Development of Laboratory Device to Simulate Roller-Compacted Concrete Placement

INTRODUCTION

Roller-compacted concrete (RCC) was a significant evolution in the construction of dams. The RCC construction method often requires a dry concrete mixture and several horizontal contraction joints. This unique methodology has raised doubts about the accuracy of conventional laboratory tests to characterize the concrete's workability parameters and the strength in the field. These facts, together with the inherent variability of the concrete materials at each site, demand the use of conservative concrete mixtures at the beginning of construction. Only as the construction proceeds and field results become available can the concrete mixture proportions be improved. This solution adds to the overall cost of the project as it does not permit an optimization of the concrete mixtures during the early stages of the project.

Concrete has been used as a construction material for dams since the last decades of the nineteenth century.1 The structural models, construction methods, and quality control are well established and have been successfully implemented, as demonstrated by several old dams still in operation. The development of RCC in the 1980s had a major impact in the construction of concrete dams due to the substantial improvement in the speed of placement and compaction.2-4 Due to very dry consistency of RCC, most of the traditional tests methods used to characterize concrete are not applicable. Several tests, such as using drilled cores or constructing experimental layers of RCC in the field, have been proposed; however, few laboratory tests were developed and had their validity verified in real field conditions.5 The concrete expert must have tools and methodologies (that is, accurate laboratory tests) that allow the proper selection of materials to obtain an economical mixture proportion that satisfies the design criteria of strength, durability, and constructibility. Often, for economical reasons, local materials are used that may not have an ideal shape or granulometric distribution.3 To overcome this problem, preliminary laboratory tests are performed, such as strength and density, to optimize the granulometry of the aggregate. These tests provide relevant information for the design of construction facilities at the site, and even for the design of dam. After the laboratory tests are performed, final adjustments are made by using materials obtained from the crushing plant at the site and analyzing the results obtained at fills near the site using actual equipment and construction labor.

Another important aspect of the RCC technology is the continued need to increase the understanding of the material behavior. Up to now, tests performed during construction have been relevant to the improvement of the concrete mixture, but other problems arise during the normal construction process because of the short time between placing concrete layers.3 In addition, field results can only be used after a large amount of concrete is placed, delaying the optimization of the mixture proportions. This paper presents a method that performs test fills in laboratory conditions that simulate the site conditions to study any influencing factors (such as cure, temperature, treatment of joints in RCC dams designs, and construction techniques). To calibrate and to verify this methodology, compressive strength results and density from the proposed laboratory test were compared with the results obtained from field tests conducted at Dona Francisca Power Plant Dam, Brazil.

RESEARCH SIGNIFICANCE

Present laboratory tests do not have the capacity to predict the behavior of RCC in field conditions. The RCC technology has been employed in many recent projects, yet there is no appropriate methodology to simulate the real placement and compaction processes to optimize the mixture proportions.

To address this issue, a large-scale laboratory device was developed to better simulate field conditions. This study first presents the laboratory results, which are then verified through field work.

DESCRIPTION OF NEW EQUIPMENT

The equipment was designed to simulate (in the laboratory) actual site conditions for RCC, including the commercial compaction equipment and cure processes. The apparatus is composed of a rail system on which the roller-compacting structure moves. In an area at the center of the rails there is a pit onto which a mold is fixed. The mold remains entirely below floor level. The apparatus (Fig. 1) basically comprises three systems: one for horizontal movement, another for vertical movement, and a third for load application, all of which are described in the following.

The test specimens were compacted inside a mold, and at specified ages removed from the pit by means of a traveler. The testing of molds of various sizes is also possible, up to 3.0 m (9.84 ft) in length, 1.2 m (3.94 ft) in height, with a constant width of 0.90 m (2.95 ft). The ability to test forked molds was also considered, with the objective of simulating joints.

Given that the width of the roller is 650 mm (2.13 ft), that is, smaller than the width of the molds, this implies that only the central region equivalent to the roller's width can be considered useable or roller-compacted; therefore, the edges are compacted using a manual compactor (Fig. 2). Likewise, the roller cannot compact the initial and final sections that correspond to the roller's radius of 450 mm (1.48 ft). This section may be used for tests with surface concretes, thus simulating conventional surface concrete or upstream facing RCC (enriched with cementitious material or cement paste).

The horizontal movement of the roller is possible by means of the system of chains and sprockets, powered by an electric servo engine attached to a reducer. The chain system was designed so as to not only produce a back-and-forth movement, but also to allow for the roller to rotate in synch with this movement, thus preventing slippage between the roller and the concrete. This configuration best simulates the action of a commercial roller-compacter. The use of an electric engine allows for control of speed in both directions, between 0 and 0.5 km/h (0 and 0.31 mph), with continuous variation.

The vertical movement system aims to adjust the roller/ hydraulic actuator assemblage to the best height of the different layers. This movement is possible by using an electric servo engine with a speed reducer. The useable track of 1400 mm (4.59 ft) allows for the execution of test specimens of up to 1200 mm (3.94 ft) in height.

Load is applied using a high-performance hydraulic actuator with a 150 kN (33.7 kip) maximum capacity. By setting the servo valve directly over the actuator, the loads can be applied with a frequency up to 70 Hz. The hydraulic actuator is also partly responsible for the vertical movement of the roller, to follow the roughness on the concrete's surface along the length of the test specimen. For this reason, an actuator with a 200 mm (0.66 ft) track was used, allowing for the absorption of roughness on the concrete's surface of up to 200 mm (0.66 ft).

The control system is entirely digital and centralized in a single CPU, using a compatible PC, ensuring that the operator may choose the position control for positioning of the actuator or the load control for performing compaction. In load control, the system establishes a preset static and dynamic power condition, automatically making the necessary corrections for the maintenance of these conditions, that is, position variation and amplitude of the actuator resulting from greater or lesser hardness of the region being worked or the roughness of the surface.

The specimens are constructed as follows: the concrete is mixed in a 1.5 m^sup 3^ (1.96 yd^sup 3^) mixer, with a hydraulic operational system that tilts the concrete. Curing is performed in a custom-built wet chamber with a hinged lid, so that the test specimen may be placed in its entirety through the top by means of a traveler. Moisturizing is conducted by two moisturizers placed inside the chamber that also features a closed hydraulic water recycling system (Fig. 3).

The test specimens used in this test were taken from the molded block. Cross sections are exacted by means of a cutting device featuring a diamond-tipped blade. For aggregate processing and production, the laboratory contains an aggregate producing plant, comprising a jaw crusher, a vertical shaft impact (VSI) crusher, and a ball mill. The aggregates are classified by a sieving system in four stages, offering the possibility of four bands of aggregate sizes with coarse aggregate maximums of 100, 50, 25, and 4.8 mm (3.94, 1.97, 0.98, and 0.19 in.), or 76, 38, 19, and 4.8 mm (2.99, 1.50, 0.75, and 0.19 in.), depending on the characteristics specified in the project.

MATERIALS, MIXTURE PROPORTIONS, AND EXPERIMENTAL METHODS

To accurately simulate the site conditions at the Dona Francisca Hydro Power Plant, laboratory samples were made using the same materials and mixture proportions: crushed basalt as the aggregate, including artificial sand. Tables 1 and 2 show the aggregate's characterization and the concrete admixture. The artificial sand had an average of 14% filler in its weight. All mixtures used cement with pozzolan addition. The cement characterization data are shown in Table 3.

The laboratory tests were validated in the field conditions of Dona Francisca Power Plant, located in the state of Rio Grande do Sul, Brazil. The plant, shown in Fig. 4, has a 125 MW powerhouse and used 170,000 m^sup 3^ (222,350 yd^sup 3^) of concrete. The dam is 51 m (167 ft) high, 550 m (1804 ft) long, and incorporates a 300 m (984 ft) long ungated spillway defined for a maximum design flow of 7300 m^sup 3^/second (9548 yd^sup 3^/second) corresponding to a 10,000-year return period. The thickness of the concrete layers is 0.30 m (0.98 ft).

During several compaction cycles, the efficiency of the compaction was verified by measuring the automatic layer top settlement (refer to Fig. 5), and by the nuclear densimeter results. These results have been compared with the field results and to the maximum theoretical density. The RCC consistencies were obtained from a Canon test, using a transparent cylindrical container. The cylindrical molded samples were compacted using a large vibrating table. In the laboratory test fills, the sample size permitted cutting prismatic samples and also permitted obtaining cylindrical drilled samples. Figures 6 to 8 show the procedures executed on the test fill samples to prepare the samples for the characterization tests. The final compaction was verified using nuclear densimeter devices, which measured the density and total water content in three fixed points of the layer thickness. The compressive and tensile strength, the elasticity modulus, and Poisson's ratio were obtained from cylindrical molded samples, removing the coarse aggregate sizes larger than 25 mm (0.98 in.), and from drilled cores samples.

Presented as follows are results from 10 laboratory test fills. All the compaction samples had the same site mixture proportioning. To verify the process calibration, the results obtained in the laboratory test fills were compared with the quality-control site results. The density and concrete compressive strength were chosen as the comparison parameters. The main research objective was to compare the site quality-control results with laboratory simulations under similar conditions.

LABORATORY STUDIES AND FIELD VALIDATION

The laboratory test fills were performed on four separate layers (or sections) numbered from I to IV. The first layer was used as a benchmark; all the tests were performed on Layers II to IV. Tables 4 and 5 give the experimental results for drilled cores obtained in the laboratory compaction device. Table 6 presents the averages, coefficients of variation, and number of tests for density and 180-day compressive strength for molded and drilled core samples for laboratory test fills. At the Dona Francisca Dam Power Plant, concrete cylinders were cast and fill tests were made with 300 mm (0.98 ft) thick lifts each. The results of compressive strength and density tests of molded samples versus dam core drilled samples were compared. During the experiments, a full-scale laboratory test was performed to verify the effect of the layer position on the core drilling samples properties; the results show no statistical differences.

Table 7 shows the density results obtained for molded samples comparing laboratory test fills results with the site results.

The compressive strength of the molded concrete samples at 180 days was compared to the test field results (Table 8); similarly, the density results are compared in Table 9. Finally, Table 10 compares the compressive strength tests results for drilled core samples with the site results. The settlement curves, the density tests results, and the compressive strength tests results obtained for both the laboratory and site are very similar. Data analysis indicates a consistent accuracy of results, demonstrating the robustness of the laboratory setup to simulate site conditions.

The Dona Francisca Dam results obtained very low sample variance, showing a surprising homogeneity. During dam construction, several RCC mixtures were used (some of them using plasticizers), varying the cement content from 85 to 100 kg/m^sup 3^ (111 to 131 lb/yd^sup 3^). The test results for the mixtures cited previously used 90 kg/m^sup 3^ (118 lb/yd^sup 3^) and included plasticizer. This mixture was used in 13,742 m^sup 3^ (17,973 yd^sup 3^) RCC for the Dona Francisca Dam. During the construction, several core drills were executed to provide a compaction quality overview from the dam top to the foundations.

All the tests were performed using the same procedures in the laboratory and during construction of the Dona Francisca Dam, with crews with similar training, to avoid variability due these factors. To conduct an analysis of the modeling quality, assuming that the tests conditions were very similar, it was necessary to compare the average results for each studied parameter, its coefficient of variation, and the demonstrate that both the laboratory and site mixtures are equivalent, with coefficients of variation quite similar and coherent. In the case of core drilled samples, for the compressive strength, although the average results are very similar (7.8 MPa [1131 psi] for laboratory test fills and 7.9 MPa [1145 psi] for the site results), the site results' coefficient of variation is larger than the laboratory's, which is probably due to the small number of tests performed.

To statistically compare the site and laboratory results, two analyses of variance were made. The first one analyzed the sample age and the sample origin (site or laboratory) as a two-factor factorial experiment for the molded samples. Table 11 summarizes the analysis of variance. The F test shows that the significance level for the origin is close to 50%, indicating that there are no statistical differences in the site and laboratory results. The same procedure was developed for the core drilling samples for the compressive strength throughout a single factor analysis of variance. This study is summarized in Table 12, with the conclusion that are no statistical differences in the site and laboratory results. In summary, test results (the settlement curves, density tests results, and compressive strength tests) show no difference between those results obtained in laboratory versus those obtained at the construction site. Data analysis indicates a consistent accuracy of results, demonstrating the robustness of the laboratory setup to simulate site conditions, such as effect thickness of the layers, bedding mortar, and humidity.

CONCLUSIONS

The laboratory apparatus described previously will enable further research in RCC technology by successfully simulating, in a laboratory setting, compacting similar to commercial products in the market, thus reducing the number of necessary field adjustments.

The initial tests, carried out with densities measured by means of a nuclear densimeter, proved that it is possible to achieve an adequate compaction on the test site with a compaction degree varying between 97 and 100%, and the possible adjustment of this parameter to the desired range.

The studies may precede the field production of RCC, as they make its simulation in advance possible by means of aggregates and mixture proportions processed in the laboratory. Based on experimental results, it will now be possible to reduce the cement content in the earlier construction phases to minimize the compaction energy, to verify the benefit of admixtures, and to optimize the aggregate gradation.

This apparatus will facilitate any necessary adjustments that need to be made in laboratory RCC mixture proportions, as well as the characterization of its properties, in this way supplying the necessary data for dam or rigid pavement projects. Nonetheless, this does not eliminate the need for a test site (full-scale trial) to adjust field production apparatus and for consequent training of production staff.