You are on page 1of 10

Materials and Structures/Mat6riaux et Constructions, Vol.

34, December2001, pp 619-628

Permeability and pore structure evolution of silico-


calcareous and hematite high-strength concretes
submitted to high temperatures
C. Gall6 andJ. Sercombe
Direction de I EneN~eNuclda~reD PC/S C CME CEA Saclay, 91191 Gif-sur-Yuette, France.

Paper received:January2, 2001; Paperaccepted:April 17, 2001

A B S T R A C T R I~ S U M I~

Two high-strength concretes (HSC) only differing in Deux b~tons a hautes pe~rmances se distinguant uni-
their aggregates - silico-calcareous and hematite - were quement par leurs granulats - silico-calcaires et h6matites -
heated at temperatures up to 450~ (l~ The evo- ont subi des traitements thermiques allant jusqu'a 450~
lution of their microstructural parameters - porosity, pore (l~ L'&olution de leurs propri~t& microstructurales
structure, permeability - were analysed. Both concretes -porosit6, structure des pores, perm&bilit~ - a ~tg exami-
showed equivalent initial microstructural characteristics. n&. Initialement, les caract&istiques microstructurales des
From 60~ heating generated a large capillary porosity deux b6tons sont identiques. D& 60~ le chauffage pro-
characterized by pore accesses around 0.1 gm. The inten- voque l'apparition d'une importante porosit~ capillaire dont les
sity and especially the width of the porosity peaks acc& de pore se situent au voisinage de 0,1 btm. L'intensit~ et
increased with temperature. For silico-calcareous HSC, surtout la largeur des pics de porosit~ augmentent avec la tem-
macropores - 50 to 0.3 g m - were detected by MIP stud- p&ature. Pour le b6ton silico-calcaire, des macropores -dans
ies at 250~ and especially at 450~ They were corre- la gamme 50 a 0.3 btm - ont ~galement ~t~ mis en &idence
lated to microcracks visually observed at the surface of the par la porosim&ie mercure a partir de 250~ et surtout
probes. Up to 250~ the intrinsic permeability increased 450~ Ceux-ci ont ~t~ corral& a des microfissures d~tect&s
similarly for both concretes. Between 250 and 450~ visuellement a la su~ce des ~prouvettes. Jusqu'a 250~ on
permeability remained stable for hematite HSC while, for observe pour les deux b&ons une augmentation de la perm&-
silico-calcareous HSC, a major change was noticed. A bilit~ au gaz. Entre 250 et 450~ ce param~tre n'&olue
good correlation between permeability and total water pas pour le b6ton d'h~matite alors qu'une &olution majeure
porosity was observed. At 450~ influence of the micro- est de nouveau observe'epour le b&on silico-calcaire. Ces &o-
cracks on permeability was greater than the impact of the lutions sont bien corr~l&s a celles de la porositd totale a l' eau.
increase of capillary pore size. As both concretes showed En tenant compte de la similitude microstructurale initiale de
similar initial microstructural features, conclusion was la phase cimentaire des deux b6tons, cette difl&ence de com-
reached that the differential behaviour can mainly be portement a 6t~ principalement attribu& a des diff&ences de
attributed to internal thermal gradients discrepancies gradients thermiques internes li~es au type de granulats, l'h~-
related to the type of aggregates: hematite allows to limit matite permettant de limiter les gradients thermiques et par
thermal gradient and thus, thermo-mechanical stresses. It cons6quent les contraintes d'origine thermo-m&aniques.
was globally observed that damage due to high tempera- Globalement, l'endommagement lid aux traitements ther-
ture thermal treatments was lower for hematite HSC. miques est moins cons6quent pour le b6ton d'hdmatite.

1. INTRODUCTION material [1, 2], but in other specific environments, such


as the storage of high level nuclear waste packages, con-
For several decades now, high-strength concretes crete structures could be sujected to relatively high tem-
(HSC) have been used more and more frequently in var- peratures - 200~ - in case of failure of cooling systems.
ious industrial domains. As a result, its extended applica- At high temperatures, high-strength concretes are sub-
tion has been associated to new using conditions such as ject to significant damages like microcracking, fractur-
severe thermal environment. Fire remains one of the ing, spalling [3-9]. Due to their limited porosity (com-
most serious potential risk for any building or structural pactness) and low permeability, high-strength concretes

1359-5997/01 9 RILEM 61 9
Materials and Structures/Mat6riaux et Constructions,Vol.34, December2001

in severe thermal environment often behave in a brittle Table 1 - Thermo-mechanical properties of aggregates
way [7]. It is generally admitted that HSC behaviour at Properties Silico-cal. Hematite
high temperature is governed by thermo-mechanical
processes related to temperature gradients which gener- Specific density (kg/m 3) 2500 5000
ate thermal expansion gradients [10, 11] and by thermo- Thermal conductivity (W/m/~ 2.5 11
hydric processes linked to water movements in the Thermal capacity (J/kg/~ 800 650
material porosity [9, 12, 13]. For the first aspect, con-
Thermal expansion coefficient (1/~ 11.10 -6 8.10 -6
crete thermal conductivity and expansion - widely
dependent on the characteristics of its components: Elastic modulus (GPa) 70 209
cement paste and aggegates - are major influential fac-
tors [1, 13, 14]. For the second aspect, permeability is
one of the key parameters controlling internal fluid Table 2 - Concretes mix design (kg/m 3)
transfer and thus the generation of potential over-pres-
Ordinary HSC
sure gradients during temperature heating. It still
remains difficult nowadays to know which process is Silico-calcareous aggregates 8/20 mm 800
predominant. It obviously depends not only on material Silico-calcareous aggregates 4/12 mm 400
properties but also on the geometry of the structure and Silico-calcareous sand 0/4 mm 750
the heating rates. Water 150
In this paper, a comparative study of the thermal Cement CEM I 350
behaviour of two HSCs which differ only by the type of Plasticizer (Gienium 27) 1.75
aggregates is presented. It is carried out in the frame- Hematite HSC
work of a research programme on long-term surface Hematite aggregates 6/25 mm 1700
storage c o n d u c t e d by the French Atomic Energy Hematite aggregates 0/8 mm 900
Commission (CEA). The mix-design properties of the Hematite sand O/1 mm 1000
two HSCs are chosen in order to dissociate the thermo- Water 150
hydric and thermo-mechanical processes occuring dur- Cement CEM I 350
ing heating: similar cement content and water/cement Plasticizer (Glenium 27) 5.25
ratios are used to obtain close transfer properties -
reflected by porosity, pore size distribution and gas per-
meability - (and thus thermo-hydric processes of similar
magnitude), hematite aggregates are used in one case to Ordinary and hematite HSC mean compressive strength
increase the global thermal conductivity of the material at 60 days were of 63 and 78 MPa, respectively.
and thus limit thermo-mechanical phenomena during
heating. To evaluate the consequences of the differential
thermo-mechanical behaviour of the two HSCs, the 2.2 Experimental heating conditions
experimental programme focuses on the evolution of the
materials pore structure identified by MIP in close rela- High-strength concrete probes were heated at 60,
tionship with their permeability. Results are presented 110, 250 and 450~ in a Carbolite oven during 75, 40,
and discussed in this paper for specimen heated to tem- 20 and 15 days on average, respectively. Durations of
peratures up to 450~ thermal treatments were adapted to obtain a stable mass
loss (< 0.01%/24h) in order to ensure that water was
totally removed for each temperature. During heatings,
2. EXPERIMENTAL PROGRAMME heating rates were kept at l~ At the end of the
heating tests, temperature was cut off and probes were
2.1 Materials cooled slowly in the ovens until reaching 60~ Then
materials were stored in sealed containers until the
The materials used for this study were two CEM I beginning of the experimental characterization.
(OPC) concretes. Specimens were French "11 x 22"
normalized cylindrical probes (Ol13 mm, H220 mm).
Ordinary HSC was prepared with silico-calcareous 2.3 Equipments and measurements
aggregates while hematite HSC was prepared with
hematite aggregates. At present, these concretes are stud- Mercury intrusion porosimetry (MIP) tests were per-
ied as structural materials in the context of long-term formed with a Micromeritics Autopore III 9420 appara-
surface storage [15]. Average thermal and mechanical tus. This technique, w h i c h is widely used for
properties of the two types of aggregates [16] are given in microstructural investigation of cement-based materials,
Table 1. Hematite aggregates were used for their high consists in injecting, isostatically, incrementally and
thermal conductivity and their reduced thermal expan- under very high pressure, a non wetting fluid (mercury)
sion coefficient. into the evacuated pores of the porous material. The test
The materials were cured in sealed bags for a period of is analyzed according to the Washburn equation in
two months. Concretes mix design are given in Table 2. which the size of penetrated pore accesses, assimilated to

620
Gall~,Sercombe

cylindrical capillaries, are inversely proportional to the


applied pressure:
p= 4ycos0 (1)
d
where P is the mercury injection pressure (Pa), y is the
surface tension of mercury (N/m), 0 is the contact angle
between solide and mercury (o) and d is the pore access
diameter (m).
For calculations, a contact angle of 130 ~ was
assumed. MIP tests were achieved with a minimum and
maximum pressures of 0.006 and 413 MPa, respectively.
For each temperature, MIP investigation was Carried out
with at least ten specimens. In addition, for non heated
specimens, the removal of water was achieved by using
the freeze-drying technique.
Gas permeability measurements were achieved with a
Hassler apparatus (Fig. 1). This device is a constant-head
permeameter type [17] originally developed for the oil
industry. It allows to apply gas injection pressures up to 5
MPa. Permeability tests were performed on "11 x 22"
cylindrical probes using dry nitrogen gas. Gas flow rates
were measured with bubble flowmeters.
Effective gas ~ermeability
r k g was calculated with
Darcy's law as modified by the Hagen-Poiseuille relation:
2QLP2~
kg -- S(P12 _ p22) (2)
Fig. 1 - S c h e m a t i c o f the Hassler gas permeameter.
where k_ is the effective gas permeability (m2), Q is the gas
flow rate measured at pressure P2 (m3 s-l), L is the speci-
men length (m), ~t is the dynamic gas viscosity (Pa s), S is
the specimen cross sectional area (m2), Pl is the absolute Table 3 - HSC's initial properties (20~
injection pressure (Pa) and P2 is the exit pressure (Pa). Ordinanj Hematite
Intrinsic permeability k was determined using the
Specific density (kg/m 3) 2378 _+8 4067 _+58
Klinkenberg approach [18, 19]. This widely-used
Total porosity (%) 10.4 + 0.4 10.0 + 0.5
method allows to estimate an intrinsic permeability from
Effective permeability (10 -19 m 2) 1.6 _+ 1.9 3.9 + 4.7
effective permeability measured at different pressures,
Intrinsic permeability (10 -20 m 2) 4.5 _+ 7.2 8.1+ 7.3
according to the following equation:
Thermal conductivity (W/m/~ 2.7_+0.1 7.3 + 0.4
Compressive strength (MPa) 63_+ 4 78_+ 5
kg=k(l+p) (3)
Elastic modulus (GPA) 49_+ 4 81+2
Poisson's ratio 0.23 + 0.05 0.26 _+0.04
where k_ is the effective gas permeability (m2), k is
the mtransac permeabilty (m2), P is the average pressure
(PI + P2)/2 and [3 is the constant of Klinkenberg.
The intrinsic permeability, as defined in the these conditions, the initial microstructural properties of
Klinkenberg approach, is regarded as independent from the concretes cementitious phases were expected to be
the average pressure and as a characteristic of the porous almost identical. This assumption was confirmed by the
medium alone. Gas permeability measurements were measurements of mass loss and, from the microstructural
achieved on five specimens for each temperature with at point of view, by porosity and permeability data.
least three gas injection pressures per specimen. The total amount of water - free water, adsorbed
water on pores and hydrates bound water - removed after
the different thermal treatments were close (Fig. 2) and
3. EXPERIMENTAL RESULTS showed that the dehydration stages were comparable for
both concretes. Moreover, average cumulative (Fig. 3) and
3.1 Initial characteristics differential (Fig. 4) pore volume distributions were simi-
lar for both HSCs. A bigger discrepancy is observed on
Initial microscopic and macroscopic properties for cumulative curves. But it is rather limited if we compare
both concretes are given in Table 3. The two HSCs were the standard deviation of total cumulative volumes for
prepared with different aggregates but with the same ordinary and hematite HSC, 0.050 + 0.011 ml/g and
CEM I (OPC) cement and water/cement ratio (0.43). In 0.031 + 0.015 ml/g,respectively. These results proved that

621
Materials and Structures/MaMriaux et Constructions, Vol. 34, December 2001

16 0.1
0 Ordinary HSC X Hematite HSC
-~ 0.09 __2o~ .... 6o~163--11O~ Ordinary HSC I
14 .'. ". "250...C. ~ 450....C .. /1111111 IIIILI4'
A

- 12
~ 0.08 IIIIIIlll
0.08
0.07
IIIIIIll
IIIIIII
IIIIlll
IIIIII iiJiiiill IIIIIII ~TIII
Illlll IIIIllll I IIIIIIb Ililll,r
O
~. .05
IIIIIIII IIIIIII IIIIII IIIIIIIII IIII1,ti -'Ill,U,I*
81o 0.04 11111111 IIIIIII 111111 IIIIIIIII IZli.l,' ~11~
.~ 0.03 IIIIIIII IIIIIII IIIIII IIIIIII1~ All
8 0.02 IIIIIII1: IIIIIII IIIIII lluv,.t'rl ,,.t1'llLI. "1111111
~0.01 11111111 IIIIIII ~ ~ 1 1 IIIIIII
o 0 !1111111.....Ultld~=~-.......i.. ZII ZII
0 50 100 150 200 250 300 350 400 450 500 1000 100 10 1 0.1 0.01 0.001
Temperature (~
Pore access diameter ( p m )

Fig. 2 - Total mass loss evolution versus temperature. "~ 0.1


g 0.09 ~j...... 20~163 ' -- -. (30;'C m l i0';C ~ Hematite HSC[
250~ 45~176 Jlrllll i iiill i '
0.08 i! ............
0.06 ,liliii
~=0.07 I,IIIII ,,ll ,,17m,llll m,ii i,,llll
,,,lll,l, lll,,I]ll II, II
,,11
.... Cumulative - Ordinary HSC]
E
>
0.05 0.08
-- Cumulative" Hematite HSCI IW'I ~. 0.05
0.04 => 0.04
11111111 IIIIII IIIIIIIII IIIiill i IIIIIIJ,I-" II11.1.1
,1t11 0.03 11111111 IIIIII Ililllllllllllll ! II~IL,.-:I~.I.H
== 0.03
E0.02
IIIII IIIIIIIII IIIIIII i ZILI,,J,'P''r I~
>= 0.02 0.01 11111111 IIIIII IIIIIIIII IIIIIII "~IIL.-.TIIIII
0 !1111111 lltiil !!!!!!!! !~ "lllllTII IIIIII
0.01 gglttt.... Igl 1000 100 10 1 0.1 0.01 0.001
o 0 Pore access diameter (IJm)
1000 100 10 1 0.1 0.01 0.001
Pore access diameter (pm)
Fig. 5 - Average cumulative pore volume versus temperature.

Fig. 3 - hfitial HSC pore structure: average cumulative pore volume.


0.08
I o n ' ; nia~] L,a "- g'{~J
0.06 0.07
0.05 ~ .... Differential-Ordinary HSC ] A 0.06 IIIII I iMBiMIi
0.05 ;!llllglllllllll
== 0.04
I[ .-.'--.'.Differential
. - Hematite HSC [
0.04 I ,,,,i m,lllll,,Hi i l[l!!!llllllrlt
E
0.03 0.03 IIIIIIII IIIIIIIII IIIIIII ! ,.t;tilllMI
o
0.02 IIIIII IIIIIIII IIIIIIIII IIIIIII i: Illl!i~iiiid311
0.02
0.01 IIIIII IIIIIIU JIIllli~q~, IIl~l'-'t=i~iilltll
OOl Illllll lltl.tl".,!
LH;4 0 IIIIII =[tt~ ~..~-- iilllllllllllllll
0
.. 4#1" IIIIIII 1000 100 10 1 0.1 0.01 0.001
Pore access diameter (pm)
1000 100 10 1 0.1 0.01 0.001
Pore access diameter ( p m )
0.08 ....... 60~ -- 11o~ I Hematiie Hsc]
Fig. 4 - lnJdal HSC pore structure: average cumulative pore volume. 0.07
0.06
--450~ ,,,,,, IIliiiii iiiiiiii
g [lllilll iillllll llllllll
_g' 008 iii[iii iiiiiiii ~L!tt iiiiiiii
r"t

the cement paste porous space structuring of the two 004 lltlltl ZII I~.I !!!!!!#
HSCs was equivalent. This point is important for further lo
003 ~,',',~,~,', J!!!!!!V11111F~.~!~!
interpretation of the concretes differential behaviour. 002 """" iiiiiiJ/;/llllllr~.~.~
On the contrary, thermal conductivities of the two 0.01 ;~;;;;; ;
0 ~""" ~ r " ,-~ ~11T]
HSCs differ by a factor of 3. Elastic moduli are also
strongly dependent on the mechanical characteristics of 1000 100 10 1 0.1 0.01 0.001
aggregates and thus differ by nearly a factor 2. As Pore access diameter (pro)

expected, hematite aggregates provide a high thermal


Fig. 6 - Average differential pore volume versus temperature.
conductivity and rigidity to hematite HSC.

initial pore size distributions was already mentioned.


3.2 Pore structure Without thermal treatment, both HSCs showed no sig-
nificant capillary porosity (Fig. 4) - macroporosity (pore
Pore structure evolution versus temperatures for both access around and up to 0.1 I.tm). Two main classes of
HSCs are given in Figs. 5 and 6 (average data based on at pore access corresponding to C-S-H intrinsic porosity,
least ten specimens each). The similarity of the concretes are generally detected in regions between 0.01-0.02 ~tm

622
Gall~, Sercombe

and around a few nm. These pore access classes can be


associated with outer C-S-H (microporosity) and inner A 14
I-<~ Ordinary HSC'-U-'Hematite fiSC
C-S-H (nanoporosity), respectively [20]. Cumulative =~13
intrusion curves globally showed a constant increase of
the total intruded volume as temperature rises. Similar /F
/
results were observed by other authors for mortars and /
concretes heated at high temperatures [6, 7, 21, 22]. This
can be related to the alteration of the porous space and .... 41k.~176

thus to the development of additional porosity. At


equivalent temperatures, ordinary HSC intrusion vol- 8
umes are significantly higher than those of hematite 0 50 100 150 200 250 300 350 400 450 500
HSC. This reveals a higher preferential damaging for Temperature (~
ordinary HSC. However, for heterogeneous materials Fig. 7 - Total water porosity evolution versus temperature.
like concrete, the total intruded volume does not reflect
the real porosity of the samples as the amount of non
porous aggregates and paste can vary from a sample to Table 4 - Relative increase (%) in porosity associated to
another. Additional comparative interpretations will be macropores (> 0.3 ~tm), capillary pores (0.3 - 0.02 ~tm)
provided with water porosity data analysis at the end of and hydrate pores (< 0.02 ~tm) versus temperature
this section. OrdinaryHSC Macropores Capillarypores Hydratespores
Differential curves provided more interesting results (> 0.3 p m) (0.3 - 0.02 p m) (< 0.02 p m)
concerning pore space thermal degradation. For the two 60~ -2.2 +55.4 -32.4
HSCs, the various thermal treatments generated an impor-
110~ -6.6 +55.5 -25.7
tant capillary porosity in which major pore access classes
are centered around 0.1 Hm. This phenomenon is detected 250~ -0.9 +59.0 -26.0
early for quite moderate temperatures such as 60~ On 450~ +22.6 +69.5 -27.8
the contrary, pore accesses associated to C-S-H porosity
Hematite HSC Macropores Capillarypores Hydratespores
seem to be influenced by temperature treatments in a (> 0.3 p m) (0.3 - 0.02 p m) (< 0.02 p m)
moderate way. For both HSCs but more especially for
60 ~ -6.0 +62.8 -32.1
hematite concrete, nanopores representativeness is smaller
after heating. This is probably linked to a reduction of pore 110~ -3.0 +62.3 -39.2
network connectMty. Similar observations were reported 250~ -4.8 +78.7 -26.7
for hardened cement pastes [23].
450~ +10.0 +116.8 -32.9
The evolution of the created capillary porosity is char-
acterized by an enlargement of the porosity peak as tem-
perature rises. This phenomenon was observed for both until stable mass loss. Then, the probes were water resat-
HSCs. At 60~ peak boundaries are located around 0.3 urated under vacuum and oven-dried again at 60~
and 0.02 tim while, at 250~ these boundaries vary Ordinary HSC total porosity increases as heating
between 0.9 and 0.009 btm. Consequently, the pore access temperature rises. This phenomenon is significant for
domain which is thermally affected, seems to extend from temperatures over 110~ The porosity increase is about
capillary porosity to outer C-S-H porosity as temperature 10% at 250~ and about 30% at 450~ This evolution
rises. It was also observed that the intensity of distribution was not observed for hematite HSC which porosity
peaks tends to increase with temperature. Comparable remained stable whatever the level of temperature
behaviours were observed by other authors with high- (around 10% with a small increase between 250 and
performance concretes and mortars heated up to 800~ 450~ With the pore size distributions obtained
[21, 22}. It was also observed that, after temperature expo- through MIP investigation, it seems that the develop-
sures, hematite HSC porosity peaks were significantly ment of this additional porosity is mainly related to the
smaller than that of ordinary HSC. formation of macropores, ranging from 50 to 0.3 l.tm,
Additional comments should be made concerning and that in such conditions, the increase in size of the
larger macropores. At 250~ and especially at 450~ capillary pores plays a minor role. To confirm these
pores accesses located approximately between 50 and results, the pore size distribution curves obtained by MIP
0.3 ~tm are detected for ordinary concrete (Fig. 6). This analysis were split in 3 classes : "macropores" (> 0.3 gm),
was not observed for hematite concrete. They are proba- capillary pores (0.3-0.02 gm) and micropores related to
bly associated with cracks that showed up at the surface the hydrates (< 0.02 lim). In Table 4, the increase (or
of heated probes of ordinary concrete. Cracking was not decrease) in porosity relative to the initial porosity at
observed for hematite concrete probes. 20~ (obtained by MIP) is given for each class of pores
Total water porosity measurements provided com- and temperature treatments.
plementary information about concrete porous space From Table 4, several points can be underlined:
thermal degradation (Fig. 7). Porosity was estimated by 9 The macroporosity remains almost stable for both
water mass difference on 40 m m in diameter and HSCs till 250~ and then increases at 450~ particu-
80 mm-high probes subjected to different temperatures larly for ordinary concrete (+ 23%). This is certainly due

623
Materials and Structures/Mat6riaux et Constructions, Vol. 34, December2001

to internal or external stresses related to the expansion of the type of hydrates. Sinfilar conclusions were reached
the aggregates (more important for siliceous aggregates through previous studies [21, 23, 25]. In addition, total
than for hematite) or structural thermal gradients during water porosity measurements showed a significant
heating. increase in total porosity for ordinary HSC. This phe-
9 As reflected by Fig. 6, the capillary porosity n o m e n o n can be mainly associated to generation of
increases drastically for both HSCs (+ 55% for ordinary microcraks which were not observed for hematite HSC.
concrete and + 62% for hematite concrete), for tempera-
tures as low as 60~ Surprisingly, this increase remains
almost stable with higher temperatures. This increase in 3.3 Gas permeability
porosity could be associated with the removal of free and
adsorbed water from the material, which may induce The evolution of intrinsic permeabilities versus tem-
damaging of the C-S-H, owing to their structure [8] perature are provided in Fig. 8 (a similar trend was
(the removal of the inter-layer water could induce a col- observed for the evolution of the effective permeability
lapse of the C-S-H sheets). It should be added that min- with temperature). The intrinsic gas permeabilities of
eralogical transformations of C-S-H hydration products ordinary and hematite concretes - measured at 60~
owing to hydrothermal conditions during heating at after complete drying - proved to be similar, around 2
temperatures exceeding 80-100~ may contribute to and 1.5 10-17 m 2, respectively. Up to 250~ ordinary
microstructural changes of cement paste [14]. However, and hematite HSC permeability evolutions follow a sim-
no XRD-analyses were performed on heated materials ilar trend and are thus characterized by a comparable
in this study. behaviour. After drying at 110 and 250~ permeability
9 The microporosity decreases for both HSCs (- 32% increases are, respectively, of about an average factor
for ordinary and hematite concrete), also for temperatures three (= 5 10-17 m 2) and a little more than one order of
as low as 60~ As for capillary pores, this increase does magnitude (= 4 10 -16 m2). Ordinary HSC is always
not evolve with higher temperatures and could be related slightly more permeable than hematite HSC.
to the collapse of the C-S-H sheets, leading to a reduced The most important changes in permeability were
amount of voids of this size. observed at 450 ~ C. At that temperature, intrinsic per-
These results are in good agreement with those meability discrepancy between the two concretes
obtained through water porosity measurements showing reached two orders of magnitude. From 250 to 450~
the major role of macropores in the generation of extra the permeability increase for hematite HSC is very lim-
porosity for the ordinary HSC. In fact, all the samples used ited (from 3 to 4 10-16 m 2) while, for ordinary HSC, the
for water porosity measurements were oven-dried at 60~ permeability change is much more significant (from
before resaturation. The small evolutions of capillary 5 10 -16 to 8 10 -14 m2). From a general point of view, the
porosity and microporosity between 60 and 250~ for the permeability change observed up to 250~ for both
two HSCs (+ 5 - 10% for the capillary porosity and - 5% HSCs can be linked to the removal of free water from
for the microporosity) in comparison to their evolutions the capillary porosity (mainly below 110~ to adsorbed
between 20 and 60~ (+ 55 - 60% for the capillary poros- water release, to cement hydrates (C-S-H) dessiccation
ity and - 30% for the microporosity) explains the fact that (dehydration) and probably to the generation of fine
no increase (or nearly no increase) in water porosity was cracks. All these processes lead to the increase in the
detected between reference samples (heated at 60~ and porous network connectivity and in the porous space
samples heated up to 250~ The increase in water poros- volume available for gas transport (especially the capillary
ity between 250 and 450~ observed particularly for ordi- porosity). Permeability data also showed that microc-
nary HSC seems therefore rather related to the increase in racks observed at the surface of ordinary HSC probes -
macroporosity (+ 23% for ordinary concrete and + 10% for obviously correlated to macropores (50 - 0.3 ~tm)
hematite concrete). Part of it could also be related to an detected by MIP analysis - have a greater influence on
additional increase in capillary porosity due to the begin-
ning of the Portlandite decomposition. It is generally con-
sidered that the stability of calcium hydroxide decreases
from 400~ [5] and that its decomposition mainly occurs 1.E-12
in the 450-550~ temperature range [23]. O ()rdirlaryI~tSC' --'Average values
I v a . ~ e s~~
E 1.E-13 9 HematiteHSC --Average
In conclusion, MIP microstructural investigations
showed that the high temperatures thermal treatments of 2 1.E-14
both HSCs are responsible for the increase in capillary ~ 1.E-15
pore sizes. This p h e n o m e n o n can be related to the
~ 1.E-16
release of adsorbed water physically bound to the solid
surface of capillary pores and to C-S-H microporosity 9~ 1.E-17 , I '
[24] and to water intrinsically bound in C-S-H and to 1.E-18
the other cement hydrates (portlandite). Dehydration 50 100 150 200 250 300 350 400 450 500
processes thus lead to the formation of an additional void Temperature(~
space in the porous medium. The size of enlarged pores
depends on the level of applied temperature and thus on Fig. 8 - Intrinsic permeability evolution versus temperature.

624
GalIGSercombe

ordinary HSC transport characteristics than the increase


1.E-12
in the capillary pore size (0.1 btm). This is reflected by O Ordinary' HSC 9 Hematite HSC ]
the results obtained for ordinary HSC at 450~ 1.E-13
The cross-plot of intrinsic permeability and of the
measured total porosity (Fig. 9) showed a good correlation ~ 1.E-14
between these two parameters. These results confirm that ~1.E-15
extra porosity detected for ordinary HSC at 250~ and o II 9 0
above all at 450~ can be mainly associated to the genera- 1.E-16
tion of large cracks which apparently, at this scale, is absent _= immo
for hematite HSC. In that configuration, the generation 1.E-17
of cracks is the main factor controlling permeability evo- 9 10 11 12 13
Total porosity (%)
lution. These results meet those obtained by Tsimbroska
on HSC heated up to 400~ [7]. Fig. 9 - Relationship b e t w e e n intrinsic permeability and total
water porosity.

4. D I S C U S S I O N 5O
- - Ordinary HSC
This study was the opportunity to experimentally 40 Hematite HSC d
compare the microstructural behaviour at high tempera- d 9
#, i
tures of two high-strength concretes only differing in 30 i

their aggregates. Under similar thermal sollicitations, the


two concretes showed quite different responses, espe-
cially for thermal treatments exceeding 200~ Through A ,I
experimental investigations, the initial intrinsic perme- I f \ !
Q}" 9 9 ,..
ability and the pore structure of the two HSCs cement 0
paste proved to be identical. Since thermo-hydric 0 1 2 3 4 5
processes (internal vapor overpressure gradients) are Time (hours)
mainly related to transfer properties, this tends to indi-
cate that the observed behaviour discrepancy of the two ~. 0.8 -- Ordinary HSC I
HSCs is rather linked to thermo-mechanical processes. m Hematite HSC I
Thermo-mechanical damage during heating of concrete 0.6 - - 7 - - / ~
is known to originate from either structural thermal gra-
dients or thermal incompatibility between the aggregates i /
and the cement paste. The former depend mainly on the
macroscopic thermal conductivity of concrete, but are E 0.2 1/ :
also related to vaporization and dehydration (due to their
endothermal nature). The large difference in thermal 0.0
conductivities between ordinary HSC (2.7 W/m/~ 0 1 2 3 4
and hematite HSC (7.3 W/m/~ supports this interpre- Time (hours)
tation. The latter depends mainly on the thermal expan-
sion coefficient of the aggregates (11 10-6/~ for silico- Fig. 10 - Maximum thermal gradients and pore pressures.
calcareous aggregates and 8 10-6/~ for hematite
aggregates), the elastic moduli of the aggregates (70 GPa design properties and that of the aggregates (respectively
for silico-calcareous aggregates and 210 GPa for 2 and 8 W/m/~ for the thermal conductivity, 790 and
hematite) and the mechanical properties of the cement 660 J/kg/~ for the thermal capacities of silico-calcare-
paste (elastic modulus, thermal shrinkage). Since the ous and hematite aggregates).
cement type and water/cement ratios were identical for Maximum pore pressures and thermal gradients were
both HSCs, one may assume that the observed behav- found at the centre of the probe and are given in Fig. 10.
iour discrepancy comes rather from the aggregates. The maximum pore pressures were obviously not influ-
To check the validity of this assumption, the coupled enced significantly by the type of HSC (0.54 MPa for
thermo-hydric behaviour of the probes during heating ordinary HSC and 0.64 MPa for hematite HSC). It was
was modeled using the mathematical description pro- observed earlier in hematite HSC since temperature rises
posed by Ba~ant et al. [12, 13]. Thermal treatments at faster at the centre of the probe, owing to its high ther-
250~ and 450~ for both HSCs were simulated mal conductivity. The small difference in the maximum
(1~ with the finite element code CASTEM2000 pressure between the two HSCs is induced by the faster
[26]. The same intrinsic permeability and total porosity rise of temperature in hematite HSC which, in turn,
were considered in the calculations for the two HSCs leads to an increased quantity of dehydrated water. This
(respectively 10-19 m 2 and 10%). Only the thermal con- additional water contributes to increase the pore pres-
ductivities and the heat capacity of the dried materials (at sure. Calculations confirmed that thermo-hydric phe-
110~ were taken different, according to their mix- nomena should be of similar magnitude in both HSCs.

625
Materials and Structures/Mat&iaux et Constructions,Vol. 34, Decem ber 2001

Thermal gradients between the centre and the exter-


nal surface of the probes were found three times more ~ Cementpaste
important for ordinary HSC (40~ than for hematite
HSC (12~ Independently of the HSC considered, the
endothermal nature of vaporization and dehydration Fig.n -
leads to peak values exceeding by nearly a factor of three Elementary
Aggregate spherical par-
the thermal gradients obtained without hygral couplings ticle model.
(respectively 15 and 5~ for ordinary and hematite
HSC). These peaks (which could induce significant
damage) occur however for the same local temperature
of 160~ (at the centre of the probes) but for different Table 5 - Characteristics of aggregates and cement paste
external temperatures (195~ for ordinary HSC and Siliceous aggregates
175~ for hematite HSC). These results were confirmed Thermal expansion coefficient at 250~ (1/~ 15.10 -6
experimentally on ordinary HSC probes heated at Thermal expansion coefficient at 450~ (1/~ 21.10 -6
l~ till 220~ (a maximum thermal gradient of Elastic modulus at 250 and 450~ (GPa) 80
37~ was obtained for a local temperature of 170~ and Hematite aggregates
are correlated to tests performed by other authors [9, 11].
Consequently, this means that the maximum "structural Thermal expansion coefficient at 250 ~ (1/~ 8.10 -6
thermal stresses" will not differ for the tests performed Thermal expansion coefficient at 450 ~ (1/~ 11.10 -6
till 250 and 450~ This cannot therefore explain the Elastic modulus at 250 and 450 ~ (GPa) 210
pronounced cracking observed at 450~ and not at Cement paste
250~ for ordinary HSC. Shrinkage strain at 250~ -0.002
Thermal incompatibility between the aggregates and Shrinkage strain at 450~ -0.005
the cement could be at the origin of additional damage Elastic modulus at 250~ (GPa) 7.0
during heating. The differential behaviour of the two Elastic modulus at 450~ (GPa) 3.5
components induces tensile stresses within the cement
paste which are maximum at the interface between
aggregates and cement paste. According to several shrinks leading to shrinkage strains es h. To simplify the
authors [7, 8, 23], crack initiation in silico-calcareous system, the Poisson's ratios (v) are assumed identical for
HSC starts for temperatures over 300~ This could the cement paste and the aggregates. The following sys-
explain the important evolution of the intrinsic perme- tem is thus obtained:
ability of ordinary HSC between 250 and 450~ This
increase in permeability (related to internal cracking) was E
r~r - (1 + v)(1 - 2v)[(1 - v)e r + 2ve, - (1 + v)~aisl
not observed for hematite HSC. Instead of being related
to structural thermal gradients during heating, the E (5)
observed different behaviour of the two HSCs could be [vgr + e, - (1 + v)edil]
induced by internal stresses of higher magnitude in ordi-
%- (1+v)(1- 2v)
nary HSC than in hematite HSC. To check this assump- with E as the modulus ofelastidty of the aggregates or the
tion, a two-phase model consisting of spherical particles cement paste, er = du / dr and et = u / r, the radial and tan-
was used to obtain comparative values of internal tensile gential strains (with u(r) the radial displacement) and ca,y
stresses at different temperature levels. In this approach, the strains associated with thermal expansion of the aggre-
the material consist of aggregate spherical particles gates (eth = ~tkT) or shrinkage of the cement paste (esh).
recovered by spherical layers of cement paste, such that Equations (4) and (5) are solved analytically to obtain an
the ratio between the radii of the aggregate particle (a) estimation of the tangential stresses within the cement
and that of the aggregate + cement paste particle (b) paste at the interface with the aggregates (for r - a). The
remains constant, whatever the size of the aggregate par- characteristics of aggregates and cement paste considered
ticles [27]. The matter is assumed to be completely filled are summarized in Table 5.
up with these composite spherical particles, such that the Average values concerning the thermal expansion
analysis can be carried out only on an elementary parti- coefficient of siliceous and hematite aggregates as well as
cle, represented in Fig. 11. the shrinkage strain of Portland cement paste were taken
The stresses (radial stress % and tangential stress %) from [13, 16]. Siliceous aggregates are primarily consid-
within the aggregates and the cement paste must verify ered for ordinary HSC since they are known to be more
the spherical conditions of equilibrium (4): sensitive to temperature than calcareous aggregates [13].
Elastic moduli for the cement paste were estimated from its
a{~r -~-2 ( ~ r
-- ~ t ) = 0 (4) value at ambient temperature (14 GPa) assuming that it
dr r decreases by 50 and 75% at 250 and 450~ respectively. A
The stress-strain relations are obtained assuming that mean value of 0.8 for both HSCs was considered for the
the materials behave elastically and include the following maximum aggregate packing density (g* = a3 / b3).
phenomena: the aggregates expand with temperature The maximum orthoradial thermal stresses within the
leading to thermal strains e th = aAT, the cement paste cement paste obtained for both HSCs are given in Fig. 12.

626
Gall~,Sercombe

4O
~ OrdinaryHSC
35 "" "HematiteHSC
30
"-"co25 __Ordinary
HSC

25
~, 20 .......
18 il5~:
5~u)
~ 200c.
~10 . . . 50
e-

0
0 1O0 200 300 400 500
Temperature(~
Fig. 12 - Maximum orthoradial thermal stresses within the Fig. 13 - Maximum orthoradial shrinkage stresses within the
cement paste. cement paste.

Values are of course not realistic but can be used to com- 450~ macropores - ranged globally from 50 to 0.3 gm
pare the behaviour of both concretes. Fig. 12 shows that - were detected for ordinary HSC and correlated to
tensile stresses resulting from the thermal expansion of microcracks observed at the surface of heated probes.
aggregates are more important in ordinary HSC than in Such damage was not observed for hematite HSC. On
hematite HSC. The difference is proportional to the ratios the contrary, at high temperature, finer C-S-H porosity
between the thermal conductivities of the aggregates seems to be reduced for both materials. Total water
(almost identical and equal to 1.9 at 250 and 450~ The porosity measurements and MIP investigations con-
elastic moduli of aggregates were found to interfere little firmed that, at 250~ and especially at 450~ ordinary
on the maximum orthoradial stresses. HSC porous space is much more degraded than that of
The maximum orthoradial shrinkage stresses within hematite HSC. Similar trends were observed with per-
the cement paste obtained for both HSCs are given in meability data. Up to 250~ the intrinsic permeability
Fig. 13. Tensile stresses associated with shrinkage were evolution is similar for both HSCs. Permeability increase
found almost identical for ordinary and hematite HSCs. is related to water removal from the porous network, to
The small difference (less than 5%) can be associated adsorbed water release and to cement hydrates dehydra-
with the differences in the elastic moduli of the aggre- tion. These phenomena contribute to the increase in cap-
gates. Combining thermal expansion of the aggregates illary pore size and to the generation of fine cracks. On
and shrinkage of the cement paste, the maximum ortho- the contrary, data provided by heating tests at 450~
radial stresses in the cement paste are the sum of the clearly showed that microcracks formation is the main
independent components resulting from each phenome- factor controlling the permeability evolution and that the
non. Obviously, the magnitude of these interfacial increase of capillary pore size is less influent. The differ-
stresses are mainly related to the thermal expansion coef- ential behaviour of ordinary and hematite HSC was
ficient of the aggregates and are therefore more impor- attributed to the thermo-mechanical discrepancy related
tant for ordinary HSC than for hematite HSC. They mainly to structural thermal gradients and differential
depend little on the magnitude of cement paste shrink- expansion of the cement paste and the aggregates. Owing
age and on the elastic moduli of the aggregates. This to its reduced thermal expansion coefficient and
could therefore explain the higher damage of ordinary increased thermal conductivity, hematite is obviously
HSCs probes with respect to hematite HSC, especially at responsible for the limitation of thermal stresses during
temperatures exceeding 250~ heating. Thermo-mechanical phenomena in heated
HSCs are therefore of great importance for the prediction
of damage at high temperatures.
5. CONCLUSION

The main purpose of this study was to evaluate the ACKNOWLEDGEMENT


microstructural property changes of two high-strength
concretes only differing in their aggregates - silico-cal- The study was carried out in the framework of the
careous and hematite - submitted to high temperatures. 1LEPO research programme at the CEA Saclay. The
The study of pore structure through MIP investigation authors acknowledge G. Arcier and M. Pin for achieving
showed that between 60 and 450~ the thermal treat- experimental tests.
ments were responsible for the generation of a large capil-
lary porosity for both HSCs. The intensity and especially
the width of the porosity peaks increased with tempera-
REFERENCES
ture. The increase in capillary pore size was related to the
release of adsorbed water of capillary pores and of cement [1] Riley, M. A., 'Possible new method for the assessment of fire-
paste hydrates bound water. Over 250~ and above all at damaged concrete', Mag. Corer. Res. 43 (155) (1991) 87-92.

627
Materials and Structures/Mat&iaux et Constructions,Vol. 34, December 2001

[2] Phan, L. T., Carino, N. J., Duthin, D. and Garboczi, E., 'Fire 'Behaviour of high performance concrete under high tempera-
performance of high-strength concrete', NIST Proceedings of ture (60-450~ for surface long-term storage: thermo-hydro-
the Int. Workshop, NIST Special Publication 919, Gaithersburg, mechanical residual properties', MRS Conference, Sydney,
Maryland, USA (1997). Australia, 27-31 August, 2000.
[3] Castillo, C. and Durrani, A.J., 'Effect of transient high temperature [16] Bouniol, P., 'Structures des r&cteurs nucl&ires. B&ons sp&iaux
on high-strength concrete', ACIMat.J. 87 (1) (1990) 47-53. de protection', in 'Les Techniques de l'Ing~nieur' (in French),
[4] Sanjayan, G. and Stocks, L. J., 'Spalling of high-strength silica Gdnie Nucl&ire, Article BN 3740.
fume concrete in fire', ACIMat..J. 90 (2) (1993) 170-173. [17] Gall~, C. and Daian, J.-F., 'Gas permeability of unsaturated
[5] Saad, M., Abo-E1-Enein, S. A., Hanna, G. B. and Kotkata, M. F., cement based materials: application of a multiscale network
'Effect of silica-fume on the phase composition and microstruc- model', Mag. Concr.Res. 52 (4) (2000) 251-263.
ture of thermally treated concrete', Cem. Contr. Res. 26 (10) [18] Klinkenberg, L. J., 'The permeability of porous media to liquids
(1996) 1479-1484. and gases',API Drilling and Production Practices (1941) 200-213.
[6] Chan, Y. N., Peng, G.-F. and Chan, K. W., 'Comparison [19] Perraton, D., A/tcin, P.-C. et Carles-Gibergues, A., 'Mesure de
between high strength concrete and normal strength concrete la perm&bilit~ aux gaz des b&ons : perm&bilitd apparente et per-
subjected to high temperature', Mater. Struct. 29 (1996) 616-619. m&bilitd intrins~que. Partie I - Validation des concepts de
[7] Tsimbrovska, M., Kalifa, P., Quenard, D. and Daian, J.-F., 'High Carman et de Klinkenberg dans le cas d'un BHP', LCPC Bulletin
performance concretes at elevated temperature: permeability and (in French) 221 (4241) (1999) 69-78.
microstructure', Transactions of the 14th International [20] Bentz, D. P., Snyder, K. A. and Stutzman, P. E., 'Hydration of
Conference on Structural Mechanics in Reactor technology portland cement: the effect of curing conditions', Proceedings of
(SmiRT 14), Lyon, France, 1997, 475-482. the 10th International Congress on the Chemistry of cement, H.
[8] Kalifa, P., Tsimbrovska, M. and Baroghel-Bouny, V., 'High-per- Justnes Ed., Vol. 2, Gothenburg, Sweden (1997).
formance concrete at elevated temperatures - An extensive [21] Rostasy, F. S., Weib, 1k. and Wiedemann, G., 'Changes of pores
experimental investigation of thermal and hygral properties and structure of cement mortars due to temperature', Cem. Contr.
microstructure', Proceedings of Int. Symp. on High- Res. 10 (1980) 157-164.
Performance and Reactive Powder Concretes, Sherbrooke, [22] Chan, Y. N., Luo, X. and Sun, W., 'Compressive strength and
Canada, 1998, 259-279. pore structure of high-performance concrete after exposure to
[9] Kalifa, P., Mennenteau, F.-D. and Quenard, D., 'Spalling and high temperature up to 800~ ', Gem. Concr. Res. 30 (2) (2000)
pore pressure in HSC at high temperatures', Cern. Concr. Res. 30 247-251.
(2000) 1-13. [23] Piasta, J., Sawicz, z. and Rudzinski, L., 'Changes in the struc-
[10] Ulm, F.-J., Acker, P. and Levy, M., 'The chunnel fire. II: analysis ture of hardened cement paste due to high temperature', Mater.
of concrete damage', ASCEJ. Eng. Mech. 125 (3) (1999) 283-289. Struct. 100 (1984) 291-296.
[11] Noumow4, A., 'Effet de hautes temp&atures sur le b&on. Cas [24] Ulm, F.-J., Acker, P. and Le W, M., 'The chunnel fire. I:
particulier du b&on h hautes performances', PhD-Thesis (in chemosplatic softening in rapidly heated concrete', ASCE.J. Eng.
French), INSA Lyon, France, (1995). Mech. 125 (3) (1999) 272-282.
[12] Ba~ant, Z. P. and Thongutai, W., 'Pore pressure and drying of [25] Harmathy, T. Z., 'Thermal properties of concrete at elevated
concrete at high temperature',J. Eng. Mech. Div. ASCE 104 temperatures', ASTMJ. ofMat. 5 (1) (1970) 47-44.
(1978) 1058-1080. [26] Sercombe, J., GallS, C., Durand, S. and Bouniol, P., 'On the
[13] Ba~ant, Z. P. and Kaplan, M. F., 'Concrete at high tempera- importance of thermal gradients in the spalling of high-strength
tures. Material properties and mathematical models', (Longman concrete', 14th Engineering Mechanics Conference, ASCE,
House, Burnt Mill, England, 1996). Austin, Texas, USA, 21-24 May, 2000.
[14] Khoury, G. A., 'Compressive strength of concrete at high tempera- [27] Le Roy, R., 'D~formations instantan&s et diff&&s des b&ons
tures: a reassessment',Mag. Concr.Res. 44 (161) (1992) 291-309. hautes performances', Research Report from the Laboratoire
[15] Galld, C., Sercombe, J., Pin, M., Arcier, G. and Bouniol, P., Cennal des Ponts et Chauss&s (in French), N o OA22, Paris, 1996.

628

You might also like