INVESTIGATION OF SHOCK WAVE INTERACTION USING LASER-INDUCED IODINE FLUORESCENCE Eric Cecil and James C. McDaniel Department of Mechanical & Aerospace Engineering University of Virginia Abstract Detailed measurements of flow behavior have been taken in a rarefied slip-flow regime hypersonic flow (Mach ) over a sharp edged flat plate with a circular-cylindrical appendage designed to produced shock wave interactions. The flat plate is presented parallel to the wind tunnel freestream to generate a weak shock which impinges on the nearly-normal cylinder bow shock wave. Tests were conducted in a cold free jet wind tunnel facility, using nitrogen gas expanded from ambient temperature. Detailed velocity measurements were made from the Doppler shift of the iodine B-X absorption spectra produced by laser-induced fluorescence (LIF) of a trace amount of iodine seeded in the gas. A planar grid of local velocity with a resolution on the order of one mean-free-path in the flow free stream was obtained using a wide sheet-beam from a frequencytunable multiwatt argon laser to excite LIF and a low-noise CCD camera to record multiple images. In addition to the velocity measurements, these spectra provide information on the local translational and rotational temperature in the flow. Introduction Shock wave interference occurs when the bow shock of a supersonic vehicle impinges on the bow shock formed on an appendage such as a wing, fin, or cowling. Such interactions result in high localized heat transfer rates and peak surface pressures on the appendage near the impingement, with the details depending the configuration of the shock waves. Shock wave interference was first studied systematically in the 960 s during the development of supersonic flight vehicles. In the work of Edney [] six shock wave interference patterns were first identified, based on the location of impingement on a curved bow shock with respect to its upper and lower sonic lines. Shock wave interference studies such as Edney s have mostly employed wind tunnel facilities operating in the continuum regime with which the investigators were concerned. These studies have involved use of surface pressure taps and heat transfer gages, and spark or Schlieren system flow visualization to locate shock wave, shear flow, and expansion wave structures, and peak heating and pressure locations. Those visualization methods depend on variation in the index of refraction due to density variations and are not sensitive enough for useful application to rarefied gas flows. Therefore a laser induced fluorescence technique which can generate detailed planar grid of velocity measurement in the flow presents a worthwhile new approach to investigating a shock interference flow field under the rarefied conditions associated with the hypersonic vehicles travelling in the upper atmosphere. A schematic of the shock interaction test flow model is shown in figure. A sharp edged flat plate is employed to generate a weak shock which impinges upon the bow shock of a cylindrical fin. Under the rarefied conditions of the tests, flow slips at the plate surface and the thickness of shock waves is not negligible. The boundary layer is thick and interacts with the shock layer near the leading edge of the plate. The interaction of the shocks shown in the figure produces a supersonic jet which impinges on the fin causing peak heating. Iodine fluorescence measurement technique This paper presents results of a technique based on narrowband excitation of fluorescence from nuclear hyperfine components in the B X electronic transition of molecular iodine. This approach was developed to study low temperature flows of I -seeded Cecil
Fluorescence signal [arbitrary units] T= K T=0 K T= K T=0 K T= K R P Fig. : Schematic of shock wave interaction model in slip flow regime. N gas. A tunable single frequency laser beam is expanded and collimated into a thin sheet and passed into a low density wind tunnel flow, causing I molecules to fluoresce within a selected planar section of the flow-field. A series of images are taken of the fluorescence with a charge-coupled device (CCD) array as the laser frequency is incrementally tuned over selected lines in the I absorption spectrum. The fluorescence signal versus frequency is thereby obtained for any point in the plane of the laser sheet, within the resolution of the digital image. These absorption spectra can provide information on the local thermodynamic conditions in the flow. The fluorescence signal produced by I vapor (in a stationary, low pressure carrier gas) from a single transition line i, centered at frequency ν 0,i, due to excitation at frequency ν, is given by [] S fi (ν) = CIn I f v J (T rot,t vib ) q v v g hf S hf,i S J J J + V (ν, ν 0,i,P,T tran ), () ν D (T tran ) where C is a constant representing signal collection efficiency and molecular constants, I is excitation source intensity, n I is I number density, P is carrier gas pressure, T tran, T rot,andt vib are respectively translational, rotational, and vibrational temperature, f v J is the Boltzmann population of the ground vibrational-rotational state v J, q v v is the Franck-Condon factor, g hf is the hyperfine degeneracy, S hf,i is the hyperfine line strength, S J J is the Hönl-London rotational line strength, and ν D (T tran ) is the Doppler line width. The Voigt function V (ν, ν 0,i,P,T tran ) is a convolution integral of Gaussian and Lorentzian lineshape contributions. Collisional quenching of the excited state is negligible in low pressure conditions, and therefore the fluorescence efficiency has been taken to be unity in Eq. ()... Relative frequency, GHz Fig. : Computed P,R hyperfine lineshape at low temperatures, plotted at very low constant pressure. Line broadening mechanisms cause contributions from closely spaced transitions to overlap. Pressure (Lorentz) broadening due to I N collisions is negligible at the low pressures encountered in this study. The observed broadening is almost entirely Doppler broadening due to the random, translational thermal motion of the gas molecules. The measured fluorescence signal is the resultant of fluorescence from all excited absorption lines, each of which contributes an amount given by Eq. (). The net signal due to excitation at laser frequency ν from N t excited transitions is S f (ν) = N t i S fi (ν). Due to nuclear electric quadrupole coupling and magnetic spin-rotation interactions in the I molecule, rotational-vibrational lines exhibit hyperfine splitting. The P and R rotational lines of the ( 0) vibrational band each have hyperfine structure consisting of components[, ]. The frequencies of these hyperfine components span. GHz and are interleaved, as shown in the lower part of figure. The heights of the vertical lines in the figure indicate relative strengths of individual components. In I vapor at room temperature (90 00 K) and low pressure, the P and R lines appear as a single blended line with a Gaussian-like profile, due to significant Doppler broadening. Below 0 K, the Doppler broadening is sufficiently low to exhibit some of the underlying hyperfine structure. The computed (low pressure) P,R lineshape S f (ν) is plotted at several temperatures from K to K in figure. Because of the irregular, close spacing of the hyperfine components, only four distinct peaks are discernable at K. At 0 to K, seven distinct peaks can be distinguished. Further peaks appear Cecil
st order nuclear electric quadrupole + magnetic spin-rotation interaction model of hyperfine splitting for blended P(-0), P0(9-0) and P(-0), R(-0) transitions Fluorescence signal E-09.E-09 E-09 E-0 T=7 K T=00 K T=0 K T=00 K T=00 K 0-6 - - - - - 0 (P=0 - atm) Relative frequency, GHz Fig. : Computed P,P0 (at left) and P,R hyperfine lineshape up to stagnation temperature, plotted at very low constant pressure. in the lineshape at K. In a free jet expansion from room temperature, this combination of low pressure and temperature is typical. The temperature sensitivity of the transition lineshape makes possible temperature determination by nonlinear least-squares regression fitting of the hyperfine spectral model to measured spectral data. Rotational temperature information is obtained by scanning an additional transition having a large rotational population at higher temperatures closer to ambient. For this purpose, the P(9 0)/P0( 0) transition is suitable. The Boltzmann population of the ground vibrationalrotational state v J, f v J which appears in equation causes the intensity of the P/P0 line to increase relative to the P/R line at increasing rotational temperatures as can be seen in figure. When the excited gas has a mean speed in the direction of laser propagation, u, the observed lineshape will be Doppler-shifted by an amount u/λ, where λ is the excitation wavelength. (At low pressure, the impact shift from collisional perturbation of molecular energy levels may be neglected.) The velocity component u is found by measurement of the frequency shift, ν, of the lineshape relative to a reference spectrum measured simultaneously in a stationary cell of I vapor at room temperature. The determination of ν from typical data is illustrated in figure. In practice, ν is determined as part of the nonlinear regression fitting process, along with the temperature. In order to establish two independent velocity components at each point, it is necessary to obtain two spectral data sets in which the laser sheet, while remaining in the same plane, propagates in different directions. Normalized Signal [arbitrary units] Normalized I static cell data, room temperature (00 K) Gaussian fit to static cell data Unshifted hyperfine spectral model: T=. K Measured LIF signal in I -seeded N flow Hyperfine spectral model fitted to data: T=. K ν Relative Frequency, GHz Fig. : Determination of temperature from lineshape and velocity component from frequency shift ν. Test facility and measurement apparatus An underexpanded free jet was produced by the expansion of I -seeded N gas across a thin circular orifice of diameter D = mm into a continuously pumped vacuum chamber. Flow stagnation properties are measured in a settling chamber immediately upstream of the orifice. Gas issuing from the sonic condition at the orifice expands radially into a source-like hypersonic flow, forming a barrel-shaped shock wave that terminates at a normal shock known as the Mach disk. The hypersonic core of the free jet serves as the wind tunnel test section. The interacting shock test flow is produced by introducing a model inside the free jet core, upstream of the Mach disk location in the undisturbed free jet. The model consists of a flat plate which has a small circular rod of diameter d =. mm projecting perpendicular to its surface. The model was constructed of aluminum plate and stainless steel rod and was painted flat black to minimize the amount of reflected and scattered laser light, since the laser sheet is incident on the model during the tests. A schematic of the PLIF imaging set-up is shown in fig.. The flat surface of the model was centered along the symmetry axis of the undisturbed free jet, so as to be at zero incidence. The leading edge of the model was located approximately mm (0. D) downstream of the orifice, where the effective freestream Mach number at the center of the leading edge, M LE.9, based on the correlation of Ashkenas and Sherman []. Cecil
Fig. : Experimental set-up for LIF imaging of the flow over the model in the low density wind tunnel. The laser beam was expanded and collimated into a thin, cm wide sheet and passed through a window in the wind tunnel to excite fluorescence in a plane passing through the flow. A schematic of the laser apparatus is shown in figure 6. A Spectra-Physics BeamLok 00A- argon ion laser operating at nm was used to excite I fluorescence in the tests. A temperature-stabilized, air-spaced intracavity etalon eliminated all but a single longitudinal mode of the laser. The etalon tilt angle, which can shift the etalon loss curve to select the longitudinal mode, was controlled via a geared stepper motor attachment. Etalon tuning between longitudinal modes of the laser ( MHz increments) is insufficient to resolve fine detail in the hyperfine I spectrum, which has thermally broadened line widths on the order of 0 MHz at 0 K. Fine frequency scanning of the laser (typically in 0 MHz steps) was accomplished by varying the laser cavity length. The BeamLok system of the laser incorporates a piezoelectric translator which can move the laser output mirror up to. µm, thereby changing the cavity length and allowing continuous scanning over a range of a few hundred MHz as the applied voltage is varied. Because that range is limited, fine scanning is done in conjunction with intermittent etalon tilt adjustments to scan the full range of the laser gain profile. For frequency measurement, the laser beam is sampled by a Fabry-Perot scanning interferometer (Spectra-Physics model 0) and passed through a room temperature I cell. Relative frequency increments between camera exposures are determined from the location of transmission fringes spaced Fig. 6: Laser set up. GHz apart in the interferometer scan output, in conjunction with a cumulative fringe count. Fluorescence in the I cell is measured with a photodetector to provide an absolute frequency reference. Images were taken with a Photometrics CH0 liquid nitrogen-cooled camera system using a PM CCD having a 6 6 array of 0 µm square pixels. This system allows acquisition of low noise, highly linear, -bit images under low light conditions. An mm f/. Nikon lens was employed with a. teleconverter, an mm extension ring (to reduce the minimum focusing distance), and a 70 nm long-pass Schott glass filter to block scattered laser light. The camera was placed with a subject-to-image plane distance of about 76 cm, resulting in a spatial resolution of 0 pixels/mm. Due to exposure times on the order of 60 sec. for each image, only timeaveraged fluorescence is recorded. The free-running laser jitters over an optical frequency bandwidth of at least MHz, at temporal frequencies between 0 00 Hz, which causes it to be averaged over a single exposure. This frequency jitter produces undesirable instrumental broadening in the absorption spectra. Measurements Measurements were taken in the symmetry plane passing through the cylinder centerline (normal to the plate) as well as parallel to the plate at three distances from the plate surface. Temperature and velocity measurements were acquired in three parallel planes and one perpendicular plane. Each of these was typically broken up into measurements on multiple days to acquire three different laser sheet Cecil
incidence angles, because there was partial blockage of the laser sheet by the cylinder. Figure 7 shows the in-plane velocity vectors and contour lines determined from measurements in the plane parallel to the plate, one cylinder diameter from the surface. Regions in which the laser sheet was blocked by the cylinder (where no data could be recorded) resulted in the two blank strips in the plot. The flow is from top to bottom with the plate leading edge located at x = 0 and the center of the cylinder is located mm downstream. The incident shock front generated by the leading edge of the plate appears at about x = mm. Streamlines and temperature contours measured in the center plane of the flow are shown in figure 7. Temperature contours corresponding to the shock wave structures (impinging shock, cylinder bow shock, and transmitted bow shock) previously outlined in figure are evident. These temperature contours were determined assuming thermal equilibrium in the spectral model. Work is now being completed to allow independent determination of translational and rotational temperatures in the flow, for better comparison with DSMC numerical flow simulations. This nonequilibrium is characteristic of the flow in regions having large gradients such as the shock waves, which are here thick enough to resolve some of the shock structure. This work is ongoing and will be supplemented by numerical simulations being made by the author. x, mm -0-0 0 0 0 7 7 6 6 6 6 6-0 0 0 z, mm Dimensions given from center of plate leading edge. V xz (m/s) 7 00 6 70 700 60 600 0 00 0 0 9 00 0 7 00 6 0 00 0 00 0 0 70 m/s Fig. 7: Velocity vectors and contours in plane located d from plate surface. Acknowledgments This research was supported by grants from the Virginia Space Grant Consortium, the NIA, and Ball Aerospace. Peter Gnoffo at NASA Langley Research Center served as a technical monitor. References [] D. E. Cecil, Planar Laser-Induced Iodine Fluorescence Measurements in Rarefied Hypersonic Flow Over a Reaction Control Jet Model in a FreeJetWindTunnel, M.S. Thesis, University of Virginia, Jan. 00. COL 70 0 90 00 0 0 0 0 0 60 70 7 6 9 6 7 7 00 0 00 ROW 9 T(K) 00 7 0 0 9 00 7 7 6 0 00 7 0 [] Barry Edney, Anomalous Heat Transfer and Pressure Distributions on Blunt Bodies at Hypersonic Speeds in the Presence of an Impinging Shock, FFA Report, The Aeronautical Research Institute of Sweden, Stockholm 96 Fig. : Measured streamlines and temperature contours in center plane of flow. Cecil
[] M. D. Levenson and A. L. Schawlow, Hyperfine Interactions in Molecular Iodine, Phys. Rev. A 6, pp. 0 0 (97). [] D. J. Ruben, S. G. Kukolich, L. A. Hackel, D. G. Youmans, and S. Ezekiel, Laser-Molecular Beam Measurement of Hyperfine Structure in the I Spectrum, Chem. Phys. Lett., pp. 6 0 (97). [] H. Ashkenas, and F. S. Sherman, The Structure and Utilization of Supersonic Free Jets in Low Density Wind Tunnels in Advances in Applied Mechanics, Supplement, Vol., Academic Press, New York, 966. Cecil 6