Static and fatigue test on lightweight UHPC-OSD composite bridge deck system subjected to hogging moment

A cost-effective Lightweight Composite Bridge Deck (LCBD) system, including Orthotropic 14 Steel Deck (OSD) and lightweight Ultra-High Performance Concrete (UHPC) layer is 15 proposed to increase the stiffness and fatigue performance of conventional OSD. Static and 16 fatigue tests on two full-scale strip models subjected to four-point bending were carried out. 17 The static nominal cracking stress of the UHPC layer with reinforcement spacing of 80 mm is 18 24.59 MPa, while it increases to 35.68 MPa when the reinforcement spacing is reduced to 19 half (40mm); both values are far greater than the nominal stress of 12.7 MPa obtained in the 20 prototype bridge. Increasing the reinforcement ratio can increase the bending stiffness of 21 LCBD and decrease the tensile strain of the UHPC layer, while the change in range is relative 22 slight. Furthermore, the flexural strength of UHPC and the reinforcement ratio are important 23 factors affecting the fatigue life of the UHPC layer. When the reinforcement spacing 24 increases from 40 mm to 80 mm, the fatigue life of the UHPC layer still satisfies related code 25 requirements. Thus, for reduction in the engineering cost and construction complexity, the 26 reinforcement spacing can be set as 80 mm. However, the application of the UHPC as the

slight. Furthermore, the flexural strength of UHPC and the reinforcement ratio are important 23 factors affecting the fatigue life of the UHPC layer. When the reinforcement spacing 24 increases from 40 mm to 80 mm, the fatigue life of the UHPC layer still satisfies related code 25 requirements. Thus, for reduction in the engineering cost and construction complexity, the 26 reinforcement spacing can be set as 80 mm. However, the application of the UHPC as the 27 steel deck pavement, the rib-to-diaphragm welded joint is still prone to fatigue cracks. In 28 addition, the existing S-N curves are hard to directly use for fatigue life prediction of the 29 UHPC layer because of the great differences in the definition of stress level and evaluation 30 index of failure in the fatigue test, which need to be modified in further studies. 31

37
Orthotropic steel deck (OSD) is a typical bridge deck system composed of stiffeners 38 (longitudinal and transverse rib) which are perpendicular to each other in longitudinal and 39 transverse directions and together with deck plate (see Fig. 1) [1][2][3][4][5][6][7][8] . OSD has the advantages of 40 high strength-to-weight ratio, superior integrity, large loading capacity and short construction 41 period. These merits have resulted in a widespread application of conventional OSDs in been widely recognized [9][10][11][12][13][14][15][16][17][18][19][20] as one of the effective means to reduce the fatigue stress 48 amplitude in an OSD. In 2002, Buitelaar [9] proposed to strengthen OSD with 50 mm 49 Reinforced High Performance Concrete (RHPC) connected via an epoxy resin bonding layer. 50 In addition, it was successfully applied to Caland Bridge in the Netherlands. Different from 51 ordinary concrete, Ultra-high Performance Concrete (UHPC) [21][22] is a type of advanced 52 cement-based composite with superior compressive (>120 MPa) and tensile strength (>8 53 MPa), high elastic modulus (40 -50 GPa) and excellent durability. In 2011, Shao et al. [10] 54 have proposed a new type of composite bridge deck system composed of OSD and 55 lightweight UHPC layer, in which the reinforced UHPC and OSD were connected via stud 56 shear connectors. In such new type of Lightweight Composite Bridge Deck (LCBD), the 57 thickness of UHPC layer is generally no more than 50 mm, and the thickness of the steel 58 plate is from 12 to 16 mm.  62 Although LCBD has been developed for nearly 10 years in China, there is limited 63 research existing on the basic mechanical properties and fatigue performance [10][11][12][13][14][15][16][17][18][19]23,24] of 64 such system. Previous push-out test results show that the LCBD connected by shear studs has 65 good mechanical performance [10,12] . Finite Element Analysis (FEA) and full-scale model 66 tests have shown that the reduction of the peak stress in the steel deck of LCBD is about 50 to 67 80 % compared with the peak stress in OSD without UHPC pavement, resulting in a 68 reduction of the risk of fatigue cracking [10,12,24] . In addition, increasing the reinforcement 69 ratio of the UHPC layer can significantly reduce the potential cracking of the UHPC layer 70 [12,14] . Moreover, previous research has shown that the spacing of the reinforcement should 71 not exceed 50 mm [10][11][12][13][14]23,24] . The LCBD has been applied in many bridges in China, such as 72 the Foshan-Fuchen Bridge [12] , the Zhuzhou-Fengxi Bridge [12] , and the Second-Dongting- 73 Lake Bridge [13] , etc., where the diameter of the reinforcing mesh in the UHPC layers is 10 74 mm, and the reinforcing spacing is 35 to 50 mm. In addition, the technical standard for light-75 weighted composite deck system [25] stipulates that the spacing of UHPC reinforcement in the 76 LCBD shall be less than 67 mm. 77 However, due to the small reinforcement spacing in the UHPC layers, the difficulties 78 associated with UHPC pouring and fibre dispersion are increasing, and the quality control 79 becomes more challenging. Furthermore, a dense reinforcement spacing will also increase the 80 construction cost. To the best of the authors' knowledge, the major reason for the above 81 requirements or regulations on the reinforcement spacing may be that the flexural and  UHPC with high tensile capacity, it is possible to increase the reinforcement spacing of   87   UHPC layer, improve the construction conditions and reduce the fabrication cost.   88   So far, previous research mainly focused on the static and fatigue behaviour of LCBDs   89 with the UHPC pavement under sagging bending moment [10,13,15] . There is still lack of 90 knowledge on the static and fatigue behaviours of the LCBD under hogging bending moment 91 [16] . Since both the UHPC layer and the reinforcements resist tensile stresses under hogging 92 moment, it is necessary to investigate the static and fatigue performance of the UHPC layer 93 on the LCBD. Furthermore, although some scholars have carried out a small number of 94 fatigue tests on composite beams [26,27] and the LCBD [11,14,16,23,24] subjected to hogging 95 moment, there were no fatigue cracks observed in the UHPC bridge deck until the end of 96 most fatigue tests [11,16,23,24] . Consequently, the crack propagation process, failure mode and 97 fatigue life under cyclic loading were not identified in their tests. In particular, for the UHPC 98 layer with different reinforcement ratios, the failure mode of LCBD under fatigue load is not 99 well understood. Therefore, it is necessary to carry out more studies to understand the fatigue   The dry mixture used in this study is composed of Portland cement, silica fume, mineral 144 powder, and quartz sand [29] , and the mix proportion of UHPC is shown in Table 1   In addition, the shear studs are in linear elastic stage at this moment [19] . Therefore, for 211 reducing the computing cost and improving computing efficiency, the bond slip between steel 212 and UHPC was not considered in this FEM, and a tie constraint was adopted between UHPC 213 layer and deck plate [10,15] . The boundary conditions at the beam ends are that the three 214 degrees of freedom of nodes were restricted at one end, while only the vertical freedom of 215 nodes was restricted at the other end. In addition, symmetric boundary conditions were 216 applied to the symmetry plane.
where σ is the nominal tensile stress of the UHPC layer under hogging moment; M and I is 270 the sectional moment and the inertia moment of the composite bridge deck, respectively; P is 271 the applied load in the test; l is the span of simply supported girder (l=900 mm); ne is the 272 Young's modulus ratio of steel to concrete (Es=206 GPa, Ec=44.4 GPa); h and b is the height 273 and width of the UHPC layer; beq is the equivalent width of the UHPC layer; yi is the distance 274 from the cross-sectional centroid of each part to the x-axis (i=1, 2, 3, z); and I1, I2, and I3 is 275 the inertia moment of the OSD, steel reinforcement, and UHPC layer, respectively. 276    these curves gradually tend to be nonlinear. Take L3 as an example, the load-deflection 320 curves of the LCBD can be divided into four stages (Fig. 14b): (1) Table 2, and the constitutive model of UHPC in tension and 343 compression [29] is shown in Fig. 15. The material model for steel rebars and OSD was 344 defined using an elastic-plastic constitutive law with elastic modulus of 206 GPa and 345 Poisson's ratio of 0.3 [28] . Assuming a complete bond between the UHPC layer and shear 346 studs, a tie constraint was used between the UHPC layer and shear studs. The contact 347 between the deck plate and UHPC layer was defined as surface-to-surface hard contact, in 348 which the coulomb friction coefficient was 0.5 [33] . The global mesh size of OSD, UHPC 349 layer and reinforcement was 15 mm, 10 mm, and 10 mm, respectively. The mesh size of the 350 concerned area of OSD was refined to 1.0 mm. The details of the FEM are shown in Fig. 16. In addition, as shown in Fig. 17a, the bending stiffness of the LCBD increases with the 379 reinforcement ratio, but the increment is rather small. Fig. 17b shows that the effect of the 380 reinforcement on the strain of UHPC layer is insignificant in the elastic stage. In the 381 nonlinear stage, when the reinforcement spacing increased from 40 mm to 80 mm, the 382 maximum difference of the applied load is no more than 6% under the same strain, and when 383 the reinforcement spacing increased from 80 to 120 mm, the maximum load difference is no 384 more than 3%. In Fig. 17b, the cracking strains in the test are much larger than the elastic 385 strain εct but are smaller than the cracking strain εcc in the direct tension test. It seems that the 386 higher the reinforcement ratio is, the larger the actual cracking strain is. During the fatigue test, when the fatigue load reached 1.142 million cycles in the first 407 phase, a fatigue crack was observed at the rib-to-diaphragm welded joint (Fig. 19a), where 408 the reinforcement spacing in the UHPC layer is 80 mm. To prevent the OSD from further 409 damage, the fatigue crack at the rib-to-diaphragm welded joint was repaired by penetration 410 welding as shown in Fig. 19b. After the repair, in the second phase of the fatigue loading, the 411 specimen experienced 3.229 million loading cycles before the repaired welding part at the 412 diaphragm cut-outs cracked again (Fig. 19c), and the base metal of the diaphragm also 413 cracked. At the end of the test, the maximum crack length at the rib-to-diaphragm welded 414 joint was 47.6 cm (Fig. 19d).  Fig. 7. According to existing research [1][2][3][4][5][6][7] , five typical fatigue-prone details were 491 selected as the monitoring points (point-A to point-E) (see Fig. 7b and c). In comparison with 492 the nominal stress method, the hot spot stress method is regarded as a more appropriate 493 method to predict the fatigue life of complex geometries such as welded connections in OSDs 494 [34,35] . Based on the IIW [34] , the hot spot stresses at the monitoring points can be obtained by 495 applying extrapolation approach, as shown in Eq. (6) and Fig. 24. Therefore, the maximum 496 hot spot stress of the monitoring points can be obtained, as illustrated in Table 6  For identifying the fatigue performance of the unfavorable fatigue-prone details, the 516 calculations were conducted according to related Eurocodes [31,36,37] and Siwowski et al. [7] . 517 For the fatigue limit state, the safety level can be given by Eq. 7 to 10.
where fat  is the fatigue safety level;    the requirement in the specification [36,38] . Moreover, the fatigue life of the UHPC layer is 99 576 times that of the rib-to-diaphragm welded joint. Therefore, the reinforcement spacing of the 577 UHPC layer can be expanded to 80 mm, so as to reduce the engineering cost and construction 578 complexity. 579 580 Table 8 The stress amplitude and equivalent fatigue cycles of the UHPC layer.   [14,[39][40][41][42][43][44][45] , and obtained corresponding S-N 590 curves, respectively, under their own experimental conditions, as depicted in Table 9. 591 However, based on the summary of existing data, it is found that there are few research on 592 the flexural fatigue properties of reinforced high performance FRC structures [14,40,41,43,45] , 593 especially for the research on the S-N curves of LCBD. Besides, because of the great where D is the damage index, ni is the number of cycles associated to the stress range of Δσp,i, 644 and Ni is the number of cycles calculated by S-N curve. 645 According to the Palmgren-Miner rule, when D≥ 1, the fatigue failure on the monitoring 646 point occurs [36,38] . Based on the calculation results, D of the Spacing-80MM is 12.79, and D 647 of the Spacing-40MM is 1.241. As a result, the difference between the test results of the 648 Spacing-80MM and the theoretical value is significant. In addition, for the Spacing-40MM, 649 the test result is also larger than the theoretical value of 1. It was found that the fatigue life of 650 the UHPC layer in our test is larger than that of in the reference [14] . In order to understand the 651 reason for this difference, the main differences of fatigue test parameters between this paper 652 and Li [14] are listed in Table 11. The compressive strength and flexural strength of UHPC 653 material used in this test are larger than those in Li [14] . In particular, the ratio of the flexural 654 strength in present test is 1.846 times of that in reference [14] .  and cracking strain of the LCBD, and reduce the tensile strain of the UHPC layer, but these 670 variations are small. The actual cracking strain in test is larger than the elastic strain but is 671 smaller than the cracking strain in the tension constitutive model. It seems that the higher the 672 reinforcement ratio is, the larger the actual cracking strain is. 673 • Application of UHPC as a steel deck pavement, the stress amplitude of the typical 674 fatigue-prone details in OSD is significantly reduced, while the rib-to-diaphragm welded joint 675 is still prone to fatigue cracks. Besides, when the reinforcement spacing increases from 40 676 mm to 80 mm, the influence on stress amplitude of OSD can be ignored (on more than 5%).

677
• Increasing the reinforcement ratio of the UHPC layer can improve its fatigue resistance.