STATISTICAL MODELS TO PREDICT MECHANICAL AND VISCO-ELASTIC PROPERTIES OF SCC DESIGNATED FOR PRECAST/PRESTRESSED APPLICATIONS

Transcription

1 STATISTICAL MODELS TO PREDICT MECHANICAL AND VISCO-ELASTIC PROPERTIES OF SCC DESIGNATED FOR PRECAST/PRESTRESSED APPLICATIONS Wu Jian Long (1) and Kamal Henri Khayat (1) (1) Department of Civil Engineering, Université de Sherbrooke, Sherbrooke, Québec, Canada Abstract The proportioning of self-consolidating concrete (SCC) often involves the adjustment of several mixture parameters to achieve some compromise between properties affecting fresh and hardened concrete properties. A factorial design was carried out to model the influence of key mixture parameters on several properties affecting mechanical and visco-elastic properties of SCC used for the construction of precast of prestressed structural beams. Mixture parameters modeled in this study included the binder content, binder type,, sand-to-total aggregate ratio (S/A), and dosage of thickening-type, viscosity-modifying admixture (VMA). In total, 16 SCC mixtures were developed to establish a factorial design with five main factors. Three replicate SCC mixtures were prepared to estimate the degree of the experimental error for the modeled responses. The mixtures were evaluated to determine several key responses that affect the performance of precast, prestressed concrete, including compressive strength development, modulus of elasticity, flexural strength, autogenous shrinkage, drying shrinkage, and creep. The derived statistical models enable to quantify the level of significance of each of the five investigated parameters on the mechanical and visco-elastic properties of SCC, which can simplifying the test protocol needed to optimize SCC. Key words: self-consolidating concrete; statistical models; mechanical properties; visco-elastic properties 1. INTRODUCTION Self-consolidating concrete (SCC) is highly flowable, non-segregating concrete that can spread into place, fill the formwork, and encapsulate the reinforcement without any mechanical consolidation [1]. The use of SCC for the manufacturing of precast, prestressed elements provides the benefits of increased rate of production and safety, reduced labor demand, and reduction in noise level at manufacturing facilities. SCC has been successfully used in various complex formwork sections and difficult concreting sections with high density of reinforcement [2-8]. Precast, prestressed bridge elements with congested reinforcement and tight dimensional geometry can then benefit from the use of such novel construction materials. 506

2 Key workability characteristics of SCC are described in terms of filling ability, passing ability, and resistance to segregation [9]. SCC used in precast, prestressed structural applications should satisfy these criteria to secure proper flow through densely reinforcement and complete filling of complex formwork systems. Typically, SCC used in precast, prestressed applications is made with relatively low of 0.32 to 0.40 [1]. Such low coupled with high cementitious materials content can provide high resistance to segregation and adequate mechanical and visco-elastic properties. For the successful design of SCC for precast, prestressed applications, it is important to understand the material properties and mix design parameters that can influence the workability, early-age properties, and mechanical properties. For example, proper estimate of compressive strength development, elastic modulus, and flexural strength of concrete as well as creep and shrinkage for a set of potential mix designs is important to estimate camber of prestressed members at the release of the prestressing load, as well as to determine prestress losses. In SCC, the paste content is typically higher than that in conventional concrete with normal consistency. Furthermore, SCC usually has lower coarse aggregate content, thus leading to greater creep and shrinkage potential than conventional concrete. Low coupled with high binder content can also lead to greater degree of autogenous shrinkage. Again, these complex mixture parameters should be taken into consideration in the design of SCC and the detailing of the prestressed elements. In order to gain better understand the influence of key mix design parameters and material constituents on the behaviour of SCC designated for the construction of prestressed, precast structural elements, a fractional factorial design was employed in this study to identify the relative significance of these primary parameters and their interactions on mechanical and visco-elastic properties of the concrete that are critical for precat, prestressed applications. Better understanding of the main mixture parameters and their interactions on the performance of SCC and knowledge of trade-offs among the various mixture parameters on various SCC properties can indeed simplify the test protocol needed to optimize SCC given a certain set of performance requirements [10]. 2 EXPERIMENTAL PROGRAM 2.1 Factorial design approach A fractional factorial design was used to evaluate the influence of mixture proportioning and constituent material characteristics on mechanical and visco-elastic properties of SCC. Four independent mixture proportioning parameters and one raw material parameter considered at five levels were used in the experimental design. The five modeled mixture parameters included the binder content (BC), binder type (BT), water-to-cementitious materials ratio (), dosage of thickening-type VMA, and volume of the sand-to-total aggregate ratio (S/A). The 16 mixtures used in this experimental design are summarized in Table 1. Three replicate central points were prepared to estimate the degree of experimental error for the modeled responses and to establish the degree of significance for each variable. All of the investigated mixtures considered in the experimateal design had an initial slump flow of 680 ± 20 mm that was obtained by adjusting the dosage rate of the high-range water reducing admixture (HRWRA). Absolute and coded values of the modeled parameters are presented in Table 2. The coded values are calculated as the difference between the absolute values and values 507

3 corresponding to the central points divided by the spread between the absolute values corresponding to 0 and 1, as shown below: Coded BC = (absolute BC 470) / 30 Coded = (absolute 0.37) / 0.03 Coded VMA = (absolute VMA 50) / 50 Coded S/A = (absolute S/A 0.50) / 0.04 Two types of Portland cement, i.e. Type MS and Type HE (similar to ASTM C 150 Type I/II cement and Type III cement) and one SCM (Class F fly ash) were used. The specific gravities of the Type MS cement, Type HE cement, and Class F fly ash are 3.14, 3.15, and 2.53, respectively. The Blaine fineness values are 390, 530, and 410 m 2 /kg, respectively. Two different binders were considered, which include Type MS cement and Type HE cement with of 20% Class F fly ash replacement, by mass. The binder content used in this study varied from 410 to 530 kg/m 3. All concrete mixtures were prepared with crushed coarse aggregate with 12.5 mm maximum size aggregate. A natural siliceous sand with a specific gravity of 2.66 conforming to AASHTO T 27 specifications was used. The particle-size distributions of the sand and coarse aggregate are within the AASHTO recommended limits. A polycarboxylate-based HRWRA complying with AASHTO M 194 (Type F) and an organic, thickening-type viscosity modifying admixture (VMA) were used in the SCC mixtures. Such admixtures are widely used in the precast industry in North America. The specific gravities of these two chemical admixtures are and 1.0, respectively. The solid content of the HRWRA and WMA are 20.3 and 6%, respectively. 508

4 Mix No. Table 1 Details of experimental program Coded values Absolute values Type Binder VMA* Binder type S/A** Binder (kg/m 3 ) VMA (ml/100 kg CM) Binder type S/A (%) Factorial mixtures Non AEA SCC (680 ± 20 mm slump flow) Fractional factorial points Central points MS HE*** MS HE MS HE MS HE MS HE MS HE MS HE MS HE MS-HE MS-HE 0.50 * Thickening-type VMA ** Crushed aggregate with MSA of 12.5 mm and natural sand *** Type HE cement + 20% Class F fly ash MS-HE 0.50 Table 2 Absolute and coded values of modeled parameters (-1 to +1) Coded Absolute Binder content (BC), kg/m³ Binder type (BT) Type MS Type MS + HE Type HE + 20%FA VMA content, ml/100 kg CM Sand-to-aggregate ratio (S/A), by volume

5 2.2 Mixing sequence and curing methods The SCC mixtures were prepared in 110-L batches using a drum mixer. The mixer was modified to promote greater shearing action of the concrete and was equipped with a speed gear to enable the simulation of concrete agitation at a low rotational speed after the end of the mixing cycle. The mixing sequence consisted of wetting the sand and coarse aggregate with half of the mixing water, followed by the addition of the binder. Once the aggregate particles were coated by a layer of cement paste, the HRWRA and the VMA diluted with the remaining mixing water were introduced over 30 seconds, and the concrete was mixed for 2.5 minutes. The concrete remained at rest in the mixer for 2 minutes for fluidity adjustment and to enable any large air bubbles entrapped during mixing to rise to the surface. The concrete was then remixed for 3 minutes. The fresh properties of SCC were measured at the ages of 10 and 40 minutes after cement and water contact. During that period, the concrete was agitated at low rotational speed of 6 rpm; the opening of the drum mixer was covered to prevent water evaporation. Concrete cylinders measuring and mm were sampled 10 minutes after the end of mixing to evaluate mechanical properties under three different curing conditions summarized in Table 3. All sampled specimens were cast in one lift without any mechanical consolidation. Some of the samples were covered and remained in the laboratory at 23 ± 2 C to air cure until the time of testing, while others were steam-cured according to the regime described in Fig. 1. Based on AASHTO, CSA, and PCI specifications, the maximum curing temperature in the concrete should not exceed 71 and 70ºC to prevent the occurrence of delayed ettringite formation. The standards also stipulate that the increase in temperature-time ratio should not exceed 22 and 20ºC/hr (depending on the standard), and the decrease in temperature-time ratio should be lower than 22 and 15ºC/hr for the AASHTO and CSA, respectively. In this investigation, the targeted release compressive strength after 18 hours of steam curing was 35 MPa. In some cases, the curing regime was slightly modified to achieve the targeted release compressive strength. The 56-day design compressive strength varied between 55 and 70 MPa, depending on the mix design. In addition to compressive strength, samples were subjected to steam curing to determine the 18-hour modulus of elasticity and to initiate creep and shrinkage testing at the prestress release time. Fig. 1 Steam curing regime specified by AASHTO and CSA 510

6 Curing methods Steamcured Stages I II III IV V Table 3 Curing conditions employed Details Ambient temperature for 2 hours after water-cement contact Temperature raised for 2 hours Concrete temperature maintained for 10 hours Temperature decreases over 2 hours to ambient temperature Air-curing until age of testing at 18 hours I 18 hours in molds with wet burlap at 23 ± 2 C Moist-cured II Moist-cured at 23 ± 2 C until testing age Air- I 18 hours in molds with wet burlap at 23 ± 2 C cured II Air-dried at 23 ± 2 C until testing age 2.3 Test methods As indicated in Table 4, the responses evaluated in the experimental design included compressive strength development, flexural strength, elastic modulus, autogenous shrinkage, drying shrinkage, and creep. In addition to standard methods to determine mechanical properties of SCC presented in Table 4, a number of special tests were used to investigate visco-elastic properties. Autogenous shrinkage was measured on prisms mm. The prisms were sealed immediately after removal from the molds at 18 hours using two layers of self-adhesive aluminum sheet to prevent any mass loss due to sample drying. The samples were stored at 23 ± 2 C until the end of testing. Autogenous shrinkage was monitored using embedded vibrating wire strain gages until stabilization. The autogenous shrinkage was obtained by subtracting the total shrinkage from the thermal deformation. The thermal expansion coefficient of the concrete was determined from the slope of the total deformation-temperature curve of concrete prisms subjected to control temperature changes. Two prisms were immersed in water at an approximate temperature of 50 C. Once the temperature of the samples was stabilized, the water was allowed to cool down to approximately 20 C. The resulting deformations were used to estimate the coefficient of the thermal expansion/contraction of the concrete. 511

7 Table 4 Testing program SCC behavior Property Test method Test age Number of samples per mixture Comments 18 hours 3 air-cured 3 steam-cured Mechanical properties Compressive strength ( mm cylinders) Modulus of elasticity ( mm cylinders) AASHTO T 22 ASTM C days 3 moist-cured 28 days 3 moist-cured 56 days 3 moist-cured 18 hours 2 air-cured 2 steam-cured 28 days 2 moist-cured Air curing - 50% ± 4% RH, 23 ± 2 C Moist curing - 100% RH, 23 ± 2 C Steam curing for 16 hours 56 days 2 moist-cured 7 days 3 moist-cured Flexural strength ( mm prisms) AASHTO T days 3 moist-cured Moist curing - 100% RH, 23 ± 2 C 56 days 3 moist-cured Autogenous shrinkage ( mm prisms) Embedded vibrating wire gages Over 1 month 2 per mixture Sealed prisms after removal from molds at release time Visco-elastic properties Drying shrinkage ( mm cylinders) AASHTO T 160 Over 11 months 3 per mixture Same curing regime used for release strength Creep ( mm cylinders) ASTM C 512 Over 11 months 3 per mixture Loading at release time Six mm cylindrical test specimens were cast to monitor creep and drying shrinkage. The specimens were steam cured until the age of 16 hours then demolded to grind their ends and fix external studs for deformation measurements. A digital-type extensometer was used to determine drying shrinkage and creep. Creep and shrinkage testing were started at the age of 18 hours. The applied creep loading corresponded to 40% of the 18-hour steam-cured compressive strength of the corresponding concrete. Initial elastic deformations were measured directly after creep loading. Creep and shrinkage specimens were tested at 23 ± 2 C and 50% ± 4% relative humidity for a period of 11 months; the long-term deformations were all stabilized at the completion of the testing. 512

8 3 TEST RESULTS AND DISCUSSION 3.1 Statistical models for mechanical properties Statistical models were established by multi-regression analyses. The estimate for each factor refers to the contribution of that factor to the modeled response. The coefficient and Prob.> [t] values of the derived models for compressive strength, modulus of elasticity (MOE), and flexural strength are presented in Table 5. Probability values less than 0.1 were considered as significant evidence that the factor has significant influence on the modeled response. Student s tests were run to evaluate the significance of the model factors and their second-order interactions on a given response. For each modeled response, the single-operator relative error corresponding to 90% confidence limit was used to perform the significance evaluation. Single-operator relative errors were determined using mixture corresponding to the central point of the models. The derived models are summarized in Table 6. As expected, the had the greatest influence on these properties. The HRWRA demand decreases with the increase in. The content and type of binder had considerable effect on mechanical properties. The MOE, and flexural strength responses appeared to be affected by S/A. In most cases, the use of thickening-type VMA did not have any significant effect on mechanical properties. The main findings from the models derived for mechanical properties are given as follows: The HRWRA demand of SCC decreases with the increase in and binder content. The use of Type HE cement and 20% Class F fly ash necessitates higher HRWRA demand than that of Type MS cement. The increase in binder content leads to higher 56-day compressive strength but lower 18-hour MOE and 7-day flexural strength. The increase in S/A decreases the MOE both at 18 hours (steam curing) and 56 days (moist curing). On the other hand, the increase in S/A leads to an increase in 7- and 56-day flexural strength. SCC made with Type HE cement and 20% Class F fly ash replacement exhibits higher compressive strength and MOE at 56 days but lower mechanical properties at 18 hours; this is mainly due to delayed setting resulting from higher HRWRA demand compared with concrete prepared with Type MS cement. 513

9 Table 5 Parameter estimates of derived models for mechanical properties ' 18-hour f c, MPa 56-day ' 18-hour MOE, MPa f c, MPa (steam-cured) (steam-cured) R² = 0.87 R² = 0.96 R² = 0.89 Model type Linear model Linear model Linear model Parameters Prob.> t Prob.> t Prob.> t Intercept Binder content NS* NS VMA NS NS NS NS Binder type S/A NS NS NS NS BC NS NS VMA NS NS NS NS BT NS NS NS NS VMA S/A NS NS NS NS BT S/A NS NS NS NS day MOE, MPa R² = day flexural, MPa R² = day flexural, MPa R² = 0.83 Model type Linear model Linear model Linear model Parameters Prob.> t Prob.> t Prob.> t Intercept Binder content NS NS NS NS Binder type NS NS NS NS S/A BC NS NS NS NS BC BT NS NS BT NS NS NS NS NS NS * NS: Not significant 514

10 Table 6 Derived statistical models for mechanical properties Property Age Derived equations R² Compressive strength*, MPa Modulus of elasticity, GPa Flexural strength, 18 hours 56 days 18 hours 56 days 7 days ( BT) 0.88 (VMA S/A) 0.77 BT 0.67 ( VMA) 0.56 VMA (BC ) BT BC (BC ) 0.71 BT 0.66 (BT S/A) 0.59 S/A 0.54 BC S/A (BC BT) BT (BC ) S/A 0.62 BC MPa 56 days S/A (BC BT) 0.83 [HRWRA demand] 0.5** (L/100 kg CM) * Linear model ** Power model, Y = X a BC BT ( BT) VMA Exploitation of statistical models for mechanical properties Figure 2 illustrates the 18-hour compressive strength contour diagram where the effect of variations in and S/A on strength are highlighted. In general, SCC prepared with Type MS cement exhibited higher 18-hour steam-cured compressive strength compared with concrete made with Type HE and 20% FA. For example, SCC made with the same 0.48 S/A and 0.35 but different binder type of Type MS and Type HE + 20% FA had 18-hour compressive strength of 33.5 MPa and 36.5 MPa, respectively. Furthermore, for a given, mixtures made with lower S/A had higher compressive strengths. For example, with 0.35 and Type MS binder, SCC made with 0.46 S/A developed 18-hour compressive strength of 37 MPa compared to 35 MPa exhibited by SCC made with 0.54 S/A. The binder content and contour diagram of the 56-day compressive strength is presented in Fig. 3. Binder content and had a direct influence on 56-day compressive strength. For the same and binder content, SCC made with Type HE cement and 20% FA exhibited approximately 4 MPa higher compressive strength than SCC made with Type MS cement. On the other hand, for lower than 0.39, higher binder content is desired to achieve higher compressive strength. 515

11 h compressive strength (MPa) h compressive strength (MPa) S/A ratio S/A ratio (a) Type MS (b) Type HE + 20% FA Fig. 2 - S/A contour diagrams of 18-hour compressive strength (MPa) d compressive strength (MPa) d compressive strength (MPa) Binder content (kg/m³) Binder content (kg/m³) (a) Type MS (b) Type HE + 20% FA Fig. 3 - binder content contour diagrams of 56-day compressive strength (MPa) 3.3 Validation of statistical models for mechanical properties Eight additional SCC mixtures proportioned in the range of the factorial design were used to validate the statistical models for mechanical properties. Again, the new mixtures had initial slump flow values of 680 ± 20 mm. The three central points were also used to evaluate the error in the 90% confidence limit. As shown in Fig. 4, these mixtures were chosen to cover a wide range of mixture proportioning within the modeled region. Mixture proportioning of these SCC mixtures is presented in Table

12 1 1-1 #3 # S/A VMA -1 1 #1 #2 Binder content VMA 1-1 # 9, 10, 11 Binder content Central points 1 S/A #8 1 #4 #6 1 #7-1 Mix no Fig. 4 Eleven SCC mixtures used to validate derived models Table 7 Mixture proportioning used to validate the derived statistical models Coded values Absolute values Binder VMA Binder type S/A Binder (kg/m³) VMA (ml/100 kg CM) Binder type 1-1/ / MS /3-1 1/ HE* / / MS /3-1/ HE /3 1/ MS / / HE /3 1-1/ MS / / HE Central points MS-HE MS-HE MS-HE 0.50 * HE = HE + 20% FA The responses of the statistical models included compressive strength and elastic modulus at 18 hours and 56 days, as well as the flexural strength at 7 and 56 days. Estimated relative errors corresponding to the 90% confidence limit shown in Table 8 were determined from the central point mixtures. S/A (%) 517

13 Table 8 Mean values and relative errors of central points for mechanical properties (90% confidence limit) Relative error, Mean 90% confidence limit % Value 18-hour compressive strength, MPa day compressive strength, MPa hour modulus of elasticity, GPa day modulus of elasticity, GPa day flexural strength, MPa day flexural strength, MPa Based on the relative error estimates, comparisons of predicted-to-measured mechanical properties are illustrated in Fig. 5. Data points above the solid diagonal 1:1 line indicate that the derived statistical model overestimates the real values, while those below the line indicate an underestimation of the actual values. Most of the data are located within the error limit, thus indicating good estimate of the 18-hour compressive strength. The mean of predicted-to-measured values for 18-hour compressive strength was 1.0, as shown in Fig. 5 (a). Relationships between measured and predicted values of 56-day compressive strength are presented in Fig. 5 (b). The evaluated SCC mixtures exhibited 56-day compressive strength ranging from 55 to 76 MPa. Seven data points are inside the error range, while four points are close to the diagonal lines. The mean predicted-to-measured ratio of the 56-day compressive strength was 1.01 with R 2 of The derived model is therefore shown to provide accurate predicting of the 56-day compressive strength. Comparisons between predicted and measured values for the modulus of elasticity determined at 18 hours and 56 days are illustrated in Figs. 5 (c) and (d), respectively. As can be observed, all of the data are located within the 90% confidence limit error range. The mean predicted-to-measured MOE ratios at 18 hours and 56 days were 0.99 and 1.0, respectively. Comparisons between measured and predicted 7- and 56-day flexural strength are shown in Figs. 5 (e) and (f), respectively. As illustrated in this figure, most of the data points are located within the 90% relative error range. The predicted-to-measured 7- and 56-day flexural strength ratios were 0.99 and 1.0, respectively. These models show good accuracy in estimating flexural strength. 518

14 Predicted 18-h compressive strength (Mpa) N = R 2 = Predicted 56-d compressive strength (MPa) N = 11 R 2 = Measured 18-h compressive strength (MPa) Measured 56-d compressive strength (MPa) (a) 18-hour compressive strength (b) 56-day compressive strength Predicted 18-h modulus of elasticity (MPa) N = 11 R 2 = 0.87 Predicted 56-d modulus of elasticity (MPa) N = 11 R 2 = Measured 18-h modulus of elasticity (MPa) Measured 56-d modulus of elasticity (MPa) (c) 18-hour modulus of elasticity (d) 56-day modulus of elasticity Predicted 7-d flexural strength (MPa) N = 11 R 2 = 0.85 Predicted 56-d flexural strength (MPa) N = 11 R 2 = Measured 7-d flexural strength (MPa) Measured 56-d flexural strength (MPa) (e) 7-day flexural strength (f) 56-day flexural strength Fig. 5 Comparison between predicted and measured mechanical properties 519

15 3.4 Statistical models for visco-elastic properties As in the case of mechanical properties, the is shown to have the highest influence on visco-elastic properties. The type of binder also had considerable effect on autogenous shrinkage. On the other hand, drying shrinkage and creep varied mainly with the binder content. In most cases, the value of the S/A and incorporation or not of VMA had minor effects on the measured responses of the hardened SCC. The coefficient and Prob.> [t] values of the derived models are presented in Table 9, and the derived models are summarized in Table 10 with the mixture variables expressed as coded values. The main findings from the models derived for visco-elastic properties can be summarized as follows: Autogenous shrinkage of SCC decreases with the increase in. The increase in binder content leads to an increase in creep and drying shrinkage. Theoretically, for a given binder content, drying shrinkage increases with increase in ; however, the derived statistical models showed an opposite trend since the measured drying shrinkage also include the autogenous shrinkage component that decreases with the increase in. SCC made with higher S/A is shown to exhibit higher drying shrinkage after 112 days of drying. SCC made with Type HE cement and 20% Class F fly ash can develop higher autogenous shrinkage and creep in compression compared with concrete prepared with Type MS cement. However, for the tested binder types, it is shown that the binder type has no significant effect on drying shrinkage. 3.5 Exploitation of statistical models for visco-elastic properties The contour diagrams of the 56-day autogenous shrinkage in Fig. 6 illustrate the trade-offs between binder content and for mixtures made with Type MS and Type HE with 20% of fly ash binder. The S/A and VMA are set to the central point values. For the same binder content and, SCC proportioned with Type HE cement and 20% of fly ash exhibited higher autogenous shrinkage than those prepared with Type MS cement. For example, for a binder content of 450 kg/m 3 and of 0.363, SCC made with Type HE cement and 20% of fly ash and Type MS cement can develop approximate 56-day autogenous shrinkage values of 265 and 130 μstrain, respectively. As indicated in Fig. 6 (a), the increase in from to 0.37 enables the increase of binder content from 450 to 465 kg/m 3 to maintain a constant autogenous shrinkage of 132 μstrain. 520

16 Table 9 Parameter estimates of derived models for visco-elastic properties Autogenous shrinkage at 7 days (μstrain) R² = 0.96 Autogenous shrinkage at 56 days (μstrain) R² = 0.93 Drying shrinkage at 28 days (μstrain) R² = 0.78 Model type Linear model Linear model Linear model Parameters Prob > t Prob > t Intercept Prob > t Binder content NS* NS NS NS Binder type NS NS BC NS NS BC BT NS NS NS NS BC S/A NS NS NS NS VMA NS NS NS NS BT NS NS NS NS VMA BT NS NS NS NS Drying shrinkage at 112 days (μstrain) R² = 0.96 Creep at 28 days (μstrain) R² = 0.75 Creep at 112 days (μstrain) R² = 0.89 Model type Linear model Linear model Linear model Parameters Prob > t Prob > t Intercept Prob > t Binder content NS NS NS NS Binder type NS NS S/A NS NS NS NS BC VMA NS NS NS NS BC S/A NS NS NS NS VMA NS NS NS NS BT VMA BT NS: Not significant 521

17 Table 10 Derived statistical models for visco-elastic properties Property Age Derived equations R² Autogenous shrinkage, μstrain 7 days 56 days BT 21.6 (BC ) 20.1 ( BT) 15.9 (BC BT) BT (BC ) (BC S/A) Drying shrinkage, μstrain 28 days 112 days ( VMA) + 35 BC (VMA BT) BC ( VMA) ( BT) 40.6 (BC VMA) S/A (VMA BT) Creep, μstrain 28 days 112 days BT (VMA BT) ( BT) BT (VMA BT) BC ( BT) 32.9 (BC S/A) d 56-d autogenous shrinkage (μstrain) (µstrain) d autogenous shrinkage (μstrain) (µstrain) 326 Binder content (kg/m³) Binder content (kg/m³) (a) Type MS (b) Type HE + 20%FA Fig. 6 Binder content contour diagrams of 56-day autogenous shrinkage (S/A = 0.50, VMA = 50 ml/100kg CM) Drying shrinkage contour diagrams are presented in Fig. 7 to show trade-offs between binder content and S/A for mixtures made with Type MS cement and with Type HE and 20% fly ash replacement. As expected, the increase in binder content and S/A are shown to increase 522

18 the drying shrinkage. Moreover, for the same binder content and S/A, SCC made with Type HE and 20% of fly ash binder exhibited approximately 20% higher 112-day drying shrinkage values compared with those proportioned with Type MS binder. For example, for the binder content of 470 kg/m 3 and 0.50 S/A, mixtures prepared with Type HE cement and 20% of fly ash and Type MS cement had 112-day drying shrinkage values of 540 and 445 μstrain, respectively d Drying drying shrinkage (μstrain) (µstrain) d 112-d Drying drying shrinkage shrinkage (μstrain) (µstrain) Binder content (kg/m³) Binder content (kg/m³) S/A ratio (a) Type MS S/A ratio (b) Type HE + 20%FA Fig. 7 Binder content S/A contour diagrams of 112-day drying shrinkage ( = 0.40, VMA = 55 ml/100kg CM) 4 CONCLUSIONS The statistical models established using a factorial design approach can be used to quantify the effect of mixture parameters and their coupled effects on mechanical and visco-elstic properties of SCC. The statistical models are valid for a wide range of mixture proportioning and provide an efficient means to determine the influence of key variables on SCC properties. Based on the statistical models derived from the factorial design in this study, the major conclusions derived from the statistical models are as follows: The exhibited the most significant effect on the investigated mechanical properties, autogenous and drying shrinkage, and moderate impact on creep. The binder type had considerable effect on mechanical properties and autogenous shrinkage and creep. The binder content had considerable effect on mechanical properties and the greatest effect on drying shrinkage The S/A had significant effect on modulus of elasticity and flexural strength and minor effect on visco-elastic properties. 523

19 The use of thickening-type VMA did not have any significant effect on mechanical properties, except in the case of 18-hour compressive strength, and minor effect on visco-elastic properties. Based on the above results, recommendations for the proportioning of SCC in terms of mechanical and visco-elastic properties are given in Table 11. Darkened areas indicate better performance for each property. For example, SCC made with Type MS cement is shown to develop less creep and shrinkage than similar mixtures prepared with Type HE cement and 20% Class F fly ash replacement; however, the use of the latter binder system can secure higher mechanical properties. Table 11 Recommendations for proportioning SCC for precast prestressed applications Binder type Binder content S/A MS HE+20% Fly ash 440 kg/m³ 500 kg/m³ hour 56-day ' f c ' c 18-hour MOE f 56-day MOE 7-day flexural strength 56-day flexural strength Autogenous shrinkage Drying shrinkage Creep * Darkened areas indicate better performance for each property. ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support provided by the Transportation Research Board (TRB) of the National Academies (NAS-NRC) of the United States of America for NCHRP Project (NCHRP Report 628). The assistance of Dr. Soo-Duck Hwang and Mr. Guillaume Lemieux during this project is especially acknowledged. 524

20 REFERENCES [1] ACI Committee 237R-07, Self-Consolidating Concrete, 2007, 30 pages. [2] Juvas, K., The European Experience of Working with Self-Compacting Concrete in the Precast Concrete Industry, Proceedings of the Combining the Second North American Conference on the Design and Use of Self-Consolidating Concrete and the Fourth International RILEM Symposium on Self-Compacting Concrete, Vol. 2, pp , Chicago, USA, October 30-November 2, [3] Juvas, K., Experiences of Working with Self-Compacting Concrete in the Precast Industry, Proceedings of the 5 th International RILEM Symposium on Self-Compacting Concrete, pp , Ghent, Belgium, September 3-5, [4] Pellerin, B., Lamotte, J., Gnagne, C., and Canevet, C., Use of Dedicated Admixtures Eases the Implementation of SCC in the Precast Industry, Proceedings of the Combining the Second North American Conference on the Design and Use of Self-Consolidating Concrete and the Fourth International RILEM Symposium on Self-Compacting Concrete, Session J-3, Chicago, USA, October 30-November 2, [5] Naito, C., Hoover, M., Applicability of Self-Consolidating Concrete for Use in Precast Bridge Beam Construction, Proceedings of the Combining the Second North American Conference on the Design and Use of Self-Consolidating Concrete and the Fourth International RILEM Symposium on Self-Compacting Concrete, Session E-3, Chicago, USA, October 30-November 2, [6] Camacho, R., Afif, R., Corona, G., Roman, H., and Sanchez, M., Applications of SCC Technology for Precast/ Prestressed Elements: the Mexican Experiences, Proceedings of the 5 th International RILEM Symposium on Self-Compacting Concrete, pp , Ghent, Belgium, September 3-5, [7] Hwang, S. D., Khayat, K. H., Bonneau, O., and Mayen-Reyna, D., Workability Requirements of Self-Consolidating Concrete Used in Casting Highly Restricted Structural Sections, Proceedings of the 4 th International Conference on Concrete Under Severe Conditions (CONSEC 04), pp , Seoul, Korea, [8] Ouchi, M., Nakamura, S., Osterson, T., Hallberg, S., and Lwin, M., Applications of Self-Compacting Concrete in Japan, Europe and the United States, ISHPC, pp 1-20, [9] Khayat, K. H., Workability, Testing, and Performance of Self-Consolidating Concrete, ACI Materials Journal, Vol 96 (3), pp , [10] Khayat, K. H., Ghezal, A., and Hadriche, M., Factorial Design Models for Proportioning Self-Consolidating Concrete, RILEM Materials and Structures, Vol 32, pp , November