Materials and Structures (2008) 41:12831290 DOI 10.

1617/s11527-007-9326-x

ORIGINAL ARTICLE

Gas permeability and capillary porosity of self-compacting concrete

V. Boel K. Audenaert G. De Schutter

Received: 14 July 2006 / Accepted: 15 October 2007 / Published online: 21 November 2007 RILEM 2007

Abstract Self-compacting concrete (SCC) can be placed without any external compaction avoiding some health risks as well as environmental problems. In order to obtain its key properties, a large amount of ne particles and a new generation of superplasticizers can be used. Earlier research by means of mercury intrusion porosimetry, already pointed out an important difference in pore structure between SCC and traditional concrete (TC). Since the transport properties of concrete are strongly depending on its pore structure, the question rises to what extent the gas permeability of SCC gets affected by the change in mixture design. In this paper the gas permeability of 16 mixtures SCC and 4 mixtures TC are being evaluated with special attention to the difference between SCC and TC, and the inuence of the following parameters: water/cement ratio, powder content, type of ller, neness of the ller, type of aggregate and cement/powder ratio. It was concluded that the gas permeability of SCC is about 5 times lower than the gas permeability of TC. The parameter with the largest impact on the gas permeability seems
V. Boel (&) Faculty of Applied Engineering Sciences-Construction University College Ghent, Ghent, Belgium e-mail: veerle.boel@Hogent.be K. Audenaert G. De Schutter Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University, Ghent, Belgium

to be the water content and secondly the powder content. The capillary porosity has been estimated and a rather good correlation has been found with the gas permeability. sume Le be ton auto-plac tre Re ant (SCC) peut e sans besoin de tassement externe. En tant que coule viter des risques pour la sante tels il est possible de `mes denvironnement. aussi bien que des proble te s principales Pourquon puisse obtenir les proprie ton auto-plac du be ant, on ajoute une grande quantite de particules nes et du superplastiant. Au moyen de trie au mercure, on a de ja ` constate quil y la porosime rence importante concernant la structure a une diffe ton auto-plac ton poreuse entre le be ant et le be nome `nes de transport traditionnel (TC). Les phe pendent largement de la structure poreuse du de ton. Il est donc inte ressant de tudier dans quelle be abilite au gaz sera inuence par le ampleur la perme che es de changement de la composition. Vingt ga ton ont e te pre pare es, repre sentant quatre be tons be tons auto-plac traditionnels et seize be ants. Les essais te realise s an de tudier linuence de quelques ont e `tres : le rapport eau/ciment (W/C), le rapport parame ciment/poudre (C/P), le type de ller (ller calcaire ` cendres volantes), le type dagre gats, le type de ou de poudre. Aussi la diffe rence ciment et la quantite ton traditionnel et le be ton auto-plac entre le be ant est tudie e. On peut conclure que la permeabilite au gaz e est plus ou moins 5 fois plus bas pour le SCC que `tre le plus important semble pour le TC. Le parame

1284

Materials and Structures (2008) 41:12831290

tre la quantite deau et deuxie `mement la quantite de e capillaire a e te estime e et on a pu poudre. La porosite lation pluto t bonne avec la permtrouver une corre abilite au gaz. e Keywords Capillary porosity Gas permeability Self-compacting concrete capillaire Perme abilite au gaz s Porosite Mots-cle ton auto-plac Be ant

Rilem TC 116-PCD: Permeability of concrete as a criterion of its durability (Recommendations). Besides the comparison between SCC and TC, also the inuence of the following parameters is studied: water/cement ratio (W/C), powder content, type of ller, neness of the ller, type of aggregate, cement/powder ratio (C/P) and type of cement. Based on the theory of Powers, for these 20 mixtures an estimation of the capillary porosity is made and its relationship with the gas permeability is investigated.

1 Introduction 2 Mixture proportions The introduction of self-compacting concrete (SCC) in the 1980s brought along some important advantages compared to traditional concrete (TC). As this kind of concrete can be placed without any compaction, some health risks as well as environmental problems can be avoided. Moreover, durability problems related to insufcient compaction of the concrete can be avoided. The three essential properties of SCC are a high owability, a high passing ability and a high segregation resistance. This can be obtained by a large amount of ne particles and the use of superplasticizers. In the most recent international conferences on SCC, the issues which have been studied the most, seem to be the mixture design, workability, rheology and mechanical properties [13]. By means of case studies where SCC has been used in the eld, also the applicability has been demonstrated. However still a lot of questions need to be solved when it comes to the actual performance of the material in general, and the durability in particular. The change in pore structure due to a different mixture design has an inuence on the transport properties and the durability of the material. To what extent this difference matters, has to be investigated thoroughly. The transport mechanisms in SCC, as well as its relation with durability are studied in the Magnel Laboratory for Concrete Research [46]. Regarding the transport properties, the gas permeability is one of the parameters giving an indication of the resistance of concrete against several kinds of deterioration. In this article results on the gas permeability of 16 mixtures of SCC and 4 mixtures of TC is experimentally determined and the results are discussed. The tests are performed by means of the Cembureau-method according to At the Magnel Laboratory for Concrete Research, the gas permeability of 16 mixtures of SCC and 4 mixtures of TC is determined experimentally. Table 1 shows the mixture proportions together with the compressive strength at 28 days measured on cubes 150 mm. The amount of superplasticizer (polycarboxylic ether) was determined in order to obtain a suitable owability without visible segregation. The grading curves of the different types of cement and ller are presented in Figs. 1 and 2 respectively. The rst four SCC mixtures have exactly the same proportions for cement, limestone ller (BETOCARB P2), sand (0/5), gravel (4/14) and water, but the type of cement is varied: Portlandcement (CEM I 42.5 R, CEM I 52.5 and CEM I 52.5 HSR) and blast furnace slag cement (CEM III/ A 42.5 LA). The amount of powder P (cement and ller) is 600 kg/m3. In the next three mixtures (SCC5, SCC6 and SCC7), the amount of water, gravel, sand and powder is kept the same, but the C/ P ratio is varied. The inuence of type of ller is studied by mixtures SCC8 and SCC9 where respectively a limestone ller with a ner grading (superne S) and y ash are used. In the next three mixtures SCC10, SCC11 and SCC12 the total amount of powder is varied (500, 700 and 800 kg/ m3). The amount of water is varied in mixtures SCC13, SCC14 and SCC15 (144, 198 and 216 kg/ m3). The inuence of the type of coarse aggregate is studied by replacing river gravel by calcareous rubble (SCC16). When considering W/C ratio, cement type and cement content, the four TC mixtures correspond with respectively mixtures SCC1, SCC3, SCC6 and SCC4.

Table 1 Mixture design CEM I 52.5 HSR [kg/m3] Sand 0/5 [kg/m3] 853 853 853 853 853 853 853 853 853 923 782 712 865 835 825 816 165 165 165 360 165 640 640 626 640 1225 1225 1200 1225 698 755 640 583 707 683 675 734 698 698 698 698 698 698 698 2.5 2.3 2.2 2.2 2.9 3.0 2.8 2.8 3.4 2.6 2.7 3.6 1.8 2.0 1.8 698 2.3 0.46 0.46 0.46 0.46 0.55 0.41 0.37 0.46 0.46 0.46 0.48 0.49 0.40 0.55 0.60 0.46 0.46 0.46 0.41 0.46 240 240 240 360 300 200 150 240 240 200 300 350 240 240 240 240 216 165 198 144 220 137 192 165 165 165 165 165 240 165 165 165 165 Limestone lter S [kg/m3] Limestone lter P2 [kg/m3] Fly ash [kg/m3] Water [kg/m3] River gravel 4/14 [kg/m3] Calcareous rubble 2/14 [kg/m3] Glenium 51 [l/m3] W/C [-] C/P [- ] Compressive strength [MPa] 0.60 0.60 0.60 0.60 0.50 0.67 0.75 0.60 0.60 0.60 0.57 0.56 0.60 0.60 0.60 0.60 1.00 1.00 1.00 1.00 57.3 68.0 66.1 70.1 46.5 64.2 68.7 56.9 66.2 60.1 55.9 50.9 68.7 46.6 40.3 74.7 48.6 49.7 53.7 50.2

CEM I 42.5 R [kg/m3]

CEM II 52.5 [kg/m3]

CEM III 42.5 LA [kg/m3]

SCC1

360

SCC2

360

Materials and Structures (2008) 41:12831290

SCC3

360

SCC4

SCC5

300

SCC6

400

SCC7

450

SCC8

360

SCC9

360

SCC10 SCC11

300 400

SCC12

450

SCC13

360

SCC14

360

SCC15

360

SCC16

360

TC1

360

TC2

360

TC3

400

TC4

1285

1286
100
CEM I 42.5 R

Materials and Structures (2008) 41:12831290

% passing the sieve

80 60 40 20 0 0.0000

CEM I 52.5 CEM III/A 42.5 LA CEM I 52.5 HSR

0.0001

0.0010

0.0100

0.1000

1.0000

Sieve diameter [mm]

Fig. 1 Grading curve of cement

100
limestone filler P2

% pa ssi ng the si ev e

80 60 40 20 0 0.0000

limestone filler S fly ash

0.0001

0.0010

0.0100

0.1000

1.0000

Sieve diameter [mm]

Fig. 2 Grading curve of the llers

3 Gas permeability of concrete 3.1 Test method The gas permeability is a measure of the ow of gas through a porous material caused by pressure head. It depends on the open porosity prevailing in this material. For this reason it is important to take into account the moisture content of the concrete since the acting gas pressure in the pores is not sufcient to
Fig. 3 Gas permeability meter and detail of a testing cell

move the water and the pores will remain blocked not being able to let the gas pass. Besides the open porosity, other parameters like pore connectivity, tortuosity and isotropy of the pore network have an inuence. For each mixture, one prism 400 9400 9 400 mm3 is made and stored in a climate room at 20 2C and more than 90% R.H. At the age of 28 days, three cores of 150 mm diameter were drilled for each concrete mixture and from the centre of these cores, samples with 50 mm height were taken. The test is performed on the age of 3 months. Until the beginning of the test the samples are stored in a climate room at 20 2C and more than 90% R.H. The testing procedure is carried out following the Cembureau-method according to Rilem TC 116PCD: Recommendations [7]. The testing equipment and a detail of a testing cell are shown in Fig. 3. This apparatus can be used for measuring values of the gas permeability in the range of 10-19 to 10-14 m2. In order to avoid a change of the pore structure during the test, the applied gas should not react chemically with the concrete. For this reason oxygen is used [8]. As mentioned before, the saturation degree of the samples is an important issue with regard to the determination of the gas permeability. The test is performed at three different saturation degrees S (1) of which one is the completely dry state of the sample. The drying of the samples is reached according to a xed procedure, starting from vacuum saturation of the samples after drying them at 80C. At this moment the concrete is completely lled with water (S = 100%) and any ow of gas through the samples is blocked. Further drying of the samples makes it possible to attain two different saturation degrees S1 and S2. Finally the samples are dried at 105C to reach a saturation degree of 0% (S3). The

Materials and Structures (2008) 41:12831290

1287
[m]
-16

Apparent gas permeability kapp, x 10

test is performed at three different levels of inlet pressure for oxygen, namely 3, 4 and 5 bar. In this paper, all the results of the tests with an inlet pressure of 3, 4 and 5 bar and at saturation degree S3 (0%) will be discussed. Sx Msx Ms3 100 % B Ms3 1

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 TC3 TC4 SCC3 TC1 TC2 SCC2 SCC1 SCC4
SCC15 SCC16

3 bar 4 bar 5 bar

Apparent gas permeability kapp, x 10-16 [m]

with Sx the saturation degree at measurement x of the gas permeability [volume%], Msx the mass of the sample at measurement x of the gas permeability [g], Ms3 the mass of the oven dry sample, measured at the last measurement of the gas permeability [g] and B the mass of the vacuum saturated sample weighted on air [g]. 3.2 Test results The apparent gas permeability kapp can be calculated by means of relationship (2) based on the HagenPoiseuille relationship for laminar ow of a compressible uid through a porous body with small capillaries under steady-state conditions [9]. kapp 4.04 P2 Q L 1016 2 m 2 A(P2 1 P2 ) 2

Fig. 4 The apparent gas permeability (dry samples)part 1

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 SCC6 SCC5 SCC7 SCC8 SCC12 SCC13 SCC1 SCC11 SCC9 SCC14 SCC10 3 bar 4 bar 5 bar

Fig. 5 The apparent gas permeability (dry samples)part 2

with Q the volume ow rate of the uid measured during the test with a bubble ow meter [ml/s], L the thickness of the specimen in the direction of the ow [m], A the cross-sectional area of the specimen [m2], P1 and P2 the inlet and outlet pressure of oxygen [bar]. This formula is valid for tests performed at 20C for which the viscosity of oxygen is 2.02.10-5 Nsm-2. The coefcients of apparent gas permeability achieved by performing the test with an inlet pressure of 3, 4 and 5 bar and on completely dry specimens, are plotted in Figs. 4 and 5.

3.3 Inuence of the concrete composition From Fig. 4, it follows that the apparent gas permeability of all SCC mixtures is only about 20% of the gas permeability of the corresponding TC mixtures. The difference is pronounced the most when the saturation degree is zero, because the higher the saturation degree, the more the trespassing gas is blocked by water-lled pores. The important

difference between SCC and TC is due to a different pore structure. A higher porosity and critical pore size were found by means of mercury intrusion porosimetry tests on cement pastes which were designed according the mixture proportions of concrete mixtures presented in Table 1 [10]. This difference might have an inuence on the assumption of laminar ow through the pore system, as is explained in [5]. More research however on this topic is in progress. When blast furnace slag CEM III A 42.5 LA or Portlandcement CEM I 52.5 HSR are used instead of Portlandcement CEM I 42.5 R, a higher apparent gas permeability is noticed for traditional concrete. This effect is not noticed for the SCC mixtures, merely the opposite is observed. The inuence of the type of cement (blast furnace slag cement versus portland cement) seems to be somewhat lower than the inuence of the neness of the cement (42.5 vs. 52.5). When the amount of cement is increased from 360 kg/m3 to 400 kg/m3, there is a signicant decrease of the apparent gas permeability for traditional concrete, induced by a lower porosity because

1288

Materials and Structures (2008) 41:12831290

Apparent gas permeability kapp, x 10-16 [m]

of the lower W/C ratio [10]. For the corresponding SCC mixtures SCC1 and SCC6, there is only a small difference in apparent gas permeability. For mixtures SCC5, SCC1, SCC6 and SCC7 (Fig. 5) the C/P ratio is varied from 0.50 to 0.75 with a constant amount of powder and water. This means that an increasing C/P ratio equals decreasing the W/C ratio, which densies the microstructure by diminishing the capillary pores [10]. As such, the gas permeability decreases with an increasing C/P ratio. Using a ner ller of the same type (SCC8) gives a more dense microstructure and for this a lower gas permeability. When replacing limestone ller by y ash (SCC9), the puzzolanic properties of y ash cause a more dense microstructure in favour of a lower gas permeability. When the powder content is increased from 500 kg/m3 (SCC10), over 600 kg/m3 (SCC1) and 700 kg/m3 (SCC11) to 800 kg/m3 (SCC12), the gas permeability increases. The C/P, W/C and W/P do not change that much, but the total amount of paste per m3 increases, making the concrete more porous. Mixtures SCC13, SCC1, SCC14 and SCC15 have an increasing W/C ratio, from 0.40 to 0.60. The amount of cement is kept the same, but the amount of water is varied. The higher the amount of water, the bigger the surplus of water not used by the hydration process, causing a higher capillary porosity [10]. According to the experiments, increasing the W/C ratio indeed gives an increase in gas permeability. Mixtures SCC5 and SCC14, have the same W/C ratio (0.55), achieved by changing the cement content as well as the amount of water. The apparent gas permeability of SCC14, having the highest water content of both mixtures, has a much higher apparent gas permeability. Using calcareous rubble 4/14 (SCC16) instead of round river gravel 4/14 (SCC1) gives a lower gas permeability. Possibly the pore structure of the matrix around the calcareous rubble is ner due to the difference in surface texture compared with gravel.

the moisture in the sample is homogeneously distributed. The drying procedure mentioned above was carried out in such way that this assumption was met. kr (s,P) ksx ks0 3

with k(s=x) and k(s=0) respectively the apparent gas permeability at a certain inlet pressure at a certain saturation degree (s = x) and at the completely dry state (s = 0) of the sample. Van Genuchten, mainly active in the eld of soil physics, proposed the following equation to relate the relative gas permeability with the saturation degree [11]. p kr (s) (1 s) (1 sb )2=b 4 with b a constant parameter and s the saturation degree. It was investigated whether the results obtained on the gas permeability for the 20 mixtures also correspond to Eq. 4. For every mixture, at three different inlet pressures, the four different values of the relative gas permeability were calculated and plotted in a graph (Fig. 6). When the model according to (4) was plotted on the same graph, a resemblance of the shape of the experimentally determined curve was noticed. The values of b typically are in the order of 1 and tend to have slightly higher values for self-compacting concrete than for traditional concrete. From these results follows that it is possible to determine the gas permeability at a certain saturation degree once the gas permeability at the dry state is known, according Eq. 5.

0.6
3 bar - calc 3 bar - exp 4 bar - exp 5 bar - exp

0.5 0.4
P1

4 bar - calc 5 bar - calc

0.3 0.2 0.1 0

3.4 Inuence of the saturation degree The inuence of the saturation degree at a given inlet pressure can be investigated by means of the introduction of the relative gas permeability kr(s,P) (3) [11]. For this purpose the assumption is made that

0.2

0.4

0.6

0.8

Saturation degree [%]

Fig. 6 Apparent gas permeability at 3 different inlet pressures P1 as a function of the saturation degree

Materials and Structures (2008) 41:12831290

1289

ksx ks0

p (1 s) (1 sb )2=b

For this reason, in the next chapter the potential to theoretically determine the gas permeability as a function of the mixture design, the hydration degree and the inlet pressure is investigated.

4 Capillary porosity A strong link has been found between the capillary porosity and the water permeability for the mixtures mentioned in Table 1 [4]. This fact is also described in literature, namely that the transport properties of concrete are mainly determined by the capillary pores [12, 13]. In this paper it is veried whether there is also a possible link between the calculated capillary porosity and the gas permeability. The calculation of the capillary porosity is based on Powers model. Free water is the initial amount of water (water) minus the gel-water and the bounded water. Vcap capillary pores free water capillary pores water gelwater bounded water Ch W 0.28 0.185 qc qw 0.72   Ch 0.23 C h 0.23 C h 1 0.185 qc qw qw Ch 1 0.1319 W 0.3194 C h qc qw

sand [kg], F the amount of ller [kg] and qagg the mass density of aggregate [kg/m3]. From the mixture composition, the parameters W, C, A, S and F are known. For the mass densities, a value of 1000 kg/m3 is used for water, 2650 kg/m3 for the aggregates, sand and ller, 3100 kg/m3 for Portland cement and 2950 kg/m3 for the blast furnace slag cement. The density of the llers was measured and turned out to be 2650 kg/m3. The densities of river sand, river gravel and calcareous rubble equal respectively 2625 kg/m3, 2625 kg/m3 and 2689 kg/m3. The test specimens are stored until the testing age (90 days) in a climate room at 20 2C and at least 90% R.H., and are taken from the centre of a prism. This means that the degree of hydration will not strongly differ from the ultimate degree of hydration that could be determined, in case of Portland cement, by the formula of Mill [14]: hultim 1.031 W=C 0.194 W=C 9

with Vcap the volume of capillary pores [m3], C the amount of cement [kg], W the initial amount of water [kg], h the degree of hydration [-], qc and qw the mass density of respectively cement and water [kg/ m3].

The predicted value is used for all compositions based on CEM I 42.5 R. CEM I 52.5 has a higher neness, which leads to a slightly higher degree of hydration. For this mixture a 5% higher value is used (i.e. from 0.72 to 0.77). Due to the higher neness of blast furnace slag cement and the prolonged hydration of this kind of cement, a 10% higher value is used for this kind of cement (i.e. 0.82 instead of 0.72). For the mixture with y ash as ller material, the W/C ratio in formula (9) is replaced by the effective W/C ratio with an efciency factor of 0.3. In Fig. 7, the gas permeability for completely dry specimens at an inlet pressure of 3 bar is plotted against the calculated capillary porosity. A good
4.0

Vconcrete Vwater Vcement Vcoarse aggregate Vsand Vfiller 3.0 W C A S F 2.5 qw qc qagg 2.0 7
1.5 1.0 0.5 0.0 0

Apparent gas permeability kapp, x 10-16 [m]

3.5

SCC TC

R2 = 0.92

Vcap capillary porosity Vconcrete

10

12

14

Capillary porosity [%]

with Vconcrete the volume of concrete [m3], A the amount of coarse aggregate [kg], S the amount of

Fig. 7 Apparent gas permeability versus the calculated capillary porosity, for SCC and TC

1290
2.0

Materials and Structures (2008) 41:12831290

References
3 bar 4 bar 5 bar
P1

Apparent gas permeability kapp, x 10-16 [m]

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 4

10

12

Capillary porosity [%]

Fig. 8 Apparent gas permeability at 3 different inlet pressures P1 versus the calculated capillary porosity (dry specimens)

correlation (R2 = 0.92) could be noticed for selfcompacting concrete. Unlike the results on the water permeability, the results of the TC mixtures do not depend in the same way on the capillary porosity as the SCC mixtures. Graphs based on the results with different inlet pressures give the same conclusions as can be seen in Fig. 8. There is a downward shift of the curve when the inlet pressure increases. Further research on modelling the gas permeability as a function of capillary porosity is carrying on.

5 Conclusions Based on an experimental programme, carried out on 16 self-compacting concrete mixtures and 4 traditional concrete mixtures, with variation of C/P ratio, type of ller, type of cement, total amount of powder, W/C ratio and type of aggregate, the following conclusions can be obtained. The most important ndings are that self-compacting concrete is a cementitious material with an important different behaviour in gas transport. The parameter with the largest impact on the gas permeability seems to be the water content and secondly, the powder content. The capillary porosity has been estimated and a rather good correlation has been found with the gas permeability for the SCC mixtures. There is also a possibility to determine the gas permeability as a function of the saturation degree.

1. Proceedings of the rst North American conference on the design and use of self-consolidating concrete, Evanston, USA, 1213 November 2002 2. Proceedings of the 3rd international conference on SCC (RILEM), Reykjavik, 17/8/200320/8/2003 3. Proceedings of the second North American conference on the design and use of self-consolidating concrete and the 4th international conference on SCC (RILEM), Chicago, USA, 31 October3 November 2005 4. Audenaert K, Boel V, De Schutter G (2004) Water permeability of SCC. In: Proceedings of the conference advances in concrete through science and engineering, Evanston, CD-ROM, RILEM, paper 39 in session Durability, 9 pp 5. Audenaert K, Boel V, De Schutter G (2005) Relation between gas permeability and carbonation of self-compacting concrete. In: Proceedings of the international conference on concrete and reinforced concrete, Moscow 6. Boel V, Audenaert K, De Schutter G, Heirman G, Vandewalle L, Vantomme J, dHemricourt J, Ndambi JM (2005) Experimental durability evaluation of self-compacting concrete with limestone ller. In: Shah SP (ed) Proceedings of the second North American conference on the design and use of self-consolidating concrete and the fourth international Rilem symposium on self-compacting concrete, Chicago, 31/10/20052/11/2005, Hanly Wood Publication, pp 297304 7. Rilem TC 116-PCD (1999) Permeability of concrete as a criterion of its durability, recommendations. Mater Struct 32:174179 8. Kropp J, Hilsdorf HK (eds) Rilem report 12: performance criteria for concrete durability 9. Kollek JJ (1989) The determination of the permeability of concrete to oxygen by the Cembureau methoda recommendation. Mater Struct 22:225230 10. Boel V, Audenaert K, De Schutter G (2006) Pore size distribution of hardened cement paste in self-compacting concrete. In: Proceedings of the 7th CANMET/ACI international conference on durability of concrete, Montreal ` s M (2004) Equivalent gas permeability 11. Verdier J, Carcasse of concrete samples subjected to drying. Mag Concrete Res 56(4):223230 12. CEB-FIP MODEL CODE 1990 (1993) Design Code, Thomas Telford, London, England 13. Marsh B, Day R, Bonner D (1985) Pore structure characteristics affecting the permeability of cement paste containing y ash. Cement Concrete Res 15:10271038 14. Van Breugel K (1991) Simulation of hydration and formation of structure in cement based materials, PhD thesis, TUDelft, pp 5457