Benefits of blast furnace slag and steel fibers on the static and fatigue performance of prestressed concrete sleepers

Engineering Structures 134 (2017) 317–333

Contents lists available at ScienceDirect

Engineering Structures journal homepage: www.elsevier.com/locate/engstruct

Benefits of blast furnace slag and steel fibers on the static and fatigue performance of prestressed concrete sleepers Jun-Mo Yang a, Hyun-Oh Shin b, Young-Soo Yoon c,⇑, Denis Mitchell d a

Steel Structure Research Group, POSCO, 100, Songdogwahak-ro, Yeonsu-gu, Incheon 21985, South Korea New Transportation Systems Research Center, Korea Railroad Research Institute, 176 Cheoldobangmulgwan-ro, Uiwang-si, Gyeonggi-do 16105, South Korea c School of Civil, Environmental and Architectural Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, South Korea d Department of Civil Engineering and Applied Mechanics, McGill University, 817 Sherbrooke Street West, Montreal, Quebec H3A0C3, Canada b

a r t i c l e

i n f o

Article history: Received 2 February 2016 Revised 20 December 2016 Accepted 23 December 2016

Keywords: Prestressed concrete sleeper Ground granulated blast furnace slag Steel fiber Static performance Fatigue performance

a b s t r a c t This study investigates the influence of the use of steel fibers and ground granulated blast furnace slag (GGBFS) on the static and fatigue performance of prestressed concrete sleepers. Two series of sleepers with a partial replacement of type III cement by GGBFS (56% by weight of the binder) were tested and the performances of these sleepers were compared to conventional railway sleepers. One series of sleepers were reinforced with the same number of prestressing strands (sixteen strands) as conventional railway sleepers, while the other series of sleepers had a reduced number of strands (fourteen strands). Each series consisted of two types of sleepers with (vf = 0.75%) and without steel fibers. Static and fatigue tests causing positive moments at the rail seat section and additional static tests causing negative bending moments at the center section of the sleepers were carried out. The sleepers produced with GGBFS showed improved static flexural and fatigue performance at the rail seat section compared to conventional sleepers with Type III cement. The addition of steel fibers instead of conventional stirrups resulted in increased flexural and fatigue capacity at the rail seat section by controlling crack propagation and by preventing brittle shear failure. The performances of all sleepers at the center section were quite similar. The combination of the reduced number of strands together with steel fibers improves the performance and achieves greater economy. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Prestressed concrete sleepers are widely used for application in railway construction [1–3]. However, increasing concerns about CO2 emissions have led to efforts to use alternative materials for the manufacture of prestressed concrete sleepers [4]. Furthermore, premature failures of prestressed concrete sleepers caused by various types of loading and concrete deterioration have been reported [5–8]. The use of ground granulated blast furnace slag (GGBFS) as a concrete binder and the addition of steel fibers in concrete mixes can be used to address these issues. Previous studies [9–11] have shown that the use of GGBFS results in higher long-term strength, improved durability, as well as reduced CO2 emissions and energy consumption. The benefits of steel fibers, including reduced and delayed spalling of concrete, control of initiation and growth of cracks, impact resistance and durability, have also been reported [12]. While the benefits of GGBFS and steel ⇑ Corresponding author. E-mail address: [email protected] (Y.-S. Yoon). http://dx.doi.org/10.1016/j.engstruct.2016.12.045 0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.

fibers may provide solutions to the problems encountered in current prestressed concrete sleepers, their performance needs to be evaluated in order to satisfy code requirements for their intended applications. The major role of railway sleepers is to transfer and distribute rail loads to the substructure [13,14]. The flexural capacity of sleepers at the rail seat section is a predominant issue because a single sleeper typically carries 45–65% of the wheel load directly above it [5]. Furthermore, cracking at the top center location of sleepers, which is called ‘‘center binding”, has been reported and this is known as one of the three primary failure mechanisms of prestressed concrete sleepers [5,6]. It is noted that this crack occurs when large negative moments at the sleeper center section, caused by ballast deteriorations, exceed the cracking moment of the sleepers. These facts demonstrate critical sections for positive and negative moments, corresponding to the rail seat and center sections, respectively. It should be noted that the magnitude and distribution of the moments vary depending on the wheel loading conditions, ballast and sub-grade conditions, as well as sleeper spacing [5,15–17].

318

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Nomenclature dp Ap db As fpi fpe l d l/d Mdr Mdcn Lr Lc

diameter of prestressing wire nominal area of three wire strand diameter of reinforcing bar nominal area of reinforcing bar initial stress in prestressing strand effective stress in prestressing strand after allowance for all prestress losses length of steel fiber diameter of steel fiber aspect-ratio of steel fiber positive design bending moment at rail seat negative design bending moment at center section clear span length of static and fatigue tests at the rail seat section clear span length of negative bending moment tests at the center section

Previous studies [18,19] have shown that the rail loads can be considered to be static and quasi-static under low to moderate train speeds, while a dynamic impact pulse is in general due to increased speeds of the continual moving ride over track irregularities. Previous studies [8,13] have also indicated that the failure of a railway sleeper is more likely due to the damage from cumulative impact conditions rather than due to a single impact event, which might occur due to derailment. Although the dynamic effects are evident in failures of the prestressed concrete sleepers, most of the design concepts are based on the static capacity of the sleepers. Numerical modeling of the sleepers also requires the properties obtained from static tests [18,19]. Therefore, the static performances at the rail seat and center sections of the sleepers were investigated in this study. Furthermore, fatigue tests at the rail seat section were performed. One of the objectives for the fatigue tests was the simulation of an exceptionally high load to create an initial crack, followed by fatigue loading, to study the effect of trains running continuously on cracked sleepers. 2. Experimental program 2.1. Test specimens and variables Eight prepressed concrete sleepers were produced for each of the five variables under the same manufacturing process as conventional railway sleepers, resulting in a total of 40 sleepers (see Fig. 1). Sleepers were designed by the allowable stress design method based on the Korean Railway Standard (KRS TR 0008) [20,21], that are classified as ‘Prestressed concrete sleepers for 60 kg K rail’. It is noted that KRS TR 0008 is equivalent to BS EN 13230-1 (general requirements) [22] and 13230-2 (prestressed

Fig. 1. Prototype prestressed concrete sleepers.

Fr0 Frr FrB Fru wFr0 wres Fc0n Fcrn FcBn TP,max TP,fail TD,20mm k2s

positive initial reference test load for the rail seat section positive test load produced the first crack formation at the rail seat section maximum positive test load at the rail seat section lower test load for the dynamic test of the rail seat section crack width at a load Fr0 after fatigue loading residual crack width after fatigue loading negative initial reference test load for the center section negative test load produced the first crack formation at the rail seat section maximum negative test load at the center section toughness up to the maximum load (FrB) toughness up to the load of the sleeper failure (FrB) toughness up to a deflection of 20 mm static coefficient to be used for the calculation of FrB

monoblock sleepers) [23]. However, the material requirements of KRS TR 0008, which are based on the Korean Standard (KS) for construction materials, are marginally different from BS EN Standard to accommodate availability of materials. All sleepers had the same geometry, a total length of 2400 mm, but were constructed with three different concrete mixes (CC, BS, and BSF) and reinforced with two different numbers of strands (sixteen and fourteen strands) as shown in Fig. 2. The specimen names indicate the mix type, including the type of binder and the steel fiber content, and the number of strands. For example, BS16 is the sleeper constructed with a mix BS without steel fibers and containing sixteen strands. While, BSF14 is the sleeper constructed with mix BS with 0.75% steel fibers and pretensioned with fourteen strands. It is noted that sleepers with steel fibers (BSF16 and BSF14) were constructed without any stirrups, whereas sleepers without steel fibers contained seven stirrups in each rail seat region which is a typical sleeper design currently used in Korea (see Fig. 2). Sleepers CC16 represent conventional prestressed concrete sleepers which are currently used in the Korean railway system. These sleepers are produced with Type III Portland cement (highearly strength) to acquire sufficient early-age strength for the release of the strands. These sleepers are reinforced with sixteen strands along the length and seven stirrups at each rail seat, resulting in a total of fourteen stirrups per specimen. Sleepers BS16 had the same geometry as well as reinforcement details as sleepers CC16, but eco-friendly concrete mix (56% replacement by GGBFS) was used to fabricate these sleepers. Sleepers BSF16 also represent sleepers with an eco-friendly concrete mix, and these sleepers contained 0.75% of steel fibers in the mix instead of stirrups. A comparison of the behavior of sleepers in series BS16 and BSF16 demonstrates the beneficial effects of steel fibers in terms of crack control as well as replacement for the stirrups. It is noted that the mix design for the concrete with GGBFS and the volume fraction of steel fibers were determined from numerous trial batches [2,24]. The flexural capacity of the prestressed concrete sleepers were calculated in accordance with the International Union of Railways (UIC) Code 713R [17], and the results indicated that the current railway sleeper design is conservative [24]. In order to provide an understanding of the beneficial effects of steel fibers and GGBFS for the performance of sleepers, a reduced number of prestressing strands were considered as a variable of the testing. As a result, sleepers BS14 and BSF14 were designed to have a reduced number of prestressing strands (fourteen of three-wire strands) compared to sleepers BS16 and BSF16. It is noted that removing the 2nd

319

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Fig. 2. Specimen details.

showed sufficient workability for casting of the concrete and easily achieved the strength requirements of 35 MPa at 16 h and 50 MPa at 28 days specified in 2011 KRS TR 0008 [20] and in the manufacturing guideline [25], respectively. It should be noted that the strength requirement at 16 h in the 2015 KRS TR 0008 [21] has been relaxed to 30 MPa from 35 MPa. Mixes BS and BSF exhibited a higher strength at early-ages compared to the mix CC, resulting in advantages in terms of earlier strands release and form stripping. These mixes exhibited improved durability performance and were shown to reduce CO2 emissions compared to mix CC [2]. Detailed mix design procedure and mechanical properties of the concretes for the sleepers are given in a previous study [2]. Three-wire strands, specified as ‘SWPD 3’ in KS D 7002 [26], were used for the prestressing steel for all specimens. Each wire

bottom prestressing strands (see Fig. 2) was the best approach in order to minimize the change in eccentricity of the prestressing. 2.2. Materials Three different types of concrete mixes were used to produce prestressed concrete sleepers: 1) mix CC for conventional railway sleepers (CC16); 2) mix BS for the sleepers BS16 and BS14; and 3) mix BSF for the sleepers BSF16 and BSF14. Table 1 summarizes the mix proportions and material properties of these concretes. It is noted that slump tests were carried out on random batches of concrete, while the compressive strengths and flexural strengths were determined from the tests on three cylinder specimens and two prismatic specimens, respectively, for every batch. All of the mixes

Table 1 Mix proportions and material properties of concretes. Mix

CC BS BSF

Mix design

Material properties

W/B [%]

S/a [%]

Water [kg/m3]

Cement [kg/m3]

GGBFS [kg/m3]

Sand [kg/m3]

Gravel [kg/m3]

AD [%]

vf [%]

Slump [mm]

f ci [MPa]

f c [MPa]

fr [MPa]

32.3 30.0 30.0

44.7 44.7 44.7

134 147 147

415 196 196

– 294 294

814 764 764

1006 943 943

0.9 1.0 1.2

– – 0.75

75 80 75

37.7 (0.60) 41.5 (0.46) 44.9 (1.12)

60.5 (2.86) 52.5 (1.06) 57.9 (0.87)

5.25 (0.74) 6.88 (0.30) 7.36 (0.56)

0

0

W/B: water-binder ratio, S/a: sand-to-aggregate ratio, Cement: Type III Portland cement  high early strength cement, GGBFS: ground granulated blast furnace slag, AD: i i polycarboxylic acid based water reducing admixture, vf: steel fiber volume fraction, f c : concrete compressive strength at 16 h, f c : concrete compressive strength at 28 days, fr: modulus of rupture at 28 days. Items in parentheses () = standard deviation.

320

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Table 2 Properties of reinforcing bar. Reinforcing bar (designation)

Type

Wire or bar diameter [mm]

Nominal area [mm2]

fy [MPa]

ey [e]

fpu or fult [MPa]

eult [e]

Prestressing strand (SWPD 3) Stirrup (SWM-B, 4.00)

3-wire strands Smooth bar

2.9/wire 4.0/bar

19.82 12.57

1705 594

– 0.003

1927 606

– 0.021

had a diameter of 2.9 mm, resulting in a nominal area of a threewire strand, Ap, of 19.82 mm2. D4 smooth reinforcing bars (db = 4.0 mm, As = 12.57 mm2), specified as ‘SWM-B, 4.00’ in KS D 3552 [27] which is equivalent to ISO 10544 [28], were used for the stirrups in the sleepers CC16, BS16, and BS14. The rectangular-shaped stirrups had 40 mm lap splices for anchorage in their top legs (see Fig. 2). The mechanical properties of the strand and stirrup reinforcement, specified by the manufacturer, are summarized in Table 2. The bundled-type hooked-end steel fibers with a length, l, of 30 mm and diameter, d, of 0.5 mm, resulting in an aspect-ratio, l/d, of 60, were used for the concrete in sleepers BSF16 and BSF14. The density, tensile strength, and elastic modulus of the steel fibers used in this test were 7.85 kg/m3, 1100 MPa, and 200 GPa, respectively. These mechanical properties of the steel fibers were specified by the manufacturer. 2.3. Manufacturing and inspection of the sleepers The prestressing strands were tensioned to an initial stress, fpi, of 1455 ± 19 MPa (0.75fpu) in the pretensioning bed prior to the casting of the concrete. Concrete was then cast in steel forms, followed by a 3 h of ambient curing, followed by 10 h of steam curing. The strands were released 16 h after concrete casting. The effective prestressing stress in these strands after allowing for all losses, fpe, was 873 MPa ( 0.6fpi). After the manufacturing process of the sleepers, visual and dimensional inspections were carried out by certified inspectors. All of the sleepers, including the sleepers with fourteen strands (BS14 and BSF14), satisfied the dimensional tolerances specified in KRS TR 0008 [20,21]. This result indicates that all of the sleepers produced in this study, including the sleepers with a decreased number of strands, are suitable for service in track systems without concerns about eccentricity and serviceability. In addition, the electrical resistance tests of the sleepers were also performed using 500V insulation-resistance multi-meter. All of the sleepers, even including sleepers with steel fibers, exhibited higher electrical resistance that required by KRS TR 0008 [20,21] (minimum 5 MX). 2.4. Test setup and procedure 2.4.1. Pull-out tests of rail fastening shoulder As shown in Fig. 3, rail fastening shoulder pull-out tests were carried out in accordance with KS TR 0014 [29] which is equivalent to BS EN 13481-2 [30] as follows: (1) the shoulder was loaded upwards using a hydraulic jack until 60 kN was attained; (2) the load was maintained for three minutes and the sleeper was inspected for cracks; and (3) the load was increased up to the fracture or pull-out of the shoulder (localized sleeper failure). The pullout loading was applied at a loading rate of 50 ± 10 kN/min. 2.4.2. Static bending tests at rail seat section The static bending performance at the rail seat section were evaluated according to the British Standard (BS EN 13230-2) [23] with three sleepers being tested for each variable. Fig. 4a shows the test setup and instrumentation. The prestressed concrete sleepers were supported on articulated supports and resilient pads with a center-to-center span length, Lr, of 500 mm. This span

Fig. 3. Pull-out test setup for the rail fastening shoulder.

length was determined in relation to Lp (design distance between the center line of the rail seat to the edge of the sleeper at the bottom, see Fig. 4a) in accordance with BS EN 13230-2 [23]. The load was applied through a tapered plate and rail pad in order to provide a uniform stress distribution at the top surface of the sleepers as well as to prevent local failure. The loading protocol for the testing was determined from the design moment (Mdr = 15.4 kN m) and geometry of the test setup (Lr = 500 mm) in accordance with BS EN 13230-2 [23] and UIC Code 713R [17]. Details of the calculations are given in a previous study [24]. From these calculations, the positive initial reference test load for the rail seat section, Fr0, was obtained as 154 kN. The loading protocol for the static bending test at the rail seat section was determined and applied as follows (see Fig. 4b): (step 1) increasing the load to Fr0 at a loading rate of 120 kN/min; (step 2) 10 kN of load increments were applied up to first crack formation, Frr; and (step 3) 20 kN of load increments were applied up to the maximum test load at the rail seat section, FrB. At steps 2 and 3, the crack propagation and crack widths were measured. It is noted that the first cracking load, Frr, was defined as the load to cause a crack length of more than 15 mm from the bottom surface of the sleepers. The first cracks of the sleepers were determined visually with the aid of a video camera focused on the region of expected cracking. A linear variable differential transducer (LVDT) was installed underneath the sleeper to measure the displacement at the rail seat section. The crack width of the sleepers was measured with a crack width microscope having 0.01 mm scale divisions.

2.4.3. Fatigue tests at rail seat section Fatigue testing at the rail seat section was carried out at each of the randomly selected sleeper for each test variable, in accordance with the British Standard (BS EN 13230-2) [23]. The test setup, instrumentation, and the loading protocol for the fatigue tests are summarized in Fig. 5. The same test setup as the static bending tests was used to investigate the fatigue performance of the sleepers. The loading was first increased up to the value that produces the first crack at the bottom of the rail seat section, Frr. A crack opening displacement transducer (COD gauge) was located on both sides of the sleeper near the bottom surface such that it crossed the initial crack to accurately measure changes of the crack width

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

321

Fig. 4. Static bending test at rail seat section: (a) test setup and instrumentation; (b) loading protocol.

during the fatigue loading cycles. After the first crack occurred, 2  105 cycles of repeated loading between Fr0 and Fru were applied at a frequency of 3 Hz (see Fig. 5b). It is noted that the same Fr0 value of 154 kN that was used in the static bending tests, and a lower dynamic test load for the rail seat section, Fru, of 50 kN (as specified in BS EN 13230-2 [23]) were used for the cycles of the fatigue loading. After the fatigue tests, the sleepers were loaded to Fr0 and then unloaded to 0 kN to measure the maximum crack width, wFr0, and the residual crack width, wres, respectively. Then, the loading was applied again at a loading rate of 120 kN/min up to their failure in order to investigate the post-fatigue performance of the sleepers. The same LVDT instrumentation as used for the static tests was used to measure the displacement at the rail seat section. 2.4.4. Static bending tests at center section In order to evaluate the static performance at the center section of the prestressed concrete sleepers, three negative bending moment tests (also known as the hogging moment test) were carried out in accordance with the British Standard (BS EN 13230-2) [23]. The test setup and the loading protocol are summarized in Fig. 6. The prestressed concrete sleepers were positioned upsidedown on the articulated supports and resilient pads with a center-to-center span length, Lc, of 1500 mm. From the calculations based on the design moment (Mdcn = 12.64 kNm) and geometry of the test setup (Lc = 1500 mm) [24], the negative initial reference test load for the center section, Fc0n, was obtained as 36.1 kN. A loading was applied up to Fc0n with a loading rate of 120 kN/min, followed by incremental loads of 5 kN up to the first cracking load at the center section, Fcrn, and maximum negative test load at the center section, FcBn, respectively. The crack propagation and crack widths were measured during the test using the

crack width microscope. An LVDT was installed underneath the sleeper, at the center of the support, in order to measure the midspan deflection of each sleeper. It should be noted that all of the test series addressed above are for benchmarking and do not represent the behavior of the sleepers in real track systems. However, in order to use different sleepers that are not allowed in the current standard, they should exhibit a minimum level of structural performance for qualification during the manufacture of the sleepers. For this purpose, structural performance of newly developed sleepers, produced with new materials and modified design concepts, such as decreased number of strands, was evaluated in accordance with international and domestic standards. It is highly recommended that field tests of the sleepers be carried out under actual train loading before they are installed in service.

3. Experimental results and discussion 3.1. Pull-out tests of rail fastening shoulder Table 3 summarizes the pull-out test results for the rail fastening shoulder, and Fig. 7 shows the appearance of the sleepers after the pull-out tests. No cracks were observed in all sleepers when the 60 kN pull-out load was applied and maintained for three minutes. Therefore, all sleepers produced in this test program acquired a sufficient pull-out resistance as that required by the KS TR 0014 [29]. As the pull-out load was increased, all of the sleepers exhibited concrete cracking, followed by tensile failure of the shoulder. The first crack in the concrete occurred at an average load of 82.4 kN, ranging from 74.0 to 89.8 kN. Tensile failure of the shoulder occurred at an average load of 89.7 kN, ranging from 85.0 to

322

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Fig. 5. Fatigue test at rail seat section: (a) test setup and instrumentation; (b) loading protocol.

91.6 kN. It is noted that the concrete cracking and tensile failure of the rail fastening shoulder occurred simultaneously in sleeper BSF14. The shoulder failure loads are dependent on the tensile capacity of the shoulder, therefore, the standard deviation of these loads was relatively small compared to the standard deviation of the first cracking loads. However, it is interesting to note that only sleeper BS16 exhibited a much smaller shoulder fracture load (approximately 6 kN less) compared to the other sleepers. Failure of the rail fastening shoulder in BS16 occurred near the top surface of the sleepers, however, the failure occurred at the mid-height of the shoulder in the other sleepers (see Fig. 7). Concrete cracking loads are directly related to the concrete strength, therefore, the cracking loads were normalized with respect to the concrete strength of each specimen (Pcr/fc0 ), as shown in Table 3. These ratios for the sleepers without steel fibers were very similar, ranging from 1.41 to 1.42, whereas the corresponding ratios for the sleepers with steel fibers were 1.50 and 1.58. This result demonstrates the beneficial effects of steel fibers on the crack control of concrete sleepers and the improved resistance to the pull-out of the shoulders.

3.2. Static bending tests at rail seat section 3.2.1. Flexural capacity Table 4 summarizes the experimental results of the static bending tests at the rail seat section, including the performance acceptance criteria for the response at loads Frr and FrB, specified in BS EN 13230-2 [23]. Table 4 also includes the toughness up to the maximum load, FrB, and the load defined as the failure of the sleepers where the applied load decreased to 80% of the capacity (TP,max

and TP,fail, respectively). The toughness, TP,max and TP,fail, can be obtained as the area under the load-deflection curve up to the maximum load and to the failure load, respectively. The key loads for determining the performance at the rail seta section are:

Frr > Fr 0 ¼ 154 kN

ð1Þ

FrB > k2s  Fr 0 ¼ 2:5  154 ¼ 385 kN

ð2Þ

where k2s is the static coefficient or impact factor applied to accidental load cases that can be taken as 2.5 in accordance with the UIC Code 713R [17]. The ratios of the test results to the specified acceptance criteria for Frr and FrB ranged from 1.52 to 1.74 and 1.25 to 1.52, respectively, which means that the test results are significantly higher than the performance acceptance criteria. This result indicates that all sleepers produced in this test program have sufficient flexural capacity at the rail seat section. 3.2.2. Observed behavior and failure mode Fig. 8 shows the applied load versus midspan deflection responses for all sleepers (three specimens for each variable) and Fig. 9 shows typical failure modes of the sleepers at the end of testing. The failure modes of each specimen are also summarized in Table 4. Three specimens in each variable showed almost the same load-deflection response until first cracking of the sleepers. However, the post-cracking response and the failure modes varied, even though all three specimens were designed to have the same capacity and they were produced with the same materials. This result was attributed to the complex stress distributions in the short spans that can result in different failure modes. All three specimens in the CC16 series exhibited first cracking at the same load (Frr = 234 kN). The stiffness decreased after cracking

323

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Fig. 6. Static bending test at center section: (a) test setup and instrumentation; (b) loading protocol.

Table 3 Shoulder pull-out test results.

a b

0

0

Sleeper

f c [MPa]

Crack inspection at a load of 60 kNa

Concrete cracking load, Pcr, [kN]

Pcr/f c [kN/MPa]

Shoulder fracture loadb [kN]

CC16 BS16 BS14 BSF16 BSF14 Average Standard Dev.

60.5 52.5 52.5 57.9 57.9 56.3 3.59

No No No No No – –

85.6 74.0 74.8 87.0 89.8 82.2 7.32

1.41 1.41 1.42 1.50 1.58 1.46 0.06

91.6 85.0 91.5 90.6 89.8 89.7 2.73

crack crack crack crack crack

A load 60 kN maintained for three minutes. Load at tensile failure of rail fastening shoulder.

and resulted in maximum loads, FrB, ranging from 465.2 to 504.0 kN. After the peak load, specimens 1 and 3 exhibited a sudden drop in load (see Fig. 8a) due to shear failure between the supports and the tapered loading plate at a deflection of about 6.0 mm (see Fig. 9a). Specimen 2 maintained a higher load compared to the other two specimens until a deflection of 7.1 mm, followed by a

drop in the load with the formation of a flexural-shear crack. It is noted that this series contained stirrups, indicating the vulnerability to shear failures, even in the presence of shear reinforcement. The first cracks (vertical flexural cracks) in the BS16 series initiated at the bottom of the sleepers at an average load of 244 kN and these cracks propagated towards the loading plate (diagonal

324

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Fig. 7. Examples of rail fastening shoulder failures: (a) BS16; (b) BS14.

Table 4 Results of static bending tests at rail seat section. Sleeper series

Test

First cracking

Maximum load a

Toughness a

Failure mode

Frr (kN) > 154 kN

dFrr (mm)

FrB (kN) > 385 kN

dFrB (mm)

TP,max (kN mm)

TP,fail (kN mm)

CC16

No.1 No.2 No.3 Avg.

234.0 234.0 234.0 234.0 (0.0)

1.53 1.70 1.57 1.60 (0.09)

465.2 469.0 504.0 479.4 (21.4)

4.61 4.71 4.72 4.68 (0.06)

– – – 1357.5

– – – 2223.5

Shear failure Flexural-shear failure Shear failure

BS16

No.1 No.2 No.3 Avg.

244.0 254.0 234.0 244.0 (10.0)

1.43 1.43 1.34 1.40 (0.05)

527.8 496.4 525.4 516.5 (17.5)

4.84 3.83 4.44 4.37 (0.51)

– – – 1382.5

– – – 3382.5

Flexural-shear failure Flexural-shear failure Flexural-shear failure

BS14

No.1 No.2 No.3

234.0 214.0 224.0

1.28 1.32 1.23

515.3 529.7 491.9

5.22 5.75 5.13

– – –

– – –

Flexural-tension failure Flexural-tension failure Flexural-shear failure with a minor shear crack

Avg.

224.0 (10.0)

1.28 (0.05)

512.3 (19.1)

5.37 (0.34)

1792.3

3442.3

BS16F

No.1

264.0

1.50

587.5

4.66

–

–

No.2

254.0

1.59

549.1

4.77

–

–

No.3

284.0

1.74

622.9

5.98

–

–

Avg.

267.3 (15.3)

1.61 (0.12)

586.5 (36.9)

5.14 (0.73)

1695.3

5286.0

No.1 No.2 No.3 Avg.

234.0 234.0 234.0 234.0 (0.0)

1.22 1.27 1.22 1.24 (0.03)

531.9 553.0 546.2 543.7 (10.8)

5.92 6.65 7.32 6.63 (0.70)

– – – 2340.3

– – – 4085.5

BS14F

Flexural-shear failure with multiple cracks Flexural-tension failure with multiple cracks Flexural-shear failure with multiple cracks Flexural-tension failure Flexural-tension failure Flexural-tension failure

(): items in the parentheses = standard deviation. a Performance acceptance criteria specified in BS EN 13230-2.

cracks). The maximum loads varied from 496.4 to 527.8 kN and all of the three specimens in this series were able to sustain these loads until relatively large deflections (up to 8 mm) without a significant drop in load. All three specimens in the BS16 series exhibited flexural-shear failures (see Figs. 8b and 9b). This emphasizes the fact that the design method for current prestressed concrete sleepers (sixteen strands with stirrups) is vulnerable to shear failure. The BS14 series exhibited early first crack formation at an average load of 224 kN, ranging from 214 to 234 kN, and significantly decreased post-cracking stiffness due to the reduced amount of strands (see Fig. 8c). The average maximum load was 512.3 kN, varying from 491.9 to 529.7 kN. Two of the three specimens in the BS14 series exhibited a flexural-tension failure rather than a shear failure or a flexural-shear failure. Sleepers having 0.75% steel fibers (BSF16 and BSF14) showed better crack control and exhibited multiple cracks (see Fig. 9) as

well as a higher flexural capacity than companion specimens with stirrups (BS16 and BS14) (see Table 4). They were also able to maintain significant loads over larger deflections. The slope of the descending branch of the load-deflection curve for these sleepers was smoother (see Fig. 8d and e), an indication of a slower rate of strength decay. It is interesting to note that except for the dominant flexure failure which occurred directly below the rail seat, the shear failures and flexural-shear failures occurred between the support at the end region of the sleepers and the rail seat (see Fig. 9). This result may be attributed to the reduced prestressing stresses near the end region due to the transfer length (about 50dp) of the pretensioned strand. 3.2.3. Comparison of the results Fig. 10 compares the applied load versus crack width responses for all sleepers. The crack widths reported are the average values

325

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

750

750 FrB

600

FrB

Load (kN)

Load (kN)

600 450 Frr

300

Specimen 1

450 Frr

300

Specimen 1

Specimen 2

150

Specimen 2

150

Specimen 3

Specimen 3

Average

Average

0

0 0

2

4

6

8

10

12

0

2

Deflection (mm)

4

6

8

10

12

Deflection (mm)

(a)

(b)

750

750 FrB FrB

600

Load (kN)

Load (kN)

600 450 Frr

300

Specimen 1

450 Frr

300

Specimen 1

Specimen 2

150

Specimen 2

150

Specimen 3

Specimen 3 Average

Average

0

0 0

2

4

6

8

10

12

Deflection (mm)

0

2

4

6

8

10

12

Deflection (mm)

(c)

(d)

750 FrB

Load (kN)

600 450 Frr

300

Specimen 1 Specimen 2

150

Specimen 3 Average

0 0

2

4

6

8

10

12

Deflection (mm)

(e) Fig. 8. Load versus deflection responses for static bending tests at rail seat section: (a) CC16; (b) BS16; (c) BS14; (d) BSF16; (e) BSF14.

Fig. 9. Typical failure modes of sleepers at rail seat section: (a) shear failure; (b) flexural-shear failure; (c) flexural-tension failure.

326

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

500 450

Load (kN)

400 350

CC16 BS16

300

BS14

250

BSF16 BSF14

200 0 0

0.5

1

1.5

2

2.5

3

Crack Width (mm) Fig. 10. Load versus crack width responses for static bending tests at rail seat section.

obtained from the three specimens in each series. The cracks were not visible up to Frr, which ranged from 234 to 284 kN, in all sleepers. However, the crack widths increased with increasing loads after the initial crack formation. The crack width of sleepers BS14, which was reinforced with a reduced amount of strands (fourteen strands), was larger than that of sleepers with sixteen strands (BS16). The same sleepers with steel fibers (BSF14) showed smaller crack widths compared to the sleepers BS14 and the crack widths are comparable to those in the sleepers with sixteen strands (CC16 and BS16). Sleepers BSF16 exhibited smaller cracks than the other sleepers, demonstrating the beneficial effects of steel fibers on crack control. The results also indicate that a lower amount of strands can achieve the same level of cracking resistance as sleepers with sixteen strands when steel fibers are added. The average load-deflection curves for the three specimens are also presented in Fig. 8 and these average curves were compared in Fig. 11 to investigate the effects of each variable on the overall

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 11. Comparison of responses for static bending tests at rail seat section: (a) effect of concrete mix; (b) effect of amount of strands (with stirrups); (c) effect of amount of strands (with fibers); (d) effect of steel fibers (with 16 strands); (e) effect of steel fibers (with 14 strands); (f) effect of steel fibers (partial substitution for strands).

327

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

Fig. 12. Typical load versus time history for the fatigue tests at rail seat section (BS16).

0.10

At a load of Fr0 (wFr0) < 0.10 mm : Criteria

Crack Width (mm)

After unloading (wres) < 0.05 mm : Criteria

0.05 0.03 0.02 0.01 0 CC16

BS16

BS14

BSF16

BSF14

Sleeper Fig. 13. Crack width results after fatigue tests and acceptance criteria.

response of the sleepers. Fig. 11a summarizes the average loaddeflection responses of the sleepers with the same reinforcement details but with different concrete mixes. Sleepers BS16, constructed with partial replacement of Type III cement by GGBFS, exhibited 4% and 8% increases in the first cracking load (Frr) and maximum load resistance (FrB), respectively, compared to sleepers CC16. Sleepers in the CC16 series are constructed with 100% of Type III cement and represent railway sleepers currently used in the Korean railway system. Sleepers BS16 had a significant 52% increase in toughness up to failure compared to sleepers CC16. This result can be attributed to the premature shear failure of sleepers CC16 (see Fig. 9a). This comparison indicates that sleepers produced by partial substitution of GGBFS can achieve not only lower CO2 emissions but also improved flexural capacity at the rail seat section. Given that the current design was shown to be conservative, the number of strands was reduced from sixteen to fourteen for a more

economical design and the results are compared in Fig. 11b and c. Sleepers with fourteen strands exhibited first cracking at decreased loads of 8–12% (see Table 4) and had a reduced post-cracking stiffness compared to sleepers with sixteen stands. However it is noted that sleepers with fourteen strands also exhibited 46–52% higher first cracking loads than the acceptance criteria specified in the Standard [23] (see Section 3.2.1). The sleepers with the reduced number of strands and without steel fibers (BS14) exhibited a comparable maximum load and post-cracking response compared to sleepers BS16 (see Fig. 11b). Fig. 11c compares the responses of the sleepers BSF16 with those in the BSF14 series. The sleepers with reduced number of strands and with steel fibers (BSF14) exhibited decreases in maximum load by 7% and toughness at failure by 23% compared to sleepers BSF16. Fig. 11d and e compares the average load-deflection responses of the sleepers with and without steel fibers. It is noted that sleepers BS16 and BS14 were reinforced with stirrups at the rail seat section and at the sleeper end region which provides improved shear resistance as well as improved anchorage of the strands, respectively. Sleepers BSF16 and BSF14 were constructed without stirrups but contain 0.75% of steel fibers (see Fig. 2). The addition of 0.75% steel fibers instead of stirrups increased the load at first cracking by 4 to 10% and increased the maximum load by 6 to 14% compared to the specimens with stirrups (no fibers). Furthermore, the sleepers with steel fibers were able to attain 1.18–1.24 times as much deflection before they reached maximum loads and they also had higher toughness both at the peak load (23– 31% increase) and failure load (19–56% increase) compared to sleepers with stirrups. The addition of steel fibers led to more ductile failures with shear failures transformed to flexural-shear failures, and flexural-shear failures transformed to flexural-tension failures. These results indicate that steel fibers increase the cracking resistance and the load carrying capacity as well as the toughness of prestressed concrete sleepers. The results also demonstrate that 0.75% of steel fibers can result in better performance of the railway sleepers than conventional stirrups. Thus, the addition of a sufficient amount of fibers can be used to replace traditional shear reinforcement in the railway sleepers. Sleepers BSF14, that contained 0.75% steel fibers, maintained a higher load carrying capacity in almost the entire deflection regions than sleepers CC16 and BS16, although they were reinforced with a reduced amount of strands (see Fig. 11f). This comparison indicates that steel fibers can compensate for the smaller number of strands. It is concluded that the combination of a reduced number of strands and steel fibers can be effectively used as an alternative design method for current railway sleepers. Fewer strands and no stirrups can significantly reduce the labor cost for producing the prestressed concrete sleepers. The beneficial effects of steel fibers were more pronounced in the sleepers with sixteen strands than those with fourteen strands. This observation can be attributed to the higher shear demand in the sleepers with sixteen strands. Specimens in series BS14 exhibited flexural-tension failures with minor concrete crushing or

Table 5 Results of post-fatigue tests at rail seat section. Sleeper

CC16 BS16 BS14 BS16F BS14F

First cracking

Maximum load

Comparison

Failure mode

Frr (kN) > 154 kNa

dFrr (mm)

FrB (kN) > 385 kNa

dFrB (mm)

Fr r;fatigue Fr r;static

Fr B;fatigue Fr B;static

224.0 244.0 234.0 254.0 234.0

1.46 1.40 1.47 1.45 1.33

501.3 519.7 516.5 575.0 519.1

6.56 7.85 7.74 8.01 6.83

0.96 1.00 1.04 0.95 1.00

1.05 1.01 1.01 0.98 0.95

(): items in the parentheses = standard deviation. a Performance acceptance criteria specified in BS EN 13230-2.

Shear failure Flexural-shear failure Flexural-tension failure Flexural-tension failure Flexural-tension failure

328

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

500

Load (kN )

450 400 350

CC16 BS16

300

BS14 BSF16

250

BSF14

200 0 0

0.5

1

1.5

2

2.5

3

Crack Width (mm)

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 14. Comparison of the response for post-fatigue tests at rail seat section: (a) load versus crack widths; (b) effect of concrete mix; (c) effect of amount of strands (with stirrups); (d) effect of amount of strands (with fibers); (e) effect of steel fibers (with 16 strands); (f) effect of steel fibers (with 14 strands).

minor shear cracks, whereas specimens in series BS16 exhibited more brittle failures, such as flexural-shear failure, compared to the sleepers in the BS14 series due to the higher shear demand of the sleepers BS16. The addition of 0.75% fibers in the BS16 series can control shear cracking and leads to the formation of multiple smaller cracks, resulting in the transformation of relatively brittle failure modes to more ductile failure modes. 3.3. Fatigue tests at rail seat section 3.3.1. Cracking of the sleepers caused by fatigue cycles Fig. 12 shows the typical load versus time history obtained from the fatigue tests. The fatigue testing machine simulated the loading history specified in the Standard. The measured crack widths after the fatigue tests as well as the crack width criteria specified in BS EN 13230-2 [23] are shown in Fig. 13. The maximum crack width after the fatigue tests, wFr0, ranged from 0.023 to 0.031 mm which were much smaller than the specified crack width criterion of

0.10 mm. After unloading, the crack width was reduced by approximately 0.006 mm, resulting in a residual crack width, wres, of 0.018–0.025 mm. These values were approximately 37 to 49% of the crack width criteria (0.05 mm) required by BS EN 13230-2 [23]. The results demonstrate that all sleepers produced in this study had sufficient fatigue performance during the 2  105 cycles of repeated loading. The sleepers with conventional concrete (CC16) and with slag concrete having a reduced number of strands (BS14) were more vulnerable to the fatigue loads compared to other sleepers. Whereas, sleepers with steel fibers (BSF16 and BSF14) exhibited smaller crack widths caused by fatigue loads (see Fig. 13). 3.3.2. Flexural strength capacity of the post-fatigue tests and code requirements The post-fatigue performance of the sleepers was evaluated (‘Stage C’ in Fig. 12) after the fatigue tests, using the sleepers damaged by fatigue loading. The post-fatigue test results are

329

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

(a)

(b)

(c)

(d)

(e) Fig. 15. Comparison between static tests and post-fatigue tests at rail seat section: (a) CC16; (b) BS16; (c) BS14; (d) BSF16; (e) BSF14.

summarized in Table 5 and these results are compared to the performance criteria of BS EN 13230-2 [23]. It should be noted that the first cracking load, Frr, was obtained from the fatigue tests (‘Stage A’ in Fig. 12) but the maximum flexural capacity, FrB, was measured from the post-fatigue tests (‘Stage C’ in Fig. 12). The values of Frr and FrB were much higher, ranging from 1.45 to 1.65 and 1.30 to 1.49, respectively, compared to the limits specified in the Standard [23]. Therefore, the sleepers showed sufficient flexural capacity even after the fatigue loading cycles.

3.3.3. Comparison of the results and crack width propagation Fig. 14a compares the load versus maximum crack width at each loading step during the post-fatigue tests and Fig. 14bf compare the post-fatigue performance of the sleepers considering the influence of several key parameters. Sleeper CC16 showed similar post-fatigue performance to the sleeper BS16 until a displacement

of 6.5 mm. However, this sleeper exhibited a sudden drop in its capacity due to the dominant shear crack formation (see Fig. 14b). The sleepers with reduced number of strands (BS14 and BSF14) exhibited decreased post-cracking stiffness compared to the sleepers with higher amounts of strands (BS16 and BSF16) (see Fig. 14c and d). They also had smaller cracking loads and maximum capacities compared to the sleepers with sixteen strands. These results coincide with the results from the static bending tests and crack width results also confirm these findings, i.e., that the sleepers reinforced with sixteen strands exhibited smaller crack widths at the same loading stages than sleepers with fourteen strands (see Fig. 14a). The sleepers with steel fibers exhibited smaller crack widths at the same loading stages (see Fig. 14a), higher post-cracking stiffness and cracking loads regardless of the number of strands (see Fig. 14e and f). This is attributed to the steel fibers in the concrete matrix bridging the cracks and limiting their growth as loading is

330

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

150

150 FcBn

FcBn

120

Load (kN)

Load (kN)

120 90 60

60

Specimen 1

30

Specimen 3

Specimen 2

Fcrn

Specimen 1

Fcrn

30

90

Specimen 3 Average

Average

0

0 0

5

10

15

20

25

30

0

5

10

Deflection (mm)

(a)

20

25

30

(b) 150

150

FcBn

FcBn

120

Load (kN)

120

Load (kN)

15

Deflection (mm)

90 60

Specimen 1 Specimen 2

Fcrn

30

90 60 Specimen 1 Fcrn

30

Specimen 3

Specimen 2 Specimen 3 Average

Average

0

0 0

5

10

15

20

25

0

30

5

10

15

20

25

30

Deflection (mm)

Deflection (mm)

(c)

(d)

150

150 FcBn

120

Load (kN)

Load (kN)

120 90 60

Specimen 1 Fcrn

30

90 CC16

60 TP,max

Specimen 2

BS16

TD,20mm

BS14

30

Specimen 3

BSF16

Average

BSF14

0

0 0

5

10

15

20

25

Deflection (mm)

(e)

30

0

5

10

15

20

25

30

Deflection (mm)

(f)

Fig. 16. Load versus midspan deflection responses for static bending tests at center section (negative bending): (a) CC16; (b) BS16; (c) BS14; (d) BSF16; (e) BSF14; (f) comparison of the average load-deflection response

increased. Fig. 14a, e and f clearly show the combined effects of steel fibers and the numbers of strands on the post-fatigue performance of the prestressed concrete sleepers. It is observed that the influence of the test variables on the postfatigue performance of the sleepers was similar to the influence of these variables on the static tests at the rail seat section. 3.3.4. Comparison of static tests and post-fatigue tests Fig. 15 compares the load-deflection responses of the sleepers obtained from post-fatigue tests and static tests at the rail seat section. The first cracking load, Frr, and the maximum flexural capacity, FrB, from those tests are also compared in Table 5. The ratios of Frr and FrB from the post-fatigue tests to those from the static tests ranged from 0.95 to 1.04 and 0.95 to 1.05, respectively. In addition, the complete load-deflection responses

are similar (see Fig. 15). Minor reductions of the post-cracking stiffness were observed for the sleepers which had experienced fatigue loading cycles and this reduction is more evident for the sleepers reinforced with reduced number of strands (BS14 and BSF14), as shown in Fig. 15c and e. The results demonstrate that there was no significant deterioration from the fatigue cycles but the sleepers reinforced with fourteen strands are somewhat more vulnerable to fatigue cycles compared to the sleepers with sixteen strands. 3.4. Static bending tests at center section The applied load versus midspan deflection responses of a set of three sleepers for each variable, including the average loaddefection curve, are shown in Fig. 16. The results of these tests, including the failure mode, cracking loads, maximum loads and

331

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333 Table 6 Results of static bending tests at center section. Sleeper series

Test

CC16

BS16

BS14

BS16F

BS14F

First cracking

Maximum load

Toughness

Fcrn (kN) > 36.1 kNa

dFcrn (mm)

FcBn (kN) > 90.4 kNa

dFcBn (mm)

No.1 No.2

56.1 56.1

1.74 1.80

112.2 104.0

13.13 12.75

No.3

61.1

2.41

101.7

12.55

Avg.

57.8 (2.9)

1.98 (0.37)

106.0 (5.5)

12.81 (0.29)

No.1

56.1

1.89

106.9

12.49

No.2

61.1

1.88

117.0

13.64

No.3 Avg.

61.1 59.4 (2.9)

1.91 1.89 (0.02)

111.6 111.8 (5.1)

12.55 12.89 (0.65)

No.1

56.1

1.81

107.8

12.93

No.2

56.1

1.77

107.8

13.22

No.3 Avg.

56.1 56.1 (0.0)

1.81 1.80 (0.02)

107.5 107.7 (0.2)

13.85 13.33 (0.47)

No.1

66.1

2.29

117.8

11.71

No.2 No.3 Avg.

61.1 61.1 62.8 (2.9)

2.13 2.01 2.14 (0.14)

105.9 108.5 110.7 (6.3)

11.16 10.70 11.19 (0.51)

No.1 No.2 No.3 Avg.

61.1 56.1 61.1 59.4 (2.9)

1.99 1.76 2.00 1.92 (0.14)

112.1 108.8 105.1 108.7 (3.5)

12.34 9.79 10.67 10.93 (1.30)

TP,max (kN mm)

Failure mode TP,D20mm (kN mm) Flexural-shear cracking with concrete crushing Flexural-shear cracking with minor concrete crushing Flexural-shear cracking with bond splitting along the strand

1041.7

1792.9 Flexural-shear cracking with minor concrete crushing Flexural-shear cracking with minor concrete crushing Flexural-shear cracking with concrete crushing

1072.6

1871.9 Flexural-shear cracking with minor concrete crushing Flexural-shear cracking with bond splitting along the strand Flexural-shear cracking with concrete crushing

1078.0

1805.7 Flexural-shear cracking with minor concrete crushing Flexural-tension failure with multiple cracks Flexural-tension failure with multiple cracks

968.6

1887.0 Flexural-tension failure with multiple cracks Flexural-tension failure with multiple cracks Flexural-tension failure with multiple cracks

930.0

1837.9

(): items in the parentheses = standard deviation. a Performance acceptance criteria specified in BS EN 13230-2.

corresponding deflections are summarized in Table 6. This table also gives the performance acceptance criteria for the response at loads, Fcrn and FcBn as specified in the BS EN 13230-2 Standard [23]. These key loads for determining the performance are:

Fcrn > Fc0n ¼ 36:1 kN

ð3Þ

FcBn > k2s  Fc0n ¼ 2:5  36:1 ¼ 90:4 kN

ð4Þ

The same static coefficient of k2s as that used for the static tests at the rail seat section, which considering accidental load cases, was used for this test series. The toughness up to a maximum load and up to a deflection of 20 mm (TP,max and TD,20mm, respectively) were calculated and are compared in Table 6 as well. All sleepers behaved in a similar manner in the ascending branch of the loading response before first cracking occurred (see Fig. 16). The sleepers with steel fibers (BSF16 and BSF14) showed slightly higher (about 6%) loads at first cracking compared to companion sleepers without steel fibers (BS16 and BS14) (see Table 6). The sleepers with sixteen strands (BS16) also exhibited a slightly higher (6%) loads at first cracking than those with a reduced number of strands (BS14). Sleepers with only fourteen strands and having 0.75% of steel fibers (BSF14 series) showed comparable and even slightly better crack control compared to the sleepers with sixteen strands and no fibers (BS16 and CC16 series) (see Table 6). This result indicates that steel fibers compensated for the reduced number of strands in controlling the cracks. The ratio of the first cracking load measured from the tests, Fcrn, to their performance

acceptance criteria specified by the BS EN 13230-2 Standard [23] ranged from 1.55 to 1.69. Therefore, all sleepers produced in this test program had sufficient cracking resistance at the center section of the sleepers. After the formation of the first crack, the loading was increased to the maximum load, FcBn. The average maximum load, FcBn, for all of the sleepers ranged from 106 kN to 112 kN with a standard deviation of 2.35 kN, indicating that the difference in the variables had little effect. The ratios of these maximum loads to their performance acceptance criteria specified by the BS EN 13230-2 Standard [23] ranged from 1.17 to 1.24. The average load deflection response and toughness of all sleepers were similar. However, the load at larger deflections tended to drop somewhat faster in the sleepers with Type III Portland cement (CC16 series currently used in Korean railway system) than the sleepers with slag concrete. Typical failure modes of the sleepers at the center section are shown in Fig. 17. All flexural cracks were initiated from the bottom surface (i.e., top surface of sleepers in service) near midspan, followed by three different major failure modes: 1) flexural-shear cracking with significant concrete crushing; 2) flexural-shear cracking with bond splitting cracks along the strands; and 3) flexural-tension failure. Sleepers without steel fibers exhibited flexural-shear cracking with either concrete crushing or bond splitting failures (see Figs. 17a and b). In contrast, all sleepers with steel fibers exhibited flexural-tension failures while controlling flexure-shear cracks, concrete crushing and bond splitting (see Fig. 17c).

332

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333

(a)

(b)

(c)

Fig. 17. Typical failure modes of sleepers at center section: (a) flexural-shear cracking with concrete crushing; (b) flexural-shear cracking with bond splitting along the strand; (c) flexural tension failure.

4. Conclusions Static and fatigue performance of prestressed concrete sleepers produced with ground granulated blast furnace slag (GGBFS) and steel fibers were investigated in this study, resulting in the following conclusions: (1) All sleepers produced in this study showed shoulder pull-out resistances exceeding the minimum requirement of the Korean Railway Standard (KRS TR 0008). The sleepers incorporating 0.75% steel fibers had higher loads at first cracking, demonstrating the ability of the steel fibers to control concrete cracking. (2) All sleepers produced in this study showed much higher cracking and maximum loads, at both the rail seat and center sections of the sleepers, than the corresponding performance criteria specified in the Standard (BS EN 13230-2). In addition, the crack widths after the fatigue cycles also satisfied the limits of this Standard. Therefore, it is possible to use sleepers produced with GGBFS and steel fibers. (3) Sleepers with partial replacement of Type III cement by GGBFS (56%) showed improved performance at the rail seat section, such as delayed first cracking, higher flexural capacity, and more ductile failure modes, compared to sleepers with Type III cement, which are currently used in Korean railway systems. Therefore, GGBFS can be used possibly as an alternative binder for prestressed concrete sleepers to achieve not only lower CO2 emissions but also improved flexural capacity at the rail seat section. (4) Sleepers with reduced amounts of strands (fourteen strands) exhibited a decreased first cracking load and reduced postcracking stiffness. Nonetheless, these sleepers showed a comparable maximum load to the sleepers with sixteen strands and exhibited more ductile failure modes due to the smaller shear demand at flexural ultimate. (5) Sleepers with steel fibers instead of stirrups showed a superior flexural performance at the rail seat section, which means that steel fibers can substitute for stirrups in prestressed concrete sleepers. Additionally, steel fibers compensated for the reduced flexural capacity of the sleepers with only fourteen strands. Therefore, a reduced number of strands together with the use of steel fibers can result in an economical solution for constructing sleepers in a precast plant. (6) The post-fatigue performance of the sleepers was similar to the static performance at the rail seat section, which means that there was no significant deterioration from the fatigue cycles.

(7) All of the sleepers showed similar bending moment capacities at their center sections (negative bending), indicating that the effects of the variables on the flexural strength at the center section were not significant. The presence of steel fibers provided some beneficial effects such as higher loads at first cracking as well as prevention of concrete crushing and bond splitting cracks.

Acknowledgements This work was supported by the Industrial Strategic Technology Development Program (10063488, Development of Earthquake Resisting Reinforced Concrete using Grade 700 MPa Reinforcing Bars for Enhancement of Seismic Safety) funded by the Ministry of Trade, Industry & Energy (MI, Republic of Korea) and a grant from R&D Program of the Korea Railroad Research Institute, Republic of Korea. References [1] Ferdous W, Manalo A. Failures of mainline railway sleepers and suggested remedies – review of current practice. Eng Fail Anal 2014;44:17–35. [2] Shin H-O, Yang J-M, Yoon Y-S, Mitchell D. Mix design of concrete for prestressed concrete sleepers using blast furnace slag and steel fibers. Cement Concr Compos 2016;74:39–53. [3] Manalo A, Aravinthan T, Karunasena W, Ticoalu A. A review of alternative materials for replacing existing timber sleepers. Compos Struct 2010;92:603–11. [4] Shojaei M, Behfarnia K, Mohebi R. Application of alkali-activated slag concrete in railway sleepers. Mater Des 2015;69:89–95. [5] Lutch RH, Harris DK, Ahlborn TM. Prestressed concrete ties in North America. Michigan Tech; 2009. [6] Taherinezhad J, Sofi M, Mendis P, Ngo T. A review of behaviour of prestressed concrete sleepers. Electron J Struct Eng 2013;13:1–16. [7] Zhao J, Chan AHC, Burrow MPN. Reliability analysis and maintenance decision for railway sleepers using track condition information. J Operat Res Soc 2007;58:1047–55. [8] Remennikov AM, Kaewunruen S. Experimental load rating of aged railway concrete sleepers. Eng Struct 2014;76:147–62. [9] Mithun BM, Narasimhan MC. Performance of alkali activated slag concrete mixes incorporating copper slag as fine aggregate. J Clean Prod 2016;112:837–44. [10] Yazıcı H, Yardımcı MY, Yig˘iter H, Aydın S, Türkel S. Mechanical properties of reactive powder concrete containing high volumes of ground granulated blast furnace slag. Cement Concr Compos 2010;32:639–48. [11] Won J-P, Kim H-H, Lee S-J, Choi S-J. Carbon reduction of precast concrete under the marine environment. Constr Build Mater 2015;74:118–23. [12] ACI Committee 544. Report on fiber reinforced concrete. Farmington Hills, United States: American Concrete Institute; 1996. [13] Kaewunruen S, Remennikov AM. Progressive failure of prestressed concrete sleepers under multiple high-intensity impact loads. Eng Struct 2009;31:2460–73. [14] Kaewunruen S, Remennikov AM. Dynamic flexural influence on a railway concrete sleeper in track system due to a single wheel impact. Eng Fail Anal 2009;16:705–12.

J.-M. Yang et al. / Engineering Structures 134 (2017) 317–333 [15] Kaewunruen S, Remennikov AM. Effect of a large asymmetrical wheel burden on flexural response and failure of railway concrete sleepers in track systems. Eng Fail Anal 2008;15:1065–75. [16] Mohammadzadeh S, Vahabi E. Time-dependent reliability analysis of B70 prestressed concrete sleeper subject to deterioration. Eng Fail Anal 2011;18:421–32. [17] UIC Code 713R. Design of monoblock concrete sleepers. Paris, France. International Union of Railways; 2004. [18] Kaewunruen S, Remennikov A. Post-failure mechanism and residual loadcarrying capacity of railway prestressed concrete sleeper under hogging moment. Proceedings of the International Conference on Structural Integrity and Failure (SIF2006). Sydney, Australia. 2006:331–6. [19] Kaewunruen S, Remennikov A. Rotational capacity of railway prestressed concrete sleeper under static hogging moment. In: Proceedings of the 10th East Asia-Pacific Conference on Structural Engineering and Construction (EASEC-10). Bangkok, Thailand; 2006:399–404. [20] KRS TR 0008-11(R). Prestressed concrete sleeper. Gyeonggi Province, South Korea. Korean Railway Standards; 2011. [21] KRS TR 0008-15(R). Prestressed concrete sleeper. Gyeonggi Province, South Korea. Korean Railway Standards; 2011. [22] BS EN 13230-1:2009. Railway applications – Track – Concrete sleepers and bearers – Part 1: General requirements. London, United Kingdom. British Standards Institution; 2009.

333

[23] BS EN 13230-2:2009. Railway applications – Track – Concrete sleepers and bearers – Part 2: Prestressed monoblock sleepers. London, United Kingdom. British Standards Institution; 2009. [24] Yoon I-J. Structural benefits of using fiber and ground granulated blast-furnace slag for PSC railway ties [Maters Thesis]. Seoul, South Korea: Korea University; 2013. [25] Taemyung Industrial Co., Ltd. KNR sleeper for 60kgK rail (KRS TR 0008). Gyeonggi Province, South Korea; 2008. [26] KS D 7002:2011. Uncoated stress-relieved steel wires and strands for prestressed concrete. Seoul, South Korea. Korea Standard Association; 2011. [27] KS D 3552:2014. Low carbon steel wires. Seoul, South Korea. Korea Standard Association; 2014. [28] ISO 10544:1992. Cold-reduced steel wire for the reinforcement of concrete and the manufacture of welded fabric. International Standard Organization; 1992. [29] KRS TR 0014-09. Rail Fastening System. Gyeonggi Province, South Korea. Korean Railway Standards; 2009. [30] BS EN 13481-2:2012. Railway applications – Track – Performance requirements for fastening systems – Part 2: Fastening systems for concrete sleepers. London, United Kingdom. British Standards Institution; 2009.