Erosional processes of solid materials have been the focus of many researchers around the world. Erosion can commence within a wide range of material strengths depending on the amount of water-driven energy and material properties. Erosion could also occur due to Aeolian effects as well as chemical weathering but these forcings are not of the focus of this research. Instead, the focus here is on rock erosion in waterways and in particular downstream of dams. Rock erosion mostly takes place at the downstream of dams where the water conveys through the spillbays from upstream to the downstream during an extreme event. This phenomenon threatens both the structural soundness of the dam with implications to the public safety. It usually occurs when the applied hydrodynamic forces (average and fluctuating) exceed the strength of the rock mass formation. Rock scour at the downstream of dams due to high velocity impinging jet is a complex and highly dynamic process. So a deeper understanding of the process is crucial to determine the rock scour rate and extent. Hydraulic laboratory models have been employed to investigate hydraulic processes and proved to be reliable tools for testing soil/sediment erosion; however, the study of rock scour remains challenging. The prototype rock formation cannot be utilized in the laboratory models because the flowing water in the scaled model contains much less energy and exerts less forcing. On the other hand, the use of granular sediment (non-cohesive), as a standalone approach to mimic the rock formation is not a precise method, since it will most probably lead to inaccurate results. The idea of using a mixture of granular and cohesive sediment is investigated here to adequately simulate the rock erosion process in the laboratory scaled models. The granular sediment represents the rock blocks while the cohesive additive is a binder to keep the granular sediment together. The rock scour process can occur through four mechanisms; fracture failure, block removal, fatigue failure and abrasion. In this study, because the focus is on the hydrodynamic forcing effects on rock erosion, we assume that in the completely and intermittently jointed rock, erosion is mostly governed by fracture, block removal and fatigue failure. Abrasion is triggered by collisional effects and is not the focus here. So, we hypothesize that if the rock formation considered being pre-fractured, it can be simulated using a mixture of non-cohesive sediment with cohesive additive. This method was utilized to assess the rock scour process at the downstream of the Priest Rapids Dam. The Priest Rapids Dam project was part of a series of projects that was conducted at IIHR-Hydroscience & Engineering at The University of Iowa and sponsored by the Public Utility District No. 2 of Grant County, Ephrata, Washington (GCPUD) to investigate juvenile salmonid migration at the Wanapum/Priest Rapids Development. It is a hydroelectric, concrete gravity, and mid-elevation dam owned and operated by Public Utility District No. 2 of Grant County, Washington (the "District"). To aid the District in their evaluation of fish passage, IIHR-Hydroscience & Engineering constructed comprehensive three-dimensional physical models of the forebay and tailrace of Priest Rapids Dam and a third model of spillbays 19-22 and powerhouse Unit 1 (sectional model). As part of the last phase of the project, it was crucial to assess the effects of the newly designed fish bypass system on the downstream rock foundation scour. To investigate this process, the 1:64 Froude-based scale tailrace model of the dam was utilized. The mixture of gravel, bentonite clay, and water was employed to mimic the rock formation and simulate the bedrock scour process in the model. Series of preliminary experiments were conducted to find the optimum mixture of gravel, bentonite and water to accurately replicate an existing scour hole observed in the prototype tailrace. Two scenarios were considered. First, tests were conducted to estimate the scour potential downstream of the fish bypass, which is currently under construction. Second, the scour potential downstream of the dam was also assessed for the Probable Maximum Flood (PMF) with the fish bypass system running. Based on the model tests results and observations, the simulated bedrock (mixture of gravel and cohesive bentonite) was able to replicate the rock scour mechanisms, i.e. fracture process, block removal and fatigue observed in nature. During the fish bypass scour tests, it was observed that the erosion process occurs in the form of block removal and fatigue failure. During the PMF scour test, instead, it was observed that the mixture is eroded in chunks of substrate. This process can be representative of fracture failure in rock which occurs when the induced pressure fluctuation exceeds the fracture strength or equivalently toughness of the rock. In the preliminary phase of this work it was recognized that a prerequisite for replicating the processes in the laboratory is the proper preparation of the mixture. There is limited information available in the literature about how much cohesive additive is required to simulate the erosional strength of the prototype rock formation. For this reason, in this study the effort has been made to develop a method to simulate the rock formation for studying rock scour process in the laboratory analog scaled models. To simulate the bedrock formation, various combination of granular sediment (gravel), cohesive additive, and water were created and tested. Choosing an appropriate cohesive additive concentration is critical and nearly a balancing act. An appropriate cohesive additive concentration should be cohesive enough to bind the material and not too strong to be eroded by the flowing water in the scaled models. Moreover, its properties should not change over time. Various cohesive additives can be mentioned i.e. kaolin clay, bentonite clay, cement, grease, paraffin wax. Among all of them, bentonite clay was chosen as the appropriate cohesive additive due to its swelling characteristic. When bentonite is mixed with granular sediment, it is restricted by the non-cohesive sediment grains. The bentonite expands to fill the voids and forms a tough, leathery mineral mastic through which water cannot readily move. In order to assess the erodibility of the mixture the Jet Erosion Test (JET) apparatus was used. The JET apparatus is a vertical, submerged, circular, turbulent impinging jet which is widely accepted and utilized to assess cohesive soil erosion through flow impingement. There are devices such as flumes which could be effectively used for bank erosion where the flow shear action is prevalent. In this study, it was sought important that the forcing replicated in the experiments was of the same nature (normal impinging forcing instead of shear forcing) as observed in the downstream end of a dam. For this reason, JET was chosen as it provided a larger range of stresses (ranging between 100-1000 Pa) comparing to the flume device. The apparatus was designed based on the device developed by Hanson and Hunt (2007) and built at the IIHR-Hydroscience & Engineering. Various replicate samples were made with different combinations of gravel, sodium bentonite clay, and water. To determine the erosional strength of the samples (critical stress) they were tested using the JET apparatus. The critical stress was determined as the stress associated with zero eroded mass. The results revealed that the erosional strength of the simulated bedrock mixtures highly depends on the amount of adhesive component (bentonite clay). The mixtures with the higher percentage of bentonite clay are less susceptible to erosion. The erosion threshold plot - similar to Annandale's plot - for the simulated bedrock mixtures was developed. Using the erosional strength of the simulated bedrock mixtures, a step-by-step systematic method was developed to determine the optimum combination of weakly cohesive substrate in order to simulate the strength of the prototype bedrock. The method is based on the Annandale's erodibility index method and requires information about the prototype bedrock strength (erodibility index). The method is explained in conjunction with the Priest Rapids Dam project example. The old trial and error method to establish an optimum weakly cohesive substrate is costly and time consuming especially in the case of large scale laboratory models. Also, the applicability of the method would be questionable when there is not enough information or a past data set that can be used as a baseline (witness) test. The new method eliminates these problems and the optimum mixture can be established using the geological information of the prototype bedrock formation.

Thesis (MScEng (Civil Engineering))--University of Stellenbosch, 2010. / ENGLISH ABSTRACT: The scour process of unlined spillways is an important research topic of value in engineering practice. In South Africa numerous unlined spillway dams have experienced severe erosion. This led, in some cases, to the costly concrete lining of spillways for erosion protection. On the other hand, the erosion of unlined spillways can lead to damage to, and even failure of dams and consequently can affect public safety, properties, infrastructure and the environment. In this regard, methods to predict erosion of unlined spillways are therefore needed as tools in the risk management and design of existing unlined spillways as well as future spillway structures. The prediction of the rock scouring process is challenging and empirical formulas have been established to predict incipient conditions for scour. These empirical methods however do not predict the rate of scour or the ultimate equilibrium rock scour. The key objective in this study was to investigate the applicability of a non-cohesive two-dimensional (2D) sediment transport hydrodynamic mathematical model to simulate unlined spillway scour. A physical model flume test was set up to simulate rock scour represented by uniformly sized polyethylene cubes. The flume slope and discharge were varied in the different test scenarios. The 2D mathematical model correctly predicted the extent and location of scour as observed in the laboratory tests. Temporal changes in the scour formation were also predicted with reliability. This was achieved by only calibrating the hydraulic roughness of the 2D model, and by specifying the ”rock ” particle settling velocity and material density. The simulation results were satisfactory, providing an accurate and detailed erosion prediction. From this, the mathematical modelling was validated by using a field case study. The results obtained with the mathematical model were promising for non-cohesive cases and could be applied to field prototype cases if the rock joint structure is known. This would typically apply in fault zones, where the joints would give an idea of the rock size to be used in the mathematical model. In general, where jointed rock is more massive and acts ”cohesive ”, rock parameters describing critical scour conditions for the rock in terms of stream power are required to be built into the mathematical models. / AFRIKAANSE OPSOMMING: Die uitskuringsproses van onbelynde oorlope is ’n belangrike navorsingsonderwerp in die ingenieurspraktyk. In Suid-Afrika word ernstige erosie in baie onbelynde oorlope van damme ondervind. In sommige gevalle het dit gelei tot die duur belyning van oorlope met beton, om die oorlope te beskerm. Aan die anderkant kan die erosie van onbelynde oorlope lei tot groot skade en selfs tot die faling van ’n dam. Dit kan weer lei tot skade aan eiendom, infrastruktuur en die omgewing, en die publiek in die gevaar stel. Daarom is dit nodig dat daar besin word oor metodes om erosie in onbelynde oorlope te voorspel, sodat die risiko bestuur kan word en om te sorg vir die beter ontwerp van onbelynde oorlope in die toekoms. Dit is moeilik om die uitskuringsproses te voorspel, maar empiriese formules is bestaan om die aanvang van uitskuring te voorspel. Hierdie empiriese metodes voorspel egter nie die snelheid waarteen die uitskuring sal plaasvind of die uiteindelike mate waartoe dit sal gebeur nie. Die hoofdoelwit van hierdie studie was om die toepasbaarheid van ’n nie- kohesie, twee-dimensionele (2D) hidrodinamiese wiskundige model te ondersoek, om sodoende die uitskuring van onbelynde oorlope te simuleer. ’n Fisiese model om die uitskuring van rots te simuleer is ook gebou. Die rots is deur polietileen blokkies van dieselfde grootte gemodelleer. Verskillende kanaalhellings en deurstromings is in verskillende toetse gebruik. Die 2D wiskundige model het volgens die waarnemings in laboratorium toetse, die mate en ligging van die uitskuring korrek voorspel. Veranderinge wat met verloop van tyd in die uitskuring formasie plaasgevind het, is ook betroubaar voorspel. Dit is gedoen deur die hidrouliese ruheid van die 2D model te kalibreer en deur te spesifiseer hoe vinnig die ”rots ” deeljies afsak en wat die digtheid van die materiaal is. Die uitslag van die simulasie was bevredigend en het die erosie akkuraat en in detail voorspel. Die wiskundige modellering is gevalideer deur middel van ’n gevallestudie. Die uitslae wat met die wiskundige model verkry is, is belowend, en geld vir nie-kohesie gevalle. Dit kan op prototipe gevalle in die veld toegepas word as die rots se naatstruktuur bekend is. Dit kan toegepas word in foutsones waar die nate ’n aanduiding sal gee van die grootte van ”rotse ” wat in die wiskundige model gebruik moet word. Maar as die rots baie groot is en die kohesie goed is, is dit nodig om meer parameters betreffende uitskuringstoestande gekoppel aan stroomdrywing, in die wiskundige model te gebruik.