STEEL FIBER REINFORCED CONCRETE FOR THE BARCELONA METRO LINE 9 TUNNEL LINING

Transcription

1 STEEL FIBER REINFORCED CONCRETE FOR THE BARCELONA METRO LINE 9 TUNNEL LINING R. Gettu 1, B. Barragán 1, T. García 1, G. Ramos 1, C. Fernández 2, R. Oliver 2 1 Universitat Politècnica de Catalunya, 2 UTE Línia 9. Barcelona, Spain Abstract This work presents the experimental studies carried out for evaluating the possible application of steel fiber reinforced concrete, without any conventional reinforcement, in the precast/prefabricated tunnel lining segments of the new Metro Line 9 under construction in Barcelona. The TBM-excavated tunnel has a diameter of about 12 m and a length of more than 40 km, and is located 30 to 70 m below the surface. According to the original design project, the reinforced concrete segments are of steel fiber reinforced concrete, with a high quantity of conventional reinforcement. However, the mechanical contribution of the fibers is not taken into account and the fibers are incorporated only to control the cracking that could occur during handling and placing of the segments. Here, an alternative solution was evaluated without any conventional reinforcement through the characterization of the material properties, from which a specific type of steel fiber was chosen from five different possibilities, and structural scale tests to study the behavior under critical loading conditions: flexure, concentrated in-plane loads, and possible stress concentrations in the joint between segments of the same ring. It was concluded that segments reinforced with 60 kg/m 3 of appropriate steel fibers could satisfy the requirements of the project, without necessitating any conventional reinforcement. 1. Introduction A new Metro line (Line 9) is currently being built in the Mediterranean city of Barcelona in Spain. The project is record-breaking, with a tunnel of almost 12 m diameter and 44 km length that will accommodate the tracks and platforms in a double deck configuration. The tunnel is being excavated, at a depth of 30 to 70 m below the surface, with an EPB-type Tunnel Boring Machine (TBM). As the excavation progresses, a robot arm of the TBM places the prefabricated segments of the tunnel lining to complete it ring by ring. As the excavation progresses, the TBM applies a reaction force through

2 hydraulic jacks as the shield cuts into the rock and/or soil ahead of it. This reaction, which can reach values of up to 140 MN, acts on the lining ring placed last. The construction of two sections of the tunnel for the Barcelona Metro Line 9 began in May 2003 and is being undertaken by two consortia of major Spanish companies, one of them being UTE Línia 9, which is first constructing a 4.5 km long section of the tunnel called Bon Pastor - Can Zam. In this section, the excavated diameter of the tunnel is m and the 35 cm thick lining has an internal diameter of m. The space between the lining and the excavated surface is grouted in order to provide a uniform load transfer and waterproofing. The lining is made up of 1.8 m wide rings, each consisting of 8 segments (7 segments of about 4.70 m, and 1 key segment of half that length), as seen in Figure Fig. 1. Segments of tunnel lining The concrete segments were originally designed with steel fiber reinforced concrete (with 30 kg/m 3 of fibers) and a significant amount of conventional reinforcement. However, the mechanical contribution of the fibers was not taken into account in the 142

3 original design, and the fibers were incorporated basically to control possible cracking induced by: Flexural stresses during demolding and stacking. Unintentional impact and high concentrated loads applied during the placing of the segments and the application of the reaction forces by TBM actuators. In this context, the constructor considered that it would be reasonable to take into account the mechanical contribution provided by steel fibers in the structural design. Moreover, in addition to the innovative application, the use of fibers would also reduce labor and construction time, thus increasing productivity. To evaluate these aspects and the possibility of totally replacing the conventional reinforcement by steel fibers, a collaborative project was undertaken by the Structural Technology Laboratory of the Universitat Politècnica de Catalunya and UTE Línia 9, where numerical and experimental studies, and quality control and other technological aspects were considered. The present paper discusses the experimental studies performed within the project at the material and structural levels. The results have been used to justify the construction of an experimental section of the tunnel with steel fiber reinforced concrete (SFRC) segments, eliminating the conventional bar reinforcement completely. 2. Material Testing Program The objective of the tests at the material level was to establish criteria and procedures for the selection of fiber type, mix design of the concrete and quality control during routine production. The Belgian NBN standard was used as the basis of the toughness evaluation, following which prisms of mm were tested under 4-point bending. A base concrete was chosen with a characteristic 28-day cylinder strength of 50 MPa and an early-age strength of 20 MPa at 6 hours (after heat curing at 40ºC). Fiber concretes with this base composition and 5 types of steel fibers were first evaluated using a dosage of 45 kg/m 3. From these tests a specific fiber was chosen for further evaluation. All the fibers that were tested had lengths of 45 to 60 mm and diameters of mm. In the first phase of the study, the specimens were cured during 6 hours at 40ºC and 80% R.H. (to reproduce the fabrication process of the segments) and were afterwards maintained during 7 days at 20ºC and 50ºC Toughness evaluation with different types of fibers Figure 2 shows the average load-deflection responses of specimens made with concrete incorporating the different types of fiber. In all cases, there is a significant residual strength with notable variations in the shape of the curve indicating the different effectiveness of the fibers. 143

4 From the load-deflection response, the equivalent flexural strength (f f,n ) can be calculated by normalizing the area under the curve up to a certain deflection limit by the deflection range, and then using the value obtained as an average load. Here, the equivalent flexural strengths at the deflection limits of 1.5 and 3.0 mm have been determined, and referred to as f f,300 and f f,150, respectively. The mean values of f f,300 and f f,150 for the concretes reinforced with 45 kg/m 3 of the different types of steel fibers are presented in Table 1, together with the mean values of the modulus of rupture (f r ) at the maximum load. Table 1 also shows the results of control tests carried out to obtain the compressive strength (f c ), the density of fresh concrete ( f ), and the modulus of elasticity in compression (E). All values correspond to the average of 3 measurements. The data corresponding to 6 hours were obtained using heat-cured specimens and those corresponding to 28 days were obtained with standard fog room curing. 50 Load (kn) Fiber A Fiber B Fiber C Fiber D Fiber E Deflection (mm) Fig. 2. Load-deflection curves for concrete reinforced with 45 kg/m 3 of different types of steel fibers 144

5 Table 1. Properties of the different concretes (mean values and coefficients of variation in %) Fiber f (kg/m 3 ) A 2410 B 2440 C 2440 D 2450 E 2410 f c 6 hours 23.7 (±2%) 25.4 (±3%) 23.5 (±1%) 24.5 (±1%) 22.3 (±1%) E 6 hours (GPa) 21.2 (2%) (1%) - f c 28 days 61.5 (±1%) 63.4 (±1%) 61.5 (±2%) 62.0 (±1%) 57.2 (±2%) E 28 days (GPa) (±2%) - f r 5.01 (±10%) 5.19 (±8%) 5.27 (±10%) 4.91 (±8%) 4.47 (±4%) f f, (±12%) 3.85 (±23%) 3.78 (±22%) 3.61 (±6%) 3.18 (±26%) f f, (±13%) 3.78 (±28%) 3.68 (±31%) 3.20 (±8%) 2.55 (±24%) From these results, and taking into account technical and economical aspects, fiber C was chosen for the subsequent material and structural-scale tests Toughness evaluation of concretes with different dosages of fiber C In the second phase of testing all the specimens were cured at 20ºC and 98% R.H. during 7 or 28 days (to obtain values that conformed to standard practice). Tests were performed on concretes with dosages of 30, 45 and 60 kg/m 3 of fiber C. Figure 3 shows the load-deflection responses obtained at the age of 28 days. The toughness parameters and other properties are given in Table 2 for the two ages. All values correspond to the average of 3 measurements. As it can be observed from the load-deflection curves in Figure 3, where the error bars indicating variability are also plotted, the residual strengths increase significantly and the variability tends to decrease with fiber content, especially at 28 days. 145

6 60 Fiber C, 60 kg/m 3 Load (kn) Fiber C, 45 kg/m 3 Fiber C, 30 kg/m Deflection (mm) Fig. 3. Load-deflection curves for concretes reinforced with different dosages of fiber C The toughness values and the corresponding variability were used as references in the material specifications. Table 2. Properties of concretes with different dosages of fiber C (mean values and coefficients of variation in %) Age 7 days 28 days Fiber dosage (kg/m 3 ) f c f r f f,300 f f, (±0.6%) 5.79 (±4%) 3.39 (±14%) 3.37 (±17%) (±3.1%) 5.47 (±3%) 4.38 (±9%) 4.30 (±11%) (±1.6%) 6.07 (±11%) 4.47 (±20%) 4.08 (±22%) (±1.6%) 6.59 (±9%) 3.72 (±23%) 3.55 (±26%) (±1.1%) 7.08 (±5%) 4.78 (±9%) 4.73 (±4%) (±0.9%) 7.18 (±1%) 6.09 (±9%) 5.98 (±8%) 146

7 3. Structural-scale test program Four types of real-scale tests have been performed on segments and specimens fabricated in the UTE Línia 9 plant. The fiber dosage was limited to 60 kg/m 3 due to economical and practical aspects (in terms of mixing, homogenization and placing of the SFRC) Flexure testing of the segments The objective of the flexural tests was to evaluate the behavior of segments with different reinforcement (with both conventional steel bars and fibers) in order to compare the responses of the original design and the segments with only fibers. The case of flexure was studied since it is the most critical case after the segment has been placed. The segment is expected to be only under circumferential compression and not under flexure. However, inadequate filling of the space between the lining and the excavated surface could lead to local flexure, as shown in Figure 4. Inadequate filling Soil/Rock Tunnel Segmental ring Filling grout Fig. 4. Motivation for the flexure tests The general flexural test configuration is shown in Figure 5. The segment is placed on two continuous and parallel metal supports with neoprene pads, of 200 mm width and 40 mm thickness. Flexure is produced with a mid-span line load and a free span of 1.80 m. Due to practical limitations in terms of transportation and loading equipment, tests were performed on the smaller key segment (Fig. 1). A Temposonics transducer was placed over the middle of each side of the segment, with a base-length of 720 mm (approximately two times the height of the element). They measured the elongation of the bottom edge of the segment, which can be considered as the sum of the crack widths within the base-length, once crack initiation occurs. The midspan deflection was measured with an LVDT placed at the shorter side of the segment. Also, a Temposonics transducer was placed on the opposite side of the segment to measure the oblique deformation near the supports, with a base-length of 300 mm and inclined 45 with respect to the horizontal plane. 147

8 Two tests were performed for each of the 4 reinforcement cases, which were 45 and 60 kg/m 3 of steel fibers, conventional reinforcement combined with 30 kg/m 3 of fibers, and only conventional reinforcement. Failure occurred due to the development of flexural cracking in the central part of the segment, as expected. Multiple cracking was observed in the segments reinforced with steel bars, as can be seen from Figure 6a. On the other hand, single localized cracks occurred in the segments reinforced only with fibers, as in Figure 6b. Fig. 5. Configuration of the flexure test a) b) Fig. 6. Cracking in the segments: (a) with conventional reinforcement combined with 30 kg/m 3 of fibers and (b) with only 60 kg/m 3 of fibers 148

9 In every case the first crack that was initiated, developed as the principal crack. In the segments reinforced with steel bars, there was secondary cracking in the middle of the span. Segments reinforced only with fibers did not exhibit any capacity to redistribute the stresses and had localized cracking. Figs. 7 and 8 show the load-crack opening and load-deflection responses of the eight segments. In both figures, the plot on the left gives the complete response obtained during the test (pre- and post-peak), and the plot on the right shows the response obtained up to a crack opening of 1.2 mm (in Fig. 7) or a deflection of 2.0 mm (Fig. 8), which are considered to be the limits of practical interest for the segments. It can be seen in the figures that segments 43 and 44, which are reinforced only with steel bars, exhibit a lower load carrying capacity and limited plastic behavior when compared with segments 41 and 42, which are reinforced with rebars and 30 kg/m 3 of fibers. The shapes of the curves are, however, similar. On the contrary, the behavior of the segments reinforced only with steel fibers is qualitatively different. Though a nonlinear zone exists before reaching the peak load, which reflects the material toughness, the plastic regime is small, and the load carrying capacity decreases progressively for crack openings beyond 1 mm, approximately. Segments with 60 kg/m 3 of fibers (i.e., segments 47 and 48) exhibited higher load carrying capacity than those reinforced with 45 kg/m 3 (i.e., segments 45 and 46), as expected. However, the loads are significantly lower than the corresponding values of the segments reinforced with both conventional bars and 30 kg/m 3 of fibers. Nevertheless, considering the initial response (i.e., plots on the right), which corresponds to service conditions, it can be seen that the behavior of the segments reinforced with only 60 kg/m 3 of fibers is close to that of the segments reinforced only with conventional steel bars, up to a crack opening of 1 mm, approximately. When an even smaller range is considered, i.e., up to 0.2 mm of crack opening, the behavior of all the segments is similar segment 41 (bars+fiber30) segment 42 (bars+fiber30) segment 43 (bars) segment 44 (bars) Load (kn) Load (kn) Crack opening (mm) * segment 45 (fiber45) segment 46 (fiber45) segment 47 (fiber60) segment 48 (fiber60) Crack opening (mm) Fig. 7. Load-crack opening response of all the segments: Complete (left) and initial (right) behavior 149

10 segment 41 (bars+fiber30) segment 42 (bars+fiber30) segment 43 (bars) segment 44 (bars) Load (kn) Load (kn) * segment 45 (fiber45) segment 46 (fiber45) segment 47 (fiber60) segment 48 (fiber60) Deflection (mm) Deflection (mm) Fig. 8. Load-deflection response of all the segments: Complete (left) and initial (right) behavior 3.2. Stacking of the elements After demolding, the segments of each ring are piled one on top of the other, with two blocks of wood between them. Usually, only 3 segments are piled after the first day and all eight segments of the ring are piled after 7 days. However, eccentricities between the wood blocks and the accidental piling of more than 3 segments at an early age could result in cracking. To evaluate the risks involved, piling tests were carried out with the objective determining safe eccentricity and over-piling limits. During piling, the most unfavorable case is that of the second segment from the bottom (since the bottom-most segment is placed on wider blocks fixed to a base) that is loaded eccentrically due to badly placed blocks below and/or above it, as seen schematically in Figure 9. The combination of exterior eccentricity at the bottom (e e ) and interior eccentricity at the top (e i ) could lead to high flexural stresses in the segment. In the present tests, the individual eccentricities were always equal to each other, and the total eccentricity is the sum of the four eccentricities. The instrumentation used in these tests consisted of Temposonics transducers placed on either side of the second segment, as in the flexural tests (see Fig. 5). 150

11 Segment 4 Segment 3 Transducer 2 Transducer 1 Segment 2 Segment 1 Fig. 9. Piling test configuration Three segments were tested: one segment reinforced according to the original design (with conventional bars and 30 kg/m 3 of fibers), and two segments reinforced only with 60 kg/m 3 of fibers. To study the behavior of the segments under the worst possible situations, exaggerated eccentricities were applied. The segment with conventional reinforcement and fibers, and one segment with only fibers were tested with e i = e e = 50 cm, which corresponds to total eccentricity of 2 m. Another fiber reinforced concrete segment was tested with e i = e e = 25 cm, which corresponds to total eccentricity of 1 m. The maximum total eccentricity expected in reality was about 40 cm. All tests were conducted at the age of 4 days. In the first test, corresponding to the segment 559 reinforced with 60 kg/m 3 of fibers, the placing of the third segment did not produce any significant displacements. This corresponds to the typical situation at the age of 3-4 days, when 3 segments are piled. Continuing with the piling, a progressive increase in the displacements was observed. When the sixth segment was placed, abrupt failure of the second segment occurred with the development of a single crack beneath one of the lumber pieces. Visual inspection of the surface did not reveal any other crack. Taking into account that at early ages only 3 segments are piled, the result of the test can be considered satisfactory since failure did not occur until the piling of the sixth segment. Table 3 shows the individual measurements and the mean values of the crack opening measured during the piling test. In the second test, the same loading procedure was applied to segment 568, which had conventional reinforcement with 30 kg/m 3 of fibers. As it can be seen from Table 3, the crack openings in this segment are slightly lower than those observed in segment 559 until the placing of the fifth segment. After the sixth segment is placed the crack openings become significant, indicating the propagation of cracks of considerable widths, though complete failure did not occur. Visual inspection of the surface using a magnifying lens revealed 7 cracks, with widths between 0.1 and 0.2 mm, distributed along the bottom face, within and outside of the central zone. 151

12 Table 3. Crack openings for the stacking test e Crack opening (mm) Segment e = e i Piled Load step (cm) segments short side long side mean (60 kg/m 3 of 50 fibers) Failure Failure Failure (bars kg/m 3 of fibers) (60 kg/m 3 of fibers) The third test with individual eccentricities of 25 cm was performed on segment 561, which was reinforced only with fibers. In this test, the complete ring was piled and no failure occurred. The measured crack openings can be observed from Table 3. When compared with the results of the first test (segment 559), the crack openings are much smaller. Moreover, no significant crack opening is observed and this was confirmed later by visual inspection of the surface. Also, for a total eccentricity of 1.0 m the crack opening of the segment reinforced only with fibers is well below 0.05 mm Simulation of the effect of the TBM actuator reaction As mentioned earlier, the TBM actuators exert a high compression force on the ring as the excavation progresses. The maximum possible load is 140 MN, applied by 30 actuators on 15 equally-spaced plates of mm (see Fig. 10a). Due to the impossibility of performing a test with the actual elements, a half-scale panel test was carried out to evaluate the effect of a single jack. The panels were 900 mm high (i.e., half of the depth of the segments), 520 mm wide (i.e., one-eighth of the width of the segments or half of the influence width of one actuator) and 175 mm thick (i.e., half the thickness of the segment). Tests were conducted on panels with conventional reinforcement and 30 kg/m 3 of fibers and on panels with only 60 kg/m 3 of fibers. They were fabricated with the same material as the segments, and those with conventional bars had half the reinforcement to maintain the same density. All tests were made at the age of 21 days. 152

13 The general test configuration is shown in Figure 10. Two loading cases were considered: centered-load uniformly-distributed along the entire width of the panel with a 120 mm width platten and load applied with an eccentricity of 28 mm. Four Temposonics transducers were placed around the perimeter of the mid-plane of the panel, as seen in Figure 10b, to measure the opening of longitudinal cracks. The base lengths were 420 mm along the width and 120 mm along the thickness. The tests were carried out, under load control, in a servo-hydraulic testing system with 4.5 MN capacity. Load was applied in two steps: first up to the maximum stress (i.e., load of 1.25 MN) and then up to an overload of 50% (i.e., a total load of 1.83 MN). None of the panels failed, though cracks appeared at the contact zone between the panel and the loading platten. In the case of panels subjected to central loading, the crack widths do not surpass 0.02 and 0.03 mm for the 1.25 and 1.83 MN loads, respectively. The panel reinforced only with steel fibers had higher transversal deformations than the panel with conventional reinforcement and fibers, as expected. In the case of panels subjected to eccentric loading, the deformations were higher. Nevertheless, the crack widths were less than and 0.05 mm for the 1.25 and 1.83 MN loads, respectively. From the results obtained in these tests, it was concluded that the segments would resist the concentrated loads transmitted by the TBM actuators without failing, and that the behavior of the segment reinforced only with 60 kg/m 3 of steel fibers would be similar to that of the segments with both conventional reinforcement and fibers. 153

14 (a) (b) Loading plate (eccentric) 540 x 120 x 10 mm Temposonic molded face Temposonic 1 lateral Temposonic 2 lateral Temposonic finishing face Continuous loading plate 540 x 180 x 10 mm Fig. 10. In plane compression due to (a) the TBM actuators simulated through (b) panel tests 154

15 3.4. Segment-to-segment compression test One of the critical aspects to be considered in the design of the concrete lining is the effect of the radial pressure of the soil/rock. In the present case, a maximum pressure of 1 MPa was considered in the design project. Though this does not lead to excessive compressive forces, non-uniform contact between the adjacent segments of a ring or adjacent rings could result in local crushing or cracking. Tests were, therefore, conducted to observe the extent of damage induced by such local effects. Blocks of mm were cut from the end of segments fabricated according to the usual procedures, and placed against each other to simulate the contact between two adjacent segments. The blocks were extracted from segments reinforced with conventional reinforcement and 30 kg/m 3 of fibers, and from segments reinforced with just 60 kg/m 3 of fibers. A compressive load of 4.3 MN was applied, which corresponds to a circumferential compressive stress of 20.3 MPa. The blocks were placed and loaded as seen in Figure 11. Possible misalignment of the segments or rings was taken into account by introducing an eccentricity of approximately 25 mm along both of the axes. Testing was performed at an age of more than 28 days. A servo-hydraulic testing system with 4.5 MN capacity was used and the loading applied under piston displacement control. After the application of the maximum load, the blocks were separated and examined to identify the cracks. No crushing and/or spalling of concrete occurred. In all cases, the non-uniformity of the loading led to crack initiation at the contact area between the loading platens and the blocks. Less cracks were seen in the blocks reinforced with both bars and fibers, which can be attributed to the higher load carrying capacity of the conventionally reinforcement segments. Fig. 11. Segment-to-segment compression 155

16 4. Conclusions This work presents an overview of the experimental studies carried out in the framework of a joint project between UPC and the construction consortium UTE Línia 9 with the objective of verifying the adequacy of steel fibers as only reinforcement for the segments of the tunnel lining of the new line 9 of the Metro of Barcelona, and consequently proposing the elimination of conventional reinforcement in some sections of the tunnel. The material characterization in compression and flexure led to a preliminary choice of the mix design and the methodology for the determination of toughness. For the selection of fiber type, toughness tests were performed with a dosage of 45 kg/m 3. Based on the analysis of the results and economical considerations, one fiber was chosen for further testing to evaluate the effect of fiber dosage on the toughness and variability of test results. Regarding the structural-scale tests, tests were performed in flexure and in-plane compression. It was seen that the segments reinforced only with 60 kg/m 3 of fibers have a similar behavior, as segments with rebars and 30 kg/m 3 of fibers up to crack widths of 0.2 mm. However, for higher crack widths beyond 1.0 mm, the plastic deformation and load-carrying capacity of the segments with rebars are much higher than segments with fibers as the only reinforcement. In the stacking tests, at the age of 4 days, it was verified that the segment reinforced with 60 kg/m 3 of fibers has sufficient capacity to resist cracking even when there is an accident overload due to eccentric piling. In the compression tests of panels, which were performed to simulate the concentrated loads produced by the actuators of the TBM, and in the compression tests on the segmentsegment interface, it was observed that the segments reinforced with 60 kg/m 3 of fibers have sufficient strength to avoid failure. However, it has been observed that the elements reinforced with both conventional bars and fibers exhibit a lesser cracking than those reinforced only with fibers due to a higher load-carrying capacity. Since fibers do not become active until cracking initiates, the presence of cracks is inevitable. From the results of this study, it was concluded that the use of segments with 60 kg/m 3 of appropriate fibers (and without any conventional reinforcement), is viable in terms of the requirements of the present project. Acknowledgements UTE Línia 9 is thanked for the support during the project and the permission to publish the results of the collaborative work. Prof. Antonio Aguado initiated the cooperation with UTE Línia 9, and his advice and guidance throughout the work is gratefully appreciated. The experimental work could not have been completely on schedule without the able assistance of the laboratory technicians, Carlos Hurtado and Camilo Bernad. 156