SELF-DESICCATION AND ITS IMPORTANCE IN CONCRETE TECHNOLOGY

1. INTRODUCTION

Bertil Persson, M.Sc., Lie. Tech. Lund Institute of Technology, Div. Building Materials, Lund University, P OBox 118,221 00 Lund, Sweden.

ABSTRACT

Self-desiccation occurs in all types of concrete due to the chemical shrinkage that takes place when water is attached to the cement. In High Performance Concrete the low porosity makes the effect of self-desiccation much more pronounced. This paper presents a summary of a Nordic Seminar on Self-Desiccation in Concrete held in Lund. For this purpose aspects of mix design, curing conditions, desiccation, micro-cracking, self-stresses, shrinkage and frost resistance related to self-desiccation are detailed. Initially a background is given on the principal effects of self-desiccation. This Nordic Seminar took place in Lund on 10 June 1997.

Keywords: Autogenous shrinkage, Chemical Shrinkage, Frost resistance, High Performance Concrete, Hydration, Internal relative humidity, Moisture, Self-desiccation, Selfstresses.

1.1 High Performance Concrete

Prestressed High Performance Concrete (HPC) with water-cement ratio, w/c < 0.38, has been used in Sweden since 1948 for the fabrication of water pressure pipes. The Swedish invention was also licensed abroad in more than 100 factories. The pipes exhibit superior durability compared with pipes made of normal strength concrete. Besides pipes, columns for electrical power lines and self-desiccating concrete slabs (more than 1 million square metres used to

date) became the first large applications of HPC in Sweden. Self-desiccation is studied during moisture-insulated conditions (constant weight) and at constant temperature conditions. No exchange of moisture takes place to or from the concrete specimen during the test period.

1.2 Self-desiccation

The fundamental cause of self-desiccation is the chemical shrinkage that takes place during hydration of water to cement. As compared with the specific volume of water in the capillary pores the specific volume of the hydrated water in the gel of concrete is reduced by about 25%. Due to the decreased size of the capillary pores especially at low w/c < 0.38, the effect of self-desiccation becomes more pronounced for HPC. Self-desiccation influences the properties of the young concrete as well as the long-term behaviour of the concrete, i.e. deformations

- 120 -

caused by self.:.generated stresses, stability and durability (frost resistance and corrosion). Concrete with low w/c deforms even with sealed curing owing to self-desiccation, free of imposed stresses (autogenous shrinkage). Silica fume concrete exhibits a very low long-term increase of the compressive strength due to self-desiccation, which may influence the long-term stability and durability. Low internal relative humidity close to the reinforcement bars is a favourable parameter related to self-desiccation and perhaps decreases the rate of corrosion. The air-filled volume due to self-desiccation created by the chemical shrinkage clearly improves the frost resistance of materials and structures. There is no significant effect of self­desiccation in normal strength concrete (NSC), and it has thus little or no effect on the properties of NSC or on NSC design. Low-w/c concrete has a shortage of water already from casting compared with the amount required for the chemical reactions to come to an end. This means that a pronounced self-desiccation takes place, which is an advantage for solving problems with moisture in the concrete during the time of construction, i.e. the amount of built-in moisture will be reduced since most of the moisture is consumed during the hydration process. The mechanisms behind self-desiccation and the design criteria of the stresses that occur in the concrete owing to the self-desiccation are largely unknown. One purpose of this Nordic Seminar was to compile the available knowledge on self-desiccation and to edit a state­of-the-art report in the field related to self-desiccation in concrete. Another purpose was to initiate new applications of HPC and new ways to estimate the effect of self-desiccation.

2. MIX DESIGN - EFFECT OF W /C AND AIR CONTENT

Since the effect of self-desiccation was more generally applied in Finland and Sweden in order to solve problems related to built-in moisture in the concrete during the time of construction, intensive development of the mix design took place. Too high an amount of built-in moisture caused additional costs as high as SEK 3 billion yearly in Sweden. A good part of this cost is related to built-in moisture from concrete. The high moisture content in the pores of the concrete affected adhesives or wood placed in direct contact with the concrete. One solution to avoid built-in moisture problems for NSC is to increase the drying time, often as much as 1 year. However, owing to economic requirements this solution is not feasible. Another solution is the use of HPC with very low content of built-in moisture. The initial development of the mix design led to the conclusion that it was favourable to use air-entrainment and silica fume in the concrete to make the self-desiccating HPC more workable /2/. The governing parameter in mix design was the water-cement ratio, w/c, which should not exceed 0.38 if an internal relative humidity, RH< 0.85, was required even under wet outer conditions such as rain or snow /3-5/. If w/c was larger than 0.38 the time of desiccation of the concrete became very long when the concrete was wetted at early ages, which was recently confirmed during a large­scale field test, Figure 1 /6/. In Finland the effect of extreme air-entrainment has recently been studied, up to 11 % air content /7 /. However, w/c still had the largest effect on the self­desiccation, Figure 2 /7 /. If the strength was held constant, the air-entrainment became favourable since lower w/c was then required, Figure 3 /7/. Another demand was to lower the amount of superplasticiser in the HPC, which was possible after use of an ideal grading curve of HPC in the fresh state /8,9/. The high cost of the superplasticiser was a reason for the required reduction. An ideal semi-linear logarithmic grading curve also improved the stability of the air-entrainment / 10/.

- 121 -

100

95

-~ 90 :c a::

i 85 '.'2 E :::i .c: 80 <I)

> ~ ~ 75 m C: ,_ JB 70 C:

65

.. '

\ O.SA

0.4A

60 --1---+----1-----1----1

0 10 20 30 40

Age (weeks)

111 w/c=0.4 A o w/c=0.4 W • w/c=0.5 A

o w/c=0.5 W & w/c=0.7 A A w/c=0.7 W

Figure 1. Development of RH in field tests of 125 mm floating concrete slabs /6/. Depth of measurement: 50 mm. A= air curing; W= air

curing after water curing for 1 months.

3. EARLY CURING

100 A

-~ 95 :i: a:: :;; ~

" 90 E ::l .c: Q)

> j 85 ~ m C: .... Q)

i: 80

0

w/c=0.36 - 11 % air

5

.•.. ..... .........

10

Age (weeks)

..... I

•

15

111 w/c=0.34 +12 °C ow/c=0.34 +20 °C

• w/c=0.36 +12 °C <>w/c=0.36 +20 °C

& w/c=0.75 +12 °C A w/c=0.75 +12 °C

Figure 2. Internal relative humidity, RH, in

100 mm cubes versus time at varying water­cement ratio, w/c, and air-entrainment. Raw

data from /7/.

Early drying desiccation from the surface of fresh concrete results in so-called plastic shrinkage

II II, which can be controlled by sealing or addition of moisture to the surface /12/. HPC is

more vulnerable owing to the lack of bleeding water and to a more pronounced early chemical

shrinkage compared with NSC /13/. Lately also water-based plastic solutions have been used in

order to control the plastic cracking due to early shrinkage /10/. If required, the surface with

the sealing solution may be ground away 1 or 2 days after the casting. As an alternative

additional water may be supplied to the surface by the channels of a special type of plastic

fleece /12/. Also the sides and the bottom of the formwork may be covered with fleece in order

to avoid possible surface drying through the formwork. Studies have shown that this method

of controlling the plastic shrinkage also improved the durability of the surface since the amount

of microcracking and the diffusivity decreased.

- 122 -

4. AUTOGENOUS SHRINKAGE

Autogenous shrinkage occurs when RH in a sealed concrete decreases due to the chemical shrinkage that takes place when the water is attached to the cement during hydration /14/. The relationship between RH and shrinkage was nearly linear, Figure 4 /15/. The aggregate in the concrete restrained the shrinkage substantially compared with the free shrinkage that occurred in cement paste. Since the autogenous shrinkage in low-w/c HPC is of the same order as the drying shrinkage, the risk of cracking even of sealed structures must be prevented by suitable working joints or by a sufficient amount of reinforcement. Use of prestressing may be an alternative crack-controlling measure. However, most of the autogenous cracking occurs within 3 months, which makes it possible for the contractor to repair any cracks within the construction time. The magnitude of the autogenous shrinkage also was dependent on the type of the cement /16/. Low-alkali cement causes less autogenous shrinkage than normal alkali cement; a slowly hardening cement has less shrinkage than a normal cement /16/. Tests show that the autogenous shrinkage in concrete with light-weight aggregate, L WA, perhaps 1s eliminated /13/. L WA seemed to perform like a reservoir for water supply in the concrete.

100

-"#. 95 -:c 0:::

i :E 90 E ::s .c

~ ~ 85 f iu E Q.l c 80

l K20-4%

\

II 'ol K:.t1 % I <> ........ <> ..

~ ~-----~. I K30-4% och 8% I 'a"- --

75 -1-----1------+----i

0 10 20

Age (weeks)

11 K20-4% a K30-4% • K30-8%

<> K35-1% A K50-1% A K50-4%

30

Figure 3. RH in 100 mm cubes versus time at varying strength, and air-entrainment. K = cube strength (MPa). 1% = 1% of air­entrainment. Raw data from /7 /.

-E ,_ Q.l c. -

4

3

Q.l 2 Cl CU .:.:: C: 't: .c: Cl) 1

' '' "\. • <>

' '

0 +----------1c------w--,

0 50

Internal relative humidity (%)

11 Concrete w/c=0.26;s/c=0.1

a Concrete w/c=0.48

• Cement paste w/c=0.19;s/c=0.1

<> Cement paste w/c=0.34

100

Figure 4. Shrinkage versus RH in concrete and cement paste. c = cement content; s= content of silica fume; w = water content. Raw data from /15/.

- 123 -

5. SELF-GENERA TED STRESSES

A large number of studies have been performed concerning the effect of autogenous shrinkage on the self-generated stresses in concrete /13-18/. During the measurements of self-generated stresses in HPC a stress rig with controlled length is most often used /13/. Owing to external restraint, self-generated compressive stresses in reinforcement reaching as much as 40 MPa have been observed in HPC /17 /. In a most interesting method to m~asure the internal self­generated stresses in the cement paste, a porcelain ball with manganin wire around it is used /19/. The manganin wire displayed changes in the pressure of the cement paste surrounding it. Alternatively a mercury thermometer may be used cast in the concrete. Small volume changes in the concrete are transferred to the mercury in the thermometer and then read as a change in temperature /19/. Both types of detectors were first calibrated in oil subjected to hydraulic pressure. Figure 5 shows self-generated stresses in cement pastes with w/c= 0.30 at different amounts of silica fume /19/. The amount of silica fume is significant with regard to the development ·of self-generated stresses in the concrete. This effect is in turn explained by the pore distribution in the concrete. The pore distribution may be described as the degree of saturation, So, related to RH, Figure 6 /14/. For example So= 0.83 gives RH= 0.90 in Portland cement concrete but RH = 0.80 in concrete with 10% silica fume. According to the well­known Kelvin equation, lower RH causes greater underpressure in the pore water and thus also larger autogenous shrinkage.

6. SURFACE MOISTURE AND VOLATILE EMISSIONS

Large volatile emissions from the adhesive between the concrete and a plastic carpet may occur when the RH = 0. 90 both at the surface of the concrete and at the critical depth provided that the concrete did not carbonate /20-23/, i.e. the concrete was cured with aluminium foil. Due to the very dense structures of silica fume concretes, the available volume is less in silica fume concretes than in Portland cement concretes with RH held constant, Figure 6 /14/. The required degree of saturation to absorb the glue is more or less constant in silica fume concretes and Portland cement concretes. For example, at a degree of saturation, So= 0.83, RH = 0.90 is obtained in Figure 6. In concrete silica fume with 6% silica fume RH = 0.84 according to Figure 6. If adhesive and a plastic carpet are placed on a concrete with 6% silica fume (at RH = 0.84 in the surface) the critical degree of saturation most certainly will be exceeded, which explained the large volatile emissions /20-23/.

The solution to the problem is perhaps to create a free pore volume in the surface by drying, with a volume that is large enough for the glue to be absorbed /24/. The required surface drying time is about 1 month. During the surface drying time carbonation will also take place, which substantially lowers the pH in the surface of the concrete, which is beneficial in order to decrease the reactions with the adhesives /25/. In low-w/c HPC carbonation takes place provided that no silica fume is used /26/. After two months of carbonation the volatile emissions from a plastic placed with adhesives on concrete were reduced to one sixth compared with emissions from a surface that did not carbonate before the adhesive was applied /27/.

- 124 -

18

16 -(tl a..

14 :a= -CII (1) 12 (I) CII ~ - 10 CII ,:, (1) 8 -e a, 6 C a, Cl

I 4 -© (/)

2

0

0

Iii

D a_ -o- -

cJ"o-- --9' I

/Afl<a-n- .. .. ~ ... .. -A- .. - .s\ II

A11 g

·100 200 300 400

Age (h)

111 Porselain ball - 20% silica fume

a Termometer - 10% silica fume

A Termometer- no silica fume

Figure 5. Self-generated stresses in cement pastes with w/c= 0.30 at different amounts of silica fume /19/.

7. FROST RESISTANCE

1

:I: 0.95 0::: ~

:!:: ,:,

ocfJ;\ E :::, 0.9 .c

D f D QJ >

/4113011 .::: a:s e 0.85

cu C i.. QJ ... C 0.8

0.75 0.8 0.85 0.9

Degree of saturation

11 Concrete with 10% silica fume

o Portland cement concrete

D

0.95

Figure 6. Degree of saturation, S0, versus internal relative humidity, RH /14/.

Even after several years water-cured HPC with low w/c maintains a very low internal relative humidity, RH, due to self-desiccation, at least in the inner part of the concrete a couple of centimetres from the surface /28/. The self-desiccation thus creates a free pore volume in HPC where the ice may expand during freezing. (The total amount of water is also less in HPC than in NSC, which also in turn increases the frost resistance.) Young HPC tolerates freezing at earlier age than a NSC, Figure 7 /28/. Furthermore, the scaling from HPC with low w/c seemed to be very small provided that no silica fume was used /28,27/. The scaling of silica fume concrete shows the opposite after 30 cycles of testing. Perhaps the porosity of silica fume concretes was altered, decreasing the available volume for ice to expand, cp. Figure 6 /14/.

8. SUMMARY AND CONCLUSIONS

A Nordic Seminar on Self-Desiccation and Its Importance in Concrete Technology was held in Lund on 10 June 1997. This article summarised the Seminar and gave some practical aspects on the effect of self-desiccation. The following conclusions were drawn:

- 125 -

:2 50 -g> 40 'i: :I (.)

"C >, c-.!.~ 30 ~ t; - <1) C C. .2 l8 e i.. 20 "C >, .c: 11--0 Q)

~ 10 C) (1)

C

/ /0

/5 6

I

I I

~

I

0 +------------0.3 0.5 0.7 0.9

Water-cement ratio, w/c

a Degree of hydration(%) o Curing time (h)

Figure 7. Required degree of hydration (%)

and curing time (h) in order to avoid frost damage. Raw data from /28/.

-N E -C) -

10000

1000

100

10

0 50 100

Number of freezing cycles

-•-w/c=0.35

--o-w/c=0.35 + 5% silica fume

-•-Vet= 0.40

--:-<>--w/c= 0.40 + 5% silica fume

Figure 8. Scaling from Portland cement and silica fume concrete respectively. Raw data from /28,27/.

• The fundamental cause of self-desiccation was the chemical shrinkage that takes place

during hydration of water to cement.

• The effect of self-desiccation was clearly observed in low-w/c concretes, i.e. in concrete

with w/c < 0.38. • Autogenous shrinkage, one effect of self-desiccation, was nearly linearly dependent on the

internal relative humidity, RH, ofHPC.

• Autogenous shrinkage also caused self-generated compressive stresses up to 40 MPa in the

reinforcement in HPC. • Self-desiccation may have a beneficial effect on the frost resistance of Portland cement

based HPC (primarily given that no silica fume is used in the mix design).

• Self-desiccation in concrete with low w/c decreased the risk of built-in moisture in

structures since a great part of the capillary water was hydrated to the cement.

ACKNOWLEDGEMENT

The Swedish Council of Building Research financed part of the Seminar, which is hereby

gratefully acknowledged. I am also most grateful to Professor Goran Fagerlund for his review.

- 126 -

REFERENCES

Ill B. Persson, G. Fagerlund. Self-desiccation and Its Importance in Concrete Technology. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. (1997).

/2/ M. Karlsson. Rapid-Desiccating Concrete. Cementa 2.92. Cementa. Danderyd (1992).

/3/ G Hedenblad. Desiccation of Moistur~ during the Construction Period - Drying Time and Measurement of Moisture. Report Tl2: 1995. ·Swedish Council of Building Research. Stockholm (1995).

/4/ B. Persson. DRY for Choice of Water-Binder Ratio in Concrete Free of Moisture during the Construction Period. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund (1997).

/5/ M. Gerlam. SBUF Guide for Drying of Moisture in Concrete. Sabema. Kfillered (1996).

/6/ B. Persson. Effect of Grinding on Surface Alkalis of Concrete. Report 97.07. Div. Building Materials. Lund Institute of Technology. Lund (1997).

/7/ V. Penttala, L Wirtanen. Drying of Concrete with Low Water Binder Ratio and High Air Content. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 209-226 (1997).

/8/ B. Persson. Ideal Grading Curve in Fresh Concrete. BETONG 3/95. Stockholm (1995).

/9/ B. Persson. Concrete in Efficient Construction. Bygg & Teknik 7/96. Stockholm (1996)

/10/ J. Persson. Private communication. Sydsten LTD. Malmo (1996).

/11/ A. Radocea. Autogenous Volume Change of Concrete at Very Early Age - Model and Experimental Data. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 56-71 (1997).

/12/ U. Guse, H.K. Hilsdorf. Surface Cracking of High Strength Concrete - Reduction by Optimization of Curing Regimes. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 239-249 (1997).

/13/ 0. Bj0ntegaard, T.A. Hammer, E. Sellevold. High Performance Concrete at Early Ages: Self-Generated stresses due to Autogenous Shrinkage and Temperature. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 1-7 (1997).

/14/ B. Persson. Experimental Studies of the Effect of Silica Fume on the Chemical Shrinkage and Self-Desiccation in Portland Cement Mortars. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 116-131 (1997). ·

.. 127 -

/15/ V. Baroghel-Bouny. Experimental Investigation of Self-Desiccation in High­Performance Materials - Comparison with Drying Behaviour. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 72-87 (1997).

/16/ E-I. Tazawa, S Miyazawa. Effect of Self-Desiccation on Volume Change and Flexural Strength of Cement Paste and Mortar. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 8-14 (1997).

/17 / F. Tomozawa, T Noguchi, K.B. Park. Experimental Determination and Analysis of Stress and Strain Distribution of Reinforced High-Strength Concrete Column Caused by Self-Desiccation and Heat of Hydration. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology. Lund, 99-115 (1997).

/18/ H Hedlund, G Westman. Measurement and Modelling of Volume Change and Reactions in Hardening Concrete. Report TVBM-3075. Div. Building Materials, 174-192 (1997).

/19/ B. Dela, H. Stang. Eigenstresses in Concrete due to Autogenous Shrinkage. Report TVBM-3075. Div. Building Materials. Lund Institute of Technology, 46-51 (1997).

/20/ H.W. Johnsson. Chemical Emissions from Flooring - Effect of Concrete Quality and Moisture. Publication P95:4. Chalmers University of Technology. Gothenburg (1995).

/21/ M. Fritsche, A. Sjoberg, H.W. Johnsson. Chemical Emissions from Combined Glued Flooring on Self-Desiccating Concrete - Effect of Method of Gluing, Type of Drying, Type of Cement, Glue and Plastic Carpet. Publication P97: 1. Chalmers University of Technology. Gothenburg (1997).

/22/ A. Sjoberg. Ongoing Research concerning Flooring System, Moisture and Alkali. AMA­News - Ground - Housing 1/97. Swedish Building Service. Solna, 15-17 (1997).

/23/ A. Sjoberg. Ongoing Research. Floor System, Moisture and Alkali. Contribution at HB/IAQ'97. Washington (1997).

/24/ B. Persson. Drying of Surface in Building-Moisture Free Concrete. AMA-News -Ground - Housing 1/96. Swedish Building Service. Solna, 20-23 (1997).

/25/ 0. Peterson. Priv. corn. Div. Building Materials. Lund Institute of Technology. (1997).

/26/ B. Persson. Long-term Shrinkage in High Performance Concrete. Contribution 2i073. 10th International Congress on the Chemistry of Cement. Gothenburg ( 1997).

/27/ B. Persson. Carbonation and Emissions. BETONG 2/98. Stockholm (1998).

/28/ G. Fagerlund. Effect of Self-Desiccation on the Internal Frost Resistance of Concrete. TVBM-3075. Div. Building Materials. Lund Institute of Technology, 227-238 (1997).

/29/ P.E. Petersson. Salt-frost Resistance of Concrete - Field Tests. Report SP 1995:73. The Swedish Testing and Research Institute. Boras (1995).

- 128 -

APPENDIX: LIST OF PARTICIPANTS AT THE NORDIC SEMINAR IN LUND

Baroghel-Bouny, Veronique, Laboratoire Central des Ponts et Chaussees, LCPC, Paris.

Bentz, Dale; National Institute of Standards of Technology, NIST, Gaithersburg.

Bj@ntegaard, @yvind; The Norwegian University of Science and Technology, Dept. Structural Engineering, Trondheim.

Dela, Birgitte Fries, Dept. Structural Engineering and Materials, DTU, Lyngby.

Fagerlund, Goran, Div. Building Materials, Lund Institute of Technology, Lund.

Hammer, Tor Arne, The Research Institution SINTEF, Civil and Environmental Engineering, Trondheim.

Hansen, Kurt Kielsgaard, Jensen, Ole Mejlhede, Dept. Structural Engineering and Materials, Technical University of Denmark, DTU, Lyngby.

Hedenblad, Goran, Div. Building Materials, Lund Institute of Technology, Lund.

Hedlund, Hans; Div. Structural Engineering, Lulea University of Technology, Lulea.

Juvas, Klaus; Partek Concrete Development Ltd, Pargas.

Mejlhede Jensen, Ole, Dept. Structural Engineering and Materials, DTU, Lyngby.

Koenders, Eddie A B; van Breugel, Klaas, Faculty of Civil Engineering, Delft University of Technology, Delft.

Leivo, Markku, Building Technology, Technical Research Centre of Finland, Espoo.

Miyazawa, Shingo, Ashikaga Institute of Technology, Dept. Civil Engineering, Ashikaga.

Mjornell, Kristina, Dept. Building Materials, Chalmers University of Technology, Gothenburg.

Persson, Berti!, Div. Building Materials, Lund Institute of Technology, Lund.

Radocea, Adrian, Dept. Building Materials, Chalmers University of Technology, Gothenburg.

Tomosawa, Fuminori, Nogushi, T, Park KB, Dept. Architecture, Faculty of Engineering, The University of Tokyo, Hongo, Tokyo.

Westman, Gustaf, Div. Structural Engineering, Lulea University of Technology, Lulea.

Wiens, Udo, Institut fur Bauforschung, Westfalische Technische Hochschule, Aachen, Germany

Wirtanen, Leif, Concrete Technology, Helsinki University of Technology, Espoo, Finland

- 129 -