Rayleigh-Taylor Driven Mixing in a Multiply Stratified Environment

Similar documents
Inertial-Range Dynamics and Mixing: Isaac Newton Institute for Mathematical Sciences, Cambridge, UK, 29 September to 3 October 2008

Shear instabilities in a tilting tube

Plumes and jets with time-dependent sources in stratified and unstratified environments

University of Bristol - Explore Bristol Research. Link to publication record in Explore Bristol Research PDF-document.

Buoyancy Fluxes in a Stratified Fluid

Simultaneous Velocity and Concentration Measurements of a Turbulent Jet Mixing Flow

Investigation of an implicit solver for the simulation of bubble oscillations using Basilisk

Grid-generated turbulence, drag, internal waves and mixing in stratified fluids

Atrium assisted natural ventilation of multi storey buildings

Kinematic Effects of Differential Transport on Mixing Efficiency in a Diffusively Stable, Turbulent Flow

American Physical Society Division of Fluid Dynamics Meeting, San Antonio, TX, November 23 25, 2008.

Validating the FLASH Code: Vortex Dominated Flows

Prototype Instabilities

Instabilities due a vortex at a density interface: gravitational and centrifugal effects

Donald Slinn, Murray D. Levine

Los Alamos National Laboratory Hydrodynamic Methods Applications and Research 1 LA-UR

A finite-volume algorithm for all speed flows

The Johns Hopkins Turbulence Databases (JHTDB)

Atwood number effects in buoyancy driven flows

2013 Annual Report for Project on Isopycnal Transport and Mixing of Tracers by Submesoscale Flows Formed at Wind-Driven Ocean Fronts

9 Fluid Instabilities

An Introduction to Theories of Turbulence. James Glimm Stony Brook University

Eddy viscosity. AdOc 4060/5060 Spring 2013 Chris Jenkins. Turbulence (video 1hr):

SIMULTANEOUS VELOCITY AND CONCENTRATION MEASUREMENTS OF A TURBULENT JET MIXING FLOW

Mush liquid interfaces with cross flow

Turbulent mixing with physical mass diffusion

arxiv:astro-ph/ v1 4 Mar 2004

10. Buoyancy-driven flow

NONLINEAR FEATURES IN EXPLICIT ALGEBRAIC MODELS FOR TURBULENT FLOWS WITH ACTIVE SCALARS

THE RESPONSE OF A PLUME TO A SUDDEN REDUCTION IN BUOYANCY FLUX

VORTICITY FIELD EVOLUTION IN A FORCED WAKE. Richard K. Cohn Air Force Research Laboratory Edwards Air Force Base, CA 92524

PHYS 432 Physics of Fluids: Instabilities

METHODOLOGY (3) where, x o is the heat source separation and α is the. entrainment coefficient α.

A two-fluid model of turbulent two-phase flow for simulating turbulent stratified flows

Fluctuation dynamo amplified by intermittent shear bursts

Direct numerical simulation of salt sheets and turbulence in a double-diffusive shear layer

Measurements of the three-dimensional scalar dissipation rate in gas-phase planar turbulent jets

LES of turbulent shear flow and pressure driven flow on shallow continental shelves.

Analysis of Turbulent Free Convection in a Rectangular Rayleigh-Bénard Cell

Mixing at the External Boundary of a Submerged Turbulent Jet

Double-diffusive lock-exchange gravity currents

PIV measurements of turbulence in an inertial particle plume in an unstratified ambient

TURBULENCE IN STRATIFIED ROTATING FLUIDS Joel Sommeria, Coriolis-LEGI Grenoble

PASSIVE SCALAR MIXING IN A TURBULENT JET

Turbulence: Basic Physics and Engineering Modeling

1. Comparison of stability analysis to previous work

Scattering of internal gravity waves

ANALYSIS OF IMPLICIT LES METHODS

For example, for values of A x = 0 m /s, f 0 s, and L = 0 km, then E h = 0. and the motion may be influenced by horizontal friction if Corioli

The Evolution of Large-Amplitude Internal Gravity Wavepackets

Implicit LES of Low and High Speed Flows Using High-Resolution and High-Order Methods

INTERFACIAL WAVE BEHAVIOR IN OIL-WATER CHANNEL FLOWS: PROSPECTS FOR A GENERAL UNDERSTANDING

CFD Analysis of Vented Lean Hydrogen Deflagrations in an ISO Container

Lecture 3. Turbulent fluxes and TKE budgets (Garratt, Ch 2)

meters, we can re-arrange this expression to give

arxiv:cond-mat/ v1 [cond-mat.stat-mech] 11 Dec 2002

Capillary-gravity waves: The effect of viscosity on the wave resistance

Turbulence Modeling I!

General introduction to Hydrodynamic Instabilities

2 In this paper, we give a brief summary of the front tracking algorithm for axisymmetric ow. We report simulations of spherical shock refraction by a

Natural Convection Heat Loss from A Partly Open Cubic Enclosure Timothy N Anderson 1,a * and Stuart E Norris 2,b

THE EFFECT OF SAMPLE SIZE, TURBULENCE INTENSITY AND THE VELOCITY FIELD ON THE EXPERIMENTAL ACCURACY OF ENSEMBLE AVERAGED PIV MEASUREMENTS

Effects of inclination angle on a shock-accelerated heavy gas column

Impact of numerical method on auto-ignition in a temporally evolving mixing layer at various initial conditions

Study of Forced and Free convection in Lid driven cavity problem

Internal boundary layers in the ocean circulation

Influence of Lock Aspect Ratio upon the Evolution of an Axisymmetric Intrusion

Before we consider two canonical turbulent flows we need a general description of turbulence.

Diffusive Transport Enhanced by Thermal Velocity Fluctuations

Tilting Shear Layers in Coastal Flows

Experiments on the perturbation of a channel flow by a triangular ripple

The Reynolds experiment

Application of the immersed boundary method to simulate flows inside and outside the nozzles

CHAM Case Study CFD Modelling of Gas Dispersion from a Ruptured Supercritical CO 2 Pipeline

Flux Enhancement By Shear Free Surfaces In A Turbulent Convection

Intuitive Introduction To Acoustic-gravity Waves

Roughness Sub Layers John Finnigan, Roger Shaw, Ned Patton, Ian Harman

5. 3P PIV Measurements

DNS STUDY OF TURBULENT HEAT TRANSFER IN A SPANWISE ROTATING SQUARE DUCT

28. THE GRAVITATIONAL INTERCHANGE MODE, OR g-mode. We now consider the case where the magneto-fluid system is subject to a gravitational force F g

DYNAMICS OF LIQUEFIED SEDIMENT FLOW. Advances in Natural and Technological Hazards Research Vol. 19

Relaxation methods and finite element schemes for the equations of visco-elastodynamics. Chiara Simeoni

Visualization of high-speed gas jets and their airblast sprays of cross-injected liquid

Theoretical Advances on Generalized Fractals with Applications to Turbulence

Atm S 547 Boundary Layer Meteorology

Density Field Measurement by Digital Laser Speckle Photography

Oblique shock interaction with a cylindrical density interface

Computation of turbulent natural convection with buoyancy corrected second moment closure models

Internal Wave Generation and Scattering from Rough Topography

centrifugal acceleration, whose magnitude is r cos, is zero at the poles and maximum at the equator. This distribution of the centrifugal acceleration

Boundary-Layer Theory

An evaluation of a conservative fourth order DNS code in turbulent channel flow

Dimensionality influence on energy, enstrophy and passive scalar transport.

Numerical Simulations of a Stratified Oceanic Bottom Boundary Layer. John R. Taylor - MIT Advisor: Sutanu Sarkar - UCSD

Colloquium FLUID DYNAMICS 2012 Institute of Thermomechanics AS CR, v.v.i., Prague, October 24-26, 2012 p.

Probability density function (PDF) methods 1,2 belong to the broader family of statistical approaches

Experiments at the University of Minnesota (draft 2)

Contents. Parti Fundamentals. 1. Introduction. 2. The Coriolis Force. Preface Preface of the First Edition

Fluid Animation. Christopher Batty November 17, 2011

VERTICAL TURBULENT BUOYANT HELIUM JET CFD MODELING AND VALIDATION

Transcription:

Rayleigh-Taylor Driven Mixing in a Multiply Stratified Environment Abstract Andrew George Weir Lawrie and Stuart Dalziel Department of Applied Mathematics and Theoretical Physics, University of Cambridge A.G.W.Lawrie@damtp.cam.ac.uk S.Dalziel@damtp.cam.ac.uk Mixing is ubiquitous in both the natural environment and industrial applications, and its consequences are far-reaching. This paper focuses on the details of the small scale mixing processes which are driven by buoyancy in a gravitational field. In particular we explore the evolution of an interleaved heavy-light-heavy miscible liquid system with one Rayleigh- Taylor unstable density interface, and one statically stable interface. Experiments are performed using LIF illumination of a chemically reactive but dynamically passive tracer, and we seek to quantify mixing induced across both interfaces. In a complementary numerical study, adaptive mesh refinement is used to target computational capacity at the small scales in the region surrounding the stable density interface. It is of particular interest whether simulations accurately capture mixing across stable density interfaces in liquid systems, when frequently such codes operate with Schmidt numbers of order unity. 1. Introduction The driving instability in the present study has been a focus for scientific curiosity since Rayleigh (1883) when the problem of dense fluid above less dense fluid in a gravitational field was first considered. Such instability is observed in a variety of fluids with applications ranging from geophysical and astrophysical to industrial fluid systems. Understanding the suppression of turbulent mixing by stable density stratification also has implications for geophysical systems and potential industrial application, and the interleaved heavy-light-heavy problem configuration (Jacobs and Dalziel (2005)) considered here is an appropriate case study, since turbulence generated by the development of the unstable interface induces mixing across the stable interface. The ability of modern numerical techniques to capture the flows with such varied mixing characteristics is not yet confirmed, and this is a focus of current investigation. 1.1. Computational Approach Despite the approximately exponential growth in computing power over time, numerical simulation of Rayleigh-Taylor instability remains a challenging problem. An approach called Implicit Large Eddy Simulation (ILES) is used in this paper, with the code of Almgren et al. (1998). Historically, the approach originated from the surprisingly successful application of modern numerical methods for compressible gas dynamics to problems involving intense mixing. Total Variation Diminishing (TVD) methods (eg. Leveque (1992)) were developed as a (non-linear) means of controlling truncation error terms in numerical algorithms. Dispersive error (odd order truncation error terms) manifests itself as spurious oscillations around discontinuities and high gradients. Such errors could be eliminated by using a numerical scheme which interpolates, using some empirically derived function,

between low and high order. Increased diffusive error (even order truncation error terms), although undesirable in regions of high gradient, is accepted as a consequence, since it has a physical analogue. The physical relevance of diffusion associated with high gradients motivated the change in application from wave-dominated to highly turbulent flows. It is well known that high gradients are smoothed by viscous effects in a real flow, and the observation that TVD methods also smoothed gradients led to the birth of ILES. Without explicitly applying viscous terms, real flows could be simulated with plausible results (eg. Youngs (1994); Dalziel et al. (1999); Ramaprabhu et al. (2005)). Effectively, the numerical error replaces a conventional eddy-viscosity sub-grid scale model, and performs a similar function. However, a better understanding of the numerical analysis of ILES for mixing problems is only now being developed (Margolin et al. (2006)). 1.2. Experimental Method While in a broad range of fields much experimental work has focussed on the physics of mixing, obtaining direct visualisation of the mixing process is much less common. Light/Laser-Induced Fluorescence (LIF) techniques have been used in the past for this purpose in miscible liquid shear flows, eg. Koochesfahani and Dimotakis (1985), and more recently also a hybrid fluorescence/phosphorescence technique in gaseous shear flows (Hu and Koochesfahani (2002)), but in Rayleigh-Taylor driven mixing there seem not to be comparable experiments. Experiments using ph indicators in the Rayleigh-Taylor context (Linden et al. (1994)) to mark mixed fluid have been performed before; the present work extends this concept by using a fluorescent ph indicator, hence permitting detailed LIF visualisation of the actual mixing within the mixing region and furthering understanding of its structure. The experiments in the current study were conducted using developments of apparatus and techniques which have been applied previously by Dalziel (1993),Dalziel et al. (1999) and Jacobs and Dalziel (2005). The principal feature of the experimental rig is a horizontal barrier to provide initial separation of fluids across an unstable interface. Using a technique devised by Lane-Serff (1989), the barrier can be removed from the tank without applying the resulting shear to the fluid. The imaging technique used in this study is Light-Induced Fluorescence, using xenon arc lamps to provide incident light. The fluorescent dye used in these experiments is 2,3,5,6-dibenzo pyridine, which absorbs light in the ultraviolet wavelength range, but its emission wavelength is ph dependent. Above ph 5 (Weast (1971)), the emission is violet blue, but below ph 5 the emission is cyan-green. By appropriately filtering the emitted light before reaching the CCD camera, a signal is only received in areas of mixed fluid. The light sheet is thin relative to the dynamics of the flow, so the internal structure of the mixing in a two-dimensional crossection can be examined. The ratio of advection to reaction timescales is very large, so the reaction plays no dynamic role in the experiment. Acid is added to one (salt solution) fluid layer, the fluorescent dye to another. Refractive indices are matched using alcohol. The density is set by the combination of acid and salt. 2. Observations on Rayleigh-Taylor mixing As a starting point for validation of the numerical simulation, detailed experimental work was focussed on the Rayleigh-Taylor component of the flow, since ILES has been found to

Figure 1: Time series charting the flow evolution in experiment and simulation, using a comparable diagnostic. The white bar indicates the progress of nondimensional simulation time from 0 to 5. The experimental time origin has been shifted to account for the initialisation differences between experiment and simulation. work well for this problem. Direct comparison of experiment and simulation is not strictly appropriate because there are significant unsimulated effects in the experiments, such as asymmetries and large scale motions introduced in the experiment by the barrier removal, and discrepancies due to the random perturbation which initialises the simulation. However, a virtual time origin can be derived from the comparison of growth profiles. The experimental results have been spatially filtered down to the same resolution as the simulation, and an equivelant diagnostic used to present the simulations. A dual time-sequence of images in figure 1 compares the evolution of the flow. As is evident from the figure, the simulation does not explicitly capture the smallest scales of the flow that are visible even in the filtered experimental time series. However, the mesh size (160x80x200) in the simulation is chosen such that the energy dissipation by numerical error (the implicit sub-grid scale model) reasonably approximates the energy dissipation by turbulent diffusion (nature s sub-grid scale model). Thus the salient mixing statistics could be expected to compare well between experiment and simulation. A good indication of the rate of mixing can be obtained by considering the surface area over which mixing takes place, and the enclosed volume of mixed fluid which grows in time as the mixing front advances. The planar analogue of the surface area (the length by pixel count of the enclosing contour) is shown in 2 and is defined as the boundary of a region where pixel intensity is greater than a threshold. The corresponding volume is shown in figure 3. As before, the processing is performed on the filtered experimental data, for valid comparison with the simulation. The threshold is chosen for convenience, to avoid contaminating the image analysis with digital noise. However, signal intensity (in the absence of other effects) is linear in the local dye concentration, so as further mixing takes place towards later times, the signal

Figure 2: Evolution of mixing surface area in time. The length of the 2-D crossection through the surface is measured by pixel count. Close agreement between experiment and simulation is achieved. Figure 3: Evolution of mixed volume in time. The independent variable is measured as a fraction of total area. Agreement is good until the dilution of fluid in the experiment is such that recovered signal intensity is of the order of the noise threshold.

Figure 4: Numerical simulation of fluorescent dye behaviour when visualising mixing across the stable density interface. The orthogonal lines show a crossection of the grid patches used in the simulation. intensity of this mixed fluid reduces, until eventually falls below the noise threshold. Since this effect is not include in the simulation diagnostic, the experimental and simulation curves diverge at late time. 3. Mixing across the stable interface Adaptive Mesh Refinement is a numerical technique which enables arbitrarily selected regions of a flow to be investigated in considerably more detail, without the associated computational cost of resolving to this level in all regions. The mixed fluid around the stable interface is used to mark the region over which more detail is required, and the resulting region is shown in figure 4. As identified by George et al. (2002) the numerical diffusion inherent by cell averaging conserved quantities in finite volume codes applies not only to momentum, but also to density and tracer advection. This implies that ILES models have a numerical Schmidt number ( µ κ) of O (1). While this is close to the physical Schmidt number in gasses, the values for liquids are frequently two orders of magnitude higher. Quantification of the error is achieved by comparing a simulation which locally well resolves the dynamical scales and in which the Schmidt number is explicitly set by computing the diffusion terms, with a standard ILES simulation. 4. Conclusions An experimental technique has been presented for directly visualising regions of molecularly mixed flow in miscible liquid flows. Particular attention has been paid to the early time evolution of the Rayleigh-Taylor instability, obtaining quantitative measurements from a novel perspective on a widely studied problem. The present study develops previous work on chemical indicators of molecular mixing by gaining a first insight into the detailed structure of the mixing region. ILES numerical simulations have been performed which show comparable mixing behaviour to the experiments, when considering fractal dimension, mixing surface and mixed volume. Adaptive Mesh Refinement techniques are used to simulate in more detail the mixing across the stable interface in the heavy-lightheavy configuration.

Acknowledgements The authors would like to thank the UK Natural Environment Research Council (NERC) as primary funding source, and the Co-operative Awards in Science and Engineering (CASE) collaborator, AWE Ltd. Useful discussions with John Bell, Ann Almgren and Nikos Nikiforakis are gratefully acknowledged. References Almgren, A. S., Bell, J. B., Colella, P., Howell, L. H., and Welcome, M. L. (1998). A conservative adaptive projection method for the variable density incompressible Navier- Stokes equations. J. Comp. Phys., 142:1 46. Dalziel, S. B. (1993). Rayleigh-Taylor instability: experiments with image analysis. Dyn. Atmos. Oceans, 20:127 153. Dalziel, S. B., Linden, P. F., and Youngs, D. L. (1999). Self-similarity and internal structure of turbulence induced by Rayleigh-Taylor instability. J. Fluid Mech., 399:1 48. George, E., Glimm, J., Li, X.-L., Marchese, A., and Xu, Z.-L. (2002). A comparison of experimental, theoretical and numerical simulation Rayleigh-Taylor mixing rates. Proc. Natl. Acad. Sci., 99:2587 2592. Hu, H. and Koochesfahani, M. M. (2002). A novel method for instantaneous, quantitative measurement of molecular mixing in gaseous flows. Exp. Fluids, 31:202 209. Jacobs, J. W. and Dalziel, S. B. (2005). Rayleigh-Taylor instability in complex stratifications. J. Fluid Mech., 542:251 279. Koochesfahani, M. M. and Dimotakis, P. E. (1985). Mixing and chemical reactions in a turbulent liquid mixing layer. J. Fluid Mech., 170:83 112. Lane-Serff, G. F. (1989). Heat Flow and air movement in buildings. PhD thesis, DAMTP, University of Cambridge, UK. Leveque, R., editor (1992). Numerical Methods for Conservation Laws. Birkhauser. Linden, P. F., Redondo, J. M., and Youngs, D. L. (1994). Molecular mixing in Rayleigh- Taylor instability. J. Fluid Mech., 265:97 124. Margolin, L. G., Rider, W. J., and Grinstein, F. F. (2006). Modeling turbulent flow with implicit LES. J. Turbulence, 7:1 27. Ramaprabhu, P., Dimonte, G., and Andrews, M. J. (2005). A numerical study of the influence of initial perturbations on the turbulent Rayleigh-Taylor instability. J. Fluid Mech., 536:285 319. Rayleigh, L. (1883). Investigation of the character of the equilibrium of an incompressible heavy fluid of variable density. Proc. Lond. Math. Soc., XIV:70 177. Weast, R. C., editor (1971). Handbook of Chemistry and Physics. CRC Press, 52 edition. Youngs, D. L. (1994). Numerical simulation of mixing by Rayleigh-Taylor and Richtmyer- Meshkov instabilities. Laser and Particle Beams, 12:725 750.