total pore space of the non-air-entrained unsalted mortar was filled with ice ... The pressures due to the expansion of non-freezing pore water causing .... /kg). 460. During the cooling test the strains of the wet and dry samples were ... of thermal expansion and density of the super cooled water and ice ..... is in balance with.
ROLE OF NON-FREEZING PORE WATER IN FREEZE-THAW DAMAGE OF CONCRETE MORTARS Vesa Penttala Helsinki University of Technology, Department of Civil and Environmental Engineering, Espoo, Finland Abstract A novel theory to explain the mechanism of freeze-thaw damage in concrete mortars is presented. The theory takes into consideration the density changes of ice and unfrozen pore water during cooling and warming phases of the freeze-thaw cycle. Two mortar sample types having 25 MPa compressive strengths were produced for the tests. The other mortar type was air-entrained while the other was produced without admixtures. Half of the test samples were cured in a sodium chloride solution before the water curing regime. During the freeze-thaw tests the temperature cycle in a low-temperature calorimeter decreased from +20 to -70 oC and back to 20 oC. The strains of the mortars were measured by strain gauges and the evolved ice amount was calculated from the calorimeter data. The volumes of ice, non-freezing pore water, and air-filled pore space were calculated during the freezing cycle. In the salted mortars the total pore space was filled with ice and non-freezing water quite soon after the initial freezing temperature. Thereafter the expansion of non-freezing water caused water to be squeezed out of the samples and to be frozen on the surface of the test prisms. Also the total pore space of the non-air-entrained unsalted mortar was filled with ice and unfrozen pore water but only at the temperature of -57 oC. About 15 % of the pore volume of the airentrained unsalted mortar was air filled at the lowest temperature. The pressures due to the expansion of non-freezing pore water causing large tensional stresses into the test mortars is proposed to be counterbalanced by a larger negative pore water pressure originating from the entropy difference between the surface of ice and the nonfreezing pore water and the curved water menisci. The theoretical negative pore water pressure on the surface of the prisms was calculated to be nearly 150 MPa and the positive pore water pressure caused by water expansion was estimated to be below 50 MPa in the nonsalted mortar prisms. 1.
INTRODUCTION
The freezing mechanism and freeze-thaw damage of concrete and other moist porous materials has been studied extensively for the last 70 years. The pioneering works of T. C. Powers and his co-workers [1-4] have laid basis on the physical behaviour of moist paste and concrete during freeze-thaw loads. C. G. Litvan [5-9] and his co-authors should also be mentioned because of their differing views of the freezing phenomenon in which the freezing of the pore water is suggested to take place on the surface of the test sample. G. Fagerlund
Page
1
100
Unfrozen pore water amount [%] -
Unfrozen pore water amount [%]
[10] introduced the concept of critical degree of pore water saturation which governs the severity of the freeze-thaw damage. Of the more recent references studying the subject the publications of M. J. Setzer [11] and G. W. Scherer [12] should be mentioned. A somewhat more refined literature review is presented in reference [15]. However, the author of this article is unaware of any theoretical or empirical research results in which the changes in volume of the unfrozen pore water and ice due to density changes have been taken into consideration in freeze-thaw damage of binder paste or concrete. Large portion of pore water in concrete does not freeze even at -70 oC. In the average, only some 30% of the evaporable pore water amount of mortars having compressive strengths from 30 to 60 MPa and cured in relative humidity from 85 to 100 % freezes at the mentioned temperature, Fig. 1 [13]. Even the fact that salt has been introduced into the pore water by immersing the test samples in NaCl-solutions has very small effect on the non-freezing relative pore water amount which is about the same as in normal mortars. 80 60 40 M30
20
MS30 0 80
85
90
95
100
Relative humidity [%]
100 80 60 40 M50 MS50
20 0 80
85
90
95
100
Unfrozen pore water amount [%]
Relative humidity [%] 100 80 60 40 M60A MS60A
20 0 88
90
92
94
96
98
100
Relative humidity [%]
Figure 1: Unfrozen pore water amounts compared to the evaporating pore water amount at -70 o C of different mortars cured at different relative humidity. Notation M60A denotes an airentrained mortar having compressive strength of 60 MPa while MS50 represents test results of non-air-entrained 50 MPa mortars preserved in a NaCl-solution for 6 weeks before the curing regime [13] The aim of this research article is to present some results of a larger project in which the unfrozen pore water amounts and strains of mortar prisms were studied in a low temperature calorimeter in the temperature range of +20…-70 oC. The test mortars were preserved in different relative humidity prior to the freezing and thawing test. In contrast to the previous tests [13-18] a low strength mortar having compressive strength of 25 MPa was used. Both air-entrained and non-air-entrained mortars were studied and half of the test samples were preserved in saturated NaCl-solutions for six weeks before the curing measures in different relative humidity or under water. Only the test results of the mortars which were cured under water will be presented here.
Page
2
2.
TEST SPECIMENS AND ARRANGEMENTS
Two mortar sample types M25 and M25A were produced for the tests. The composition of the mortars was chosen to represent the ingredients of normal concrete in which the aggregate fractions from 2 to 16 mm have been omitted. This explains the large binder amounts of the test mortars presented in Table 1. Test mortar M25 was a normal mortar in which no air-entraining admixture was used while in test mortar M25A extra air was introduced into the mix with an air-entraining agent. When the samples were immersed into a NaCl-solution for six weeks the results are presented by using abbreviations MS25 and MS25A, respectively. The air content of mortars M25 was 2.0% and of mortars M25A 9.5%. The binder was Portland cement CEM II A 42,5 of local origin produced by Finnsementti Oy, Table 2. The aggregates were sieved into 4 fractions and their petrographic composition was mostly granite. The air-entraining admixture was a vinsol resin. When the mortars have been cured in different relative humidity the humidity is presented after the before mentioned notation e. g. MS25ARH100 as all mortars to be presented in this paper have been cured under water. Table 1. Composition and compressive strength results of the test mortars Test mortar M25 and MS25 M25A and MS25A Aggregates (kg/m3) Total 1306 1290 - # 1-2 mm 285 252 - # 0.5-1.2 mm 321 283 - # 0.1-0.6 mm 392 346 - # < 0.125 mm 308 409 Cement (kg/m3) 463 391 3 Water (kg/m ) 327 304 Air-entraining admixture (kg/m3) 0.326 Air content (%) 2.0 9.5 Compressive strength (MPa) 25.5 23.4 -100-mm cubes at 28 days After mixing for 4 minutes the test mortars were poured into 2 dl polyethylene bottles which were rotated for 5 hours to prevent any segregation. At the age of 3 days the bottles were cut open and 10•10•55 mm3 prisms were cut. Two strain gauges were glued on the opposite sides of the prisms and they were placed into water. A comparison prism was dried in oven at 105 o C temperature for 5 days before the dilation test. On the test prisms in which no additional salts were introduced two strain gauges were glued on the dry comparison sample at the ages of 30-36 days, wires were soldered to the strain gauges and it was placed into the reference chamber of a low-temperature calorimeter. At the same time the wet sample was taken from the storage environment and wires were similarly soldered to the strain gauges and the sample was placed into the other measuring cell of the calorimeter. The wiring procedure of the wet sample lasted from 7 to 10 minutes and during this time some amount of water evaporated from the wet mortar prisms. When test prisms were cured in saturated NaClwater solutions the glueing of the strain gauges and the freeze-thaw test started at the mortar age of 82-102 days.
Page
3
Table 2 : Chemical composition and physical properties of cement CEM II A 42,5 CEM II A 42,5 Chemical composition (w-%) CaO 59.9 SiO2 20.2 Al2O3 5.4 Fe2O3 2.3 MgO 3.3 K2O 0.64 Na2O 0.76 SO3 3.5 Loss of ignition (950o C) 0.77 Physical properties Compressive strength (MPa) 1d 24 7d 48 28 d 56 Specific area (Blaine) (m2/kg) 460 During the cooling test the strains of the wet and dry samples were measured every 40 seconds. The strain values of the strain gauges were calibrated by steel samples from which the coefficient of temperature expansion was known. The difference in the heat capacities of the wet and dry test samples was measured by the low-temperature calorimeter produced by Setaram. The cooling rate of the test was 3 oC/hour and the following heating rate 4.2 o C/hour. The test started from +20 oC and the temperature dropped linearly to -70 oC and again increased back to +20 o C after a few hours waiting at -70 oC. 3.
TEST RESULTS
The strains and relative ice amounts of the four mortar types M25RH100, M25ARH100, MS25RH100, and MS25ARH100 are presented in Figs 2-5. The pore water, ice, and evaporating water amounts of the test mortars are shown also in Fig. 6. In the figure the calculated gel pore water amounts are also presented. In the calculation the hydration degree of the different mortars has been estimated from previous tests. The results of the relative humidity measurements of the chamber air in the calorimeter are shown in Figs 7 and 8. 4.
DISCUSSION
As can be noticed in the strain results of the unsalted test mortars of Figs 2 and 3, the initial freezing did not cause noticeable expansion in the test mortars. A remarkable contraction can be noticed during the freezing phase which continues for some 20 degrees lower from the initial freezing temperature. Only in the vicinity of temperature -40 oC a small expansion can be seen in the unsalted test mortars. The test mortars preserved in NaCl-solution prior to the water curing phase behaved quite differently. Quite soon after ice was noticed in the samples a remarkable expansion took place, Figs 4 and 5. In order to analyze the relative volumes of the different phases in the mortar pore structure Figs 10-13 were drawn up. The coefficients
Page
4
of thermal expansion and density of the super cooled water and ice presented in Fig. 9 were used in the calculation. The density values have been taken from references [19-23]. The total pore volumes of the test mortars were calculated from duplicate test mortars by preserving them for 24 hours in a water pressure of 15 MPa and weighing the samples in air and under water. Thereafter the total water amount was measured by heating the test samples at 105 oC.
0 -70
-60
-50
-40
-30
-20
-10
0
10
20 -100 -200 -300 -400 -500 -600
Ice evolution M25RH100 freezing M25RH100 thawing
-700 -800 -900
3
100
Strain [µm/m] and ice evolution [*10 g/gsample]
200
-1000 Temperature (°C)
Figure 2: Strain and ice evolution results of test mortar M25RH100
400 200 0 -70
-60
-50
-40
-30
-20
-10
0
10
20 -200 -400
Ice evolution M25ARH100 freezing M25ARH100 thawing
-600 -800
4
600
Strain [µm/m] and ice evolution [*10 g/gsample]
800
-1000 Temperature (°C)
Figure 3: Strain and ice evolution results of air-entrained test mortar M25ARH100
Page
5
Ice evolution 1400 MS25RH100 freezing 1200
4
MS25RH100 thawing
Strain [µm/m]and ice evolution [*10 g/gsample ]
1600
1000 800 600 400 200 0 -70
-60
-50
-40
-30
0
-10
-20
10
20 -200 -400
Temperature (°C)
Figure 4: Strain and ice evolution results of test mortar MS25RH100
MS25ARH100 freezing
400
3
MS25ARH100 thawing 300 200 100 0 -70
-60
-50
-40
-30
-20
-10
0
10
Strain [µm/m] and ice evolution [*10 g/gsample ]
500
Ice evolution
20 -100 -200
Temperature (°C)
Figure 5: Strain and ice evolution results of air-entrained test mortar MS25ARH100
Page
6
Pore water and ice amounts [g/g sample]
Evaporating water Ice at -65 C Gel water Unfrozen water at -65 C
0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 M25RH100
M25ARH100
MS25RH100
MS25ARH100
Figure 6: Pore water and ice amounts of the test mortars 100 M25RH100 thawing
95
M25ARH100 freezing M25ARH100 thawing
90 85
RH [%]
M25RH100 freezing
80 75 70 -70
-60
-50
-40
-30
-20
-10
0
10
20
Temperature [°C]
Figure 7: The relative humidity data of the test mortars M25RH100 and M25ARH100 95 MS25RH100 freezing MS25RH100 thawing
90
MS25ARH100 thawing
85
80
RH [%]
MS25ARH100 freezing
75
70 -70
-60
-50
-40
-30
-20
-10
0
10
20
Temperature [°C]
Figure 8: The relative humidity data of the test mortars MS25RH100 and MS25ARH100
Page
7
The volume of air filled pores has been calculated by applying Equations 1-3 Va (T ) = V p (T ) − Vw (T ) − Vi (T ) Va (T ) = V p 20 ⋅ e α (T − 20 ) −
(1)
m w 20 − mi (T ) mi (T ) − ρ w (T ) ρ i (T )
(2)
α = 3 ⋅ α lin Va Vp Vw Vi T Vp20 mw20 mi
ρw ρi α αlin
is the volume of the air filled pores is the total pore volume is the volume of pore water is the volume of ice is temperature [o C] is the total pore volume at 20 o C is the mass of evaporable pore water is the evolved ice amount is the density of water is the density of ice is the volumetric thermal expansion coefficient of the mortar is the linear thermal expansion coefficient of the mortar
Literature Density of water Literature Density of ice
0.002
1.01
Density [g/cm3]
0 -70 -60 -50 -40 -30 -20 -10
0.96
0
10
20 -0.002 -0.004
Ice Water
0.91
-0.006 -0.008 -0.01
0.86
-0.012 -0.014
0.81 -70
-60
-50
-40
-30
-20
-10
Temperature [o C]
0
10
-0.016
20
Volumetric thermal expansion coefficient
in which
(3)
-0.018 o
Temperature [ C]
Figure 9: The densities and thermal coefficients of volumetric expansion of ice and supercooled water As can be seen in the results of Figs 10 and 11 a small part of the frozen ice amount seems to melt in the temperature range -50…-65 oC which is considered to be caused by the entropy paradox [24-26] when the entropy difference between ice and non-freezing pore water changes sign at -49 oC and the density of water is smaller than the density of ice. As a consequence, when temperature increases from the lowest temperature to -49 oC during the thawing phase, ice amount increases. Pore water of the non-air-entrained mortar M25RH100 is squeezed out of the specimen at the temperature of -57 oC and the whole pore space is filled with unfrozen pore water and ice. This is the other reason for the increase of relative humidity in the calorimeter chamber below the temperature of -45 oC of these concretes in Fig. 7. The
Page
8
other reason for the increase in relative humidity in the chamber is the before mentioned entropy paradox. In the air-entrained mortar M25ARH100 there is still about 15% of the total pore volume air filled at -65 oC. M25RH100
Total pore volume
100
Relative pore volumes [%]
Total pore volume Pore water volume Ice volume Pore air volume
90 80 70
Water
60 50 40
Ice
30 20 10
Air -70
-60
-50
-40
-30
-20
0 -10 -10 0
10
20
10
20
o
Temperature [ C]
M25RH100
Relative pore volumes [%]
Ice volume Pore air volume
-70
-60
-50
-40
-30
-20
-10
11 10 9 8 7 6 5 4 3 2 1 0 -1 0 -2 -3
Temperature [o C]
Figure 10: Relative volumes of the pore phases of test mortar M25RH100 The situation in the salted mortars MS25RH100 and MS25ARH100 is remarkably more severe as can be seen in Figs 12 and 13. Due to the expansion of pore water the pore space is filled with water and ice quite soon after the initial freezing temperature of ice at temperatures of -28 and -27 oC, respectively. Thereafter, the out squeezing pore water freezes on the surface of the test prism and the whole sample is covered and sealed with ice. The forming ice makes the pore system more impermeable and simultaneously the increasing pore water
Page
9
pressure exceeds the tensional capacity of binder paste and large expansive strains are evident in both test mortars in Figs 4 and 5. The decrease of relative humidity in Fig. 8 takes place at the same time as the pore volume air content changes sign in Figs 12 and 13 and pore water is squeezed into the chamber where it freezes on the surface of the mortar prism. Relative humidity increases again below -49 oC as the entropy difference between ice and non-freezing pore water changes sign and previously formed ice begins to melt due to the entropy paradox. It is also noteworthy, that in the temperature range -49…-65 oC during the freezing phase the relative humidity in the sample chamber is above the relative humidity of the thawing phase. The situation is similar both in the salted and unsalted mortars. This is due to the entropy paradox which causes melting of ice during freezing and ice formation during thawing. M25ARH100
Total pore volume
100 90 80 70 60 50 40 30 20 10 0
Relative pore volumes [%]
Total pore volume Pore water volume Ice volume Pore air volume Water Ice Air -70
-60
-50
-40
-30
-20
-10
0
10
20
0
10
20
Temperature [oC]
M25ARH100 Relative pore volumes [%]
30 25 20 15 10
Ice volume Pore air volume
5 0
-70
-60
-50
-40
-30
-20
-10
o
Temperature [ C]
Figure 11: Relative volumes of the pore phases of test mortar M25ARH100
Page 10
It is also interesting to notice that during the freezing phase the maximum strain value of the salted mortars which are sealed by an ice layer is taking place approximately at the same temperature as the entropy difference between ice and non-freezing pore water changes sign when the entropy paradox begins. When the lowest temperature is reached and the thawing phase starts in the salted mortars MS25RH100 and MS25ARH100 there can be noticed an increase in strain caused by the fact that now the coefficient of thermal expansion of ice is about six times larger than that of the mortar. During the thawing phase ice expands six-fold compared to the expansion of the mortar and additional cracking is taking place. It must be noted that when the thawing phase begins the whole test prism is covered with a thick ice layer. Most of this bulk ice melts only in the latter part of the thawing phase. MS25RH100
Relative pore volumes [%]
Total pore volume Water
Ice
-70
Air
-60
-50
-40
-30
-20
100 90 80 70 60 50 40 30 20 10 0 -10 -10 -20
Total pore volume Pore water volume Ice volume Pore air volume
0
10
20
-2 0
10
20
o Temperature [ C]
MS25RH100 4
Relative pore volumes [%]
2 0 -70
-60
-50
-40
-30
-20
-10
-4 -6
Ice volume Pore air volume
-8 -10 -12 -14
o
Temperature [ C]
Figure 12: Relative volumes of the pore phases of test mortar MS25RH100
Page 11
The strain curve of the unsalted test mortars M25RH100 and M25ARH100 is quite opposite when the thawing phase starts. Now the unfrozen pore water of the non-salted mortars contracts much faster than ice expands when temperature rises and the strain curve is below the freezing strain curve. In normal mortars the ice amount compared to the unfrozen pore water amount is smaller as in the situation in the salted mortars and the samples are not covered by a thick ice layer, Fig. 6 and Figs 10-13, and therefore, the expansion of ice does not cause expansion in the test mortar during the first half of the thawing phase of the cycle. Similarly, as the ice amount is increasing during the temperature rise from -65…-49 oC in the thawing phase contraction is taking place in the mortar. MS25ARH100 100 90 80 70 60 50 Total pore volume
Relative pore volumes [%]
Total pore volume
Water
Pore water volume
Ice
-70
Air
-60
-50
-40
-30
-20
-10
40 Ice volume 30 Pore air volume 20 10 0 -10 0 10 20 -20
o
Temperature [ C]
MS25ARH100 6
Relative pore volumes [%]
4 2 0 -70
-60
-50
-40
-30
-20
-10
-2 0
10
20
-4
Ice volume Pore air volume
-6 -8 -10 -12 -14
o
Temperature [ C]
Figure 13: Relative volumes of the pore phases of test mortar MS25ARH100
Page 12
The pore water pressure is calculated by Equation 4. The temperature θ in Equation 4 is presented in Kelvins. o o o pv ∆hwi 1 θ θ c pw − c pi + p − p0 = dθdθ ln (θ − θ 0 ) + ⋅∫ ∫ T ν w pv 0 ν w ⋅ θ 0 ν w θ0 θ0
100
M25RH100 50
-70
-60
-50
-40
-30
-20
-10
2
1
10
0
20 -100 -200 -300 -400 -500
3
-600
Ice evolution M25RH100 freezing
4
-700 -800 -900
25
Stress caused by water expansion
3
0
(4)
0 -70
-60
-50
-40
-30
-20
-10
Mortar stress
0 -25 -50
Tension in water calculated from test result
-75
Stress [MPa]
200
Strain [µm/m] and ice evolution [*10 g/gsample]
R ⋅θ
-100 -125
-1000
Theoretical tension in water -150
Temperature (°C) o Temperature [ C]
p-po=(Rθ/νw) ln pv/pvo
2
1 Water in small tension or compression if pore space is filled
If air pore space is filled then water squeezes to surface
p-po=Rθ/νw ln pv/pvo+∆hwio(θ-θo)/θo/νw+pw,exp
3 If air pore space is filled then water squeezes to surface
p-po=Rθ/νw ln pv/pvo+∆hwio(θ-θo)/θo/νw
Ice
Ice
Water in even larger tension but a relative decrease around -40 oC due to a decrease in permeability
If air pore space is filled then water squeezes to surface
Ice
Ice
Water in larger tension and the positive pore water pressure pw,exp has not become fully effective yet. This is normally the situation when the largest tensional stresses occur on the surface due to Kelvin equation.
p-po=Rθ/νw ln pv/pvo+∆hwio(θ-θo)/θo/νw+pw,exp
4 If air pore space is filled then water squeezes to surface
Ice
Ice
Water pressure caused from the water expansion is in balance with the pressure caused by the entropy difference between ice and nonfreezing pore water. Water underpressure is still increasing.
Figure 14: Freezing mechanism of concrete and an example of the tensions induced into water The density of super cooling water (non-freezing water) in the non-salted mortars is decreasing fast as temperature decreases and the sum of ice and non-freezing water is increasing compared to the total pore volume. Due to the large expanding non-freezing water amount large tensional stresses could be expected to be generated into the mortar matrix.
Page 13
However, also the negative pore pressure which is caused by the entropy difference between ice and unfrozen pore water and the Kelvin equation is very large and this negative pressure always exceeds the positive pressure created by the increasing pore pressure caused by expanding non-freezing pore water and the mortar prism is compressed after the initial ice formation. Expansion of non-freezing pore water due to temperature decrease and the entropy difference-induced negative pore water pressure are hydrodynamic and effect on all unfrozen water, also gel water. This causes a three dimensional hydrostatic contraction into freezing concrete and the phenomenon resembles the thermal strains when homogeneous materials expand or contract due to temperature change, now with a different coefficient of thermal expansion. When the initial freezing takes place the negative water pressure caused by the entropy difference and Kelvin equation dominate and the non-salted mortars contract remarkably and a compressive stress state is induced into the test specimens. It should be noted that the forming stress state of the cross section of the prism is self-equilibrating in the sense that the resultant of the stresses must always equal zero because there are no external loads on the specimen. Therefore the first term on the right side of Equation 4, the so called Kelvin equation, is essentially the reason why freezing induced tensional stresses are created in freezing moist concrete. The value of the Kelvin equation is a function of relative humidity and this humidity and consequent pressure difference between the surface and inside mortar is the cause of the freezing-induced stresses in freezing concrete. The largest tensional stresses are always on the surface of the test sample during freezing [15]. It is very common that the Kelvin equation is ignored in the pressure equations of concrete when freeze-thaw pressures are derived as if these pressures caused by curved water menisci would suddenly disappear when some ice is formed into the concrete or mortar sample even though a large amount of non-freezing water is still in the pore system. If the Kelvin equation is not taken into consideration very small freezing-induced tensional stresses would be generated into freezing concrete and these could by no means cause cracking in concrete. As a good approximation in calculating the stresses induced into freezing concrete the Kelvin equation is quite sufficient. Then the relative humidity in the concrete and in the environment of the specimen should be known. At around -37 oC the permeability of the test mortars has decreased in such an extent that the expansion of non-freezing pore water together with the additional ice amount increase which is taking place at that temperature range causes a small expansion into the mortars. Below -45 oC the expansion is levelled off and these mortars contract rather linearly in a similar manner as the dry reference mortars in the other chamber of the calorimeter. Because of the page restriction it is not possible to present here the calculated pore water pressures caused by the expansion of non-freezing pore water and the negative water pressure originating from the entropy difference and the Kelvin equation by using Equation 4. However, the water pressures of test mortar M25RH100 are presented in Fig. 14 together with the proposed freezing mechanism. It should be emphasized that the stress state in mortar is not governed by the pore pressure as such but the stresses in the freezing mortar or concrete test prism is a function of the pressure difference between surface and middle part of the test specimen. This can be seen in the stress equations presented in Equations 12-19 in reference [15]. Therefore the negative pressure p-po which is increasing with decreasing temperature in Fig 14 does not describe the stress state, for example, on the surface of the specimen. The relative humidity inside the small test mortar should be known during the freeze-thaw cycle.
Page 14
Only then the largest tensional stresses could be calculated. This has been done for test mortar M25RH100 in Fig. 15 by assuming that the relative humidity inside the mortar is 25%-units lower than the chamber relative humidity. However, this is not the situation when the initial freezing takes place because the measured relative humidity values [15] inside concrete decrease much faster and the humidity difference is larger than 25 %. Therefore the theoretical strain curve does not fit very well to the test results when freezing begins. The small expansion at -40 oC is predicted quite well by the theory. The problems arising in the calculations by the proposed theory are the lack of reliable density test results of super cooling water below -34 oC and the lack of data of the water permeability changes in mortars when ice amount is increasing. Actually, the density of super cooling water should be known also as a function of pore water pressure because after the initial freezing temperature there is a large tension in the pore water. The results presented in Figs 10-13 are calculated by using the curve presented in Fig. 9 which is based on data obtained in atmospheric pressure. 100 0 -70
-60
-50
-40
-30
-20
-10
0
10
20
-100
-300 -400 -500 -600
Strain (µm/m)
-200
-700
M25RH100 freezing Theory
-800 -900 -1000
Temperature (°C)
Fig. 15. Theoretical strains in test mortar M25RH100 during the cooling phase. The theoretical strain curve has been calculated by assuming that the relative humidity inside mortar is 25%-units lower than in the chamber. The schematic pictures of Fig. 14 present the proposed freeze-thaw mechanism by a piston and cylinder model in which the unfrozen pore water situated in pores having dimensions under 15 nm is presented schematically in the middle part of the cylinder between ice layers. The proposed freeze-thaw mechanism is self-equilibrating in the sense that, for example, when pressure in the non-freezing pore water decreases in temperatures below the initial freezing temperature this of course changes the density curve of super cooling water to the right in Fig. 9 and, similarly, the curve of volumetric thermal coefficient of expansion also to the right. This balances the pore water pressure until equilibrium has been reached with the negative pore water pressure originating from the entropy difference between ice and nonfreezing pore water together with the negative pore water pressure caused by curved water menisci and calculated by the Kelvin equation. The entropy difference-induced negative pore water pressure behaves like a pre-tensioning or post-tensioning reinforcement in tensioned concrete structures. It causes a hydrostatic compressive stress state which is nearly evenly distributed in the lateral direction of the cross section into freezing concrete so that the test
Page 15
specimen can endure the large expansive forces originating from the expansion of nonfreezing pore water which is also nearly evenly distributed in the cross section perpendicular to the axis of the test prism when concrete freezes. The second term in the right part of Equation 4 represents the hydrostatic negative pore pressure which is caused by the entropy difference between ice surface and the non-freezing pore water. It is a function of only temperature and, therefore, nearly evenly distributed in the cross section because there is only a few degrees temperature difference between the surface and the middle part of the test sample. These two hydrostatic pressures are inner pressures and because there are no external loads acting on the test prism they nearly vanish in the equilibrium equations of the stress state. Kelvin equation is causing large stresses into the specimen. The hydrostatic stresses change the axial strain so that after freezing starts the volumetric contraction has another volumetric ‘thermal’ coefficient of expansion. After ice is formed into the mortar this new contraction is also linear if the Kelvin equation is not taken into consideration. 5.
CONCLUSIONS
A novel freeze-thaw mechanism of porous materials is proposed. The actual cause of the freeze-thaw damage of concrete is not the ice formation in the pores nor the anomalous behaviour of pore water in sub-zero temperatures. Unfrozen pore water expands as temperature decreases. During freezing this fills the pore space and squeezes pore air out of the material. If the initial pore filling degree of the concrete sample is high enough also pore water is squeezed out of the test sample and it freezes on the surface of the test prism. This freezing behaviour can explain the surface scaling phenomenon in contemporary freeze-thaw tests. This squeezing of pore water is also seen in the rise of the relative humidity in the sample chamber. In the freeze-thaw situation the role of ice is to provide an almost evenly distributed compressive stress state into concrete. However, at the same time it decreases the permeability of concrete and this increases the pressure build up in the unfrozen pore water. When thawing phase begins the salted test specimens are covered by a thick ice layer and the relative ice amount is higher than the unfrozen pore water amount compared to the situation in normal unsalted mortars. Therefore, an additional expansion is taking place during the first half of the thawing cycle. This causes further damage and cracking into the mortars. In normal concretes in which there are no salt loads this additional damage is lacking because of the smaller ice amount compared to the unfrozen pore water amount at -65 oC. The proposed freeze-thaw damage mechanism explains also the efficiency of air-entraining in preventing freeze-thaw damage. If the large protective air pores are not filled with water they provide space into which the expanding unfrozen pore water can flow and freeze there freely without causing any freeze-thaw damage. What is then the cause of freeze-thaw damage in concrete? According to the test results the non-salted mortars are in a compressive stress state during the freezing part of the freeze-thaw cycle. However, the stress state in the test mortar prisms cannot be even on the radial direction of the cross section because on the sample surface there is no stress. There exists a similar stress state in the prisms which is presented in reference [15]. This stress state is in equilibrium over the cross section of the prism. The inside of the prisms is in an even larger compression than the surface and, therefore, there can be tension on the surface of the test prism because the relative humidity inside the prism is smaller than in the calorimeter
Page 16
chamber. This is obvious when pressure equation 4 is examined. As a good approximation in calculating the stresses induced into freezing concrete the Kelvin equation is quite sufficient. Then the relative humidity in the concrete and at the surface of the specimen should be known. If the relative humidity inside mortar is, for example, 35 %-units lower than in the chamber then the tensional stress on the surface is about 5 MPa. Unfortunately, the prisms are so small that it is not possible to measure the relative humidity inside mortar, at least using contemporary measuring devices. At -49 oC the density of non-freezing pore water passes below the density of ice and an entropy paradox is taking place when the entropy difference between ice and the non-freezing water changes sign. Below this temperature some of the previously formed ice begins to melt. This can be seen in the ice amounts of the non-salted mortars but not in the salted mortars because the freezing of the expanding pore water at the surface of the test sample exceeds the amount of melting ice. When thawing phase begins the entropy paradox causes an additional increase in ice amount during temperature rise. During the freezing phase of the salted mortars the entropy paradox temperature coincides approximately with the maximum strain value of these mortars covered and sealed by ice. During the temperature range between -15 to 0 oC in the thawing phase of the unsalted mortars the strains curve slightly upwards and as most of the ice melts in this temperature range it is reasonable to assume that this expansion is caused by ice. The thermal expansion coefficient of ice is about six-fold compared to the mortar matrix and, therefore, during the thawing phase ice expands much faster compared to mortar and this can cause tensional stresses into mortar and concrete. Similarly, gel water is drawn out from the CSH-structure due to the large negative pore pressures in the unfrozen pore water. Gel pores have a large surface area and this drying effect causes shrinkage crack formation into the hydrated binder paste. In the salted test mortars the air filled pore space is filled with unfrozen pore water and ice quite soon after ice is crystallizing and the test samples are covered and sealed with ice. Then the expansion of forming ice together with the expansion of unfrozen pore water is so large that this causes large tensional stresses into concrete so that freezing damage takes place already then. The results of this research project show that in three of the four studied test concretes unfrozen pore water and ice fill the total pore space during the freeze test and, thereafter, pore water freezes on the surface of the test specimen. This is caused by the volume change of unfrozen pore water and ice due mainly to the density change of unfrozen pore water. This kind of freezing of pore water on the surface of the test specimen has been previously suggested by Litvan [9] but the reason for this was explained to be caused by the fact that the vapor pressure of super cooled water is higher than the vapor pressure of ice which causes migration of pore water on the surface of the test specimen where it freezes. This is considered rather controversial assumption and it has been largely criticized. In this article unfrozen pore water is considered to be squeezed out of the pore structure only after the whole pore volume is filled with unfrozen water and ice due to the volume increase of unfrozen pore water and only thereafter, water freezes on the surface of the test specimen. Before this the unfrozen super cooled pore water is considered to migrate to the nearest ice surface in the pore system, not to the surface of the test sample. According to the presented freeze-thaw theory in the unfrozen pore water there exists a large tensional stress state which was calculated to be about 150 MPa at -65 oC. The unfrozen
Page 17
pore water is in a double metastable state and it is both super cooled and superheated. At this stability limit the unfrozen pore water is near to freeze and boil at the same time. REFERENCES [1] Powers, T.C., ‘A Working hypothesis for further studies of Frost Resistance of Concrete’, Journal of the American Concrete Society 16 (4) (1945) 245-271. [2] Powers, T.C., ‘The Air Requirements of Frost-Resistant Concrete’, in ‘Proceedings of the Highway Research Board’, Bulletin 33, (Portland Cement Association, 1949) 1-28. [3] Powers T.C. and Brownyard T.L., ‘Studies of the Physical Properties of Hardened Portland Cement Paste’, Journal of the American Concrete Institute 18 (5) (1947) 549-602 and 18 (8) (1947) 933-969. [4] Powers T.C. and Helmuth R.A., ‘Theory of Volume Changes in Hardened Portland-Cement Paste During Freezing’, Proceedings of the Highway Research Board 32 (1953) 285-297. [5] Litvan C.G., ‘Phase transitions of adsorbates III: Heat effects and dimensional changes in nonequilibrium temperature cycles’, Journal of Colloid and Interface Science 38 (1) (1972), 7583. [6] Sidebottom E.W. and Litvan C.G., ‘Phase transitions of adsorbates – Vapour pressure and extension isotherms of the porous-glass+water system below 0o C’, Transactions of the Faraday Society N:o 585 67 (9) (1971) 2726-2736. [7] Litvan C.G, ‘Phase transitions of adsorbates: IV, Mechanism of frost action in hardened cement paste’, Journal of The American Ceramic Society 55 (1971) 38-42. [8] Litvan C.G., ‘Phase transition of adsorbates: V. Aqueous sodium chloride solutions adsorbed on porous silica glass’, Journal of Colloid and Interface Science 45 (1973) 154-169. [9] Litvan C. G., ‘Freeze-thaw durability of porous building materials’ in ‘Durability of building materials and components’, ASTM STP 691 Eds. Sereda P. J. and Litvan C. G., American Society for Testing and Materials (1980) 455-463. [10] Fagerlund, G., ‘The critical degree of saturation method of assessing the freeze/thaw resistance of concrete’, Materials and Structures 58 (1977) 217-229. [11] Setzer, M. J., ‘Micro ice lens formation in porous solid’, Journal of Colloid Interface Science 243 (2001) 193-201. [12] Scherer, G. W., ‘Freezing gels’, Journal of Non-Crystalline Solids 155 (1993) 1-25. [13] Penttala V., ‘Strains and pressures induced by freezing mortars exposed in sodium chloride solution’, Concrete Science and Engineering 1 (1) (1999) 2-14. [14] Penttala V., ‘Freezing-induced strains and pressures in wet porous materials and especially in concrete mortar’, Advanced Cement Based Materials 7 (1) (1998) 8-19. [15] Penttala, V. and Al-Neshawy, F., ‘Stress and strain state of concrete during freezing and thawing cycles’, Cement and Concrete Research 32 (2002) 1407-1420. [16] Penttala, V., ‘Effects of freezing rate on the strains and ice formation in concrete mortar’, in Proceedings of the 2nd Int. Conference on Concrete under Severe Conditions. Tromsø, Norway, June, 1998, 478-488. [17] Penttala, V., ‘Freezing Dilations and Ice Formation in High Performance Concrete Mortars’ in ‘The International Symposium on High-Performance and Reactive Powder Concretes’. Sherbrooke, Canada, August, 1998. Proceedings. Université de Sherbrooke, 245-260. [18] Penttala, V. and Al-Neshawy, F., ‘Ice Formation and Pore Water Redistribution During 2-Cycle Freezing and Thawing of Concrete Mortars’ in ‘Frost Damage in Concrete’, Minneapolis, USA, June 1999. France, Rilem Publications Sarl, RILEM Proceedings PRO 25, 115-126. [19] Landolt–Börnstein, ‘Zahlenwerte und Funktionen aus Physik, Chemie, Astronomie, Geophysik, Technik’. Bd II, 1. Teil, 6. Aufl., 449-453. [20] Hobbs, P. V., ‘Ice Physics’, (Oxford University Press, 1974).
Page 18
[21] Angell, C. A., ‘Supercooled water’ in ‘Water, a comprehensive treatise’, Vol 7, ‘Water and Aqueous Solutions at Subzero Temperatures’ Ed. Franks, F. (Plenum Press, New York, 1982) [22] Zhelesnyi, B. V., ‘The density of supercooled water’, Russian Journal of Physical Chemistry 43 (9) (1969) 1311-1312. [23] Kell, G. S., ‘Density, thermal expansivity, and compressibility of liquid water from 0 o to 150 oC : Correlations and tables for atmospheric pressure and saturation reviewed and expressed on 1968 temperature scale’, Journal of Chemical and Engineering Data 20 (1) (1975) 97-105. [24] Kauzmann, W. ‘The nature of the glassy state and the behavior of liquids at low temperatures’, Chemical Reviews 43 (1948) 219-256. [25] Hendersson, S. J. and Speedy, R. J., ‘Temperature of Maximum Density in Water at Negative Pressures’, The Journal of Physical Chemistry 91 (11) (1987) 3062-3068. [26] Hendersson, S. J. and Speedy, R. J., ‘Melting Temperature of Ice at Positive and Negative Pressures’, The Journal of Physical Chemistry 91 (11) (1987) 3069-3072.
Page 19
