Collapse Analysis of Utatsu Ohashi Bridge Damaged by Tohuku Tsunami Using Applied Element Method Hamed Salem, Suzan Mohssen, Kenji Kosa, Akira Hosoda

Collapse Analysis of Utatsu Ohashi Bridge Damaged by Tohuku Tsunami using Applied Element Method Hamed Salem, Suzan Mohssen, Kenji Kosa, Akira Hosoda

Journal of Advanced Concrete Technology, volume 12 ( 2014 ), pp. 388-402

The Micro Truss Model: An Innovative Rational Design Approach for Reinforced Concrete Hamed M. Salem Journal of Advanced Concrete Technology, volume 2 ( 2004 ), pp. 77-87

Computer-Aided Analysis of Reinforced Concrete Using a Reﬁned Nonlinear Strut and Tie Model Approach Hamed M. Salem, Koichi Maekawa Journal of Advanced Concrete Technology, volume 4 ( 2006 ), pp. 325-336 Journal of Advanced Concrete Technology Vol. 12, 388-402, October 2014 / Copyright © 2014 Japan Concrete Institute 388

Scientific paper Collapse Analysis of Utatsu Ohashi Bridge Damaged by Tohuku Tsunami using Applied Element Method Hamed Salem1*, Suzan Mohssen2, Kenji Kosa3 and Akira Hosoda4

Received 1 March 2014, accepted 29 September 2014 doi:10.3151/jact.12.388 Abstract The 2011 Tohuku tsunami on the east coast of Japan resulted in killing more than 15,000 people and missing more than 2,500 people, washing away of more than 250 coastal bridges and loss of US$235 billion. Collapse of coastal bridges due to tsunami impact represents a huge obstacle for rescue works. Therefore, in the current study, the collapse of Uta- tsu Ohashi bridge is numerically studied. The analysis is carried out using the Applied element Method due to its advan- tages of simulating structural progressive collapse. The AEM is a discrete crack approach, in which elements can be separated, fall and collide to other elements in a fully nonlinear dynamic scheme of computations. The Utatsu Ohashi bridge collapse was successfully simulated using AEM. It was numerically found that the amount of trapped air between deck girders during tsunami had a significant effect on the behavior of the bridge. This is attributed to the buoyant force accompanied with the trapped air. A simplified method for estimating trapped air was assumed and proved to give reasonable results compared to reality. Three different solution examples for mitigating collapse of similar existing bridges were introduced and applied to Utatsu Ohashi bridge case and found to be efficient for preventing collapse.

1. Introduction Bay suffered extensive damage by tsunami as shown in Fig. 1, where most of the bridge decks were washed On March 11th, 2011, a powerful tsunami, Tohuku tsu- away by the tsunami forces (Kawashima et al. 2011; Fu nami, with 10m-high waves swept over the east coast of et al. 2012). Japan. The tsunami was produced by a 9.0 Richter mag- The objective of the current study is to numerically nitude earthquake that reached depths of 24.4km mak- investigate the collapse mechanism of the Utatsu bridge ing it the fourth-largest earthquake ever recorded. The and propose structural design enhancements to avoid Japanese National Police Agency confirmed 15,884 collapses of similar bridges in future under tsunami ac- deaths, 6,150 injured, and 2,640 people missing across tion. The choice of the numerical method to do such twelve prefectures, as well as 126,631 buildings totally investigation was very important because of the signifi- collapsed, with further 272,653 buildings 'half col- cant need to simulate the collapse of different parts of lapsed', and 743,492 buildings partially dam- the bridge to the end. Although the FEM is a robust and aged. Approximately 452,000 people were relocated to well-established structural analysis method, it is not the shelters. The violent shaking resulted in a nuclear emer- optimum solution for the scope of the current study. gency, in which the Fukushima Daiichi nuclear power Many drawbacks are associated with the FEM progres- plant began leaking radioactive steam. The World Bank sive collapse analysis; the element damage separation, estimates that it could take Japan up to five years to falling and collision with other elements are very diffi- financially overcome US$235 billion damages. cult (Hartmann et al., 2008). Therefore, in the current The 2011 Tohuku tsunami also caused extensive and study, the numerical analysis was carried out using the severe structural damage to various infrastructures in Applied Element Method. The Applied Element Method north-eastern Japan, especially in the coastal area of is based on discrete crack approach and is capable of Iwate, Miyagi, Fukushima and Ibaraki Prefectures. following the structure's behavior to its total collapse More than 250 bridges were washed away. As an exam- (Tagel-Din and Meguro 2000; Meguro and Tagel-Din ple, Utatsu Bridge at Minami-Sanriku Town over Irimae

1Professor, Structural Engineering Department, Cairo University, Egypt. *Corresponding author, E-mail: [email protected] 2Structural engineer, Applied Science International, Cairo, Egypt. 3Professor, Department of Civil Engineering, Faculty of Engineering, Kyushu Institute of Technology. 4Associate professor, Faculty of Urban Innovation, Fig. 1 Destroyed Utatsu Ohashi bridge in Miyagi Prefec- Yokohama National University, Japan. ture due to Tohuku Tsunami. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 389

2001; Tagel-Din 2002; Meguro and Tagel-Din 2003; original position. It is noted that S3 and S4, and S5, S6 Tagel-Din and Rahman 2004; Galal and ElSawy 2010; and S7 flowed out together. This was explained that Sasani and Asgitoglu 2008; Salem et al. 2011; Park et al. they were tied by cable restrainers for preventing exces- 2009; Helmy et al. 2009; Helmy et al. 2012; Helmy et sive superstructure response under a large seismic exci- al. 2013; Sasani 2008; Wibowo 2009; Salem 2011; Sa- tation as shown in Fig. 3. On the other hand, S8, S9 and lem and Helmy 2014) S10 overturned during being floated as shown in Fig. 4. Figure 5 shows the top of a pier after superstructures 2. Collapse of Utatsu Ohashi bridge were washed away. Two types of steel devices were set as an unseating prevention device in this column; one is Utatsu Ohashi Bridge consisted of 12 spans with the devices aiming to increasing seat length, and the prestressed simply supported girders. It included 3 types other is the devices which were set for preventing ex- of superstructures with spans ranging from 14.4m to cessive deck displacement in the longitudinal direction. 40.7m as shown in Fig. 2. Piers consisted of both circu- It is important to observe that none of those devices lar and rectangular RC columns supported on pile foun- were detached from the pier, which must have happened dations. Bridge columns have been retrofitted by RC if the decks were simply washed away laterally. Only jacketing and an extension for the seat length was in- some of those devices were rotated, as shown in Fig. 6, stalled at the top of pier. The superstructures from S3 to and therefore, It is likely that the decks were uplifted by S10 were completely washed away from their supports tsunami buoyancy force and then they were washed in the transverse direction due to tsunami while the su- away. Steel plate bearings used in this bridge was very perstructures S1, S2, S11 and S12 were not. It was simple as shown in Fig. 7 such that both uplift and lat- found that concrete and steel shear keys, installed at the eral force capacities were limited. This is also the case pier girder, were damaged and some damage took place at a column shown in Fig. 6, in which four stoppers did at the land side of the pier girders. not tilt. Nevertheless, a RC side stopper at the land side The outflow displacements of S3~S10 are shown in collapsed probably due to a transverse force which ap- Fig. 2. The spans located at the center such as S5-S7 plied from the deck. It is likely that due to tsunami force and S8 were flowed 28 m and 41 m away from the the deck uplifted at the sea side first being supported only at the land side, which resulted in larger tsunami force. Thus the side stopper at the land side collapsed due to excessive concentration of tsunami force.

Fig. 2 Outflow of Superstructure of Utatsu Ohashi bridge in Miyagi Prefecture due to Tohuku Tsunami (Kawashima et al. 2011).

Fig. 4 An overturned superstructure in Utatsu Ohashi bridge.

Fig. 5 Steel devices for extending seat length (short) and Fig. 3 Effective Restrainers to tie adjacent decks to- steel stoppers for preventing excessive longitudinal deck gether in Utatsu Ohashi bridge (Kawashima et al. 2011). response due to ground motions (tall). H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 390

Fig. 7 An upper steel bearing after superstructure col- Fig. 6 Failure of a RC side stopper, with non-damaged lapse (Kawashima et al. 2011). four steel stoppers used for preventing excessive longitudinal deck response. springs are responsible for transfer of normal and shear stresses among adjacent elements. Each spring repre- 3. The applied element method (AEM) sents stresses and deformations of a certain volume of the material as shown in Fig. 8. Each two adjacent ele- The AEM is an innovative modeling method adopting ments can be completely separated once the springs the concept of discrete cracking. In AEM, structures are connecting them are ruptured. modeled with elements assembly as shown in Fig. 8. Fully nonlinear path-dependant constitutive models The elements are connected together along their sur- are adopted in the AEM as shown in Fig. 8. For con- faces through a set of normal and shear springs. Those crete in compression, elasto-plastic and fracture model

Fig. 8 Modeling of a structure with the AEM. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 391

C s on Contact p ta rin c normal spring g t S h ea r r a e sh ct ta n g o n C ri sp

Shear spring in y Shear spring in X Normal Spring (a) Corner-to-face or corner-to-ground contact (b) Edge-to–edge contact Fig. 9 Different types of elements contact. is adopted (Maekawa and Okamura 1983). When con- each other if the matrix springs connecting them are crete is subjected to tension, linear stress-strain relation- ruptured. Elements may automatically separate, re- ship is adopted until cracking, where the stresses drop to contact or contact other elements. Figure 9 illustrates zero. Since the method adopts discrete crack approach, the different types of element contact, where contact the reinforcing bars are modeled as bare bars for the springs are generated at contact points of elements. In envelope (Okamura and Maekawa 1991) while the this study, the Extreme Loading for Structures (ELS) model of Ristic et al. (1986) is used for the interior software (www.appliedscienceint.com), which is based loops. on the AEM, is used. An interface material model is used for modeling The AEM was proven to be capable of following the bearings. The interface material model is a pre-cracked deformations of a structure subjected to extreme loads model where the material is initially cracked and can to its total collapse (Sasani and Sagiroglu 2008; Sasani not carry tensile stresses. As for compression, the stress- 2008; Park et al. 2009; Wibowo et al. 2009; Galal and strain relation is linear up to compression failure stress ElSawy 2010; Salem et al. 2011; Helmy et al. 2009; as shown in Fig. 8. The relationship between shear Helmy et al. 2012; Helmy et al. 2013; Sasani 2008; Sa- stress and shear strain is assumed linear till the shear lem 2011; Salem and Helmy 2014 ). Therefore, and stress exceeds μ σn (coefficient of friction times normal since the goal of the current study is to investigate the stress) where the shear stress remains with this value (μ collapsing behavior of Utatsu Ohashi bridge under se- σn) as long as normal stresses are not changed. Again, vere loads resulting from tsunami action, it was decided increasing the compressive stresses will lead to an in- that the AEM would be the most appropriate numerical crease in shear stresses again till shear stresses reach tool for such investigation. Although the Finite Element (μσn). The shear stiffness is set as minimum if the crack Method (FEM) is a robust and well established struc- is open or during sliding. tural analysis method, it is not the optimum solution for The AEM is a stiffness-based method, in which an the scope of progressive collapse analysis. Many draw- overall stiffness matrix is formulated and the equilib- backs are associated with the FEM progressive collapse rium equations including each of stiffness, mass and analysis. Hartmann et al. (2008) showed that the com- damping matrices are nonlinearly solved for the struc- putations associated with the simulation of collapses of tural deformations (displacements and rotations). The real world structures based on conventional FEM are solution for equilibrium equations is an implicit one that very costly, and therefore followed another approach adopts a dynamic step-by-step integration (Newmark- based on multibody models. beta time integration procedure) (Bathe, 1982 and Cho- pra, 1994). 4. Bridge analytical model In the AEM, two adjacent elements can separate from 4.1 Structural model The bridge was modeled using Extreme Loading for Structures software (ELS, www.appliedscienceint.com). The model included concrete and reinforcement details of the bridge superstructure, i.e, slabs, girders and piers. All bearings, concrete restrainers and steel restrainers for lateral displacement were explicitly included in the model. The substructure was not modeled and the piers were assumed totally fixed to their foundations. This

Perspective View assumption is not far from reality because the deep foundations of the bridge insured high rigidity and also because no soil scouring was recognized or reported. Figure 10 shows the whole modeled bridge, while Fig. Top View 11 shows individual models of its constituent elements. Fig. 10 Full AEM model of Utatsu Ohashi bridge. A mesh sensitivity analysis was carried out for differ- H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 392

(a) Model of S1 and S2 (b) Mesh discretization for S1, S2 and Pier P1

(e) Model of Abutment A1 (f) Model of Pier P2

(g) Model of Abutment A2 (h) Model of Pier P7 Fig. 11 AEM model of different parts of Utatsu Ohashi bridge. ent structural components. For example, Fig. 12 shows in compression but behave linearly. the mesh sensitivity analysis carried out for deck S10. A lateral static pressure of 3 kN/m2 was applied to the side 4.3 Main assumptions faces of the girders and slabs and the mesh size sensitiv- The followings are main assumptions adopted in the ity was investigated. It was found that a mesh with 1500 analysis; elements was good enough to get a convergence for the 1. Due to lack of some structural Data, the real shape of values of deformations of the model. A total of 21,430 the bearings was not known to the authors. Therefore, elements were used for the whole bridge model. the bearings were modeled as shown in Fig. 11(g), where the cylinders were constrained with the bottom 4.2 Material properties bearing plate in hinged support while unconstrained Table 1 shows the material properties adopted in the in the roller support to allow for cylinder rotation, AEM analysis. The bearing interface is given a rela- and hence allow for the longitudinal motion of the tively high compressive strength so that it does not fail bridge deck. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 393

Table 1 Material properties Young's modulus Compressive strength Tensile strength Yield stress Ultimate strength Material (MPa) (MPa) (MPa) (MPa) (MPa) Concrete 26715 30 2 ------Steel and Reinforcing 2.03E5 ------360 504 bars Bearing Interface 2.03E5 1500 0 ------

4 4.4 Tsunami loads acting on the bridge 3.5 Tsunami loads considered to be acting on the bridge are the drag forces (hydrodynamic forces) and the buoyant 3 forces (uplift forces). Surge (impulsive) forces are of (mm)

2.5 low effect because they are caused by the leading edge

2 of the water surge and at that stage the water height is low and have little effect on the superstructure. Displacement 1.5

Lateral 1 4.5 Hydrodynamic force 2 2 = /3 mkNp = /3 mkNp Hydrodynamic (drag) forces act when water flows 0.5 around the bridge. They include frontal impact on the 0 upstream face, drag along the sides, and suction on the 300 500 700 900 1100 1300 1500 1700 downstream side. These forces are induced by the flow Number of elements Fig. 12 AEM mesh sensitivity analysis for S10. of water moving at moderate to high velocity, and are a function of fluid density, flow velocity and structure 2. The non-structural handrails in bays S3~S7 are mod- geometry. Hydrodynamic forces can be computed as eled to take into account the water pressure applied to follows (JSCE, 2007 and FEMA ,2008): those handrails since those pressures are eventually 1 2 transferred to the superstructure Fdsd= ρ CAV (1) 3. Water velocity is adopted from calculations of Li et al. 2 (2013). Figure 13 shows water velocity, water direc- where tion, and water height for the whole bays of the Fd = Horizontal drag (hydrodynamic) force bridge. Cd = Drag coefficient ≅ 2.0 (FEMA ,2008)

Inundation Velocity (m/s) Section 1 (39 min.) height (m)

max height Section 2

Section 1

Time (min.)

Water Direction Water velocity Section 1 Section 2 Water Direction (m/s)

Time (min.)

Max velocity in Max velocity in Section 1 Section 2 Water Velocity Water Velocity Fig. 13 Water velocity, direction and height as determined by Li et al. (2013). H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 394

Fig. 14 Iterative Procedure for calculations of hydrodynamic water pressure.

25 0.00 velocity velocity given by Li et al (2013). From the AEM analy- lateral pressure 20 -0.20 sis, the velocity of the bridge deck is obtained, and ) 2 hence the relative velocity between the water flow and 15 -0.40 /sec)

m Simplified pressure the bridge deck is recalculated and the hydrodynamic for analysis trial (2) 10 -0.60 pressure is also recalculated. With the new calculated hydrodynamic pressure, the AEM analysis is carried out Velocity obtained 5 -0.80 again and the velocity of the moving deck is recalcu- Deck velocity (

from analysis trial (1) (t/m Lateral Pressure lated. These steps are repeated till reaching a non- 0 -1.00 0 5 10 15 20 25 changeable pressure values. Figure 14 shows the itera- tion scheme for such calculations, while Fig. 15 shows a -5 -1.20 Time (sec) sample for pressure calculation for bays S3-S7. Table 2 Fig. 15 A sample for pressure simplified calculations for shows values of lateral hydrodynamic pressure calcu- bays S3-S7. lated for different bays in the bridge.

ρs = density of sea water including sediments (= 1.2 4.6 Buoyant force density of sea water) The buoyant force’s magnitude equals to the weight of V= velocity of water the volume of water displaced by the submerged body. A= Area of structural member normal to flow Buoyant forces are calculated as follows: It is herein important to understand that this equation F = ρ gV (2) assumes that the object around which water flows is a bs D stationary object and the water flows around with a ve- where locity “V”. In the case of Utatsu Ohashi bridge analysis, F = Buoyant Force the bridge deck could be stationary in the beginning of b V = volume of water displaced by the submerged ob- the tsunami attack but when it starts to slide and move, D ject (bridge deck) it will be a moving object. Therefore, the velocity “V” ρ = density of sea water including sediments (= 1.2 would be the relative velocity between the water and the s density of sea water) bridge deck. Since the ELS is pure structural analysis g= gravitational acceleration software and has no coupled fluid-structural dynamics The buoyant force was considered in the model by solver, a simplified iterative method has been carried reducing the unit weight of the bridge concrete by a out to consider the correct pressure acting on the bridge. magnitude equals the unit weight of the sea water. First of all, the pressure is calculated based on the water Bricker et al. (2012) carried out two-dimensional computational fluid dynamic analysis to the deck of Table 2 Hydrodynamic pressure for different bays. Utatsu Ohashi bridge and found out that, air is trapped Bay Number S1~S2 S3~S7 S8~S10 S11~S12 between girders during motion of water as shown in Fig. Hydrodynamic Pressure 16. This trapped air causes additional buoyant forces 2 1.8 9.7 12.5 1.9 (kN/m ) that are essential to be considered in the analysis. The H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 395 magnitude of those additional buoyant forces equals to the weight of the volume of water displaced by the trapped air and is calculated as follows:

FbT= ρ sgV AT (3) where FbT = Buoyant Force due to trapped air VAT= volume of water displace by the trapped air Fig. 16 Trapped air between Utatsu Ohashi bridge gird- ρs = density of sea water including sediments (= 1.2 ers; CFD calculations by Bricker et al. (2012). density of sea water) g= gravitational acceleration These forces are considered by being directly applied to the deck slab in the upward vertical direction as shown in Fig. 17(a). For the cases where bridge decks are overturned, analysis is repeated considering those buoyant forces to reduce to zero when the deck rotation h reaches 90 degrees. It is believed that at this stage the trapped air would be released.

5. Result of AEM analysis Trapped Air Additional Uplift Trapped air height = Due to Trapped Air 0, 0.5 h, 0.75 h, 1.0 h 5.1 Results neglecting trapped air (a) Additional uplift due to (b) Investigated cases for When neglecting trapped air between girders, the nu- trapped air trapped air merical analysis did not show any failure in the bridge Fig. 17 Study cases for assumed trapped air between deck as shown in Fig. 18(a). These results reflect the girders.

(a) Trapped air height = 0 and 0.5 h

(b) Trapped air height = 0.75 h

(c) Trapped air height = h Fig. 18 Effect of amount of trapped air between girders on Utatsu bridge behavior under the effect of tsunami loads. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 396 importance of including the effect of the trapped air in 800 analysis which will add more buoyant forces on the 700 bridge deck, and thus, reducing the overall deck weight and leading to possible wash away of the deck as ob- 600

served in reality. (kN) 500

400 Force 5.2 Effect of the amount of trapped air In addition to the case of no trapped air, three cases with 300 different amounts of trapped air were studied as shown Lateral Trapped air height= 100% H 200 in Fig. 17(b). The height of trapped air for the three Trapped air height= 75% H cases was 50%, 75% and 100% of the clear depth of the 100 Trapped air height= 0& 50%H girder, respectively. The results of analysis for the four 0 cases are shown in Fig. 18. As seen in Fig. 18(a), when 00.511.52 the height of the trapped air is 50% of the clear depth of Deck Lateral Movement (m) the girder, no collapse was observed for the bridge (a) S10 decks. On the other hand, for the cases with trapped air 450 depth of 75% and 100% of the clear depth of the girder, 400 collapse of the bridge decks took place with higher level of outflow distances with the 100% case. This is ex- 350 plained by the effect of the additional buoyant force due 300 (kN) to trapped air on reducing the overall deck weight, thus 250 facilitating the mission of the wave pressures to wash Force away the bridge decks. 200

Figure 19 illustrates the effect of amount of trapped 150 Lateral air between girders on the stability of decks S10 and S5. Trapped air height= 100% H The total applied lateral loads due to tsunami for S10 100 Trapped air height=75% H and S5 are 690 kN and 379 kN, respectively. As seen in 50 Trapped air height=0& 50%H Fig. 19(a), the deck S10 remained stable with the appli- 0 cation of the full lateral load for the cases of trapped air 0 0.2 0.4 0.6 height of 0%, 50% and 75% of the girder clear depth. Deck Lateral Movement (m) However, for the case of trapped air height of 100% of (b) S5 the girder clear depth, the deck started to lose its stabil- Fig. 19 Effect of amount of trapped air between girders ity and was washed away at a lateral force of 630 kN. on deck lateral movement. On the other hand, for deck S5, it remained stable with the application of the full lateral load for the cases of tional fluid dynamics analysis for accurate investigation. trapped air height of 0% and 50% of the girder clear In the current study, a simple method is proposed and depth but lost its stability and was washed away at al- evaluated through numerical results. The amount of most full lateral load for the cases of trapped air height trapped air is assumed directly proportional to water of 75 % and 100% of the girder clear depth. wave speed so that the height of trapped air equals 100% of the clear depth of the girder for the cases of 5.3 A simplified method for amount of trapped higher water speed (> 3.13 m/sec) as shown in Fig. 20. air Table 3 shows values of uplift pressure calculated based None of the collapses obtained by the AEM for different on this assumption. This assumption was found to give a amounts of trapped is similar to the real collapse of the reasonable agreement with the field observation as bridge. In real collapse, decks S3 to S10 were washed shown in Fig. 21. Decks S1, S2, S11 and S12 did not away, while decks S1, S2, S11, and S12 were not. The collapse, while remaining decks were washed away reason that the analysis is not matching the reality could matching real collapse. However, collapsed decks did be that the amounts of trapped air in different bays are not all collapse in the same way as in real collapse. For not necessarily the same fraction of the clear depths of example, decks S4 and S5 were separated from each the girders. They might differ according to deck geome- other. Both decks were washed away in a sliding man- try, surrounding topography and water wave speed. It is ner similar to real collapse, nevertheless, deck S5 over- a rather complicated issue to estimate the amount of the turned after hitting the ground. trapped air and requires a three dimensional computa- Figures 22 and 23 show the obtained sliding and

Table 3 Uplift pressure resulting from trapped air for different bays. Bay Number S1~S2 S3~S7 S8~S10 S11~S12 Uplift Pressure (kN/m2) 7 7 15 5.2 H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 397

1.2 of the deck terminates. It is important to point out that the current simplified model is just an approximation to 1 solve the rather complicated three-dimensional dynamic structure-water interaction problem. A detailed investigation would be the best solution where computational 0.8 fluid dynamics should be coupled with nonlinear structural dynamic scheme of computations. 0.6 6. Examples for application of the 0.4 proposed method

In this section, three different solution examples for 0.2 Trapped air height/girder clear depth mitigating collapse of existing similar coastal bridges due to tsunamis are proposed. Those examples are ap- 0 plied separately to the bridge as follows; 0123 4 Water Speed (m/sec) 6.1 Example (1) Fig. 20 Proposed amount of trapped air between girders. In this example, making several small punch-outs in the deck slab is proposed for allowing air to run away overturning collapses of spans S7 and S10, respectively, through them during tsunami and hence reduce a lot the which seems reasonable and agrees with real collapse. uplift pressure resulting from trapped air as shown in Figures 24 and 25 show the time history for the deck Fig. 26. According to the numerical results shown in movement and speed in the direction of water waves for Fig. 18(a), if no trapped air was there, the bridge would decks S7 and S10, respectively. On both figures, the not collapse. The punch-outs could be with small size so points at which the deck speed reaches the water speed they do not affect the vehicles motion on the deck slab. are shown. At these points, the hydrodynamic pressure This solution may require huge construction effort and it is reduced to zero. Figures 24 and 25 shows also the is actually more convenient for newly constructed points at which the deck hits the ground. At those points, bridges. the speed of the deck decreases to zero and the motion

(a) Bridge collapse (Plan) (b) AEM results (Plan)

(c) AEM results (Isometric view) Fig. 21 Comparison of AEM results and real collapse of bridge. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 398

(a) t=13.18 sec. (b) t=14.79 sec.

(e) t=15.95 sec. (f) t= 16.2 sec. Fig. 22 Sliding collapse of deck S7.

(a) t=10.5 sec. (b) t=12.7 sec.

(e) t=14.08 sec. (f) t=14.5 sec.

Fig. 23 Overturning collapse of deck S10. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 399

45 40 35 Start of reduction of (m) 30 hydrodynamic pressure due 25 to reaching water speed 20 15 10 Start of impact Movement 5 with the ground 0 0 5 10 15 20 25 30 35 Time (sec) (a) Deck movement (m) - time relationship in direction of water waves 18 16 Start of impact 14 with the ground 12 Start of reduction of 10 hydrodynamic pressure due (m/sec) 8 to reaching water speed 6

Speed 4 2 0 0 5 10 15 20 25 30 35 Time (sec) (b) Deck speed (m/sec) - time relationship in direction of water waves Fig. 24 Movement and speed of deck S7.

35 30 Start of reduction of (m) 25 hydrodynamic pressure due 20 to reaching water speed Start of impact 15 with the ground 10

Movement 5 0 0 5 10 15 20 25 30 35 Time (sec) (a) Deck movement (m) - time relationship in direction of water waves 10 Start of impact 8 Start of reduction of hydrodynamic pressure due with the ground 6 to reaching water speed 4 (m/sec) 2 0 Speed 0 5 10 15 20 25 30 35 ‐2 ‐4 (b) Deck speed (m/sec) - time relationship in direction of water waves Fig. 25 Movement and speed of deck S10. H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 400

Fig. 28 Proposal (3): Use of U- stopper.

Fig. 26 Proposal (1): Punch-outs in the deck slab

Cables

(a) U-stopper with a cross section of 300 x 300 mm

Fig. 27 Proposal (2): Use of Anchor cables.

6.2 Example (2) The idea of this example is to resist sliding and overturning of deck slabs by means of cables firmly an- chored to the ground as shown in Fig. 27. Each deck end is connected by a steel cable inclined @ 45 degrees to the ground through a footing with properly design (b) U-stopper with a cross section of 450 x 450 mm tension piles. Through trial analysis, this solution proved to be successful if cables are designed to carry 250 kN tensile force. It is obvious this proposal would be relatively a costly one because of the construction of tension piles.

6.3 Example (3) In this example, a reinforced concrete stopper with a U- shape is added to the tip of the pier girder as shown in Fig. 28. The function of the U-stopper is to arrest the deck motion during tsunami both horizontally and verti- cally. Three sizes of the section of the U-stopper were (c) U-stopper with a cross section of 600 x 600 mm investigated; 300 x 300 mm, 450 x 450 mm and 600x Fig. 29 Behavior of bridge strengthened with U-stopper 600 mm. All had an arbitrarily longitudinal reinforce- of different sizes. ment ratio of 1%. As shown in Fig. 27, the U-stopper of 300 and 400 mm sizes collapsed and could not prevent the deck sliding or overturning. On the other hand, the 7. Conclusions 600 mm U-stopper could efficiently stop the collapse of the deck slab. This solution is suitable for existing Based on numerical investigations using the Applied bridges as well as newly constructed ones and its cost is Element Method for the Utatsu Ohashi bridge under the relatively low. Tohuku tsunami loads, 2011, the following conclusions could be drawn; 1) Applied Element Method was successfully used to simulate the Utatsu Ohashi bridge collapse H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 401

2) The amount of trapped air between deck girders dur- Code.” Journal of performance of constructed ing tsunami has a significant effect on the behavior of facilities, ASCE, 27(5), 529-539. the bridge due to the buoyant force accompanied JSCE, (2007). “Standard specifications for concrete with the trapped air structures -Design.” No. 15. 3) Assumption that trapped air amount is proportional to Kawashima, K., Kosa, K., Takabashi, Y., Akiyama, M., water wave speed was found to give reasonable and Nishioka, T., Watanabe, G., Koga, H. and Matsuzaki, close-to-reality results for the bridge collapse. How- H., (2012). “Damage of bridges during 2011 Great ever, a detailed future investigation need to be con- East Japan Earthquake.” Retrieved July, sidered using detailed three dimensional coupled http://www.nehrp.gov/pdf/UJNR-4217.pdf CFD-structural analysis Li, F., Kosa, K., Nakano, A. and Sasaki, T., (2013). 4) Three solution examples for mitigating collapse of “Detailed analysis of damages in Utatsu area by existing similar bridges were introduced and applied Tsunami.” Proceedings of Annual Conference of JCI, to Utatsu Ohashi bridge case and found to be effi- 35(2), 799-804 cient for preventing collapse Maekawa, K. and Okamura, H., (1983). “The deformational behavior and constitutive equation of References concrete using the elasto-plastic and fracture model.” Applied Science International, LLC (ASI) Journal of the Faculty of Engineering, The University www.appliedscienceint.com of Tokyo (B),1983, 37(2), 253-328. Bathe, K., (1995). Solution of equilibrium equations in Meguro, K. and Tagel-Din, H., (2000). “Applied dynamic analysis. Prentice Hall, Englewoods Cliffs, element method for structural analysis: theory and N.J. application for linear materials.” Structural Bricker, J., Kawashima, K. and Nakayama, A., (2012). Eng./Earthquake Eng., International Journal of the “CFD analysis of bridge deck failure due to tsunami.” Japan Society of Civil Engineers (JSCE), 17(1), 21- Proceedings of the International Symposium on 35. Engineering Lessons Learned from the 2011 Great Meguro, K. and Tagel-Din, H., (2001). “Applied East Japan Earthquake, March 1-4, 2012, Tokyo, element simulation of RC structures under cyclic Japan loading.” ASCE, 127(11), 1295-1305. Chopra, A., (1995). “Dynamics of structures: Theory Meguro, K. and Tagel-Din, H., (2003). “AEM used for and applications to earthquake engineering.” large displacement structure analysis.” Journal of Prentice Hall, Englewoods Cliffs, N.J. Natural Disaster Science, 24(1), 25-34. FEMA, (2008). “Guidelines for design of structures for Okamura, H. and Maekawa, K., (1991). “Nonlinear vertical evacuation from tsunamis, FEMA P646 analysis and constitutive models of reinforced Report, prepared by the Applied Technology Council concrete.” Gihodo, Tokyo. for the Federal Emergency Management Agency, Park, H., Suk., C. and Kim, S., (2009). “Collapse 2008, Redwood City, California. modeling of model RC structures using the Applied Galal, K. and El-Sawy, T., (2010). “Effect of retrofit Element Method.” Journal of Korean Society for Roc strategies on mitigating progressive collapse of steel Mechanics, Tunnel& Underground Space, 19 (1), 43- frame structures.” Journal of Constructional Steel 51. Research, 66(4), 520-531. Ristic, D., Yamada, Y. and Iemura, H., (1986). “Stress- Hartmann, D., Breidt, M., Nguyen, V., Stangenberg, F., Strain Based Modeling of Hysteretic Structures under Hohler, S., Schweizerhof, K., Mattern, S., Earthquake Induced Bending and Varying Axial Blankenhorn, G., Moller, B. and Liebscher, M., Loads.” Research report No. 86-ST-01, School of (2008). “Structural collapse simulation under Civil Engineering, Kyoto University, Kyoto, Japan. consideration of uncertainty -fundamental concept Salem H., (2011). “Computer-aided design of framed and results.” Computers and Structures, 86, 2064- reinforced concrete structures subjected to flood 2078. scouring.” Journal of American Science, 7(10), 191- Helmy H., Salem H. and Tageldin H., (2009). 200. “Numerical simulation of Charlotte Coliseum Salem, H., El-Fouly, A. and Tagel-Din, H., (2011). demolition using the Applied Element Method.” “Toward an economic design of reinforced concrete USNCCM-10 conference-Ohio-USA. structures against progressive collapse.” Engineering Helmy, H., Salem, H. and Mourad, S., (2012). Structures, 33, 3341-3350. “Progressive collapse assessment of framed Salem, H. and Helmy, H., (2014). “Numerical reinforced concrete structures according to UFC investigation of collapse of the Minnesota I-35W guidelines for alternative path method.” Engineering bridge.” Engineering Structures, 59, 635-645 Structures, 42, 127-141. Sasani, M. and Sagiroglu, S., (2008). “Progressive Helmy, H., Salem, H. and Mourad, S., (2013). collapse resistance of hotel San Diego.” Journal of “Computer-aided assessment of progressive collapse Structural Engineering, 134(3), 478-488. of reinforced concrete structures according to GSA Sasani, M., (2008). “Response of a reinforced concrete H. Salem, S. Mohssen, K. Kosa and A. Hosoda / Journal of Advanced Concrete Technology Vol. 12, 388-402, 2014 402

infilled-frame structure to removal of two adjacent Society of Civil Engineers (JSCE), 17(2), 215-224. columns.” Engineering Structures, 30, 2478-2491. Tagel-Din, H., (2002). “Collision of structures during Tagel-Din, H. and Meguro, K., (2000). “Applied earthquakes.” Proceedings of the 12th European Element Method for simulation of nonlinear Conference on Earthquake Engineering, London, UK. materials: theory and application for RC structures.” Tagel-Din, H. and Rahman, N., (2004). “Extreme Structural Eng./Earthquake Eng., International loading: breaks through finite element barriers.” Journal of the Japan Society of Civil Engineers Structural Engineer, 5(6), 32-34. (JSCE), 17(2), 123-148. Wibowo, H., Reshotkina, S. and Lau, D., (2009). Tagel-Din, H. and Meguro, K., (2000). “Applied “Modelling progressive collapse of RC bridges Element Method for dynamic large deformation during earthquakes.” CSCE Annual General analysis of structures.” Structural Engineering and Conference, GC-176-1-11. Earthquake Engineering. Journal of the Japan