Appendix A4 Autogenous shrinkage

A4 Estimating autogenous shrinkage

A4.1 Introduction

While considerable research has been carried out to measure autogenous shrinkage and many of the
influencing factors have been identified (most notably the water/binder ratio [w/b] and the binder
composition) it is still recognised that that the magnitude of autogenous shrinkage is difficult to predict.
Nevertheless, EN1992-1-1 offers a method for estimating autogenous shrinkage based solely on the
strength class of the concrete. This method is described and compared with published data. The influence
of different cementing materials has also been investigated.

A4.2 The method of EN1992-1-1

The EN1992-1-1 method for estimating autogenous shrinkage is based solely on the strength class of the
concrete according to the equation,

εca(t) = as(t) . εca(∞) (A4.1)

where:
εca(t,) is the autogenous shrinkage at time t days
εca(∞) is the ultimate autogenous shrinkage = 2.5(fck – 10) x 10-6 (A4.2)

where
fck is the characteristic cylinder strength
ds(t) is a function defining the time dependent development of autogenous shrinkage
ds(t) = 1 – exp(-0.2.t0.5) (A4.3)

Using these equations a series of autogenous shrinkage curve have been developed for different strength
classes as shown in Figure A4.1.

While it has previously been assumed that autogenous shrinkage will only occur in concretes with very
low w/b ratios, typically below about 0.40 (Pigeon et al, 2005) EN1992-1-1 predicts that autogenous
shrinkage will occur, to some degree, in all structural concretes. An investigation of published data has
been carried out to support this supposition.

The results from tests on cement pastes over a range of w/b ratios are shown in Figure A4.2. It can be
seen that in the short term the magnitude of autogenous shrinkage may be low when the w/b is of the
order of 0.45 and at higher w/b ratios swelling may occur (Nawa and Horita, 2004). However, in the long-
term, autogenous shrinkage may be significant (Baroghel-Bouny and Kheirbek, 2000) even when the w/b
ratio is as high as 0.6. The magnitude of the values for concrete will, of course, be much lower than the
values shown in Figure A4.2, as shown by the comparison with values estimated in accordance with
EN1992-1-1.

Unfortunately, because of the assumption that autogenous shrinkage is only significant in high-
performance, low w/b concrete, published data for concretes with w/b ratios above about 0.4 are both
limited and variable. Saje et al (2001) reported autogenous shrinkage at 48 hours of 180 and 0 microstrain
for concretes with w/b ratios of 0.40 and 0.52 respectively. These concretes had mean cube strengths of
67.4 and 48.4 MPa (representing class C50/60 and C33/40) with EN1992-1-1 predicting 48-hour values
of autogenous shrinkage of 25 and 14 microstrain. For the higher strength concrete the measured value
was significantly higher than predicted while for the lower strength class, EN1992-1-1 predicted higher
shrinkage than reported.

CIRIA C660 1
Appendix A4 Autogenous shrinkage

Figure A4.1 Autogenous shrinkage for different strength classes estimated using the method
of EN1992-1-1

1600
Autogenous shrinkage (x10-6)

1400 Baroghel-Bouny & Kheirbek (2000)

1200 1-year values for 3 cements

1000
800
600
400
200 Nawa & Horita (2004)
7-day values
0
-200
-400
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
w/c ratio

Figure A4.2 Reported values of autogenous shrinkage in cement pastes

Other comparisons with EN1992-1-1 can be made for concretes at lower w/b ratios. Pigeon et al (2005)
tested 5 concretes, all with w/b =0.35 but with different binders and additives. Autogenous shrinkage was
measured at 5 days and the results are shown in Figure A4.3 compared with the mean cylinder strength.
Again, all measured values were in excess of values predicted by EN1992-1-1.

CIRIA C660 2
Appendix A4 Autogenous shrinkage

Figure A4.3 Measured (at 5 days) and predicted values of autogenous shrinkage for concretes using
different binders and additives (Pigeon et al, 2005)

Scheißl et al (2000) tested high performance concretes, mostly with silica fume and again reported values
of autogenous shrinkage that are appreciably higher than predicted by EN1992-1-1. After 21 days values
typically in the range 300-400 microstrain were obtained while EN1992-1-1 predicts values which are
only about 50 per cent of those reported.

In a review of work carried out by the JCI committee on autogenous shrinkage, Tazawa et al (2000) offers
equations for ultimate autogenous shrinkage of the form:

for 0.2 ≤ w/b ≤0.5

εca0  3070.exp{7.2(w/b)} (A4.4)

for w/b > 0.5 εca0  80 (A4.5)

At a w/b ratio of 0.35, this predicts an ultimate autogenous shrinkage of about 250 microstrain compared
with values of about half this magnitude predicted by EN1992-1-1 for concretes with a strength class
associated with this w/b ratio.

The evidence suggests, therefore, that the approach of EN1992-1-1 will underestimate autogenous
shrinkage for concretes of a high strength class, but may overestimate the magnitude for concretes of
lower strength classes. The apparent underestimate for normal structural grades of concrete may be due to
the fact that EN1992-1-1 only requires that autogenous shrinkage be considered when new concrete is
cast against hardened concrete and that the values presented take into account a 50 per cent reduction for
creep. This would seem to be a reasonable explanation and is discussed in more detail in Section A4.3.

A4.3 The critical w/b ratio

CIRIA C660 3
Appendix A4 Autogenous shrinkage

In order to determine whether a critical w/b exists, below which it may be assumed that autogenous
shrinkage will not occur, and to derive a method of prediction which is more consistent with
measurements, a detailed analysis has been carried using one of the most comprehensive studies
undertaken on cement pastes with w/b ratios varying from 0.25 to 0.60 (Baroghel-Bouny and Kheirbeck,
2000). Recognising that shrinkage in concrete will be significantly less that in cement paste, the
composite model developed by Hobbs (1974) has been applied to estimate values for concrete. This takes
account of the relative volume of cement paste and aggregate and the relative stiffness of the two
components. Calculations have been carried assuming an aggregate with an elastic modulus of 55 GPa.
The elastic modulus of the cement paste is based on a relationship with w/b ratio derived from the data of
Nawa and Horita (2004). The Hobbs model predicts that the shrinkage in concrete will be between 20 and
30 per cent of that occurring in cement paste, with the higher percentage being achieved in the lower w/b
mixes.

The results from this analysis are shown in Figure A4.4, presented as the time-dependent strain at
different w/b ratios. Data were reported for a period of 12 months and the results have been shown for the
full period and on an expanded scale for the first 28 days. The results indicate that short term autogenous
shrinkage (ie within the first 7 days) is unlikely in concretes with values of w/b in excess of 0.45. Indeed,
during this early period some swelling may be expected. However, even for these higher w/b concretes,
autogenous shrinkage may need to be taken into account when assessing longer term strains.

The results shown in Figure A4.4 indicate that EN1992-1-1 may underestimate the magnitude of
autogenous shrinkage for the lower w/b mixes, particularly within the period when it will have the
greatest impact on early thermal cracking. Consider a concrete with a w/b ratio of 0.45. Using CEM I this
might be expected to achieve a class C40/50 concrete with a mean cube strength of about 60 MPa. For
this strength class, EN1992-1-1 predicts 30 microstrain autogenous shrinkage, while measurements
suggest a 7-day value of 70 microstrain. The difference in the longer term strain is greater. At 1 year,
while EN1992-1-1 predicts 75 microstrain, the value based on observations is closer to 200 microstrain.

Figure A4.4 Estimated values for autogenous shrinkage of concretes using CEM I with w/b ratios
from 0.3 to 0.6 (left – long-term; right – up to 28 days)

While for low w/b ratio concrete EN1992-1-1 would appear to underestimate the shrinkage, at higher w/b
ratios EN1992-1-1 predicts higher values than indicated by measured results. At a w/b of 0.6, the results
in Figure A4.4 suggest that autogenous shrinkage will not be apparent within the first three months, while
EN1992-1-1 predicts that for a C25/30 concrete, the shrinkage within this period will be 20 microstrain.

CIRIA C660 4
Appendix A4 Autogenous shrinkage

Hence, while the method of EN1992-1-1 for predicting autogenous shrinkage is simple, it is likely that if
these estimates are used in design, particularly for low w/b concrete, the magnitude of strain will be
underestimated.

CIRIA C660 5
Appendix A4 Autogenous shrinkage

A4.3 Dealing with autogenous shrinkage in design

Many of the observations of autogenous shrinkage indicate that the biggest impact of w/b ratio is on the
magnitude of shrinkage within the first few days. In studies on cement paste, Baroghel-Bouny and
Kheirbeck, (2000) reported that for very low w/b mixes, more than 50 per cent of the 1-year shrinkage
was achieved within the first 24 hours. It was also reported that for pastes with normal w/b ratios (0.45
and above) some swelling may occur during this early period. Others have observed this early swelling
(Scheiβl et al, 2000). The results also indicated beyond the first month or so after casting, w/b ratio had
very little effect on the rate of shrinkage. For example, between 28 days and 360 days Baroghel-Bouny
and Kheirbeck, (2000) reported an increase in autogenous shrinkage of about 400 ± 50 microstrain for
pastes with w/b ratios from 0.25 to 0.60. Interestingly the trend was for the magnitude to be higher for the
higher w/b mixes.

With regard to early-age cracking, however, it is the stress-inducing component of autogenous shrinkage
that is important. In the study by Scheiβl et al (2000) the stress development associated with autogenous
shrinkage after 7 days was measured. It was concluded that tensile stresses could not be determined
directly from autogenous shrinkage. Indeed, in some specimens tensile stresses were recorded even
though there was a reported net swelling of the concrete. A closer inspection of these results suggests that,
as assumed for early thermal stresses, much of the stress developed due to strain within the period
immediately after setting is relieved by creep and may be ignored in relation to stress development.
Where data are provided (Scheiβl et al, 2000) it would appear that the stress at seven days may be related
broadly to the strain that occurs between 1 and 7 days, indicating that strains within the first 24 hours,
whether shrinkage or swelling, did not contribute to stress development. During the period from 1 to 7
days, shrinkage strains were of the order of 100 microstrain, leading to stresses of about 3 MPa. The
estimated mean elastic modulus of these concretes derived from the stress and strain measurements is
about 33 GP, this being consistent with high performance, low w/b concrete.
Assuming that the early autogenous shrinkage will not contribute to stress development, the data from
Figure A4.4 have been normalised to show only the change in autogenous shrinkage beyond 24 hours.
The curves are shown in Figure A4.5 and are compared with the estimates of EN1992-1-1.

When taking into account the uncertainties in generating the results for concrete from tests on cement
paste may be concluded that the values estimated using the method of EN1992-1-1 are broadly consistent
with the reported data over the long-term. The principal difference is at very early-age, the period of
interest for this guide, where there is evidence of some early swelling in concretes with w/b ratios greater
than about 0.45. These early values are shown in Figure A4.6 on an expanded timescale. The results have
been used in considering an alternative set of design values for early-age autogenous shrinkage and these
are given in Figure A4.7. Using this approach autogenous shrinkage may be expressed by the equation:

εca  240t 0.15  650(w/b) (A4.6)

where: t is the time in days.

These results are consistent with current practice in the UK which has assumed that autogenous shrinkage
does not occur when w/b < 0.45. In view of the current uncertainties in predicting autogenous shrinkage it
is proposed that swelling is not taken into account in the design process.

CIRIA C660 6
Appendix A4 Autogenous shrinkage

Figure A4.5 A comparison between autogenous shrinkage estimated using the method of
EN 1992-1-1 and values derived from the results of Baroghel-Bouny and Kheirbeck

CIRIA C660 7
Appendix A4 Autogenous shrinkage

Figure A4.6 Data used as the basis for the development of alternative design values for
early-age autogenous shrinkage related to time and w/b ratio

Figure A4.7 Alternative design values for autogenous shrinkage for CEM I concrete

CIRIA C660 8
Appendix A4 Autogenous shrinkage

A4.4 Influence of binder type

The above analysis has dealt with concretes containing CEM I. Limited data are available, however, on
the influence of cement chemistry and binder type. It is apparent that the use of silica fume or ggbs will
increase the magnitude of autogenous shrinkage at a given w/b ratio, while a reduction is achieved
through the use fly ash. A summary of the data reviewed is given in Table A4.1.

Table A4.1 The influence of mineral additions on autogenous shrinkage

Source % %
Time % % % %
CP = cement paste w/b silica fly
days change change slag change
C = concrete fume ash
Baroghel-Bouny et CP 0.25 28 5 37 - - - -
al, (2000)
0.25 28 10 89 - - - -
0.35 28 5 46 - - - -
0.35 28 10 83 - - - -
Pigeon et al, C 0.35 5 8 50 25 -30 35 10
(2005)
Scheiβl et al, 2000 C 0.35 21 4 50 22 -10 - -
0.35 21 8 200 50 -50 - -
Saje et al, (2001) C 0.4 2 10 20 - - - -
Igarashi et al, CP 0.25 7 10 0 - - - -
(2001)
Poppe et al, (2001) C 0.37 6 - - - - 33-40 0
0.37 6 - - - - 66-80 70
Nawa and Horita, CP 0.32 7 - - 10 -19 10 27
(2004)
0.32 7 - - 30 -53 30 51
0.32 7 - - 60 -79 60 63

A4.4.1 Silica fume

Much of the published data on autogenous shrinkage covers concrete with silica fume (sf) as is it typically
used in low w/b, high strength concretes. However, the results are extremely variable. In work on cement
pastes at a w/b of 0.25 and 0.35, Baroghel-Bouny et al (2000) reported an increase in autogenous
shrinkage after 28 days of about 40 per cent with 5 per cent sf and 85 per cent with 10 per cent sf. In
concretes with a w/b of 0.35 Pigeon et al (2005) reported a 50 per cent increase after 5 days over CEM I
concrete with 8 per cent sf. Also using concretes at a w/b of 0.35, Schieβl et al (2000) recorded significant
increases in autogenous shrinkage after 21 days with the use of sf; about 50 per cent with 4 per cent sf,
over 200 per cent at 8 per cent sf and over 400 per cent at 16 per cent sf. Saje et al (2001) also reported an
increase in autogenous shrinkage in sf concretes at a w/b ratio of 0.4. After 48 hours the increase was
about 20 per cent using 10 per cent sf. Based on these studies it would appear, therefore, that the use of
silica fume will result in an increase in autogenous shrinkage by a significant extent although, in other
studies the effect of silica fume has been found to be much less significant. For example, Igarashi et al
(2001) reported little difference between CEM 1 and silica fume concrete at a w/b of 0.25. However, no
data have been identified that show a reduction in autogenous shrinkage with the use of silica fume and
no data have been identified for concretes at more commonly used w/b ratios.

Current evidence indicates that silica fume causes an increase in autogenous shrinkage and that the
values given in Figure A4.7 are increased by 10 per cent for every 1 per cent of silica fume
expressed as weight of the total binder content (Figure A4.8).

CIRIA C660 9
Appendix A4 Autogenous shrinkage

A4.4.2 Fly ash

Scheiβl et al (2001) tested concrete at a w/b of 0.35 with fly ash levels of 22 per cent and 50 per cent and
reported reductions after 21 days in the order of 10 per cent and 50 per cent respectively. These concretes
also contained 8 per cent sf and it should be noted that the use of fly ash partially offset the increase
caused by the silica fume. Pigeon et al (2005) investigated low w/b concrete with 25 per cent fly ash and
observed a reduction after 5 days of about 30 per cent compared with CEM I concrete. Testing cement
pastes with a w/b of 0.32, Nawa and Horita (2004) observed reductions in autogenous shrinkage after 7
days of 19, 53 and 79 per cent with fly ash contents of 10, 30 and 60 per cent. While the results are
varied, it may be concluded from the evidence available that fly ash will reduce the extent of autogenous
shrinkage.

The evidence available indicates that fly ash causes a reduction in autogenous shrinkage and that the
values given in Figure A4.7 are reduced in direct proportion to the weight percentage of fly ash used
in the mix (Figure A4.8).

A4.4.3 GGBS

In a study of self-compacting concrete with a w/b of 0.37, Poppe and Schutter (2001) reported an increase
in autogenous shrinkage after 7 days of about 70 per cent when using CEM III/B42.5 (66 to 80 per cent
blast furnace slag) when compared with identical concrete using CEM I 52.5R. Using a 50:50 blend of the
two binders (ie with 33 to 40 per cent slag) achieved a result similar to that of the CEM I 52R concrete.
Pigeon et al (2005) using 35 per cent slag, observed an increase at 5 days of about 10 per cent when
compared with CEM I concrete at a w/b of 0.35. Using binder pastes at a w/b of 0.32, Nawa and Horita
(2004) reported increases after 15 days of 27, 51 and 63 per cent with 10, 30 and 60 per cent slag
respectively. While the results are varied, it may be concluded from the evidence available that blast
furnace slag will increase the extent of autogenous shrinkage.

Current evidence indicates that ggbs causes an increase in autogenous shrinkage and that the values
given in Figure A4.7 are increased by 8 per cent for every 10 per cent of ggbs expressed as weight of
the total binder content (Figure A4.8).

Figure A4.8 Adjustments to the design curves in relation to the use of silica fume,
ggbs and fly ash

CIRIA C660 10
Appendix A4 Autogenous shrinkage

A4.5 Conclusions

1. A review of published data has indicated high degree of variability of results and a difficulty in
predicting reliably the magnitude of autogenous shrinkage for use in design. EN1992-1-1
assumes that autogenous shrinkage occurs in all structural grades of concrete and the evidence
available indicates broad agreement when considering the uncertainties involve in generating the
data. A more comprehensive review of the available data and additional research would be
needed to justify a change in the EN1992-1-1 method.

2. The magnitude of stress developed is not always in proportion to the magnitude of unrestrained
autogenous shrinkage, with early creep playing a significant role. On this basis it would appear
that deformation within the first 24 hours or so, whether shrinkage or expansion, has little direct
effect on stress development. Comparing the values estimated using EN1992-1-1 with publish
results indicates that this effect may have been taken into account.

3. There is evidence to suggest that the nature of mineral additions may also be influential in
determining the magnitude of autogenous shrinkage, at least in concretes of low w/b ratio. Both
silica fume and ggbs may result in an increase in autogenous shrinkage compared with CEM I,
while a reduction may be achieved with the use of fly ash. Additional research on the effects of
these materials at more realistic levels of w/b would be useful.

CIRIA C660 11
Appendix A4 Autogenous shrinkage

A4.6 References

Baroghel-Bouny, V and Kheirbek, A (2000)

“Effect of mix-parameters on autogenous deformations of cement pastes – microstructural
interpretations”
In: Proc Int. RILEM PRO17 workshop Shrinkage of concrete – Concrete 2000, Paris, France, 16-17
October 2000 (eds: V Baroghel-Bouny and P-C Aïtcin) Paper 8, pp 115-142

Hobbs, D W (1974)
"Influence of aggregate restraint on the shrinkage of concrete, 
Journal of ACI, Sept, 445­450, American Concrete Institute, Detroit, Michigan

Igarashi, S and Kawamura, M (2001)

“Effects of microstructure on restrained autogenous shrinkage behaviour in high strength concrete at
early-ages”
In: Proc Int. Conf. RILEM PRO 23 Early-age cracking in cementitious systems – EAC’01, Haifa, Israel,
12-14 March 2001 (eds K. Kovlerand, A. Bentur), RILEM Publications, Paper 13, 125-132

Nawa, T and Horita, T (2004)

“Autogenous shrinkage of high-performance concrete”
In: Proc. Int. Workshop Microstructure and durability to predict service life of concrete structures,
Sapporo, Japan, 2004

Pigeon, M, Bissonnette, B, Marchand, J, Boliy, D and Barcelo, L (2005)

“Stress relaxation of concrete under autogenous early-age restrained shrinkage”
ACI, Special Publication, SP-227-16, Detroit, Michigan

Poppe, A-M and Deschutter G (2001)

“Heat of hydration of self-compacting concrete”
In: Proc Int. Conf. RILEM PRO 23 Early-age cracking in cementitious systems – EAC’01, Haifa, Israel,
12-14 March 2001 (eds K. Kovlerand, A. Bentur), RILEM Publications, 13, pp 71-78

Saje, D Saje, F and Kavčcič, F (2001)

“Autogenous shrinkage as a special property of HPC”
In: Proc 3rd Int. Conf. Concrete under severe conditions CONSEC ’01, Vancouver, Canada, 18-20 June
2001 (eds N Banthai, K Satai and O E Gjorv), University of British Columbia, pp 1538-1545

Scheibl, P, Plannerer, M and Brandes, C (2000)

“Influence of binders and admixtures on autogenous shrinkage of high performance concretes”
In: Proc Int. RILEM PRO17 workshop Shrinkage of concrete – Concrete 2000, Paris, France, 16-17
October 2000 (eds: V Baroghel-Bouny and P-C Aïtcin) Paper 12, pp 179-190

Tazawa, E SATO, R Sakai, E and Miyazawa, S (2000)

“Work of JCI committee on autogenous shrinkage”
In: Proc Int. RILEM PRO17 workshop Shrinkage of concrete – Concrete 2000, Paris, France, 16-17
October 2000 (eds: V Baroghel-Bouny and P-C Aïtcin) Paper 2, pp 21-41

CIRIA C660 12