Publications

Abstract not provided.

Investigation of turbulent mixing in pipe joints has been a topic of recent research interest. These investigations have relied on experimental results with downstream sensors to determine the bulk characteristics of mixing in pipe joints. High fidelity computational fluid dynamics models have also been employed to examine the fine scale physics of the mixing within the joint geometry. To date, high resolution imaging of experimental conditions within the pipe joint has not been reported. Here, we introduce high speed photography as a tool to accomplish this goal. Cross joints with four pipes coming together in a single junction are the focus of this investigation. All pipes entering the junction are the same diameter and made of clear PVC. The cross joint was milled from clear acrylic material to allow for high resolution imaging of the mixing processes within the joint. Two pipes carry water into the joint, one with clear water and the other inlet with water containing dye and a salt tracer. Two outlet pipes are carry water away from the joint. A high-speed digital camera was used to image mixing within the joint at an imaging rate of 30 Hz. Each grey-scale (8-bit) image is 1280 x 1024 pixels in a roughly 17.8 x 14.5 cm image containing the cross joint. The pixel size is approximately 0.13 x 0.14 mm. Four experiments using the clear cross-joint have been visualized. The Reynolds number (Re) for the tracer inlet pipe is held constant at 1500, while a different Re in the clear inlet pipe is used for each experiment. The Re value in the outlets are held equal to each other at the average Re of the inlets. Re values in the clear inlet pipe values are: 500, 1000, 2000 and 5000. Visual examination of the images provides information on the mixing behavior including tracer transport along the walls of the pipe, transient variation in the amount of tracer entering each outlet, the sharpness of the clear-tracer interface and variation in the concentration of the tracer throughout the joint geometry. A sharp tracer-clear interface is visible for the clear inlet Re values of 500, 1000 and 2000, but decays to a broad gradual transition zone at a clear inlet Re of 5000. There are no visible instabilities in the clear-tracer interface at the lowest clear water Re (500), but regular periodic instabilities occur for the Re=1000 experiment and these become irregular, but still periodic at clear inlet Re = 2000 and then lose all regular structure in the Re = 5000 experiment. High speed photography applied to clear pipe joints with the necessary image processing can provide qualitative and quantitative insights into mixing processes. A limitation of this approach is that it provides two-dimensional images of a three-dimensional process. ©ASCE 2009.

This report summarizes the experimental and modeling effort undertaken to understand solute mixing in a water distribution network conducted during the last year of a 3-year project. The experimental effort involves measurement of extent of mixing within different configurations of pipe networks, measurement of dynamic mixing in a single mixing tank, and measurement of dynamic solute mixing in a combined network-tank configuration. High resolution analysis of turbulence mixing is carried out via high speed photography as well as 3D finite-volume based Large Eddy Simulation turbulence models. Macroscopic mixing rules based on flow momentum balance are also explored, and in some cases, implemented in EPANET. A new version EPANET code was developed to yield better mixing predictions. The impact of a storage tank on pipe mixing in a combined pipe-tank network during diurnal fill-and-drain cycles is assessed. Preliminary comparison between dynamic pilot data and EPANET-BAM is also reported.

An experimental study was performed to determine the fraction of the heat flux that is due to radiation (sometimes referred to as radiation partitioning of the total heat flux measurement) to a calorimeter engulfed in a large methanol pool fire to improve understanding and develop high-quality data for the validation of fire models. Diagnostics employed include Coherent Anti-Stokes Raman Spectroscopy (CARS), Particle Image Velocimetry (PIV), total and radiative thermometry, and thermocouples. Data are presented not only for the physics measurements but also for all initial and boundary conditions required as necessary inputs to computational models. The large physical scale, the experimental design (enhanced convection relative to radiation heat transfer), the use of independent measurement techniques, and the attention to data quality, provide a unique dataset that emphasizes the convective component to support numerical fire model validation for convective and radiative heat transfer in fires. © 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

An experimental program was conducted to study the multiphase gas-solid flow in a pilot-scale circulating fluidized bed (CFB). This report describes the CFB experimental facility assembled for this program, the diagnostics developed and/or applied to make measurements in the riser section of the CFB, and the data acquired for several different flow conditions. Primary data acquired included pressures around the flow loop and solids loadings at selected locations in the riser. Tomographic techniques using gamma radiation and electrical capacitance were used to determine radial profiles of solids volume fraction in the riser, and axial profiles of the integrated solids volume fraction were produced. Computer Aided Radioactive Particle Tracking was used to measure solids velocities, fluxes, and residence time distributions. In addition, a series of computational fluid dynamics simulations was performed using the commercial code Arenaflow{trademark}.

The effect of particle diameter on downward co-current gas-liquid flow through a fixed bed of particles confined within a cylindrical column is investigated. Several hydrodynamic regimes that depend strongly on the properties of the gas stream, the liquid stream, and the packed particle bed are known to exist within these systems. This experimental study focuses on characterizing the effect of wall confinement on these hydrodynamic regimes as the diameter d of the spherical particles becomes comparable to the column diameter D (or D/d becomes order-unity). The packed bed consists of polished, solid, spherical, monodisperse particles (beads) with mean diameter in the range of 0.64-2.54 cm. These diameters yield D/d values between 15 and 3.75, so this range overlaps and extends the previously investigated range for two-phase flow. Measurements of the pressure drop across the bed and across the pulses are obtained for varying gas and liquid flow rates. Copyright © 2005 by ASME.

Threats to water distribution systems include release of contaminants and Denial of Service (DoS) attacks. A better understanding, and validated computational models, of the flow in water distribution systems would enable determination of sensor placement in real water distribution networks, allow source identification, and guide mitigation/minimization efforts. Validation data are needed to evaluate numerical models of network operations. Some data can be acquired in real-world tests, but these are limited by 1) unknown demand, 2) lack of repeatability, 3) too many sources of uncertainty (demand, friction factors, etc.), and 4) expense. In addition, real-world tests have limited numbers of network access points. A scale-model water distribution system was fabricated, and validation data were acquired over a range of flow (demand) conditions. Standard operating variables included system layout, demand at various nodes in the system, and pressure drop across various pipe sections. In addition, the location of contaminant (salt or dye) introduction was varied. Measurements of pressure, flowrate, and concentration at a large number of points, and overall visualization of dye transport through the flow network were completed. Scale-up issues that that were incorporated in the experiment design include Reynolds number, pressure drop across nodes, and pipe friction and roughness. The scale was chosen to be 20:1, so the 10 inch main was modeled with a 0.5 inch pipe in the physical model. Controlled validation tracer tests were run to provide validation to flow and transport models, especially of the degree of mixing at pipe junctions. Results of the pipe mixing experiments showed large deviations from predicted behavior and these have a large impact on standard network operations models.3

A turbulence model for buoyant flows has been developed in the context of a k-{var_epsilon} turbulence modeling approach. A production term is added to the turbulent kinetic energy equation based on dimensional reasoning using an appropriate time scale for buoyancy-induced turbulence taken from the vorticity conservation equation. The resulting turbulence model is calibrated against far field helium-air spread rate data, and validated with near source, strongly buoyant helium plume data sets. This model is more numerically stable and gives better predictions over a much broader range of mesh densities than the standard k-{var_epsilon} model for these strongly buoyant flows.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Contaminant dispersal models for use at scales ranging from meters to miles are widely used for planning sensor locations, first-responder actions for release scenarios, etc. and are constantly being improved. Applications range from urban contaminant dispersal to locating buried targets from an exhaust signature. However, these models need detailed data for model improvement and validation. A small Sandia National Laboratories Laboratory Directed Research and Development (LDRD) program was funded in FY04 to examine the feasibility and usefulness of a scale-model capability for quantitative characterization of flow and contaminant dispersal in complex environments. This report summarizes the work performed in that LDRD. The basics of atmospheric dispersion and dispersion modeling are reviewed. We examine the need for model scale data, and the capability of existing model test methods. Currently, both full-scale and model scale experiments are performed in order to collect validation data for numerical models. Full-scale experiments are expensive, are difficult to repeat, and usually produce relatively sparse data fields. Model scale tests often employ wind tunnels, and the data collected is, in many cases, derived from single point measurements. We review the scaling assumptions and methods that are used to relate model and full scale flows. In particular, we examine how liquid flows may be used to examine the process of atmospheric dispersion. The scaling between liquid and gas flows is presented. Use of liquid as the test fluid has some advantages in terms of achieving fully turbulent Reynolds numbers and in seeding the flow with neutrally buoyant tracer particles. In general, using a liquid flow instead of a gas flow somewhat simplifies the use of full field diagnostics, such as Particle Image Velocimetry and Laser Induced Fluorescence. It is also possible to create stratified flows through mixtures of fluids (e.g., water, alcohol, and brine). Lastly, we describe our plan to create a small prototype water flume for the modeling of stratified atmospheric flows around complex objects. The incoming velocity profile could be tailored to produce a realistic atmospheric boundary layer for flow-in-urban-canyon measurements. The water tunnel would allow control of stratification to produce, for example, stable and unstable atmospheric conditions. Models ranging from a few buildings to cityscapes would be used as the test section. Existing noninvasive diagnostics would be applied, including particle image velocimetry for detailed full-field velocity measurement, and laser induced fluorescence for noninvasive concentration measurement. This scale-model facility will also be used as a test-bed for data acquisition and model testing related to the inverse problem, i.e., determination of source location from distributed, sparse measurement locations. In these experiments the velocity field would again be measured and data from single or multiple concentration monitors would be used to locate the continuous or transient source.

Abstract not provided.

Abstract not provided.

Abstract not provided.

An experimental program was conducted to study a proposed approach for oil reintroduction in the Strategic Petroleum Reserve (SPR). The goal was to assess whether useful oil is rendered unusable through formation of a stable oil-brine emulsion during reintroduction of degassed oil into the brine layer in storage caverns. An earlier report (O'Hern et al., 2003) documented the first stage of the program, in which simulant liquids were used to characterize the buoyant plume that is produced when a jet of crude oil is injected downward into brine. This report documents the final two test series. In the first, the plume hydrodynamics experiments were completed using SPR oil, brine, and sludge. In the second, oil reinjection into brine was run for approximately 6 hours, and sampling of oil, sludge, and brine was performed over the next 3 months so that the long-term effects of oil-sludge mixing could be assessed. For both series, the experiment consisted of a large transparent vessel that is a scale model of the proposed oil-injection process at the SPR. For the plume hydrodynamics experiments, an oil layer was floated on top of a brine layer in the first test series and on top of a sludge layer residing above the brine in the second test series. The oil was injected downward through a tube into the brine at a prescribed depth below the oil-brine or sludge-brine interface. Flow rates were determined by scaling to match the ratio of buoyancy to momentum between the experiment and the SPR. Initially, the momentum of the flow produces a downward jet of oil below the tube end. Subsequently, the oil breaks up into droplets due to shear forces, buoyancy dominates the flow, and a plume of oil droplets rises to the interface. The interface was deflected upward by the impinging oil-brine plume. Videos of this flow were recorded for scaled flow rates that bracket the equivalent pumping rates in an SPR cavern during injection of degassed oil. Image-processing analyses were performed to quantify the penetration depth and width of the oil jet. The measured penetration depths were shallow, as predicted by penetration-depth models, in agreement with the assumption that the flow is buoyancy-dominated, rather than momentum-dominated. The turbulent penetration depth model overpredicted the measured values. Both the oil-brine and oil-sludge-brine systems produced plumes with hydrodynamic characteristics similar to the simulant liquids previously examined, except that the penetration depth was 5-10% longer for the crude oil. An unexpected observation was that centimeter-size oil 'bubbles' (thin oil shells completely filled with brine) were produced in large quantities during oil injection. The mixing experiments also used layers of oil, sludge, and brine from the SPR. Oil was injected at a scaled flow rate corresponding to the nominal SPR oil injection rates. Injection was performed for about 6 hours and was stopped when it was evident that brine was being ingested by the oil withdrawal pump. Sampling probes located throughout the oil, sludge, and brine layers were used to withdraw samples before, during, and after the run. The data show that strong mixing caused the water content in the oil layer to increase sharply during oil injection but that the water content in the oil dropped back to less than 0.5% within 16 hours after injection was terminated. On the other hand, the sediment content in the oil indicated that the sludge and oil appeared to be well mixed. The sediment settled slowly but the oil had not returned to the baseline, as-received, sediment values after approximately 2200 hours (3 months). Ash content analysis indicated that the sediment measured during oil analysis was primarily organic.

An experimental program is being conducted to study a proposed approach for oil reintroduction in the Strategic Petroleum Reserve (SPR). The goal is to assess whether useful oil is rendered unusable through formation of a stable oil-brine emulsion during reintroduction of degassed oil into the brine layer in storage caverns. This report documents the first stage of the program, in which simulant liquids are used to characterize the buoyant plume that is produced when a jet of crude oil is injected downward from a tube into brine. The experiment consists of a large transparent vessel that is a scale model of the proposed oil injection process at the SPR. An oil layer is floated on top of a brine layer. Silicon oil (Dow Corning 200{reg_sign} Fluid, 5 cSt) is used as the simulant for crude oil to allow visualization of the flow and to avoid flammability and related concerns. Sodium nitrate solution is used as the simulant for brine because it is not corrosive and it can match the density ratio between brine and crude oil. The oil is injected downward through a tube into the brine at a prescribed depth below the oil-brine interface. Flow rates are determined by scaling to match the ratio of buoyancy to momentum between the experiment and the SPR. Initially, the momentum of the flow produces a downward jet of oil below the tube end. Subsequently, the oil breaks up into droplets due to shear forces, buoyancy dominates the flow, and a plume of oil droplets rises to the interface. The interface is deflected upward by the impinging oil-brine plume. Two different diameter injection tubes were used (1/2-inch and 1-inch OD) to vary the scaling. Use of the 1-inch injection tube also assured that turbulent pipe flow was achieved, which was questionable for lower flow rates in the 1/2-inch tube. In addition, a 1/2-inch J-tube was used to direct the buoyant jet upwards rather than downwards to determine whether flow redirection could substantially reduce the oil-plume size and the oil-droplet residence time in the brine. Reductions of these quantities would inhibit emulsion formation by limiting the contact between the oil and the brine. Videos of this flow were recorded for scaled flow rates that bracket the equivalent pumping rates in an SPR cavern. Image-processing analyses were performed to quantify the penetration depth of the oil jet, the width of the jet, and the deflection of the interface. The measured penetration depths are shallow, as predicted by penetration-depth models, in agreement with the assumption that the flow is buoyancy-dominated, rather than momentum-dominated. The turbulent penetration depth model provided a good estimate of the measured values for the 1-inch injection tube but overpredicted the penetration depth for the 1/2-inch injection tube. Adding a virtual origin term would improve the prediction for the 1/2-inch tube for low to nominal injection flow rates but could not capture the rollover seen at high injection flow rates. As expected, the J-tube yielded a much narrower plume because the flow was directed upward, unlike the downward-oriented straight-tube cases where the plume had to reverse direction, leading to a much wider effective plume area. Larger surface deflections were caused by the narrower plume emitted from the J-tube. Although velocity was not measured in these experiments, the video data showed that the J-tube plume was clearly faster than those emitted from the downward-oriented tubes. These results indicate that oil injection tube modifications could inhibit emulsion formation by reducing the amount of contact (both time and area) between the oil and the brine. Future studies will employ crude oil, saturated brine, and interfacial solids (sludge) from actual SPR caverns.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Experiments are presented in which electrical-impedance tomography (EIT) and gamma-densitometry tomography (GDT) measurements were combined to simultaneously measure the solid, liquid, and gas radial distributions in a vertical three-phase flow. The experimental testbed was a 19.05-cm diameter bubble column in which gas is injected at the bottom and exits out the top while the liquid and solid phases recirculate. The gas phase was air and the liquid phase was deionized water with added electrolytes. Four different particle classes were investigated for the solid phase: 40--100 {micro}m and 120--200 {micro}m glass beads (2.41 g/cm{sup 3}), and 170--260 {micro}m and 200--700 {micro}m polystyrene beads (1.04 g/cm{sup 3}). Superficial gas velocities of 3 to 30 cm/s and solid volume fractions up to 0.30 were examined. For all experimental conditions investigated, the gas distribution showed only a weak dependence on both particle size and density. Average gas volume fraction as a function of superficial gas velocity can be described to within {+-} 0.04 by curve passing through the center of the data. For most cases the solid particle appeared to be radically uniformly dispersed in the liquid.

An electrical-impedance tomography (EIT) system has been developed for quantitative measurements of radial phase distribution profiles in two-phase and three-phase vertical column flows. The EIT system is described along with the computer algorithm used for reconstructing phase volume fraction profiles. EIT measurements were validated by comparison with a gamma-densitometry tomography (GDT) system. The EIT system was used to accurately measure average solid volume fractions up to 0.05 in solid-liquid flows, and radial gas volume fraction profiles in gas-liquid flows with gas volume fractions up to 0.15. In both flows, average phase volume fractions and radial volume fraction profiles from GDT and EIT were in good agreement. A minor modification to the formula used to relate conductivity data to phase volume fractions was found to improve agreement between the methods. GDT and EIT were then applied together to simultaneously measure the solid, liquid, and gas radial distributions within several vertical three-phase flows. For average solid volume fractions up to 0.30, the gas distribution for each gas flow rate was approximately independent of the amount of solids in the column. Measurements made with this EIT system demonstrate that EIT may be used successfully for noninvasive, quantitative measurements of dispersed multiphase flows.