Experimental Study for Investigating the Impact Force on a Wooden House by Driftwood in Steady and Unsteady Surge-type Flow K. Miyahara 1 and N. Tanaka 2 1 Graduate School of Science and Engineering, Saitama University 255 Shimo-okubo, Sakura-ku, Saitama, Saitama, JAPAN 2 International Institute for Resilient Society, Saitama University 255 Shimo-okubo, Sakura-ku, Saitama, Saitama, JAPAN E-mail: tanaka01@mail.saitama-u.ac.jp Abstract: The objective of this study is to clarify the change of impact force and the duration in steady and unsteady flow when driftwood collides with houses. Flume experiments were conducted in steady and unsteady flow conditions, created by sudden opening of gate. Impact force and its duration were measured by using a force gauge. The current study helped to elucidate the behavior of impact duration for both steady and unsteady flow. In steady flow, just in front of the house model, the backwater effect reduced the velocity of driftwood by acting as a buffer zone. In contrast, the reduction effect was not marked in unsteady and supercritical flow, because the water splashed upward after hitting the house model. This study indicates the possibility that the magnitude of the driftwood impact force could change depending on the flow structure around houses before driftwood hit. Keywords: driftwood, collision, impact duration. 1. INTRODUCTION The driftwoods produced from landslide or water-borne disaster like tsunamis or floods, sometimes cause a catastrophic damage to structures like buildings or bridges. In Izu-ohshima Island, large debris flow which included large amount of driftwoods occurred by the 2013 historical heavy rain, destroyed and washed out houses on the foot of the mountain. The estimation of debris impact force and the countermeasure are of great importance. Most of formulas defining the impact force by the collision of floating debris are derived from experiments (Matsutomi 1999; Haehnel & Daly 2004) considering tsunamis and floods. However it is very difficult to apply it for debris flow because flow structure is quite different. A few research has been done on the numerical simulation of driftwood motion apart from representative studies by Goto et al (2001) and Nakagawa et al (1992). These studies analyzed driftwood movements by the Lagrangian approach. However, there is no research which analyzes the process of driftwood production, driftwood transportation, and the collision that damage houses. Moreover, it is much important to figure out the duration with accuracy in computing the impact force by using the equation of impulse-momentum approach. In the approach, impact duration is a very important parameter, but other than tsunami cases little is known for strongly nonlinear and unsteady flow. Tsunami flow is mostly considered as unsteady. However, for a very long wave around the point of maximum fluid force it is sometimes treated as steady. So the knowledge on the duration has some possibilities not to be applied for a strongly unsteady flow. Therefore, fundamental flume experiment was conducted in order to clarify the differences about the impact force and duration between steady and unsteady flow. 101
2. MATERIAL AND METHODS 2.1. Modeling Impact Force of Driftwoods The City and Country of Honolulu uilding Code (CCH) (2000) recommended the impact duration as 1.0 s for wood construction, and the U.S. Federal Emergency Management Administration (FEMA) (2005) recommended value of impact duration as 0.7 1.1 s for the collision of woody debris to wood. Although the longer impact duration is safer because the impact force is adversely correlated to the impact duration, the characteristics of impact duration of driftwood have not been elucidated yet. ecause the timing of driftwood generation from a forest depends on the fluid force and forest condition (i.e. tree diameter, substrate condition), the impact force of driftwood also changed with the location of houses. Therefore, this study needed to develop a numerical simulation model of tsunami and driftwood motion in which the impulse-momentum approach was applied. The resultant force (F xm, F ym ) acting for an infinitesimal impact duration (Δt) is equal to the change in liner momentum : mv x = F xm Δt, mv y = F ym Δt (1) where I x, I y are impulses in x, y directions, and V x, V y are the simulated velocities of the driftwood at collision in x, y directions, respectively. 2.2. Experiments to Estimate Duration of Impact between Driftwoods and House Model For the impulse-momentum approach, it is quite important to estimate Δt correctly for the collision of driftwoods and houses. This study conducted flume experiments and estimated Δt. Table 1 shows experimental conditions. For steady flow conditions, all the experiments were conducted in a water flume (constant bed slope 1/375), which is 5m in length and 0.7m in width. The Froude number (Fr =V/(gh) 0.5, where, V = the depth-averaged velocity (m/s), g = gravitational acceleration (m/s 2 ), h = water depth(m)) was set approximately 1.1 and the corresponding water depth was 4.5cm. In unsteady condition, a flume with a channel width of 0.4 m was used and the bed slope was set at 1/25 and 3/1000. The water depths were set at 1.1 cm and around 1.3 2.8 cm against each slope, respectively. Fr was set approximately 2.1 and 1.3 for the slope of 1/25 and 3/1000, respectively. In addition, flat slope experiments were also conducted (Fr = 1.5). From these experiments, the change of impact duration was tested by the effects of Fr and the timing at which driftwoods collides on houses. A rectangular-shaped wooden house model which was 5, 5, and 10 cm in stream-wise length, crossstream width, and height, respectively, was set in a flume. These were the maximum possible dimensions kept considering that the reflection from the sidewall should not affect the driftwood motion in front of the house model. After washing out, pieces of driftwood are assumed to mutually interact and combine with each other followed by collision with the buildings. This study neglected the combining process of driftwood but used a group model of driftwood since released to float. The group model of driftwood was made by connecting cylinders (diameter = 0.4 cm and length = 12.5 cm) using thread. The spacing of each cylinder was 0.5 cm and the number of cylinders were 5. In steady flow conditions, before release, a driftwood model was positioned at the center of the channel and 1.5 m upstream from the house model location. On the other hand, in unsteady conditions, driftwood model was released in two positions, one parallel to flow (30cm U/S of house model) and the other perpendicular to flow (10cm U/S of house model). Moreover, the throw timing 102
was delayed when a driftwood model placed in some cases. Impact duration and force acting on the house model were directly measured by a two-component force gauge (SSK Co., model L60-1N, measuring time: 100 s, measuring time interval: 100 Hz). For measurement of the drag force, gap is required between the channel bottom and the house model. The experiments were conducted with the smallest possible gap, 0.001 m, similar to previous studies (Takemura & Tanaka 2007; Iimura & Tanaka 2012). Table 1 Experimental conditions Case Number Flow condition Inclination Tank water depth Way to place a driftwood model 1 steady i=0.0027 A 2 5cm 3 C i=0.003 4 10cm 5 C 6 5cm 7 C unsteady i=0.04 8 10cm 9 C Length from front edge of a wave to driftwood model placed 0cm 10 10cm 11 C 20cm 12 i=0 10cm 40cm 13 D 0cm A: To place a driftwood model at the center of the channel and 1.5 m upstream from the location of a house model. : To place a driftwood model in parallel to the flow at 10 cm upstream from the location of a house model. C: To place a driftwood model in parallel to the flow at 30 cm upstream from the location of a house model. D: To place a driftwood model in cross-stream direction at 10 cm upstream. 103
3. RESULTS AND DISCUSSION 3.1. Experimental Results of Impact Force Figure 1 shows the relationship between the impact force measured by a force gauge and the force calculated by using mass, driftwood velocity and impact duration. The measured impact force was proportional to the mass, indicating the applicability of the impulse momentum approach. The measured values are almost equal to the calculated values. This indicates that the force and impact duration measured by the experiment in this study and by definition are same and hence correct. Measured Force (N) 2 1 0 0 1 2 Calculated Force (N) i=0.0027 (steady condition) i=0.04 (unsteady condition) i=0.003 (unsteady condition) additional experiment (unsteady) Figure 1 Comparison of the calculated and measured impact force on a house model 3.2. Differences about the Duration between Steady and Unsteady Flow The knowledge on the debris impact duration is quite important for the numerical analysis on the mudflow which contains lots of driftwoods. Figure 2 indicates that the impact duration is not affected greatly with change in Froude number. y contrast, the duration in unsteady flow was approximately a quarters to a tenths as short as that in steady flow with same Froude number. This difference was caused by the splashed flow which rose a driftwood model just when it collided on a house model in unsteady flow. 104
Duration of impact (s) 0.04 0.02 0.00 i=0.003 (unsteady condition) i=0.04 (unsteady condition) i=0.0027 (steady condition) additional experiment (unsteady) 0 1 2 3 Froude number Figure 2 Relationship between the duration of impact and Froude number 3.3. Comparison of the Effect of ackwater in Steady and Unsteady Flow Some differences have been clarified about the impact force between steady and unsteady flow. In steady flow, the backwater effect was observed on the upstream of house model. This backwater acted as a buffer zone and it reduced the driftwood velocity. y contrast, although the backwater effect formulated well in unsteady flow, however, the flow structure was different from that in steady flow. A thin layer of backwater was generated; not upstream but upward along the wall of house model. Moreover, velocity reduction did not have a major effect and driftwood model was raised by the splashed water. Therefore, in unsteady flow, the impact duration was shorter and impact force was larger than that in steady flow. Furthermore, in some cases, it was difficult to measure impact force using force gauge because the water splashed upward fell down and the water surface oscillated quite intensively when driftwood models collided on house model. 3.4. Effect of the Timing when Driftwood Collide with Houses Figure 3a shows the time series of force acting on a house model when the unsteady surge with a driftwood model collided with a house model almost at the same time. In this case, the effect of the reduced speed of driftwood in the backwater region was not significant. Therefore, the peak value of the force was just at the moment of collision. In that case, the impact force is additional to the fluid force and hence increases the total force. On the other hand, Figure 3b shows the time series of force when driftwood collided with a house model after the surge collided. The impact force was affected by the backwater effect, compared with Figure 3a. There was no apparent peak when the driftwood collided. It suggests that the water splashed upward fell down in front of a house model while driftwoods collided. In that case, water cushion effect was supposed to act strongly. For that reason, the cases of which impact force was not apparent at the moment of collision are about 19 % of the total cases. 105
Measured Force (N) 1.4 1 0.6 0.2-0.2 0 1 2 3 Time (s) (a) Case 10 1.4 Measured Force (N) 1 0.6 0.2 Case 11_2 2.5 3.5 4.5 5.5-0.2 Time (s) (b) Figure 3 Time series of force acting on a house model in (a) Case 10 and (b) Case 11 4. CONCLUSIONS (i) (ii) In front of a house model, the backwater was observed both in steady and unsteady flow. However, the different flow structures caused a big difference on the impact duration and force between steady and unsteady flow. The impact duration in unsteady flow was approximately a quarters to a tenths as short as that in steady flow. It is very important to pay attention to the duration in the estimation of the impact force in unsteady and especially super critical flow where water is just splashed upward. 106
(iii) In unsteady surge type flow, the timing of collision did not greatly affect the magnitude of impact force as long as the collision of driftwood occurs with the surge front. Moreover, the peak value of total force (fluid force and impact force) appears when the driftwood model collided with the house model (approximately 81 %). For the rest, about 19 % cases have lower values under peak of fluid force. In that case, even when driftwood model collided on the house model, water cushion effect occurs by the splash motion in front of a house model. ACKNOWLEDGMENTS This study is funded by Research and Development Promotion Program on River and Sabo Field, Ministry of Land, Infrastructure, Transport and Tourism (MLIT). REFERENCES CCH, 2000, City and Country of Honolulu uilding Code, Department of Planning and Permitting of Honolulu Hawaii, Honolulu, Hawaii. FEMA, 2005, Coastal Construction Manual, FEMA 55 Report, Edition 3, Federal Emergency Management Agency, Washington, D.C. Goto, T., Sakai, T. & Hayashi, M., 2001, Lagrangian particle method for analysis of dam-up process by drift timbers (in Japanese). Annual Journal of Hydraulic Engineering, JSCE 45, 919-924. Haehnel, R.. & Daly, S.F., 2004, Maximum impact force of woody debris on floodplain structures Journal of Hydraulic Engineering, 130(2), 112-120. Limura, K. & Tanaka, N., 2012, Numerical simulation estimating the effects of tree density distribution in coastal forest on tsunami mitigation Ocean Engineering, 54, 223-232. Matsutomi, H., 1999, A practical formula for estimating impulsive force due to driftwoods and variation features of the impulsive force (in Japanese), Proceedings of the Japan Society of Civil Engineers, 621, 111-127. Nakagawa, H., Takahashi, T. & Ikeguchi, M., 1992, Numerical simulation of drift wood behavior. Annuals Disas, Prev. Res. Inst., Kyoto Univ., No. 35-2, 1-18. Takemura, T. & Tanaka, N., 2007, Flow structures and drag characteristics of a colony-type emergent roughness model mounted on a flat plate in uniform flow, Fluid Dynamics Research, 39, 694-710. 107