Model to calculate the quantity of water to avoid self-desiccation ...........56 .... to the water in the aggregate within a reasonable time (this limitation can be ... The lightweight aggregate used was pumice sand, which was sieved and divided into ...... 4.75 mm). To obtain a flat and smooth interface with the paste, the aggregates ...
NATIONAL BULDING RESEARCH INSTITUTE
Founded by MINISTRY OF CONSTRUCTION AND HOUSING TECHNION ISRAEL INSTITUTE OF TECHNOLOGY
Faculty of Civil & Environmental Engineering
017-674/3
United States-Israel Binational Science Foundation (BSF)
Autogenous Curing of High-Strength Cementitious Materials by Fine Uniformly Distributed Lightweight Aggregate Water Reservoirs Final Report Principal Investigators: Konstantin Kovler and Arnon Bentur Technion – Israel Institute of Technology David A. Lange University of Illinois, USA Participating Investigators: Dale Bentz National Institute of Standards, USA Klaas van Breugel Delft University of Technology, the Netherlands Pietro Lura Technical University of Denmark Semion Zhutovsky and Andrey Souslikov Technion – Israel Institute of Technology
Copyright © 2004 by K. Kovler, A. Bentur and D. Lange, United States-Israel Binational Science Foundation (BSF), the Technion Research and Development Foundation, Ltd, Haifa and University of Illinois, Urbana-Champaign, USA
Haifa
October 2004
FORWARD The present report summarizes the results of the joint Technion – University of Illinois research project. The research was carried out by close cooperation between the two teams, Israeli team led jointly by Prof. Arnon Bentur and Prof. Konstantin Kovler, and US team led by Prof. David A. Lange. The objective of the research included: Proof of the feasibility of the internal curing approach Identification of the mechanisms and quantification of the parameters which control them Setting the base for technological development of this concept by optimization of the whole process with minimum amount of internal curing agents. The report consists of four parts contributed by the Israeli and US teams, which were streamlined and coordinated to achieve the overall objectives of this study. Feasibility: Part 1, by the Israeli team. Fundamental mechanisms: Part 2 (by the Israeli team) addressing and modeling the mechanisms for the macro-scale point of view and Part 3 (by the US team) based on experimental evaluation of processes on the micro-scale. The two approaches were studied in close cooperation to complement each other. Optimization of the technology, based on the fundamental mechanisms: Part 4 (by the Israeli team), in which different types of lightweight aggregates were evaluated to maximize the internal curing with minimization of the curing agent through control of size and pore structure. Some of the results have already been published, and the references are given at the end of this report.
ACKNOLWEDGEMENTS The principal investigators acknowledge the US-Israel bi-national science foundation for supporting this research. The personal contribution of Mr. Dale Bentz (National Institute of Standards, USA), Prof. Klaas van Breugel (Delft University of Technology, The Netherlands), and Dr. Pietro Lura (Technical University of Denmark), Mr. Semion Zhutovsky (Technion) and Mr. Andrey Souslikov (Technion) is highly appreciated. The authors would like to thank the NIST – National Institute of Standards, USA, for its cooperation to enable the use of equipment and hosting Dr. Pietro Lura.
CONTENTS PART 1: FEASIBILITY OF AUTOGENOUS CURING OF HIGH STRENGTH CEMENTITIOUS MATERIALS .................................................................................3 1. Theoretical Concepts ........................................................................................3 2. Experimental ....................................................................................................5 2.1. Lightweight aggregate for internal curing..................................................5 2.2. Variables studied.......................................................................................6 2.3. Concrete composition................................................................................6 2.4. Testing procedures ....................................................................................7 3. Results..............................................................................................................8 3.1. Free shrinkage of reference mixes .............................................................8 3.2. Effect of the size of lightweight aggregate.................................................9 3.3. Effect of internal curing water content..................................................... 10 3.4. Strength .................................................................................................. 10 4. Discussion ...................................................................................................... 11 4.1. Spacing and efficiency ............................................................................ 11 4.2. Strength .................................................................................................. 12 4.3. Conclusions ............................................................................................ 13 5. References ...................................................................................................... 14 PART 2: MECHANISMS: ASSESSMENT OF WATER MIGRATION DISTANCE IN INTERNAL CURING BY MODELING ............................................................... 15 1. Introduction .................................................................................................... 15 2. Experimental Procedures ................................................................................ 16 2.1. Lightweight aggregates for internal curing .............................................. 16 2.2. Testing procedures .................................................................................. 16 2.3. Specimen Preparation.............................................................................. 17 2.4. Concrete composition.............................................................................. 17 3. Calculation of Paste-Aggregate Proximity....................................................... 18 4. Assessment of the Distance of Water Migration.............................................. 19 4.1. Aggregate content effect ......................................................................... 20 4.2. Grain size effect ...................................................................................... 21 4.3. Cement matrix effect............................................................................... 21 4.4. Effect of aggregate type .......................................................................... 22 5. Conclusions .................................................................................................... 23 6. References ...................................................................................................... 31 PART 3: MECHANISMS: STUDY OF WATER MIGRATION ON THE MICROSCOPIC SCALE ........................................................................................... 33 1. Introduction .................................................................................................... 33 2. Materials......................................................................................................... 34 2.1. LWA....................................................................................................... 34 2.2. Cement paste........................................................................................... 34 3. X-Ray Absorption Test and Chemical Shrinkage ............................................ 34 3.1. Methods .................................................................................................. 34 X-ray absorption................................................................................................. 34 Chemical shrinkage and non-evaporable water content ....................................... 36 3.2. Results .................................................................................................... 37 X-ray absorption................................................................................................. 37 Chemical shrinkage and non-evaporable water content ....................................... 39
3.3. Discussion............................................................................................... 40 3.4. Conclusions ............................................................................................ 41 4. Absorption-Desorption Isotherms of LWA ..................................................... 42 4.1. Methods .................................................................................................. 42 IGAsorp Moisture Sorption Analyser.................................................................. 42 Desiccators ......................................................................................................... 43 4.2. Results .................................................................................................... 43 IGAsorp Moisture Sorption Analyser.................................................................. 43 Desiccators ......................................................................................................... 44 4.3. Discussion and conclusions ..................................................................... 45 5. Low Temperature Calorimetry (LTC) Study of Pumice Aggregates ................ 45 5.1. Methods .................................................................................................. 45 5.2. Results .................................................................................................... 46 5.3. Discussion and conclusions ..................................................................... 46 6. SEM Observation of the Interface between Pumice and Cement Paste ............ 47 6.1. Methods .................................................................................................. 47 6.2. Results .................................................................................................... 48 6.3. Discussion and conclusions ..................................................................... 49 7. Mortars with Different Pumice Content .......................................................... 50 7.1. Materials ................................................................................................. 50 7.2. Methods .................................................................................................. 51 7.3. Results .................................................................................................... 52 7.4. Discussion and conclusions ..................................................................... 54 8. Conclusions .................................................................................................... 56 9. Appendix ........................................................................................................ 56 9.1. Model to calculate the quantity of water to avoid self-desiccation ........... 56 9.2. Some applications ................................................................................... 57 10. References .................................................................................................. 59 Part 4: HOW TO IMPROVE THE EFFICIENCY OF AUTOGENOUS CURING BY FINE UNIFORMLY DISTRIBUTED LIGHTWEIGHT AGGREGATE WATER RESERVOIRS ........................................................................................................... 60 1. Introduction .................................................................................................... 60 2. Experimental .................................................................................................. 61 2.1. Lightweight Aggregates .......................................................................... 61 2.2. Concrete Compositions ........................................................................... 62 2.3. Concrete Preparation and Testing............................................................ 64 3. Results............................................................................................................ 64 4. Discussion ...................................................................................................... 66 5. Conclusions .................................................................................................... 71 6. References ...................................................................................................... 72 TECHNION – UNIVERSITY OF ILLINOIS RESEARCH COOPERATION............ 74 CONCLUSIONS........................................................................................................ 76
PART 1: FEASIBILITY OF AUTOGENOUS CURING OF HIGH STRENGTH CEMENTITIOUS MATERIALS
Self-desiccation is a phenomenon inherent to high strength concretes, HSC, with low water/binder, w/b, matrix. As a consequence, HSC exhibits substantial autogenous shrinkage, which can lead to early-age cracking in restrained components. In practice, components are restrained, to one degree or another. An effective strategy to overcome this problem is the use of pre-soaked lightweight aggregates as internal water reservoirs. This concept was explored by several investigators [1-7] and was assessed quantitatively with respect to strength development and elimination of autogenous shrinkage and the resulting stresses under restrained conditions. In these studies the primary emphasis was placed upon an investigation of the effects of the replacement level of normal weight coarse aggregates by saturated lightweight ones, and the degree of water saturation of the lightweight aggregate. These parameters provide the means to control the effectiveness of autogenous curing.
1. Theoretical Concepts It is well known that self-desiccation is induced by the emptying of pores due to chemical shrinkage of the hydrated water. Hence, the amount of water required in the internal reservoirs of the lightweight aggregates to completely eliminate self desiccation can be calculated from chemical shrinkage as follows [8]: Wcur
C
max
CS
where: Wcur - water content (kg/m3); C - cement content (kg/m3); max -maximum degree of hydration; CS - chemical shrinkage (kg water/kg cement hydrated). Studies reported in recent years suggested that the amount of water within the aggregates which was required for the elimination of self desiccation and autogenous shrinkage was considerably higher than that calculated from eq.1. Calculations of the content of internal water curing based on eq. 1 provide values in the range of 18 to 23 kg/m3 for the systems reported in references [3, 4, 5, 6, 7]. However, in none of them were this content sufficient and the levels required were at least 30 to 40 kg/m3 or more (Table 1).
(1)
K. Takada et al
0.37
A. Bentur et al 0.33 P. Schwesinger 0.28et al 0.32 K. van Breugel 0.34,0.39 et al P. Lura et al 0.37
13.0
Successful autogenous shrinkage elimination Concrete still exhibits marked autogenous shrinkage
Calculated value
Internal curing water content, kg/m3
Replacement level, %
Degree of aggregate saturation (S)
Water/binder ratio (w/b)
Authors
Aggregate water absorption by weight ( )
Table1: Calculated internal curing water required to eliminate autogenous shrinkage derived from data in other studies
10, 17.5, 25, 100
42, 70
11.0 13.0
0.2, 0.6, 1 0.5, 1 0.5
7, 12, 14, 17 19-23
25, 100 31, 49, 54
24, 64 31, 45, 70
13, 33 -
18-19 19-20
14.0
0.5, 1
25, 100
38, 76
22
22-23
15.2
1
100
83
-
19-23
The explanation for this apparent discrepancy is that not all the water in the aggregates can become effective to counteract self desiccation. Several factors can be considered, such as: (i) Aggregate pore size: if it is very fine water may not migrate readily into the surrounding paste, and (ii) The spacing between the aggregate particles: if it is too large the paste surrounding the aggregates may not be accessible to the water in the aggregate within a reasonable time (this limitation can be quantified in terms of the diffusivity of the water into the surrounding matrix which becomes denser with time). These influences may be expressed in a simplified engineering approach in terms of an efficiency term, , which is a factor in the range of 0 to 1, describing the portion of water in the aggregate which can become available for internal curing. In the view of the above, this factor is the result of a complex function and is not only dependent on the properties of the aggregate. Accordingly, the content of lightweight aggregate, LWA, in units of kg per m3 of concrete required to eliminate self desiccation by the internal water in the aggregates can be calculated: LWA
Wcur S
where: - aggregate water absorption by weight (kg water per kg of dry aggregate); S - degree of saturation of aggregate; - efficiency factor (i.e. the fraction of water absorbed in a saturated aggregate that can become effective to counteract self desiccation). Ideally, one would like to develop aggregates for internal curing where the efficiency factor, , is 1, and the water absorption is as high as possible, to minimize the aggregate content required to obtain effective internal curing. To achieve this, one would resort to aggregates of a small size and a pore structure which is large and coarse. Small size of particles would minimize the distance between the reservoirs, making the paste volume more accessible to the water in the internal reservoirs. Coarse pore structure of the aggregate is beneficial to make the water more readily
(2)
dischargeable into the surrounding paste. A large pore volume in the aggregate may have two opposing influences: it may enable to minimize the lightweight aggregate content, but at the same time it will increase the spacing between the aggregate. Thus, from the point of view of the effectiveness of the aggregate, there may be an optimal pore volume for a given aggregate size. The object of the present study was to explore the possibility to obtain lightweight aggregate having maximum internal curing efficiency (i.e. =1), that could effectively eliminate autogenous shrinkage, but with an aggregate content which is as small as possible, without detrimental effects on strength. in view of the considerations outlined above the size range explored was that of fine aggregates, smaller than 5 mm.
2. Experimental 2.1.
Lightweight aggregate for internal curing
The lightweight aggregate used was pumice sand, which was sieved and divided into three fractions: 0.15 mm4.75 mm
25
20
15
10
5
0 80
82
84
86
88
90
92
94
96
98
100
RH [%]
Figure 9. Desorption of pumice fractions at 25 ºC from vacuum-saturated conditions.
1
Absorption/max absorption
0.6-1.18 mm 0.9
1.18-2.36 mm
0.8
2.36-4.75 mm >4.75 mm
0.7
0.6
0.5
0.4
0.3
0.2 80
82
84
86
88
90
92
94
96
98
100
RH [%]
Figure 10. Relative desorption of pumice fractions at 25 ºC from vacuum-saturated conditions.
4.3.
Discussion and conclusions
In Figure 8, notice the hysteresis at lower RHs and the sharp increase of water content around 90%. The water content of the pumice depends on its moisture history. The water content at 95% is very low, below 1%. The behavior of the two aggregates is different, with the bigger one having a higher absorption at almost all RH levels. From the absorption and desorption curves, it is possible to calculate a pore size distribution of the LWAs (see, for example [15]). In Figure 9, notice that the bigger the fraction, the higher the absorption. This is in accordance with previous results on the same pumice fractions, which were saturated in boiling water [16]. The difference in this case is even more pronounced, with the biggest fraction absorbing 38% by weight and the smallest only 13%. A possible explanation of this fact, confirmed also from direct SEM observation [17], is that, because the particles are obtained by crushing, the particles break apart along the largest pores [16, 17]. Therefore, the largest particles have higher porosity and bigger pores. The lower density of the larger particles (see Part 2.2) is further evidence of this. In Figure 10, the smallest fraction loses about 80% of the absorbed water at 84% RH, while the two largest lose only about 50%. One would expect that the largest aggregates, having supposedly the largest pores, would lose proportionately more water than the smaller ones. This apparent paradox could be due to the presence of inkbottle pores: the emptying of the larger pores would not occur until the RH dropped beyond the equilibrium value for the smaller entryways. Since a great quantity of water is in fact entrapped in the internal porosity of the larger LWAs, one should consider that only about half of it will be available for internal curing. In the case of the smaller fraction the opposite seems to hold: the absorption is very low, but almost 80% of the water is lost by 85% RH. A further comment on these studies is that even if total saturation of the porosity is useful for research purposes and vacuum saturation is a very efficient way to obtain it, it is not feasible in practice. The same is true for boiling the LWAs or immersing them in water for months [16]. The “ideal” LWA is the one whose porosity is almost filled in about one day underwater, a time that might be feasible for practical use.
5. Low Temperature Calorimetry (LTC) Study of Pumice Aggregates 5.1.
Methods
LTC is a technique that can be used for probing the pore structure of porous building materials. The advantage of LTC over many other techniques is that the specimen can be evaluated in the saturated state, with no pre-drying required. Pore diameters from approximately 4 m to 30 nm are detectable using LTC. Because water in a small pore freezes at a lower temperature than bulk water, the observed freezing-point depression can be related to the size of the pore. By monitoring the heat
absorbed/released as a function of temperature in a differential scanning calorimeter (DSC), the volume of water frozen vs. temperature (or pore size), basically equivalent to a pore size distribution, can be determined. In the case of the pumice aggregates, performing LTC on both dry (in lab conditions ~50% RH) and water-saturated aggregates, can give information on the size of pores in which the water is entrained. one dry and one wet aggregate, with dry weight around 40 mg, were tested in a Differential Scanning Calorimeter. The samples were first equilibrated at a temperature of 10 ºC and then the temperature was lowered to -60 ºC at a rate of -0.5 ºC per minute. The heat flow from the sample was registered continuously.
5.2.
Results
0.006
Pumice dry Pumice saturated 0.005
Heat flow [W/g]
0.004
0.003
0.002
0.001
0 -60
-50
-40
-0.001
-30
-20
-10
0
T [oC]
Figure 11. Freezable water, evident as heat flow from the sample, for dry and vacuum saturated pumice samples. In Figure 11 the freezable water, evident as heat flow from the sample, is shown for a dry and for a saturated pumice sample. Only the portion of the curve below 0°C is shown.
5.3.
Discussion and conclusions
According to Figure 11, most of the water in the saturated LWA froze at about -8ºC, which corresponds to pores with radii around 170 nm [18]. For a comparison, capillary water in 6 h old cement paste with a w/c ratio of 0.3 [19] froze at temperatures around –17ºC, corresponding to pores with radii around 70 nm. A very small peak is visible at -40ºC, corresponding to pore sizes around 50 nm. On the other hand, even in the dry sample some water is actually present and it freezes at the same temperature as the second peak in the saturated specimen. In fact, the peak of the dry specimen is much higher than the one of the saturated one.
An explanation of this fact is once again the presence of inkbottle pores. Since ice formation proceeds from the outer surface of the specimen, and in the saturated specimen all pores are full, a path through larger pores connects the outside to the inside and almost all pores freeze at the same temperature. In the dry specimen, however, water in some large inner pores is connected to the outer surface only through smaller entries, and freezes at lower temperatures. Therefore, the peak seen at -40ºC in the dry specimen actually also includes water in pores with radii larger than 50 nm.
6. SEM Observation of the Interface between Pumice and Cement Paste 6.1.
Methods
A prismatic specimen, with cross-section 10 mm x10 mm, was cut from a sample that had been used in one trial run of the x-ray test. It contained a single LWA taken from the fraction > 4.75 mm and cement paste. The pumice had not been vacuum-saturated but only immersed in water for one day. A layer of several mm of cement paste totally covered the pumice. According to calculations (see Appendix), the water contained in the pumice was in this case insufficient to compensate for the self-desiccation of the paste. At an age of 2 weeks the specimen was dried in the oven at 105 °C. The specimen was vacuum impregnated with a low-viscosity epoxy resin and cured for one day at 60 C. The surface was cut with a diamond saw to expose the crossection of the LWA. Saw marks were removed by grinding with 400 grit followed by 600 grit sandpaper. Final polishing was done on a lap wheel with (6, 3, 1, and 0.25) m diamond paste for 45 s each. After each polishing, the specimen was cleaned using a clean cloth. The final polished specimen was coated with carbon to provide a conductive surface for viewing in the SEM. once properly prepared, the specimen was placed in the SEM viewing chamber, and signals were collected for the backscattered electrons and X-rays. Typical accelerating voltage and probe current for the backscattered electron images were 12 kV and 2 nA, respectively. For the X-ray images, the probe current was increased to about 10 nA. The contrast in the BE image is dependent on the average atomic number (Z), with higher Z phases appearing brighter than lower Z phases. Thus, for hardened cement paste, anhydrous cement appears brightest followed by calcium hydroxide, and calcium silicate hydrate; porosity filled with embedding resin appears dark. The pumice grain, composed mainly of alumino-silicates, is a gray shade similar to the calcium silicate hydrate. Phase identification is made by examination of the phase shape, relative brightness, and chemical composition as determined by qualitative Xray microanalysis. With the fairly distinct gray-level separations between anhydrous cement, CH, C-S-H, and the resin-filled porosity, the BE image could be processed and analyzed to obtain quantitative analysis data.
The SEM observation concentrated on the interface between the pumice grain and the cement paste. At two different points along the rim of the LWA, which was quite straight, a series of 4 images (~160 µm x 200 µm) was taken, proceeding from the aggregate towards the bulk cement paste. The gray levels corresponding to each phase were determined on one picture and then used to analyze all the other pictures. Phase fractions were measured as a function of distance from the aggregate surface for both sets of pictures. The degree of hydration as a function of the distance from the interface was calculated according to: CSH CH 2.15 (1) CSH CH UnhydratedCement 2.15 The results of the two sets of pictures were then averaged.
6.2.
Results
In Figure 12 a BE image of the interface between the aggregate and the cement paste is shown. The dark gray porous region at the top is the rim of the LWA. Notice the high porosity region at the interface and the absence of outer-product C-S-H gel. in Figure 13, porosity, anhydrous cement, and degree of hydration are plotted as a function of the distance from the interface. The values represent the average of two sets of BE images.
Figure 12. BE image of the interface between pumice aggregate and cement paste (160 µ x 200µm).
! 0.8
Porosity Anhydrous cement Degree of hydration
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
100
200
300
400
500
600 700 800 Distance from interface [ m]
Figure 13. Porosity, anhydrous cement, and degree of hydration as a function of the distance from the interface (average of two sets of BE images).
6.3.
Discussion and conclusions
Both the BE image (Fig. 12) and the plots in Fig. 13 show that there is a region with very high porosity at the interface with the LWA. This was observed before in the case of saturated LWAs [20]. A possible explanation is that water and air came out of the LWA, creating a high porosity zone at its outer rim. In other points along the interface (not shown in the picture) large circular air voids (~50 µm diameter) were present right at the interface. The air might have escaped from the LWA that was certainly not saturated after one day under water (see also [16]). In Fig. 12, it is also evident that the degree of hydration at the interface is very low and that a lot of unhydrated cement is present, especially if one considers that very little cement is present there at all. One would expect the degree of hydration of the cement in this region to be very high, because of the high local w/c ratio and the supply of water from the pumice. In fact, the low degree of hydration measured at the interface could be explained by water having been pulled out from the interface porosity. This explanation is confirmed by two observations: 1) From 200 µm to 350 µm from the interface, all around the LWA a denser region of paste is observed, where the degree of hydration is higher than average and very little unhydrated cement is found (Fig. 13). This region, that had dense initial packing because of being far from the interface, has probably taken advantage of the water coming from the LWA to reach a higher degree of hydration. 2) The pores observed at the interface are larger than the ones in the paste, and the largest pores are emptied first in a self-desiccating system (see Chapter 3). Thus, a possible explanation of what observed is that, because the water in the LWA is insufficient to cure the whole cement paste, water was sucked into the cement paste not only from the aggregate but also from the interface. A rim around the pumice that had a high porosity and little hydrated cement was created this way. The effect of the
" curing water was seen in a small area from 200 µm until about 350 µm from the LWA. Probably, the increased hydration in this area would not compensate, in the global properties of the composite, the fact that a very bad interface was created. The principal reason for this is the insufficient curing water provided to the system. Some other considerations emerging from the image analysis are: Evidence of penetration of cement paste into the outer pores of the LWA was found. The cement grains, though, were very little hydrated. Possible evidence of alkali-silica reaction at the interface of the LWA was found. X-ray analysis of the image revealed higher counts of potassium and calcium in zones where the LWA was lighter. This observation should be confirmed by analyzing images of the untreated LWAs to check whether this effect might be instead due to degradation of the pumice exposed to weathering.
7. Mortars with Different Pumice Content 7.1.
Materials
Three mortar mixtures with the same cement paste but different contents of saturated pumice were mixed: reference mix without LWA, ~4%, and ~8% LWAs by volume. Mix compositions are shown in Table 1. The two LWA mixes were obtained by replacing a fraction of the normal weight aggregates with the same volume of saturated pumice. The pumice was vacuum saturated according to the procedure described in Par.4.1 and contained 38% water by weight. In this case the pumice was not in surface dried conditions, but excess water was removed mechanically. The cement paste used was as described in Par. 2.2. About 1.5 l of each mortar were mixed in a Hobart mixer. The water was inserted first into the mixer, then the cement was added and finally the aggregate, ending with the pumice fraction. Total mixing time was approximately 5 minutes. Table 1. Mixture compositions of mortars with w/c ratio 0.3 Mixture composition Ref LWA1 LWA2 3 Cement 135 kg/m 775 775 775 3 Water kg/m 225 225 225 3 Sand 0-4 mm kg/m 1409 1281 1171 Pumice 1.18-2.36 mm (dry) kg/m3 --54 108 3 Superplasticizer WRDA-19 (GRACE) kg/m 9 9 9 3 Specific weight of the mix kg/m 2404 2345 2289 Water in the LWA kg/m3 --21 42 w/c ratio (without entrained water) 0.3 0.3 0.3 w/c ratio (entrained water included) 0.3 0.327 0.353
7.2.
Methods Autogenous deformation
The cement paste was cast into corrugated, tight, low-density polyethylene plastic moulds and vibrated on a vibrating table. Length of the samples was approximately 300 mm. The moulds were specially designed to minimize restraint on the paste and were watertight [21]. with the CT1 digital dilatometer, linear measurements of autogenous deformation of the cement paste from the moment of setting could be made. The device consisted of a measuring bench of stainless steel in which the specimen was longitudinally supported by two parallel cylindrical 20 mm-diameter rods. A Mitutoyo ID-C digital gauge recorded length changes to the nearest 0.001 mm. The specimens were kept in a room with a temperature of 23±1 °C. They were supported by special racks during hardening and manually transferred to the dilatometer only for measuring. Measurements started about 4 hours after casting.
Compressive strength Compressive strength was measured on 2-inch cubes at 3 days, 7 days, and 28 days. For each mixture, age and curing condition, two cubes were tested. Cubes were cast into steel molds and demolded after 1 day. The cubes were kept at 23±1 °C. After demolding, the two LWA mixes were cured in sealed conditions; for the reference mix two different curing regimes, sealed and underwater, were employed.
Non-evaporable water content Pieces of mortar were collected from the inner core of the 2 inch-cubes crushed in the compressive strength test. Non-evaporable water was measured for all mixes and curing conditions at 3 days, 7 days, and 28 days of age. The procedure for measuring the non-evaporable water content was the same as the one used for cement paste (see Par 3.1). A further correction had to be made for the loss on ignition of the normal sand and the pumice, which had also to be measured. About 9-10 g of crushed mortar was weighed in each crucible, much more than for cement paste, to ensure homogeneity.
Modeling of internal curing with HCSS The mixtures were simulated with the Hard-Core Soft-Shell (HCSS) [22] model to calculate the fraction of paste within a certain distance from the LWAs. The input of the model is the particle size distribution of the NWAs and of the LWAs. Assuming a transport distance of the curing water from the saturated LWAs (the same for all LWAs), the corresponding volume of cured paste is calculated. The HCSS model is
more accurate than the equations developed by Lu & Torquato [23, 24], since it takes into account the presence of the NWAs.
7.3.
Results 50
0
Autogenous deformation [ strain]
0
1
2
3
4
5
6
7
-50
Cem 135, w/c ratio 0.3 o Curing temperature: 23 C
-100
Reference (no LWA)
-150
LWA 1 (54kg/mc) LWA 2 (108kg/mc)
-200
-250
-300
Age [days]
Figure 14. Autogenous deformation of mortar mixes (see Table 1). In Figure 14, the autogenous deformations of the mortar mixes, measured from the time or setting, are plotted as a function of age. In Figure 15, the cube compressive strength of the mixes is plotted as a function of age. In Figure 16, the development of the degree of hydration, calculated from the non-evaporable water content, is plotted.
60 59 58 57 56
Strength [MPa]
55 54 53
Cem 135, w/c ratio 0.3 Curing temperature: 23oC
52 51 50
Reference SEALED
49
Reference UNDERWATER
48
LWA 1 (54kg/mc)
47
LWA 2 (108kg/mc)
46 45 0
7
14
21
28
Age [days]
Figure 15. Compressive strength of mortar mixes (see Table 1) measured on 2 inchcubes. The average and the standard deviation of two cubes are indicated.
Degree of hydration (from non evap. water content)
0.7
0.6
0.5
0.4
Cem 135, w/c ratio 0.3 o Curing temperature: 23 C 0.3 Reference SEALED 0.2
Reference UNDERWATER LWA 1 (54kg/mc)
0.1
LWA 2 (108kg/mc) 0 0
7
14
21
28
Age [days]
Figure 16. Degree of hydration of mortar mixes (see Table 1) calculated on the basis of non-evaporable water measurements. The average value of two samples cubes is indicated. In Figures 17-19, some results of the numerical simulation of the two LWA mixtures with the HCSS model are presented. Figure 17 shows the fraction of paste within a certain distance from the rim of the LWAs. In Figure 18, 3D images of the two LWA mixtures are shown, where the blue color indicates the internally cured volume. Finally, in Figure 19, 2D slices cut from the 3D images are shown.
1 0.9 0.8
Fraction of paste
0.7 0.6 0.5 0.4 0.3 LWA 1 (54kg/mc) 0.2 LWA 2 (108kg/mc)
0.1 0 0
500
1000
1500
2000
Distance [ m]
Figure 17. Fraction of paste within a distance from the LWAs, calculated with the HCSS model.
Figure 18. 3D view of the LWA mixtures 1 (left) and 2 (right) simulated by the HCSS model. NWAs are red, LWAs are yellow, internally cured paste is shades of blue and self-desiccating paste is white.
Figure 19. 2D slice of the HCSS model of the LWA mixtures 1 (left) and 2 (right). NWAs are red, LWAs are yellow. Lighter shades of blue indicate the regions of the paste progressively more distant from the LWAs. Self-desiccating paste is white.
7.4.
Discussion and conclusions
The two LWA mixtures studied were designed having in mind the theoretical quantity of entrained water needed to avoid self-desiccation (see Appendix). LWA2 contained a sufficient quantity to avoid it, LWA1 about half of that. Since internal RH change was not measured on these mixtures, no direct proof of the success of this approach is available. Nevertheless, both the available experimental results and the numerical simulation show a clear picture. Autogenous shrinkage of mixture LWA2 was about 1/4 of the reference and 1/3 of mixture LWA1 (Figure 14). Compressive strength of the two LWA mixtures was higher than for the reference mixtures and LWA2 had the highest strength (Figure 15). The same is true for the degree of hydration, where the two mixtures with LWAs
performed better than the cubes cured underwater, clearly showing the advantages of internal curing (Figure 16). All these data confirm that the water entrained in the LWAs was available for transport to the cement paste, or at least some part of it. The shrinkage of mixture LWA1 also confirms that the water in the LWAs was insufficient to avoid self-desiccation, as was calculated (see Appendix). It must be noticed that most of the shrinkage occurred in the first 3 days, and took place after an initial expansion. Also mixture LWA2 showed some (very minor) shrinkage, which also occurred mostly in the first 3 days. The causes of this fact could be: 1) The quantity of entrained water was insufficient to totally compensate for selfdesiccation of the paste. This means that the amount calculated in the appendix was underestimated (which is not strange because the coefficients used should vary according to the type of cement). However, there is good agreement between the final calculated value and the measured chemical shrinkage of the cement paste (see Figure 5 and the Appendix). 2) The transport distance the entrained water had to cover was too long (see Fig. 17). However, it was observed that shrinkage occurred mostly in the first 3 days, when no depercolation of the capillary pores had taken place and transport of water should be very rapid. This indication is also confirmed by the x-ray experiments (see Chapter 3). 3) Some of the entrained water (and some water present on the surface of the LWAs) might have come out during mixing. According to the calculations in the Appendix, at low w/c ratio (below 0.38) an increase of the initial w/c ratio leads to a higher amount of entrained water needed to totally avoid selfdesiccation. 4) In the first days, when shrinkage occurred, the RH drop in the cement paste was very small and possibly not sufficient to pull the entrained water out of the LWAs. It was shown in Chapter 4 that a relatively high amount of entrained water is remaining in the LWAs at high RH. The same situation might have occurred in the LWA mortars in the first days: The paste was selfdesiccating but the capillary depression in the pore water was not sufficient to pull out the water from the LWAs. 5) Another possibility is that the force that drove water from the LWAs was very small, so that the transport was not quick enough to avoid some selfdesiccation of the paste. It is pointed out that most of the curing water for the cement paste is needed in the first days, since most of the chemical shrinkage occurs at this stage (Fig. 5). Nevertheless, the degree of hydration of the LWA mortars (Figure 16) was higher than the one of the reference at all ages, indicating that some internal curing was occurring at least by the third day. Simulation results of the HCSS model (Figures 17-19) show that, since the low percentage of LWAs in the mixes (4% and 8% by volume) a transport distance of 2 mm was needed to cure the whole cement paste in mix LWA2, while it cured only 70% in mix LWA1 (Fig.17). Since most of the autogenous deformation takes place in the first 3 days, transport of water is not hampered by depercolation of the cement paste. Nevertheless, the transport through the capillary pores might be very slow, since the driving forces are very small. As a conclusion of the experiments and numerical simulations on mortars, it was observed that mortars with improved strength, enhanced degree of hydration and
reduced autogenous shrinkage could be obtained by addition of saturated pumice aggregates. Internal curing of 2-inch cubes was, for strength and degree of hydration, more efficient than curing underwater. The LWA mortar with higher pumice and entrained water content (calculated in order to avoid self-desiccation) had slightly higher strength, higher degree of hydration and much less shrinkage than the mortar with half the amount of LWAs. Most of the shrinkage occurred in the first 3 days, when depercolation of the capillary porosity is not an issue.
8. Conclusions X-ray absorption showed that most of the transport of water from saturated LWAs to hydrating cement paste occurs in the first days after casting. This is also the period when the greatest part of the chemical shrinkage takes place. On the other hand, also most of the autogenous shrinkage is concentrated in the first days. In this period, transport of water within cement paste should be a relatively fast process (as shown indeed by the x-ray study), since the capillary pores are not depercolated yet. As a consequence, to avoid self-desiccation shrinkage in the first days after casting, entraining a large amount of water in the LWAs is more important than having the LWAs well dispersed in the mix, with very short transport distances. An additional issue that emerged in this study is the importance of the pore structure of the pumice in the internal curing process. The different pumice fractions show different porosity, absorption, and desorption. The smaller fractions have lower absorption, but they release a greater percentage of the absorbed water at the RH of practical interest. Absorption-desorption isotherms and LTC indicated the presence of inkbottle pores in the pumice aggregates. To design a LWA mixture, one should also take into account the amount of entrained water that will actually remain inside the LWAs, essentially useless for internal curing. Finally, the study of an interface between pumice and cement paste showed a worstcase scenario: Due to insufficient water entrained into the pumice, the hydrating cement paste pulled water out from the interface, preventing further hydration and leaving a very porous and weak region around the LWA.
9. Appendix 9.1.
Model to calculate the quantity of water to avoid self-desiccation
The amount of entrained water needed to avoid self-desiccation equals the chemical shrinkage in saturated conditions. For Portland cement pastes, using Powers’ model, one obtains [25]: w c
0.18 e
w c
w c
0.38
(2)
For a w/c ratio of 0.3 this means w/ce=0.054. According to the CEMHYD3D simulation (Fig.5), the final value of chemical shrinkage in saturated conditions for
the cement paste considered is 0.054206 ml/gcem. Since this is exactly the volume that needs to be filled by entrained water, the agreement of the two methods is almost perfect. However, if one instead considers the chemical shrinkage measurements (Figure 5), which diverge from the simulations due to depercolation of the capillary porosity, one would conclude that only about 0.043 ml/gcem are needed. It is noticed that using different values for the parameters of Powers’ model (non-evaporable water, gel water, and chemical shrinkage) leads to a different coefficient in eq.2.
9.2.
Some applications
X-ray experiment In the x-ray experiment (see Chapter 3), the amount of water in the pumice was first measured. Then the maximum volume of cement paste that would be internally cured was calculated on the basis of eq. 2 and of the w/c ratio of the cement paste. This corresponded to a thickness of 6 mm above the pumice. However, due to imprecision in the sample configuration, the thickness of the cement paste turned out to be about 4.5 mm.
LWA mortars The amount of water entrained in mortar LWA2 (see Chapter 7) was calculated with eq. 2. Mortar LWA1 had half of the pumice and therefore half of this amount.
An extension to silica-fume modified cement pastes The addition of silica fume to a Portland cement paste can be taken into account modifying the basic set of equations of Powers’ model [25, 26]. On the basis of these equations, the quantity of entrained water to avoid self-desiccation is calculated as described in [25].
maximum [-] (saturated)
1
0.9
0.8
Plain
0.7
3% silica fume 5% silica fume 10% silica fume
0.6
15% silica fume 20% silica fume
0.5
0.4 0.2
0.22
0.24
0.26
0.28
0.3
0.32
0.34
0.36
0.38
w/c ratio [-]
Figure 20. Maximum degree of hydration in saturated conditions of pastes with different w/c ratios and silica fume addition. 0.1
w/c entrained [-]
0.09
0.08
0.07
0.06 Plain
0.05
3% silica fume 5% silica fume 10% silica fume
0.04
15% silica fume 20% silica fume
0.03 0.2
0.22
0.24
0.26
0.28
0.3
0.32
0.34
0.36
0.38
w/c ratio [-]
Figure 21. Amount of entrained water to avoid self-desiccation in pastes with different w/c ratios and silica fume addition. Both the final degree of hydration in saturated condition of the pastes and the quantity of required entrained water is changed by the addition of silica-fume. In Figures 20 and 21, the change in final degree of hydration and the amount of entrained water needed to avoid self-desiccation is shown for different w/c ratios and silica fume additions. It is noticed that, according to the model, maximum degree of hydration is decreased by the addition of silica fume at each w/c ratio. This is due to the fact that a part of the capillary water is adsorbed on the surface of the pozzolanic C-S-H gel and becomes unavailable for hydration of cement.
! However, since the pozzolanic reaction produces a great chemical shrinkage (about 3 times more than hydration of Portland cement), the quantity of necessary entrained water is strongly increased by the addition of silica fume, as shown in Figure 21.
10. References [1] Final Report, Portland Cement Proficiency Samples Number 135 and 136, Cement and Concrete Reference Laboratory at NIST, 2000. [2] D.P. Bentz, X. Feng, C.J. Haecker, P. Stutzman, Analysis of CCRL Proficiency Cements 135 and 136 Using CEMHYD3D, NIST Internal Report 6545 (2000). [3] GNI X-ray System, http://www.gni.dk/. [4] K.K. Hansen, S.K. Jensen, L. Gerward and K. Singh, Dual-energy X-ray absorptiometry for the simultaneous determination of density and moisture content in porous structural materials. Proc. 5th Symp. Build. Phys. Nordic Countries, Gothenburg, Sweden (1999). [5] D.P. Bentz and K.K. Hansen, Preliminary observations of water movement in cement pastes during curing using X-ray absorption. Cem. Concr. Res. 30 (2000), pp. 1157-1168. [6] D.P. Bentz, M.R. Geiker, K.K. Hansen, Shrinkage-reducing admixtures and earlyage desiccation in cement pastes and mortars, Cem. Concr. Res. 31(7) (2001) 10751085. [7] D.P. Bentz, Influence of Curing Conditions on Water Loss and Hydration in Cement Pastes with and without Fly Ash Substitution, NIST Internal Report 6886 (2002). [8] D.P. Bentz, K.K. Hansen, H.D. Madsen, F. Vallee, E.J. Griesel, Drying/hydration in cement pastes during curing, Mater Struct 34(243) (2001) 557-565. [9] T. Østergaard, Report of work at NIST or GNI news http://www.gni.dk/. [10] Proposed ASTM Test Method for Water Imbibition by Hydrating Cement Pastes [11] Technical Note VCCTL-01: Estimation of the Degree of Hydration of Portland Cement by Determination of the Non-Evaporable Water Content. [12] D.P. Bentz, CEMHYD3D: A Three-Dimensional Cement Hydration and Microstructure development Modelling Package. Version 2.0, NIST Internal Report 6485 (2000). [13] M.R. Geiker, Studies of Portland Cement Hydration, PhD thesis, Institute of Mineral Industry, Technical University of Denmark, 1983. [14] ASTM C128-97, Standard Test Method for Specific Gravity and Absorption of Fine Aggregate. [15] L.F. Nielsen, A research note on sorption, pore size distribution, and shrinkage of porous materials, Building Materials Laboratory, The Technical University of Denmark, Lyngby, Denmark, 1991, TR 245/91. [16] Autogenous Curing of High Strength Cementitious Materials, United StatesIsrael Binational Science Foundation (BSF), Report of First Year Study, 2002, Technion, Israel. [17] S. Zhutovsky, K. Kovler, A. Bentur, Influence of cement paste matrix properties on the autogenous curing of high-performance concrete, submitted to Cem. Concr. Comp., 2002. [18] G. Fagerlund, Determination of pore-size distribution from freezing-point depression, Materiaux et Constructions, 6(33) (1973). [19] K. Snyder & D.P. Bentz, Early age cement paste hydration at 90% relative humidity and the loss of freezable water, unpublished results.
" [20] S. Helland, M. Maage, Strength loss in Un-remixed LWA Concrete, Proc. Int. Symp. on Utilization of High Strength Concrete, Lillehammer, June 1993. [21] O.M. Jensen, P.F. Hansen, A dilatometer for measuring autogenous deformation in hardening cement paste, Mater Struct 28(181) (1995) 406-409. [22] D.P. Bentz, E.J. Garboczi, K. A. Snyder, A hard core/soft shell microstructural model for studying percolation and transport in three-dimensional composite media, NIST Internal Report 6265 (1999). [23] B.L. Lu, S. Torquato, Nearest-surface distribution-functions for polydispersed particle-systems, Physical Review A 45(8) (1992) 5530-5544. [24] D.P. Bentz, K.A. Snyder, Protected paste volume in concrete: extension to internal curing using saturated lightweight fine aggregate, Cem. Concr. Res. 29(11) (1999) 1863-1867. [25] O.M. Jensen, P.F. Hansen, Water-entrained cement-based materials. I. Principles and theoretical background, Cem. Concr. Res. 31(5) (2001) 647-654. [26] P. Lura, O.M. Jensen, K. van Breugel, Autogenous shrinkage in highperformance cement paste: An evaluation of basic mechanisms, Cem. Concr. Res. 33(2) (2003) 223-232.
Part 4: HOW TO IMPROVE THE EFFICIENCY OF AUTOGENOUS CURING BY FINE UNIFORMLY DISTRIBUTED LIGHTWEIGHT AGGREGATE WATER RESERVOIRS 1. Introduction The present chapter is intended to develop the approach of autogenous curing further, by optimizing the porosity and size of the aggregates to enable successful internal curing with only a small amount of aggregates, which could be viewed as additives rather than bulk replacement of conventional aggregates. The approach taken was to increase the porosity of the aggregates and reduce their size, to obtain a system with numerous internal reservoirs of sufficiently small spacing to allow water to be readily discharged from the aggregate and transported over the whole range of the matrix. It was shown that aggregates with porosities of about 50% by volume, size of few millimeters and contents of less than 50 kg/m3 could provide full elimination of autogenous shrinkage in concretes having w/cm as low as 0.25, with only a small effect on strength. The parameters controlling the efficiency of the aggregates were assessed, indicating that their pore structure is the most important one, and that water from within the aggregates can be readily transported into the matrix to a distance of few millimeters. The present chapter is a continuation of the previous work, in which a wider range of aggregates was studied (particularly aggregates of higher porosities), with the objective of reducing the required content to below 50 kg/m3 without affecting strength.
This chapter describes the results of the 3rd year study, addressing the feasibility of the approach and discussing the major parameters which control the efficiency of the aggregates, in particularly their pore structure and size. The notion behind the approach presented here is that the aggregates could be viewed as internal reservoirs which need to be distributed over the concrete volume with the spacing between them being sufficiently small. If this condition is met, than most of the water within the aggregates could penetrate into the paste volume between the aggregates, to eliminate self desiccation and the resulting autogenous shrinkage. This concept requires that the aggregates be sufficiently small, in the range of sand particles. Bentz and Snyder (1999) modeled such a system and addressed the aggregates from the point of view of providing a protective envelope to the paste matrix around them. They assumed that this envelope could range over several of hundreds of m, and on that basis calculated the content of fine lightweight aggregates which need to replace the sand fraction in the concrete.
2. Experimental 2.1.
Lightweight Aggregates
The lightweight aggregates (LWA) used in the present study were pumice obtained from two different locations: Greece (Yali) and Iceland (Hekla). The properties and behavior of the Yali pumice was reported in the paper (Zhutovsky et al 2003a). It was denser than the Hekla pumice which was evaluated in the present study. The Hekla aggregate consisted of two sand fractions, one which is more uniform in size and coarser (Hekla/C) and the other of continuous grading and overall finer (Hekla/S). Each of the Yali and Hekla sands was sieved to obtain three fractions of uniform size: 0.6-1.18 mm (fraction 0), 1.18-2.36 mm (fraction 1) and 2.36-4.75 mm (fraction 2). The bulk densities and the total water absorption by volume of the different aggregates are presented in Table 1. Total absorption was obtained by immersing the aggregates in boiling water, to assure better penetration of water within a short time. The total absorption by volume can be considered as a reasonable estimate for the total porosity of the aggregate. TABLE 1 — Properties of the pumice LWA Aggregate
Grain size, mm
Bulk specific gravity in oven dry conditions, kg/m3
Water absorption % by weight
% by volume
Yali0
0.15 – 1.18
1330
13.0
17.3
Yali1
1.18 – 2.36
1310
19.0
24.9
Yali2
2.36 – 4.75
1210
26.7
32.3
Hekla0/C
0.6 – 1.18
917
48.0
43.7
Hekla1/C
1.18 – 2.36
782
70.5
54.8
Hekla2/C
2.36 – 4.75
766
68.2
51.8
Hekla0/S
0.6 – 1.18
1206
32.3
38.9
Hekla1/S
1.18 – 2.36
1028
45.0
46.6
2.2.
Concrete Compositions
The concretes prepared were with a maximum aggregate size of 9.5 mm, which was crushed dolomite with water absorption of 1% by weight, and specific gravity of 2700 kg/m3. The fine aggregate was siliceous sand with water absorption of 0.4%. The cement was a product of Nesher, Israel Cement Enterprises Ltd, which is equivalent to type I Portland cement with specific surface area of 323 kg/m2. Silica fume was incorporated in one of the concretes. It had a SiO2 content of 92% and surface area of 18.2 m2/g. Two types of concretes were prepared. One with a w/cm ratio of 0.33 and the other with a w/cm ratio of 0.25, which included also 10% silica fume by weight of the cement. All the mixes contained superplasticizer, naphthalene formaldehyde sulfonate type, added at a dose of 1.5% by weight of the cement. Reference concretes, without any LWA, were prepared with the aggregates being in SSD conditions. The lightweight aggregates were added to the mixes as replacement (by volume) for part of the siliceous sand. The content of the added lightweight aggregate was such that it contained in it the amount of water required to compensate for self desiccation for the first week of hydration. The required amount of water was determined by a calculation taking into account the estimated chemical shrinkage and measured rates of hydration of the cementitious system. A detailed calculation is provided in the work (Zhutovsky et al 2003a). The water contents were 20 kg/m3 and 23 kg/m3 for the 0.25 and 0.33 w/cm concretes, respectively, containing ~600 kg/m3 cement + ~60 kg/m3 silica fume, and ~510 kg/m3 cement, respectively. These values, combined with the absorption of the aggregate (Table 1) enabled to determine their content. The mix compositions for the reference concretes and the ones containing the lightweight aggregates are presented in Tables 2 and 3. It can be seen that the replacement with Hekla LWA was much smaller than the Yali (as low as 30 kg/m3 for the Hekla, compared with about 100 kg/m3 for the Yali), which reflects the much higher porosity of the Hekla aggregate (Table 1).
TABLE 3 — Composition of concretes with 0.25 w/cm ratio, with the LWA containing water sufficient to counteract self desiccation for 7 days (23 kg of water per m3 of concrete)1. Cement 580
Silica fume 60
580
60
Yali1
580
60
Yali2
580
60
Hekla0/C
619
64
Hekla1/C
619
64
Hekla2/C
617
64
Hekla0/S
637
66
Hekla1/S
634
66
Reference WSAREF025SF Yali0
Component, kg/m3 Mix water Sand 160 472 (470) 160 117 (117) 160 228 (227) 160 286 (285) 171 307 (305) 171 338 (336) 170 339 (337) 176 221 (219) 175 259 (257)
Gravel 1162 (1145) 1162 (1145) 1162 (1145) 1162 (1145) 1241 (1222) 1240 (1221) 1236 (1218) 1277 (1258) 1271 (1252)
Pumice 0 199 (176) 144 (121) 108 (85) 76.1 (51.1) 59.6 (35.6) 60.6 (36.2) 103.5 (78.2) 80.6 (55.4)
TABLE 2 — Composition of concretes with 0.33 w/cm ratio, with the LWA containing water sufficient to counteract self desiccation for 7 days (20 kg of water per m3 of concrete)1. Cement 506
Mix water 167
506
167
Yali1
506
167
Yali2
506
167
Hekla0/C
514
170
Hekla1/C
514
170
Hekla2/C
512
169
Hekla0/S
526
174
Hekla1/S
524
173
Reference (WSAREF033) Yali0
1
Component, kg/m3 Sand 574 (572) 268 (267) 362 (361) 410 (408) 420 (418) 445 (443) 446 (444) 352 (350) 382 (380)
Gravel 1162 (1145) 1162 (1145) 1162 (1145) 1162 (1145) 1180 (1162) 1180 (1162) 1176 (1158) 1208 (1190) 1204 (1186)
The aggregate weight is in SSD conditions; the values in parentheses are oven dry weights
Pumice 0 174 (154) 125 (105) 95 (75) 62 (42) 48.6 (28.6) 49.3 (29.3) 81.9 (61.9) 55 (35)
2.3.
Concrete Preparation and Testing
The preparation of the concretes and their testing were carried out by the same procedures reported earlier (Zhutovsky et al 2003b). Immediately after mixing the concretes were placed in special molds which were sealed and were equipped with instrumentation that enabled measurement of free length changes immediately after casting, i.e. before setting. The specimens were kept in sealed conditions and in a controlled laboratory at 30oC. The specimen size was 40x40x1000 mm and the length change tests were carried out up to 168 hours. Compressive strength was determined on 50 mm cube specimens which were cured in sealed conditions at 30oC at 1, 3, 7, and 28 days. Details of the testing arrangements are provided in references (Bentur et al 2001, Zhutovsky et al 2003a, Zhutovsky et al 2003b).
3. Results Typical autogenous shrinkage curves are shown in Fig. 1. In the first few hours there is some expansion, both in the specimens with and without lightweight aggregates, which levels off or increases at a much lower rate in the concretes with lightweight aggregates, and decreases in the case of the reference concretes. This time, t0, at which early expansion peaks or levels off, was taken as a reference time from which autogenous shrinkage is measured. The point t0 occurs within less than 10 hours, and it roughly coincides with the setting time. Such a trend of early expansion in sealed conditions has been observed previously in several studies, and it can be attributed to a variety of effects, such as small temperature rise and hydration swelling. Since autogenous shrinkage is expected to initiate only after setting has occurred, it is justified to take the time t0 as a reference from which autogenous shrinkage will be determined from the length change curves. The curves in Fig. 1 clearly show the difference in behavior of the various systems: marked autogenous shrinkage in the reference concrete, with a marked reduction in this shrinkage in the concretes with soaked lightweight aggregates, or even slight expansion, which in some cases may be maintained even for the whole week period. The curves are shown for reference concretes at 0.33 and 0.25 w/cm ratio (WSAREF0.33 and WSAREF0.25SF, respectively) and concretes with soaked porous lightweight aggregates having 51.8% porosity by volume and 2.36-4.75 mm size, which were added to the 0.33 and 0.25 w/cm concretes in a quantity with sufficient soaked water to eliminate self-desiccation (Hekla 2/C (20) and Hekla 2/C (23), respectively). In order to quantify the autogenous shrinkage curves, an efficiency coefficient was defined, being the ratio by which the autogenous shrinkage of the lightweight aggregate concrete is reduced, relative to the reference concrete. The autogenous shrinkage strains were taken from the time t0, as explained above. Since in each of the systems the aggregates contained the amount of water needed to eliminate self-
desiccation during one week of hydration, than the efficiency coefficient can be used as an engineering parameter to characterize the efficiency of the lightweight aggregate. The efficiency factor was calculated here for the time period of one week. The efficiency values for the different aggregates in the two types of concretes are shown in Table 4. Values higher than one in some of the systems are indicative of concretes where expansion occurred and was maintained even under sealed curing. This implies that water in the lightweight aggregates in such concretes is readily available to support hydration to an extent which leads to slight swelling, as occurs in submerged concretes. 40
strain, 10
-6
0 Hekla2/C(23) Hekla2/C(20) WSAREF0.33 WSAREF0.25SF
-40 -80 -120 -160 0
24
48
72
96
120
144
168
time, hours
strain, 10
-6
40
Hekla2/C(23) Hekla2/C(20) WSAREF0.33 WSAREF0.25SF
20
0
-20 0
4
8
12
16
20
24
time, hours
FIG. 1 — Typical curves of autogenous shrinkage of concretes with soaked lightweight aggregates and reference concretes during seven days of sealed curing (top), and the first 24 hours of sealed curing (bottom).
TABLE 4 — Efficiency coefficients of the various systems, calculated for strains measured after 7 days of sealed curing of concretes with w/cm of 0.33. Aggregate type
Porosity, % vol.
Aggregate spacing, mm
Efficiency coefficient
Yali0 Yali1 Yali2 Hekla0/C Hekla1/C Hekla2/C Hekla0/S Hekla1/S
17.3 24.9 32.3 43.7 54.8 51.8 38.9 46.6
0.5 1.4 3.2 1.2 2.6 4.8 1.2 2.6
0.17 0.48 0.78 0.84 1.15 1.10 0.77 0.95
Compressive strength at 28 days, MPa 70.2 70.5 64.7 75.9 72.6 73.4 71.9 70.8
Trends demonstrating the effect of aggregates on strength are presented in Figure 2 in terms of the strength of the concretes with lightweight aggregates relative to the reference concrete at the same age. The data presented is for the extreme cases, namely aggregates of high and low porosities. Generally, the relative values seem to be lower in: (i) the higher w/cm ratio concrete, (ii) in the aggregates with the higher porosity, and (iii) at 1 day age. By 28 days, the strength is practically the same as in the control for the lower porosity aggregates, at both w/cm ratios, and in the higher porosity aggregate for the low w/cm ratio concrete. The higher porosity aggregate, in the higher w/cm ratio mix (0.33), achieves at 28 days about 90% of the strength of the control. These values indicate that the range of particles studied here enable to achieve the goal of eliminating autogenous shrinkage by using a small amount of aggregates, with only a minimal decline in strength, or non at all. There are however some trade off effects which need to be noted: the higher porosity aggregate is extremely efficient in counteracting self desiccation compared to the low porosity aggregate (efficiency coefficient of about 1 vs. 0.5, respectively); however, in the case of the 0.33 w/cm ratio concrete its presence is accompanied by a 28 days strength reduction of about 10%.
4. Discussion The two parameters which are expected to influence the effectiveness of the aggregate are their pore structure and the distance between the aggregates (the aggregate spacing). The pore volume of the aggregates served as a parameter characterizing the pore structure. The aggregate spacing parameter was determined by calculations in which the proximity of the paste surrounding the aggregate was characterized by calculated proximity curves, describing the portion of the volume of paste around the aggregate which lies within a distance “d” from the aggregate. The concepts for this calculation are described in reference (Zhutovsky et al 2003b). They are based on determination of the probability of finding a matrix fraction to be lying within a given distance from the surface of a lightweight aggregate, using the nearest surface distribution function developed by Lu and Torquato (1992). Typical results of the calculation are shown in Fig. 3. The distance at which 99% of the paste is within the
perimeter of the lightweight aggregate was used here as the value characterizing the aggregate spacing. 1.2
1 day 28 days
relative strength
1 0.8 0.6 0.4 0.2 0
0.33 w/cm; 54.8% por.
0.25 w/cm; 54.8% por.
0.33 w/cm; 17.3% por.
0.25 w/cm; 17.3% por.
percentage of cement paste lying within the given distance
FIG. 2 — The effect of lightweight aggregate porosity and concrete composition on the 1 and 28 days strength of concretes with lightweight aggregates relative to reference concretes. 100% 80%
Hekla2/C Hekla1/C Hekla1/S Hekla0/C Hekla0/S Yali2 Yali1 Yali0
60% 40% 20% 0% 0
1
2
3
4
5
6
distance from aggregate, mm
FIG. 3 — Typical calculated aggregate-matrix proximity curves for the mixes with Yali and Hekla pumice made at w/cm = 0.33 and containing 20 kg/m3 of internal curing water. In order to assess the significance of these parameters in controlling the efficiency of the lightweight aggregates, the efficiency coefficients were plotted against the two of them (Figs. 4, 5 and 6). It can be seen that there is a trend for the efficiency to increase with the porosity of the aggregate; the relation is clear cut in the case of the 0.33 w/cm ratio (Fig. 4), but there is a greater spread in the case of the 0.25 w/cm
ratio concrete (Fig. 5). There is also a tendency for the efficiency to increase with aggregate spacing, but the relation exhibits considerable spread (Fig. 6). One would expect the trend to be just the opposite of the one observed in Fig. 6, with higher spacing associated with lower efficiency. This apparent contradiction may be resolved if the relations between the porosity and spacing of the aggregates is considered (Fig. 7). It can be seen that for each of the aggregates studies here, Yali and Hekla, higher porosity was associated with higher spacing. This reflects the fact that the coarser aggregates are the ones with higher porosities (Table 1). Thus, the trends observed here suggest that the pore structure (as characterized by the porosity value) is the overriding parameter controlling the lightweight aggregate efficiency in the range of lightweight aggregates studied here. The apparent lack of influence of aggregate spacing may imply that once the water is out of the aggregate, it can readily be transported over distances of several millimeters. Lura et al (2002) demonstrated in microscopical studies that such a range of transport indeed occurs in these systems. 1.4
efficiency coefficient
1.2
0.60-1.18 1.18-2.36 2.36-4.75
1 0.8 0.6 0.4 0.2 0 0
10
20
30
40
50
60
porosity, % vol.
FIG. 4 — Relation between efficiency coefficients and porosity for 0.33 w/cm concrete.
!
1 0.9
0.60-1.18 1.18-2.36 2.36-4.75
efficiency coefficient
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
10
20
30
40
50
60
porosity, % vol.
FIG. 5 — Relation between efficiency coefficients and porosity for 0.25 w/cm concrete.
FIG. 6 — Relations between the efficiency coefficients and the aggregate spacing value for the 0.33 and 0.25 w/cm ratio concretes.
"
60
porosity, %vol.
50 40 30 20 Yali Hekla
10 0 0
1
2
3
4
5
aggregate spacing, mm
FIG. 7 — Relation between aggregate spacing (for 0.33 w/cm concrete) and aggregate porosity. The trends and discussion above suggest that the “bottle neck” in mobilizing all the water within the aggregate, to counteract self-desiccation, is the movement of the water from within the aggregate outward. Obviously, this is a simplistic explanation, and in order to support it and quantify it there is a need to study the pore structure of the aggregates. The trends observed here suggest that aggregates with higher porosities are the ones which allow water to move more readily. This implies a more open pore structure. Microscopical observations of the aggregates reveal an apparently open structure consisting of big pores in all of them (Fig. 8). There is a need for an in depth study of the nature of the pore structure, to characterize the connectivity of these big pores in order to understand the differences in performance observed here.
FIG. 8 — Pore structure of Hekla2/C lightweight aggregate observed by SEM. The influences on strength (Fig. 2) can be readily explained based on the concepts prevailing in lightweight aggregate concrete, namely that the strength in such systems is controlled also by the properties of the aggregate, and there is a need for high strength matrix to compensate for the weakening effect of the lightweight aggregates. The lower porosity aggregate is presumably stronger, and therefore shows a better performance at the 0.33 w/cm ratio concrete. In the 0.25 w/cm ratio concretes the two aggregates perform equally well at 28 days, which is probably the result of the high strength of the matrix, compensating for the weaker higher porosity aggregate. In the 0.33 w/cm ratio concrete, the matrix is apparently not strong enough to compensate for the effect of the higher porosity aggregate. Similar considerations can be made to account for the lower relative strength values at 1 day, where the matrix is of lower strength.
5. Conclusions (a) The results in the present study support the feasibility of the concept of using small lightweight aggregates which are highly porous, as additives to reduce or eliminate autogenous shrinkage in high strength concrete: addition of a relatively small content (less than 30 kg/m3) of aggregates having about 50% volume porosity and size of only few millimeters, could eliminate autogenous shrinkage with only a small influence on compressive strength (either keeping it unaffected or reducing it by about 10%).
(b) The effectiveness of the lightweight aggregates was largely dependent on their total porosity, but not on the aggregate spacing. Higher porosity resulted in greater efficiency. (c) The greater efficiency of the more porous aggregates may be due to a more open pore structure which is associated with the larger porosity, allowing water to be transported more readily from within the aggregate, to mobilize all the water within it. There is a need for additional study to validate this explanation. (d) The experimental results and the calculations of the aggregate spacing suggest that in the systems studied here, the water can effectively penetrate from the aggregate into the matrix, over a distance of several millimeters. This implies that the “protective envelope” (whereby the matrix can be effectively supplied with water from the aggregate), of a properly formulated lightweight aggregate, can extend to distances of several millimeters. This relatively large distance may account for the lack of sensitivity to the spacing between the aggregates.
6. References Bentur, A., 2001, “Early Age Shrinkage and Cracking in Cementitious Systems”, Concrete Science and Engineering, Vol. 3, pp. 3-12. Bentur, A., Igarashi, S., and Kovler, K., 2001, “Prevention of Autogenous Shrinkage in High- Strength Concrete by Internal Curing Using Wet Lightweight Aggregates”, Cement and Concrete Research, Vol. 31, pp.1587-1591. Bentz, D.P. and Snyder, K.A., 1999, “Protected Paste Volume in Concrete, Extension to Internal Curing Using Saturated Lightweight Fine Aggregate”, Cement and Concrete Research, Vol. 29, 1863-1867. Kohno, K., Okamoto, T., Isikawa, Y., Sibata, T. and Mori, H., 1999, “Effects of Artificial Lightweight Aggregate on Autogenous Shrinkage of Concrete”, Cement and Concrete Research, Vol. 29, pp. 611-614. Lu, B. and Torquato, S., 1992, “Nearest-Surface Distribution Functions for Polydispersed Particle System”, Physical Review, Vol. 45, pp. 5530-5544. Lura, P. and Van Breugel, K., 2000, “Moisture Exchange as a Basic Phenomenon to Understand Volume Changes of Lightweight Aggregate Concrete at Early Age”, PRO 17: Shrinkage of Concrete - 'Shrinkage 2000', Proceedings of the International RILEM Workshop, Paris, France, 16-17 October 2000, Ed. V. Baroghel-Bouny and P. C. Aïtcin, RILEM Publications, France, pp. 533-546. Lura, P., Van Breugel, K. and Maruyama, I., 2002, “Autogenous and Drying Shrinkage of High-Strength Lightweight Aggregate Concrete at Early Ages – The Effect of Specimen Size”, PRO 23: Early Age Cracking in Cementitious Systems EAC'01, Proc. of the International RILEM Conference, Haifa, Israel, 12-14 March 2001, Ed. K. Kovler and A. Bentur, RILEM Publications, France, pp. 335-342. Schwesinger, P. and Sickert, G., “Reducing Shrinkage in HPC by Internal Curing by Using Pre-soaked LWA”, 2000, Proceedings of International Workshop on Control of Cracking in Early-Age Concrete, Tohoku University, Japan, pp. 313318. Takada, K., van Breugel, K., Koenders, E.A.B. and Kaptijn, N., 1998, “Experimental Evaluation of Autogenous Shrinkage of Lightweight Aggregate Concrete”, Proceedings of International Workshop on Autogenous Shrinkage of Concrete, JCI, Ed. E. Tazawa, June 13-14, Hiroshima, Japan, pp. 221-230. Van Breugel, K., Outwerk, H. and De Vries, J., 2000, “Effect of Mixture
Composition and Size Effect on Shrinkage of High Strength Concrete”, PRO 17: Shrinkage of Concrete - 'Shrinkage 2000', Proceedings of the International RILEM Workshop, Paris, France, 16-17 October 2000, Ed. V. Baroghel-Bouny and P. C. Aïtcin, RILEM Publications, France, pp. 161-177. Zhutovsky, S., Kovler, K. and Bentur, A., 2001, “Influence of Wet Lightweight Aggregate on Autogenous Shrinkage of Concrete at Early Ages”, Proc. 6th Int. Conf. “Creep, Shrinkage and Durability Mechanics of Concrete and Other QuasiBrittle Materials”, Ed. F.-J. Ulm, Z.P. Bazant and F.H. Wittmann, Cambridge (MA), USA, August 20-22, 2001, Elsevier, pp. 697-702. Zhutovsky, S., Kovler, K. and Bentur, A., 2003a, “Influence of Cement Paste Matrix Properties on the Autogenous Curing of High-Performance Concrete”, accepted for publication, Cement and Concrete Composites. Zhutovsky, S., Kovler, K. and Bentur, A., 2003b, “Assessment of Distance of Water Migration in Internal Curing of High-Strength Concrete”, accepted for publication, ACI Special Publication on Autogenous Shrinkage of Concrete.
TECHNION – UNIVERSITY OF ILLINOIS RESEARCH COOPERATION The researchers at the Technion and the University of Illinois collaborated in the task defined in the program as “Preparation and testing of special lightweight microaggregates for autogenous curing of concrete”. These special aggregates were prepared from natural pumice sand of different fineness, according to the calculations and test simulations of water flow from a single lightweight aggregate to the surrounding cement paste. It was agreed that the aggregate size is one of the most important issues to be tested. It would be difficult to realistically construct a 2D physical specimen that would simulate aggregate diameters below 2 mm. Therefore, in a meeting of Prof. Kovler and Prof. D. Lange in Boston, at the end of August 2001, it was agreed that it would be reasonable to cut small prisms with a 5*5 mm square cross-section for laboratory simulation of water migration from the aggregate to the surrounding cement matrix. This size was selected because of the following reasons: First, the initial concept of the autogenous curing was based on using fine aggregate, and 5 mm is practically the largest size of the sand particles. Second, in the tests of the Technion group, which studied three different sizes of pumice aggregate (“Pumice0” with particles in the range of 0.15 to 1.18 mm; “Pumice1” with particles in the range of 1.18 to 2.36 mm and “Pumice2” with particles in the range of 2.36 to 4.75 mm), it was shown that the best effect was observed with the coarser aggregate (“Pumice2”). In the course of 2002 the US group characterized the Technion lightweight (LW) aggregate using Scanning Electron Microscopy (SEM) and Mercury Intrusion Porosimetry (MIP) methods. Image analysis was employed to measure pore volume and shape. It was resolved that the different sizes of LW particles have different pore size distribution (PSD) as measured by MIP. These results were discussed at the joint meeting of Prof. Kovler, Prof. Lange and Dr. Lura in the US, during the ACI Fall convention 2002, and also in the joint meeting of Prof. Bentur, Prof. Kovler, Prof. Lange and Dr. Lura during the RILEM Symposium on Concrete Science and Engineering, March, 2004, Northwestern University, Evanston, Illinois. It was found that the results obtained independently in the NIST by the American team are consistent with those of the Technion observations. Most of the pores of LWA were found to be “tubular”. This fact can be explained by the evolution of the pumice particles: when this kind of volcanic material was created in the form of magmatic fume, the hot earth gases passed through it leaving tube-like open pores, which are responsible for the effective deliberation of water in tests performed by the Technion. Both research groups agreed that the fact that different sizes of LW particles have different PSD is because the bulk material fractures along weak planes; thus the smaller grains will systematically have smaller pore sizes. As a result of this work, it was suggested that the concept of using LW particles for carrying water would only prove valid for larger particles. It will be important to define a function relating
particle size to performance. Obviously, the obtained results are valid for the pumice aggregate only.
As a result of the current research, and in addition to papers by each group, joint papers have been prepared and published: Lura, P., Bentz, D.P., Lange, D.A., Kovler, K., Bentur, A. and van Breugel, K., Measurement of water transport from saturated pumice aggregates to hardening cement paste, Proc. of the Conference ‘Advances in Cement and Concrete’, Aug. 10-14, 2003, Edited by D. Lange, K. L. Scrivener and J. Marchand, ECI, USA, 2003, pp. 89-99. Lura, P., Bentz, D.P., Lange, D.A., Kovler, K. and Bentur, A., ‘Pumice aggregates for internal water curing’, PRO 36: Proc. Int. RILEM Symp. on Concrete Science and Engineering - A Tribute to Arnon Bentur, 24 March, 2004, Northwestern University, Evanston, Illinois, USA, Eds. K. Kovler, J. Marchand, S. Mindess and J. Weiss, RILEM Publications S.A.R.L., 2004, pp. 137-151.
CONCLUSIONS The results presented in this report clearly demonstrate the feasibility of the approach of autogenous (internal) curing to high-strength / high-performance cementitious systems, and based on the current results some of the more fundamental parameters and processes can be assessed. The goal for the third year is to quantify the mechanisms identified and based on that to optimize the technology of internal curing. The results of the study have been published in the following papers: 1. Zhutovsky, S., Influence of Wet Lightweight Aggregate on Autogenous Curing of High Strength Concrete, M.Sc Thesis, Technion - Israel Institute of Technology, Haifa, 2002. 2. Zhutovsky, S., Kovler, K. and Bentur, A., ‘Efficiency of lightweight aggregates for internal curing of high strength concrete to eliminate autogenous shrinkage’, in ‘Early Age Cracking in Cementitious Systems’, K. Kovler and A. Bentur, editors, Proc. RILEM Conference, (2001) 365-374. 3. Zhutovsky, S., Kovler, K. and Bentur, A., ‘Influence of wet Lightweight fine Aggregate on Autogenous Shrinkage of Concrete at Early Ages’, in CONCREEP: Creep Shrinkage and Durability Mechanics of Concrete and other Quasi-Brittle Materials, F.-J. Ulm, Z.P. Bazant and F.H. Wittmann, editors, Proc. Int. Conf. August 2001, Cambridge, USA, Elsevier, (2001) 697-702. 4. Zhutovsky, S., Kovler, K. and Bentur, A., ‘Efficiency of Lightweight Aggregates for Internal Curing of High Strength Concrete to Eliminate Autogenous Shrinkage’, Materials and Structures, Vol. 35 (2002) 97-101. 5. Zhutovsky, S., Kovler, K. and Bentur A., Autogenous curing of high-strength concrete using pre-soaked pumice and perlite sand, Proc. of the 3rd International Seminar on Self-Desiccation and its Importance in Concrete Technology, Lund, Sweden, 14-15 June 2002, Ed. B. Persson and G. Fagerlund, Lund University, Sweden, 2002, pp. 161-173. 6. Lura, P., Bentz, D.P., Lange, D.A., Kovler, K., Bentur, A. and van Breugel, K., Measurement of water transport from saturated pumice aggregates to hardening cement paste, Proc. of the Conference ‘Advances in Cement and Concrete’, Aug. 1014, 2003, Edited by D. Lange, K. L. Scrivener and J. Marchand, ECI, USA, 2003, pp. 89-99. 7. Zhutovsky, S., Kovler, K. and Bentur A., Assessment of distance of water migration in internal curing of high-strength concrete. “Autogenous Deformation of Concrete”, ACI SP-220, American Concrete Institute, 2004, pp. 181-197. 8. Zhutovsky, S., Kovler, K. and Bentur A., Influence of cement paste matrix properties on the autogenous curing of high-performance concrete. Cement & Concrete Composites, 2004, V. 26, pp. 499–507. 9. Lura, P., Bentz, D.P., Lange, D.A., Kovler, K. and Bentur, A., ‘Pumice aggregates for internal water curing’, PRO 36: Proc. Int. RILEM Symp. on Concrete Science and Engineering - A Tribute to Arnon Bentur, 24 March, 2004, Northwestern University, Evanston, Illinois, USA, Eds. K. Kovler, J. Marchand, S. Mindess and J. Weiss, RILEM Publications S.A.R.L., 2004, pp. 137-151. 10. Kovler, K., Souslikov, A. and Bentur A., Pre-soaked lightweight aggregates as additives for internal curing of high strength concretes. Cement, Concrete & Aggregates (accepted to publication).
