A model for predicting time-dependent chloride binding. capacity of cement-fly ash cementitious system

Transcription

1 Material and Structure / Matériaux et ontruction, Vol. 37, July 2004, pp A model for predicting time-dependent chloride binding capacity of cement-fly ah cementitiou ytem T. Sumranwanich and S. Tangtermirikul Sirindhorn International Intitute of Technology, Thammaat Univerity, Pathum Thani, Thailand ABSTRAT A model for predicting time-dependent chloride binding capacity of cement-fly ah cementitiou ytem wa propoed. The propoed model took into account both chemical binding and phyical binding. hemical binding wa conidered to depend on the amount of unhydrated aluminate and aluminoferrite phae while phyical binding depended upon the quantity of hydrated and pozzolanic product. The concept of time-dependent chloride binding capacity wa introduced in the model with the conideration of curing time and chloride expoure period. The chloride binding of cement pate and cement-fly ah pate under different curing time and chloride expoure period were teted. Three type of cement and two type of fly ah were ued. From the experimental reult, time-dependent behavior of chloride binding capacity wa oberved. At the ame chloride expoure period, pate with longer curing time prior to chloride expoure bound le chloride than thoe expoed with horter curing time. Longer expoure period of pate reulted in larger chloride binding capacity. The analytical reult from the model were verified with the experimental reult from the author and other reearcher. The verification howed that the propoed model wa atifactory for predicting the chloride binding capacity of variou cement and cement-fly ah cementitiou ytem. RÉSUMÉ L article préente un modèle capable de prévoir la capacité d un mélange ciment-cendre volante de fixer le ion chlorure. Il prend en compte à la foi le liaion chimique et le liaion phyique. Le liaion chimique dépendent de la quantité d aluminate non hydraté aini que de la quantité d alumino-ferrite alor que le liaion phyique dépendent de la quantité de produit hydraté et de la quantité de pouzzolane. La variable temporelle a été introduite afin de prendre en compte la durée de cure aini que la durée d expoition aux chlorure. Troi type de ciment et deux type de cendre volante ont été utilié et pluieur durée de cure et d expoition aux chlorure ont été tetée. Pour la même durée d'expoition au chlorure, le pâte ayant un temp de cuion plu long avant d'être expoée au chlorure ont une capacité de liaion du chlorure moindre que celle expoée avec de temp de cuion plu court. Un autre réultat et que la capacité de fixer le ion chlorure augmente avec la durée d expoition aux ion chlorure. Le réultat analytique du model ont été confronté aux réultat expérimentaux de auteur de l article aini qu aux réultat publié dan la littérature. ette vérification a montré que le prédiction du model analytique ont conforme aux réultat de l expérience. 1. INTRODUTION One of the predominant caue of the corroion of teel in concrete i chloride attack. hloride ion may be preent in a concrete mixture either a a reult of uing contaminated ingredient or certain chemical admixture or a a reult of penetration from external ource uch a eawater or deicing alt. The ability of hydrating cement to bind chloride from the pore olution in concrete i one of the important factor which control the initiation of chloride-induced corroion of teel in concrete. Thi i becaue only free chloride preent in the pore olution can initiate corroion when the free chloride content around the teel reache a critical value. Therefore, chloride binding capacity i a ignificant property of concrete for prolonging the ervice life of the reinforced concrete tructure ubjected to chloride attack. There are many factor that govern the chloride binding capacity, uch a type of cement, type and proportion of cement replacement material, water to cement ratio, curing time prior to chloride attack, expoure period with chloride and o on. hloride binding capacity of variou cementitiou ytem had been tudied by many reearcher [1-13]. Some have propoed a model for predicting the chloride binding capacity /04 RILEM 387

2 Sumranwanich, Tangtermirikul of cement-ground granulated blatfurnace lag pate [7]. However, there i till no model that conider the effect of curing time and chloride expoure period in the prediction of chloride binding capacity of cement-fly ah pate. The time-dependent chloride binding capacity of pate depend on the age of the pate at the tart of chloride expoure, and the chloride expoure period. Aluminate ( 3 A) and aluminoferrite ( 4 AF) phae in cement have been found to be reponible for the chemical binding of chloride [2-6]. Thee two phae form Friedel alt (a 6 Al 2 O 6.al 2.10H 2 O) and calcium chloroferrite (a 6 Fe 2 O 6.al 2.10H 2 O). The binding capacity wa then conidered depending on the content of 3 A and 4 AF in cement. The increae of ulfate content in cement wa found to reduce the chloride binding capacity ince ulfate were more trongly bound with 3 A than were chloride [8, 9]. The content of 3 A, 4 AF and ulfate in cement were found to be ignificant parameter influencing the chemical binding of chloride [10]. While chemical binding wa dicovered to depend on the content of aluminate and aluminoferrite phae in cement, phyical binding depended upon the content of hydrated product, particularly the content of -S-H in concrete [11, 12]. Moreover, there wa evidence that calcium aluminate hydrate produced by the pozzolanic reaction of fly ah cement blend can bind the chloride [13]. The aim of thi tudy i to propoe a model for predicting chloride binding capacity of cement-fly ah cementitiou ytem baed on mixture proportion and propertie of cementitiou material. The time-dependent effect of curing time prior to chloride attack and chloride expoure period on the chloride binding capacity were conidered in the model. Both chemical binding and phyical binding were included into the model. The validity of the model wa verified by the experimental reult obtained from both the author and from other reearcher [1, 4, 8, 18]. 2. EXPERIMENTAL PROGRAM Table 1 - hemical compoition and phyical propertie of Portland cement and fly ah hemical compoition Type I Portland cement Type III Portland cement Type V Portland cement Low calcium fly ah (F-type) High calcium fly ah (-type) SiO 2 (%) Al 2 O 3 (%) Fe 2 O 3 (%) ao (%) MgO (%) SO 3 (%) Na 2 O (%) K 2 O (%) Free lime (%) Lo on ignition (%) Phyical propertie Blaine finene (cm 2 /g) 3,190 4,770 3,760 3,460 3,510 Specific gravity Bogue potential compound compoition 3 S (%) S (%) A (%) AF (%) Table 2 - Mixture Material w/b f/b uring hloride Mix deignation ement time expoure Fly ah (day) period (day) type type 1 Type I , 7, 28 28, 56, 91 2 Type III , 7, 28 28, 56, 91 3 Type V , 7, 28 28, 56, 91 4 Type I , 7, 28 28, 56 5 Type I , 7, 28 28, 56, 91 FL1 Type I F * , 7, 28 28, 56, 91 FL2 Type I F * , 7, 28 28, 56, 91 FL3 Type I F * , 7, 28 28, 56, 91 FH1 Type I ** , 7, 28 28, 56, 91 FH2 Type I ** , 7, 28 28, 56, 91 FH3 Type I ** , 7, 28 28, 56, 91 * F-type fly ah ha (SiO 2 + Al 2 O 3 + Fe 2 O 3 ) content greater than 70%, but very little in ao content. It i called low calcium fly ah in thi tudy. ** -type fly ah ha (SiO 2 + Al 2 O 3 + Fe 2 O 3 ) content le than 70%, but larger in ao content. It i called high calcium fly ah in thi tudy. 2.1 Material, mix proportion and pecimen preparation Three type of cement, type I, type III and type V Portland cement, were ued in thi tudy. Two type of fly ah correponding to ASTM F-type (low calcium fly ah) and ASTM -type (high calcium fly ah) were mixed with type I Portland cement for producing the cement-fly ah pate. The chemical compoition and phyical propertie of cement and fly ah are lited in Table 1. Eleven different mixture of cementitiou pate were prepared for thi invetigation, a hown in Table 2. There were five mixture of cement pate and ix mixture of cement-fly ah pate. The tet parameter were type of cement, type of fly ah, water to binder ratio (w/b) and fly ah to binder ratio (f/b). Specimen, 50 mm in diameter and 10 mm thick, were cat in PV mold. Thirteen pecimen were prepared for each mixture condition, ten for expreing the pore olution and three for determining the evaporable water content. The mixing procedure wa performed according to ASTM uring time and chloride expoure period After cating, pecimen were ealed with platic heet to prevent drying for 24 hour. Except for pecimen to be expoed to chloride at 1 day, all pecimen were cured in water immediately after removal from the mold. uring time were 1, 7 and 28 day a hown in Table 2. The curing temperature wa 302. At the end of water curing, pecimen were expoed to chloride by ubmerion in alt water with 3.0% chloride ion 388

3 Material and Structure / Matériaux et ontruction, Vol. 37, July 2004 concentration (30 gram per liter) for different expoure period. The expoure period were 28, 56 and 91 day a hown in Table 2. The volume of alt water (chloride olution) wa 2.0 liter. The temperature during the chloride expoure period wa Determination of chloride content At the end of the chloride expoure period, pecimen were removed from alt water. The urface of the pecimen were dried by uing tiue paper. Pore olution inide the pecimen wa obtained uing a pore expreing apparatu. The maximum loading preure for expreing the pore olution wa about 500 MPa. Two or three cycle of loading and unloading were performed in order to get 3 to 5 cm 3 of pore olution. The evaporable water content of the pecimen wa teted for ue in the determination of free chloride in the pecimen. Total chloride wa determined from the difference between initial chloride content of the alt water olution at the tart of expoure and it final chloride content at the end of expoure, and wa aumed to be hared equally by all pecimen ubmerged in the alt water. The free chloride wa determined from the chloride concentration of the pore olution expreed from the pecimen multiplied with the evaporable water. Finally, the fixed chloride of cementitiou pate wa determined by ubtracting the free chloride from the total chloride. All chloride concentration were analyzed by potentiometric titration with AgNO 3 olution and a chloride ion elective electrode. 3. MODEL OF TIME-DEPENDENT HLORIDE BINDING APAITY Since the chloride binding capacity of cement-fly ah cementitiou ytem i a reult of chemical binding and phyical binding a decribed in the introduction. Then, the model of time-dependent chloride binding capacity i formulated by taking into account both chemical and phyical binding a given in Equation (1). The aluminate phae, 3 A, and aluminoferrite phae, 4 AF, in cement were conidered reponible for the chemical binding while the hydrated product from cement and pozzolanic product from fly ah, uch a -S-H, -A-H, -A-F-H, ettringite and monoulfate were reponible for phyical binding. fix (t, t ) (t, t ) (t ) (1) e fix, chem e fix, phy where fix (t, t e ) i the total fixed chloride content in the cementitiou ytem (% by weight of binder), fix, chem (t, t e ) and fix, phy (t e ) are the fixed chloride content by chemical binding and by phyical binding, repectively (% by weight of binder), t i the age at the tart of chloride expoure which i equal to curing time (day) and t e i the age at the end of chloride expoure (day). It i noted that t e -t repreent the chloride expoure period (day). 3.1 Hydrated ma of cement and reacted ma of fly ah Hydrated ma of cement There are four major compound in Portland cement, i.e., 3 A, 4 AF, 3 S and 2 S. The ma of each major e compound i calculated baed upon Bogue equation, which related to the chemical compoition of Portland cement. The hydrated ma of compound i at age t i determined from Equation (2). i(t) Mhyd,i(t) Mi, i = 3 A, 4 AF, 3 S, 2 S (2) 100 where M hyd, i (t) i the hydrated ma of compound i at age t day (kg/m 3 of concrete), M i, i the ma of each major compound in Portland cement (kg/m 3 of concrete), i (t) i the degree of hydration of compound i of cement at age t day (%) and t i the age of the ample (day). The age i equal to zero at the time at which water i added to the mixture. The degree of hydration of each major compound i defined a the percentage of the hydrated ma of that compound at a certain age to the total ma before hydration of that compound. The degree of hydration of each major compound depend on many factor uch a water to binder ratio, temperature and time. The detail of degree of hydration are not provided in thi paper but elewhere [14, 15] ince they are not the direct cope of thi tudy. The example of degree hydration of each major compound of type I Portland cement in the pate with w/c of 0.40 i hown in Fig. 1. Degree of hydration (%) Age (day) 3 A 3 S 4 AF 2 S Fig. 1 - Degree of hydration of type I Portland cement of author (w/c=0.40, temperature=30) Reacted ma of fly ah There are mineralogical and glay compoition in fly ah. Only the glay compoition of fly ah i reactive in the pozzolanic reaction. The reacted ma of fly ah in the pozzolanic reaction at age t day i calculated according to Equation (3). fa (t) Mpoz, fa (t) Mfa (3) 100 where M poz, fa (t) i the reacted ma of fly ah at age t day (kg/m 3 of concrete), M fa i the ma of fly ah (kg/m 3 of 389

4 Sumranwanich, Tangtermirikul concrete) and fa (t) i degree of pozzolanic reaction of fly ah at age t day (%). The degree of pozzolanic reaction i defined a the percentage of reacted ma of fly ah to the total ma of fly ah. The degree of pozzolanic reaction of fly ah depend upon many factor uch a water to binder ratio, temperature, time, effective calcium oxide, finene, and o on. The detail of degree of pozzolanic reaction of fly ah i not provided in thi paper but elewhere [14, 15]. The degree of pozzolanic reaction of low and high calcium fly ahe in the pate with w/b of 0.40 and f/b of 0.30, 0.50 and 0.70 ued in thi paper are hown in Fig. 2. Degree of pozzolanic reaction (%) Low calcium fly ah, f/b=0.30 High calcium fly ah, f/b=0.30 Low calcium fly ah, f/b=0.50 High calcium fly ah, f/b=0.50 Low calcium fly ah, f/b=0.70 High calcium fly ah, f/b= Age (day) Fig. 2 - Degree of pozzolanic reaction of low and high calcium fly ahe of author (w/b=0.40, f/b=0.30, 0.50, 0.70, temperature=30). 3.2 Hydration product and pozzolanic product It i aumed here for implicity that the quantity of hydrated product of cement and pozzolanic product of fly ah are determined baed on the reaction hown in Table 3. The quantity of product i calculated baed on the reaction equation in that table and their correponding hydrated ma of cement and reacted ma of fly ah. 3.3 hemical binding In general, a part of 3 A and 4 AF in cement firt react with gypum to form ettringite and monoulfate. The remaining unhydrated 3 A and 4 AF react further with water during the curing period. It i aumed here that only a certain fraction of the remaining content of 3 A and 4 AF are efficient for chemical binding. The efficient part are thoe which react during the chloride expoure period only and form Friedel alt and calcium chloroferrite wherea thoe which react before the chloride expoure period do not contribute to chemical binding. Although it i evident that ettringite can diolve and tranform into ome chloride-bearing phae and vice-vera, depending on their tability which i influenced by the chloride and ulfate ion concentration in the pore olution [16]. However, the quantification of thi tranformation require further intenive tudie, therefore it wa not taken into account in the model. The fixed chloride content by chemical binding ( fix, chem (t, t e )) i defined a hown in Equation (4). The time-dependent effect of curing time, t, and age at the end of chloride expoure, t e, were taken into account in thi equation. fix, chem (t, t ) e fix, 3A (t, t ) e weight of fix, 4AF binder (t, t e ) 100 where fix, 3A (t, t e ) and fix, 4AF (t, t e ) are the fixed chloride content by chemical binding of 3 A and 4 AF, repectively during the expoure period of t e -t (kg/m 3 of concrete). Weight of binder i in kg/m 3 of concrete. The amount of fix, 3A (t, t e ) and fix, 4AF (t, t e ) can be determined from Equation (5) and (6), repectively. M fix, 3A(t,te) M fix, 3A (4) (t ) M (t ) (5) hyd, 3A e hyd, 3A fix, 4AF(t,te) Table 3 - Reaction of cement and fly ah [17] Material Reaction Product 1. ement 3 A 3 A + 3SH H 6 AS 3 H A + 6 AS 3 H H 3 4 ASH 12 3 A + 6H 3 AH 6 4 AF 4 AF + 3SH H 6 (A,F)S 3 H 32 + (A,F)H 3 4 AF + 6 (A,F)S 3 H H 3 4 (A,F)SH 12 + (A,F)H 3 4 AF + 9H + 4H 4 (A,F)H 13 fix, 4AF (t ) M (t ) (6) hyd, 4AF e hyd, 4AF in which 1.12 fix,3a A 3.3 e (7) and 0.6 fix,4af AF (8) 3.3 e 6 AS 3 H 32 4 ASH 12 3 AH 6 6 (A,F)S 3 H 32 4 (A,F)SH 12 4 (A,F)H 13 3 S 2 3 S + 6H 3 S 2 H 3 + 3H 3 S 2 H 3 2 S 2 2 S + 4H 3 S 2 H 3 + H 3 S 2 H 3 2. Fly ah S 2S + 3H 3 S 2 H 3 3 S 2 H 3 A 2A + 3H 3 A 2 H 3 3 A 2 H 3 Note: = ao, S = SiO 2, A = Al 2 O 3, F = Fe 2 O 3, H = H 2 O, S = SO 3 where fix, 3A and fix, 4AF are defined a the fixed chloride ratio of 3 A and 4 AF, i.e., the ratio of fixed chloride to hydrated ma of 3 A and 4 AF, repectively, and 3A and 4AF are the change of degree of hydration of 3 A and 4 AF, repectively during the expoure period (%). The relationhip between the fixed chloride ratio of 3 A and 4 AF and their repective change of degree of hydration are hown in Fig. 3. Thi relationhip i calculated from the back 390

5 Material and Structure / Matériaux et ontruction, Vol. 37, July 2004 analyi of the tet reult of fixed chloride. The fixed chloride ratio decreae with the increae of the change of degree of hydration during the expoure period ( in Fig. 3). Thi implie that more chloride can be bound chemically at early hydration of 3 A and 4 AF. 3.4 Phyical binding hloride may be phyically adorbed on the urface of -S- H gel [11, 12]. In thi model, it wa anticipated that the chloride can be phyically adorbed on the urface of other product of reaction in the cementitiou ytem, uch a -A-H, -A-F-H, ettringite and monoulfate. Thi iue, however, require future intenive reearch. The fixed chloride content by phyical binding at the end of chloride expoure ( fix, phy (t e )) i defined in Equation (9). Thi equation alo take into account the time-dependent effect of curing time plu chloride expoure period, t e. fix, phy fix M product (t e ) 100 (t e ) 100 weight of binder where fix i the fixed chloride content of hydrated and pozzolanic product (%) and M product (t e ) i the ummation of ma of hydrated product and pozzolanic product at the end of chloride expoure (kg/m 3 of concrete). Weight of binder i in kg/m 3 of concrete. For implicity, it i aumed here that all hydrated and pozzolanic product have the ame fixed chloride content. The fixed chloride content of hydrated and pozzolanic product depend on the total chloride content, water to binder ratio and finene of cement in the cementitiou ytem. The phyically bound chloride content wa derived from back computation uing tet data of chloride binding capacity. The derived equation i hown in Equation (10) fix fix (%) wb e 0.5 tot F tot wb / tot 3 A 4 AF Fig. 3 - Relationhip between fixed chloride ratio of 3 A and 4 AF and change of degree of hydration during the expoure period. (9) (10) fix (%) tot (% by wt of binder) where tot i the total chloride content (% by weight of binder), w/b i the water to binder ratio and F c i the Blaine finene of cement (cm 2 /g). A hown in Fig. 4, the phyically bound chloride content of hydrated and pozzolanic product increae with increaing total chloride content and decreaing water to binder ratio. 4. RESULTS AND DISUSSIONS Data of w/b=0.30 Data of w/b=0.40 Data of w/b=0.50 M odel of w/b=0.30 M odel of w/b=0.40 M odel of w/b=0.50 Fig. 4 - Fixed chloride content for hydrated and pozzolanic product. In thi tudy, tet of chloride binding capacity were conducted baed on the fact that in mot cae, chloride attack concrete from external ource. Thi type of chloride i referred to external chloride in thi paper. On the other hand, the chloride preent in concrete at the tart of concrete mixing i called internal chloride. The tet reult of total chloride and fixed chloride are preented by bar chart. The value in parenthei above the bar indicate the ratio of fixed chloride content to total chloride content. It can be een from Fig. 5 and 6 that the chloride binding capacity of cement pate exhibit a time dependent behavior. onidering ample with the ame expoure period, t e -t, pate with horter curing time had higher total and fixed chloride content than thoe with longer curing time. The reaon for the larger total chloride content in the horter curing time cae wa that younger pate had bigger pore diameter, o larger amount of chloride could penetrate into the pate. Fixed chloride wa alo higher becaue there were larger amount of unhydrated aluminate and aluminoferrite phae, which were acceible for chloride binding. On the contrary, conidering pate with the ame curing time, a longer expoure period in altwater reulted in higher total and fixed chloride content. Thi wa imply becaue higher total chloride wa jut of the longer expoure period while the reaon for larger fixed chloride content wa that larger amount of hydrated and pozzolanic product produced during the longer expoure period can bind chloride. By comparing Fig. 5(a) with Fig. 5(b), it can be een that at horter expoure period, type III cement pate had a higher fixed chloride content than type I cement pate. Thi wa becaue type III cement had a higher finene than type I 391

6 Sumranwanich, Tangtermirikul (0.51) (0.57) (0.62) 1.74 (0.52) (0.61) (0.58) (a) Type I cement pate (w/c=0.40) (a) Type I cement pate (w/c=0.30) (0.56) (0.61) (0.66) (0.64) (0.60) (0.69) (0.64) (b) Type III cement pate (w/c=0.40) 1.40 (0.39) (0.45) (0.49) (0.57) 0.76 (0.58) 1.30 (0.53) (0.60) 0.53 (0.60) (0.56) (0.51) (0.57) (0.61) (0.52) (0.62) (0.58) (0.30) (0.40) 0.56 (b) Type I cement pate (w/c=0.40) (0.43) 2.02 (0.61) (0.33) (0.35) (0.61) (0.29) (0.52) 0.97 (c) Type V cement pate (w/c=0.40) total, 28-day expoure total, 56-day expoure total, 91-day expoure fixed, 28-day expoure fixed, 56-day expoure fixed, 91-day expoure Fig. 5 - hloride binding capacity of variou cement pate. cement, o the hydration developed fater and a more hydration product were produced. Thi reulted in higher fixed chloride content. However, when the expoure period wa longer, the binding capacity of type I cement and type III cement wa nearly the ame. (c) Type I cement pate (w/c=0.50) total, 28-day expoure total, 56-day expoure total, 91-day expoure fixed, 28-day expoure fixed, 56-day expoure fixed, 91-day expoure Fig. 6 - hloride binding capacity of cement pate with variou water to cement ratio. When comparing Fig. 5(a) with Fig. 5(c), it can be oberved clearly that type V cement pate had a lower fixed chloride content than type I cement pate for all curing and expoure period. Thi wa mainly becaue the type V cement had a lower content of 3 A than type I cement. 392

7 Material and Structure / Matériaux et ontruction, Vol. 37, July 2004 By conidering the effect of water to cement ratio on chloride binding capacity in Fig. 6, it can be een that when the water to cement ratio wa increaed, though the total chloride content increaed, the ratio of fixed chloride content to total chloride content decreaed. Thi may be becaue chloride can be more eaily retrained, epecially phyically, in a dener pate. In addition, Fig. 7 and. 8 how that the characteritic of time-dependent chloride binding of cement-fly ah pate follow the ame trend a thoe of the cement pate. Fig. 7 and 8 illutrate that cement pate with high calcium fly ah had a higher fixed chloride content than that with low (0.37) (0.44) (0.29) (0.34) (0.40) (0.50) (0.54) (0.48) (a) f/b=0.30, w/b= (0.42) 0.68 (0.24) (0.44) 0.78 (0.38) (0.44) 0.82 (b) f/b=0.50, w/b= (c) f/b=0.70, w/b= (0.34) (0.20) (0.35) (0.22) (0.43) (0.50) (0.31) (0.13) (0.40) (0.21) 0.19 total, 28-day expoure total, 56-day expoure total, 91-day expoure fixed, 28-day expoure fixed, 56-day expoure fixed, 91-day expoure (a) f/b=0.30, w/b= (0.37) 1.81 (0.38) 1.99 (0.35) (0.53) 2.22 (0.56) (b) f/b=0.50, w/b= (c) f/b=0.70, w/b= (0.32) (0.56) (0.27) (0.39) (0.38) 0.68 total, 28-day expoure total, 56-day expoure total, 91-day expoure (0.50) fixed, 28-day expoure fixed, 56-day expoure fixed, 91-day expoure Fig. 7 - hloride binding capacity of type I cement low calcium fly ah pate with variou fly ah to binder ratio. Fig. 8 - hloride binding capacity of type I cement high calcium fly ah pate with variou fly ah to binder ratio. 393

8 Sumranwanich, Tangtermirikul calcium fly ah. Thi i becaue the high calcium fly ah uually contain ome cementitiou component which can hydrate to bind chloride and alo to increae the early pozzolanic reaction o that more pozzolanic product can be produced, epecially at the high replacement ratio. 5. VERIFIATIONS The chloride binding capacity model wa verified with tet reult obtained from both the author and other reearcher. The mix ingredient and chemical compoition of material from other reearcher were briefly hown in Table 4. The verification wa done on pate, mortar and concrete with different type of cement and fly ah, fly ah replacement ratio, water to binder ratio, curing time and chloride expoure period. Fig. 9 and 10 how the verification of the model with the experimental reult conducted by the author. In each figure, the fixed chloride content calculated from the model are compared with thoe from experiment. It can be een from the figure that the model can be ued to predict the chloride binding capacity of cement pate and cement fly-ah pate at variou curing time and chloride expoure period to a atifactory degree. Fig. 11 to 14 demontrate the verification of the model Table 4 - Mixture and propertie of material from other reearcher Reearcher Raheeduzzofar [4] Huain [8] Arya [1] Maruya [18] Specimen type ement pate ement pate ement ement mortar, fly ah pate mortar and concrete w/b f/b hloride type internal internal external external hloride content 0.30%, 0.60%, 1.20% by wt of cement 0.60%, 1.20% by wt of cement 12.1 g/l 18.2 g/l Material FA , 28, hloride expoure period (day) , 56, 84 28, 91, 182, 365 hemical compoition SiO 2 (%) Al 2 O 3 (%) Fe 2 O 3 (%) ao (%) SO 3 (%) * 2.54* 2.61* 0.50 Lo on ignition (%) Bogue potential compound compoition 3 S (%) S (%) A (%) AF(%) Note: = ement, FA = Fly ah * SO 3 content were alo raied to 4.00% and 8.00% by adding odium ulfate into the pate. fix (% by wt of cement) from model fix (% by wt of binder) from model FL1 FL2 FL3 FH1 FH2 FH3 fix (% by wt of cement) from experiment Fig. 9 Verification of model with variou cement pate in thi tudy. fix (% by wt of binder) from experiment Fig Verification of model with variou cement-fly ah pate in thi tudy. 394

9 Material and Structure / Matériaux et ontruction, Vol. 37, July 2004 fix (% by wt of cement) from model 3 A=2.43% 3 A=7.37% 3 A=9.10% 3 A=14.00% fix (% by wt of cement) from model cement pate Raheeduzzofar [4] Internal chloride fix (% by wt of cement) from experiment Fig Verification of model for variou 3 A content of cement pate. fix (% by wt of cement) from model 3 A=2.43%, SO 3 = % 3 A=7.59%, SO 3 = % 3 A=14.00%, SO 3 = % Arya [1] External chloride fix (% by wt of cement) from experiment Fig Verification of model for cement pate. fix (% by wt of binder) from model cement mortar fly-ah mortar concrete fix (% by wt of cement) from experiment with the tet reult from other reearcher. Fig. 11 and 12 how the verification of the model with reult from internal chloride tet, while Fig. 13 and 14 how the verification of model with reult from external chloride tet. A indicated in Fig. 11 and 12, the model can be ued to predict the chloride binding capacity of cement pate with variou 3 A content from 2.43% to 14.00% and SO 3 content from 1.7% to 8.0%. A illutrated in Fig. 13 and 14, the model can alo be applied to predict the chloride binding capacity of variou cement pate, mortar, fly-ah mortar and concrete. 6. ONLUSIONS Huain [8] Internal chloride Fig Verification of model for variou 3 A and SO 3 content of cement pate. Baed on the experimental reult, the model formation and the verification of model, the following concluion can be drawn. 1. The behavior of chloride binding capacity of cementfly ah cementitiou ytem wa time-dependent. Maruya [18] External chloride fix (% by wt of binder) from experiment Fig Verification of model for cement mortar, fly-ah mortar and concrete. hloride binding depended on the curing and chloride expoure period. Older pate prior to chloride attack bound le chloride than thoe expoed to chloride at younger age. Longer expoure period of pate reulted in larger chloride binding capacity. 2. ement pate with high calcium fly ah had higher fixed chloride content than thoe with low calcium fly ah. Thi i becaue the high calcium fly ah uually contain ome cementitiou component which can hydrate to bind chloride and alo to increae the early pozzolanic reaction o that more pozzolanic product can be produced epecially at the high replacement ratio. 3. A model for predicting chloride binding capacity of cement-fly ah cementitiou ytem wa propoed by conidering that the unhydrated aluminate ( 3 A) and aluminoferrite ( 4 AF) phae in cement were reponible for the chemical binding, while the hydrated product from cement and pozzolanic product from fly ah were reponible for phyical binding. 4. The propoed model can atifactorily predict the chloride binding capacity of variou cement pate and cement-fly ah pate with different mixture proportion, 395

10 Sumranwanich, Tangtermirikul propertie of cement and fly ah, curing time and chloride expoure period. AKNOWLEDGMENTS The author gratefully acknowledge the upport provided for thi reearch by the Thailand Reearch Fund (TRF). REFERENES [1] Arya,., Buenfeld, N.R. and Newman, J.B., Factor influencing chloride-binding in concrete, em. on. Re. 20 (2) (1990) [2] Ramachadran, S., Seeley, R.. and Polomark, G.M., Free and combined chloride in hydrating cement and cement compound, Mater. Struct. 17 (1984) [3] Theiing, E.M., Mebiu-Van De Laar, T. and De Wind, G., The combining of odium chloride and calcium chloride by the hardened Portland cement compound 3 S, 2 S, 3 A and 4 AF, Proceeding of 8 th International Sympoium on hemitry of ement, Rio de Janeiro, 1986, [4] Raheeduzzafar, Huain, E.S. and Al-Saadoun, S.S., Effect of cement compoition on chloride binding and corroion of reinforcing teel in concrete, em. on. Re. 21 (5) (1991) [5] Suryavanhi, A.K., Scantlebury, J.D. and Lyon, S.B., The binding of chloride ion by ulphate reitant Portland cement, em. on. Re. 25 (3) (1995) [6] izmadia, J., Balaz, G. and Tama, F.D., hloride ion binding capacity of aluminoferrite, em. on. Re. 31 (4) (2001) [7] Dhir, R.K., El-Mohr, M.A.K. and Dyer, T.D., hloride Binding in GGBS concrete, em. on. Re. 26 (12) (1996) [8] Huain, E.S., Raheeduzzafar and Al-Gahtani, A.S., Influence of ulfate on chloride binding in cement, em. on. Re. 24 (1) (1994) [9] Xu Y., The influence of ulphate on chloride binding and pore olution chemitry, em. on. Re. 27 (12) (1997) [10] Jenen, O.M.., Korzen, M.S.H., Jakoben, H.J. and Skibted, J., Influence of cement contitution and temperature on chloride binding in cement pate, Advance in ement Reearch 12 (2000) [11] Luping, T. and Nilon, L.-O., hloride binding capacity and binding iotherm of OP pate and mortar, em. on. Re. 23 (2) (1993) [12] Delagrave, A., Marchand, J., Ollivier, J.P., Julien, S. and Hazrati K., hloride binding capacity of variou hydrated cement pate ytem, Advanced ement Baed Material 6 (1) (1997) [13] Jenen, H.-U. and Pratt, P.L., The binding of chloride ion by pozzolanic product in fly ah cement blend, Advance in ement Reearch 2 (7) (1989) [14] Nipatat, N. and Tangtermirikul, S., ompreive trength prediction model for fly ah concrete, Thammaat International Journal of Science and Technology 11 (2000) 1-7. [15] Tangtermirikul, S. and Saengoy, W., Simulation of free water content of pate with fly ah, Reearch and Development Journal of the Engineering Intitute of Thailand 13 (2002) [16] Damidot, D. and Glaer, F.P., Thermodynamic invetigation of the ao-al 2 O 3 -aso 4 -al 2 -H 2 O ytem and the influence of Na 2 O, 10 th ongre on the hemitry of ement, Gothenburg, Sweden, 1997 [17] Minde, S. and Young, J.F., oncrete, (Prentice-Hall, New Jerey, 1981). [18] Maruya T., Tangtermirikul, S. and Matuoka Y., Modeling of chloride movement in the urface layer of hardened concrete, oncrete Library International of JSE 32 (1998) Paper received: February 9, 2003; Paper accepted: July 8,