Moisture in concrete subjected to different kinds of curing

Transcription

1 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997, pp Moisture in concrete subjected to different kinds of curing Bertil Persson Lund Institute of Technology, Division Building Materials, University of Lund, P O Box 118, Lund, Sweden SCIENTIFIC REPORTS Paper received: May 13, 1996; Paper accepted: July 17, 1996 A B S T R A C T This article outlines an experimental and numerical study of moisture in concrete subjected to air, water or sealed curing. For this purpose, columns of eight qualities of concrete were studied at 3 different ages each over a period of 450 days. Parallel studies of hydration, internal relative humidity, self-desiccation and strength were carried out. Finally the article presents an analysis of the internal relative humidity and the mechanisms of moisture transfer by modelling the diffusivity and the capillary conditions of 4 of the concretes. The project was carried out between the years 1989 and R É S U M É Cet article présente une étude expérimentale et numérique de l humidité du béton soumis à l air, à l eau ou à une cure sous enduit de protection. À cet effet, des colonnes faites de huit qualités de béton ont été étudiées à 3 âges différentes sur une période de 450 jours. Des études parallèles ont également été menées sur l hydratation, l humidité interne relative, l autodessication et la résistance. Enfin, l article présente une analyse de l humidité relative interne et des mécanismes de transfert d humidité par modélisation de la diffusion et de la capillarité de 4 de ces bétons. Ce projet de recherche a été mené entre 1989 et INTRODUCTION 1.1 Self-desiccation and strength For normal concrete with a water-cement ratio, w/c > 0.39, sufficient water is available for the hydration of the cement to be finalised. Self-desiccation hardly affects the moisture of the concrete at all. Normally, concrete is regarded as a porous material affected by the ambient climate. However, for concretes with w/c 0.39, the rate of hydration is decreased substantially due to self-desiccation [1]. The volume created in the concrete, due to the chemical shrinkage that takes place when the water reacts with the cement [1], decreases the internal relative humidity, Ø, which can be as low as 0.72 at low w/c. This might then affect the strength of the concrete. Furthermore, a concrete with low w/c has very few capillary pores internally. The concrete might be more porous close to the surface. The few capillary pores in a concrete with low w/c probably influence the effect of water curing on both hydration and strength development of the concrete. It was thus considered important to measure the internal relative humidity of the concrete, especially at low w/c. 1.2 Effect of moisture on the durability of the concrete Durability factors such as corrosion of reinforcement and freeze-thaw resistance are clearly affected by the moisture conditions in the concrete. The chloride diffusion is dependent on Ø in the concrete as well as on diffusion of gas [2]. Above a certain degree of pore saturation, a substantial amount of surface concrete will spall due to ice lenses created in freeze-thaw periods [3, 4]. A freeze-thaw resistant concrete must either be air-entrained or contain a sufficient air-filled pore volume related to chemical shrinkage during the hydration. In this case, too, knowledge of the moisture situation in the concrete is of the utmost importance. 1.3 Factors influencing the human environment In dwelling houses, the maximum ambient Ø is normally restricted to Ø < For houses made of concrete, it was then considered essential to estimate the time required for the drying of the concrete to reach this level of Ø. Editorial note Mr. Bertil Persson is a RILEM Senior Member. He participates in TC 161-GMC on Modelling the behaviour of Concrete in service: a Guide for the engineer /97 RILEM 533

2 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997 During the drying period, the ventilation must be sufficient to reduce the moisture content in the house. When wood is placed directly on the concrete, the Ø of the concrete must not exceed 0.75 [5], or else the moisture of the concrete will cause mould between the concrete and the wood. These organisms will secondarily cause a bad smell in the house, allergic reactions, etc. When Ø > 0.80, the wood starts to rot [5]. Finally, when Ø > 0.90, glued carpets may loosen from the concrete due to too high a saturation of the pores (no space for the glue resin in the concrete pores) [5]. At Ø > 0.90, mould may also occur between the plastic carpet and the concrete [5]. 2. EXPERIMENTAL METHODS 2.1 Specimen The specimen was a disc of 1 m diameter and a thickness of 0.1 m. To simulate a long column, the flat sides of the disc were sealed by thick layers of epoxy resin. The watertightness of the epoxy resin was carefully studied over a period of 6 years [6]. Compared to the porous concrete, the diffusivity of the 2 mm layer of epoxy resin was very low. In this respect the epoxy resin was considered as watertight. At different distances from the exposed circular surface of the column, cast-in plastic tubes were placed, 50, 150 and 350 mm from the edge, Fig. 1. Parallel to the cast-in items, thermocouples were placed in the concrete, Fig. 2. The measurement points were protected by a cover made of expanded plastic insulation in order to minimise the effects of variations in the ambient climate in the laboratory, Fig. 3. Fig. 1 Cast-in plastic tube (RH-pipe) in the specimen (diameter 25 mm). Fig. 2 Thermocouple in air-cured and sealed specimens. Fig. 3 Plan (downturned cross-section) of the studied specimen (diameter: 1 m), cast-in plastic tubes (RH-pipes), thermocouples and protective cover. 2.2 Studied material A low-alkali cement was used. Table 1 shows the chemical composition of the cement [7]. Table 1 Chemical composition of the cement Analysed properties (%): CaO 64.6 SiO Al 2 O Fe 2 O MgO 0.84 K 2 O 0.62 Na 2 O 0.07 Alkali 0.48 SO CO Free CaO 1.13 C 2 S 22.5 C 3 S 53.0 C 3 A 1.42 C 4 AF 13.4 Physical properties: Ignition losses 0.63 Blaine (m 2 /kg) 325 Density (kg/m 3 ) Mix proportion of concrete Eight types of concrete were studied, 3 columns of each quality, in all 24 concrete columns. Using optimised mix proportions, an ideal grading curve and a correct mixing order, it was possible to carry out the research on concretes with good workability and with 450-day cylinder strength of up to 142 MPa. The aggregate consisted of crushed quartzite sandstone 8-12 mm (compressive strength: 333 MPa, split tensile strength: 15 MPa, Young s modulus: 60 GPa [8] and ignition losses: 0.25% [7]) together with natural gravel 0-8 mm (granite, ignition losses: 0.85% [7]). The silica fume was granulated powder (ignition losses: 534

3 Persson Table 2 Composition (kg/m 3 dry material) and properties of the concretes [7] Quartzite Gravel Cement, C Silica fume, S Superplasticiser Density Water-cement ratio, w/c Air content (%) Workability (vebe) day strength (cylinder, MPa) day strength (cylinder, MPa) day strength (cylinder, MPa) Measurements of internal relative humidity, Ø The measurement probe was entered into the cast-in tube (RH-pipe) and tightened by an expanding rubber seal. The measurement of Ø was carried out for 22 h in order to stabilise the moisture content between the pores in the concrete and the air in the plastic pipe close to the capacity probe. The temperature of the concrete and the air in the cast-in item were also measured. In the case of submerged concrete, the specimen was lifted above the water before the measurement was carried out Sources of error 2.25% [7]). The superplasticiser was naphthalene sulphonate. In Table 2, the composition (kg/m 3 dry material) of the concretes and the properties in the fresh state are stated. The elastic modulus of the concretes varied between 27 GPa and 50 GPa. 2.4 Casting and curing The specimen was cast in a steel mould and sealed by a double plastic covering in the fresh state. The maximum temperature during the early age was 30 C. At 16 h age, one flat side of the specimen was sealed by 2 mm epoxy resin. The other flat side of the specimen was kept in the steel mould and sealed at 40 h age. One specimen of each quality was also sealed at the circular surface (rim) in order to obtain a totally sealed specimen; one specimen was aircured at the edge starting at 40 h age, and finally the remaining specimen was subjected to water curing after 66 h age. In Fig. 4, the ambient temperature is shown. Fig 5 shows the ambient relative humidity. Calibration faults of the measurement probes, the temperature difference between the probe and the concrete, and too short a measurement period were the main sources of error. The accuracy of the measurements will be discussed below. 3. RESULTS, ANALYSIS AND ACCURACY OF INTERNAL RELATIVE HUMIDITY 3.1 Air curing Figs. 6, 7 and 8 give the internal relative humidity, Ø, at a distance of 50, 150 and 350 mm respectively from the exposed surface. The mix proportions 1, 4, 6 and 8 contained 10% silica fume which is indicated by S in Figs. 6, 7 and 8. It was of great interest, from a practical point of view, to describe the internal relative humidity, Ø, versus the water-cement ratio, w/c, and the age of the concrete. However, for each distance from the exposed surface and each curing condition, different values of Ø were obtained dependent on the presence of silica fume. Formulas of Ø versus w/c and time make sense when used in, for example, computer programs. Fig. 4 Ambient temperature as a function of time. Fig. 5 Ambient relative humidity as a function of time. 535

4 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997 Fig. 6 Ø 50 mm from the surface with air curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). Fig. 7 Ø 150 mm from the rim with air curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). From Fig. 6 the following equations were obtained for Ø 50 mm from the exposed surface: Ø(w/C, t) a50 = 1.24 ( Ln(t)) (w/c) ( Ln(t)) {R 2 = 0.94} (1) Ø(w/C, t) Sa150 = 1.62 ( Ln(t)) (w/c) 0.52 ( Ln(t)) {R 2 = 0.91} (2) a denotes air curing t denotes age (28 < t < 450 days) w/c denotes the water-cement ratio (0.22 < w/c < 0.48) Ln denotes the natural logarithm R 2 = 1 - (SSE)/(SST) (3) SSE= (Y i -Y i ) 2 (4) SST= ( Y 2 i )-( Y i ) 2 /n (5) S denotes 10% silica fume Y i denotes the measured value Y i denotes the mean value From Fig. 7, the following equations were obtained for Ø 150 mm from the exposed surface: Ø(w/C, t) a150 = 1.04 ( Ln(t)) (w/c) (1.24 Ln(t)-1) {R 2 = 0.94} (6) Ø(w/C, t) Sa150 = 1.29 ( Ln(t)) (w/c) 0.28 ( Ln(t)) {R 2 = 0.76} (7) From Fig. 8, the following equations were obtained for Ø 350 mm from the exposed surface: Ø(w/C, t) a350 = ( Ln(t)) (w/c) (0.96 Ln(t)-1) {R 2 = 0.98} (8) Ø(w/C, t) Sa350 = 1.29 ( Ln(t)) (w/c) 0.28 ( Ln(t)) {R 2 = 0.80} (9) The symbols in equations (6) to (9) are given above. Fig. 8 Ø 350 mm from the rim with air curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2) Self-desiccation Fig. 9 shows the internal relative humidity, Ø, at selfdesiccation as a function of time. The mixed proportions 1, 4, 6 and 8 contained 10% silica fume which is indicated by S in Fig. 9. Each mark in Fig. 9 represents the mean value of 6 measurements of the internal relative humidity, Ø. From Fig. 9 the following equation was obtained for Ø: Ø(w/C, t) s = 1.08 ( t) (w/c) 0.16 ( t) {R 2 = 0.86} (10) Ø(w/C, t) Ss = 1.37 ( Ln(t)) (w/c) 0.33 ( Ln(t)) {R 2 = 0.89} (11) 536

5 Persson Fig. 10 Efficiency factor for silica fume related to self-desiccation [7]. Fig. 9 Ø with sealed curing. Symbols: d = days age, S = 10% silica fume [6]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). s denotes sealed curing t denotes age (28<t<450 days) w/c denotes the water-cement ratio (0.22 < w/c < 0.48) Ln denotes the natural logarithm R 2 denotes the parameter given by equations (3) to (5) S denotes 10% silica fume The efficiency factor for silica fume related to selfdesiccation was defined as (w/c) eff = w/(c+k s S) (12) (w/c) eff denotes the efficient (eff) water-cement ratio related to self-desiccation w denotes all the mixing water (kg/m 3 ) C denotes the cement content (kg/m 3 ) k s denotes the efficiency factor for silica fume related to self-desiccation S denotes the content of silica fume (= 10% of the cement content for mixed proportions 1, 4, 6 and 8) Fig. 10 shows the efficiency factor, k se, for silica fume related to self-desiccation as a function of the watercement ratio. The efficiency factor was described by the following equation: k se = 17.2 (0.004 t - 1) (w/c) t {R 2 = 0.98} (13) k se denotes the efficiency factor for silica fume related to self-desiccation t denotes the age of the concrete (days) w/c denotes the water-cement ratio R 2 denotes the parameter given by equations (3) to (5). At a water-cement ratio w/c= 0.39, the efficiency factor, k se = 2.7, was observed to be fairly independent of age. However, at a lower w/c, k se was larger at 28 and 90 days of age and smaller at 450 days of age; the contrary was observed at a higher w/c. This observation will be discussed below. 3.3 Water curing Figs. 11, 12 and 13 give the internal relative humidity, Ø, at a distance of 50, 150 and 350 mm respectively from the exposed surface. The mix proportions 1, 4, 6 and 8 contained 10% silica fume which is indicated by S in Figs. 11, 12 and 13. From Fig. 11, the following equations were obtained for Ø 50 mm from the exposed surface: Ø(w/C, t) w50 = 1.02 ( Ln(t)) (w/c) ( Ln(t)) {R 2 = 0.79} (14) Ø(w/C, t) Sw50 = 1.16 ( t) (w/c) 0.24 ( t) {R 2 = 0.89} (15) Fig. 11 Ø 50 mm from the rim with water curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). 537

6 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997 Fig. 12 Ø 150 mm from the rim with water curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). Fig. 13 Ø 350 mm from the rim with water curing. Symbols: d = days age, S = 10% silica fume [7]. From the left mix proportions: 1, 8, 7, 5, 6, 3, 4 and 2 (given in Table 2). : t denotes age (28<t<450 days) w denotes water curing w/c denotes the water-cement ratio (0.22 < w/c < 0.48) Ln denotes the natural logarithm R 2 denotes the parameter given by equations (3) to (5) S denotes 10% silica fume From Fig. 12 the following equations were obtained for Ø 150 mm from the exposed surface: Ø(w/C, t) w150 = 1.1 ( t) (w/c) 0.18 ( t) {R 2 = 0.78} (16) Ø(w/C, t) Sw150 = 1.19 (w/c) 0.29 ( t) {R 2 = 0.65} (17) The symbols of equations (16) and (17) are given above. From Fig. 13 the following equations were obtained for Ø 350 mm from the exposed surface: Ø(w/C, t) w350 = 1.13 ( t) (w/c) 0.22 ( t) {R 2 = 0.90} (18) Ø(w/C, t) Sw350 = 1.16 ( t) (w/c) 0.26 ( t) {R 2 = 0.80} (19) t denotes age (28<t<450 days) w denotes water curing w/c denotes the water-cement ratio (0.22 < w/c < 0.48) R 2 denotes the parameter given by equations (3) to (5). S denotes 10% silica fume. 3.4 Accuracy Fig. 14 Difference in temperature between concrete (c) and the measurement probe (p) [7]. The calibrations were carried out for 22 h above extremely pure salts in a climate chamber [6, 9]. The calibration fault was within 0.01 Ø. Fig. 14 shows the difference in temperature between the probe and the concrete (less than 0.1 C). The maximum temperature fault was also 0.01 Ø. The required time to obtain stability in humidity between the air in the pores of the concrete and the air around the probe was investigated. Fig. 15 shows the internal relative humidity, Ø, as a function of measurement time. After 14 h, the required stability was obtained. The measurement time was then set at 22 h. The total accuracy was then estimated as ± 0.02 Ø. 538

7 Persson The slope of the isotherm at desorption, dw e /dø, (indicated above) was studied within the project on exactly the same type of material as utilised for the studies of diffusivity [10]. 4.2 Experimental methods utilised in the diffusivity studies Fig. 15 Internal relative humidity, Ø, in air around the capacity probe as a function of time [7]. 4. DIFFUSIVITY 4.1 General At an early stage of the project, it was observed that concrete with a low water-cement ratio attained an internal relative humidity substantially lower than 1 when it was submerged. It was then of general interest to investigate the diffusivity and the capillarity of the concrete to explain the properties of the referred material. The moisture diffusivity was calculated according to: δ w = F 0 h 2 /t (20) δ w denotes the moisture diffusivity (m 2 /s) when 80% of the moisture had dried from the concrete at an ambient relative humidity, Ø = 0.33 F 0 denotes the Fourier constant h is half the thickness of a disc that is drying in both directions (m) t is the drying time (s). The moisture permeability was calculated according to: δ c = ( T) δ p /( ) (21) δ c denotes the moisture permeability (m 2 /s) T denotes the temperature ( C) δ p = δ w (dw e /dø) 1/p 0 (m 2 /s) (22) δ w denotes the moisture diffusivity (m 2 /s) given in equation (20) dw e /dø is the slope of the isotherm at desorption (the ratio of decline in evaporated water, dw e, to decline in internal relative humidity, dø) p 0 denotes the pressure of water vapour at saturation Concrete cylinders with a diameter of 100 mm and a length of 200 mm were cast from the mix proportions 1, 3, 5, and 7. Discs (slices) were cut from one cylinder the day after pouring. Other cylinders of these concretes were cured in water for 250 days and then cut in slices with a thickness of both 15 mm and 45 mm. The circular sides of the slices were sealed by 2 mm epoxy resin. Both sides of the slices were dried, starting directly after preparation (at age 1 day or 250 days). The ambient drying climate had a constant relative humidity of Ø = 33% provided by a salt in solution and was held free from carbon oxide by barium hydroxide. The specimen was weighed continuously for 250 days (thickness 15 mm) or for 500 days (thickness 45 mm). After the diffusivity studies, the same specimen was used for capillarity studies. 4.3 Results and analysis of diffusivity studies Fig. 16 shows the moisture diffusivity as a function of the water-cement ratio. The following equation was obtained for concrete at early ages (drying between 1 and 250 days age): δ we = 7.1 [(w/c) (w/c) ] {R 2 = 0.99} (23) Fig. 16 Moisture diffusivity of concrete as a function of the watercement ratio. Thickness and age of concrete is indicated [10]. 539

8 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997 δ we denotes the early age moisture diffusivity (m 2 /s) when 80% of the moisture had dried from the concrete at an ambient relative humidity, Ø = 0.33 w/c denotes the water-cement ratio (0.22 < w/c < 0.48) R 2 denotes the parameter given by equations (3) to (5). The following equation was obtained for mature concrete (drying between 250 and 750 days age): δ wm = 0.4 (w/c ) {R 2 = 0.95} (24) δ wm denotes mature moisture diffusivity (m 2 /s) when 80% of the moisture in the concrete had dried at an ambient relative humidity, Ø = 0.33 w/c denotes the water-cement ratio (0.22 < w/c < 0.48) R 2 denotes the parameter given by equations (3) to (5). The low diffusivity was one explanation of the fairly slow drying of the column. The moisture diffusivity decreased, especially at a higher water-cement ratio, which further prolonged the drying time of concrete, which already from the start of drying contained more water than was required for the hydration of the cement to come to an end (water-cement ratio, w/c > 0.39). Fig. 17 shows the moisture permeability of the concrete as described in equation (23) CAPILLARITY 5.1 General Fig. 17 Moisture permeability of concrete versus water-cement ratio at varying age and Ø [10]. At early ages, the moisture permeability decreased with the water-cement ratio as it increased with the water-cement ratio in mature concrete, Fig. 17. However, the drying time for mature concrete was much greater than for young concrete, especially with a thickness of 45 mm, Fig. 18. The final weight of the 45 mm slices had then to be estimated based on the drying of the 15 mm slices. The value of resistance to water penetration was defined according to: m = t u /(4 h 2 ) (25) m denotes the value of resistance to water penetration (s/m 2 ) t u denotes the time required for the water to reach the upper side of the specimen h is half the thickness of a disc absorbing water in one direction (m). 5.2 Experimental method used in the capillarity studies Concrete cylinders with a diameter of 100 mm and a length of 200 mm were cast from the mix proportions 1, 3, 5, and 7. Discs (slices) were cut from one cylinder the day after pouring. Other cylinders of these concretes were cured in water for 250 days and then cut in slices with a thickness of both 15 mm and 45 mm. The circular sides of the slices were sealed by 2 mm epoxy resin. Drying of both sides of the slices started directly after preparation (at age 1 day or 250 days). The ambient drying climate had a constant relative humidity of Ø = 33% provided by a salt in solution and was held free from carbon oxide by barium hydroxide. The drying was carried out continuously for 250 days (thickness 15 mm) or for 500 days (thickness 45 mm). After the diffusivity studies, the specimen was placed horizontally on supports submerged about 1 mm below the surface of the water basin, which was sealed. The water vapour in the air above was thus saturated. The specimen was weighed continuously for 250 days. Before the weight was taken, the specimen was released from free water on the surface. Afterwards, the surface was wet completely before it was placed back in the water basin to avoid air bubbles between the water and the concrete. 5.3 Results and analysis of capillarity Fig. 19 shows the water suction of 6 slices (15 mm) as a function of the square-root of time. As can be observed

9 Persson Fig. 18 Relative evaporated water versus drying time for slices of varying water-cement ratio, w/c and varying thickness given in the figure [10]. Young or mature concrete as indicated. in Fig. 19, the curve was distinctly bent when the water reached the upper side of the specimen at the suction time t u. From Fig. 19, the value of resistance to water penetration (s/m 2 ) was easily calculated by equation (25). However, it was of more general interest to study the value of resistance to water penetration versus the capillarity porosity of the cement paste, (P c ) p, defined as: (P c ) p = (w/c α)/( w/c) (26) (P c ) p denotes the capillary porosity of the cement paste (-) w/c denotes the water-cement ratio α denotes the degree of hydration defined as w n /C w n denotes the water-losses of ignited concrete between 105 C and 1050 C [11] C denotes the cement content in the concrete, Table 2. The degree of hydration was studied parallel to the capillarity [10] and it was thus possible to estimate the capillary porosity of the cement paste according to equation (26). Fig. 20 shows the value of resistance to water penetration versus the capillarity porosity of the cement paste. Fig. 19 Water suction of 6 slices (15 mm) as a function of square-root of time ( s) [10]. 541

10 Materials and Structures/Matériaux et Constructions, Vol. 30, November days and 450 days of age, Fig 10. At w/c= 0.47 more water was available than required for the hydration of the cement to proceed, which also gave the silica fume the amount of calcium hydroxide needed for the puzzolan reaction to proceed. The self-desiccation caused an excess pressure in the pore water of the concrete to occur, which in its turn resulted in a compressive stress in the aggregate and the cement paste of the concrete [14]. A substantial autogenous shrinkage took place [12]. However, the self-desiccation was also beneficial for solving moisture problems during the construction time [10, 15]. It seemed as if the compressive stress in the concrete caused by the self-desiccation also made the concrete less permeable to water. Fig. 20 Resistance to water penetration versus the capillarity porosity of the cement paste [10]. From Fig. 20, the following equation was obtained [10]: m = 26 [0.3 - (P c ) p ] 10-7 {0.08 < (P c ) p < 0.22; R 2 = 0.83} (27) m denotes the value of resistance to water penetration (s/m 2 ) (P c ) p denotes the capillary porosity of the cement paste (-) R 2 denotes the parameter given by equations (3) to (5). Equation (27) was based on studies on 15 mm slices. When slices with a thickness of 45 mm were studied, the suction time became much larger, especially at a lower water-cement ratio, Fig DISCUSSION 6.1 Self-desiccation About 150 measurements were carried out related to sealed concrete columns. A substantial self-desiccation of the concrete was observed at a lower water-cement ratio, w/c. The self-desiccation was caused by the chemical shrinkage of the water that occurred when it was attached to the cement [1]. A gas-filled pore volume appeared in the concrete, causing the internal relative humidity, Ø, to decline [12]. However, at a lower w/c, the rate of self-desiccation was initially fast, which in turn decreased the rate of hydration [13]. At w/c = 0.22 a decrease of Ø of only 0.06 between 28 days and 450 days of age was observed, Fig. 9. Since the reaction between silica fume and calcium-hydroxide (the puzzolan reaction) was dependent on the hydration of the cement, the so-called efficiency factor of the silica fume related to self-desiccation also decreased from about 5 to about 1 between 28 days and 450 days of age, Fig. 10. The opposite development of the so-called efficiency factor was observed at w/c = The efficiency factor then increased from 1 to 4 between Fig. 21 Ratio of absorbed water to evaporated water versus time. w/c= water-cement ratio. 6.2 Air curing About 150 measurements were carried out related to drying concrete columns. The measurements were carried out 50, 150 and 350 mm from the surface of the column (rim). Fifty mm from the surface of a concrete with a low water-cement ratio, w/c, the decrease of internal relative humidity, Ø, was very slow even when the column was subjected to air curing. Between 28 days and 450 days of age, Ø dropped only 0.09 at w/c = 0.22 (in the interior of the same concrete column, Ø dropped 0.07), Figs. 6 and 8. However, in a concrete with w/c = 0.58 the comparable decrease of Ø was 0.17 and 0.05 respectively, Figs. 6 and 8. These observations were probably explained by the moisture transport coefficients of the concrete which were also studied, Figs. 16 and 17. The moisture diffusivity of a very dense concrete with w/c = 0.22 was about a third of the moisture diffusivity in a concrete with w/c = The time of drying was thus much longer at a low w/c, especially when the dimension of the studied layer of concrete was increased, Fig. 18. The moisture permeability of the concrete was dependent on the water-cement ratio. A minimum of 542

11 Persson the moisture permeability was observed at w/c = At lower w/c the moisture permeability increased, Fig. 17, most probably due to the configuration of the isotherm [10]. In a concrete with a low w/c, a very small alteration of the moisture content caused large alterations in the internal relative humidity, Ø. 6.3 Water-cured concrete About 150 measurements were carried out on the internal relative humidity of submerged concrete. Even as close as 50 mm from the water-exposed surface, a continuous decrease of the internal relative humidity, Ø, was observed at a low water-cement ratio, w/c = 0.22, Fig. 11. At w/c = 0.58, Ø remained at even after 450 days of curing, Figs The measured difference to saturation was most probably due to the so-called alkali effect [16]. However, at low w/c the fault of the measured internal relative humidity was of the same size as at w/c = 0.58, since the degree of hydration was also lower [17], Fig. 22. At constant w/c, the decrease of Ø was larger in concrete without silica fume than in concrete with silica fume. Fig. 22 Reduction of measured internal relative humidity in concrete, Ø, due to alkalis in normal-alkali cement compared to low-alkali cement [16-17]. The noticeable decrease of Ø in submerged concrete was partly explained by the self-desiccation, partly by the dense structure of the concrete at low w/c. The value of resistance to water penetration, m, was less than a third at low w/c compared to high w/c, Fig. 20. The low value of m reflected the low capillary porosity in a concrete of low w/c, Fig. 20. The thickness of the studied concrete was also of importance: in a 15 mm concrete layer, the water suction was very rapid, Fig. 19, while the required time became much larger at a thickness of 45 mm, Fig. 21. The time required to absorb half the amount of formerly evap- orated water was more than 10 times longer at w/c = 0.47 than at w/c = 0.22, Fig. 21. Even after 250 days of suction, only 0.9 of the concrete was water-filled in a concrete with w/c = 0.22, while only 14 days was needed at w/c = 0.47 to reach the same saturation of the concrete. Most probably the autogenous shrinkage of the concrete caused a compressive stress in the concrete with a low w/c, especially at the water-cured surface [14]. The concrete became, in a way, prestressed in the surface layer due to the autogenous shrinkage in its interior, which further prevented water transport through the surface layer. The water was rapidly absorbed in 15 mm thick slices of concrete independent of the w/c. Most probably the assumed surface prestressing occurred at low w/c and a thickness of 45 mm, since the time of absorption was substantially prolonged, Fig CONCLUSIONS This article began by presenting the scope of problems related to moisture in concrete. A summary was made of the experimental method, the studied materials, measurement procedure, etc. In the original experimental results presented in this article, the drying and selfdesiccation of 8 concretes were studied for simulated concrete columns over a period of 450 days. The watercement ratio varied between 0.22 and Half of the concretes contained 10% silica fume. More than 450 measurements of 22 h each were carried out. The experiments were supplemented by studies on the capillarity, diffusivity and the permeability of 4 concretes over a period of 750 days. The principal results obtained were: 1) Due to a high value of resistance to water penetration of the low-w/c concrete and also due to a kind of prestressing of the low-w/c concrete in the surface layer, the internal relative humidity, Ø, decreased to around 0.8 close to the surface even when the concrete was submerged. 2) The self-desiccation of the concrete was dependent on time, w/c and content of silica fume. 3) The effect of silica fume on the self-desiccation was time-independent at w/c = At a lower watercement ratio the effect of silica fume was more pronounced at an early age. The opposite effect of silica fume was observed for mature concrete. 4) Due to the very low diffusivity, the internal relative humidity, Ø, decreased only about 0.10 at a distance of 50 mm from the drying surface in concrete with a low w/c, but about 0.20 in concrete with a higher w/c (higher diffusivity). 5) Formulas were devised to describe the development of Ø in the concrete, based mainly on type of curing, age, w/c and content of silica fume. ACKNOWLEDGEMENTS Financial support from the Swedish Council for Technical Development, NUTEK, is gratefully acknowledged. The presentation of this article was possible after 543

12 Materials and Structures/Matériaux et Constructions, Vol. 30, November 1997 financial support from the Norwegian-Swedish Consortium of High-Performance Concrete (Cementa, Elkem, Euroc Beton, NCC, Skanska, Strängbetong, BFR and NUTEK), which is also gratefully acknowledged. I am also most grateful to Professor Dr. Göran Fagerlund at Lund Institute of Technology, Division Building Materials, University of Lund, Sweden, for his critical reviews. REFERENCES [1] Powers, T. C. and Brownyard, T. L., Studies of physical properties of hardened Portland cement paste, Research Laboratories of the Portland Cement Association, Bulletin 22, Journal of the American Concrete Institute, Oct April 1947, Proceedings, Vol. 43. (1947), , [2] Tuutti, K., Corrosion of steel in concrete, The Institute of Cement and Concrete Research, Report Fo 4:82. (CBI, Stockholm, 1982) , [3] Fagerlund, G., The critical degree of saturation method - A general method of estimating the frost resistance of materials and structures, The Swedish Institute of Cement and Concrete Research, Report Fo 12:76. (CBI, Stockholm, 1976). [4] Fagerlund, G., Influence of Environmental factors on the Frost Resistance of Concrete, TVBM-3059, Lund Institute of Technology (Division of Building Materials, Lund, 1994) [5] Nilsson, L-O., Moisture Problems at Concrete Floors, TVBM (Lund Institute of Technology, Division of Building Materials, Lund, 1980) [6] Hedenblad, G., Moisture Permeability of Mature Concrete, Cement Mortar and Cement Paste, Doctoral Thesis, TVBM- 1014, Lund Institute of Technology, Division of Building Materials, (1993) 44, 45, [7] Persson, B., Hydration, structure and strength of High Performance Concrete, Laboratory data and calculations, Report TVBM-7011, Lund Institute of Technology, Division of Building Materials, (1992) , , 211. [8] Hassanzadeh, M., Fracture mechanical properties, Report M4:05 (Lund Institute of Technology, Division of Building Materials, Lund, 1992) [9] ASTM E Standard Practice for Maintaining Constant Relative Humidity by Means of Aqueous Solutions (The American Society for Testing and Materials. Philadelphia, 1985) 33-34, 637. [10] Persson, B., Hydration, structure and strength of High Performance Concrete, Licentiate Thesis, Report TVBM-1009, Lund Institute of Technology, Division of Building Materials, Lund, 1992, , [11] Persson, B. Hydration and Strength of High Performance Concrete, Advanced Cement Based Material (Elsevier Science Inc., New York, 1996) [12] Persson, B. Self-desiccation and its importance in concrete technology, Mater. Struct. 30 (199) (1997), [13] Norling Mjörnell, K., Self-desiccation in concrete, Report P- 94:2, Division of Building Materials, Chalmers University of Technology, Gothenburg, 1993, [14] Persson, K., (Early) basic creep of high-performance concrete, Proceedings of the 4th International Symposium on Utilization of High-strength/High-performance Concrete, Paris, 1996, [15] Persson, B., Self-desiccating High-Strength Concrete Slabs, Proceedings at the Symposium of High-Strength Concrete, Lillehammer, Norway 1993, Edited by Holand and Sellevold, [16] Hedenblad, G. and Janz, M., Effect of alkali on the measured internal relative humidity in the concrete, Report TVBM-3057, Lund Institute of Technology, Division of Building Materials, Lund, 1994, [17] Jonasson, J-E., Modelling of Temperature, Moisture and Stresses in Young Concrete, Doctoral Thesis, Report 153D, Division of Structural Engineering, Department of Civil and Mining Engineering, Luleå University of Technology, Luleå, 1994,