Experimental study of scouring process and flow behavior around a spur dike during the surge pass

Experimental study of scouring process and flow behavior around a spur dike during the surge pass Tomasz Marek Mioduszewski and Shiro Maeno Okayama University, Department of Civil Engineering Okayama, Japan ABSTRACT Detailed measurements of pressure and scouring depth were performed to improve understanding of the influence of varied phenomena on the riverbed around the spur dike. Two sets of hydraulic conditions, pressure-measuring points and scouring recorded during the surge combined together gave the image of how the liquefaction occurs around the structure. The liquefaction was found as a facilitating process during the scouring action. KEY WORDS: spur dike; flow measurement; hydrodynamic pressure; excess pore water pressure; scouring; liquefaction INTRODUCTION Spur dikes are seen as one of the best types of hydro-engineering structures for preserving the desired water depth or deflecting the main current in the harbor channel or river. This type of construction has been used for a long time in many cases in the river hydraulics as well as in the marine hydraulics, proving its usefulness and good features. In the sandy riverbeds, placement of every kind of structure results in scouring. This process poses a very serious danger for the stability of a structure. The primary target of the research is to develop better structures and improve understanding of the interaction between artificial structures and the natural environment. That is the reason why many researchers conducted experiments and analysis on flow model and scouring action around obstacles in water. Most of them focus on the scouring phenomenon (Sumer, 99, ; Elawady, ; Rahman, 99) the velocity profiles (Mia, ; Sumer, 99; Elawady, ) near the structures. The scouring phenomenon is very important because of its crucial point in a structure s stability. The main parameter influencing the scouring process is shear stress caused by velocity. Because the velocity profiles are complicated and characterized by 3-dimensional particle movement, many researchers dealt with this topic. Different configurations of structures like spur dikes, bridge abutments, piles etc. were investigated which improved understanding of the complicated water flow around single and groups of obstacles. On the other hand, many researchers investigated scouring near the structures under attack of waves. Some of them (Sumer, ; Mia, ) combined the scouring caused by waves with the scouring caused by current. This flow condition was investigated especially in a case of several coastline structures (Sumer, 99, ) and in the river hydraulics in a case of bridge piers (Mia, ). Only few researchers measured other factors influencing the scouring action and sand behavior during the water flow. To improve the understanding of other phenomena in the vicinity of the spur dike, the authors of this article took the liquefaction phenomenon in consideration. The liquefaction phenomenon, though assumed not to be the reason for the scouring process, may facilitate the process. What is more, this phenomenon is responsible for the reduction of the bearing capacity for the structures founded on the soil, where liquefaction occurs. Because the liquefaction phenomenon is most visible during the rapid flow, the surge type flow was investigated. This type of flow may occur in different cases. Most of them are like breakdown situations. One example may be the collapse of a weir (see Photo ), which may be the reason for the flow that is never expected upstream and downstream from the damaged structure. Different example of this type of flow in marine usage may be experienced while the storm surge or tsunami wave is coming to the harbor. Especially when harbor is located in the river s estuary, many spur-dike-like structures may be in danger. Another example of this type of flow happens when a heavy rain falls in the vicinity of an almost fully filled reservoir so that the operators of the dam have to release big amounts of water. The above situations are unusual and should be considered as emergency cases. It is a good purpose, therefore, to conduct investigations if one breakdown might be a reason for a chain reaction of other breakdowns. The wave, as a result of such a breakdown, may be observed both upstream and downstream from the damaged structure. The subject of Photo Damaged weir may cause the unpredictable flow Paper No. 3-JSC- Mioduszewski

SIDE VIEW 7 PLAIN VIEW Sand level 7 UW UW UW Upstream Head wall UH H UH H U UH H DH DH D DW DW DW3 DW 3 3 PROJECTION. Fig. Experimental setup this paper is to clarify the influence of a surge upstream from the damaged area on the spur dikes installed in the river. EXPERIMENTAL SETUP Experiments were conducted in the laboratory flume of Okayama University. The laboratory flume is a m long,. m wide and. m deep steel construction with its front wall made of glass. The channel inclination was /. The pit for experimental installation is located in the middle part of the flume (see Fig. ). The depth of the pit is 7 cm and its length is cm. Its width is equal to the channel width. During the experiment, the pit was filled with movable sand. The sand used for this experiment is characterized by an average grain diameter d =. mm. The spur dike was constructed as a full height structure, cm thick and cm wide. As shown in Fig. twenty pressure sensors were placed on it: eight on the upstream, four on the head wall and eight on the downstream. Pressure sensors were configured into five vertical columns and four horizontal rows. The highest level of the sensors during all the experiments was above the ground level. The second and third rows from the top in some cases emerge in the water during the surge while the lowest row of sensors was still in the sand. All pressure sensors used in the experiment were the transducer type. The sensors were connected to the amplifiers and to the computer to record the data. Sampling frequency was set to Hz. Measurement was performed at s with the first s as the stable flow time without initializing the surge. Those seconds at the time of data analysis were used in adjusting the time scale of all cases and in adjusting the zero value at the beginning of each case and data channel. In the graphs, this time delay is not shown to make them clear time zero equals opening the gate. The scour depth during the surge was recorded using underwater camera located near the dike. For purposes of recording, the surface of the walls Unit: cm Upstream was covered by the special foil with the chessboard-type pattern (see Photo ). It allowed a precise measurement of the scouring depth during movies postproduction process. HYDRAULIC CONDITIONS The experimental wave came from the downstream side of the dike. The surge was generated by opening the gate located at the end of the channel. Investigations were performed with two different initial water depths: cm and 3 cm. The depth of the water after the experiment was stabilized at the level of about cm. In both cases the water discharge was set to the value Q =. m 3 /s. The Froude number of the flow varied from. to.. Nomenclature of observation results depends on the initial water depth, and it is as follows: initial depth h = cm Case ; initial depth h = 3 cm Case. Head wall H UH UW Fig. Pressure sensor location PLAIN VIEW Because of the big number of pressure sensors, and the video recording using the foil, each experimental case was conducted several times. The obtained results were fitted with time scales to give a uniform image of the phenomena. For this purpose, each measurement was started with s of stable flow, after that time, the gate was opened, and the wave was initialized. The videos of scouring process were time adjusted using recoded simultaneously audio track where the time of opening the gate was recorded. DH DW Unit: cm EXPERIMENTAL RESULTS Pressure characteristics Photo Spur dike during scouring depth recording Pressure characteristics are presented in the Fig. 3 and Fig. for Case and respectively. All measurements (shown in all of the graphs presented in this paper) were conducted recording only the alteration of the pressure. At the beginning of each experiment, the pressure value for Paper No. 3-JSC- Mioduszewski

each sensor was set to zero. What one can see in the graphs and what is described in the text are only the pressure change in time without considering the value of the pressure at the time before experiment started. This type of measurement, though it calculates the excess water pressure, does not include the initial hydrostatic pressure. It presents only the difference in pressure. Using this method of calculation results in a considerably small error because comparison is done between two pressure values. Pressures, as said before, were recorded at s. The biggest changes in pressure were observed at the first s of the experiments. For the reason of good legibility, the time axis of the graphs was truncated to s. It is easy to see the very big value of oscillating pressure in each graph that registered just after beginning of the experiment. It is supposed to be the effect of the process of opening the gate. The time of these oscillations is about 3 seconds. This assumption is confirmed by the fact that, during these oscillations, the average value of the pressure was not changing. Other parameters (like water depth and velocity) also during this time remained unchanged. Fig. 3 shows the pressure measured at several points in the vertical column in Case. The general tendency of falling pressure level corresponds with the changing water level. Nevertheless, the differences in pressure change at different levels in the sand. What is obvious is that, in different columns, the pressure was changing in different ways. The pressures measured in the columns located on the upstream of the dike are changing less rapidly than the pressures recorded at the downstream. What is more, the pressures at all levels in the sand are changing almost simultaneously. The downstream side pressures look slightly different. Bigger alterations between particular values are also evident. The most rapid changes and the biggest fluctuations are observed in the H column located at the head wall of the spur dike. The difference of the pressure in this column is also the biggest. Like what has been assumed before, a kind of pressure change delay was observed in the area of rapid changes the pressure in the lower layers has changed more slowly than in upper layers. This phenomenon is a reason for the appearance of excess pore water pressure. Particularly in the column H the dissimilarity between points in the water and in the sand is visible. During the experiment the points H and emerge in the water. Before emerging, those points followed the same pattern as the others in the sand. But just after emerging, the change of the pressure value was very rapid. During the time of to 3 seconds the pressure value changed, and followed the pattern as points in the water. In other columns it was also visible. Case shows a good agreement with Case. Pressures measured in Case are shown in Fig.. Naturally, the pressure change is bigger in this case since the initial water level was also higher. The differences of pressure values in one column are larger as well, but the characteristic of the curves remains the same. All processes described above for Case took place in Case. The run of some of them was more rapid, and the time of occurrence was shorter than in Case. water pressure Observed excess pore water pressure for Case is presented in Fig.. Pressure sensors located above the sand level have measured the total pressure in a column namely the hydrostatic and hydrodynamic pressure. Because of the rapid changes in the pressure above the sand, the pressure measured in the sand showed some difference according to the limited velocity of water flow in the sand layer. The observed difference of pressures between values in the sand and above it is an excess pore water pressure. For each point in the sand, where pressure measurements were conducted, the excess pore water pressure was calculated as a difference between the pressure value at this point and at the point located in the water above the sand in the same column. Usage of this method of calculation results in a considerably small error, because the comparison is done DW3 DW - 3 3 - - DW DW DW DW DW3 DW DH D - 3 3 - DH DH DH - D H H - 3 3 - H H H - H UH U - 3 3 - - UH UH UH U UH UW UW - 3 3 - UW UW UW - UW Fig. 3 Pressure in vertical columns Case - 3 3 - - - - DH D DH - D - DH DH - 3 3 - H H - H H H H - 3 3 UH - UH - UH UH UH U - U UH - Fig. Pressure in vertical columns Case Paper No. 3-JSC- Mioduszewski 3

DH D D DH - 3 3 H H H H - 3 3 - between two pressure values located in the same column. The highest level of excess pore water pressure is in the column H. The biggest difference between pressures above the sand and in the layer of sand is in the column H (fig.3). The pressure measured in the point above the sand is changing most rapidly (Fig. 3 and Fig. ). Similar pattern is visible in the pressures recorded at the downstream of the spur dike. As a result, the excess pore water pressure in those columns has a big, positive value. It means that the pressure in the ground is bigger compared to the pressure in the water. A slightly different situation was recorded at the upstream. Because not only the change of the pressure is slower but also the hydrodynamic pressure is acting on this, the pressure in the water changed slower than in the sand. The difference is almost insignificant but the graph for the upstream shows a small negative value of excess pore water pressure (Fig. and Fig. ). In Case, the excess pore water pressure (see Fig. ) follows the same rules as in Case. However, the characteristic of the curves remains very similar, calculated excess pore water pressure is higher (about cm of water in the higher point) than for Case. The big difference is seen in the time of occurrence. The highest value of excess pore water pressure in Case is observed in the time of 3 s, while in Case, the highest value is observed in the time of s. Liquefaction UH U UH 3 3 UW UW UW - 3 Fig. pressure Case Usually the occurrence of excess pore water pressure is accompanied by the liquefaction process. Nago (9) on the experimental basis proposed Eq. that allows calculation of the dimensionless effective stress at specified depth in the sand according to the scouring depth and the excess pore water pressure. The dimensionless was calculated according to Eq.. In this article, was calculated for several points arrayed at the spur dike. Results are presented in Fig. 7 and Fig. for Case and respectively. In those graphs, the vertical line with an arrow shows the moment that point DH D DH D - 3 3 - H H H H - 3 3 - UH U UH UH U - 3 3 UH - Fig. pressure Case H. H H H. H 3 3 DH. D. DH D 3 3 Fig. 7 Case.. H H 3 3 DH. D DH DH. H H 3 3 Fig. Case H D Paper No. 3-JSC- Mioduszewski

emerged in water. The identification of the emerging point is indicated inside arrows. ρw g h σ z = () ρ ρ g Z z λ ( ) ( ) ( ) s w where: σ z - dimensionless h - excess pore water pressure g - acceleration due to gravity ρ s - sand density ρ w - water density z - scouring depth above the point of calculation Z - initial sand layer thickness above the point of calculation λ - porosity The liquefaction process near the specified point happens, as the dimensionless calculated for this point decreases to the zero value, while the point is still submerged in the sand. For almost all points that emerge in water, the liquefaction process takes place. The time of liquefaction vary in different points. Times of the liquefaction process are presented in Table. Table. Times of the liquefaction process in different points (unit: s) Case Case H.9. 9. 3. DH -. The longest time of liquefaction in Case is seen for the points located near the head wall. In Case, situation changes and the longest time of liquefaction is observed for the points, located at the downstream. In Case for the downstream the reduces (Fig. ). The liquefaction is observed for the point at level. In other cases, even if the liquefaction was not observed, the reduction of the is noticeable. Liquefaction is deepest in column H. In both Cases, the liquefaction was observed even in point H, which did not emerge in the water during the experiment. The time of sand liquefaction at this level lasted about 7 to seconds. After that time, gradually reached the initial value. Scouring depth During the experiments, the scouring depth was recorded. The final scouring form is presented for Case in Fig. 9 and for Case in Fig.. Naturally, in Case, the scouring area and depth was bigger than the one in Case. The biggest scouring depth observed after the experiment was. cm (in Case ) and 9.9 cm (in Case ). The area of the scouring process in Case reached cm upstream from the zero point set at the downstream of the structure, and 3 cm downstream. The width was limited by the experimental channel width. The area of the very intense scouring was limited to cm upstream, cm downstream and cm from the head in a crosswise direction (the total width of scouring was 3 cm). In Case the area was much bigger. The scouring reached cm upstream and cm downstream. The cone shaped scour hole reached cm upstream, cm downstream and cm, from the head, crosswise. The sand washed out in the spur dike area deposited in the region sheltered by the spur dike. Change of the sand level in time may be seen in Fig. and Fig. for Case and Case respectively. The observed scouring during the surge was different from what was described in the literature (Elawady, ) and what was observed in a former research for the case of a stable flow. For the stable flow, the scouring was very intense at the start, and then 3 7 3-3 -3 - - 3 Fig. 9 Scouring depth Case (unit: cm) 3 y (cm) y (cm) 9 7 3-3 - - 3 Fig. Scouring depth Case (unit: cm) gradually became weaker. The total scouring process lasted for a very long time. For the surge, during the whole time, when scouring took place, the process was characterized by equal intensity. The fading of the scouring intensity in time described in the references for stable flow was not observed during the experiment with the surge. The scouring starts about 3 to seconds after opening the gate at the upstream headwall corner. In columns where pressure was measured the scouring actions began after 7 seconds in Case and seconds in Case. The scouring action was very intensive and finished after 3 seconds ( seconds for Case ). The scouring process in Case, with bigger surge lasted in a shorter time compared to Case. In both cases, in column H, the final scouring depth was shallower than what was recorded during the experiment. The difference between maximum and final scouring depth in both cases equaled to cm. Seepage force Figures and present pressure levels on the walls of the dike. The dark areas in those figures represent regions where the pressure was lowest. The pressure configuration at the s and headwall of the spur dike suggests water flow in the ground in directions marked by arrows in the figures. According to the pressure gradients, the seepage water flowed horizontally and upward from the upstream side area to the downstream side area. The liquefaction process in the direct vicinity of the headwall and downstream described above enabled the flow out of the ground water. The seepage force was acting on the sand grains in the liquefied area promoting scouring process by water flow. - -3 - - - - x (cm) x (cm) Paper No. 3-JSC- Mioduszewski

Upstream UW Head H DW t=s Upstream UW Head H DW t=s UH DH UH DH DW3 DW3 initial bed level initial bed level t=s t=s t=s t=s t=s final sandbed profile t=3s t=s final sandbed profile t=3s - - -9 - -7 - - - -3 - - - -9 - -7 - - - -3 - - Fig. Pressure on the spur dike walls and during the experiment Case (unit: cm) - - -9 - -7 - - - -3 - - - -9 - -7 - - - -3 - - Fig. Pressure on the spur dike walls and during the experiment Case (unit: cm) CONCLUSIONS The conducted experiment proved that scouring action around the spur dike may be influenced by many factors like occurrence of excess pore water pressure and liquefaction. Observed sand liquefaction phenomenon combined with the seepage forces has an effect on the scouring. The obtained results enriched the image of phenomena taking place near the spur dike. This type of flow enabled researchers to investigate in a detailed way the occurrence of excess pore pressure and liquefaction. The experimental study indicated the influence of those as two of the factors. Nevertheless, more detailed theoretical work and numerical simulations, especially of the storm surge or big waves which are the reason for many unexpected processes, are scheduled to be done in order to improve the knowledge of these phenomena. Their simulation will enable researchers to understand and analyze it in detail which is the main aim of this research. ACKNOWLEDGMENT This study was partly supported by a Grant-in-Aid for Scientific Research (A) (Project No.7, Chief: Prof. Sakai Tetsuo), Japan Society for the Promotion of Science. The authors express their gratitude for the support. REFERENCES Elawady, E., Michiue, M. and Hinokidani, O.: Experimental Study of Flow Behavior Around Submerged Spur-Dike on Rigid Bed, Annual Journal of Hydraulic Engineering, JSCE, vol., pp. 39-,. Elawady, E., Michiue, M. and Hinokidani, O.: Movable Bed Scour Around Submerged Spur-Dikes, Annual Journal of Hydraulic Engineering, JSCE, vol., pp. 373-37,. Mia F. Sand Bed Behavior Under Water Pressure Variation and Local Scour at Bridge Piers, Doctors Thesis, Okayama University,. Maeno, S., Magda, W., Nago, H.: Floatation of Buried Submarine Pipeline under Cyclic Loading of Water Pressure, Proceedings of the Ninth International Offshore and Polar Engineering Conference, Brest France, pp. 7-, 999. Nago, H.: Liquefaction of Highly Saturated Sand Layer under Oscilating Water Pressure, Memoirs of the School of Engineering,, Okayama University, 9 Rahman, M., Murata, H., Nagata, N., Muramoto, Y.: Local Scour Around Spur-Dike-like Structures and their Countermeasures Using Sacrificial Piles, Annual Journal of Hydraulic Engineering, JSCE, vol., pp. 99-99, 99. Sumer B.M., Whitehouse J.S., Torum A.: Scour Around Coastal Structures: a summery of recent research, Coastal Engineering, vol. pp. 3-9,. Sumer B.M., Fredsoe J., Christiansen N., Hansen S.B.: Bed Shear Stress and Scour Around Coastal Structures, Proceedings of th International Coastal Engineering Conference ASCE, Kobe, Japan, vol., pp. 9-9, 99. Thorne C.R., Abt S.R., Barends F.B.J., Maynord S.T., Pilarczyk K.W.: River, Coastal and Shoreline Protection. Erosion Control Using Riprap and Armourstone, John Whiley&Sons Ltd, 993. Paper No. 3-JSC- Mioduszewski