Professional Documents
Culture Documents
A R T I C L E I N F O A B S T R A C T
Keywords: The diagrid structural system has seen significant uptake in medium to high rise buildings because of the
Diagrid structural system architectural and resource advantages that it provides. These arise mainly as a result of flexibility in the topology
Steel structures that this particular structural solution provides. However, as a result of the way in which diagrids carry both
Structural fire resistance
horizontal and vertical loading, the diagrid structure itself may be susceptible to fire in ways which are not
Numerical modelling
immediately obvious on the basis of our understanding of more traditional rectilinear construction forms. This
study addresses this to improve our understanding of diagrid structures’ response to fire. A comprehensive
structural analysis on 45 fire load cases is conducted, considering different fire locations and sizes, using
parametric design tools and finite element analysis software. The results provide valuable insights into the load
redistribution and collapse mechanisms of diagrid structures in fire conditions.
* Corresponding author.
** Corresponding author.
E-mail addresses: zhiruoyu.wang@uq.edu.au (Z. Wang), anwar.orabi@connect.polyu.hk (M.A. Orabi), zhuojun.nan@connect.polyu.hk (Z. Nan), wywang@cqu.
edu.cn (W. Wang), matthew.mason@uq.edu.au (M. Mason), d.lange@uq.edu.au (D. Lange).
https://doi.org/10.1016/j.firesaf.2023.103924
Received 18 June 2023; Received in revised form 13 August 2023; Accepted 21 August 2023
Available online 26 August 2023
0379-7112/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
2
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
Fig. 2. The details criteria and structure design of model (a) mechanical loads (b) series numbers of the diagrids’ members and connection nodes.
to the floors which coincide with diagrid chords as diagrid nodal floors, Cuthill-McKee numbering technique, a transformation constraints
such as N67, N68 and N69 as shown in Fig. 2 (b). handler, and the Full Newton-Raphson algorithm to solve the system of
Diagonal members and chord beams were modeled as displacement- equations. The dynamic interval, on the other hand, used a load-
based beam-column components with fibre-based sections. 8 discrete controlled integration approach with the UmfPack system of equa
elements were used per member, resulting in a total of 3630 discrete tions. The reverse Cuthill-McKee numbering technique was employed,
elements to represent the 176 structural members in the model. The along with a transformation constraints handler, to facilitate the inte
boundary conditions implemented are limited to the bottom nodes of the gration process. The integration process was performed through the
diagrid structure system, effectively prohibiting all translational and application of the Newmark method with an average acceleration
rotational movements, as shown in Fig. 2(a). approach by using the parameters γ = 0.5 and β = 0.25 [10] and the
Krylov Newton algorithm was utilized to solve the system of equations.
The fire scenario could be condensed by a factor of 1/1000 for the im
2.3. Fire scenarios
plicit dynamic analysis in OpenSEES [18]. The analysis was performed
with a step size of 0.005 s and a total duration of 3.6 s.
A total of 45 fire scenarios were created by applying an idealised
uniform fire to various locations of the diagrid system model. The
3. Results
location of the fires was chosen to allow a study of the response of the
diagrid with one or more members on one floor heated; as well as with
3.1. Response to single floor fires
one or more members plus chords on multiple floors heated. Thus,
combinations of heated elements represent fires on one floor or on
In the discussion of results we use load case 0 as a reference case. This
multiple floors. A summary of these cases is given in Table 1. Since the
is the scenario with only one diagrid member span heated on one floor.
model is comprised of steel, none of the behaviour is likely to be influ
Displacements of all diagrid nodes in the model are shown in Fig. 3. Of
enced by rate of heating, thus results are presented as a function of
particular relevance are the displacements of Nodes N67, N68 and N69
average steel temperature. In each case listed in Table 1,diagrid mem
in this analysis, since these represent the nodes at the tops of the indi
bers and chords are heated continuously and simultaneously.
vidual diagrid member spans which comprise the heated diagrid mem
ber in this analysis. As is shown in the plot of displacement, after the
2.4. Numerical analysis details static loading step and when heating starts, heating to approximately
500 ◦ C of the lower diagrid member span results in an expansion and
The structural analysis of the diagrid system was performed using upwards displacement of all floors above the 19th floor, as indicated in
both static and dynamic intervals. The static interval adopted a load- Fig. 3(a) and (b). The thermal expansion phase of the Diagrid Member
controlled integration approach utilizing the Band General system of span No. 67 is followed by a buckling phase after steel temperature
equations. This integration method was coupled with the reverse
3
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
4
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
Table 1 (continued ) the rest of this diagrid member, which occurs as the load applied at the
Fire Diagrid member span Chords Description connections with the floor structures pushes this member downwards.
scenario heated heated This is clearly seen in Fig. 3 (a). Fig. 3 (b) also shows the displaced shape
case of the structure at various stages in its response.
number The load redistribution throughout the diagrid can be explored in
43 67 + 68+69 + 85+86 151 + Twelve diagrid member in reference to the section forces in the diagrid member spans. This is
+ 87+128 + 129+130 150+ 149 four diagrid element spans provided in Fig. 4 (a) for Diagrid Member spans 67–69, 85–87, 128–130,
+ 171+172 + 173 on 19F,20F,21F and 3
76–78, and 58–60, with visualisations of the section force shown
connecting diagrid chord
44 67 + 68+69 + 85+86 151 + Fifteen diagrid member in inprocessing models in Fig. 4 (b).
+ 87+128 + 129+130 150+ 149 five diagrid element spans As can be seen, the increase in temperature of one member span up to
+ 171+172 + 173+91 on 19F,20F,21F and 3 ca. 570 ◦ C results in an increase in the section force in all three spans of
+ 92+93 connecting diagrid chord the heated diagrid member. Note that the section force in the other
diagrid member that connects at the node with the heated member span
reaches approximately 570 ◦ C, corresponding with a reduction in yield is also subject to the same behaviour, a slight increase in the section
stress of approximately 50% of the ambient value of this member span. force upon heating of the adjacent member. This is accompanied by a
Failure of this member is accompanied by a downwards displacement of reduction in section force in the surrounding diagrid members, however
when the heated diagrid member span fails all of the accompanying
Fig. 3. (a) Z-axis Displacement of Connection Nodes between 19 and 21F, and (b) Five-times scaled displacement structural members at the 100th, 200th, and 720th
Intervals of Dynamic steps.
Fig. 4. (a) Section force plot, (b) Visualized section force post-processing models.
5
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
spans on the diagrid member and in the adjacent member are subject to partial restraint of this structure, the distribution of loads, and the
a significant reduction in the section force, and this force is transferred reduction in strength of Diagrid Member span 68 after reaching a tem
to the adjacent members. perature of 600 ◦ C. Note that the magnitudes of these displacements are
As can be seen, the majority of the load reduction incurred in the significantly lower than those observed in load case 0.
triangle diagrid structure, comprised of Diagrid Member 67, has been In the scenario of Case 002, with the diagrid member span below the
redistributed among Diagrid Members No. 128–130 and 58–60. These nodal floor heated, the structural behaviour of Diagrid Member span 69
diagrid members align in the same direction as Diagrid Members No. when heated exhibits similarities to that observed in load Case 0.
67–69, rather than being transferred to the nearest diagrid members, However, it is important to note that the expansion primarily resulted in
76–78 and 85–87. A smaller load redistribution occurs to member spans downwards displacement of Diagrid Member spans 68 and 67, as
85 to 87, the magnitude of which corresponds with the load reduction on depicted in Fig. 6 (a). This deviation from the response observed in Case
member spans 76 to 78, which it is aligned with. 000 can be attributed to the presence of the major connection node N69,
In the scenario of Case 001, the structural response of Diagrid which provides substantial restraint at the top of Diagrid Member span
Member No. 68, on an intermediate floor between two of the diagrid 69.
nodal floors is exposed to fire. The behaviour is similar to that observed It is important to emphasize that the load reduction and load dis
in the previous load case, with a pre-buckling phase upon heating to a tribution observed in Case 000 were consistent across both Case 001 and
temperature of ca. 600 ◦ C and a post-buckling phase characterized by Case 002, as indicated by the section forces presented in Fig. 6(b).
similar behaviour as that observed above. Thermal expansion in this A progression from the initial scenario of Case 000 was established
initial phase causes Diagrid Member span 69 to move upwards and by incrementally increasing the number of diagrid members subjected to
Diagrid Member span 67, below, to move downwards, as shown in Fig. 5 fire on same floor. This resulted in the establishment of four additional
(a). However, both Diagrid Member spans 68 and 67 recover these scenarios, load cases 005, 009, 012, and 019, which correspond to the
displacements after the temperature of the heated span exceeds horizontal fire exposure of 2, 3, 4, and 5 diagrid member spans,
approximately 600 ◦ C. respectively, on the same floor, as listed in Table 1. As shown in Fig. 7
This recovery of the displacement can be attributed to the clockwise (a), 7(b), and 7(c), the structural responses of Cases 005, 009, and 012
rotation at the major connection node N69 as shown in Fig. 5 (b). This closely followed the pattern observed in Case 000, however, with a
rotation was likely the result of a combination of factors, including the progressively increasing magnitude of maximum displacement in the Z
geometry of the diagrid, an increasing P-delta moment caused by direction. However, upon exposing 5 diagrid members to fire in Case
displacement of the diagrid element and thermal expansion against the 019, the diagrid system collapsed, as shown in Fig. 7 (d), initiated when
Fig. 5(a). Z-axis Displacement of Connection Nodes between 19 and 21F, and (b) the nodal displacements and rotations.
6
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
Fig. 6. (a) Z-axis Displacement of Connection Nodes between 19 and 21F, and (b) Visualized section force post-processing models.
Fig. 7. Z-Axis displacement of connection nodes between 19 and 21F of (a) Case 005(b) Case 009 (c) 012 (d) Z-axis displacement of connection nodes between 19th
to 21st floor of Case 019 with interval steps on X-Axis.
the steel reached a temperature of 400 ◦ C–500 ◦ C. member was subjected to fire, have also been demonstrated to hold true
The findings of Case 000 and the subsequent “Case 000-type" group when multiple diagrid members on the same floor were exposed to fire.
cases (005, 009, 012, and 019) were also observed to apply to the rest of The section force plot and visualized section force post-processing
the series of the two “Case 001-type" group cases and the “Case 002- models of Case 006 and Case 009 show the impact of fire on a diagrid
type" group cases. member and its resultant effect on the entire triangular diagrid struc
The findings regarding load reduction and load distribution observed ture. The results show that when a diagrid member is exposed to fire, it
in the cases of Case 000, Case 001, and Case 002, where a single diagrid results in a reduction of section force values for all three diagrid member
7
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
spans and a significant amount of load reduction and redistribution redistribution throughout the triangular Diagrid structure and 2) the
throughout the triangular diagrid structure, as shown in Fig. 8 (a) and load reductions incurred in the triangular Diagrid structure are pri
Fig. 8 (b). Furthermore, the results support the earlier conclusion that marily redistributed among Diagrid members oriented in the same di
the load reductions incurred in the triangular diagrid structure are pri rection, rather than being transferred to neighboring Diagrid members.
marily redistributed among diagrid members aligned in the same di
rection, rather than being transferred to the nearest diagrid members, as 3.3. Comparative Analysis of the model with concrete slabs
shown in Fig. 8 (c) and Fig. 8 (d).
The approach to modelling the floors, as outlined in Section 2.2,
introduces a representative floor structure that simulates the membrane
3.2. Response to multi-floor fires
action of the floor system for transferring horizontal loads between
diagrid members. This represents a novel contribution to the study of
The results of the multi-floor fire simulation study indicate that the
diagrid structures under fire conditions. This modification focuses on the
displacement of the Diagrid members (No. 67, No. 68, and No. 69)
behaviour of the diagrid components under fire exposure without the
subjected to fire in Case 008 is significantly greater than that observed in
need to model the entire floor system, without substantially increasing
the benchmark case, Case 000. The primary cause of this increase in
the computational cost. These radial beams serve to connect the diagrid
displacement can be attributed to the thermal expansion of the Diagrid
to the core of the building. Because of the concave topology of the
members. This expansion leads to a significant upward displacement of
diagrid structure chosen in this case study these beams are in tension as
the connection node N69, while simultaneously accelerating the
the vertical loading on the diagrid increases. The modelling approach
downward displacement of all three Diagrid members, as depicted in
thus has the advantage of being able to account for the contribution of
Fig. 9 of the study.
these beams (although this will form a part of a future study).
Despite the differences in displacement and expansion observed
In this section, a variety of cases including the concrete slab, have
between single-floor and multi-floor structures subjected to fire, the
been analysed and presented to validate the proposed simplified floor
analysis of the section force of the Diagrid members in Case 000 and
structure models. As the shown in Fig. 11, the general behaviour and
Case 008 reveals a substantial degree of consistency with respect to the
magnitudes of the displacements correspond to those shown in Fig. 7. In
overall pattern of the curve and the section force of each Diagrid
addition, it is noteworthy to mention that both modelling methodologies
member at the final interval step of the simulation, where the steel
predict a global collapse happens for case 019.
temperature reaches 934 ◦ C as,depicted in Fig. 10 (a) and Fig. 10 (b).
Further empirical evidence is provided in Fig. 12 that the load
The similarities in load reduction and distribution observed in multi-
reduction and distribution patterns in the model including the concrete
floor fire Case 008 and single-floor fire Case 000, as depicted in Fig. 10
slabs exhibits a high degree of congruence with the structural modelling
(c) and (d) respectively, are indicative of the validity of two established
approach that excludes concrete slabs, as shown in Fig. 8.
principles of load reduction and distribution. These principles, estab
lished through previous analysis, state that: 1) exposure of a Diagrid
member to fire results in a reduction of section force values for all three
Diagrid members and a significant amount of load reduction and
Fig. 8. The diagrid members’ section force plot of (a)Case 006, (b)Case 009 and Visualized section force post-processing models of (c) Case 006, (b)Case 009.
8
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
Fig. 9. Z-Axis displacement of connection nodes between 19 and 21F of (a)Case 000 (b)Case 008.
Fig. 10. The diagrid members’ section force plot of (a)Case 000, (b)Case 008 and Visualized section force post-processing models of (c) Case 000, (b)Case 008.
Fig. 11. Comparative analysis model: Z-Axis displacement of connection nodes between 19 and 21F of (a) Case 005(b) Case 009(c) Case 012 (d) Case 019.
4. Overall response and damage to diagrid system from fire inflicted upon the Diagrid structure due to fire. Within this context, the
term ‘non-collapse’ denotes that indirectly fire-exposed members retain
Table 2 summarises the comprehensive response and damage their structural integrity and do not fail to collapse. ‘Local collapse’
9
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
Fig. 12. Comparative Analysis of the Model: the diagrid members’ section force plot of (a)Case 006, (b)Case 009 and Visualized section force post-processing models
of (c) Case 006, (b)Case 009.
Table 2
Summary of global collapse, local collapse, and non-collapse cases.
Cases DiaE Beam Collapse Cases DiaE Beam Collapse
0 67 / N 11 69 + 87+130 150 N
1 68 / N 12 67 + 85+128 + 171 / N
2 69 / N 13 68 + 86+129 + 172 / N
3 67 + 68 / N 14 69 + 87+130 + 173 151 + 150 N
4 68 + 69 / N 15 67 + 68+86 + 85 / N
5 67 + 85 / N 16 68 + 69+86 + 87 151 + 150 N
6 68 + 86 / N 17 67 + 68+69 + 85+86 151 N
7 69 + 87 / N 18 67 + 68+69 + 86+87 151 + 150 N
8 67 + 68+69 151 N 19 67 + 85+128 + 171+91 / Y
9 67 + 85+128 / N 20 68 + 86+129 + 172+92 / Y
10 68 + 86+129 / N 21 69 + 87+130 + 173+93 151 + 150+149 Y
22 67 + 68+85 + 86+128 / N 33 68 + 86+129 + 172+92 + 14+101 / Y
23 67 + 68+85 + 86+129 / N 34 69 + 87+130 + 173+93 + 15+102 151 + 150+149 + Y
148
24 68 + 69+87 + 86+129 151 + 150 N 35 67 + 68+85 + 86+128 + 129+171 / Y
25 68 + 69+87 + 86+130 151 + 150 N 36 67 + 68+85 + 86+128 + 129+172 / Y/N(Local
Collapse)
26 67 + 85+128 + 171+91 + / Y 37 68 + 69+87 + 86+128 + 129+172 151 + 150 Y/N(Local
13 Collapse)
27 68 + 86+129 + 172+92 + / Y 38 68 + 69+87 + 86+128 + 129+173 151 + 150 N
14
28 69 + 87+130 + 173+93 + 151 + Y 39 67 + 68+69 + 85+86 + 87+128 151 + 150 N
15 150+149
29 67 + 68+85 + 86+128 + / N 40 67 + 68+69 + 85+86 + 87+129 151 + 150 N
129
30 68 + 69+86 + 87+129 + 151 + 150 N 41 67 + 68+69 + 85+86 + 87+130 151 + 150 N
130
31 67 + 68+69 + 85+86 + 87 151 + 150 N 42 67 + 68+69 + 85+86 + 87+128 + 129+130 151 + 150 N
32 67 + 85+128 + 171+91 + / Y 43 67 + 68+69 + 85+86 + 87+128 + 129+130 + 151 + 150+149 Y
13+100 171+172 + 173
44 67 + 68+69 + 85+86 + 87+128 + 129+130 + 151 + 150+149 Y
171+172 + 173+91 + 92+93
10
Z. Wang et al. Fire Safety Journal 141 (2023) 103924
11