Studies on Hydrodynamic Erosion with Soil Protrusion Apparatus

34 th IAHR World Congress - Balance and Uncertainty 26 June - 1 July 2011, Brisbane, Australia 33 rd Hydrology & Water Resources Symposium 10 th Hydraulics Conference Studies on Hydrodynamic Erosion with Soil Protrusion Apparatus Sarik Salim 1 and Ravindra Jayaratne 2 1,2 School of Computing, Information Technology & Engineering University of East London, Docklands Campus 4-6 University Way, London E16 2RD UNITED KINGDOM E-mail: s.salim@uel.ac.uk 1, r.jayaratne@uel.ac.uk 2 Abstract: The current paper discusses an extended investigation on the hydrodynamic behaviour of soil erosion under the fully controlled laboratory conditions studied in previous experiments (Jayaratne et al., 2010a, 2010b). The Ahlborn sediment mobile bed tank at the University of East London hydraulics laboratory was purposefully re-fabricated to install a Soil Protrusion Apparatus (SPA) for investigating sediment properties on soil erodibility (Jayaratne et al., 2010a). A plastic sediment bed was used to allow hydraulic measurements to be carried out with clear visual observation. An HD digital video camera was used to record the soil-hydrodynamic phenomena and digital images were analysed to obtain accurate measurements of the spreading area and volume with a computer aided design software. Extended physical modelling of crescent zones also included analysing their grain size distribution. Measured sediment and flow data were used to develop a simple empirical relationship for erosion rate for each crescent zone using dimensional analysis and best-fit techniques. Keywords: Soil-hydrodynamic behaviour, Soil erodibility, Crescent zones, Empirical relationship, Dimensional analysis, Best-fit technique. 1. INTRODUCTION Soil erosion due to fast flowing water is the main case of river and coastal structure failure around the world. Erosion of sediment is attributed as the cause of failure in hydraulic structures such as bridges, levees and embankments adopted in flood protection schemes in river and coastal environments. These structures need to withstand the severe flow conditions of natural disasters such as floods, hurricanes and storm surge. In the UK, incidents such as the collapse of the railway bridges in Glanrhyd (1987), Inverness (1989), and Northside Bridge in Workington (2009), and coastal erosion of sea cliffs at Happisburgh, Norfolk (2006) have enhanced understanding of how hydraulic structures are swept away due to soil erosion [see Happisburgh (2010) and BBC News (2010)]. Many research studies in the past revealed that most hydraulic structures have collapsed due to the failure attributed to erosion associated with a soil-hydrodynamic interaction phenomenon. To minimize future structural damage due to erosion and ensure public safety, developing and implementing improved procedures for designing hydraulic structures and inspecting them for erosion have prime importance. The Ahlborn sediment mobile bed tank (4.0 0.6 0.2 m) at the University of East London (UEL) hydraulics laboratory was purposefully re-fabricated to install a Soil Protrusion Apparatus (SPA) for investigating of sediment properties on soil erodibility. The influence of sediment properties (Density, porosity, average grain size, grain shape, uniformity coefficient and gradation coefficient) on soil erodibility is discussed in detail by the authors. Erosion characteristics such as sediment deposition patterns, longitudinal (l 1 ) and lateral (l 2 ) spreading length and area for three different sediment diameters under wet and dry conditions are studied (see Jayaratne et al., 2010a). A simple empirical relationship was proposed for the dimensionless erosion rate in terms of the average shear stress, average grain diameter, soil protrusion and erosion spreading length ratio (l 1 /l 2 ) using the dimensional analysis and a best-fit technique (see Jayaratne et al., 2010b). Model parameters were calibrated using measured hydrodynamic data. The current accepted practice for erosion prediction is the further extension to the work of laboratory investigation reported in Jayaratne et al. (2010a, 2010b). For this purpose, a plastic sediment bed was used allowing hydraulic measurements to be carried out with a clear visual observation. Detailed ISBN 978-0-85825-868-6 3761 Engineers Australia

physical modelling of crescent zones was discussed considering grain size distribution. Experimental data were used to develop a simple empirical relationship for the erosion rate of each crescent zone using dimensional analysis and a best-fit technique. Finally, a discussion is made for the field application with respect to the integration of multiple crescent zones. 2. SEDIMENT TRANSPORT MECHANISM Sediment is fragmental material, primarily formed by the physical and chemical disintegration of rocks from the Earth s crust. Such particles range in size from large boulders to colloidal size fragments and vary in shape from rounded to angular. Erosion is the process of moving and removing of sediments from their original source or resting place. Usually, three modes of particle transportation are distinguished: bed load, suspended load and wash load transport (Van Rijn, 1993). Generally, the transport of particles by rolling, sliding and saltating is called bed load transport and it provides the major process relation between the hydraulic and material conditions that govern sediment bed erosion. An understanding of bed-load movement is required not only to reveal the causes and consequences of changes in fluvial form but also to help make decisions regarding the sustainability of a hydraulic structure. In this context, the physics and relevant existing formulae for erosion of the bottom bed without any obstacles are explained based on previous research works (e.g. crescent zone development) to predict the behaviour of erosion in natural rivers and coastal areas. Figure 1 illustrates the involved sediment transport mechanism where the movement of the sediment grains due to collisions is assumed to be governed by a fluid lifting force and the force of gravity. Based on this mechanism, the physics of soil erosion in smooth plastic bed is investigated as described in Section 5.1. Water flow Soil-water interaction Turbulent eddies of water flow Grain collisions Lifting force> Gravity force Sediment transport Turbulent eddies of water flow Grain collisions Gravity force> Lifting force Deposition (Equilibrium Stage) Figure 1. A simplified model flow chart demonstrating the considered bed load transport mechanism involved during the experimental investigations 3. SOIL PROTRUSION APPARATUS (SPA) The Soil Protrusion Apparatus (SPA) described in this paper was developed at the University of East London with the following specific goals in mind; (1) to be able to perform site specific erosion studies, (2) to minimise disturbance, (3) to study the hydrodynamic characteristics of the soil s, and (4) to incorporate the test results in an erosion prediction method. The tank was purposely re-fabricated in order to install a Soil Protrusion Apparatus (SPA) that takes 100.0 mm diameter sediment core s (see Figs. 2a, 2b and 2c). The apparatus was placed at a distance of 1.5 m from the upstream of the flow, and along the centre line of the tank. The tube 3762

was fitted with a moveable piston which enabled the to be pushed to form a protrusion between z=1.0-10.0 mm in 1.0 mm intervals (see Fig. 2d). The constructed channel was made of smooth plastic with a flume dimension of 4.0 0.6 0.2 m. A thin plastic 5.0 5.0 mm X-Y grid system was attached to the bottom of the modified plastic flume to measure the sediment spreading lengths (see Fig. 2h and 2i). Flow visualisation was carried out using Potassium Permanganate (KMnO 4 ) and the resulting patterns were recorded using the HD Sony video camera, under normal lighting conditions (see Fig. 2e and 2f). Vortex eddies were also observed by sprinkling of Potassium Permanganate (KMnO 4 ) on the water surface, which gave a good indication of areas with re-circulating flow (see Fig. 2g). Ø100.0 mm Soil protrusion tube 5.0 mm thick protrusion (b) (c) (f) (g) (a) (d) (e) Ahlborn sediment bed tank at UEL Current meters Current Meter Soil Sample Current Meter Water Flow Sediment Plume Water flow Water depth X-Y Grid System Mobile Bed Model Tank (h) Mobile (i) tank bed Protruding portion of soil (i) SPA Figure 2. (a) Modified Ahlborn sediment bed tank, (b) Ø100.0 mm Soil protrusion tube, (c) 5.0 mm thick protrusion in the soil tube, (d) Soil Protrusion Apparatus (SPA) with a dial gauge (0.25 mm precision), (e,f) Mounted HD Sony digital camera, (g) Unidirectional flow visualised after sprinkling Potassium Permanganate (KMnO 4 ), (h) Thin plastic X-Y grid system attached at the bottom of the modified plastic flume, (i) Conceptual diagram of experimental set-up Experiments were repeated at least 3 times for each test condition to ascertain the reproducibility and representability of the observations. Longitudinal (l 1 ) and lateral (l 2 ) erosion spreading of the sediment plumes were recorded via an HD Sony digital video camera (1920 1080, 24 frames per second). The flow rate was adjusted as desired through control of the voltage input into the pump. Propeller type current meters (0.06-1.5 m/s) were used in the upstream and downstream sections of the tank to monitor the uniformity of the flow and its measurement. 3763

Manual Data Source Spreading length Soil protrusion Deposition time Eroded weight Water depth Instrumental Data Source Upstream and downstream velocities Grain size Grain shape Erosion profile Start Complete Dataset Digital Data Source Spreading length (Longitudinal/Lateral) Spreading area Erosion time Erosion pattern Figure 3. Data collection process During the experiments, data were collected from three sources: Manual data source, Instrumental data source, and Digital data source (see Fig. 3). 4. PREVIOUS INVESTIGATION 4.1. Physical Model Previous laboratory experiments on soil-water interaction for 3 different diameters of soil s under wet and dry conditions were carried out in the same moveable sediment tank (Jayaratne et al., 2010a). Based on the experimental results, it was found that soil of the same diameter under wet and dry conditions gives significant changes in sediment transport rate and to its erosion pattern. Transport rates were much slower in dry conditions than in wet conditions. It was observed that for the same flow condition, different soils give different long term equilibrium deposition patterns due to the grain size distribution and particle shape. Soil Water flows Poorly graded Erosion starts Well graded Erosion stops Equilibrium stage Erosion Poorly graded Well graded Figure 4. Identification of different erosion stages during experiments Figure 5. A photographic view of crescent deposition zones developed on smooth bed (d50=0.30 mm, wet soils, z=5.0 mm, Q=3.57 l/s) Wake vortices were generated behind the soil s which resulted in the formation of a series of crescent zones. The process of sediment deposition during scour process was identified through the laboratory observations on a smooth plastic sediment bed (see Figs. 4 & 5). 3764

4.2. Empirical Relationship A simple empirical relationship was developed for the erosion rate in terms of measured flow and sediment parameters using dimensional analysis and a best-fit technique (Jayaratne, 2004). A calibration coefficient for this model is applicable for smooth bed flow conditions and model dimensionless parameters are given in Eq. (1). 0.40 0.21 q s 5 τ s z = 1 10 (1) d 50 ( 1) 50 d50 g( s 1) ρgd s d50 where: q s = Volumetric sediment transport rate (m 2 /s/m) d 50 = Median particle size of soil (m) g = Acceleration due to gravity (m/s 2 ) s = Specific density (ρ s /ρ) τ s = Average bed shear stress (N/m 2 ) ρ = Density of water (kg/m 3 ) z = Protrusion of the soil (m) 5. NEW INVESTIGATION 5.1. Physical Modelling of Soil Erodibility with Crescent Zones The modes of sediment transport in rivers depend on the grain size, shape and density of the material, settling velocity and flow velocity (Marriott, 2009). Hence, it is important to study the influence of those parameters on soil erodibility. One of the vital parameters is gradation of soil which represents the soil engineering properties such as compressibility, shear strength, and hydraulic conductivity. The tested soils were a series of uniform soils of known particle size distributions with uniformity coefficients and effective sizes. Apart from the Scanning Electron Microscope observations, further classification of the experimental soil s based on the uniformity and gradation coefficients and the corresponding values are discussed in Jayaratne et al. (2010a). The crescent zone phenomenon is described based on this classification in order to find a correlation between erosion and soil/water parameters, it is essential to study the scour and deposition patterns around the soil s. Spreading length and area is an indication of the sediment transport mechanism acting on such soils under particular flow conditions. Bed scour determines the degree of deformation of the original soil protrusion. (a) d 50 =0.15 mm at t=21.0 s (b) d 50 =0.26 mm at t=23.0 s (c) d 50 =0.30 mm at t=25.0 s (d) d 50 =0.75 mm at t=27.0 s Figure 6. Equilibrium stage of 4 different soil s with soil protrusion, z=5.0 mm and water discharge, Q=3.57 l/s Figure 6 shows that for the same experimental conditions, equilibrium settling patterns of soil s in the downstream section of the tank form a specific shape named crescent zone as described in Jayaratne et al. (2010b). According to Fig. 6a, crescents are dense and smaller in size and all are connected to each other. In Fig. 6b, crescents are less dense than those shown in Fig. 6a. In Fig. 6c, the crescents demonstrate large gaps between adjacent zones. Figure 6d shows a similar phenomenon but as the gravitational force is higher than the lifting force crescents are formed less frequently than in the previous case. For the case of soils with d 50 =0.30 mm, it was noticed that the propensity to develop such zones was more common in comparison to the other soil s. This phenomenon is primarily governed by the particle shape and size. The visual and video observations revealed that the wake vortices generated at the upper and lower ends in those zones further refined the deposition shape. Repeated observations confirmed this phenomenon. 3765

For example, a time dependent analysis is described in Fig. 7. During the erosion process, crescent zones maintain a sequential order where zone A was first observed then zone B, C and D as shown in Fig. 7. The reason for this is related to the water vortex developed by the crescents for each zone. Sediment particles in each crescent zone attempt to settle down as group of same grain sizes (d 50 ). A B A C B A D C B A Erosion at t= 0 s Erosion at t= 5.0 s Erosion at t=10.0 s Erosion at t=15.0 s Figure 7. Time dependent development of crescent zones for d 50 =0.30 mm at Q=3.57 l/s During the experiments, it was noticed that average grain size decreases with the increment of water discharge because high water force reduces the fixed packing sediment which causes more spreading and results in decrement of d 50. Moreover, investigation of bed profiles shows when the d 50 is small, deposition height increases from its protrusion that is used in experiments. This is because the lifting of particles of smaller diameter is easier than that of larger sized particles. Figure 8 shows the comparison of crescent zone development in smooth and rough beds to simulate field conditions. (b) (a) Figure 8. Comparison of crescent zone developed on (a) smooth bed, and (b) rough bed 5.2. Empirical Relationships for Erosion Rate Prediction of Crescent Zones In this experiment series, measured sediment and flow data were used to develop a simple empirical relationship for the erosion rate of each crescent zone using dimensional analysis and a best-fit technique. Different non-dimensional parameters were included to verify significant aspects of soil transport processes. Comparisons were made with respect to Reynolds number (Re) as it is a nondimensional parameter which provides better a representation of the discharge and flow velocity. d 50 crescent/ d50 1 0.8 0.6 0.4 0.15mm 0.2 0.26mm 0.30mm 0.75mm 0 10000 12000 14000 16000 18000 20000 Re A crescent / A 5 (a) (b) (c) 4 3 2 0.15mm 1 0.26mm 0.30mm 0.75mm 0 10000 12000 14000 16000 18000 20000 Re q s /q c 3 2.6 2.2 1.8 0.15mm 1.4 0.26mm 0.30mm 0.75mm 1 10000 12000 14000 16000 18000 20000 Re Figure 9. (a) Relationship of dimensionless average grain size with Re, (b) Relationship of dimensionless spreading area with Re and (c) Relationship of dimensionless erosion rate with Re The main focus of this analysis is to produce a model for the prediction of the erosion rate for each crescent zone. The average grain size of each developing crescent was found to be a dominant parameter therefore it is included in the proposed empirical model. Using the recorded video footage of each experiment, it was also found that spreading area of eroded soil plays a vital role for the development of crescent zones. According to Fig. 9, a dimensionless average grain diameter decreases incrementally with the Reynolds number. Furthermore, a dimensionless parameter for the spreading area and erosion rate 3766

per unit width are found to be directly proportional to the Reynolds number. This is because, under the same flow conditions, the maximum extent of crescent zone varies with the gradation of soil. When the is well graded, the lateral length of crescents is high on the other hand when it is poorly graded the lateral length of crescents is low. An empirical relationship is developed using the dimensionless parameters given in Eq. (2). d 50, Acrescent zone q cs = f, (2) d 50, crescentzone A where: q cs d 50, d 50,crescent zone = Volumetric total sediment transport rate of each crescent zone (m 2 /s/m) = Median particle size of soil (m) = Median particle size of each crescent zone (m) A crescent zone = Area of each crescent zone (m 2 ) A = Area of soil in the SPA tube core (m 2 ) Figure 10 illustrates the relationship between the erosion rate of each measured crescent with average grain diameter and crescent area. Using the best-fit technique, a relationship for total erosion rate (q ts ) was obtained as; q ts 0.15 0.47 n d = 50, Acrescent zone = k (3) i 1 d50, crescentzone A where numerical coefficient, k=1.36 and has a dimensions of m 2 /s/m. q cs 5 4 3 2 1 0 R² = 0.92 1 1.5 2 2.5 3 (d 50 crescent zone /d 50 ) (A crescent zone / A ) Figure 10. Relationship of dimensionless erosion rate with dimensionaless average grain diameter and spreading area 6. CONCLUSIONS In this investigation, laboratory-developed Soil Protrusion Apparatus (SPA) was used to help better understand the fundamental physics of sediment transport and to develop an empirical relationship to predict the soil erosion rate of crescent zones. For this purpose, 6 sets of laboratory experiments were conducted on each two average grain diameters (d 50 =0.15, 0.26 mm) and 5 sets of similar experiments were carried out for d 50 =0.30 mm comprising a total of 330 experiments. Apart from the scale effects, based on the results it was found that for the same water discharge protruded s formed crescent zones maintaining a sequential order. For the small grain sized s, more crescent zones are formed and these are found to be closer to each other compared to the s with larger grain sizes. Present empirical equation shows a relationship between dimensionless average grain diameter and spreading area with erosion rate of each crescent zone per unit width for smooth bed which could be used with further modification to predict the erosion rates in field conditions. 7. FUTURE DIRECTIONS Present investigation represents an improvement to an initial step, filling the knowledge gap on the fundamental understanding of soil-water interaction. The empirical relationship for erosion rate has been established in terms of smooth bed condition whereas rough bed analysis is desirable to validate the formula. Investigations need to be compared with full-scale scour measurements such as field data in hydraulic structure failure sites. Moreover, two aspects of the SPA can be improved. Firstly, the water depth needs to be increased to a reasonable depth. Secondly, the turbulence intensity generated by the water velocity and the friction of the wall in test section could not be controlled 3767

therefore by placing adjustable obstacles on the side walls to control the turbulence intensity while measuring it qualitatively. Areas likely to receive attention for an extended investigation on soil-hydrodynamic behaviour include increasing the mobile bed roughness to values greater than those used in earlier experiments by the authors. A low cost technique involving spraying separate solutions on to a drained flume will be used to freeze the soil bed geometry (Benson et al., 2001). Proposed experimental work will lead to the comparison, calibration and development of predictive curves of erodibility for different soils and verify the proposed empirical relationship. New improved process models will be developed analysing the experimental data using suitable computer programmes for the simulation of interrelated changes in bed topography. 8. ACKNOWLEDGEMENTS The authors wish to thank the financial support provided by the Graduate School of University of East London (UEL) to carry out the laboratory experiments in line with the extensive research project Mathematical and Physical Modelling of Hydrodynamic Erosion of Soils (HES) set up under UEL Promising Researcher Fellowship (2009/10). Special appreciation goes to Prof. D C Wijeyesekera at UEL for his pastoral support to initiate the HES project. 9. REFERENCES Benson, I.A, Valentine, E.M, Nalluri, C. and Bathurst, J.C. Stabilising the sediment bed in laboratory flumes, Journal of Hydraulic Research, Vol. 39, No. 3, 2001, pp. 279-282. Briaud, J. L., Ting, F.C.K., Chen, H.C., Gudavalli, R., Perugu, S., and Wei, G. Erosion function apparatus for scour rate prediction, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 125(4), 2001, pp. 105-113. Jayaratne, M.P.R. Modelling of suspended sediment concentration and cross-shore beach deformation model, PhD Thesis, Yokohama National University, 2004, pp. 133. Jayaratne, R, Salim, S. and Wijeyesekera, C. Sediment transport with a hydrodynamic perspective, 8 th International Conference on Geotechnical and Transportation Engineering (Geotropika 10), Kota Kinabalu, Sabah, 2010a, CD-ROM (Paper No: 117). Jayaratne, R, Salim, S. and Wijeyesekera, C. Scour rate prediction through laboratory observations of sediment transport in transitional turbulent flow regime, 3 rd International Conference on the Application of Physical Modelling to Port and Coastal Protection (CoastLab 10), Barcelona. 2010b, pp.119-120. Marriot, M. Civil Engineering Hydraulics, 5 th Edition, Wiley-Blackwell Publishers, Chapter 14, 2009, pp. 327-359. Van Rijn, L.C. Principles of Sediment Transport in Rivers, Estuaries and Coastal Seas, Aqua Publications, Amsterdam, Chapter 3, 1993, pp. 3.5-3.7. Internet Sources: Happisburgh, http://www.happisburgh.org.uk/ (Last accessed: 15/04/2010). BBC News: Can you stop bridges collapsing in floods? http://news.bbc.co.uk/1/hi/magazine/8374616.stm (Last accessed: 12/04/2010). p 3768