The safety of migrating salmon, especially salmonids, in the Pacific Northwest has been a concern for decades. With the advent of fish bypass systems, and safer turbines the focus of salmon safety has turned to total dissolved gases. Produced by entrainment of air into tailrace waters, total dissolved gases (TDG) can cause gas bubble disease, a harmful and potential lethal disease in fish. Avian predators are another danger for migrating salmon. In some areas of the world birds common in the Pacific Northwest can account for as much as 65% of salmon smolt losses. The goal of this thesis is to determine the effects of changing operational conditions at McNary dam on fish exposure to predator habitats and TDG. Computational fluid dynamic models were implemented to predict the hydrodynamics, TDG distribution and inert particle trajectories in the tailrace of McNary dam for varying operational conditions. A 3D volume of fluid (VOF) model was used first to capture the free surface shape in the tailrace. A rigid-lid model was then used to simulate the hydrodynamics and TDG distribution within the tailrace using the free surface shape from the VOF model. This 3D two phase model utilized an anisotropic Reynolds Stress turbulence model. All grids were generated using the commercial Gridgen software. A lagrangian particle tracking model that followed Newton's laws of motion were used to track inert particles throughout the domain. Validation of the model was performed. A grid refinement study with four different refinement levels was performed. Velocities for each grid type were compared against field data taken in 2004, and TDG was compared amongst the four grids. It was determined the medium level of refinement could accurately predict the velocities, and the TDG was relatively independent of grid density; TDG averages at the grid outlets were within 1.435% of one another. The TDG distribution was then compared, using the grid of medium refinement against field data measured in 1997and were between 1.5 and 3% of error depending on the transect. After validation of the model 16 predictive simulations were run with varying levels of total river flow and operational conditions. Tailrace hydrodynamics along with TDG production and distribution were compared for simulations with comparable total river flow rates. Fish trajectories were tracked using the particle tracking model. Inert particles were injected into the domain and properties such as velocity, distance to the shore and depth about each were recorded. Statistics were then generated for the particles based on criteria that defined dangerous predation zones within the tailrace. After completion of the simulations, it was determined that existing operations consistentlyproduced higher levels of TDG due to increased entrainment of the powerhouse flows into the spillway regions. It was also found that with increasing total river flows, TDG levels increased. On average, summer operations had lower TDG than spring due to the lower total river flows. Predation zones were similar for all simulations, but particle statistics varied depending on operational conditions. In general, particles were safer for higher flowrates as fewer low velocity eddies where particles could be trapped formed in simulations with high flowrates.

The purpose of this thesis was to perform a comprehensive evaluation of proposed sluiceway deflectors in Hells Canyon Dam with the use of Computational Fluid Dynamics (CFD). A CFD model developed and validated by Politano et al. (2010) was used to assess the downstream performance of the deflectors. Relative performance is measured by effects of the deflectors on the flow field, Total Dissolved Gas (TDG) production, and probability of mechanical fish injury. The deflectors evaluated in this model included the deflector with dimensions determined from a physical model as well as three additional deflector geometries that adjusted elevation, length and transition radius based on the physical model deflector. Physical model testing, at a 1:48 scale, of deflectors on Hells Canyon Dam performed by Haug and Weber (2002) provided a baseline deflector for the deflectors modeled in this study. The physical model was built and tested by the IIHR Hydroscience and Engineering. The performance study that this thesis focuses on was performed at two different tailwater elevations, established with two different total river flowrates of 25 kcfs and 45 kcfs. Each deflector was evaluated considering the spillway jet regime, tailrace flow pattern, and total dissolved gas (TDG) production. According to the model, decreasing the deflector length or increasing the transition radius results in more TDG production at all tailwater elevations. At 45 kcfs, the height of the deflector does not appreciably affect the spillway jet regime or the TDG distribution in the tailrace. However, increasing the deflector elevation at this river flow increases the amount of powerhouse entrainment and induces a recirculation in the western region of the tailrace. The baseline deflector performed best because it had the smallest impact on the tailrace flow pattern and produced the least TDG. The performance of the selected deflector was further evaluated for additional river flow rates of 37 kcfs, 45 kcfs and a 7Q10 flow condition of 71.5 Kcfs, with the 7Q10 condition being tested with and without the deflector. Although the deflector was able to prevent the spillway flow from creating a large amount of downstream TDG the 7Q10 flow condition significantly increased the TDG values downstream of the deflector relative to the other tested conditions. With the chosen deflector TDG values returned to forebay levels after 1 and 3.5 miles for the 37 kcfs and 45 kcfs river flowrates, respectively. With the deflector installed the 7Q10 flow condition creates considerable TDG production however the deflectors are able to reduce TDG production by 10% from the test without a deflector installed. For all evaluated river flows, with the chosen deflector, entrainment from the powerhouse is observed in the simulations; this entrainment is caused by the sluiceway surface jets. As powerhouse flow increases there is an observed decrease in entrainment. This is due to the increase of flow velocity in the streamwise direction, or perpendicular to the direction of entrainment. An important western recirculation that is prominent in the 7Q10 flow condition is also caused by the introduction of deflectors onto the spillways. Reversed flows near the fishtrap region and water directed back into the aerated section of the spillway are consequences of this recirculation. The effect causes a 25% percent increase of entrained flow relative to the no deflector 7Q10 flow. Injury of fish traveling over the spillway and through the sluiceway was estimated with the use of inert spherical particles and the computed flow field. Acceleration and strain experienced by the particles was calculated over the length of the spillway region. Numerical results were compared against literature values published by Deng (2005). Including the deflectors in the design increases the probability that fish will be injured. The most extreme cases of fish injury probability were 37 kcfs and the 7Q10 kcfs flowrates. For these cases, injuries experienced by the fish were 10% and 3% for minor and major injuries respectively. With comparison of the 7Q10 flows it appears that the inclusion of the deflector increases the induced minor injury induce from 5% to 10% and the major injury from 1% to 3%. Fish tailrace residence time was calculated using inert particles introduced to the computed fluid flow field. These particles were tracked for 650 feet past the sluiceway inlets and their time to completion was recorded. Particles were released from the sluiceways as well as the powerhouses for the 37 kcfs, 45kcfs and 7Q10 flow conditions. Particles released from the sluiceways reduced in residence time with an increase in sluiceway flowrate. With some amount of powerhouse entrainment increasing the residence time of the particles released from the powerhouse. These particles follow the entrainment to the deep low velocity region in the stilling basin. As the lateral flow increases some of the particles released from the spillway will join the high speed jets produced by the deflectors and their residence time will be reduced. According to the model, deflectors consistently reduce overall residence time and are therefore not expected to increase fish migration time. Water surface elevation near the fishtrap was measured for the 25 kcfs, 37 kcfs, 45 kcfs and 7Q10 flow conditions. The wave height near the fishtrap for the 7Q10 deflector case was predicted to be about one foot above the estimated water surface elevation. According to the model the inclusion of the deflector reduces the wave height.

Here the nature of gas phase processes and their implications for flow and transport were examined using a pilot-scale, 2-dimensional, laboratory tank instrumented for direct, in situ trapped gas measurements. Experimental conditions mimicked an unconfined, homogeneous sand aquifer with horizontal flow. Key areas of investigation included i) trapped gas dissolution following a water table fluctuation; and ii) gas phase dynamics within a hydrocarbon plume experiencing dissolved gas production via biodegradation. In the first experiment, dissolution occurred as a diffuse, wedge-shaped front propagating down-gradient in the tank over time, with enhanced dissolution at depth. Front advancement at the deepest monitoring point was 4.1 - 5.7x faster. This dynamic, depth-dependent pattern was mainly attributed to increased dissolved gas solubility. An estimated 12% increase in quasi-saturated hydraulic conductivity (Kqs) also contributed to greater dissolution at depth. Overall, the dissolution front near the water table advanced 1 m down-gradient in 344 days, suggesting that gas trapped shallowly will likely persist for significant periods of time. The utility of total dissolved gas pressure sensors for simple in-well measurements to detect trapped gas and monitor its dissolution were also demonstrated. During the second experiment, biodegradation occurred under variable redox conditions, ranging from denitrification to methanogenesis. Significant in situ increases in trapped gas were observed within the tank over 330 days. Maximum gas saturations never exceeded 27% of pore volume even during continued dissolved gas production, indicating ebullition upon reaching a gas phase mobilization threshold. Consequently, associated reductions in Kqs were restricted to a factor of 2 or less, but still appeared to alter the groundwater flow field. While trapped gas increases within the biodegradation plume were expected, declines in gas saturations were also observed. Thus, the overall pattern of trapped gas growth exhibited high spatial and temporal variability. Influencing factors included changes in hydrocarbon inputs and microbial controls on redox zonation, in addition to ebullition and changes in groundwater flow; emphasizing that gas phase growth in contaminant plumes will be highly complex and dynamic in the natural systems. Given the impacts on hydraulic conductivity, and the fate and transport of volatile compounds, an improved understanding of quasi-saturated conditions will be beneficial for various groundwater applications, from recharge and paleoclimate studies to site characterizations and remediation strategies. / Dissertation / Doctor of Philosophy (PhD)