See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/319798190

Investigation of the cause of failure of the Omo River Bridge

Article · August 2017

DOI: 10.2749/101686617X14881937384729

CITATIONS READS

3 1,764

3 authors:

Girma Zerayohannes Esayas Gebreyouhannes

Addis Ababa Institute of Technology Addis Ababa University, Addis Ababa Institute of Technology
5 PUBLICATIONS   3 CITATIONS    33 PUBLICATIONS   307 CITATIONS   

SEE PROFILE SEE PROFILE

Adil Zekaria
Addis Ababa Institute of Technology
9 PUBLICATIONS   11 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fatigue Performance of Concrete Pavement View project

Masters Thesis For Concrete Column Design View project

All content following this page was uploaded by Esayas Gebreyouhannes on 13 August 2018.

The user has requested enhancement of the downloaded file.

Investigation of the Cause of Failure of the Omo River
Bridge
Girma Zerayohannes, DR. -Ing Civil Eng.; Esayas Gebreyouhannes, Dr. Civil Eng.; Adil Zekaria-Abdullahi, Dr. -Ing, School
of Civil and Environmental Engineering, Addis Ababa University, Addis Ababa, Ethiopia. Contact: adilzekaria@yahoo.com
DOI: 10.2749/sei.1356

Abstract constructed from standard-type steel Description of Launching

truss bridge parts through a process Sequences
The reason for the failure of the Omo similar to actions involved in stages 1
River Bridge that collapsed and par- to 11 prior to the launching of the A typical launching cycle is described in
tially sank into the river during construc- main bridge. To connect the main the erection manual1 using launching
tion was investigated using standard, bridge and the launch nose during the stage 12 as an example. The same is
structural and analytical finite element launching operation, special erection repeated below to highlight the steps
model (FEM) and design review proce- link angles were provided. undertaken during each cycle and their
dures according to Eurocode 3. The significance with respect to the required
bridge was initially constructed by a The Ethiopian Road Authority had structural models (more than one) for
launching construction technique. The commissioned the Addis Ababa any given cycle. A typical launching
span design was based on a Callender- Institute of Technology (AAiT) and cycle consists of:
Hamilton bridge type with Warren flown senior faculty members to
the site on 15 May 2011 to investi- • Adding a D8M panel (D8M is the
trusses of the D8M type classification. designation used to identify one par-
In all 40 stages were required to put the gate the cause of failure of the
bridge. In situ inspection during the ticular type of bridge with a depth of
bridge in place. The bridge collapsed 9414 mm and 8000 mm center to
and partially sank into the river during site visit indicated that the main
center distance of the joints where
construction, between stages 34 and 35. bridge did not show any significant
the main diagonal elements meet the
From in situ inspection, it was observed distress. Furthermore, the mode of
top and bottom chords)
failure, which resembles a mechanism
that the main bridge did not show any • Jacking up the nose and moving the
significant distress. Moreover, the mode of a simply supported beam with a
launch skate to the next node
plastic hinge at the critical section,
of failure suggested a weak link between • Pulling the bridge forward by half a
the main bridge and the launch nose. suggested a weak link between the
panel length
The results of the investigation showed main bridge and the launching nose. • Jacking up the bridge and nose and
that the diagonal link angles in the criti- The focus of the investigation was moving the skates to the next node
cal region where the main truss was con- therefore directed towards checking points
nected with the launching nose were the resistances of the members in the • Pulling the bridge forward by an addi-
found to be the most critically loaded critical region. tional half a panel length so that at
members. These members were stressed The investigation was based on the the end of each cycle, the bridge is
to their ultimate limit state, causing material strengths determined from lab- moved by a distance equivalent to
them to buckle and trigger the sudden oratory tests, with average yield strength span of one D8M panel as illustrated
collapse of the bridge at the end of of 517 MPa and action effects for load in Fig. 2
launching stage 34. case unfactored self-weight plus addi-
tional permanent loads used as The construction was proceeding suc-
Keywords: girder truss bridge; launch- cessfully until launching of stage 34.
ing construction technique; battened counterweight.
built-up members; launch nose; buck-
ling failure; failure analysis.

Introduction
Main bridge
The Omo River Bridge was intended
for construction in the Southern
Launch
Nations, Nationalities and People’s nose
Region of Ethiopia, approximately
800 km southwest of the capital Addis
Ababa. In 2011, the bridge collapsed
into the river while still under con-
struction (Fig. 1). The launching
method was used for the construction
and the main bridge was constructed
behind a light-weight “launch nose”
and pushed or pulled into place, the
bridge being supported on rollers or
sliding skates. The launching nose was Fi g . 1: Picture of the collapsed Omo River Bridge near Omarate

Structural Engineering International Nr. 3/2017 Technical Report 1

 D8M Bridge Erection Link Steel B15M Launch Investigation of the Cause of
Failure
(1) (2) Stage 11 Completed A three-dimensional (3-D) finite ele-
ment centreline model was developed
for the investigation of the cause of fail-
ure of the bridge during construction.
Stage 12 (i): Add Panel of D8 Bridge All members of the main bridge and the
launch nose are uniform battened built-
up members with different cross-
sections, as shown in Table 1. The link
Stage 12 (ii): Jack up Nose at (2) and move angles that are used to connect the main
Launch Skates to next node point
bridge with the launch nose are also uni-
form battened built-up members.
Therefore, substitute truss members
Stage 12 (iii): Pull forwards with effective stiffnesses have been used
to develop the structural model.
As the bridge had collapsed during con-
Stage 12 (iv): Jack up Bridge and Nose at (1) and (2) struction, the governing load case is self-
and move Skates to next node point weight plus additional permanent load
used as a counter-weight at launching
stage 34. Thus, the assessment of self-
Skates Stage 12 (v): Pull Forwards weight and stiffnesses of truss members
Launch Beams
as accurately as possible was critical for
F ig . 2: Illustration of the launching procedure during stage 121 obtaining more reliable values of the
action effects in the structural members.
This was facilitated by available design
1 2 3 4
documents in general and the schedule
of parts for the main bridge and nose3 in
Stage 15
particular. The self-weights of the plates
and bolts were taken from the schedule
and distributed as corresponding nodal
Stage 16
loads in the structural model, while the
self-weights of truss members were cal-
Stage 17
.
.
culated using the software. The recom-
. mended counter-weight in the original
.
design for launching stage 34 is 25 t per
Stage 30
panel applied on the outer four panels
as shown by the bold lines in Fig. 3.
Stage 31
The result of the analysis showed that
Stage 32 for the load case self-weight plus the
additional permanent load applied
as a counter-weight, the maximum
Stage 33
responses in the erection link members
(link angles, diagonal link angles and top
Stage 34
chord link angles) occurred at the end of
launching stage 34. Moreover, the
Stage 35
design of these members were checked
according to the recommendation by
Stage 40 EN1993-1-1:2005;2 the results showed
that the most critical member had
Stage 41 -Complete reached the ultimate limit state of buck-
ling at the end of launching stage 34.
F ig . 3: Summary of the launching sequence1 Figure 4 shows the action effects in
the critical region at launching stage
34 for load case unfactored self-weight
Figure 3 shows successfully executed To provide a clear launching path for plus unfactored permanent load of a
launches up to stage 34 and the remain- the main bridge and the launching magnitude of 25 t per panel, applied
ing unfinished launching sequences as nose, the concrete side-span abutments in the outer four panels of the main
illustrated in the erection manual.1 were not constructed until after the bridge as a counter-weight. The most
Grid lines 3 and 4 represent the loca- main bridge was in place, supported by critically loaded members in the criti-
tions of the pier heads. the pier heads. cal region—link angles, top chord link

2 Technical Report Structural Engineering International Nr. 3/2017

 Truss IT OT IT OT IT OT IT OT
No. L sections 1 1 2 1 2 2 3 2
Top chord

Diagonal

Bottom chord

Truss IT OT IT OT IT OT
No. L sections 3 3 4 3 4 4
Top chord

Diagonal

Bottom chord

IT, internal truss; OT, outer truss; make-up, number of L-sections used to build up the sections for internal or outer trusses.

Table 1: Built-up sections3

angles and diagonal link angles—are the left of pier 3 and ten panels (Nose) plus counter-weight at the end of
shown in Fig. 4 with the member des- to the right of pier 4 for intermediate stage 34.
ignations used in the schedule of step and four panels (Main) to the left (2) Design checking using the Euro-
parts.3 pier 3 and 11 panels (Nose) to the code 3 (EN 1993-1-1:2005)2 and the
right pier 4 for final step-, resulted in action effects for the above-
Table 2 shows the magnitudes of the
internal axial forces in the diagonal mentioned load case and material
action effects, the types of the cross
link angles that were less than the cor- strengths determined from labora-
sections, the lengths of the members
responding value at the end of launch- tory tests (σ m = 517.0 MPa) showed
and the batten spacings in the critical
ing stage 34. Fig. 5 shows the internal that the diagonal link angle member
members.
forces in the critical members for reached the ultimate limit state of
launching staged 32 up to 36. buckling about the y-axis at
Discussion of Results the critical action effect of (P/
More than one structural model is
The following were observed: M = −1661.6 kN/2.9 kNm).
required for a single launching stage (1) The analysis result showed maxi- Therefore, the diagonal link angles
because a typical cycle of the launching mum values of the axial compres- had been stressed to their ultimate
process consists of a number of sub- sive forces for load case self-weight limit state, causing them to buckle
stages, as shown in Fig. 2. The models
for the different stages have been devel-
oped and analysed to determine the
state of stress in the critical members
Link angles
during the initial stages of launching
prior to stage 33. The analysis results
showed action effects that were insignifi-
cant compared with those at critical
stages 33, 34 and 35. Internal forces are
more critical in other members during
the initial stages of launching. However,
the member resistances are all adequate
at the initial stages. Thus, the launching
continued without a problem until stage
34, when it suddenly collapsed after Top chord
link angles
stage 34 was completed and stage 35 was
about to be implemented. Diagonal
link angles
At the end of launching stage 34, there
were five panels (Main) to the left of
pier 3 and nine panels (Nose) to the
right of pier 4. Analysis of the interme-
diate and final steps during launching
stage 35 -with five panels (Main) to Fi g . 4: Action effects in the weak link region and the critical members

Structural Engineering International Nr. 3/2017 Technical Report 3

 Length of member and Action effects elements, causing all of them to fail
Member Cross-sections batten spacing (mm) P(kN)/M33(kNm) suddenly and, as a result, the collapse
750 mm
of the entire bridge.
Link l = 7680 −3087.3/−25.4
angle a = 2560
y
Conclusions
z Detailed analytical investigation was
double angle flanges made conducted to determine the cause of
of 2 × L 200/200/20 failure of the Omo River Bridge dur-
Top chord 750 mm l = 4572 −3159.5/14.4 ing construction. Of the three highly
link angle a = 1524 stressed members (link angles, top
y chord link angles and diagonal link
angles) in the region where the main
truss was connected with the launch-
z
ing nose, the diagonal link angles
4-angle and 3-angle flanges
were found to be the most critically
made of 4 × L 150/150/10
loaded members.
and 3 × L 150/150/10
Diagonal 750 mm l = 5112 −1661.6/2.9 The investigation is repeated for load-
link angle a = 1704 ing and boundary conditions corre-
y sponding to launching stage 33 to be
able to determine the state of stress
before collapse and reflect on availa-
z
ble strength reserve and/or signs of
double angle flanges made
imminent failure at that stage. The
of 2 × L 150/150/10
results of the investigation showed that
P, Axial; M33, Moment about the major axis. the most critical member, that is, the
Table 2: Critical members’ cross-sections, lengths and action effects3 diagonal link angle, had not reached
its ultimate limit and possessed some
3500 strength reserve at launching stage 33.
Linkangle It was thus concluded that the Omo
3000
Diaglinkangle River Bridge had collapsed during
construction because the buckling
Topcrdlinkangle
2500 resistance of the diagonal link angles,
determined on the basis of the mean
2000
value of the yield strength and accord-
ing to the Eurocode 32 for built-up
sections, was less than the action
1500 effects in the members determined
using unfactored loads coming from
1000 the self-weight and other permanent
loads applied as a counter-weight at
launching stage 34.
500
The result of the study indicated that
0
the failure of the bridge could have
32 33 34 35 36 been prevented with proper routine
engineering analysis prior to the
F ig . 5: Internal axial compressive forces (kN) in the critical members for launching launching of the bridge.
stages 32–36

Members P (kN) M33 (kNm) M22 (kNm) References

Link angle −28 866.8 2338.8/−2053.7 32.8/−38.8
[1] XXXX. Callender-Hamilton Bridge—
Link top chord −2638.2 1566/−430.9 −16.2/2.8 Erection Manual, 1988
Link bottom chord −2704.8 −1795.5/2405.4 2.5/−8.4 [2] CEN. Eurocode 3: Design of steel structures—
Part 1–1: General rules and rules for buildings,
Table 3: Action effects in critical members after buckling of diagonal link angle member European Standard EN 1993-1-1:2005, European
Committee for Standardization, Brussels, 2005.
and trigger the sudden collapse of the (conservatively) that these elements [3] XXXX. Callender-Hamilton Bridges, Drawing
entire bridge. The adverse effect of become ineffective once they have Nos. AO-CHB-8701-2/3/4; AO-CHB-S112-15,
the failure of the diagonal link ele- buckled. Internal forces increase Schedule of Parts for 128.0 m Span Omo River
ments can be illustrated by assuming many fold (Table 3) in all link Bridge and 82.3 m Launch Nose-Ethiopia, 1986/87

4 Technical Report Structural Engineering International Nr. 3/2017

View publication stats