavior of the dome was assumed.
Conclusion However, in the experimental tests, this was not observed.
The experimental results showed a load × displacement
Geodesic domes have been used in many important situ- curve with nonlinear change of slope for loads above cer-
ations, generally large covertures. In order to improve tain value. This may suggest a structural accommodation
safety conditions and achieve cost reductions, the knowl- or effect of nodal semi rigidity. This issue should be fur-
edge about the actual behaviour of this kind of structure ther investigated in future work.
must be enhanced. Therefore, the present work developed One relevant aspect is the joint connections between the
a geodesic dome in laboratory scale and tested it for the members. In this work, the connectors were done by small
Experimental Techniques (2022) 46:823–834 833

Fig. 10  Load displacement

curves with 0%, 5%, 15% and
20% of global stiffness loss –
Load Case 1

disks, screws, and knots, carefully tightened. Some diver- References

gences between experimental and numerical results sug-
gest a loss of stiffness of connections. The transmissibility 1. Ramaswamy GS (2002) Analysis, design, and construction of
of vibrations through the nodes is affected by the degree steel space frames. Londres: Thomas Telford
2. Bysiec D (2013) The Investigation of Stability of Double-Layer
of rigidity of member connections. An additional difficult Octahedron-Based Geodesic Domes. Faculty of Civil and Envi-
is due to the spatial orientation of the joints in the dome ronmental Engineering, Kielce, Poland
whose rigidity under ver tical loads are different node to 3. Santos LB (2005) – A contribution for steel structures study (In
node becoming more expensive a semi-rigid approach for portuguese: Contribuição ao Estudo de Estruturas Metálicas).
Doctorate Thesis, University of São Paulo
the numerical model. 4. Abatan AO, Holzer SM (1978) Degree of stability of geodesic
The results here obtained may support the structural domes with independent loading parameters. Comp & Struct
designers and researchers who are trying to optimize 9:43–51
geodesic domes and other metallic covers, highlighting 5. Fan F, Yan J, Cao Z (2012) Elasto-plastic stability of single-
layer reticulated domes with initial curvature of members. Thin
the importance to consider the stiffness reduction for Walled Struct 60:239–246
the whole structure. In the present work, the stiffness 6. Ma H, Fan F, Zhong J, Cao Z (2013) Stability analysis of single-
reduction was done by changing the elasticity modulus layer elliptical paraboloid lattice shells with semi-rigid joints.
although, evidently, this is not the best way to be solved. Thin-Walled Structures 72:128–138
7. Yan J, Qin F, Cao Z, Fan F, Mo YL (2016) Mechanism of cou-
However, in the case of the present dome, a reduction pled instability of single-layer reticulated domes. Eng Struct
of 20% in the whole stiffness was enough to approxi- 114:158–170
mate the numerical and experimental results. This value 8. Wei J, Tian L, Hao J (2018) Improving the progressive collapse
indicates the total amount of stiffness loss. Semi-rigid resistance of long-span single-layer spatial grid structures. Const
Build Mat 171:96–108
connectors models may improve the numerical results. 9. Xu Y, Han Q, Parke GAR, Liu Y (2017) Experimental study and
Therefore, near future works will examine the nodal loss numerical simulation of the progressive collapse resistance of
of stiffness for dome connections. This subject is cru- single-layer latticed domes. J Struct Eng 143(9)
cial for developing an efficient finite element model for 10. Tian L, Li Q, Zhong W, Wei J (2019) Effects of the rise-to-span
ratio on the progressive collapse resistance of Kiewitt-6 single-
dome structures considering the nonlinear loss of nodal layer latticed domes. Eng Fail Anal 106
rigidity. 11. Grünitz L (2003) The effect of welding on the buckling behavior
of a spherical cap. Int J Press V Piping 80(4):237–241
Acknowledgements The authors would like to thank the Brazilian 12. Chen Z-H, Wang X-X, Liu H, Yu Y (2019) – Failure Test of a
National Council for Scientific and Technological Development – Suspendome due Cable Rupture. Advanced Steel Construction
CNPq, for the financial support for this research. 15(1):23–29
13. Piotrowski R, Szychowski A (1944) Lateral Torsional Buckling
of Steel Beams Elastically Restrained at the Support Nodes. Appl
Declarations Sci 2019:9. https://​doi.​org/​10.​3390/​app90​91944
14. Saka MP (2007) Optimum topological design of geometrically
Conflict of Interests The authors declare that they have no conflict of nonlinear single layer latticed domes using coupled genetic algo-
interest. The research did not involve any human participants and/or rithm. Comp & Struct 85:1635–1646
animals.
834 Experimental Techniques (2022) 46:823–834

15. Pokusinski B, Kaminski M (2018) Various response functions 30. Kato S, Mutoh I, Shomura M (1998) Collapse of semi-rigidly
in Lattice Domes reliability via analytical integration and finite jointed reticulated domes with initial geometric imperfections. J
element method. Int J Appl Mech Engng 23:445–469 Const Steel Res 48:145–168
16. Pokusinski B, Kaminski M (2019) Lattice domes reliability by the 31. Morris NF (1991) Effect of Imperfections on Lattice Shells. J
perturbation-based approaches vs. semi-analytical method. Comp Struct Engng 117(6):1796–1814
& Struct 221:179–192 32. Yamada S, Takeuchi A, Tada Y, Tsutsumi K (2001) Imperfec-
17. Kubicka K, Radon U, Szaniec W, Pawlak U (2017) Compara- tion-Sensitive Overall Buckling of Single-Layer Lattice Domes.
tive Analysis of the Reliability of Steel Structure with Pinned J Engng Mech 127:382–386
and Rigid Nodes Subjected to Fire. IOP Conf. Series: Materials 33. Nanhai Z, Jihong Y (2014) Structural Vulnerability of a Single-
Science and Engineering 245, 022051. https://​doi.​org/​10.​1088/​ Layer Dome based on its Form. J Engng Mech 140:112–127
1757-​899X/​245/2/​022051 34. API-650: 2012 (2012) American Petroleum Institute – Welded
18. Ye J, Lu M (2020) Design optimization of domes against instabil- Tanks for Oil Storage
ity considering joint stiffness. J. Construct. Stell Research, 169 35. Kenner H (1976) Geodesic Math and how to use it. California,
19. Zabojszcza P, Randon U (2019) Effect of Increased Density of CA, USA: University of California Press
Nodes in Geodesic Dome on its Critical Load Capacity. IOP Conf. 36. Clinton J (1965) Structural Design Concepts for future Space Mis-
Series: Materials Science and Engineering 471, 052051 IOP Pub- sions, Progress Report – NASA
lishing https://​doi.​org/​10.​1088/​1757-​899X/​471/5/​052051 37. Shirley WF (1984) Geodesic dome analysis. 1984.131 f., Master
20. Szaniec W, Biernacka K (2013) Modal analysis of selected bar thesis – University of Arizona, Tucson
domes. Technological University of Kielce. Structure. 15–20 38. Diniz JAV (2006) Geodetic structures: retrospective studies and
21. Kubik M (2009) Structural Analysis of Geodesic Domes. 2009. proposal for an environmental education space. 2006. 00 f (143).
68 f. Final Course Work. Durham University, Durham Dissertation (In portuguese: Estruturas geodésicas: estudos ret-
22. Barbieri N, Machado RD, Barbieri L, Lima KF, Rossot D (2016) rospectivos e proposta para um espaço de educação ambiental.),
Dynamic Behaviour of the Geodesic Dome Joints. Int J Comp Master Dissertation, Federal University of Ouro Preto
Appl 140(6) 39. Rossot D (2014) Geodesic Dome – Numerical and experimental
23. Yan S, Zhao X, Chen Y, Xu Z, Lu Y (2018) A new type of approaches in laboratorial model (In portuguese: Domos Geo-
truss joint for prevention of progressive collapse. Eng Struct désicos – Abordagem numérica e experimental em um modelo de
167:203–213 laboratório). Pontifical Catholic University, PUC-PR
24. Chen WF (2000) Practical Analysis for Semi-Rigid Frame Design, 40. Chopra AK (1995) Dynamics of structures Theory and Appli-
World Scientific Publishing Co. Pte. Ltd., Singapore cations to Earthquake Engineering. University of California at
25. López A, Puente I, Serna MA (2007) Direct evaluation of the Berkeley
buckling loads of semi-rigidly jointed single-layer latticed domes 41. Rao SS (2011) – The Finite Element Method in Engineering. 5th
under symmetric loading. Eng Struct 29:101–109 Edition. Elsevier
26. de Alvarenga AR (2010) Semi-rigid connections in advanced 42. Bathe KJ (1996) Finite element procedures. Prentice-Hall
analysis with plastic zone of plane steel frames (In Portuguese: 43. Cadregeo - https://​www.​cadre​analy​tic.​com/​cadre​geo.​htm
As ligações semirrígidas na análise avançada com zona plástica 44. Orlando D, Castro CHL, Gonçalves PB (2018) Nonlinear vibra-
de portais planos de aço), 534 p., Doctoral Thesis, Federal Uni- tions and instability of a bistable shallow reticulated truss. Non-
versity of Ouro Preto. Minas Gerais, Brazil linear Dyn 94:1479–1499
27. Ma H, Fan F, Wen P, Zhang H, Shen S (2015) Experimental and 45. Wei J, Tian L, Hao J (2018) Improving the progressive collapse
numerical studies on a single-layer cylindrical reticulated shell resistance of long-span single-layer spatial grid structures. Const
with semi-rigid joints. Thin-Walled Structures 86:1–9 Building Mat 171:96–108
28. Han Q, Liu Y, Xu Y (2016) Stiffness characteristics of joints and 46. El Ouni MH, Kahla NB (2014) Active tendon control of a Gei-
influence on the stability of single-layer latticed domes. Thin ger dome. J Vib Control 20(2):241–255. https://​doi.​org/​10.​1177/​
Walled Struct 107:514–525 10775​46312​458944
29. Silva LS, Oliveira S, Costa R, Gentili F (2020) – Design and
Analysis of Steel Structures Considering the 3D Behaviour of the Publisher's Note Springer Nature remains neutral with regard to
Joints. Advanced Steel Construction 16(2):137–145 jurisdictional claims in published maps and institutional affiliations.