John M. Budzinski, Robert F. Benjamin and Jeffrey W. Jacobs- Influence of initial conditions on the flow patterns of a shock-accelerated thin fluid layer

  • Upload
    ponmij

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

  • 8/3/2019 John M. Budzinski, Robert F. Benjamin and Jeffrey W. Jacobs- Influence of initial conditions on the flow patterns of a

    1/3

    Influence of initial conditions on the flow patterns of a shock-acceleratedthin fluid layer

    John M. Budzinski and Robert F. BenjaminDynamic Experimentation Division, Los Alamos National Labo ratory Los Alamos, New Mexico 87545

    Jeffrey W. JacobsDepartment of Aerospace and Mechanical Engineering, University of Arizona, lkson, Arizona 85 721

    (Received 5 May 1994; accepted 22 July 1994)

    Previous observations of three flow patterns generated by shock acceleration of a thin perturbed,fluid layer are now correlated with asymmetries in the initial conditions. Using a different diagnostic(planar laser Rayleigh scattering) than the previous experiments, upstream mushrooms, downstreammushrooms, and sinuous patterns are still observed. For each experiment the initial perturbationamplitude on one side of the layer can either be larger, smaller, or the same as the amplitude on theother side, as observed with two images per experiment, and these differences lead to the formationof the different patterns.

    This study is directed toward a better understanding ofthe flow of an accelerated heavy layer imbedded betweenlower-density fluids. As an example, this phenomenon may

    occur in inertial confinement fusion targets when the abla-tively driven impulse causes implosion of a fuel-containingshell. A potential difficulty is the mixing of shell materialwith the fuel, as a consequence of interfacial fluid instability.

    A shock-accelerated layer involves the Richtmyer-Meshkov (RM) instabilityr of each interface between fluidsof different density. A perturbed interface between two fluidssubjected to a normal, impulsive acceleration is unstable anddistorts after the interaction. If the downstream fluid isdenser than the upstream fluid the perturbation grows. How-ever, if the upstream fluid is denser than the downstreamfluid, the perturbation first reverses phase and then grows,

    and the original peaks in the perturbation will become thevalleys and the valleys will become the peaks. In both casesthe early-time growth rate is constant and proportional to theamplitude of the initial perturbation. At later times thegrowth rate decreases, and mushroom shapes form from thepeaks in the distorting interface.

    We study a shock interaction with a layer of heavy gasbounded on both sides by a lighter gas, where both interfacesof the layer have nearly sinusoidal perturbations. Previousresearch34 observed that three distinct flow patterns evolvefrom similar initial conditions, and speculated that differ-ences in initial conditions might influence which flow pattern

    occurs during each event. These experiments observed onlythe dynamic flow condition for each event at one preset timebecause of limitations that precluded multiple image acqui-sition. We now report experiments where both initial anddynamic flow conditions are recorded for each event. Thesame three flows are observed, and have a strong correlationwith asymmetries in the initial flow profile.

    The present experiments use planar laser Rayleigh scat-tering (PLRS) to capture two frames per event, but use thesame shock tube and test section as in the earlier work.3*4Our PLRS images clearly show regions of high SF, concen-tration because SF6 scatters about six times more light thanair. Iwo pulsed dye lasers illuminate the test region with

    coplanar sheets of laser light, and a cooled CCD camerarecords each image. We convert the images to maps of SF,concentration by calibrating the system with scattering from

    pure air and pure SF,.The nozzle above the test section produces a vertically

    flowing, laminar gas jet6 (i.e., the gas curtain) with a varicosecross section and diffuse boundaries; the wavelength is about6 mm and the peak-to-peak amplitude is up to 2 mm on eachside. This flow system has been shown to produce a two-dimensional flo~.~ The shock tube is fired when the gas cur-tain appears to be stable as observed with a real timeschlieren system, and drives a Mach 1.2 shock wave in air.Small, unsteady, uncontrollable fluctuations in the SF, flowcause asymmetries which vary from experiment to experi-ment and are correlated with the three different shock-

    accelerated flows. We find that the thicker regions of thelayer contain about 50% SF, and the thinner regions about40% SF,. The SF, concentration across the layer width hasan approximately Gaussian profile.

    We observe the same three patterns after a Mach 1.2shock interaction as seen in the earlier work, and find thatasymmetries in the initial condition correlate with the threepostshock flow patterns in over 90% of the lOO+ experi-ments performed. Typical examples of the initial and dy-namic conditions for the three flow patterns are shown inFig. 1, where the initial condition was taken about 100 wbefore the shock interaction and the dynamic condition at

    400-450 ,us after the shock interaction. Upstream mush-rooms formed in about 46% of the experiments, a sinuouspattern formed in about 41% of the experiments, and down-stream mushrooms developed in 13% of the experiments.Mostly we find that an upstream mushroom pattern developswhen the perturbations on the upstream side (i.e., on the sidefirst interacting with the shock wave) are larger than the per-turbations on the downstream side. A sinuous pattern formswhen the perturbations on the downstream side are about thesame or slightly larger than upstream side. A downstreammushroom develops when the perturbation on the down-stream side is much larger than the perturbation on the up-stream side. About 8% of the experimental images were

    3510 Phys. Fluids 6 (ll), November 1994 1070-6631/94/6(11)/3510/3/$6.00 @ 1994 American Institute of Physics

    Downloaded 26 Aug 2005 to 128.165.51.58. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp

  • 8/3/2019 John M. Budzinski, Robert F. Benjamin and Jeffrey W. Jacobs- Influence of initial conditions on the flow patterns of a

    2/3

    Just AfterShock

    Early time150p.WC

    Later time450psec

    FIG. 1. Composite of images from three typical experiments showing theinitial and dynamic condi tion on each experiment. The shock wave movesleft to right. Th e distance shown in the figure between the initial and dy-namic condi tion of each pair is considerably l ess than the actual distance inthe experiment. Each pair of initial and dynamic conditi on was recorded onthe same event. (a) Sinuous, dynamic image recorded at ~450 p after theshock interaction; (b) upstream mushroom, dynamic image t=450 F; (c)downstream mushr oom, dynamic image t = 400 w.

    anomalous and did not follow the correlation describedabove.

    The postshock flow pattern was classified visually as oneof the three flow patterns based on the asymmetries presen tin the spacing of the pattern lobes and the SF, mass distri-bution, using the criteria described in Fig. 2. The preciseperiodicity shown in the patterns in Fig. 2 was usually notpresent in the data; irregularities often occurred as shown inthe typical data, Fig. 1.

    When the initial conditions fell be tween those that pro-duced one of the standard patterns, typically the postshockpattern had characteristics of more than one pattern. Whenthe initial conditions had a downstream perturbation that wasnot large enough to produce a full downstream mushroom,and yet not small enough to produce a pure sinuous pattern,then by 0.45 ms after the shock interaction, well-defined

    stems and caps would not be present, but the pattern wouldhave some characteristics of the early-time downstre ammushroom pattern. For example, the dimension a in Fig. 2would be greater than b, but not to the extent as found in thedevelopment of a standard downstream mushroom. This pat-tern was categorized as one of the 41% sinuous patterns.

    We also find that the upstream mushrooms are alignedwith the thicker parts of the initial layer whereas the down-stream mushrooms are aligned with the thinner regions. Thusin the early development of downstream mushrooms, theheavy fluid moves laterally (i.e., normal to the shock direc-tion) from the thicker to the thinner regions; there is lesslateral movement of the heavy fluid during the early devel-

    MSb d

    FIG. 2. Differences in the different patterns throughout their development

    and a simple model to explain why they might occur. (a) Sinuous pattern:perturbations are about equal on both sides of the layer, at early timesa-b,at latter times c-e,d--f, and the SF6 is evenly distributed throughout thelayer. Just after the shock interaction the center of vorticity (shown by thecircular arrows) is roughly in the center of the layer inducing the sinuouspattern to develop. (b) Upstream mushroom pattern: The initial upstreamperturbation a mplitude is larger than the downstream, at early times b>a, atlater times d>e, cb, at later times df. The highest SF, concentrations and most ofthe SF, ma ss is in th e mushroom caps. Just after the shock interaction mostof the vorticity is to the right of the layer. As the vorticity entrains the fluidthe mushrooms form.

    opment of upstream mushrooms. When a sinuous patternforms the downstream lobes are aligned with the thin regionsof the initial layer and the upstream lobes are aligned withthe thicker regions.

    Our interpretation of these flow patterns is based on aone-dimensional (1-D) calculation, Richtmyer-Meshkov in-stability, and vortex dynamics. The wave dynamics of theinteraction of a planar shock wave with a layer is complexbecause of multiple wave-interface interactions. The wavesreverberate within the heavy layer after the initial interac-

    tions that produce a Mach 1.3 shock wave in the SF6 and aMach 1.17 shock in the air downstream of the layer. Theinterfaces attain the same velocity after a few reverberations.

    The observed nonlinear Bow patterns can be interpretedqualitatively as the independent RM-instability growth ofeach interface, followed by strong coupling in the nonlineargrowth regime. Richtmyers impulse model reasonably pre-dicts the early-time growth rate of a single interface (shownrecently in analyses7s). For our layer, the coupling betweeninterfaces is initially weak7 when perturbation amplitudes a resmall, so the growth of each interface occurs independently.The impulse model predicts that the ratio of growth rates of

    the interfaces is equal to the ratio of initial perturbation am-Phys. Fluids, Vol. 6, No. 11, November 1994 Letters 3511

    Downloaded 26 Aug 2005 to 128.165.51.58. Redistribution subject to AIP license or copyright, see http://pof.aip.org/pof/copyright.jsp

  • 8/3/2019 John M. Budzinski, Robert F. Benjamin and Jeffrey W. Jacobs- Influence of initial conditions on the flow patterns of a

    3/3