William A. Perkins

Pore-Scale and Multiscale Numerical Simulation of Flow and Transport in a Laboratory-Scale Column

Pore-scale models are useful for studying relationships between fundamental processes and phenomena at larger (i.e., Darcy) scales. However, the size of domains that can be simulated with explicit pore-scale resolution is limited by computational and observational constraints. Direct numerical simulation of pore-scale flow and transport is typically performed on millimeter-scale volumes at which X-ray computed tomography (XCT), often used to characterize pore geometry, can achieve micrometer resolution. In contrast, laboratory experiments that measure continuum properties are typically performed on decimeter-scale columns. At this scale, XCT resolution is coarse (tens to hundreds of micrometers) and prohibits characterization of small pores and grains. We performed simulations of pore-scale processes over a decimeter-scale volume of natural porous media with a wide range of grain sizes, and compared to results of column experiments using the same sample. Simulations were conducted using high-performance codes executed on a supercomputer. Two approaches to XCT image segmentation were evaluated, a binary (pores and solids) segmentation and a ternary segmentation that resolved a third category (porous solids with pores smaller than the imaged resolution). We used a multiscale Stokes-Darcy simulation method to simulate the combination of Stokes flow in large open pores and Darcy-like flow in porous solid regions. Flow and transport simulations based on the binary segmentation were inconsistent with experimental observations because of overestimation of large connected pores. Simulations based on the ternary segmentation provided results that were consistent with experimental observations, demonstrating our ability to successfully model pore-scale flow over a column-scale domain.

Pore-scale and continuum simulations of solute transport micromodel benchmark experiments

Four sets of nonreactive solute transport experiments were conducted with micromodels. Each set consisted of three experiments with one variable, i.e., flow velocity, grain diameter, pore-aspect ratio, and flow-focusing heterogeneity. The data sets were offered to pore-scale modeling groups to test their numerical simulators. Each set consisted of two learning experiments, for which all results were made available, and one challenge experiment, for which only the experimental description and base input parameters were provided. The experimental results showed a nonlinear dependence of the transverse dispersion coefficient on the Peclet number, a negligible effect of the pore-aspect ratio on transverse mixing, and considerably enhanced mixing due to flow focusing. Five pore-scale models and one continuum-scale model were used to simulate the experiments. Of the pore-scale models, two used a pore-network (PN) method, two others are based on a lattice Boltzmann (LB) approach, and one used a computational fluid dynamics (CFD) technique. The learning experiments were used by the PN models to modify the standard perfect mixing approach in pore bodies into approaches to simulate the observed incomplete mixing. The LB and CFD models used the learning experiments to appropriately discretize the spatial grid representations. For the continuum modeling, the required dispersivity input values were estimated based on published nonlinear relations between transverse dispersion coefficients and Peclet number. Comparisons between experimental and numerical results for the four challenge experiments show that all pore-scale models were all able to satisfactorily simulate the experiments. The continuum model underestimated the required dispersivity values, resulting in reduced dispersion. The PN models were able to complete the simulations in a few minutes, whereas the direct models, which account for the micromodel geometry and underlying flow and transport physics, needed up to several days on supercomputers to resolve the more complex problems.

Two-Dimensional Modeling of Time-Varying Hydrodynamics and Juvenile Chinook Salmon Habitat in the Hanford Reach of the Columbia River

The Hanford Reach is the only remaining unimpounded reach of the Columbia River in the United States above Bonneville Dam. Discharge in the Hanford Reach is regulated by several dams and is often subject to rapid changes. Sharp flow reductions have led to the stranding or entrapment, and subsequent mortality, of juvenile chinook salmon (Oncorynchus tshawytscha) and other important fish species within the Hanford Reach. A multi-block two-dimensional depth-averaged hydrodynamic model was used to simulate time-varying river velocity and stage in a 37~km portion of the Hanford Reach. Simulation results were used to estimate time-varying juvenile chinook salmon habitat area, and the part of that habitat affected by discharge fluctuations. Affected habitat area estimates were made for the chinook salmon rearing period of four years. These estimates were used, along with other important factors, to establish a statistical relationship between discharge fluctuation and juvenile chinook salmon mortality.

Efficient calculation of dewatered and entrapped areas using hydrodynamic modeling and GIS

River waters downstream of a hydroelectric project are often subject to rapidly changing discharge. Abrupt decreases in discharge can quickly dewater and expose some areas and isolate other areas from the main river channel, potentially stranding or entrapping fish, which often results in mortality. A methodology is described to estimate the areas dewatered or entrapped by a specific reduction in upstream discharge and applied in a case study. A one-dimensional hydrodynamic model was used to simulate steady flows. Using flow simulation results from the model and a geographic information system (GIS), estimates of dewatered and entrapped areas were made for a wide discharge range. The methodology was applied to the Hanford Reach of the Columbia River in central Washington State. Results showed that a 280 m^3/s discharge reduction affected the most area at discharges less than 3400 m^3/s. At flows above 3400 m^3/s, the affected area by a 280 m^3/s discharge reduction (about 25 ha) was relatively constant. A 280 m^3/s discharge reduction at lower flows affected about twice as much area. The methodology and resulting area estimates have been used to identify discharge regimes, and associated water surface elevations, that might be expected to minimize adverse impacts on juvenile fall chinook salmon (Oncorhynchus tshawytscha) that rear in the shallow near-shore areas in the Hanford Reach.

GridPACK ™ : a framework for developing power grid simulations on high performance computing platforms

This paper describes the GridPACKTM framework, which is designed to help power grid engineers develop modeling software capable of running on high performance computers. The framework makes extensive use of software templates to provide high level functionality while at the same time allowing developers the freedom to express whatever models and algorithms they are using. GridPACKTM contains modules for setting up distributed power grid networks, assigning buses and branches with arbitrary behaviors to the network, creating distributed matrices and vectors and using parallel linear and non-linear solvers to solve algebraic equations. It also provides mappers to create matrices and vectors based on properties of the network and functionality to support IO and to manage errors. The goal of GridPACKTM is to substantially reduce the complexity of writing software for parallel computers while still providing efficient and scalable software solutions. The use of GridPACKTM is illustrated for a simple powerflow example and performance results for powerflow and dynamic simulation are discussed.

Two-Dimensional Simulation of Hydrodynamics, Water Quality, and Fish Exposure in the Columbia/Snake River System

The U.S. Army Corps of Engineers (USACE) Dissolved Gas Abatement Program (DGAS) is studying options to reduce dissolved gas supersaturation (DGS) associated with spill at hydroelectric dams on the Lower Columbia and Snake rivers. In order to address the complex nature of DGS exposures numerical models of gas transport, gas mixing, and dynamic gas bubble trauma were developed for the Walla Walla and Portland Districts of the USACE. These models couple hydrodynamics, temperature, DGS production, DGS transport, and fish distribution information with a dynamic gas bubble trauma mortality model which will provide an estimation of cumulative mortality in fish populations passing Columbia and Snake River dams.

Simulating Collisions for Hydrokinetic Turbines. FY2010 Annual Progress Report

Computational fluid dynamics (CFD) simulations of turbulent flow and particle motion are being conducted to evaluate the frequency and severity of collisions between marine and hydrokinetic (MHK) energy devices and debris or aquatic organisms. The work is part of a collaborative research project between Pacific Northwest National Laboratory (PNNL) and Sandia National Laboratories , funded by the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy Wind and Water Power Program. During FY2010 a reference design for an axial flow MHK turbine was used to develop a computational geometry for inclusion into a CFD model. Unsteady simulations of turbulent flow and the moving MHK turbine blades are being performed and the results used for simulation of particle trajectories. Preliminary results and plans for future work are presented.

Characterizing the Fish Passage Environment at The Dalles Dam Spillway: 2001-2004

The spill environment at The Dalles Dam in 2001-2004 was characterized using a field-deployed autonomous sensor (the so-called Sensor Fish), computational fluid dynamics (CFD) modeling, and Lagrangian particle tracking. The sensor fish has a self-contained capability to digitally the record pressure and triaxial accelerations it was exposed to following its release into the spillway. After recovery downstream of the tailrace, the data stored in the memory of the sensor are downloaded and stored for analysis. The spillway, stilling basin, and tailrace hydrodynamics were simulated using an unsteady, free-surface, three-dimensional CFD code that solved the Reynolds-averaged Navier-Stokes equations in conjunction with a two-equation turbulence model. The results from the CFD simulations were then used in a Lagrangian particle tracking model that included the effects of mass, drag, and buoyancy in the particle equation of motion. A random walk method was used to simulate the effects of small-scale turbulence on the particle motion. Several operational and structural conditions were evaluated using the Sensor Fish, CFD, and particle tracking. Quantifying events such as strike and stilling basin retention time characterized exposure conditions in the spill environment.

GridPACKTM: A framework for developing power grid simulations on high-performance computing platforms

This paper describes the GridPACKTM framework, which is designed to help power grid engineers develop software capable of running on high-performance computers. The framework makes extensive use of software templates to provide high-level functionality while still providing flexibility to easily implement a broad range of models and algorithms. GridPACKTM contains modules for setting up distributed power grid networks, supporting application-specific bus and branch models, creating distributed matrices and vectors and using parallel linear and non-linear solvers. It also provides mappers to create matrices and vectors based on properties of the network and functionality to support Input/Output (IO) and to manage errors. The goal of GridPACKTM is to substantially reduce the complexity of writing software for parallel computers while still providing efficient and scalable software solutions. The use of GridPACKTM is illustrated for a simple powerflow example and performance results for the powerflow and dynamic simulations are discussed.

The GridPACK™ toolkit for developing power grid simulations on high performance computing platforms

This paper describes the GridPACK#8482; framework, which is designed to help power grid engineers develop modeling software capable of running on high performance computers. The framework contains modules for setting up distributed power grid networks, assigning buses and branches with arbitrary behaviors to the network, creating distributed matrices and vectors, using parallel linear and non-linear solvers to solve algebraic equations, and mapping functionality to create matrices and vectors based on properties of the network. In addition, the framework contains additional functionality to support IO and to manage errors. The goal of GridPACK#8482; is to provide developers with a comprehensive set of modules that can substantially reduce the complexity of writing software for parallel computers while still providing efficient and scalable software solutions.