Case studies of some concrete structural failures R bi Whittle Robin Whittl

Design Errors Design errors alone are seldom the cause of the failure of a structure. structure When failure does occur it is usually the result of errors in at least three different aspects of construction (e.g. design, detailing, and construction errors).

Case Study 1 Collapse of a reinforced concrete structure,

This factory building included concrete columns and a steel truss for the roof.

Concrete gutter to factory building

Gutter collapse

Edge beam and column connection

Edge beam Concrete gutter

Column

Loading at failure

Sand 18.6kN

L4

K4

J4

Crane cage g and bricks 6kN

I4

H4

Mechanism of failure  (1)

Piles of sand and bricks Steel truss.

Beam and gutter start to rotate

Cracks start to open up

Mechanism of failure (2)

Sand and bricks collect into corner Steel truss.

Cracks start to open up

Beam and gutter continue to rotate and start to move out and down.

Th outer The t column l b bars stay attached to the beam and drag the outer face of the column away.

The full length links fail as the column bars pull out

Mechanism of failure  (3) Steel truss.

Column outer bars remain attached to the beam.

The outer links fail as the column bars pull out and down.

Beam and gutter fall, dragging the outer column bars with them. Cracks extends down the column l until il either i h the h column bars fail or pull out at a lap.

Reinforcement layout 2T20s T8 links@130

2T8s

Steel truss. (stopped off off at at beam beam cage) cage) T8 links @ 300 (stopped

T10s@100

T20s 2T16s

T8 links @ 300

The beam had been designed to take the torsion from the gutter loading but the joint  with the column had not been designed to take the torsion.

The top two links in the column had been detailed to enclose all the column  main bars but the contractor had reduced their length for ease of construction.   Even if these links had been constructed correctly they would have been Even if these links had been constructed correctly they would have been  inadequate to support the loading. 

Simple model of failure mechanism cracks form

T1 tension

applied torque

G tt slab Gutter l b compression

Applied torque  from  self weight  plus sand and brick pile                =   76 kNm p p Resistance from tensile strength ( y ) of concrete  (say 2MPa)                =   74 kNm Applied load exceeds resistance

tension T2

Column

Summary • Two people were killed • Design error. No consideration was given in the calculations of how forces were transmitted through the joint between the edge beam and column. • Detailing error. Inadequate connection between beam and column reinforcement. • Construction error. Column links were excluded for convenience

Case Study 2 Widespread p cracking g in post-tensioned/reinforced concrete frame This structure did not collapse but the cause of the cracking took a long time to establish and the subsequent litigation was very costly. costly The design was of a five storey car park for which the construction period was extremely short (8 months) months). In consequence the prestressing work was carried out under a very tight schedule.

The post-tensioned prestressed beam was cast in one operation and fully stressed a few days later. Five Fi e days da s after stressing cracks appeared in many parts of the structure.

Problem Whilst the problem was debated, in order to proceed with construction t ti with ith minimum i i delay, d l th the prestressing t i was altered to a two-stage process – only 50% applied at first and the remaining prestress after two weeks weeks. After much discussion it was concluded that when earlyy thermal effects were included with the other shortening effects, the total shortening of the prestressed beams was sufficient ffi i t tto cause th the cracking. ki

Temperature – Time effect on concrete As the chemical action of the cement takes place the concrete heats up. g this p period the During concrete is plastic and the increase in volume results in a fatter rather than a longer beam. However when the cooling phase starts the concrete has hardened and is no longer plastic. The length shortens.

Summary • The cracking of the parts of the structure resulted in a change in the programme of work causing extra cost to the Main Contractor. •

Although the contract was completed on time the ensuing dispute was very costly.

•

When the case came to court it was agreed that it was reasonable to assume that, h at that h time, i a normall engineer i would ld not h have b been expected to include early thermal effects in the analysis. The code of practice stated that ‘unless the lesser section dimension is greater than 600mm and the cement content is greater than 400kg/m3 there is no need to consider early thermal effect’.

•

The result of the investigation brought about a change to normal design procedures. It is now common to consider early thermal effects in the design of long length continuous reinforced and prestressed t d beams b and d slabs. l b

Case Study 3

Temperature effects on a long-span hybrid structure • The ground level of a two storey underground car park slab was not covered. covered Ambient temperature changes caused continual movement. • The structure consisted of 16m spanning hollowcore units bearing on precast concrete beam nibs. • Movement joints had been shown on the drawings but these did not function correctly for a variety of reasons reasons.

Causes of cracking Hard material can prevent movement Movement

Rotation can cause spalling

Friction can cause cracking

(a)

Rotation

Rotation

(b)

(c)

The upper surface of the slab was exposed to the weather and in particular to large variations in temperature temperature. The latter caused movement and rotation of the units and their supports. This resulted in severe cracking of the supporting nibs and in some places cracking at the end of the hollowcore units. units Even after repair of the cracks they reappeared each subsequent year for more than five years.

Possible failure mechanisms for the hollowcore units Shear tension crack

Anchorage slip

Large crack close to support

(a)

(b)

• H Hollowcore ll units it are inherently i h tl vulnerable l bl tto th the effects ff t off cracking ki close to the support as there is no shear reinforcement and they rely on the tension strength of the concrete. • Anchorage bond failure can occur when cracks occur close to the support. This can cause the prestressing strands to slip. The crack size increases until the unit fails either in anchorage bond (a) or in shear (b).

Summary • The reoccurrence of cracking of the supporting nibs each year eventually required a temporary support structure to be built. Protection from falling concrete was also required. •

Concern that the hollowcore slabs might g eventually y fail in shear or anchorage has led to the possibility of a rebuild.

•

The cost of remedial work and litigation fees have escalated each year as the decision on what action to take is delayed.

•

Lesson: It is essential to name a single designer or engineer who retains overall responsibility for the stability of the structure, the compatibility of the design, and details of the parts and components, even where h some or allll off the th d design i iincluding l di d detailing t ili off th those parts t and components are not carried out by this engineer.

Case Study 4 Piled raft for tower block The design of the raft assumed that the walls of the two level basement car parks would act with the raft over the piles to transmit the shear and bending forces to the outer piles piles. The walls of the basement had almost full height openings, placed one above the other, and contained only nominal reinforcement.

Original design The combined strength of these walls plus the 1.5m thick slab was quite inadequate to transmit the loads.

Core walls

Columns

The mistake was discovered whilst the tower block was being g constructed.

Possible line of shear failure

Basement car parks Piled raft

Schematic arrangement of new raft The remedial work required a new raft to be constructed beneath the existing one

Existing raft 3.5m

New supplementary raft to take shear

Piles scabbled to take new concrete

Remedial work Placing of concrete for the lower part of the new raft was carried out conventionally. IIn order d tto achieve hi good db bond d with ith th the existing i ti raft ft th the upper partt off the new raft was packed with single sized aggregate and then grouted with a retarded and fluid cement paste. The grout was introduced under pressure to the back of the pour through a complicated system of metal pipes, pinned to the underside of the existing raft raft. The method produced a wall of grout that extended from top to bottom of the pour and that flowed forward towards the peripheral shutters with the top surface behind the bottom. bottom Pipes were so placed to let the air our in front of the grout surface, grout had arrived, and then to then to indicate where and when the g allow grouting to continue from immediately behind the advancing wall of grout. Grouting was continuous until the work was complete.

Inappropriate Use of Code of Practice Clauses

Case Study y5 Ferry Bridge Cooling Tower Collapse This incident should really be in a chapter of its own as the failure resulted in a change of philosophy in the design code of practice, CP 114. Although there were some defects in wall thickness, it was concluded that the main cause of the collapse was because the design value chosen for the wind load was too small small. No account had been taken of the venturi effect of the wind passing through the towers upwind. This collapse ensured the early adoption of a limit state code of practice in the UK resulting in a completely new approach to design. The first draft of the Unified code appeared in 1968 and in 1972 it was published as CP110. This was the first comprehensive limit state code of practice ever published.

•

In November 1965, three out of  eight cooling towers collapsed  during a high wind. 

•

Each tower was 375 feet high

•

The wind load in the initial design  was seriously underestimated. i l d ti t d

•

CP 114 – Permissible Stress Design  was used.

•

This collapse ensured the early  adoption of limit state design.

•

In 1968 the first draft of the  unified code appeared.

•

In 1972 this was published as  In 1972 this was published as CP110.   

The wind speed passing between the upwind towers increased significantly.  This caused a higher i ifi l hi d hi h wind force on the downwind towers.

It is possible that consideration of the load combinations required for the load combinations required for  limit state design might have been sufficient  to prevent the collapse.

Load combination

Permissible Stress Design: ((CP114)) G x b/2  W x hw New Limit State Design: (CP110) 1 0 (G x b/2)  1.4 1.0 1 4 (W x hw)

W

G

hw

b

Inadequate q Assessment of Critical Force Paths

Case Study 6 Shear wall with holes and corner supports A multi-storey shear wall required so many openings (windows, doors etc) that the load path became very complicated doors, complicated. The designer assumed that the load would flow to the corners at y down the edge g of the wall. each floor and then track vertically Since the wall was built insitu as a homogenous structure, strain compatibility caused the load to flow back into the full width of wall. The result was that several storeys of load were supported by a deep beam at the bottom of the wall, which transferred the load to its end supports at first floor level.

Deep beam failure

Design

Behaviour

The design assumed  that the load from  the wall would be the wall would be  transferred to the  corner columns at  each floor level.

The actual load  transfer to the  corner columns took corner columns took  place at the bottom  of the wall. The height of the  The height of the natural arch was  only  0.6 x the Span.

The tie forces at  each floor level were  small.

This caused large  horizontal tie forces  at the bottom of the  wall. ll No o wall a

(a)

No wall

(b)

Model of force path

The assumed force path  Th df th down the edges would  not require ties at top and not require ties at top and  bottom.   However without these  the actual force path  would cause large cracks  ld l k to open up at the top and  bottom surfaces bottom surfaces 

tie assumed force path

actual force path

tie

without tie reinforcement large cracks form

Case Study 7 Design of boot nibs The conventional assumption taken for the effective depth and lever arm for a short cantilever is unsafe for a boot nib. The design compression zone for such a model would be close to the bottom face of the beam and likely to fall outside the beam reinforcement (both the links and main reinforcement). Strut and tie modelling is helpful to understand why this is so. The strut must be supported mechanically by the reinforcement of the supporting beam. The effective lever arm becomes much smaller and the tension force in the nib top reinforcement much larger than assumed by the short cantilever approach.

Design of boot nibs Traditional design of cantilevers and nibs assumes an effective depth, dc, between the  outer compression face and the centroid of tension reinforcement.   This could cause  the bottom layer of concrete in the cover zone to spall off.   The  strut should be designed to shed its load on to the corner bar of the beam.  The  The strut should be designed to shed its load on to the corner bar of the beam The vertical component of the force is then taken by the link leg.

db Fc = FEd x ac/zb Force in link leg: Ft2d = Fc + Fed

Fc

zb

ac Ft2d

FEd

HEd

This is addition to any  shear that the link is shear that the link is  carrying.

Ft1d

zn dc z c

Poor detailing

Case Study 8

Failure of cellular wall structure p in an offshore oil platform

Sleipner offshore oil platform collapse The platform included a large cellular  concrete structure below the three  towers. During construction the platform  underwent submerging for deck  mating after which the plan was to  raise it again  and tow it to its final  position in the oil field. h l f ld It was during the submersion that one  of the tri‐cells in the cellular structure  f th t i ll i th ll l t t failed.   This caused uncontrollable sinking   This caused uncontrollable sinking that led to an implosion of the  structure and complete collapse.

Plan form of the cellular structure

see detail

Tri-cell wall shape

550 800 Water pressure 58 00

Original design shape  with cylindrical walls i h li d i l ll

Actual shape of construction Actual shape of construction

The  natural arch action provided by the geometry  was not present in the modified form. The quadrilateral finite elements for the analysis were distorted in the region of the tri‐cell  corners from the ideal square shape.  This led to errors in the results.

Tri-cell joint detail • The critical shear section was  reinforced with T‐headed bars. f d h h d db • The design required that the  length of the  T‐headed bars to  g extend across the full width of  section. • As As they were difficult to fix  they were difficult to fix through the outer layer of  reinforcement it was decided to  reduce their length. reduce their length. • A cracked formed at a corner of  the cell and spread to the end of  the T bar the T bar.

compression failure

initial cracking

• Water pressure became active in  the crack. • A shear crack developed into the  compression zone and this failed  in a brittle manner. in a brittle manner.

'T' headed bar as required

'T' headed bar as fixed water p pressure

Summary This catastrophic failure was the result of a number of errors: •

The analysis program was set up with a finite element mesh that was too coarse to provide accurate shear results.

•

The ‘T’ headed bars were too short and allowed the shear resistance t become to b unsafe. f This Thi was probably b bl the th primary i cause off the th failure. f il

• There was minimal checking of the design and detailing. • In previous designs the geometry of tri-cells had been formed by intersecting cylinders. The geometry of tri-cells was altered on this project in order to make the formwork simpler to construct. construct Unfortunately the new form did not allow arching action to take place. • The rebuild retained the cylindrical geometry in the tri-cells tri cells and the reinforcement was detailed to ensure mechanical linkage. ‘T’ headed bars were extended to the outer reinforcement.

Case Study 9 Camden School for Girls Assembly hall roof collapse This disaster could also be called the miracle of the decade. On 13 June, 1973 late in the evening g the roof of the assembly y hall crashed to the ground. In the words of the caretaker, he heard a loud rumble, went to investigate by torch light light, and found the whole roof weighing many tons had collapsed. Twenty four hours before this event some five hundred parents had attended a meeting in the hall, chairs were still laid out.

Camden School for Girls Edge beam which had supported the precast beams

Part of the roof which had collapsed on to the chairs below

Collapsed roof lying on the floor below

Summary •

The principal cause of the collapse was inadequate bearing for beam seatings and deterioration of concrete att b beam ends. d Thi This was one off the th first fi t b buildings ildi ffound d to have suffered from the effects of high alumina cement.

•

This was an example of inadequate design and poor detailing g of the end bearing g nibs built into the supporting pp g beam for the precast beams. The reduction in strength caused by HAC left no margin for temperature effects.

•

The combination of these effects was likely to have triggered the collapse. collapse

Case Study 10 Ronan Point collapse Precast concrete panel building This collapse was a significant event for the industry in the UK and marked the partial demise of the precast industry. Large precast panel and frame construction became much less popular in the following two decades. Information g gathered from the incident led to major j changes to the UK Building Regulations (1970) and codes of practice (starting with the precast concrete code, CP 116, in 1970) with regard to progressive collapse and robustness. More recently the Eurocodes have included accidental load and robustness clauses clauses.

In the early hours of 16 May 1968 a g gas explosion p in a bathroom on an upper floor shook the building, resulting in the h iinstantaneous collapse ll off part of one wing of the building. Four people were killed.

The cause of collapse: •

The possibility of unusual, and hence non-codified, loads occurring was not considered in design. g

•

The structure was inadequately mechanically tied together.

Comment Traditional two-storey housing, pre World War 2, would not have had any engineering input; brick wall thicknesses and ti b floor timber fl joist j i t sizes i were prescribed ib d b by th the L London d Cit City Council Building By-laws, and similar regulations outside London. London There had been gas explosions before this incident in similar types of dwelling dwelling, but the damage damage, and any casualties, had usually been limited to one household, and the risk was accepted p as a ‘fact of life’. There were no precedents for progressive collapse, when ‘system y building’ g was introduced.

Poor Construction

Case Study 11 Pipers Row Car park collapse The e ca car pa park was as co constructed s uc ed us using g the e lift sslab ab method. e od This s involved o ed casting the slabs one on top of another on the ground. Precast columns were positioned and then the slabs were jacked up the columns until at the correct level level. The slabs were then held in place by the use of wedges wedges. This form of construction had been used in many places in the UK during the 1970s and 80s and has been a common form of construction in the USA. It has provided reasonably robust structures. The very nature of the construction method focuses attention on the column/slab j i t In joint. I some situations it ti th the structure t t has h relied li d on th the momentt resistance of these joints, i.e. an unbraced frame. In other situations separate insitu core structures have been built to take the sway forces.

In March 1997 a 120 tonne section of the roof of the car park collapsed onto the floor below . This occurred at 3am when, fortunately, y nobodyy was around. It was immediately clear from the debris that a punching shear failure had taken place.

Final connection between the slab and the column was made via a steel collar in the slab and a steel insert in the column into which wedges were fixed. The steel collar supported the slab on angles that formed an “H” in plan.

Summary • The 230mm thick slab was constructed with concrete of highly variable quality. •

Areas of low quality concrete deteriorated probably through freeze thaw action.

•

In some places this deterioration had occurred to a depth of 100mm and this had been repaired.

• The repair was poorly bonded to the parent material. • This left a slab which was effectively split into two layers with the only connection being the longitudinal steel passing through the repair into the original concrete. • Further deterioration of the original concrete, and in particular its bond strength to the top steel, reduced what composite action existed until failure occurred.

Case Study 12 Flat slab construction for a hotel For a short time in the early 1970s the Government provided loans for the construction of hotels. In order to be eligible the construction period had to be very tight. The workmanship of some of the hotels hotels, built at this time time, was shoddy shoddy. For one such hotel this was not discovered until twenty years later j refurbishment was taking g place p when a major

Hotel of the 1970s

This shows the structural layout of a typical floor, flat slab. The depth of the slab was 250mm. The spans along the building were 7.2m and across the building were 6.1m and 7.4m.

Problem The top surface of the slabs was very uneven and did not appear to have been levelled (by hand or power float). In some places boot marks had been left. Cracks (generally not larger than 0.3mm width) had occurred on the upper surface radiating from the corners of the columns with one or two small cracks running g tangentially. g y Large g cracks ((up p to 1mm width)) had occurred at some of the construction joints. The deflection of one of the slab bays of an upper floor was large, over 75mm. 75 An independent adviser decided that: • the punching shear was close to its limit limit. Additional steel shear heads were constructed and fitted to all column slab intersections. • the bending strength of the longer spanning bays was at its limit. After a year of measuring the deflection of one bay of the slab it was found that no movement had occurred. The reason for such large deflections was not understood understood.

A second independent check revealed: In one of the bays the skirting board between two edge columns had been made in two equal lengths split in the middle (as shown) shown). Deflections of between 15 and 20mm had occurred below each half of the skirting board. This represented an edge deflection of up to 50mm.

Since the skirting board was attached to the wall it was likely that it was fitted this way and that much of the deflection had taken place before construction of the wall. This was confirmed by finding that the bottom courses of the external wall had been laid on the sagging shape of the slab and the following courses adjusted so that they were level at the window sills above the floor.

Summary The second independent check concluded: • Punching shear capacity: Both BS 8110 and BS EN 1992-1-1) provide reliable methods for predicting punching shear capacity using characteristic values for the concrete strength instead of the factored design values and the ‘as built’ information concerning the reinforcement ((i.e. the size,, spacing p g and cover to the bars). ) An assessment of the safety can be made by comparing the ‘worst credible’ loads with the resistance. For thi F this situation it ti the th calculations l l ti showed h d th thatt th the worstt credible dibl lloads d could be carried with a sufficient safety factor. • Bending capacity: The top cover to the reinforcement near the column supports was found to be on average 30mm more than specified. Once reasonable moment redistribution had been included in the calculations there was still sufficient overall moment capacity in the slab without requiring any reduction to the design safety factors.

Conclusion Although the construction workmanship had been very poor the structure was not in danger of collapsing. A great deal of money had been spent unnecessarily.

Case Study 13 Precast concrete tank A liquid storage tank was constructed with precast wall panels. The diameter and height of the tank were 12.2m and 7m. The tank collapsed suddenly within two years of construction.

Liquid storage tank

Anchor unit

.2m 2 1

The vertical panels were held in place by unbonded prestressed tendons threaded through horizontal PVC ducts, embedded in the concrete and fully encircling the tank at set levels throughout the height.

Section through precast panel Interface with adjacent unit

23mm PVC duct

• In order to achieve watertight construction of the edges of the wall units required to be built with very small tolerances. A rubber strip was inserted within the joint between each set off adjacent panels. The water tests showed leaks. Several attempts were made to seal these before watertightness was achieved. • Plastic sheathing and grease around the tendons was intended to provide protection from corrosion corrosion.

Detail at anchorage of tendons Anchorage cast into concrete 7-wire greased tendon

Screw in cap filled with grease

PVC d ductt castt into concrete

Sheath Sheathover overtendon tendon cut cutback backfrom fromend end

• The grease used in this particular type of unbonded tendon (12.5mm diameter Tyesa y 7-wire strand)) was found to emulsify y when in contact with water. This allowed any water that had penetrated the anchor zone not only to come into contact with the bare part of the tendons but also to penetrate into the sheathing.

Summary The alloy steel of the particular prestressing tendons used d iin thi this structure t t h had d a microstructure i t t susceptible to stress-corrosion cracking, and the stress in the tendons was greater than 50% of the yield strength. Moisture in contact with the tendons provided a corrosive environment. On examination after the collapse collapse, it was found that stress-corrosion cracking had taken place in many parts of the unbonded tendons.

Poor Management

Case Study 14 Placing of precast units Floor collapse

precast slab jacked into position

precast slabs

spine i b beam

PLACING OF PRECAST UNITS The spine beam beam, carrying precast planks, lost its bearing because a labourer, in trying to wall supporting pp g jac jack o one eo of the e final a p planks a s into o spine beam position, actually levered out the wall panel supporting the end of the spine p beam. This caused the spine beam to lose its bearing which led to the collapse of the floor floor.

wall shifted outwards causing spine i b beam tto fall f ll off ff its it bearing b i

precastt slab l b jjacked k d into position

Plan The lacer bars had not been inserted at the time of erecting and laying of the floor elements.

lacer bars not in place l att ti time off jjacking ki

Section

Summary This is an example where the management should have had more control on how the erection and placing of precast units took place. More importantly, it should have ensured that the lacer bars at the ends of the spine beam were in place before the erection of floor units took place.

Poor Planning

Case Study 15 Power station on the river Thames The power station was constructed on the north bank of the Thames in the early 1960s 1960s. Originally it was to be coal fired to produce 1500MW. The foundations Th f d ti off th the power station t ti sit it on 20 20,000 000 reinforced i f d concrete piles.

Special on site casting yard

Each pile was 430mm square, 18m long.

Piling rig Several pile rigs were set up with diesel driven hammers. A pile il was h hoisted i t d iinto t position iti and d then given a tap by the hammer to get the point of the pile through the top crust of the marshland. Then under its own weight the pile dropped 15m through the mud! Each pile was then driven into the gravel to a specified g p set.

Procedure Piling g commenced from the edge g of the site closest to the river and continued inland 250m placing piles at 1.5m centres (on average) average). This took about eighteen months. Excavation for the foundations started from the same end and commenced six months after the start of piling. Concreting of the foundations also started from the same end of the site. After a year after the start of piling piling, when concreting of the foundations had reached about a 1/3 of the way along l th the site, it it was discovered di d that th t the tops of the piles that were still exposed were moving. Measurements showed that this movement was up to 1.5m !

Remedial work a) Additional 600 vertical piles to compensate the reduction in vertical capacity of the existing piles b) Additional 200 raked piles to compensate the horizontal force component caused by the bent piles. The resulting remedial and extra work caused by this movement was very large. For example the existing piles no longer followed the plan layout for the eight inlet and outlet culverts that wound their way through the site bringing cooling water to the condensers and returning it to the River Thames. On site decisions decisions, making changes to the design design, had to be taken each day

Summary The programme for the contracts on this project did not foresee the problems caused by progressing the work from one end of the site to the other. In previous similar projects there had been a significant delay between piling and the start of excavation which allowed enough time for much of the soil pressure to dissipate. In order to keep a tight programme one possible solution might have been to start the piling from both ends of the site.

Deliberate Malpractice

Case Study 16 Floor with excessive deflection The building in question was a telephone exchange and had been built in the mid 1970s, ten years earlier than the i investigation. ti ti Th The slab l b off a ttypical i l end db bay h had db been d designed i d as single way spanning between two shallow haunched beams. The span was 9m with a slab only 250mm thick which many engineers would consider to be too thin. Ten years after the building had been completed the operators were complaining that the deflection was still increasing and causing some of the switch-gear to become faulty. The designers asked for a second opinion on the design of the slab. The calculations and drawings were checked and no major flaw was found. It was just conceivable that creep and shrinkage effects were still increasing increasing. A site visit was arranged arranged.

Typical end bay layout 9m

A 600

250 thick slab 300

Excessive deflection (still increasing after 10 years)

A A-A

Summary (1) The visit to site included the inspection of the slab close to a column. The screed had been removed to expose the top surface of the structural t t l slab. l b A As a crude d check h k off th the h hardness d off the th concrete t surface, a penknife was used. Quite unexpectedly the blade of the knife penetrated into the concrete surface right up to the hilt ! A further check of the soffit of the slab gave a similar result. An additional interesting feature of the soffit was the presence of a number of shallow disc shaped (‘flying ( flying saucers’) saucers ) pieces of concrete (150mm diameter) which were separating from the surface. One such piece came away as it was being examined. Although the slab had been designed to span one-way, the supporting beam was sufficiently flexible for the slab to behave more like a flat slab The ‘flying slab. flying saucers’ saucers had occurred in the compression areas of the soffit and were considered to be the effect of spalling.

Summary (2) It was clear that the slab in question required immediate additional support and the rest of the building required core testing. After cores had been taken throughout the building it was discovered that the concrete cube strength, which should have been 25MPa, was on average only 5MPa 5MPa. The sub-contractor had deliberately y reduced the cement content in the specified mix. Major remedial work followed.

C Case St Study d 17 Insitu columns supporting a precast building This building was constructed with precast elements above ground. Below ground the foundations, columns and beams were constructed insitu insitu. Construction had reached an advanced stage when cracks appeared d iin th the iinsitu it columns l jjustt b below l th the connection ti with ith th the precast columns.

Layout of elements

Precast beams, columns and slabs

See detail of column connection Street level Existing retaining wall

Transfer beams Insitu beams and columns

Intended construction procedure Column cast with large polystyrene box-out

Column reinforced as normal

Polystyrene totally removed; CHS 114 dia dowel cast in with fresh concrete filling box-out box out

As constructed CHS dowel pushed into polystyrene

Only thin layer of concrete cast in top of column

Only top layer of polystyrene removed Existing insitu column

First sign of imminent failure Load from 7 floors above Precast column

Grouting tube

Severe cracking of i it column insitu l wallll

Load from precast unit supported on thin outer shell of insitu column

Insitu column

Remedial work • In order to repair the top of the insitu columns the load from the precast building had to be removed removed. • This was achieved byy providing p g props p p and jjacks close to the existing precast columns at each floor level and creating a new load path to the ground. • This released the load in the insitu columns below, and allowed the required remedial work to take place - reconstruction of the top of the insitu columns.

Thank you for your attention