Author:
Load bearing capacity analysis of collapsed Maxima shopping center steel trusses.
On November 21.11.2013 Maxima shopping center roof collapsed in Riga. This technical
report is intended to verify the maximum load bearing capacity of steel truss due to failure of tension flange connection. This paper is a second step on collapse analysis, preceded by preliminary
analysis based on photographic and video evidence. The report will be succeeded by more elaborate
calculations, identifying all connection capacities and maximum load bearing capacity of the steel
truss, given that tension connection would have been sufficiently strong.
This report is based on technical documents, provided by steel truss manufactures SIA
Vikom in Latvia. This report is impartial and independent and aims to find the true collapse load
and mechanisms behind it.
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
1/9
Number of bolts
10 pcs
At
244.794 mm2
fu
800MPa
fy
640MPa
Nt.Rd
141
Nt.rupture
195.8 kN
(EN 1993-1-8)
kN
(14 t)
(19.8 t)
Diagonals:
-
Figure 1. Dimensions of truss, profiles: HEA260 (blue), CFRHS 160*5 (red), HEA300 (pink)
Span of truss:
Lspan
16
Spacing of trusses
cc
heff.steel
1.7 m
heff.composite
1.88 m
gsw
2 kN/m
ghollow
33.6kN/m
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
2/9
gtoping
8 kN/m
Figure 2. Dimensions of tension connection end plate, location of bolt hole, red holes represent
(NFT = non-functional bolts), load is carried through truss with eccentricity of 26.76mm. End
plate 20mm, yields safely due to bolt forces and redistributes forces in safe manner between 8 effective bolts.
Bolts are initially assumed to be loaded in statically determined fashion with eccentric tensile load
from HEA 260 steel beams, with load eccentricity calculated from FEM model.
Assuming the 20mm plate infinitely rigid can the load capacity of bolts group be found as:
NRd.connedtion =
(94.29mm+40.71mm)/( 94.29mm)*4*141kN
807
kN
(1)
Nconnection,final rupture =
(94.29mm+40.71mm)/( 94.29mm)*4*195.8kN
1121
kN
(2)
To ensure that no steel is wasted and structure exhibits ductile behaviour, the bolt connection shall
be able to carry yield force of the truss lower chord HEA260 profile. Load is applied with eccentricity of 26.76mm, leading to reduced axial yield capacity of HEA260 profile.
Nyield.profile,ecc
2801.5 kN
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
(3)
3/9
Connection is brittle and efficiency is low (steel beams HEA 260 strength reserves cannot be used)
(4)
0.28805 < 1
In final stages before approaching collapse connection could have been carrying 1121kN. Based on
span of trusses the applied distributed load acting on truss can be calculated as:
qcollapse
1121kN*1.88m*8/(16m)2
In moment before final collapse, steel truss would have been loaded with 6585 kg/m.
The extra loads on hollow cores can be calculated using trivial algebra calculus:
qextra
2.78 kN/m2
(22.25kN/m)/cc
(7)
Hence the average load on roof before collapse was 278kg/m2 including heat insulation.
Based on visual inspections from photographic evidence the loading was most likely 400kg/m2.
This means that secondary effects not included here, helped carry the load and relieved stresses on
bolts. Secondary continuity of truss top chord helped increase the final collapse load.
Structure behaved better, than structural engineer could have imagined after blind calculations.
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
4/9
Fact: 20mm S235 end plate will form yield lines and form plastic mechanism. This reduces axial
loading eccentricity. Appendix 1 contains calculation of yield line and required bolt forces for that
mechanism (justification). However, according to manufacturer, the steel grade was increased to
S355, which makes the end plate stronger, but connection in general more brittle.
Bolts were observed to not break in tension, but shearing the threads. This could mean that bolts
themselves were performing in a ductile manner, which is a very good sign. It allowed for redistribution of internal bolt forces before the collapse, allowing structure to obtain its maximum peak
load carrying capacity.
In any case, both the plate and bolts performed better that one can expect from blind calculations.
Utilizing the fact of bolt internal force redistribution connection strength can be assumed to be
on higher levels as follows:
NRd.connedtion
8*141kN
1128
kN
(8)
Nconnection,final rupture
8*195.8kN
1566.4 kN
(9)
To ensure that no steel is wasted and structure exhibits overall ductile behaviour, the bolt connection shall be able to carry at least yield limit of the truss lower chord HEA260 profile with centric
loading.
Nyield.profile,centric
(10)
3083.5 kN
0.365 < 1
(11)
Based on span of trusses the applied distributed load acting on truss can be re-calculated.
Composite action between hollow cores and steel truss is neglected in this calculation, partly because there is little evidence for effective and reliable composite action in current structural system.
qcollapse
1566.4 kN*1.7m*8/(16m)2
83.2 kN/m
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
(12)
5/9
In moment before final collapse, steel truss would have been loaded with 8320 kg/m.
The extra loads on hollow cores slabs can be calculated using algebra calculus:
qextra
39.615 kN/m
(13)
4.95 kN/m2
(14)
(39.615kN/m)/cc = (39.615kN/m)/8m =
Results seem to match with visual inspection of photos. Hence the probable loading due to gardening would have been 495 kg/m2 on average, just a moment before collapse.
1.5
permanenload
1.15/1.35
additional.load
1.151.5
q total
q total
1128*1.7m*8/(16m)2
59.925 kN/m
(15)
Using algebra and safety factors, maximum allowable extra load on structure would have been:
qallowable.extra
(16)
Distributing allowable load for truss, to slab yields suitable and allowable extra load on truss:
q allowable gardening + heat insulations
qallowable.extra /cc
0.543kN/m2.
(17)
Structure could have been considered safe only for 54.3kg/m2 of variable load like snow. Beyond
this limit, structure can be considered exceeding it suitable design limits.
It could be seen, that real loading involved in collapse exceeded safe allowable value almost 10
times.
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
6/9
Additional loads acting on top of hollow core slabs specified on design drawings:
Characteristic permanent load acting on roof has been specified as: 6.38 kN/m2 (638kg/m2)
Characteristic snow load acting on roof has been specified as:
Characteristic garden soil pressure
1 kN/m2 (100kg/m2)
20 kN/m2 (2000kg/m2
(18)
(19)
(20)
Design loads
Design loads
(21)
Design loads exceed the observed performance of structure. Structure was underperforming as
compared to characteristic and design load levels. Design extra load levels of 3883kg/m2, would
have required much more robust solution. Structure resisted maximum of 500kg/m2 on average.
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
7/9
Conclusions:
From a Euro-norms design point of view, structure could not have been considered safe
even for normal snow loads (100kg/m2) due to bad connection behaviour.
Before immediate collapse the extra load on roof panels was in range of 400 to 500 kg/m2
according to calculations and visual inspection.
Composite action between steel trusses and hollow core slabs could have increased precritical load levels. However failure of longitudinal shear between hollow cores and steel
truss could have led to sudden increase in bolt forces in later stages, causing final bolts
snapping (rupture).
20mm HEA 260 bolt connections end plate could have allowed for more warning, if S235
material would have been used instead of S355. Redistribution of bolt forces would have
been safer.
M20 8.8 bolts were most likely behaving in relative ductile manner and allowing for sufficient redistribution hence higher plastic capacity of connection.
2 bolts out of 10, where not functional at all (figure 2). Moving these ineffective bolts more
close to HEA 260 section webs, would have yielded 2 more effective bolts in equation.
Number of bolts is too low for this connection or instead size of bolts is too small. M36 size
bolts would have given sufficient warning of overload, due to excessive and visible plastic
deformation of 20mm end plate.
Bolts were performing much better than could be expected from blind engineering calculations. Bolts were probably behaving in relative ductile fashion.
Design and characteristic loads specified in drawings, exceed the real collapse loads. Hence
the structure was most likely underperformer.
The weight of debris from collapse could be measured to verify the exact average load carrying capacity of trusses. It cannot be known otherwise, since the composite action between
hollow cores and steel trusses shall be taken into account as well. Measuring the weight of
total debris, can give reasonable verification.
Structures overall performance was brittle and it did not give any warning before sudden
collapse.
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
8/9
This accident and tragic loss could have been avoided by safe steel connection design. The
capacity reserves of steel sections could not be used due to weak connection. Hence resources were used with very low rate of return on investment. Engineering community can
and should improve the performance of designs. Random overdesign does not lead to necessarily safer structures.
Engineers can use lots of steel, without gaining anything in real performance.
This is a lesson, engineering community in Finland, Estonia, Latvia, etc should learn. Brute
force methods do not solve problems.
Disclaimer: This report is concerning technical details of a collapse only. It does not take a position of what and how has happened during initial construction and later in retrofitting new winter
garden
Appendixes:
Appendix 1: Yield line method analysis of end plate
Literature:
Yield line method, for steel end plate connections
http://scholar.lib.vt.edu/theses/available/etd-07212009040533/unrestricted/LD5655.V855_1995.B674.pdf
Steel and concrete composite structures
http://219.223.252.149/service/courseware/%E5%A5%BD%E4%B9%A6/Composite%20St
ructures%20of%20Steel%20and%20Concrete.pdf
Rak Tek Solutions Oy Linnoitustie 4A, 02600 Espoo, Finland Puh. +358 44 5575 835 info@rakteksolutions.fi
9/9
20 mm end plate bolt load carrying capacity using yield line method
MSc Toomas Kaljas
Rak Tek Solutions Oy
11.2013
Figure2. Plastic mechanism, yield lines and bolt holes of connection plate and rotations yield lines
a, b and c. Unit (1) deformation in corner. Bolt experiences 0.5 times the corner displacement
Dimensions
a 115
b
55.49
c 60.45
d 45 mm
e 225
f 135
g 170.49
1.
11.2013
Appendix 1
1/2
t 20mm
t
Wpl
4
2.
kN m
Mpl Wpl fy 23.5
m
2 a b 2 a b 2 e e 2 M
pl
d
d
b
Eint
3.
kN m
Eint 482.1
m
External energy
1 Fext
4.
Eint
482.1 kN
1
External force is made out of 4 bolts, hence in requires one bolt to have loading of
Fbolt.forplateyield
Fext
120.5 kN
4
End plate of S235 would have form a plastic mechanism before the bolt ruptures. This would
have been beneficial factor. It adds possibility to redistribute bolt forces and alleviate
prying forces
Given the manufacturer statement, that the plate material was changed to stronger S355
Calculations follow:
355 Fext
Fbolt.forplateyield
182.1 kN
235 4
It becomes less likely that the end plate would have yielded, but it could have done so
when the bolts were finally rupturing. Increasing the steel strength reduced all possible
warnings signs and made structure more brittle
11.2013
Appendix 1
2/2