Damage Assessment and Repair Techniques
Content
Damage Assessment and Repair Techniques Used in Bomb and
Fire Damaged Central Bank Building in Sri Lanka
K.L.S. Sahabandu, Central Engineering Consultancy Bureau, Sri Lanka
Abstract
The paper describes the methods adopted to assess and classify the bomb and fire
damage on structural members, which include both visual and quantitative
techniques. Repair techniques which can be broadly categorized into two types (viz,
replacement of damaged concrete with proprietary cement mortars and restoration of
the affected elements by various strengthening methods) are explained together with
relevant details.
1. Introduction
The Central Bank Head Office Building, a prominent land mark for over 30 years of Colombo,
capital of Sri Lanka, was subject to a suicide truck bomb attack on 31st January 1996, which claimed
the lives of at least 85 persons and injured some 1500 more, while damaging severely the Central
Bank Building and the buildings in the vicinity and which sent shock waves throughout the island
nation.
The Central Engineering Consultancy Bureau (CECB) who had been earlier awarded the
consultancy for the Central Bank Extension Building Project in 1994, were entrusted with the
consultancy for the rehabilitation work on the damaged building. Following the explosion, CECB was
assigned to carry out a detailed survey of damages caused by the blast and the resulting fire. The
report, which was based on a comprehensive structural assessment of the entire building included the
results of non-destructive testings and load testings, was handed over to the Central Bank within four
weeks of commencement.
Meanwhile, CECB who was involved in the clearing operations on the damaged building, also
designed a temporary support structure which was put up within a couple of weeks of the explosion
and which prevented further collapse of the severely damaged area of the building and made the site
safe to work in.
In the wake of the survey of damages to the Bank Building, CECB was also given the task of
preparing documents for international tendering for the rehabilitation work on the building. This
assignment was completed within four months but subsequently the Central Bank took a decision to
award the contract to Ms. Ed Zublin AG of Germany, who already held the contract for the Extension
Building Project. The reason for this was the time factor involved. Negotiating a deal with Zublin, who
were already on site was a lot faster than going through the process of international tendering.
2. Building and its Salient Features
2.1 Building - general
The building which has a total floor area of 26,000 square metres consists of three tower blocks of
9 floors each which are interconnected by two service cores in front and rise above the top of the
mezzanine and ground floors, which floors extend over the full length and combined width of all three
towers. The building has a split level basement consisting of main vault, day vaults, office and service
areas and mechanical and electrical plant rooms. Each tower block is approximately 48m long and
11.5m wide, with approximately 10m space between them. The building is approximately 54m tall
from its foundation level, and occupies an area of 45m x 68m at ground level.
2.2 Building structure
The building is of reinforced concrete construction and stands on a pile foundation consisting of
some 260 cast-insitu piles. Entire basement up to and including the major part of ground floor have
been constructed of conventional in-situ reinforced concrete as a single monolith. The central tower
and a large part of the two service shafts on either side form one structural entity. Part of the two
service shafts close to the two end towers are integral with the respective towers and separated from
the main service shaft and central tower by two movement joints.
Each tower is supported on two rows of tapering columns spaced approximately 11.5m apart with
columns at 5.2m intervals. Columns are 1600mm deep up to the first floor level tapering down to
610mm at top with a uniform width of 305mm throughout.
Each pair of columns, supports a 508mm deep 'preflex' beam at each floor level. The beams
were factory produced and preflexed for use as composite beams. Ribs of the floor slab which are
placed at 685mm centres, are 100mm wide where they join the slab and taper down to 75mm at
bottom. Rib beams are reinforced with a single 25mm dia. mild steel bar in the longitudinal direction
at the bottom, with 6mm dia. shear links at a spacing of 150mm and 300mm at supports and mid span
respectively. Slab between ribs varies in thickness from 65mm at ribs to 50mm at mid span, and is
reinforced with 6mm diameter mild steel bars at 225mm centres in both directions. Ribs rest on
bottom flange of the preflex beams and are encased in concrete with 250mm square openings
coinciding with 203mm dia. holes in the preflex beam web, which were required for air conditioning
ducts to pass through.
3. Damages to the Building
3.1 Blast damage
The bomb resulted in massive damage to the building structure in the immediate vicinity of the
blast. Though severe, blast damage was localized and visible. Blast disintegrated the ground floor
slab at the entrance over the basement and made a two metre deep crater just outside the basement
perimeter wall. One of the four supporting columns of the mezzaninie floor was destroyed causing the
collapse of nearly two third of the cantilevered mezzanine floor with its roof. A corner column in the
centre tower was sheared at ground level and broken at two other places causing it to buckle and
settle by 300mm with all the upper floors in that corner.
Mezzanine floor in the vicinity of the blast was completely destroyed. The blast created an
upward bulge on the first floor of the centre tower. Spalled concrete on preflex beam flanges showed
that part of the slab of the first floor lifted upwards under the blast pressure. The ground floor slab
along with beams in the front area of the south tower sagged under the blast, and the mezzanine floor
above it heaved upwards.
Among all blast damages referred to above, the most critical was the damage to the corner
column in the centre tower, which column therefore had to be strengthened quickly to prevent further
collapse of the building and to be able to continue the clearing operations. Two methods of temporary
restoration were considered, viz,
1. encasing the damaged column from basement floor, up to the first floor level, where the
transfer of load would take place from column to the encasing and back to the column by the bond
between the new concrete and the existing concrete.
2. transferring the load at the damaged column to the two columns on either sides of the
damaged column by ties at different levels.
After a close examination of the damaged structure, following shortcomings were recognized in
the proposal of encasing the damaged column.
1. The proposed encasement had to be large to accommodate the excessive buckling of the
damaged column.
2. To construct the encasement the area had to be cleared by removing the debris etc. and
such operation had to be carried out while the structure was in an unstable condition.
3. During the permanent restoration of the building, restoration procedure would be
complicated by the presence of the encasement.
Due to these reasons the second proposal was accepted and implemented. Reinforced steel
bars 32mm diameter were used to transfer the load from damaged columns to the adjoining columns
and 350x350 'H' sections were used to counterbalance the horizontal force exerted on the adjoining
columns by the tie action.
3.2 Fire damage
Damage caused by the fire that spread from floor to floor of the centre and south towers was widespread and insidious. It was more difficult to assess damages caused by fire and therefore most of
the effort of the survey was directed at ascertaining the extent of damage caused by fire.
3.2.1 Effect of fire on reinforced concrete
Residual strength of concrete that has been exposed to high temperatures depends on
temperatures reached, mix proportions of the concrete and loading conditions at the time of exposure.
It is generally believed that temperatures up to 300°C do not seriously affect residual strength of
structural concrete. However, at temperatures greater than 500°C, compressive strength of concrete
reduces very rapidly and to only a fraction of its original value on reaching temperatures in the order
of 650°C. The concrete under compressive stress retains a higher compressive strength after
exposure to high temperatures in comparison with unstressed concrete. Development of red or pink
colouration in concrete containing natural sands or aggregates having appreciable iron oxide content
occurs at 300°C. This property is especially significant in assessment of fire damage, since 300°C is
the temperature at which concrete begins to deteriorate.
Elastic modulus of concrete drastically reduces if concrete is heated to temperatures in excess of
300°C and cooled. Ultrasonic pulse velocity in concrete has a direct relationship to its elastic modulus
and can therefore be used as a means of ascertaining the thermal history of concrete.
Two types of spalling are known to occur in concrete when exposed to heat. Explosive spalling
occurs in the first 30 minutes of exposure to heat, and proceeds with a series of disruptions, each
locally removing layers of concrete to a shallow depth. Instances of this type of spalling were
observed on the underside of some preflex beams. Another type of spalling, often referred to as
'sloughing off' is a gradual separation which occurs in beams and columns, where cracks form parallel
along arises or along any other plane of weakness such as along reinforcement bars. This type of
spalling was observed in the columns, beams, slabs and rib beams of the fire damaged areas of the
building.
At high temperatures, unrestrained expansion of steel reinforcement is greater than that of most
concrete in heavily reinforced members, which can result in bursting stresses and cracking of
concrete around the steel.
Exposure to high temperatures also weakens bond strength of reinforcement bars with concrete.
In the case of ribbed bars, loss of bond occurs at a temperature at which compressive strength of
concrete is also lost, whereas in the case of plain bars, loss of bond occurs at much lower
temperatures. Loss of bond directly affects crack width control and consequently reduces durability of
the structure.
Steel also loses its strength at high temperatures and is usually the reason if excessive
deflections are observed after a fire. Exposure of mild steel to temperatures less than 600°C does not
significantly affect yield strength after cooling.
3.2.2 Assessment of fire damage
Fire damage assessment was carried out in two stages namely, Visual Assessment and
Quantitative Assessment.
3.2.2.1 Visual assessment
Visual assessment of damage to various structural components was based on recommendations
of the Concrete Society, Technical Report No. 33, "Assessment of Repair of Fire-damaged Concrete
Structures and Repair by Gunite', 1990 Edition. Purpose of this assessment was to obtain an overall
picture of the fire damage, in order to make an initial assessment of the repairs likely to be required,
and to provide some guidance for second stage of survey that was to follow.
Classification of damages suffered by various components, i.e. columns, main beams (preflex
beams), secondary beams (web of ribbed slab) and the slabs was carried out generally in accordance
with the guidelines given in the Concrete Society Report. These guidelines are reproduced in Table 1
Class of
Damage
Element
Surface appearance of concrete
Condition of
plaster/finish
0
Any
1
Column
Wall
2
3
4
Colour
Structural condition
Crazing
Cracks
Deflection/
Distortion
none
,,
none
,,
,,
,,
,,
,,
Unaffected or beyond extent of fire
some peeling
,,
normal
,,
slight
,,
minor
,,
Floor
,,
,,
,,
,,
Beam
,,
,,
,,
,,
Column
Exposure and condition of main
reinforcement (1)
Spalling
sustainable loss
pink (2)
moderate
Wall
,,
,,
,,
,,
Floor
,,
,,
,,
,,
Beam
,,
,,
,,
,,
Column
total loss
none exposed
,,
,,
,,
,,
very minor exposure
localized corners
upto 25% exposed, none buckled
none
none
localized to patches
upto 10% exposed, all adhering
,,
,,
,,
,,
,,
,,
minor
none
small
not
significant
,,
,,
,,
,,
localized to corners,
minor to soffit
upto 25% exposed, none buckled
extensive
considerable to corners
,,
considerable to surface
upto 50% exposed, not more than one bar
buc k l ed
upto 20% exposed, generally adhering
Wall
,,
,,
whitish
grey
,,
Floor
,,
,,
,,
,,
considerable to soffit
Beam
,,
,,
,,
,,
considerable to corners,
sides, soffit
upto 50% exposed, not more than one bar
buckled
,,
Almost all surface spalled
over 50% exposed, more than one bar
buc k l ed
over 20% exposed, much separated from
concrete
,,
,,
major
Column
destroyed
buff
surface lost
Wall
,,
,,
,,
,,
,,
Floor
,,
,,
,,
,,
,,
Beam
,,
,,
,,
,,
,,
,,
,,
over 50% exposed, more than one bar
exposed
Notes:(1) In the case of beams and columns the main reinforcement shall be assumed to be in the corners only unless other information exists
(2) Pink discolouration due to ferric salts in aggregates. Not always present and seldom in calcareous aggregate.
Table 1 : Visual damage classification for reinforced concrete elements
,,
severe and
significant
,, ,,
,,
,,
,,
,,
,,
any
distortion
severe and
significant
,, ,,
,,
,,
,,
and Table 2. Results of this visual survey together with damage classification were indicated for each
member in tabular form and in diagrammatic form in the survey report.
Class of
damage
Repair
Classification
Repair Requirements
Class 0
Decoration
Redecoration if required
Class 1
Superficial
Superficial repair of slight damage not needing fabric
reinforcement
Class 2
General repair
Non-structural or minor structural repair restoring cover
to reinforcement where this has been partly lost. The
gunite or repair material reinforced with a nominal light
fabric
Class 3
Principal repair
Strengthening repair reinforced in accordance with the
load-carrying requirement of the member. Concrete
and reinforcement strength may be significantly
reduced requiring check by design procedure.
Class 4
Major repair
Major strengthening repair with original concrete and
reinforcement written down to zero strength, or
demolition and recasting
Table 2 : Repair Classification
History of a fire can be roughly estimated by visual observation of debris found at the site. Some
of the debris found in the Central Bank Building after the fire are shown in Table 3 with their condition
and corresponding temperature.
Material
Glass
Aluminium Partition
Brass Fittings
Copper Wires
Condition
Melted
Melted
Melted
Not Melted
Melting Temperature
600°C
650°C
850°C
1100°C
Table 3 : Condition of debris found in Central Bank Building
From the above observations, it can be inferred that maximum temperature attained during the
fire was below 1100°C but above 850°C.
3.2.2.2 Quantitative assessment
Purpose of the second stage assessment was to determine the extent to which performance of
various structural component had been affected by fire.
One method of determining the extent of fire damage is by deducing the temperature history of
building elements on the basis of fire characteristics. Severity of a fire is influenced by three
parameters. These are, (1) Fire load (quantity, type and distribution); (2) Ventilation (area, height,
location); (3) Compartment (floor area, surface area, shape, thermal characteristics). Fire load can be
expressed in terms of equivalent weight of wood having calorific value of 16-18 MJ/kg or directly, in
terms of MJ or M Cal. 'Specific fire load' or 'Fire load density' of a conventional office space is
generally taken as 25 kg/m2 of wood or 500 MJ/m2. Ventilation is generally expressed as the 'opening
factor', F, defined as Av.hv½/At, where Av is area of ventilation (m2), At is surface area of
compartment (m2) and hv is window height (m).
In building fires, there are two regimes of combustion, identified as 'Ventilation controlled fire' and
'Fuel surface controlled fire'.
'Ventilation controlled' fires occur where availability of air is limited whereas in 'fuel surface
controlled' fires, air is freely available so the limit is imposed by availability of combustible materials.
The fire at the Central Bank Building was of 'fuel surface controlled' type due to the large extent of
glass frontage that had been damaged by the explosion giving free access to air.
In the case of 'fuel surface controlled' fire, the period of fully developed burning has been shown
to be equal to 151/Ø seconds, where Ø is the specific surface of combustible material. Value of Ø for
furniture in an office environment has been determined to be typically 0.13 m2/kg. Therefore the
maximum expected duration of a fully developed 'fuel surface controlled' fire is about 1160 seconds or
19 minutes. Making allowance for variation in combustibles, period of 'fully developed burning' was
predicted to be no longer than thirty minutes in the Central Bank fire.
Using the above information (ie. peak temperature of 850°C and maximum fully developed
burning duration of 30 minutes) temperature history of various structural members was deduced using
available charts produced by standard BS furnace tests and by computer generation. In this analyses
peak temperature distribution of the cross section of each structural component was predicted and
residual compressive strength of each layer was determined from Compressive Strength Vs Exposure
Temperature charts. These results were used to determine the extent of the deterioration of concrete
and steel and bond between steel and concrete.
Schmidt Hammer tests were carried out to measure hardness of surface of fire damaged
concrete. Although there is no direct relationship between surface hardness and strength, an
empirical relationship does exist. Due to the convenience and the low cost involved in using this
apparatus, concrete members affected by fire were extensively checked using Schmidt Hammer for
correlation with results obtained from other tests.
Use of ultrasonic pulse velocity (UPV) measurement for estimation of concrete strength is
covered in BS 6089 and BS 1881 Part 203. Since there is no fundamental relation between pulse
velocity and compressive strength, estimation can be obtained by correlation with other tests, and this
is used as a basis for comparison of known sound concrete with concrete that is suspect. This
method was found especially useful in the assessment of ribs of ribbed slabs. Table 4 compares the
typical values of compressive strengths estimated from Schmidt Hammer with UPV test results for
various structural elements in areas affected by fire with those in unaffected areas.
Structural Element
North Tower
(Unaffected by fire)
Schmidt Hammer UPV Test
(N/mm2)
(km/s)
South & Central Tower
(Affected by fire)
Schmidt Hammer
UPV Test
(N/mm2)
(km/s)
Preflex Beam
Bottom Flange
Web
50-60
40-46
-
<10-25
<10-46
-
Rib Beam
Side
Soffit
32-44
-
4.0-4.3
-
20-35
<10-35
0.6-4.3
-
Ribbed Slab
Top
Bottom
28-35
40-48
4.0-4.8
-
<10-25
<10-35
0.9-3.8
-
Solid Slab
Top
Bottom
-
-
14-25
10-15
0.9-3.2
-
Columns
40-55
4.0-4.5
<10-50
0.85-4.0
Table 4 - Comparison of typical compressive strengths estimated from Schmidt hammer and
UPV test for various structural elements in areas affected by fire with those in
unaffected areas.
It can be observed that ultrasonic pulse velocity in sound concrete is in excess of 4 km/s as
predicted by standard tests, whereas concrete from all fire affected areas yielded UPV values in the
range of 1 to 4 km/s depending on the degree of deterioration.
The most direct method of estimating strength of in-situ concrete is by testing cores cut from the
structure. However, only a limited number of test cores were extracted from the fire damaged areas
in order to minimize further damages. Cores of sound concrete were extracted from the blast
damaged structural components for comparison purposes.
Table 5 compares compressive strengths of ribbed floor concrete obtained from Schmidt Hammer
Tests and UPV Tests (both are indirect methods to determine the compressive strength of concrete)
with core test results. Average maximum temperatures attained by concrete estimated by UPV
methods are also indicated in the table.
Structural
Component
Slab
Rib-beam
Beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Sample Condition
Sound (unaffected
by fire
- do - do Affected by fire
- do - do - do - do - do - do -
Pulse
velocity
km/s
4.35
4.09
4.08
2.47
2.47
1.69
2.26
2.90
3.29
3.25
UPV Test Results
Temperature
Comp.
attained
Strength
°C
N/mm2
38
525°
525°
575°
550°
500°
475°
475°
37
36
21
21
15
20
23
27
27
Schmidt
Hammer
Results
N/mm2
42
Core
Test
Results
N/mm2
38
44
43
26
25
28
32
34
30
24
24
33
-
Table 5 : Comparison of UPV and Schmidt Hammer estimations with core test results
This comparison reveals that average temperature attained by the concrete under study is in the
order of 500°C which is consistent with the value estimated from fire load calculations explained
before. It further discloses that UPV tests underestimate the residual strength of fire affected concrete
by 10% - 20% while Schmidt Hammer results overestimate the residual strength by 10% - 20%. Over
estimation of residual compressive strength by Schmidt Hammer tests could be due to skin hardening
effect that is believed to occur on fire damaged concrete. However, it can be concluded that both
UPV and Schmidt Hammer tests give decisive results which can be used to determine the state of
damage of fire affected concrete.
Where real behaviour of a structural member is not amenable to calculation, it is generally
ascertained by a load test. Study of a fire-affected structure where doubt exists about its residual
properties is one such case. A load test was especially thought to be useful for checking the
performance of ribs of the floor slab since bond failure of bottom reinforcement had occurred in many
locations. Besides, the effect of loading on cracks in the order of 0.4 mm, observed in a large number
of ribs had to be reliably ascertained.
A load test was carried out on a selected rib beam by loading it up to the designed service load.
Despite the suspected bond loss the beam performed satisfactorily during the test with regard to
deflection with no significant increase of crack widths.
Samples of reinforcing bar extracted from fire-affected and unaffected members were tested and
a loss in tensile property was observed. Loss in tensile property of rebars was observed in fire
affected members where the concrete cover to rebars had been damaged during the fire. However,
the loss of tensile strength in these rebars was not significant to change load bearing capacities of the
fire affected concrete components.
4. Structural Repair
Detailed classification of structural damage according to the guidelines given in Table 2 was given
in the form of drawings in the survey report.
4.1 Repair of blast damage
As stated earlier, explosion damage was intense but concentrated around the area of blast.
Restoration of the structure to its original form required the most affected parts to be demolished and
rebuilt and therefore these damages were categorized under class 4 in the survey report.
Buckled and settled corner column of the centre tower could not be restored by jacking since
settlement had caused plastic extension of reinforcement in slab and floor beams. Therefore, entire
first bay of the centre tower had to be demolished and rebuilt. Major part of the ground floor slab in
front of the building, cantilevered mezzanine floor and the interior mezzanine floor in front area of the
building were demolished and rebuilt.
Cracks in the main columns supporting the tower structure were repaired by epoxy grout injection.
Two front columns (in the vicinity of the bomb blast) with large cracks and small lateral deformation
were encased with reinforced concrete after injection of epoxy grout.
4.2 Repair of fire damage
Fire damage was worst on the top (ninth) floor of both south and centre towers due to the higher
fire load provided by timber partitions that had been installed on these floors. Roof slabs over the
ninth floor of both these towers had been damaged to the extent that they had to be demolished and
rebuilt.
A summary of fire damage to various structural elements along with the repair methods is given in
the following paragraphs.
4.2.1 Columns
All columns were thoroughly checked by 'hammer sounding' followed by Schmidt Hammer testing
of suspect areas. It was apparent that the 40mm thick plaster on the inner faces of columns in
general office areas had provided substantial protection for columns from fire damage. However,
hammer soundings on columns indicated that more than fifty percent of columns in fire affected areas
had been subjected to debonding of main reinforcement from surrounding concrete. Concrete of the
top third of some columns had deteriorated relativity more extensively due to higher temperatures
experienced at upper levels. Aforesaid deterioration was mainly observed in the interior columns
while deterioration decreased in severity outwards, with sound concrete being encountered outside
the window line. Above observations were confirmed by ultrasonic pulse velocity measurements.
Fire damage to most of columns was categorized under class 2. All affected concrete was
removed and replaced with repair mortar or with gunite after sandblasting of reinforcement.
4.2.2 Preflex beams
Preflex beams are essentially GRAY 40 girders of type A52 steel with bottom flange encased in
reinforced concrete to act as a composite beam. 'Preflex' beams were factory produced by intense
artificial flexing of the steel girder in the same direction of flexure as when in use, until maximum
stress in use was attained or exceeded over the whole length, followed by encasing bottom flange in
concrete, and retaining the flexed state until concrete had hardened. On release of the flexed
condition a pre-compression was induced in the concrete encasement. In use, concrete will be
decompressed, but without reaching tensile state. Improved load carrying capacity and stiffness are
given to the steel girder by the above means, which are further improved by composite effect of the
slab concrete and the end restraint provided by monolithic casting of the slab and column concrete at
the column and beam junction with some rebars connecting column and beam sections.
Most preflex beams which had been covered with sooty deposit had not suffered any significant
damage or deterioration due to the lower temperature of exposure. Spalling of the concrete
encasement, severity depending on degree of exposure, was observed in beams which had been
exposed to higher temperatures. Spalling had occurred on the underside of flanges and along
longitudinal reinforcement. In-situ concrete in rib and preflex beam joints had cracked in a few places,
resulting in loss of integrity between ribs of floor slabs and beams. Schmidt Hammer tests on the
underside of preflex beams invariably gave low values despite concrete itself appearing to be
reasonably sound. These results and hammer soundings confirmed that concrete encasements of
preflex beams had debonded from the steel girder in a large number of beams.
Further it was observed that preflex beam and column junctions had been severely damaged
during the fire. This damage would adversely affect the load carrying capacity of the preflex beam
and the lateral stability of the building under the wind loading due to the loss of bending moment
capacity of the joint. It appeared that in the original design bending moment capacity of the joint was
achieved with the top reinforcement bars (25mm dia., 4 Nos.) which had been extended from the
column to the insitu slab over the preflex beam, dowel action of the embedded steel beam in the
column concrete and the bond between the web of the beam and the column concrete. Top
reinforcement bars contribute only to the hogging bending moment capacity and the other two factors
contribute equally to the hogging and sagging moment capacities of the joint. Sagging moment
capacity of this joint is important due to the reversible bending moment induced by the change of wind
direction.
It was assumed that the bond between the steel and concrete was lost completely and dowel
action was reduced by fifty percent due to the fire damage at the beam-column joint.
Most of the fire damaged preflex beams were categorized under class 3. Therefore the repair of
preflex beams was carried out by restoring the damaged beam-column joints and debonded concrete
encasements in order to achieve the original bending moment capacities and stiffnesses.
The restoration of the beam-column junction was achieved by introducing a haunch 1300mm long
and 300mm deep at each end of the beam to improve the bending moment capacities. Sagging
moment capacity was provided by installation of two 16mm diameter rebars at the bottom of the
haunch having sufficient anchorage length into the column .
The restoration of the bottom concrete encasement was important because it provides the
required additional stiffness and fire protection to the steel beam. However, replacement of debonded
concrete alone would not give the required stiffness as well as the fire protection due to the tensile
stresses and cracks which would be induced under the live load application. Therefore it was decided
to 'preflex' the beam in-situ so as to induce required compressive stress in the new concrete
encasement. This was achieved by applying two 40 kN loads at distances of 2.60m away from the
interior face of each column. These loads were applied using two hydraulic jacks which were
connected to one hydraulic circuit. After the loads were applied, two steel supports capable of holding
the required load of 40 kN were installed next to the hydraulic jacks. Steel supports were tightly fitted
into the space between preflex beam in the upper floor and the beam subject to pre-flexing. The jacks
were removed after installation of steel supports firmly in place. The bottom flange was encased to
the original profile with two haunches at each end by using a proprietary type of concrete. The
vertical steel supports were removed once the new concrete encasement achieved its required
strength.
4.2.3 Ribbed slab
General deterioration of concrete in a ribbed slab when exposed to fire is greater in comparison to
that of a flat slab due to its thin sections and greater area of exposure.
Observation of colour change in concrete and UPV measurements confirmed that floor terrazzo
tiles, their mortar bedding and underlying sand cushion had provided effective fire insulation for floor
slabs from above. Test results showed an average residual strength of 20 N/mm2 in both precast rib
beams and in-situ slab concrete (ie. 60% of its original value). Extensive cracking with a range of
crack widths, but most between 0.1mm and 0.5mm were observed on the insitu slabs. This was to be
expected in view of the wide spacing (225mm) between reinforcement in both directions. Ribs had
vertical cracks which passed completely through with widths ranging between 0.1mm and 0.4mm.
Horizontal cracks appeared in concrete along the bottom reinforcement of some ribs caused by
differential expansion of steel and concrete. Even where there were no horizontal cracks, hammer
soundings revealed debonding of bottom reinforcement on a fairly large scale. Despite these defects,
the load test demonstrated that the slab could carry the required load without significant widening of
existing cracks or excessive deformation.
In instances where bottom reinforcement had debonded, or concrete had deteriorated, weak
concrete was removed to an extent well above the level of the bottom rebar and replaced with a
proprietary type of pre-mixed concrete. Since a large number of ribs were required to be restored by
this method, repair per 'run' was limited to every other rib in order to avoid the need for special
supports to carry the load of the slab. Special formwork was erected for the reprofiling work and
removed after the required strength of 25 N/mm2 was achieved within 18 hours after casting. Once a
set of rib beams were repaired as described above, the balance ribs (every other rib) were restored in
similar manner in the next 'run'.
The soffit which had been cracked significantly was sprayed with shotcrete mortar, suitable for
overhead application, having compressive strength of 55 N/mm2 after 28 days. The thickness of the
mortar layer was in the range of 10 to 20mm. The mortar layer improved the concrete cover to the
reinforcement and at the same time sealed cracks in the floor slab from underneath and thus
improved the durability of the slab. Large cracks of the slab were treated with low viscous epoxy
material by the penetration method from the top of the slab. The original debonded terrazzo floor was
replaced by well bonded 40mm thick concrete screed incorporating mesh reinforcement to improve
the load distribution. Light weight PVC tiles were used as the floor finish of the slab, to compensate
for the additional loads induced by the shotcrete layer and floor screed.
4.2.4 Solid slabs
Spalling of concrete cover accompanied by exposure and buckling of reinforcement was observed
in almost all solid slabs in fire affected areas of the south and centre towers. Most of these slabs
were categorized under class 3 of fire damage. Repair of solid slabs required all loose and fire
damaged concrete to be removed, additional bottom reinforcement to be provided according to the
design requirements and doweled into the slab and the soffit to be treated with 40mm thick gunite.
5. Conclusion
The paper describes the bomb and fire damage that occurred at the Central Bank of Sri Lanka
Head Office Building following a truck bomb attack. A comprehensive damage survey was carried out
to assess the extent of the damage, to evaluate the residual material strengths and to propose the
possible repair techniques in order to restore the building to its original load bearing capacity.
Specifications were prepared on the basis of the damage report and a contract was negotiated and
awarded for restoration after a thorough study of 'repair' and 'demolish and re-build' options. The
repair of bomb and fire damages to the building was completed successfully according to the
specified repair techniques within the estimated cost. However, additional repairs had also to be
carried out on the external facade of the building due to the corrosion damage of the reinforcement
that had taken place over a period of time since the construction of the building 30 years before.
6. Acknowledgement
The paper is based on the bomb damage survey carried out on the Central Bank building and
author's experience gained in the project from its inception to the completion of work. Author is
grateful to the management of the Central Bank of Sri Lanka and the Central Engineering
Consultancy Bureau for permission granted to present this paper. He also thanks his counterpart
engineers of Ms. Ed Zublin AG and his colleagues at CECB whose contributions have enhanced the
final quality of the restoration work.
References
[1] The Concrete Society, Assessment and Repair of Fire-damaged Concrete Structures and Repair
by Gunite. Technical Report No. 33. The Concrete Society, London 1990.
[2] ACI Committee 216, Guide for Determining the Fire Endurance of Concrete Elements, ACI 216 R89, ACI Committee 216, 1989.
[3] The Institution of Structural Engineers, Appraisal of Existing Structures, The Institution of Structural
Engineers, London, 1980.
[4] British Standards Institution, B.S. 6089, Guide to Assessment of Concrete Strength in Existing
Structures, 1981.
[5] British Standards Institution, BS 1881, Part 203, Testing Concrete, Recommendations for
Measurement of Velocity of Ultrasonic Pulses in Concrete, 1986.
[6] H.L. Malhotra, Spalling of Concrete in Fires, Technical Note 118, Construction Industry Research
and Information Association, London, 1984.
[7] K. Mani and N. Lakshmanan, Determining the extent of damage due to fire in concrete structures
by ultrasonic pulse velocity measurements, Indian Concrete Journal, July - 1986.
[8] S.C. Chakrabarti and S.K. Mital, Structural damage and remedial measures for the fire affected
library block in Chandigarh, Indian Concrete Journal, Nov - 1986.
[9] V Diederichs and U. Schneider, Bond strength at high temperatures, Magazine of Concrete
Research Vol 33, No. 115, June 1981.
[10] The Institution of Structural Engineers, The Structural Engineer's Response to Explosion
Damage, The Institution of Structural Engineers, London, 1995.
[11] British Standards Institution, BS 1881, Part 202, Testing Concrete, Recommendations for Surface
Hardness Testing by Rebound Hammer, 1986.
[12] R.D. Anchor, H.L. Malhotra and J.A. Purkiss, Design of Structures Against Fire, Elsevier Applied
Science Publishers Ltd., 1986.
[13] T.Z. Harmathy, Fire Safety Design & Concrete, Concrete Design and Construction Series, 1992.
Fire Damaged Central Bank Building in Sri Lanka
K.L.S. Sahabandu, Central Engineering Consultancy Bureau, Sri Lanka
Abstract
The paper describes the methods adopted to assess and classify the bomb and fire
damage on structural members, which include both visual and quantitative
techniques. Repair techniques which can be broadly categorized into two types (viz,
replacement of damaged concrete with proprietary cement mortars and restoration of
the affected elements by various strengthening methods) are explained together with
relevant details.
1. Introduction
The Central Bank Head Office Building, a prominent land mark for over 30 years of Colombo,
capital of Sri Lanka, was subject to a suicide truck bomb attack on 31st January 1996, which claimed
the lives of at least 85 persons and injured some 1500 more, while damaging severely the Central
Bank Building and the buildings in the vicinity and which sent shock waves throughout the island
nation.
The Central Engineering Consultancy Bureau (CECB) who had been earlier awarded the
consultancy for the Central Bank Extension Building Project in 1994, were entrusted with the
consultancy for the rehabilitation work on the damaged building. Following the explosion, CECB was
assigned to carry out a detailed survey of damages caused by the blast and the resulting fire. The
report, which was based on a comprehensive structural assessment of the entire building included the
results of non-destructive testings and load testings, was handed over to the Central Bank within four
weeks of commencement.
Meanwhile, CECB who was involved in the clearing operations on the damaged building, also
designed a temporary support structure which was put up within a couple of weeks of the explosion
and which prevented further collapse of the severely damaged area of the building and made the site
safe to work in.
In the wake of the survey of damages to the Bank Building, CECB was also given the task of
preparing documents for international tendering for the rehabilitation work on the building. This
assignment was completed within four months but subsequently the Central Bank took a decision to
award the contract to Ms. Ed Zublin AG of Germany, who already held the contract for the Extension
Building Project. The reason for this was the time factor involved. Negotiating a deal with Zublin, who
were already on site was a lot faster than going through the process of international tendering.
2. Building and its Salient Features
2.1 Building - general
The building which has a total floor area of 26,000 square metres consists of three tower blocks of
9 floors each which are interconnected by two service cores in front and rise above the top of the
mezzanine and ground floors, which floors extend over the full length and combined width of all three
towers. The building has a split level basement consisting of main vault, day vaults, office and service
areas and mechanical and electrical plant rooms. Each tower block is approximately 48m long and
11.5m wide, with approximately 10m space between them. The building is approximately 54m tall
from its foundation level, and occupies an area of 45m x 68m at ground level.
2.2 Building structure
The building is of reinforced concrete construction and stands on a pile foundation consisting of
some 260 cast-insitu piles. Entire basement up to and including the major part of ground floor have
been constructed of conventional in-situ reinforced concrete as a single monolith. The central tower
and a large part of the two service shafts on either side form one structural entity. Part of the two
service shafts close to the two end towers are integral with the respective towers and separated from
the main service shaft and central tower by two movement joints.
Each tower is supported on two rows of tapering columns spaced approximately 11.5m apart with
columns at 5.2m intervals. Columns are 1600mm deep up to the first floor level tapering down to
610mm at top with a uniform width of 305mm throughout.
Each pair of columns, supports a 508mm deep 'preflex' beam at each floor level. The beams
were factory produced and preflexed for use as composite beams. Ribs of the floor slab which are
placed at 685mm centres, are 100mm wide where they join the slab and taper down to 75mm at
bottom. Rib beams are reinforced with a single 25mm dia. mild steel bar in the longitudinal direction
at the bottom, with 6mm dia. shear links at a spacing of 150mm and 300mm at supports and mid span
respectively. Slab between ribs varies in thickness from 65mm at ribs to 50mm at mid span, and is
reinforced with 6mm diameter mild steel bars at 225mm centres in both directions. Ribs rest on
bottom flange of the preflex beams and are encased in concrete with 250mm square openings
coinciding with 203mm dia. holes in the preflex beam web, which were required for air conditioning
ducts to pass through.
3. Damages to the Building
3.1 Blast damage
The bomb resulted in massive damage to the building structure in the immediate vicinity of the
blast. Though severe, blast damage was localized and visible. Blast disintegrated the ground floor
slab at the entrance over the basement and made a two metre deep crater just outside the basement
perimeter wall. One of the four supporting columns of the mezzaninie floor was destroyed causing the
collapse of nearly two third of the cantilevered mezzanine floor with its roof. A corner column in the
centre tower was sheared at ground level and broken at two other places causing it to buckle and
settle by 300mm with all the upper floors in that corner.
Mezzanine floor in the vicinity of the blast was completely destroyed. The blast created an
upward bulge on the first floor of the centre tower. Spalled concrete on preflex beam flanges showed
that part of the slab of the first floor lifted upwards under the blast pressure. The ground floor slab
along with beams in the front area of the south tower sagged under the blast, and the mezzanine floor
above it heaved upwards.
Among all blast damages referred to above, the most critical was the damage to the corner
column in the centre tower, which column therefore had to be strengthened quickly to prevent further
collapse of the building and to be able to continue the clearing operations. Two methods of temporary
restoration were considered, viz,
1. encasing the damaged column from basement floor, up to the first floor level, where the
transfer of load would take place from column to the encasing and back to the column by the bond
between the new concrete and the existing concrete.
2. transferring the load at the damaged column to the two columns on either sides of the
damaged column by ties at different levels.
After a close examination of the damaged structure, following shortcomings were recognized in
the proposal of encasing the damaged column.
1. The proposed encasement had to be large to accommodate the excessive buckling of the
damaged column.
2. To construct the encasement the area had to be cleared by removing the debris etc. and
such operation had to be carried out while the structure was in an unstable condition.
3. During the permanent restoration of the building, restoration procedure would be
complicated by the presence of the encasement.
Due to these reasons the second proposal was accepted and implemented. Reinforced steel
bars 32mm diameter were used to transfer the load from damaged columns to the adjoining columns
and 350x350 'H' sections were used to counterbalance the horizontal force exerted on the adjoining
columns by the tie action.
3.2 Fire damage
Damage caused by the fire that spread from floor to floor of the centre and south towers was widespread and insidious. It was more difficult to assess damages caused by fire and therefore most of
the effort of the survey was directed at ascertaining the extent of damage caused by fire.
3.2.1 Effect of fire on reinforced concrete
Residual strength of concrete that has been exposed to high temperatures depends on
temperatures reached, mix proportions of the concrete and loading conditions at the time of exposure.
It is generally believed that temperatures up to 300°C do not seriously affect residual strength of
structural concrete. However, at temperatures greater than 500°C, compressive strength of concrete
reduces very rapidly and to only a fraction of its original value on reaching temperatures in the order
of 650°C. The concrete under compressive stress retains a higher compressive strength after
exposure to high temperatures in comparison with unstressed concrete. Development of red or pink
colouration in concrete containing natural sands or aggregates having appreciable iron oxide content
occurs at 300°C. This property is especially significant in assessment of fire damage, since 300°C is
the temperature at which concrete begins to deteriorate.
Elastic modulus of concrete drastically reduces if concrete is heated to temperatures in excess of
300°C and cooled. Ultrasonic pulse velocity in concrete has a direct relationship to its elastic modulus
and can therefore be used as a means of ascertaining the thermal history of concrete.
Two types of spalling are known to occur in concrete when exposed to heat. Explosive spalling
occurs in the first 30 minutes of exposure to heat, and proceeds with a series of disruptions, each
locally removing layers of concrete to a shallow depth. Instances of this type of spalling were
observed on the underside of some preflex beams. Another type of spalling, often referred to as
'sloughing off' is a gradual separation which occurs in beams and columns, where cracks form parallel
along arises or along any other plane of weakness such as along reinforcement bars. This type of
spalling was observed in the columns, beams, slabs and rib beams of the fire damaged areas of the
building.
At high temperatures, unrestrained expansion of steel reinforcement is greater than that of most
concrete in heavily reinforced members, which can result in bursting stresses and cracking of
concrete around the steel.
Exposure to high temperatures also weakens bond strength of reinforcement bars with concrete.
In the case of ribbed bars, loss of bond occurs at a temperature at which compressive strength of
concrete is also lost, whereas in the case of plain bars, loss of bond occurs at much lower
temperatures. Loss of bond directly affects crack width control and consequently reduces durability of
the structure.
Steel also loses its strength at high temperatures and is usually the reason if excessive
deflections are observed after a fire. Exposure of mild steel to temperatures less than 600°C does not
significantly affect yield strength after cooling.
3.2.2 Assessment of fire damage
Fire damage assessment was carried out in two stages namely, Visual Assessment and
Quantitative Assessment.
3.2.2.1 Visual assessment
Visual assessment of damage to various structural components was based on recommendations
of the Concrete Society, Technical Report No. 33, "Assessment of Repair of Fire-damaged Concrete
Structures and Repair by Gunite', 1990 Edition. Purpose of this assessment was to obtain an overall
picture of the fire damage, in order to make an initial assessment of the repairs likely to be required,
and to provide some guidance for second stage of survey that was to follow.
Classification of damages suffered by various components, i.e. columns, main beams (preflex
beams), secondary beams (web of ribbed slab) and the slabs was carried out generally in accordance
with the guidelines given in the Concrete Society Report. These guidelines are reproduced in Table 1
Class of
Damage
Element
Surface appearance of concrete
Condition of
plaster/finish
0
Any
1
Column
Wall
2
3
4
Colour
Structural condition
Crazing
Cracks
Deflection/
Distortion
none
,,
none
,,
,,
,,
,,
,,
Unaffected or beyond extent of fire
some peeling
,,
normal
,,
slight
,,
minor
,,
Floor
,,
,,
,,
,,
Beam
,,
,,
,,
,,
Column
Exposure and condition of main
reinforcement (1)
Spalling
sustainable loss
pink (2)
moderate
Wall
,,
,,
,,
,,
Floor
,,
,,
,,
,,
Beam
,,
,,
,,
,,
Column
total loss
none exposed
,,
,,
,,
,,
very minor exposure
localized corners
upto 25% exposed, none buckled
none
none
localized to patches
upto 10% exposed, all adhering
,,
,,
,,
,,
,,
,,
minor
none
small
not
significant
,,
,,
,,
,,
localized to corners,
minor to soffit
upto 25% exposed, none buckled
extensive
considerable to corners
,,
considerable to surface
upto 50% exposed, not more than one bar
buc k l ed
upto 20% exposed, generally adhering
Wall
,,
,,
whitish
grey
,,
Floor
,,
,,
,,
,,
considerable to soffit
Beam
,,
,,
,,
,,
considerable to corners,
sides, soffit
upto 50% exposed, not more than one bar
buckled
,,
Almost all surface spalled
over 50% exposed, more than one bar
buc k l ed
over 20% exposed, much separated from
concrete
,,
,,
major
Column
destroyed
buff
surface lost
Wall
,,
,,
,,
,,
,,
Floor
,,
,,
,,
,,
,,
Beam
,,
,,
,,
,,
,,
,,
,,
over 50% exposed, more than one bar
exposed
Notes:(1) In the case of beams and columns the main reinforcement shall be assumed to be in the corners only unless other information exists
(2) Pink discolouration due to ferric salts in aggregates. Not always present and seldom in calcareous aggregate.
Table 1 : Visual damage classification for reinforced concrete elements
,,
severe and
significant
,, ,,
,,
,,
,,
,,
,,
any
distortion
severe and
significant
,, ,,
,,
,,
,,
and Table 2. Results of this visual survey together with damage classification were indicated for each
member in tabular form and in diagrammatic form in the survey report.
Class of
damage
Repair
Classification
Repair Requirements
Class 0
Decoration
Redecoration if required
Class 1
Superficial
Superficial repair of slight damage not needing fabric
reinforcement
Class 2
General repair
Non-structural or minor structural repair restoring cover
to reinforcement where this has been partly lost. The
gunite or repair material reinforced with a nominal light
fabric
Class 3
Principal repair
Strengthening repair reinforced in accordance with the
load-carrying requirement of the member. Concrete
and reinforcement strength may be significantly
reduced requiring check by design procedure.
Class 4
Major repair
Major strengthening repair with original concrete and
reinforcement written down to zero strength, or
demolition and recasting
Table 2 : Repair Classification
History of a fire can be roughly estimated by visual observation of debris found at the site. Some
of the debris found in the Central Bank Building after the fire are shown in Table 3 with their condition
and corresponding temperature.
Material
Glass
Aluminium Partition
Brass Fittings
Copper Wires
Condition
Melted
Melted
Melted
Not Melted
Melting Temperature
600°C
650°C
850°C
1100°C
Table 3 : Condition of debris found in Central Bank Building
From the above observations, it can be inferred that maximum temperature attained during the
fire was below 1100°C but above 850°C.
3.2.2.2 Quantitative assessment
Purpose of the second stage assessment was to determine the extent to which performance of
various structural component had been affected by fire.
One method of determining the extent of fire damage is by deducing the temperature history of
building elements on the basis of fire characteristics. Severity of a fire is influenced by three
parameters. These are, (1) Fire load (quantity, type and distribution); (2) Ventilation (area, height,
location); (3) Compartment (floor area, surface area, shape, thermal characteristics). Fire load can be
expressed in terms of equivalent weight of wood having calorific value of 16-18 MJ/kg or directly, in
terms of MJ or M Cal. 'Specific fire load' or 'Fire load density' of a conventional office space is
generally taken as 25 kg/m2 of wood or 500 MJ/m2. Ventilation is generally expressed as the 'opening
factor', F, defined as Av.hv½/At, where Av is area of ventilation (m2), At is surface area of
compartment (m2) and hv is window height (m).
In building fires, there are two regimes of combustion, identified as 'Ventilation controlled fire' and
'Fuel surface controlled fire'.
'Ventilation controlled' fires occur where availability of air is limited whereas in 'fuel surface
controlled' fires, air is freely available so the limit is imposed by availability of combustible materials.
The fire at the Central Bank Building was of 'fuel surface controlled' type due to the large extent of
glass frontage that had been damaged by the explosion giving free access to air.
In the case of 'fuel surface controlled' fire, the period of fully developed burning has been shown
to be equal to 151/Ø seconds, where Ø is the specific surface of combustible material. Value of Ø for
furniture in an office environment has been determined to be typically 0.13 m2/kg. Therefore the
maximum expected duration of a fully developed 'fuel surface controlled' fire is about 1160 seconds or
19 minutes. Making allowance for variation in combustibles, period of 'fully developed burning' was
predicted to be no longer than thirty minutes in the Central Bank fire.
Using the above information (ie. peak temperature of 850°C and maximum fully developed
burning duration of 30 minutes) temperature history of various structural members was deduced using
available charts produced by standard BS furnace tests and by computer generation. In this analyses
peak temperature distribution of the cross section of each structural component was predicted and
residual compressive strength of each layer was determined from Compressive Strength Vs Exposure
Temperature charts. These results were used to determine the extent of the deterioration of concrete
and steel and bond between steel and concrete.
Schmidt Hammer tests were carried out to measure hardness of surface of fire damaged
concrete. Although there is no direct relationship between surface hardness and strength, an
empirical relationship does exist. Due to the convenience and the low cost involved in using this
apparatus, concrete members affected by fire were extensively checked using Schmidt Hammer for
correlation with results obtained from other tests.
Use of ultrasonic pulse velocity (UPV) measurement for estimation of concrete strength is
covered in BS 6089 and BS 1881 Part 203. Since there is no fundamental relation between pulse
velocity and compressive strength, estimation can be obtained by correlation with other tests, and this
is used as a basis for comparison of known sound concrete with concrete that is suspect. This
method was found especially useful in the assessment of ribs of ribbed slabs. Table 4 compares the
typical values of compressive strengths estimated from Schmidt Hammer with UPV test results for
various structural elements in areas affected by fire with those in unaffected areas.
Structural Element
North Tower
(Unaffected by fire)
Schmidt Hammer UPV Test
(N/mm2)
(km/s)
South & Central Tower
(Affected by fire)
Schmidt Hammer
UPV Test
(N/mm2)
(km/s)
Preflex Beam
Bottom Flange
Web
50-60
40-46
-
<10-25
<10-46
-
Rib Beam
Side
Soffit
32-44
-
4.0-4.3
-
20-35
<10-35
0.6-4.3
-
Ribbed Slab
Top
Bottom
28-35
40-48
4.0-4.8
-
<10-25
<10-35
0.9-3.8
-
Solid Slab
Top
Bottom
-
-
14-25
10-15
0.9-3.2
-
Columns
40-55
4.0-4.5
<10-50
0.85-4.0
Table 4 - Comparison of typical compressive strengths estimated from Schmidt hammer and
UPV test for various structural elements in areas affected by fire with those in
unaffected areas.
It can be observed that ultrasonic pulse velocity in sound concrete is in excess of 4 km/s as
predicted by standard tests, whereas concrete from all fire affected areas yielded UPV values in the
range of 1 to 4 km/s depending on the degree of deterioration.
The most direct method of estimating strength of in-situ concrete is by testing cores cut from the
structure. However, only a limited number of test cores were extracted from the fire damaged areas
in order to minimize further damages. Cores of sound concrete were extracted from the blast
damaged structural components for comparison purposes.
Table 5 compares compressive strengths of ribbed floor concrete obtained from Schmidt Hammer
Tests and UPV Tests (both are indirect methods to determine the compressive strength of concrete)
with core test results. Average maximum temperatures attained by concrete estimated by UPV
methods are also indicated in the table.
Structural
Component
Slab
Rib-beam
Beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Rib beam
Sample Condition
Sound (unaffected
by fire
- do - do Affected by fire
- do - do - do - do - do - do -
Pulse
velocity
km/s
4.35
4.09
4.08
2.47
2.47
1.69
2.26
2.90
3.29
3.25
UPV Test Results
Temperature
Comp.
attained
Strength
°C
N/mm2
38
525°
525°
575°
550°
500°
475°
475°
37
36
21
21
15
20
23
27
27
Schmidt
Hammer
Results
N/mm2
42
Core
Test
Results
N/mm2
38
44
43
26
25
28
32
34
30
24
24
33
-
Table 5 : Comparison of UPV and Schmidt Hammer estimations with core test results
This comparison reveals that average temperature attained by the concrete under study is in the
order of 500°C which is consistent with the value estimated from fire load calculations explained
before. It further discloses that UPV tests underestimate the residual strength of fire affected concrete
by 10% - 20% while Schmidt Hammer results overestimate the residual strength by 10% - 20%. Over
estimation of residual compressive strength by Schmidt Hammer tests could be due to skin hardening
effect that is believed to occur on fire damaged concrete. However, it can be concluded that both
UPV and Schmidt Hammer tests give decisive results which can be used to determine the state of
damage of fire affected concrete.
Where real behaviour of a structural member is not amenable to calculation, it is generally
ascertained by a load test. Study of a fire-affected structure where doubt exists about its residual
properties is one such case. A load test was especially thought to be useful for checking the
performance of ribs of the floor slab since bond failure of bottom reinforcement had occurred in many
locations. Besides, the effect of loading on cracks in the order of 0.4 mm, observed in a large number
of ribs had to be reliably ascertained.
A load test was carried out on a selected rib beam by loading it up to the designed service load.
Despite the suspected bond loss the beam performed satisfactorily during the test with regard to
deflection with no significant increase of crack widths.
Samples of reinforcing bar extracted from fire-affected and unaffected members were tested and
a loss in tensile property was observed. Loss in tensile property of rebars was observed in fire
affected members where the concrete cover to rebars had been damaged during the fire. However,
the loss of tensile strength in these rebars was not significant to change load bearing capacities of the
fire affected concrete components.
4. Structural Repair
Detailed classification of structural damage according to the guidelines given in Table 2 was given
in the form of drawings in the survey report.
4.1 Repair of blast damage
As stated earlier, explosion damage was intense but concentrated around the area of blast.
Restoration of the structure to its original form required the most affected parts to be demolished and
rebuilt and therefore these damages were categorized under class 4 in the survey report.
Buckled and settled corner column of the centre tower could not be restored by jacking since
settlement had caused plastic extension of reinforcement in slab and floor beams. Therefore, entire
first bay of the centre tower had to be demolished and rebuilt. Major part of the ground floor slab in
front of the building, cantilevered mezzanine floor and the interior mezzanine floor in front area of the
building were demolished and rebuilt.
Cracks in the main columns supporting the tower structure were repaired by epoxy grout injection.
Two front columns (in the vicinity of the bomb blast) with large cracks and small lateral deformation
were encased with reinforced concrete after injection of epoxy grout.
4.2 Repair of fire damage
Fire damage was worst on the top (ninth) floor of both south and centre towers due to the higher
fire load provided by timber partitions that had been installed on these floors. Roof slabs over the
ninth floor of both these towers had been damaged to the extent that they had to be demolished and
rebuilt.
A summary of fire damage to various structural elements along with the repair methods is given in
the following paragraphs.
4.2.1 Columns
All columns were thoroughly checked by 'hammer sounding' followed by Schmidt Hammer testing
of suspect areas. It was apparent that the 40mm thick plaster on the inner faces of columns in
general office areas had provided substantial protection for columns from fire damage. However,
hammer soundings on columns indicated that more than fifty percent of columns in fire affected areas
had been subjected to debonding of main reinforcement from surrounding concrete. Concrete of the
top third of some columns had deteriorated relativity more extensively due to higher temperatures
experienced at upper levels. Aforesaid deterioration was mainly observed in the interior columns
while deterioration decreased in severity outwards, with sound concrete being encountered outside
the window line. Above observations were confirmed by ultrasonic pulse velocity measurements.
Fire damage to most of columns was categorized under class 2. All affected concrete was
removed and replaced with repair mortar or with gunite after sandblasting of reinforcement.
4.2.2 Preflex beams
Preflex beams are essentially GRAY 40 girders of type A52 steel with bottom flange encased in
reinforced concrete to act as a composite beam. 'Preflex' beams were factory produced by intense
artificial flexing of the steel girder in the same direction of flexure as when in use, until maximum
stress in use was attained or exceeded over the whole length, followed by encasing bottom flange in
concrete, and retaining the flexed state until concrete had hardened. On release of the flexed
condition a pre-compression was induced in the concrete encasement. In use, concrete will be
decompressed, but without reaching tensile state. Improved load carrying capacity and stiffness are
given to the steel girder by the above means, which are further improved by composite effect of the
slab concrete and the end restraint provided by monolithic casting of the slab and column concrete at
the column and beam junction with some rebars connecting column and beam sections.
Most preflex beams which had been covered with sooty deposit had not suffered any significant
damage or deterioration due to the lower temperature of exposure. Spalling of the concrete
encasement, severity depending on degree of exposure, was observed in beams which had been
exposed to higher temperatures. Spalling had occurred on the underside of flanges and along
longitudinal reinforcement. In-situ concrete in rib and preflex beam joints had cracked in a few places,
resulting in loss of integrity between ribs of floor slabs and beams. Schmidt Hammer tests on the
underside of preflex beams invariably gave low values despite concrete itself appearing to be
reasonably sound. These results and hammer soundings confirmed that concrete encasements of
preflex beams had debonded from the steel girder in a large number of beams.
Further it was observed that preflex beam and column junctions had been severely damaged
during the fire. This damage would adversely affect the load carrying capacity of the preflex beam
and the lateral stability of the building under the wind loading due to the loss of bending moment
capacity of the joint. It appeared that in the original design bending moment capacity of the joint was
achieved with the top reinforcement bars (25mm dia., 4 Nos.) which had been extended from the
column to the insitu slab over the preflex beam, dowel action of the embedded steel beam in the
column concrete and the bond between the web of the beam and the column concrete. Top
reinforcement bars contribute only to the hogging bending moment capacity and the other two factors
contribute equally to the hogging and sagging moment capacities of the joint. Sagging moment
capacity of this joint is important due to the reversible bending moment induced by the change of wind
direction.
It was assumed that the bond between the steel and concrete was lost completely and dowel
action was reduced by fifty percent due to the fire damage at the beam-column joint.
Most of the fire damaged preflex beams were categorized under class 3. Therefore the repair of
preflex beams was carried out by restoring the damaged beam-column joints and debonded concrete
encasements in order to achieve the original bending moment capacities and stiffnesses.
The restoration of the beam-column junction was achieved by introducing a haunch 1300mm long
and 300mm deep at each end of the beam to improve the bending moment capacities. Sagging
moment capacity was provided by installation of two 16mm diameter rebars at the bottom of the
haunch having sufficient anchorage length into the column .
The restoration of the bottom concrete encasement was important because it provides the
required additional stiffness and fire protection to the steel beam. However, replacement of debonded
concrete alone would not give the required stiffness as well as the fire protection due to the tensile
stresses and cracks which would be induced under the live load application. Therefore it was decided
to 'preflex' the beam in-situ so as to induce required compressive stress in the new concrete
encasement. This was achieved by applying two 40 kN loads at distances of 2.60m away from the
interior face of each column. These loads were applied using two hydraulic jacks which were
connected to one hydraulic circuit. After the loads were applied, two steel supports capable of holding
the required load of 40 kN were installed next to the hydraulic jacks. Steel supports were tightly fitted
into the space between preflex beam in the upper floor and the beam subject to pre-flexing. The jacks
were removed after installation of steel supports firmly in place. The bottom flange was encased to
the original profile with two haunches at each end by using a proprietary type of concrete. The
vertical steel supports were removed once the new concrete encasement achieved its required
strength.
4.2.3 Ribbed slab
General deterioration of concrete in a ribbed slab when exposed to fire is greater in comparison to
that of a flat slab due to its thin sections and greater area of exposure.
Observation of colour change in concrete and UPV measurements confirmed that floor terrazzo
tiles, their mortar bedding and underlying sand cushion had provided effective fire insulation for floor
slabs from above. Test results showed an average residual strength of 20 N/mm2 in both precast rib
beams and in-situ slab concrete (ie. 60% of its original value). Extensive cracking with a range of
crack widths, but most between 0.1mm and 0.5mm were observed on the insitu slabs. This was to be
expected in view of the wide spacing (225mm) between reinforcement in both directions. Ribs had
vertical cracks which passed completely through with widths ranging between 0.1mm and 0.4mm.
Horizontal cracks appeared in concrete along the bottom reinforcement of some ribs caused by
differential expansion of steel and concrete. Even where there were no horizontal cracks, hammer
soundings revealed debonding of bottom reinforcement on a fairly large scale. Despite these defects,
the load test demonstrated that the slab could carry the required load without significant widening of
existing cracks or excessive deformation.
In instances where bottom reinforcement had debonded, or concrete had deteriorated, weak
concrete was removed to an extent well above the level of the bottom rebar and replaced with a
proprietary type of pre-mixed concrete. Since a large number of ribs were required to be restored by
this method, repair per 'run' was limited to every other rib in order to avoid the need for special
supports to carry the load of the slab. Special formwork was erected for the reprofiling work and
removed after the required strength of 25 N/mm2 was achieved within 18 hours after casting. Once a
set of rib beams were repaired as described above, the balance ribs (every other rib) were restored in
similar manner in the next 'run'.
The soffit which had been cracked significantly was sprayed with shotcrete mortar, suitable for
overhead application, having compressive strength of 55 N/mm2 after 28 days. The thickness of the
mortar layer was in the range of 10 to 20mm. The mortar layer improved the concrete cover to the
reinforcement and at the same time sealed cracks in the floor slab from underneath and thus
improved the durability of the slab. Large cracks of the slab were treated with low viscous epoxy
material by the penetration method from the top of the slab. The original debonded terrazzo floor was
replaced by well bonded 40mm thick concrete screed incorporating mesh reinforcement to improve
the load distribution. Light weight PVC tiles were used as the floor finish of the slab, to compensate
for the additional loads induced by the shotcrete layer and floor screed.
4.2.4 Solid slabs
Spalling of concrete cover accompanied by exposure and buckling of reinforcement was observed
in almost all solid slabs in fire affected areas of the south and centre towers. Most of these slabs
were categorized under class 3 of fire damage. Repair of solid slabs required all loose and fire
damaged concrete to be removed, additional bottom reinforcement to be provided according to the
design requirements and doweled into the slab and the soffit to be treated with 40mm thick gunite.
5. Conclusion
The paper describes the bomb and fire damage that occurred at the Central Bank of Sri Lanka
Head Office Building following a truck bomb attack. A comprehensive damage survey was carried out
to assess the extent of the damage, to evaluate the residual material strengths and to propose the
possible repair techniques in order to restore the building to its original load bearing capacity.
Specifications were prepared on the basis of the damage report and a contract was negotiated and
awarded for restoration after a thorough study of 'repair' and 'demolish and re-build' options. The
repair of bomb and fire damages to the building was completed successfully according to the
specified repair techniques within the estimated cost. However, additional repairs had also to be
carried out on the external facade of the building due to the corrosion damage of the reinforcement
that had taken place over a period of time since the construction of the building 30 years before.
6. Acknowledgement
The paper is based on the bomb damage survey carried out on the Central Bank building and
author's experience gained in the project from its inception to the completion of work. Author is
grateful to the management of the Central Bank of Sri Lanka and the Central Engineering
Consultancy Bureau for permission granted to present this paper. He also thanks his counterpart
engineers of Ms. Ed Zublin AG and his colleagues at CECB whose contributions have enhanced the
final quality of the restoration work.
References
[1] The Concrete Society, Assessment and Repair of Fire-damaged Concrete Structures and Repair
by Gunite. Technical Report No. 33. The Concrete Society, London 1990.
[2] ACI Committee 216, Guide for Determining the Fire Endurance of Concrete Elements, ACI 216 R89, ACI Committee 216, 1989.
[3] The Institution of Structural Engineers, Appraisal of Existing Structures, The Institution of Structural
Engineers, London, 1980.
[4] British Standards Institution, B.S. 6089, Guide to Assessment of Concrete Strength in Existing
Structures, 1981.
[5] British Standards Institution, BS 1881, Part 203, Testing Concrete, Recommendations for
Measurement of Velocity of Ultrasonic Pulses in Concrete, 1986.
[6] H.L. Malhotra, Spalling of Concrete in Fires, Technical Note 118, Construction Industry Research
and Information Association, London, 1984.
[7] K. Mani and N. Lakshmanan, Determining the extent of damage due to fire in concrete structures
by ultrasonic pulse velocity measurements, Indian Concrete Journal, July - 1986.
[8] S.C. Chakrabarti and S.K. Mital, Structural damage and remedial measures for the fire affected
library block in Chandigarh, Indian Concrete Journal, Nov - 1986.
[9] V Diederichs and U. Schneider, Bond strength at high temperatures, Magazine of Concrete
Research Vol 33, No. 115, June 1981.
[10] The Institution of Structural Engineers, The Structural Engineer's Response to Explosion
Damage, The Institution of Structural Engineers, London, 1995.
[11] British Standards Institution, BS 1881, Part 202, Testing Concrete, Recommendations for Surface
Hardness Testing by Rebound Hammer, 1986.
[12] R.D. Anchor, H.L. Malhotra and J.A. Purkiss, Design of Structures Against Fire, Elsevier Applied
Science Publishers Ltd., 1986.
[13] T.Z. Harmathy, Fire Safety Design & Concrete, Concrete Design and Construction Series, 1992.
