NONLINEAR ANALYSIS OF THREE-DIMENSIONAL LATTICE

STEEL TRUSSES

Miguel H. de O. Costa
José Guilherme S. da Silva
miguelh.o@gmail.com
jgss@uerj.br
Civil Engineering Post-graduate Programme, PGECIV
State University of Rio de Janeiro, UERJ
São Francisco Xavier St., No 524, Maracanã, 20550-900, Rio de Janeiro/RJ, Brazil
Abstract. The use of lattice structures for shoring of steel, steel-concrete composite and
reinforced concrete structures is considered an effective solution in the construction of civil
engineering systems. A change in the construction process methodology associated with costs
reduction has caused a considerable increase in the use of three-dimensional lattice steel
truss systems with greater load capacity. Unfortunately, the design of these structural systems
is based on very simplified calculations related to one-dimensional beams with constant
inertia properties. Such a very simplified modelling cannot adequately represent the actual
response of the structural models and can lead to uneconomic or even unsafe structural
design. On the other hand, these lattice steel structures are related to three-dimensional
models of complex geometry and are designed to support very high loading levels. Therefore,
this work research has proposed finite element models to represent the actual three-
dimensional character of shoring system, evaluating the static and dynamic structural
behaviour with more reliability and security. The proposed analysis methodology developed
for the non-linear static and dynamic analysis, has adopted the usual mesh refinement
techniques present in finite element method simulations, based on the Ansys program. The
present study has considered the results of a linear elastic and nonlinear geometric analysis
for serviceability actions, physical and geometrical nonlinear analysis for ultimate actions.
The numerical results were obtained and compared with those supplied by the traditional
simplified methodology of calculation and with the limits recommended by traditional design
standards.
Keywords: Three-dimensional lattice steel trusses, Nonlinear physical and geometrical
analysis, Structural behaviour.

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

1 INTRODUCTION

In many parts of the world, shoring projects of bridges, overpasses, and works of art are
made with spatial truss systems that resist and absorb the forces of steel and reinforced
concrete structures on-site. The possibility of overcoming large spans such as rivers,
overpasses, and median strips without disrupting traffic on roads and highways justifies and
intensifies the need for this type of construction (Du chateau, 1984; Costa, 2012; Huu-Tai
Thai and Seung-Eock Kim, 2011).
In the civil construction environment, the use of plane and spatial truss structures for the
shoring of permanent or temporary structural elements is the most widespread method applied
in structural engineering because of the need for technical development and to minimise costs
and timelines (Du chateau, 1984; Costa, 2012; Huu-Tai Thai and Seung-Eock Kim, 2011).
This procedure has generated structural systems for use in large spans that require high
absorption capacity of static loads from shored structures.
Thus, the objective of this work is to present a study on the structural behaviour of truss
systems through the development of numerical-computational models that simulate the actual
behaviour of these systems as closely as possible so that these structures are used as rationally
and efficiently as possible (Costa, 2012).
In this investigation, the typical discretisation techniques are employed based on the finite
elements method (FEM) and using the ANSYS program (ANSYS, 2009). The results
obtained during this investigation (displacements, stresses, and strains) are compared both
qualitatively and quantitatively with the usual methods in project design to obtain numerical
results that simulate the actual behaviour of these types of structural systems.

2 DESIGN CONSIDERATIONS

The standardised modular geometry of the M-150 truss allows for large variations with
respect to this shoring system’s use. Variations in the length of the free span by stiffening the
resistant section with the addition of new lines to adequately configure the system to
withstand various forces can be performed without difficulty (Costa, 2012).
The lower uprights are always mounted in the traction configuration with a set of screws,
nuts, and counter nuts. A wedge with angle groove geometry is positioned between the
modules, which allows for a counter shaft along the longitudinal length of the truss, thereby
controlling the deformation of the structure. When studying the bracing system, the need for
shoring on the three planes is verified, ensuring perfect alignment between the trusses.
The weakness of the system is found at the critical design points. Because of the need for
quick responses in the growing and dynamic civil construction market, designers find
themselves forced to check the configurations and sizing of modulations based on simplified
criteria such as a simple supported beam, thus restricting the economic design criteria.
On the other hand, the assembly of the modules contributes to the fragility of the system
because human error in the assembly can cause safety concerns (Costa, 2012). Based on the
previously described research, the numerical analysis developed in this study seeks to develop
and enhance the capacity for the economic use of truss system. This modelling approach is
supported by experimental tests that have been conducted to represent the actual structural
behaviour of the truss system.

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
 M.H. de O. Costa, J.G.S. da Silva

The numerical analysis method is calibrated with the vast experimental material that the
domestic shoring market possesses of the model presented based on the convergence of the
failure modes of the real structure. The Brazilian national shoring market has demonstrated
ample knowledge and experimental certification of the materials used in projects (Costa,
2012).

3 DESCRIPTION OF THE INVESTIGATED STRUCTURAL MODEL

The structural models investigated in this study are steel truss systems. The longitudinal
length consists of a span of 30 metres, and the models do not possess poles joining the support
peripheries.
Independent of the proposed model for analysis, the assembly and modulation are
composed of a module smaller than six metres that is connected with screws in their upper
and lower uprights. These screws ensure the modules are fixed longitudinally, helping to
shape a shored spatial truss system.
The investigated models (Models I and II, see Figs. 1 and 2) have two parallel rows of
trusses at two metres with five intermediate modules measuring six metres in each row, a
support system at the base of the periphery of the assembly, no post bonding to the uprights,
and horizontal bracing in tubes at every 1.5 metres. Model II also has diagonal bracing at
every 1.5 metres.
In this work, the steel possesses a longitudinal elastic modulus equal to 2 x 1011 N/m2 (E=
2 x 1011 N/m2), a Poisson’s coefficient equal to 0.3 ( = 0.3), a density of 7850 kg/m3 (γ =
7850 kg/m3), and a thermal expansion coefficient of (α= 12 x 10-6 ºC-1). The use of SAC-50,
SAE-1040, SAE-5140 and SAE-1010 commercial steels is utilized in the uprights, diagonals,
fixing screws, and brace pipes, respectively.

4 COMPUTATIONAL MODELLING

In the development of the finite element models of the investigated structural system, the
usual discretisation techniques are employed based on the finite element method using the
ANSYS software (ANSYS, 2009). The discretisation techniques are used to obtain a more
realistic numerical model of the investigated truss systems.
In the computational model, steel structures are simulated by three-dimensional finite
elements where the effects of bending and torsion are considered. In this work, the finite
element BEAM44 (ANSYS, 2009) is defined to represent the steel uprights and the PIPE16
element (ANSYS, 2009) represents the steel diagonals.
The obtained results are compared with the recommended limits in design standards
(ABNT, 2009) and their use on the market for static loads. Initially, a highly refined finite
element model is developed where the uprights, diagonals, and locking are simulated using
three-dimensional finite elements BEAM44 (ANSYS, 2009) and PIPE16 (ANSYS, 2009).
Thus, the behaviour generated by the interaction between the structural elements (uprights,
diagonal, and locking) is obtained naturally (Costa, 2012).

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

ELEMENTS: 758
BEAM44: 478
PIPE16: 280
NODE: 602
NGL: 3600

Figure 1. Structural model I

ELEMENTS: 838
BEAM44: 478
PIPE16: 360
NODE: 622
NGL: 3720

Figure 2. Structural model II

The element BEAM44 (ANSYS, 2009), according to Fig. 3, has six degrees of freedom
per node, composed of three translations and three rotations in directions x, y, and z, enabling
the use of different geometries at the opposite ends of the bar. The orientation of the element
is from node I to node J, and this is the x-axis. The principal axes y and z are located in the
cross section.
CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
 M.H. de O. Costa, J.G.S. da Silva

Figure 3. Element BEAM 44 (ANSYS, 2009)

The finite element of tube PIPE16 (ANSYS, 2009), according to Fig. 4, is defined by six
degrees of freedom at each node: three translations and three rotations associated to the axes
x, y, and z. The orientation of the element is from node I to node J, defined as the x-axis. The
main axes y and z are located in the cross section.

Figure 4. Element 3D PIPE16 (ANSYS, 2009)

With respect to the analysis of the flexibility of the supports, supports were adopted both
at the extremity of the lower upright and at the base of the post connecting the modules
according to the model description, restricting the displacement and rotation in x, y, and z.
The dimensions of the free span are altered according to the disposition of the models. For
non-linear analysis, the bi-linear isotropic material is used with a post-limit stiffness equal to
10 MPa, according to Fig. 5.

Figure 5. Post-limit stiffness

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

5 RESULTS ANALYSIS AND DISCUSSION

The main objective of this study is to verify the simplified design methods adopted by the
market on the static response (displacements and stresses) of truss structures for the
temporary shoring of structural elements. For this purpose, finite element analyses are
presented based on use of the program (ANSYS, 2009), where the models are evaluated based
on the maximum values of displacements, strains, and stresses.
The models are considered non-linear and are displayed in three groups, differentiated by
the type of horizontal and diagonal locking of the structural model. The geometric non-
linearity is considered by the updated Lagrange algorithm, and the solution for the nonlinear
system of equations used the Newton-Raphson method. In the models where convergence
presented problems, the arc-length method was used.
The numerical structural response (displacements and stresses) is investigated according
to four different situations: a simplified project model without consideration of loads, a linear
numerical model without consideration of loads, a non-linear numerical model without
consideration of loads, and according to the limits proposed by NBR-15696 (ABNT, 2009).
According to the maximum deflection value established by the NBR-15696 (ABNT,
2009), the limit vertical displacement for structural shoring elements has a maximum value of
acceptability equal to 1 + L/500, where L corresponds to the maximum span considered.
In the sequence of the investigation, Figs. 6 and 7 illustrate curves associated with the
vertical displacement obtained at the centre of the truss structure as a function of the variation
of additional loading at the top of the model investigated. Thus, for the span of 30 metres, the
maximum allowable vertical displacement to Model I and II is 61 millimetres (61 mm).
Basically, there is a linear variation of the maximum values of these displacements in the
elastic range, where the simplified design model presents displacements values according to
the limits proposed by NBR 15696 (ABNT, 2009), up to a loading limit of approximately 203
kN. Table 1 shows the critical loads of the studied models and illustrates a comparison of the
percentage increase in strength compared to the simplified project model.
The design limit loads (ABNT, 2009) of the investigated structural models are presented
in Table 1. In this table, the results are shown based on comparisons between the simplified
design model (MSP) and the developed numerical finite element models (Model I and Model
II, see Figs. 1 and 2). It must be emphasized that Models I and II, see Figures 1 and 2, have
two parallel rows of trusses at two metres with five intermediate modules measuring six
metres in each row, a support system at the base of the periphery of the assembly, no post
bonding to the uprights, and horizontal bracing in tubes at every 1.5 metres. The structural
Model II also has a diagonal bracing system at every 1.5 metres.

Table 1. Design limit loads of the analysed models (MSP, Model I and Model II)

Investigated Models Design Limit Loads (kN) Increase of Resistance (%)

MSP 203.0 -

Model I 312.5 53.94

Model II 321.0 58.12

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
 M.H. de O. Costa, J.G.S. da Silva

312.5 kN

Figure 6. Displacement as a function of concentrated load on the model I

Figure 7. Displacement as a function of concentrated load on the model II

The structure has its scaling controlled by the yield strength. To demonstrate this fact,
Figs. 8 and 9 present the Von Misses stress distribution on the bracing tubes closest to the
support, the distribution of stresses on the diagonals closest to the support, the distribution of
stresses on the truss fixing screws closest to the support, and the distribution of stresses on the
upper uprights connected to the support for four loading levels, obtained based on a numerical
analysis.
CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

a) Prior to the yielding regime: P=128.8 kN b) Yielding moment: P =129.6 kN

σy=128 MPa

c) Subsequent to the yielding regime: P = 130.1 kN

σy=375 MPa

d) Yielding of the top chords: P = 148.0 kN

Figure 8. Evolution of the Von Mises stress (MPa) on the model I

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
 M.H. de O. Costa, J.G.S. da Silva

a) Prior to the yielding regime: P=177.2 kN b) Yielding moment: P =178.6 kN

σy=128 MPa

c) Subsequent to the yielding regime P = 179.8 kN

σy=608 MPa

d) Yielding of the fixing screws: P = 1601 kN

Figure 9. Evolution of the Von Mises stress (MPa) on the model II

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

These levels correspond to the moment prior to entering in the yield regime, at the exact
moment of yielding, and subsequent to entering the yield regime, and the decline of the load
versus the displacement curve with the beginning of yielding or the failure of the materials
employed for constructing the steel trusses. These levels are shown in item “d” in Figs. 8 and
9, respectively.
In evaluating the study, the accumulation of stresses and normal forces is observed in the
diagonal near the supports, especially in the end diagonals, which are more prone to forces
from reaction of the support, corroborating the acceptance of estimating shear forces of a
simple and bi-supported beam, as adopted in the simplified design model.
All the results can be verified in Table 2, where the translational vertical displacements
(y-axis) in the investigated models are compared: the simplified design model (MSP), the
linear numerical model (MNL), and the non-linear numerical model (MNNL: Model I and
Model II, see Figs. 1 and 2). Deformation of the structure is illustrated in Figs. 10 and 11.

Table 2. Vertical translational displacements of the investigated models

Maximum vertical translational displacement (mm)

Load (kN)
Model I Model II
MSP MNL
MNNL MNNL

1 0.30 0.19 0.19 0.19

25 7.50 4.80 4.81 4.79

50 14.90 9.60 9.64 9.49

75 22.40 1.,40 14.50 14.38

100 29.90 19.20 19.37 19.18

125 37.40 24.00 24.25 23.98

150 44.90 28.80 - 28.77

175 52.40 33.60 - 33.57

200 59.90 38.40 - 38.37

225 67.40 43.20 - 43.17

250 74.90 48.00 - 47.27

275 82.40 51.90 - 51.96

300 89.90 56.70 - 56.66

325 97.40 61.40 - 61.40

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
 M.H. de O. Costa, J.G.S. da Silva

Figure 10. Deformed on the y-axis of the structural model I

Figure 11. Deformed on the y-axis of the structural model II

With respect to the quantitative question, a comparison may be highlighted between the
values of the yielding loads of Models I and II (see Figs. 1 and 2) without utilisation of the
connection poles in the support, which provide yielding loads equal to 148 kN and 1601 kN,
respectively, with differences of 1081%. This difference is related to the significant influence
of the Model II (see Fig. 2), proposed bracing system.

6 CONCLUSIONS

Spatial truss systems play a very important role in the design of steel, reinforced concrete,
and composite structures (steel-concrete). When a civil engineer has knowledge about the
structural behaviour of a shoring system and its safety concerns, shoring designs should be
more efficient, safe, and cost effective.
Another issue of great importance concerns the small number of full-scale tests
performed in the laboratory to provide a more effective calibration of the models currently
used in design practice. This lack of data leads to additional difficulties analysing the
parameters that influence the behaviour of spatial truss systems.
In this project, numerical-computational models were developed to accurately represent
the spatial truss systems used for the shoring of civil structures. This type of shoring system,
regularly employed by the construction industry, requires that its actual behaviour be

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013
Nonlinear analysis of three-dimensional lattice steel trusses

appropriately investigated based on numerical modelling, as performed in this study, to

establish a database with reliable results. The analysis of the numerical results obtained in the
study, supported by the development of numerical-computational models, permit some
important conclusions.
Based on the developed numerical model, we observed that the models accurately
reproduced the modes of failure of the investigated shoring system with respect to the non-
linear analyses.
In all the cases studied with respect to the elastic regime, the simplified project model
presented utilisation limit load values that were much smaller than the numerical-
computational models, with differences of approximately 50%. We therefore conclude that
the truss M-150 displays a large increase in strength when this structural system is compared
with the calculations used for the project. In summary, the results indicate that this system
could certainly be better investigated to economise material and/or develop a more extensive
knowledge of the system’s safety.

ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support for this work provided by the
Brazilian Science Foundation’s CAPES, CNPq and FAPERJ.

REFERENCES
ABNT: Brazilian association of technical standards, 2009. NBR-15696: Forms and shoring
for concrete structures-design, sizing and executive procedures.
ANSYS. Swanson Analysis Systems, Inc., P.O. Box 65, Johnson Road, Houston, PA, 15342-
0065, version 12.1, Basic analysis procedures, Second edition, 2009.
Costa, M.H. de O., 2012. Modelagem do Comportamento Estrutural de Sistemas Treliçados
Espaciais para Escoramentos de Estruturas de Aço, Concreto e Mistas (Aço-Concreto). MSc
Dissertation (In Portuguese). Civil Engineering Post-graduate Programme, PGECIV, State
University of Rio de Janeiro, UERJ, Rio de Janeiro/RJ, Brasil.
Du chateau, S, Structures spatiales, 1984. International conference on space structures.
Guildford, UK, Proceedings: London, New York: Elsevier Applied Science. pp. 1058-1065.
Huu-Tai Thai, Seung-Eock Kim, 2011. Nonlinear inelastic time-history analysis of truss
structures. Journal of Constructional Steel Research, vol. 67, pp. 1966-1972.

CILAMCE 2013
Proceedings of the XXXIV Iberian Latin-American Congress on Computational Methods in Engineering
Z.J.G.N Del Prado (Editor), ABMEC, Pirenópolis, GO, Brazil, November 10-13, 2013