Optimization of Mechanical Properties and Durability of Reactive Powder Concrete

INTRODUCTION

(ProQuest Information and Learning: ... denotes formulae omitted.)

The adaptation of fiber reinforced polymers (FRPs) into the construction industry provides opportunities for many applications. In the area of prestressing/post-tensioning, anchorage of high-strength carbon FRP tendons remains a problem.1 If an anchorage system could be developed that provides the flexibility associated with steel tendons and anchorages, then more rapid use of CFRP tendons would occur. In particular, if the anchorage were made from nonmetallic materials, a corrosion-free system would become feasible. To this end, concrete anchors wrapped with CFRP sheets were developed.1,2 These anchors were too large for implementation in practice, so the objective of the work described herein was to develop a concrete capable of providing the required strength and toughness for an anchorage more of the size of a standard metallic one.

Innovative portland cement concretes are categorized according to their mechanical, chemical, and physical properties. Concretes in the chemically bonded ceramics (CBC) category can be divided into macro-defect-free (MDF) and densified-with-small-particles (DSP) concretes. DSP materials include both ultra high-performance concrete (UHPC) and reactive powder concrete (RPC).3,4 These relatively new materials are characterized by an extremely dense microstructure, compressive strengths in the range of 150 to 250 MPa (22 to 36 ksi) (UHPC), or 200 to 840 MPa (29 to 122 ksi) (RPC) and superior durability performance.

Reactive powder concrete is a special type of ultra highstrength superplasticized silica fume concrete, often fiber-reinforced, with improved homogeneity because traditional coarse and fine aggregate are replaced by very fine sand with particle sizes in the range of 100 to 400 µm (3937 to 15,748 ?in.).5-8

RPC properties are attractive for the purpose in mind because compressive strengths up to 800 MPa (116 ksi) having been recorded, but more typically in excess of 200 MPa (29 ksi). Flexural strengths up to 50 MPa (7.3 ksi) have been reported-the latter achieved when stainless steel fibers were included in the mixture.6,7

Freezing and thawing is expected not to have a deleterious effect on dense compacted RPC specimens because the material is free from air and water pockets or cracks.

RESEARCH SIGNIFICANCE

Because there is little research in the area of durability and mechanical properties of RPC, this study dealing with the durability and mechanical properties of RPC will be very useful to concrete, and in particular RPC technology. The objectives of this study were to develop a tough and durable RPC. The authors assessed the influences of presetting pressure, post-setting heat treatment, and the inclusion of carbon fibers independently while seeking optimum parameters.

The authors performed tests on cylindrical specimens (50 mm [2 in.] in diameter by 100 mm [4 in.] in length) cast in a specially designed steel mold. A mixture design and post-casting treatment process was developed to attain high compressive strengths. Different mixtures of RPC were prepared (plain and fiber reinforced mixtures) with a water-cement ratio (w/c) of 0.13.

Previous freezing-and-thawing durability tests on RPC were performed on uncracked specimens. Therefore, grooves were introduced on each side of some specimens to simulate cracks or microcracks, and permit water ingress to those specimens. Tests performed herein were on precracked 279 x 279 x 406 mm (11 x 11 x 16 in.) long specimens. Twenty such prisms were cast (10 plain and 10 fibered), which included four prisms beyond testing needs to compensate for any prisms damaged in the precracking process. Six cylinders, 50 mm (2 in.) in diameter by 100 mm (4 in.) in length (three plain and three fibered RPC), were also cast and cured for 14 days for determining compressive strength. Fracture parameters were also assumed.

For these tests, 101.6 x 101.6 x 355.6 mm (4 x 4 x 14 in.) (width x depth x length) beams with a loaded span (S) of 300 mm (11.8 in.) were used.

EXPERIMENTAL METHODS

Selection of carbon fibers

The carbon fibers to be used were selected via the properties supplied by the commercial manufacturers or from available literature. The methods used to select the fiber were the general usefulness ranking (GUR) and the weighted properties method (WPM). The GUR was conducted among the different grades of each carbon fiber separately, while the WPM was conducted on the different selected grades.9

Steel mold and specimens

A specially designed steel mold was machined from AISI 3040 steel. A schematic of the steel mold is shown in Fig. 1. The mold consisted of five pieces, a base, and four wedges, tied together with 16 high-strength steel bolts. Also, eight precision pins were incorporated to ensure perfect mating of the mold parts when assembled. The mold was designed to cast a solid cylinder 50 mm (2 in.) in diameter and 100 mm (4 in.) high, according to ASTM standard requirements for the compression test.

The mold incorporated small grooves between every mating face. These grooves had an average width of 4 ?m (157.5 ?in.), and were included to allow excess water and air bubbles to escape when the presetting pressure was applied, but not the solid mixture ingredients. A picture of the steel mold is shown in Fig. 2(a). Side and bottom grooves are visible in Fig. 2(b). The inside of the mold was machined to a very smooth finish to facilitate easy release of the cast RPC cylinders. Six cylinders were cast for each optimization parameter.

Steel molds were also used to manufacture the freezing-and-thawing specimens. The molds facilitated the embedment of stainless steel studs in the ends of the prisms with a gauge length between the studs of 330 mm (13 in.). The specimens were covered with plastic and left to set at room temperature (23 ?C [73.4 ?F]) for 24 hours after casting. Twenty prisms (10 plain RPC and 10 fiber-reinforced RPC) were cast, including four extra prisms to compensate for any prisms damaged in the precracking process.

Eight cylinders, (four nonfibered and four carbon fiber-reinforced RPC), were also cast and cured for 14 days to test for compressive strength.

Two sets of specimens were tested for fracture mechanics (plain [P] and CFRRPC [CF]), involving pairs of notched and unnotched specimen beams.

RPC mixture design and specimens

Before testing for durability and mechanical properties of RPC, different mixture proportions of RPC (plain or with fibers) were considered and tested for the highest compressive strength. Mixtures were designed to have the lowest w/c and optimal carbon fiber content. The optimization process also included different curing techniques and temperatures.

Freezing-and-thawing specimens and their abbreviated annotations were:

* Plain RPC (P);

* Plain precracked RPC (P^sub c^);

* Carbon fibered-reinforced RPC (CF); and

* Carbon fibered-reinforced precracked RPC (CFC).

Mixing and consolidating

The following techniques were used to mix, cast, and cure the RPC. Special care and accuracy is needed while mixing RPC.6-8,10-13

Mixing-Dry ingredients were mixed first until a homogeneous mixture was reached (based on color and visual appearance). Carbon fibers were added to the dry mixture, in order to break the fiber pellets and avoid the entanglement or balling of the fibers that tends to occur when the mixing water is added. Fifty percent of the mixing water and high-range water-reducing admixture were added. Five minutes later, the remaining high-range water-reducing admixture and mixing water were added. The total mixing time was between 10 and 15 minutes, at a very high speed (30 revolutions/second).

Molding and placing-The steel mold was placed on a vibrating table, and the RPC added in portions while the mold was being vibrated. The vibrating or consolidating time was between 5 and 10 minutes. The RPC was very thick and viscous, as the mixture had zero slump. Consolidation was accelerated by the vibration.

Curing-The RPC cylinders were compressed during setting under 50 kN (11.2 kip) each (26 MPa [3.8 ksi]) for 24 hours, then cured under a temperature of 150 ?C (302 ?F). The presetting load was applied with an electrohydraulic closed-loop test machine.

The freezing-and-thawing prisms and accompanying cylinders were cured in the fog room for 14 days and immersed in containers containing lime-saturated water to avoid leaching. After curing, deposited lime was cleaned off these samples using tap water and the samples were dried with a cloth.

Fracture mechanics specimens were cast in triplicate according to the optimum mixture design. All specimens were subjected to 50 kN (11.2 kip) (26 MPa [3.8 ksi]) presetting load for 24 hours before being heat treated at 150 ?C (302 ?F). The weight w of each specimen was obtained prior to testing.

Demolding-No wax or demolding agents were required for the RPC as the RPC was easy to demold. The cast surface was very shiny and smooth with a surface texture close to polished marble or granite.

Precracking of freezing and thawing specimens-Only 16 specimens were tested for freezing-and-thawing durability because four specimens were destroyed during the precracking process. Compressive load was applied gradually until cracking was heard or observed (less than the compressive strength calculated). Four specimens were failed completely in this process.

Freezing-and-thawing test parameter measurements

A freezing-and-thawing test was conducted according to the ASTM C 666 standard.14 Specimens were placed in a tempering tank for 24 hours with the temperature maintained between -1.1 ?C and +2.2 ?C (-2 ?F and +4 ?F) of the target thaw temperature for the specimens in the actual freezing-and-thawing cycles.

Immediately after removing the specimens from the conditioning bath, an arrow was drawn on each side of each specimen to indicate the top and bottom of each specimen (for average length measurements), and the following data were recorded:

* Fundamental transverse frequency (acoustic transducer);

* Weight (digital scale); and

* Average length (length gauge).

The fundamental transverse frequency was always measured from the same side of the specimen. Also, the average length change was always measured the same way on the specimens. Specimens were then replaced in the thawing water to begin the thawing phase of the freezing-and-thawing cycle, to start the tests. The fundamental transverse frequency, average length, and weight were recorded after every 36 cycles, after the specimens had been dried lightly with a piece of cloth. When the length measurements were taken, the specimens were spun clockwise with approximately the same speed every time to ensure contact between the stainless steel studs in the prisms and the two measuring pins in the length measurement apparatus. The container was rinsed and clean water added before the specimen was returned to the bath. Specimens were returned to the apparatus in a random manner and rotated end over end. The apparatus was checked every day for water level and the water was topped off when water had evaporated.

Fracture mechanics

To evaluate the fracture parameters of RPC and CFRRPC, fracture toughness tests were conducted using four-point bending. Pairs of notched and unnotched 101.6 x 101.6 x 355.6 mm (4 x 4 x 14 in.) (width x depth x length) beams were tested.

A schematic diagram of the test set-up and specimen dimension parameters is shown in Fig. 3(a). The span-todepth ratio was 3.0 for all specimens. Specimens were notched using a 3 mm (0.12 in.) thick carbide saw. The crack tip was semi-circular. The load was applied downward on the specimens at the third points, as shown in Fig. 3(a) and (b). The choice of four-point bending was based on the need to determine J^sub IC^ , which requires the application of pure bending.15

A closed-loop testing system was developed using a feedback signal from a crack mouth opening displacement (CMOD) clip gauge attached to the specimen. This feedback mechanism provided stable, controlled failure of the concrete. The CMOD and the applied load were recorded using a data acquisition system sampling at 5 Hz (five readings/second). The midspan deflection was measured simultaneously with the load using a linear variable differential transducer (LVDT). Two LVDTs were placed over the supports to eliminate the error due to the deformations at the supports.16 Loading was applied in displacement control (0.01 mm/second) using an MTS machine. The machine was programmed to change the rate of loading from 0.01 to 0.00033 mm/second (393.7 to 13 µin./second) according to the CMOD feedback signal, which provided a controlled rate of increment of CMOD. The MTS machine and loading set-up are shown in Fig. 3(b).

EXPERIMENTAL RESULTS AND DISCUSSION

Tests for optimal parameters

Carbon fiber content-Carbon fibers are coated with a sizing material (1.6% per weight of fiber) to help handling of the fibers (winding, avoiding stickiness, etc.). This sizing is designated as Type 5, with a basic chemical structure of Bisphenol A, diglycidylether, and Copolyester of saturated fatty acids. This sizing is designated for use with polymer composites incorporating carbon fibers, such as epoxy and vinyl ester resin systems. Heating of the fibers leads to polymerization and eventual decomposition of the sizing material, which leads to loss of bond strength (debonding) between the fibers and RPC. Thus, heat-treating the fibers to vaporize the sizing material prior to casting was necessary in order to avoid debonding during post-setting heat treatment. To ensure full removal of the sizing material, carbon fibers were heat treated at 500 ?C (932 ?F), but only for 1 hour. The process was not carried out for more than 4 hours in order to avoid evaporating the filament binding agent as suggested by the carbon fiber manufacturer.17

Two sets of tests were conducted with and without the sizing material. A significant rise in compressive strength results was obtained when the sizing material was removed (for example, at 150 ?C [302 ?F] [the compressive strengths were 184 and 282 MPa (26.7 and 41 ksi) for non-heat-treated and heat-treated carbon fibers, respectively]). Compressive strength versus increasing amounts of heat-treated carbon fiber in the RPC is shown in Fig. 4. The optimal amount of carbon fibers for use with RPC was taken as 0.125 by weight of cement, which resulted in a maximum compressive strength (229 MPa [33.2 ksi]).

Presetting pressure-Pictures showing the matrix at a microscopic level, of samples subjected to 0 and 50 kN (0 and 26 MPa) presetting load are shown in Fig. 5(a) and (b). The air voids and water pockets present in the zero pressure samples disappear with the application of presetting pressure.

Different values of presetting load ranging from 0 to 400 kN (0 to 90 kip) were investigated in intervals of 50 kN (11.2 kip). The load was applied for 6 hours. It was found that applying a presetting force of between 50 and 100 kN (11.2 and 22.5 kip) resulted in a maximum compressive strength for the size of the specimen used. Higher values of pressure resulted in lower compressive strength values. This was thought to be due to microcracks that could be induced on the release of the presetting load. With load application, the aggregates are compressed, so on the release of load, the aggregates expand. If the expansions are large enough, they would cause microcracking in the specimen.

Heat-treating temperature and curing technique-Heat treatment of CFRRPC after setting is essential in order to accelerate both the pozzolanic reaction of the silica fume and quartz, which alters the microstructure of RPC, as well as to remove any remaining excess mixing water. Temperature can also heal microcracks produced by the application and removal of presetting load. Heat-treating temperatures from room temperature (23 to 400 ?C [73.4 to 752 ?F]) were thus investigated in increments of 50 ?C (122 ?F).

Heat treatment was performed in an autoclave oven. In order to avoid thermal shock that could lead to cracking, the temperature was increased at a rate of 50 ?C (122 ?F) per hour to the desired temperature; at higher temperature gradients, multiple cracking and disintegration of specimens was experienced. Cooling was also carried out gradually for the same reason. Specimens were heat-treated for a period of 7 days. Compressive strength versus curing temperature for 0 and 100 kN (22.5 kip) presetting load are shown in Fig. 6.

The compressive strength increases rapidly with curing temperature between 23 and 150 ?C (73.4 and 302 ?F) due to the acceleration of the hydration process. Compressive strength falls between 150 and 200 ?C (302 and 392 ?F), due to the rapid evaporation of internal water, causing incomplete hydration and an increase in porosity. Compressive strength rises again between 200 and 300 ?C (392 and 572 ?F), due to the pozzolanic reaction of quartz, which can be activated at these temperatures, causing the formation of very dense calcium silicate hydrate compounds very low numbers of water molecules (for example, Xonotlite). Another drop in compressive strength occurs for temperatures above 300 ?C (572 ?F). This drop is related to the decomposition and evaporation of the high-range water-reducing admixture and its components, which leads to an increase in the micro-porosity with a concomitant decrease in compressive strength.

Hence, temperatures between 100 to 150 ?C (212 to 302 ?F) and 200 to 300 ?C (392 to 572 ?F) resulted in the highest compressive strengths. With temperatures above 200 ?C (212 ?F), however, some surface microcracking was noticed on some of the test specimens when they were removed from the oven. The number of cracks and their width increased with increasing curing temperature and curing period. However these cracks were surface cracks, probably caused by differential expansion between the internal core and the surface of the specimen. The number and distribution of cracks varied between specimens, but the cracks appeared to have minimal effect on the compressive strength. Specimens with few cracks had compressive strength values close to those of specimens with more cracks.

Post-set heat treatment processes probably modify the chemical composition of the hydrated products by reducing the CaO/SiO^sub 2^ and H^sub 2^O/CaO ratios, thus favoring the formation of tobermorite (Ca^sub c^Si^sub 6^O^sub 16^(OH)^sub 2^ ? 8H^sub 2^O) at temperatures below 200 ?C (392 ?F) and Truscottite (Ca^sub 14^(Si^sub 24^O^sub 58^)(OH)^sub 8^ ? 2H^sub 2^O), Gyrolite (NaCa16(AlSi^sub 23^O^sub 60^)(OH)^sub 8^ ? 14H^sub 2^O), Xonotlite (Ca^sub 6^Si^sub 6^O^sub 17^(OH)2) and Hillebrandite (Ca6(Si3O9)(OH)6) at higher temperatures.10

Cracks present in specimens cured at temperatures above 300 ?C (572 ?F) were thought to be induced by residual expansive behavior attributed to the formation of hydrates exhibiting relatively low density such as tobermorite. For T = 300 ?C (572 ?F), however, the high-range water-reducing admixture begins to decompose and this continues as the temperature increases. For T = 250 ?C (482 ?F), porosity and threshold pore sizes increased, due to the formation of Xonotlite, which was accompanied by the release and evacuation of water. A minimum value of porosity, corresponding to small pore diameters, is obtained for RPC pressurized during setting and heat treated between 150 and 200 ?C (302 and 392 ?F).9,18-20

Other curing regimes investigated were:

* Fog room (60% humidity);

* Autoclave oven; and

* Room temperature (23 ?C [73.4 ?F]).

The compressive strengths obtained with each curing technique are shown in Fig. 7. Specimens heat-treated in the autoclave oven at 150 ?C (302 ?F) were the strongest (245 MPa [35.5 ksi]). Fog room specimens had the lowest compressive strengths, whether for plain specimens or mixtures with carbon fibers. The low compressive strength for specimens in the fog room can be attributed to the lack of, or at best a slow, pozzolanic reaction, and to the increase in water content.

Optimal CFRRPC mixture composition-Based on the aforementioned test results, a mixture design and curing technique were chosen to achieve the maximum compressive strength for CFRRPC. Mixture design proportions for both plain and carbon fiber RPC along with the chosen curing regime are given in Table 1.

The minimum value for a presetting load of 50 kN (11.2 kip) (26 MPa [3.8 ksi]) and a curing temperature of 150 ?C (302 ?F) were selected in order to minimize the cost of CFRRPC.

Comparative test-After all the aforementioned tests were performed, a last set of tests was conducted with ultra high modulus and strength carbon fibers.

The carbon fibers were also coated with a Type 5 sizing material. The 0.4% per weight of fiber of the sizing is much less than TORAYCA carbon fibers. Therefore, less polymerization and decomposition will occur, leading to more flexibility while heat-treating CFRRPC composites. Compressive strength results are given Fig. 8. The highest compressive strength achieved was 502 MPa (72.8 ksi) in 7 days with MR50 carbon fibers. Specimens were subjected to a presetting load of 50 kN (11.2 kip) (26 MPa [308 ksi]) for 6 hours, then were heat treated at 150 ?C (302 ?F).

Physical and mechanical properties

Density-The density of the RPC was calculated for the different mixture samples, and different curing regimes (heat-treated and/or subjected to presetting load). The average density values (three samples/mixture) for different RPC mixtures are given in Table 2. The RPC produced had a density ranging from 1.76 to 2.41 Kg/m3 (0.11 to 0.15 lb/ft3). Some densities achieved were comparable to the density of a commercially available RPC with organic fibers, which are in the range of 2.2 to 2.4 Kg/m3 (0.14 to 0.15 lb/ft3). Density increases with increasing presetting load, but decreases with increasing heat-treating temperature (Table 2). Increasing the presetting pressure decreases the air voids and excess water, thus increasing compactness.21 On the other hand, increasing the heat-treating temperature decreases water further, increasing porosity and thus decreasing specimen density.

RPC porosity-The porosity, pore size, and pore distribution of concrete are important parameters with respect to durability issues in concrete. Ingress of chemicals and water are the cause of almost all durability problems. RPC has a very low porosity, especially when pressure is applied prior to setting, which results in the loss of all entrapped air and excess water pockets (Fig. 5(a) and (b)).

Fracture parameters of RPC-A set of different parameters was derived from the load-deflection curves obtained for the notched and unnotched specimens. These parameters included the peak load (P^sub c^) and its corresponding deflection (d^sub c^), and the elastic load (P^sub e^) and its corresponding deflection ((a^sub e^)). The area under each load-displacement curve was calculated using trapezoidal integration.22 The areas were calculated up to the peak load for notched specimens, for both notched (A^sub N^) and unnotched specimens (A^sub UN^). For unnotched specimens, the total area under the whole load deflection curve was also calculated (A^sub Total^). The parameters derived from the experimental results are summarized in Table 3. The parameters in Table 3 were used to calculate, several fracture parameters using a mathematical computational program23 (Table 4).

The fracture toughness tests showed that adding carbon fibers to RPC resulted in stiffer specimens with increased fracture toughness, tensile, and compressive strength (Fig. 9).

Freezing-and-thawing test results

Characteristics of test specimens-Test specimens (prisms) had a smooth surface, before being placed in the apparatus for the first time, with very few small voids due to lack of consolidation (vibration) or shrinkage.

Precracked specimens (PC4 to PC6 and CFC4 to PC6) contained random cracks. The cracks varied in size, width, and length through the specimen. Some specimens contained more cracks than others, with a couple of specimens appearing to be severely cracked (Specimens PC4 and PC6).

No pronounced change or surface degradation was noticed during the period of the test (300 cycles), however, except for severely cracked Specimen PC6, which split in half after approximately 280 cycles, (noticed when adding water to the apparatus).

Compressive strength-Six cylindrical samples (50 mm [2 in.] in diameter by 100 mm [4 in.] in length) were cast to test for compressive strength. Cast cylinders were cured at room temperature (23 ?C [73.4 ?F]) for 14 days before being tested. Both ends of the test samples were ground smooth prior to testing. The average compressive strength of CF specimens (288 ? 5.4 MPa [41.8 ? 0.8 ksi]) was higher than that of plain specimens (243 ? 3.6 MPa [35.2 ? 0.5 ksi]).

Average weight change-No significant weight loss was measured throughout the freezing-and-thawing tests or even beyond 300 cycles. Also, when specimens were removed from test tanks for measurements, no concrete particles were found in the tank. Specimen PC6 was weighed after it split (over 280 cycles). Very few concrete particles were detected in the test tank for this specimen.

Average length change and fundamental transverse frequency-No significant length changes were measured and there was little variation in the value of transverse frequency throughout the tests. Plain specimens appeared to have greater variability in the measurements compared to the carbon fiber ones. Variability in the results for all the specimens is related to the temperature of the specimens at the time of testing, the accuracy of the testing apparatus, and the level of repeatability of the test process, especially the repeatability of locating the transducer positions. The locations of both transducers used for measuring the fundamental transverse frequency were guided by a stainless steel plate with the same diameter as the transducers.

CALCULATIONS OF FREEZING-AND-THAWING PARAMETERS

Relative dynamic modulus of elasticity

Numerical values for relative dynamic modulus of elasticity are calculated from

... (1)

where P^sub c^ equals the relative dynamic modulus of elasticity, after c cycles of freezing and thawing, percent; n equals the fundamental transverse frequency at 0 cycles of freezing and thawing; and n^sub 1^ equals the fundamental transverse frequency after c cycles of freezing and thawing.

Calculation of dynamic modulus of elasticity is based on the assumption that the weight and dimensions of the specimen remain constant throughout the test. The weight and dimensions of RPC specimens were more constant than for ordinary concrete. The results for RPC are therefore somewhat more realistic and accurate than for OPC concrete.24 Relative dynamic modulus of elasticity results for all RPC specimens were approximately [asymptotically =]100.

Durability factor

The durability factor (DF) is a measure of how durable the concrete is on a scale of 100, with 100 being the most durable. A durability factor of 100 indicates that no physical change occurred over the test period.

The DF was calculated using the following relationship

... (2)

where P equals the relative dynamic modulus of elasticity at N cycles, percent; N equals the number of cycles at which P reaches the specified minimum value for discontinuing the test or the specified number of cycles at which the exposure is to be terminated, whichever is less; and M equals the specified number of cycles at which the exposure is to be terminated.

The durability factors for RPC specimens at the end of the test (300 cycles) were approximately [asymptotically =]100 for all specimens (101.6? for RPC with carbon fiber, and 101.0 for plain RPC, excluding Specimen PC6).

Percent length change

The percent in length change was calculated using the following relationship

... (3)

where L^sub c^ equals the length change of the test specimen after c cycles of freezing and thawing, percent; l^sub 1^ equals the length comparator reading at 0 cycles; l^sub 2^ equals the length comparator reading after c cycles; and L^sub g^ equals the effective gauge length between the innermost ends of the gauge studs.

The effective gauge length between the innermost ends of the gauge studs was 330.2 mm (13 in.), measured at the time of casting. Minimal length change took place. The CF specimens appeared to shrink slightly less than the P specimens.

DISCUSSION

The new developed optimum RPC mixture showed excellent fracture parameters, especially fiber reinforced samples that had the highest values of fracture toughness.

RPC, whether fibered or plain, showed excellent resistance to rapid freezing and thawing for more than 300 cycles (maximum 600 cycles). All specimens except PC6 had a DF of 100 or more after 300 cycles. Variations in durability factor were minimal and within acceptable limits for the ASTM C 666 standard. The performance of precracked specimens was of particular interest due to the expected adverse effect. One severely precracked specimen did end up in two pieces. The minimal changes observed in other specimens, however, reflected well on the durability of RPC.

This superior durability could be further improved if RPC was subjected to presetting pressure to eliminate the air trapped during mixing or due to lack of consolidation, and to increase density and compactness. Postsetting heat treatment would improve the microstructure.

The freezing-and-thawing test was terminated at 600 cycles and no failure was observed at that time. Specimens before and after 600 cycles are shown in Fig. 10(a) and (b). These results agree with those on RPC containing 12 mm (0.47 in.) steel fiber, where the durability factor was found to be more than 100% after 300 cycles following the ASTM C 666 standard.10,25

SUMMARY AND CONCLUSIONS

A carbon fiber-reinforced reactive powder concrete composite and casting procedure have been developed that allow the manufacture of a compact nonmetallic anchorage system for post-tensioning of concrete with CFRP tendons. A compressive strength as high as 500 MPa (72.5 ksi) was reached in 7 days, for CFRRPC. The mixture was optimized assuming independence of the variables in order to achieve the best mechanical and physical properties. Better results might have been achieved with the use of metallic fibers. Use of metallic fibers was avoided, although corrosion of steel fibers would be very slow in the high density mixtures achieved, which would have very low permeability and porosity values. Therefore, RPC mixtures using steel or other metallic fibers could potentially be optimized. However, steel fibers exposed in a crack might corrode.

The CFRRPC is also durable against freezing and thawing, even when pre-cracked, permitting use in cold environments. The fracture toughness tests showed that adding carbon fibers to RPC resulted in stiffer specimens with increased fracture toughness, tensile, and compressive strength. The RPC mixture developed can be used to create a nonmetallic anchorage for CFRP tendons much smaller than that described in the literature.1,2

NOTATION

a^sub c^ = critical effective crack length

CTOD^sub c^ = critical crack tip opening displacement

E = modulus of elasticity

E^sub c^ = plain strain modulus of elasticity of imaginary beam including effective elastic crack

G^sub F^ = fracture energy

G^sub Ic^ = critical energy release rate for mode I cracks

J^sub Ic^ = critical J-Integral

K^sub Ic^ = critical stress intensity factor for mode I cracks