You are on page 1of 4

Paper - Alexander:Layout 1

29/11/06

11:56

Page 40

conference paper: alexander

Why does our concrete still crack and leak?


Synopsis
Current knowledge of the behaviour of reinforced concrete
appears to be not good enough to enable us to design
basements which do not crack and leak. The paper discusses
the possible reasons for this, and makes proposals for better
designs in future.
This starts with the life cycle of concrete and the
importance of restraint. Formulae for minimum
reinforcement contents to control cracking in both immature
and mature concrete are then derived. This is illustrated by a
flow chart. The misleading advice in codes of practice and
industry guides is discussed, and the true tensile strength of
concrete is analysed.
Although the outcome is a recommendation for higher
reinforcement contents than are currently employed, it is
based on a sound theoretical approach backed up by the
evidence of basements which still crack and leak.
Furthermore, if cracking is not controlled, all the crackcontrol reinforcement that is in the concrete has been
wasted. A reversal of traditional design philosophy is
proposed, with the concrete section first selected at
minimum thickness and then adequate reinforcement
provided.

Introduction
The question is posed with particular reference to the walls and
ground slabs of basements. Fig 1 shows a typical basement
used for car parking, while Fig 2 is a close-up of the cracks in
the foreground.
It is surprising that our knowledge of the behaviour of reinforced concrete is still not good enough to enable us to design
basements which do not crack and leak. In what follows,
reasons for this situation are discussed and proposals are made
for better designs in future.

Life cycle of concrete


Before considering these questions, it is important to appreciate the behaviour of concrete with time. Concrete typically goes
through the following life cycle:
Early-age contractions In the first 36 days, the hydration
process first causes the concrete to heat up from a casting
temperature that is usually already a few degrees above
ambient. During this period, it behaves in a plastic manner so
the only result is an increase in volume. However, when it cools
to ambient temperature it has hardened enough to go into
tension and may crack if it is restrained. At the same time,
early-age shrinkage (a complex combination of actions, see
Altoubat and Lange1) also occurs, adding to the contractions.
Strength gain In the following 34 weeks (longer in mixes
with pfa or ggbs), both compressive and tensile strength
increase.
Temperature In the first year, it will experience the seasonal
temperature and humidity cycle. Transitory (daily and weekly)
variations will also occur. Because the tensile strength of
concrete under sustained stress is only 70% of that under transitory (short-term) stress, seasonal temperature changes should
be treated as sustained while daily and weekly variations are
transitory.
Shrinkage Over a longer period, perhaps 210 years, longterm drying shrinkage will gradually occur.

Is restraint not recognised?


Clearly if concrete can contract freely no tensile stress builds
up and cracking does not occur. However, restraint is frequently
prevalent, and two particular situations need considering.
Early-age contractions are restrained where the concrete
abuts a previous pour.
Basement structures are frequently restrained by piles, pile
caps, footings, returns, even friction with the soil. Fig 3 shows
40|The Structural Engineer 5 December 2006

the section through the car park slab shown in Fig 1, from
which it can be seen that a high level of restraint is present.
Some texts recommend providing joints, generally full contraction joints with sliding dowels. However, these only eliminate
the restraint for a short distance each side, and for the structure in Fig 3 joints would be needed in each panel between lines
of columns and foundations. This would be difficult structurally
as the floor is designed as a two-way spanning flat slab.
Furthermore joints are difficult and expensive to construct and
almost impossible to repair if they fail. If it is assumed that full
restraint is present throughout, crack control can be provided
by reinforcement and all joints can be omitted. Construction
joints are formed without breaking the reinforcement and then
treated as part of a continuous structure.

Creep helps
Some alleviation of sustained tensile stresses is provided by
creep. Fig 4 shows the reduction over time of both early-age
contractions and shrinkage. Early-age contraction stresses are
reduced surprisingly rapidly. The figure is based on quite
conservative assumptions (loading at 3.5 days, section 300mm
thick exposed on one side only, relative humidity (rh) 80%), yet
the reduction is estimated to be 30% at 4 months and 60% at 2
years. Shrinkage is different. Although each increment of
shrinkage is lessened by creep, the shrinkage builds up quite
slowly so the effect of creep is even more delayed. The figure
shows that the effect is hardly significant up to 1 year, although
ultimately the reduction is about one-third.

Cumulative effect of restrained contractions


The cumulative effect of the different contractions is illustrated
in Fig 5. This is drawn for a typical basement ground slab,
although the tensile stresses at any point can only be guesses.
Line 1 is the tensile strength of the concrete, showing both the
strength under transitory stress (1a) and the reduced strength
(factor 0.7) under sustained stress (1b). Line 2 leading to 2a is
the early-age contraction stress, showing the relief from creep.
Line 2b shows the addition of long-term drying shrinkage. Not

Stuart J.
Alexander
MA, CEng, FIStructE,
FICE, MCMI
WSP Group
Keywords: Reinforced
concrete, Basements,
Cracking, Codes of
practice, Case studies,
Car parks, Design
Stuart J. Alexander
This is a compilation of
two papers: one with
the same title published
at the congress Global
construction: ultimate
concrete opportunities,
Dundee, 5-7 July 2005,
the second The
importance of time in
assessing cracks in
concrete, presented to
the 3rd International
Conference on Forensic
Engineering, London,
10-11 November 2005

Fig 1. (below left)


General view of a
typical basement car
park. Note cracks in
floor in foreground
Fig 2. (below)
Close-up of the car
park cracks

Fig 3. Section through floor construction of car park basement in Fig 1

Paper - Alexander:Layout 1

29/11/06

11:56

Page 41

conference paper: alexander

Are we misled by BS 8110?


In BS 81102, the only references to minimum reinforcement are
0.13bh% in bending and 0.45Ac% in direct tension.
This second limit is generally thought to apply to structural
tension members such as hangers and not to members where
the tensile stress arises from contraction. Neither of these is
related to the tensile strength of the concrete, whereas equation
1 shows that it should be directly proportional to it. Meanwhile,
concrete is getting stronger with grade 32/40 frequently specified and higher grades often used.
A comparison can also be made with the design of continuous reinforced concrete pavements, for which the Highways
Agency3 specifies 0.6% reinforcement in a single layer at the
mid-depth.

Is the true tensile strength of concrete


misunderstood?
The tensile strength of immature concrete, i.e. during the earlyage contractions phase, can be given very approximately by
reference 4.
fct,imm = 0.12 (fcu + 10)0.7

(2)

where fcu is the specified 28-day cube strength; the addition


of 10MPa allows for the average strength. This tensile strength
is usually taken as 1.6MPa for grade 28/35 concrete.
For mature concrete this value increases5 to
fctm = 0.30 (fcy)0.67
shown are stresses from seasonal and transitory temperature
variations, but even without them there are risks of cracking
at 3-6 days and then as shrinkage builds up over the years. It
is also clear that only quite small additional stresses from
temperature drops can be enough to trigger cracking, with the
risk increasing as time goes by.

Is control of cracking by reinforcement not


understood?
The principles of controlling cracking by reinforcement are
illustrated in Fig 6. If concrete tries to contract but is restrained
so that movement cannot occur, tensile stress gradually builds
up. When this reaches the tensile strength of the concrete
section, a crack occurs. The movement at the crack relieves
some of the stress but as the contraction continues the stress
builds up again. If the reinforcement is weaker than the force
which caused the first crack, it will yield and all future contraction will be concentrated at the same point so that the crack
simply gets wider (Fig 6a). If the reinforcement is stronger than
the cracking force, it will remain elastic and as the contraction
increases a new crack will form at the next weakest crosssection. This process continues with enough cracks forming to
absorb the total contraction (Fig 6b). The crack in the first
model is uncontrolled while the cracks in the second are
controlled.
The role of reinforcement in controlling cracking can be
derived from first principles by making the strength of the reinforcement greater than the strength of the concrete in tension.
Applying this principle, the reinforced axial capacity is:
Fs = As fs
and the uncracked capacity is:
Fc = Ac fct [1+ (m1)]

Fig 4. (top)
Reduction over time
of restrained tensile
stress due to creep.
Curve A early-age
contractions (from
100%), curve B
shrinkage (from free
shrinkage C above)
Fig 5. (above)
Tensile strength and
stress over time.
Curve 1 strength, 1a
for transitory stress,
1b for sustained
stress. Curve 2 stress,
2a early-age with
creep relief, 2b earlyage plus shrinkage.
Temperature not
shown

(3)

where fcy is the specified 28-day cylinder strength; the


formula corrects for the higher average strength and for scatter
about the average.
The lower and upper characteristic tensile strengths are
given in EN 1992-1-15 as 0.7fctm and 1.3fctm respectively. Clearly
it is the highest tensile strength that must be considered, so
average values should be multiplied by 1.3. It is also important
to derive the tensile strength from a realistic estimate of the
compressive strength. It is often assumed (as in this paper) that
the mean strength is 8/10MPa (cylinder/cube) above the specified minimum, but a higher strength may be delivered for other
reasons, e.g. early striking of formwork. Some recognition may
also need to be given to the natural increase in strength beyond
28 days, particularly for concretes containing ggbs or pfa.

Are the statistics not understood?


Controlling cracking should not be thought of as being in the
same category as most other serviceability criteria. It is fundamental to the satisfactory behaviour of structures like basements, and therefore needs to be relatively improbable. The
criterion suggested here is a 95% limit, i.e. only 1 in 20 failures.
However, if the maximum concrete tensile strength is
combined with the minimum reinforcement yield strength the
resulting probability is 1 in 20 20, i.e. 1 in 400. This is not what
is intended, and the 1 in 20 overall probability is achieved by
combining the maximum concrete tensile strength with the
mean reinforcement strength. Since January 2006, the characteristic reinforcement strength has been 500MPa with the
mean expected to be approximately 550MPa, so if we wish to

Fig 6.
Control of cracking
by reinforcement

The second term in the brackets is quite small and is


neglected at this point, although a correction for it is applied
later. Putting Fs Fc gives:

1.0 (fct / fs)

(1)

In this, is the reinforcement content As/Ac, fct is the tensile


strength of the concrete and fs is the yield strength of the reinforcement. Appropriate values of these parameters are considered below.
5 December 2006 The Structural Engineer|41

Paper - Alexander:Layout 1

29/11/06

11:56

Page 42

conference paper: alexander

retain the characteristic reinforcement strength in the formulae, a correction factor 500/550 needs to be introduced. The
other correction introduced is 1.06 for the enhancement of
tensile strength Fc given by 0.6% reinforcement.

Table 1. Minimum reinforcement contents in direct


tension for immature and mature concrete related to
strength grade
Specified strength fck / fck,cube
Expected mature average
strength fcm / fcm,cube
Immature tensile strength fct,imm
crit = 1.25 (fct,imm / fyk)
Mean mature tensile strength fctm
crit = 1.00 (fctm / fyk)

Immature concrete
This leads to the minimum reinforcement content for immature
concrete of

crit = 1.25 (fct,imm / fyk)

(4)

Values of crit for various specified concrete strengths are


shown in Table 1.

20/25

28/35

35/45

45/55

28/35
1.45
0.36%
2.2
0.44%

36/45
1.72
0.43%
2.8
0.56%

43/55
1.98
0.50%
3.2
0.64%

53/65
2.23
0.56%
3.8
0.76%

early-age contractions become uncontrolled for all further


contraction movements. However, if is greater than crit, new
cracks form in addition to the original ones and both old and
new cracks are controlled.
It is also clear that if the amount of reinforcement is insufficient to control cracking, whatever is provided for crack control
is wasted as it achieves nothing a classic case of false economy.

Mature concrete
For mature concrete a further correction is made. The dominant
cause of full-depth cracking is sustained contraction longterm shrinkage and seasonal temperature drops added to earlyage contractions although more rapid temperature drops
frequently contribute. Allocating these 2/31/3 suggests that a
reduction in fctm for long-term effects of 0.8 can be applied.
Taking this into account, the minimum reinforcement
content for mature concrete is

Should we specify a maximum concrete strength?

Values of crit for various specified concrete strengths are


also shown in Table 1.

The data above are based on the assumption that the average
compressive strength will be 8/10MPa above the specified
minimum plus an allowance for scatter. This suggests that an
upper limit perhaps fck + 16/20 should be written into specifications, as it is clear from Table 1 that the minimum reinforcement content will be underestimated if the compressive
strength is too high.

Flow chart

Do we get false security from BS 8007?

The previous analysis for elements subject to direct tension


has been consolidated into a flow chart, Fig 7. Although this is
for fully restrained structures, it is always on the safe side to
make this assumption.
Working down from the top, stage 1 deals with early-age
contractions, stage 2 covers the relatively rapid increase of
compressive and tensile strength, and stage 3 explains the
subsequent behaviour of the mature concrete. The immature
strength of the concrete is denoted fct1, the mature strength fct2.
The stresses at the corresponding stages are ct1 and ct2.
The key area not properly dealt with in any code of practice
or industry-standard publication is where is greater than crit
but less than crit. Then the cracks which were controlled for

BS 80076 considers design to resist early-age contractions. For


grade 28/35 concrete, it derives the minimum reinforcement
content as 1.6/460 = 0.35%. The analysis explained here results
in the percentages in the fourth row of Table 1 which are around
25% higher than in BS 8007, principally through allowing for
the tensile strength being significantly higher than the mean.
Importantly, BS 8007 is written for reservoirs, i.e. tanks full
of water such that there will be little subsequent shrinkage and
small temperature variations. It is assumed implicitly that
subsequent contractions will be less than the relief of early-age
contractions by creep, and the important behaviour of mature
concrete in basements and similar structures is ignored.

crit = 1.00 (fctm / fyk)

(5)

Fig 7.
Flow chart showing
possible outcomes as
concrete matures

Poor guidance for basements?


In the UK, British Standards 8007, 81027 and 8110 all
contribute relevant material, but even so basements are not
properly covered. In particular, the importance of considering
contraction in mature concrete is ignored. Industry guides such
as CIRIA reports 918 and 1399 similarly miss this important
point. The minimum percentages of reinforcement in mature
concrete are shown in the sixth row of Table 1, and can be seen
to be much higher than current general practice.
Note that EN 1992-1-1 includes a provision for minimum
reinforcement (if crack control is required), with a formula
which for rectangular sections in direct tension reduces to the
same expression as equation 5. The analysis above shows that
this is intended to provide the upper characteristic reinforcement content to control cracking in mature concrete.

Case study basement floor slab


Figs 1 and 2 show the floor slab of a single level basement 215
70m used for car parking which has cracked extensively. The
slab is 350mm thick, designed as a flat slab on pile caps at twoway centres varying between 5.5 and 6.5m. The section (Fig 3)
shows that full restraint would be a realistic assumption. Total
reinforcement content is 0.38% in the middle strips, 0.60% in
the column strips, average 0.49%. Concrete grade was specified
as 28/35, although no actual cube test results are available.
The structure was built in the mid-1990s, although the cracks
were only observed around 2000. They are roughly parallel, at
right angles to the long dimension. They start about 27m from
each end and are in groups of two to five spaced at 500 to
1100mm, roughly centred in each bay. They are now around
0.5mm wide, although filled with a dark grey precipitate,
42|The Structural Engineer 5 December 2006

Paper - Alexander:Layout 1

29/11/06

11:56

Page 43

conference paper: alexander

Fig 8.
Reduction factor
(upper) and
illustration of surface
and internal zones
(lower) against
increasing overall
depth of section

having leaked extensively over the three winters from 2000-1


to 2002-3 and been below the water table for at least part of that
time.
The author believes the explanation is that at the immature
stage was greater than crit (see Table 1), and that controlled
cracks initially formed as predicted for early-age contractions;
the presence of groups of parallel cracks is evidence for this.
However, after 5 years and with a plentiful supply of water the
concrete strength increased, became less than crit and with
cold temperatures and shrinkage the cracks widened uncontrollably rather than additional new cracks forming.
The absence of cracks at each end probably shows the
distance needed for full restraint to develop, perhaps aided by
the inward pressure of the soil on the perimeter walls. The
absence of cracks parallel to the long sides (and of any movement at the construction joints one can be seen in Fig 2) could
be for the same reason: the width is not much more than twice
27m. However, the slab was constructed in longitudinal strips
generally 5-6m wide containing only one line of columns, so an
alternative explanation is that the early-age contraction cycle
passed without restraint inducing any significant tensile stress
and that subsequent contractions have been within the remaining tensile strain capacity.

Cant understand crack width calculations?


In British codes of practice crack widths derived by two opposing theories are presented. The method in BS 8110 and
Appendix C of BS 8007 assumes that there is no slip at the
interface of the concrete and reinforcement, but that there is
shear lag so that cracks taper towards the surface. The method
in Appendix B of BS 8007 assumes the opposite, i.e. that there
is slip at the interface between the steel and the concrete but
that plane sections remain plane and that the cracks are of
uniform width. The method in Eurocode 2 combines these two
theories, resulting in a more meaningful estimate of crack width
at the surface. However, this is intended principally for sections
in bending, and the application to sections in direct tension is
open to doubt.
More importantly, these methods all assume that the reinforcement remains elastic, in other words that the cracks are
controlled.

Too much reinforcement in thick sections?


BS 8007 introduces the concept of reinforcing only surface
zones, generally 250mm thick. EN 1992-1-1 makes an
allowance for non-uniform self-equilibrating stresses in thick
members. Although expressed in different ways, the net effects
are similar. They are illustrated in Fig 8, both as a reduction
coefficient and as a concentration into surface zones. BS 8007
goes further by reducing the surface zone in the bottom of
ground slabs to 100mm.

Confused by proprietary waterproof concrete?


A number of products Caltite, Pudlo, Sika, Xypex, for example
are marketed as admixtures to make concrete watertight. The

mechanism is typically that the key ingredient blocks the microcracks and pores so making the concrete impermeable.
However, this does not stop leaks through cracks (although
Xypex forms crystals claimed to block cracks up to 0.4mm wide)
so the need to control cracks is not diminished.

Why not use tanking?


In theory, an external membrane can be provided to keep the
basement watertight and the concrete then be designed solely
for resistance to loads. However, such membranes are expensive
to apply, demand the highest quality of materials and workmanship, and can delay the construction process. Many basements are constructed within embedded pile walls, which are
particularly difficult to protect in this way.
More importantly, the principle is flawed, as explained in
Design and construction of deep basements10. This points out
that membranes prevent autogenous healing of early-age
cracks and encourage drying shrinkage cracks. For these
reasons, it is generally better to avoid tanking and rely on a
properly reinforced concrete structure.

This amount of reinforcement is too expensive?


When it was first published 35 years ago, the method in BS
8007 took the reinforcement from around 0.2% to 0.35%, a
significant increase. The approach outlined here increases it to
over 0.6%, a further very substantial increase. However it is
based on a sound theoretical approach, and is backed up by the
evidence of the large numbers of newly constructed basements
which still crack and leak. Furthermore, if cracking is not
controlled, all the reinforcement that is in the concrete has
been wasted.

A new philosophy?
In the past, the approach has been to choose the concrete thickness first, usually on the basis that the thicker the better.
Reinforcement has then been provided, generally as little as
permitted by codes of practice. In future the approach should
be the opposite. The concrete thickness should be the minimum
needed to provide the required structural capacity; the
minimum of 250mm from BS 8102 will often be enough. Then
sufficient reinforcement to control cracking should be provided,
drawn from the fourth row of Table 1 if the only restraint is to
early-age contractions or from the sixth row if the mature
concrete will be restrained.

REFERENCES
1.

Altoubat, S. A. & Lange, D. A.: Creep, shrinkage and cracking of restrained concrete at early age, ACI Materials J.,
July-August 2001, pp 323-331. (Summarised by Alexander,
S. J. with the same title in Concrete, March 2006)
2. BS 8110-1: Structural use of concrete, Part 1 Code of practice
for design and construction, 1997, British Standards
Institution
3. Design manual for roads and bridges, V7, section 2, part 3,
notes on figure 2.5, Highways Agency
4. Hughes, B. P.: Limit state theory for reinforced concrete design,
3rd edn, Pitman, London, 1980, pp 289, 377 & 589
5. EN 1992-1-1: Eurocode 2 Design of concrete structures Part
1: General rules and rules for buildings, 2002, British
Standards Institution
6. BS 8007: Design of concrete structures for retaining aqueous
liquids, 1987, British Standards Institution
7. BS 8102: Code of practice for protection of structures against
water from the ground, 1990, British Standards Institution
8. Early-age thermal crack control in concrete (revised edition),
Report 91, 1992, Construction Industry Research and
Information Association
9. Water-resisting basement construction: a guide, Report 139,
1995, Construction Industry Research and Information
Association
10. Design and construction of deep basements including cut-andcover structures, 2004, Institution of Structural Engineers
5 December 2006 The Structural Engineer|43

You might also like