Shock Waves (2000) 10: 377387

Shock tube investigation of hydrodynamic issues related to inertial connement fusion

M.H. Anderson, B.P. Puranik, J.G. Oakley, P.W. Brooks, R. Bonazza
Fusion Technology Institute, Department of Engineering Physics, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA Received 5 January 1999 / Accepted 10 July 2000 Abstract. A shock tube investigation of two hydrodynamic issues related to inertial connement fusion (ICF) is undertaken. ICF is a promising source of energy for the future. There has been a considerable increase in the interest in ICF with the development of the National Ignition Facility (NIF). However, much remains to be investigated before a useful yield is obtained from a fusion reaction for power generation. The physics involved in carrying out a fusion reaction combines hydrodynamics, plasma physics and radiation eects superimposed on each other, at extremely small scales, making the problem very complex. One such phenomenon occurring in the deuterium-tritium pellet implosion is the Richtmyer-Meshkov instability occuring at each layer of the fuel which results in the mixing of the ablator with the fuel. This causes dilution of the fuel and reduces the yield of the reaction. Another issue is the impulsive loading of ICF reactor cooling tubes due to the shock wave produced as a result of the fusion reaction. These tubes must withstand the impulse of the shock wave. A shock tube provides an ideal environment to study these issues at large geometric scales with the isolation of hydrodynamics from other eects. A new vertical, square shock tube has been designed specically for the purpose of studying these uid ow phenomena from a fundamental point of view. The shock tube is vertical, with a large square inner cross-section and is designed to allow for the release of a M = 5 shock into air at atmospheric pressure. In this paper, we describe the new shock tube and related instrumentation in detail and present a few preliminary results on the Richtmyer-Meshkov instability and shock-cylinder interactions. Key words: Inertial connement fusion, Richtmyer-Meshkov, Hydrodynamic instability, Shock diraction

1 Introduction
The use of light ion, heavy ion, electron beam or laser driven inertial connement fusion (ICF) has received recent interest with the development of the National Ignition Facility (NIF). In addition to the many neutronic hurdles inherent in ICF, there are several hydrodynamic issues that need to be resolved for the ignition of the DT fuel and for the cooling of possible ICF reactors, such as SOMBRERO, HIBALL, LIBRA, LIBRA-LiTE, LIBRASP, BLASCON, HYLIFE-I, OSIRIS (Kulcinski et al. 1994; Moir 1996). These hydrodynamic issues are well suited for exploration with a shock tube. 1.1 Ignition of DT fuel The target capsules for NIF and other proposed ICF reactors contain deuterium-tritium (DT) solid and gaseous layers inside an ablative shell (typically polycarbonate or beryllium) that is several millimeters in diameter (Lindl
Correspondence to : M.H. Anderson (e-mail: manderson@engr.wisc.edu)

Fig. 1. Conceptual implosion of DT target. The outer shell is ablated by laser or X-ray energy. This ablation propels the DT fuel inward, compressing the inner gas until ignition occurs

1995), see Fig. 1. If the solid surface of either the DT ice or the ablator is too rough, the densities and temperatures needed for ignition may not be attained because of the mixing of the fuel and ablator due to hydrodynamic instabilities at their interface. The amplitude of surface perturbations between the two dierent density materials (ablator material and DT

378

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

ice) increases upon shock acceleration leading to a distortion of the interface; this phenomenon, resulting in the dilution of the fuel, is known as the Richtmyer-Meshkov (RM) instability (Richtmyer 1961, Meshkov 1970). This is similar to the more familiar Rayleigh-Taylor (RT) instability (amplitude growth due to constant acceleration), in that the growth of the perturbations is due to a baroclinic generation of vorticity consequent to the non-zero cross product of the pressure and density gradients at the interface (Samtaney and Zabusky 1994). However, the RT perturbations grow unbounded only if the constant acceleration is directed from the light to the heavy uid, whereas in the RM case, growth occurs regardless of the direction of the shock propagation. The growth rate of the interfacial perturbations in the linear stage of the process (characterized by an amplitude much smaller than the wavelength ) is proportional to the wavenumber k of the perturbations (k = 2/), the initial amplitude 0 , and the Atwood number A = (2 1 )/(1 + 2 ) (Richtmyer 1961). The linear regime is well understood; however, when the amplitude becomes comparable to the wavelength, the growth becomes non-linear exhibiting phenomena expected to be quite dierent than during the linear stage. Among these dierences are: a decrease in the perturbation amplitude growth rate; the onset of Kelvin-Helmoltz shear instabilities (Chandrasekhar 1961); the interaction between dierent modes (dierent hydrodynamic scales aect one another) (Haan 1991); the distortion of the interface and the departure of its shape from a sinusoidal shape. Knowledge of this non-linear growth rate along with that of the fairly well developed linear growth of the perturbations is crucial in quantifying the mixing of the ablator material and the cold fuel. This mixing rate is necessary to determine whether the density and temperature of the hot center spot are sucient for ignition. In an eort to study this growth rate, the problem can be simplied to a two-dimensional perturbation between two gases of densities 1 and 2 , accelerated by a shock wave generated in a shock tube. This simplication in geometry, and the isolation of pure hydrodynamic eects from radiation and plasma eects, allow a better understanding of the mixing due to the aforementioned instability. 1.2 Impulsive shock loading on cooling tubes An additional hydrodynamic issue in ICF that is suitable for study within a shock tube is the protection of the rst structural wall in reactors. Many types of inertial fusion energy (IFE) target chambers experience signicant hydrodynamic motion. In gas-protected target chambers, the target X-rays and debris ions stop in the gas. Their energy generates a blast wave that consists of both a shock and a radiation wave, where the relative strength of each is a function of the opacity of the gas. Liquid-protected target chambers that are initially at low gas density (thick liquid and wetted-wall concepts) produce signicant gas densities by vaporization of some of the liquid. This vapor exhibits many of the same features as the gas in gas-

Fig. 2. Design of the INPORT tubes. PbLi or other suitable liquid metal ows through the tube and forms an ablative liquid layer by owing through a porous weave of SiC, C or steel

protected concepts. Therefore, interactions between hydrodynamic shocks and target chamber structures are issues for most IFE power plant design concepts. A couple of examples of liquid wall protection concepts are the INPORT (INhibited Flow in PORous Tube) design (Fig. 2) and the rigid PERIT (PErforated RIgid Tube) unit design (Kulcinski et al. 1994). The INPORT design consists of an array of hollow tubes, on the inside of the rst structural wall, which carry a liquid PbLi eutectic alloy. These tubes are constructed out of a porous orthogonal weave of SiC, C or steel which allows an ablative lm of the PbLi to form on the outer surface of the tube which absorbs X-rays and target debris while the bulk of the liquid owing through the tube absorbs the thermal energy and mitigates the isochoric heating by the neutrons. The rst layer of tubes in the PERIT system design have fan sprays, which create a liquid sheet of PbLi which performs essentially the same task as the lm on the INPORT tube design. These tubes must be able to withstand the high impact and diraction of the shock wave formed by the thermonuclear reaction of the DT fuel. To determine the pressure load and its distribution around the cooling tubes it is possible to conduct shock diraction studies within a shock tube. Bryson and Gross (1960) and Syschicova et al. (1967) conducted some initial studies recording the shock formations at dierent times after an initial shock-cylinder interaction. This information is quite valuable; however, only the formation and geometrical aspects of the reected and diracted shocks were investigated while the pressure distribution around the cylinder and its accelerations were not studied. Bishop and Rowe (1967) measured the pressure distribution around a cylinder in a blast channel with piezoelectric pressure transducers. However, due to the limitations of the blast channel, only experiments with low Mach numbers were conducted. The need for experimental pressure data to compare with numerical solutions, and the lack of a robust facility for directly measuring pressures at high Mach numbers, has prompted several studies using holo-

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

379

graphic interferometry, e.g. (Heilig (1969)). This method works well and an estimate of the pressure distribution can be obtained from the experiments using the density gradients and an assumed equation of state. Although these types of measurements are helpful for comparing with numerical simulations, they are not direct measurements of the pressure distribution and are an indirect method of calculating the impulsive force on the cylinder from the shock wave. In an eort to increase the extent of the experimental database, accelerations and pressure distributions around cylinders for strong shocks are directly measured in a large, square inner cross-section shock tube able to achieve Mach numbers of the order of 5 into atmospheric air. The time history of the horizontal and vertical accelerations and pressure distributions are recorded with piezoelectric transducers mounted ush on a cylindrical surface.

2 Experimental facility
2.1 The Wisconsin shock tube A new shock tube has been fabricated especially for the shock-interface interaction and shock diraction studies. The tube is vertical, 9.2 m long, with a large square inner cross-section (25.4 cm 25.4 cm). The square crosssection provides parallel walls suitable for ow visualization without resorting to a structurally weak tube extension (Meshkov 1970). It has a structural capability to withstand a 20 MPa pressure load. Thus, strong shocks can be red into a driven section initially at atmospheric pressure. High initial pressure in the driven section is useful for reducing the eects of the wall vortices created as a result of the interaction of the reected shock from the tubes end wall with the boundary layer that is formed behind the incident shock (Brouillette and Sturtevant 1994). These vortices exert a strain on the interface and thus introduce artifacts in the amplitude growth measurements. These eects are shown to be signicant in shock tubes with small cross-sections and also if the driven section is initially at sub-atmospheric conditions, which is often necessary, to produce strong shocks without creating pressure loads that exceed the structural capability of existing shock tubes (Houas and Chemouni 1996). The vertical orientation of the tube allows preparation of a continuous interface between gases of dierent densities utilizing gravitational stratication. Figure 3 shows the tube assembly. The driver section is made from a circular, chromeplated carbon steel pipe, 46 cm OD, 1.9 cm wall thickness and 2 m long. It is equipped with four ignition tubes mounted inside the driver section, capable of igniting a stoichiometric mixture of oxygen and hydrogen diluted in helium to produce high pressure and temperature driver conditions. To release the shock into the driven section, a metal diaphragm is ruptured either by pressurizing the driver section using compressed gas bottles, or if the hydrogen-oxygen combustion approach is used, by detonation of an explosive charge placed around the circumference of the diaphragm. Sharp knife edges, in the form

Fig. 3. Schematic of the primary features of the shock tube. The shock wave is created by high pressure gas in the driver section rupturing a diaphragm. The shock wave travels down the tube and interacts with either an interface between two dierent gases or with an object in the test section

Fig. 4. Details of tube construction: the outer circular crosssection is steel with a stainless steel square cross-section in the center. The volume between the outer steel tube and the inner wall is lled with concrete

of a cross, are placed just below the diaphragm to facilitate its rupture in the form of four petals. After the diaphragm is ruptured, the four petals remain attached to the diaphragm along its circumference. Table 1 lists diaphragm materials, diaphragm thickness, driver gases and the Mach numbers of the shock wave generated upon rupturing them. In all cases listed in the table, the driven gas is air. The repeatability of the shock strength is typically to within 0.4%. Figure 4 shows the cross-section of the driven section. An internal liner, consisting of four stainless steel plates 9.5 mm thick, welded together, is supported by a concrete matrix contained in a circular carbon steel pipe, 46 cm OD to form a segment of the driven section. These segments are then capped with a class 300# A-105 ange welded to the top and bottom of the tube. This allows any of the nine sections to be bolted together in a variety of congurations. The test section consists of four steel plates, chrome plated on their inside surface, 7 cm thick, welded together to form a box with a 25.4 cm square inside cross-section.

380

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion Table 1. Diaphragm rupture characterization data Material Al 5052 Al 5052 Al 5052 Steel A366 Al 5052 Al 5052 Steel A366 Steel A366 Steel A366 Steel A366 Steel A366 Steel A366 Description One sheet, 0.381 mm thick Two sheets, 0.381 mm thick One sheet, 1.524 mm thick One sheet, 0.864 mm thick Two sheets, 1.524 mm thick Four sheets, 1.524 mm thick One sheet, 1.397 mm thick Two sheets, 1.397 mm thick Three sheets, 1.397 mm thick One sheet, 3.038 mm thick One sheet, 1.397 mm thick Two sheets, 1.397 mm thick Rupture pressure, kPa 232 275 958 1528 1718 2571 2571 3979 4655 5282 2620 3873 Driver gas Air Air Air Air Air Air Air Air Air Air He He Mach number in air 1.25 1.28 1.66 1.85 1.86 2.00 2.00 2.15 2.23 2.28 2.77 3.08

The test section contains two circular ports 28 cm in diameter, on opposite sides. Fused quartz windows, 24 cm in diameter and 9 cm thick, are mounted in these ports for optical access. The tube also contains an interface section specically designed for the study of shock-interface interactions. It consists of four steel plates, chrome plated on their inside surface, approximately 7 cm thick, welded together to form a box with a 25.4 cm square inside cross-section. An interface between two gases of dierent densities is formed in this section either by retracting a thin metal plate initially separating the gases or by placing a thin nitrocellulose or mylar membrane (thickness of the order of one micron), held in a support frame, between the two gases. The retraction technique gives rise to a diuse, continuous interface while the membrane produces a sharp, discontinuous interface. The membrane can also be given a sinusoidal shape by placing it in a suitable supporting frame. Such an interface provides the initial perturbation for the shock-interface interaction. The modular structure of the shock tube allows the placement of the interface section, the test section and the end wall at various mutual distances, allowing for the visualization of the interface after the initial shock acceleration at various delay times. The end wall of the shock tube contains a circular, fused quartz window, 2.5 cm in diameter and 3.8 cm thick. A laser sheet can be projected upward, for planar imaging of the interface. The whole shock tube assembly is supported on three cylindrical, solid steel legs, 18 cm diameter, 30 cm long. The legs, in turn, are anchored in an isolated 8 m3 concrete foundation. A photograph of the shock tube, test section and support structure is shown in Fig. 5.

Fig. 5. Photograph of the University of Wisconsin shock tube. Shown in the photo is the driven section of the shock tube being lowered down onto the base (three solid steel legs fastened to a 8 m3 thick concrete isolation pad). The square section in the center of the photograph is the test section with the diagnostic ports used for optical access for the shock cylinder and RM images

2.2 Instrumentation The driver section is equipped with a strain gauge pressure sensor and a type E thermocouple to measure the pressure and temperature of the driver gas just before the

release of the shock. The driven section has pressure transducer ports at various locations along its length. Piezoelectric pressure transducers are mounted in these ports, ush with the inner wall of the tube, to measure the shock speed and trigger the ow imaging diagnostic hardware. The driver and driven sections can be evacuated to 13.3 kPa.

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

381

Since the velocities of the shock and the accelerated interface are of the order of 1000 m/s, imaging times on the order of tens of nanoseconds are required to obtain sharp images. In the rst series of ow visualization experiments using the shadowgraph technique, either a modied continuum (Surelite II-PIV) pulsed Nd:YAG laser or an arc discharge lamp is used as a light source for the imaging. The laser consists of two laser cavities capable of delivering 250 mJ/pulse (and hence 500 mJ/pulse when the beams are superimposed) at = 532 nm. The pulse width is 10 ns making it suitable for imaging high speed ows. The Xenon Inc. Model-437B short duration 20 ns broad spectrum, arc discharge lamp is used in some experiments to attain a more spatially uniform collimated light source than can be achieved with the laser. A 16-bit CCD camera (Spectra Video Series, by Pixel Vision) is used to capture the ow eld images. It has a back-lit, 1024 1024 pixel array, a thermo-electrically cooled sensor and a low speed (40 kHz) transfer rate to minimize the dark current and readout noise (1 electron/pixel/s and 58 electrons per readout rms respectively, at 45 C). One image is obtained per event. An HP-Innium 4 channel digital oscilloscope with a sampling rate of 1 GHz per channel is used to record the pressure traces from the piezoelectric pressure transducers. 2.3 Diagnostics In the case of shock-interface interaction experiments, typically four wall mounted pressure transducers are used in each experiment. The transducer closest to the diaphragm triggers the laser to pulse, through a variable delay timing box. The delay is set such that the laser pulses when the interface is in the test section. The delay is calculated from one-dimensional gas dynamics and previous experiments correlating the thickness of the diaphragm to the shock speeds it generates upon rupturing. The shock speed is measured from the traces obtained by two other wall mounted pressure transducers. The last pressure transducer is mounted in the end wall to measure the highest pressure generated behind the reected shock. A shadowgraph system is used to image the interface. Figure 6 shows the optical setup for the experiment. The laser beam is spatially ltered before it is expanded by a plano-concave lens (focal length 15 mm). It is then collimated by a plano-convex lens (focal length 500 mm) into a parallel beam of diameter 140 mm. The collimated beam is passed through the test section windows and projected on a screen. The CCD camera is focused on the screen and captures the image. In the case of shock-cylinder interaction experiments, in addition to the transducers that trigger the laser and measure the shock speed, four more are ush mounted on the cylinder at 0, 30, 60 and 90 degrees, as measured from the topmost point on the cylinder. The cylinder can be rotated by 90 degrees and pressures at three more locations can be measured, thus giving a total of seven locations around the cylinder (0, 30, 60, 90, 120, 150 and

Fig. 6. Optical setup for shadowgraph diagnostics. The viewing area in the test section is 212 mm, however in the current setup the illuminating light source is only expanded to 140 mm. As the light rays pass through the test section they are refracted by changes in the density gradient creating an image on the screen

Fig. 7a,b. Schematic of the pressure transducer locations. The cylinder can be rotated 90 so that the pressure distribution around the cylinder can be measured

180 degrees as measured from the topmost point on the cylinder); see Figs. 7a,b. In addition, two accelerometers are mounted on the cylinder, one in the direction of shock propagation and the other perpendicular to it, to measure the acceleration of the cylinder due to the shock loading. The ow is imaged with the shadowgraph technique. The delay for the laser pulse is set such that the shock is captured at dierent locations along the cylinder so the temporal evolution of the ow eld behind the shock and around the cylinder can be recorded. At the time of an experiment, the whole optical setup and a portion of the shock tube containing the test section are surrounded by an enclosure made from thick theater curtains so that the area surrounding the test section is completely dark. The camera shutter is opened and the CCD exposure is started prior to the rupture of the diaphragm (the diaphragm rupture pressure is known from the previous experiments). After the diaphragm is ruptured, the shock triggers the light source and an image of the ow eld of interest is obtained with the CCD camera.

382

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

3 Experimental results
Typical shock tube data are presented for the RM instability study and the ICF reactor cooling tube study. Data include pressure and accelerometer summaries for the cooling tube study and shadowgraph images for both the cooling tube and interface growth studies. 3.1 Richtmyer-Meshkov instability experiments As stated in the introduction, to achieve ignition of the DT fuel in the ICF reactor, the mixing of the ablator material, the DT ice and the hot inner gas after shock acceleration must be minimized. In an eort to study this mixing rate and the growth of the Richtmyer-Meshkov instability, an interface between two dierent density gases is created. The goal is to try to quantify the hydrodynamic mixing between these gases in a quasi two-dimensional system and then apply this knowledge to the more complex ICF capsule, where hydrodynamic, radiation and plasma effects aect the instability. The initial interface between two gases has previously been created in several dierent manners, which fall into two general categories: continuous (diuse) and discontinuous (sharp) interfaces. There are certain advantages and disadvantages to studying both types of interfaces and there have been many variations in the methodology used to create them (Rightley et al. 1998, Houas et al. 1988). The discontinuous interface allows the study of the growth of perturbations on a sharp interface with a known initial condition. The disadvantage of this method is that some type of membrane is required to separate the two gases. This membrane adds another component to the system which may aect the ow and growth rate. The continuous interface allows the study of the growth rate without the addition of a membrane, however, the initial conditions are less well known and more dicult to impose. In the Wisconsin shock tube it is possible to conduct experiments with both continuous and discontinuous interfaces; however, this paper presents results using only discontinuous interfaces. As stated earlier, the shock tube is constructed in a modular form so that the distance between the interface and the test section where the growth of the interface is measured can be changed. This allows the investigation of the growth at dierent times after the interface is accelerated by the shock wave for a given Mach number. The data presented below are taken with the interface section at a distance of 45.4 cm above the centerline of the test section viewing window. In all RM unstable interface images to follow, the shock is traveling from the top to the bottom and is below the eld of view. Figure 8 shows a representative shadowgraph of an initially at interface between air and argon when accelerated by a shock wave with a Mach number of 1.931, 1.12 ms after the initial shock acceleration. Two dierent membranes are used to create the interface: mylar (0.94 m thick) and nitrocellulose (approximately 0.5 m thick). The shadowgraph images for both of the dierent membranes are similar and indicate a thick

Fig. 8. Shadowgraph of a at interface between air and argon with an incident Mach number of 1.931 in air, 1.12 ms after shock acceleration. This shadowgraph is taken with a Nd:YAG laser as the light source

dark diuse line where the interface is calculated to be. The shock accelerated interface appears to be at as expected, since there is no initial perturbation imposed on the interface. (Any perturbations of the lm are too small to be resolved with our optical technique; however, small perturbations in the lm, along with diusion, will result in thickening of the initial interface as observed.) Additional experiments performed with series of wires to support the membrane across the frame indicate that the thick line shown in Fig. 8 is most likely a combination of an intact membrane that is ripped from the sides of the support frame and accelerated down the tube and the interface between the two gases. This explains the small bump in the middle of the interface; since, as the membrane travels down the tube it experiences aerodynamic drag resulting in a parachute eect. From these results it is concluded that it is necessary to use several wires to aid in the rupture of the membrane in order to reduce the eects of the membrane. A second interface with an initially imposed sinusoidal perturbation (with an amplitude of 0.318 cm and a wavelength of 6.35 cm) is shown in Fig. 9. The imposed perturbation of the initial interface allows the investigation of the growth of a known sine wave. The amplitude and wavelength of this initial perturbation also initially sets the interface in the non-linear regime so that non-linear growth rates can be studied a short time after initial shock-acceleration. This has been done for a few dierent post-shock Atwood numbers (0.240.87). Figure 10 shows the growth of the sinuous interface between air and CO2 1.149 ms after the acceleration by a shock with a Mach number 1.999. The interface is created by separating air and CO2 with a 0.94 m mylar lm supported on the sine wave frame. A series of 0.23 mm diameter nylon monolament wires approximately 1.27 cm apart (located at the peak, trough and approximately halfway between the peak and trough

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

383

Fig. 9. Schematic of sine wave frame and initial sine wave perturbation imposed on the interface. A mylar lm is inserted in the frame and separates the two gases to prevent mixing before the shock arrival. The inside of the frame is 25.4 x 25.4 cm2 . The viewing area is approximately 14 cm in diameter

Fig. 10. Shadowgraph of a sinuous interface between air and CO2 with a Mach number of 1.999 in air, 1.149 ms after shock acceleration. This image and the following interface images are taken with a 20 ns pulse duration arc lamp as the light source. The calculated one-dimensional gas dynamics location is used as the center of the interface and the spike (s ) and bubble (b ) growth are measured from this position

Fig. 11. Shadowgraph of a sinuous interface between helium and CO2 with an incident Mach number of 1.909 in helium, 0.892 ms after shock acceleration. Note the large amplitude spikes of CO2 into the helium. The dark, thin region in the upper right of the eld of view is a defect that developed in one of the windows. The size of the spikes and bubbles are estimated in the gure in a similar way as was done in Fig. 10

of the sine wave), are used to help initially support and maintain the mylar lm in the sine wave shape and to help break apart the membrane upon shock interaction. This produces an interface with a post-shock Atwood number of 0.248. The diameter of the eld of view in the gure is 13.54 cm and contains approximately two wavelengths of the initial perturbation ( = 6.35 cm). As can be seen, the amplitude has grown signicantly from the initial condition (approximately by a factor of ten) while the wavelength has remained relatively constant. Spikes of the heavy uid (CO2 ) can be seen to protrude into the lighter air above, whereas the lighter uid forms bubbles into the heavier uid. Figure 11 shows the growth of the instability with a post-shock Atwood number of 0.867, incident Mach number 1.909, 0.892 ms after shock interaction. This interface is created by separating helium and CO2 using the mylar membrane. The amplitude of the CO2 spikes into the lighter helium has grown signicantly as compared to Fig. 10 and the bubbles of helium into CO2 are much atter. This is primarily due to the dierent Atwood numbers

of the interface. The higher Atwood number is indicative of two gases of substantial density dierences. Figure 12 shows a sinusoidal interface between nitrogen and CO2 , 0.648 ms after acceleration by a shock of Mach number 2.834. This experiment is similar to the air/CO2 experiment discussed earlier, however, it is at a much higher Mach number and much earlier time. The growth of the amplitude of the sine wave is therefore signicantly smaller at this point. The higher Mach number results in a higher compression of the initial interface and seems to lead to a slower growth rate. The higher Mach number also resulted in visible secondary shocks o of the joints in the tube walls. In the previous gures the lower Mach number resulted in weaker secondary reections and were not as prominent as in this Fig. 12. In an eort to estimate an experimental growth rate of the initial perturbation from the shadowgraph images discussed above, it is decided to use the distance traveled by the interface for a given incident Mach number, based on one-dimensional gas dynamics calculations as the reference location. This is somewhat of an arbitrary reference,

384

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

Fig. 12. Shadowgraph of a sinuous interface between nitrogen and CO2 with an incident Mach number of 2.834 in N2 , 0.648 ms after shock acceleration. The amplitude of the spikes and bubbles is reduced (as compared to Fig. 10) due to shock compression and the shorter time of travel for the interface. Below the interface are secondary reections o of the shock tube walls

but serves as a convenient theoretical location of the center of the interface. The one-dimensional calculations place the mean location of the interface approximately halfway between the tips of the spike and bubble in all images shown. The age of the interface given above is a direct measure of the time between shock interface interaction and the pulse of the light source. The fact that the onedimensional calculation of the interface location ends up in the center of the spike and bubble is an indication that the gases were relatively pure. The spike height (s ) and bubble height (b ) are then measured from this reference location and a growth rate is estimated by the following expression: 0.5(s + b ) o d = , dt expt tage (1)

where Vshock is the incident shock velocity. As can be seen in Table 2, the calculated growth rates (obtained from the linear theory) for Figs. 10 and 12 seem fairly consistent with the experimentally measured growth rates. The prediction for test 119, where there is signicant observed non-linear growth (see Fig. 11), however, is substantially higher than the measured growth rate. This observation of a decrease in growth rate (as compared to the prediction of the linear theory) for highly non-linear perturbations is consistent with numerical and theoretical studies conducted by Holmes et al. (1995) and Zhang and Sohn (1996). The above shadowgraphic images yield data of the growth rate of perturbations and are currently being used to qualitatively validate numerical predictions from hydrodynamic codes. Future experiments are planned to use planar Rayleigh scattering, Mie scattering, PLIF (planar laser induced uorescence) and other techniques that will result in more quantitative measurements of the growth rate. In these types of diagnostics a laser sheet (approximately 7 cm wide and 0.3 mm thick) will be projected from the bottom of the tube. The Rayleigh scattered light or uorescence can be used to quantitatively determine the density eld of one of the two gases forming the interface. Using a thin light sheet also reduces errors due to boundary layers and other three-dimensional eects associated with averaging the refracted light through the entire 25.4 cm wide shock tube. 3.2 Shock-cylinder interaction experiments The cooling tubes in an ICF reactor are subject to repeated shocks from the ignition of DT fuel reactions that occur in rapid succession. To properly design them, it is necessary to understand the force load history on the cooling tubes. The cooling tubes are long cylinders that have a slight curvature along the azimuthal axis. The ICF reactor cooling tube is modeled with a 25.4 cm long cylinder, 6.35 cm in diameter, placed in the center of the test section. The 25.4 cm length completely spans the 25.4 cm cross-section of the shock tube and permits a twodimensional uid dynamics study of shock-induced transient ow around a cylinder. Two 2.54 cm diameter, xed rods are in place beneath each end of the tube to vertically support it during the experiment. Accelerometers are mounted vertically and horizontally on the interior diameter (1.27 cm) of the cooling tube model. The hollow portion of the tube also serves as a cable-way for the transducers cables. Several experiments are conducted with the cylinder in each position so that statistically consistent measurements are made, and also, a shock history image series is obtained. The presented data are for a Mach number 1.85 shock in air at atmospheric pressure and temperature. The peak initial pressure is shown in Fig. 13a,b along with the pressure history. The circular gure (Fig. 13b) presents the maximum initial pressure distribution around the cylinder while the

where tage is the measured age of the interface. Table 2 shows the experimental growth rates along with the linear theory, calculations of the growth rate obtained using Richtmyers (1961) expression: d = kVint A o , dt theory (2)

where k is the wave number (k = 2/), Vint is the interface velocity, A is the post-shock Atwood number and o is the initial amplitude corrected for shock compression using the expression initially suggested by Markstein (1957): o = 1 Vint Vshock o , (3)

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion Table 2. Experimental results and linear theory calculations of the growth rate, where o = 0.318 and = 6.35 cm Test number 110 (Fig. 10) 119 (Fig. 11) 116 (Fig. 12) Mach number 1.999 1.909 2.834 Vshock , m/s 697.4 1946.7 1000.9 Vint , m/s 396.5 531.7 662.2 Age, ms 1.149 0.892 0.648 A 0.248 0.867 0.267 o , cm 0.137 0.231 0.107 (d/dt)expt , m/s 16.20 27.17 15.39 (d/dt)theory , m/s 13.30 105.17 18.75

385

Fig. 14. Vertical and horizontal acceleration of the cylinder during shock diraction

Fig. 13a,b. Pressure traces from the transducers located on the cylinder during shock diraction. The separation in time indicates the relative position of the pressure transducers. The rst trace is from the pressure transducer on the top of the cylinder. The lower gure shows the maximum pressure generated at the dierent measurement locations around the cylinder

graph (Fig. 13a) shows the time history of the pressure distribution. Qualitatively, the gure reveals that the maximum pressure occurs at the top of the cooling tube model (normal to the shock) and then decreases as the angle (relative to the upwards vertical) increases. For impulsive force measurements needed for cooling tube design, it is necessary to integrate the pressure history data for each angular segment of the cylinder as a function of time. The maximum pressure occurs instantaneously at the top of the tube, however, the force on the tube must be determined by integrating the seven pressure time series over the surface of the tube. The maximum vertical acceleration on the tube is 13,800 gs. This result cannot be used directly because of dierences between the cooling tube prototype and the cooling tube model, both in the size of the tube and the manner in which it is

supported. However, numerical studies of the structural response of the tube model to impulsive loading are in progress and the results will be compared against the experimental data shown in Fig. 14. Upon such validation, the numerical model will represent a useful design tool for the nal cooling tube prototype. Visual diagnostics for these experiments use the shadowgraph technique (optically sensitive to the second derivative of the density gradient). Three shadowgraph results are shown in Figs. 15ac. The thick black vertical structures between the cylinder and the bottom of each of the images are the two (inline) support rods. The optical diagnostic provides a history of the incident shock moving around the cylinder. The initial shock is planar (it appears as a horizontal dark line above the cylinder in Fig. 15a.) A most signicant feature of the shadowgraph images shown in Figs. 15b,c is the bulbous shock reection from the surface of the cylinder. The incident shock remains planar as it travels vertically downward (except where it is attached to the cylinder and a Mach stem is formed); however, the reected shock o of the cylinder surface is curved (as expected). The radius of the reected shock grows in time as the incident shock travels down the cylinder. The time series of photographs reveal the behavior of both the planar shock and gas ow around the cylinder. The combination of dynamic pressure history, acceleration data and optical diagnostics provide a unique experimental characterization of the ICF cooling tube model when subjected to shock loading. This new experimental apparatus is crucial to the understanding of the structural

386

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion

have been conducted successfully and discussed briey above. The results suggest that, in the future, more quantitative data can be obtained using advanced diagnostic techniques that can serve as a set of benchmarking data for the numerical codes that simulate such hydrodynamic phenomena. It is hoped that these studies will help address issues regarding the target designs and the cooling tube designs for possible IFE reactors along with unraveling some of the hydrodynamic physics of ow instabilities.
Acknowledgements. The authors would like to acknowledge the nancial support of the University of Wisconsin-Madison and the Department of Energy (DE-FG03-98DP00207 and DEFG02-97ER54413).

References
Bishop VJ, Rowe RD (1967) The interaction of a long duration friedlander shaped blast wave with an innitely long right circular cylinder. Atomic Weapons Research Establishment, Aldermaston, Berkshire, England AWRE Report No. 038/67 Brouillette M, Sturtevant B (1994) Experiments on the Richtmyer-Meshkov instability: single-scale perturbations on a continuous interface. J Fluid Mech 263: 271292 Bryson AE, Gross RF (1960) Diraction of strong shocks by cones, cylinders and spheres. J Fluid Mech 10(1): 123 Chandrasekhar S (1961) Hydrodynamic and Hydromagnetic Stability. Oxford University Press, Oxford, pp 433 Haan SW (1991) Weakly nonlinear hydrodynamic instabilities in inertial fusion. Phys Fluids B3(8): 23492355 Heilig WH (1969) Diraction of a shock wave by a cylinder. Phys Fluids 12, Supplement 1: 154157 Houas L, Farhat A, Brun R (1988) Shock induced RayleighTaylor instability in the presence of a boundary layer. Phys Fluids 31(4): 807812 Holmes RL, Grove JW, Sharp DH (1995) Numerical investigation of Richtmyer-Meshkov instability using front tracking. J Fluid Mech 301: 5164 Houas L, Chemouni I (1996) Experimental investigation of Richtmyer-Meshkov instability in a shock tube. Phys Fluids 8(2): 614627 Kulcinski GL, Peterson RR, Moses GA, Bruggink D, Cousseau P, Engelstad RL, Lee YM, Khater HY, Lovell EG, MacFarlane JJ, Mogahed EA, Rutledge S, Sawan ME, Sviatoslavsky IN, Wang P, Wittenberg LJ (1994) Evolution of light ion driven fusion power plants leading to the LIBRASP design. Fusion Technology 26: 849856 Lindl J (1995) Development of the indirect-drive approach to inertial connement fusion and the target physics basis for ignition and gain. Phys Plasmas 2(11): 39333992 Markstein GH (1957) Flow disturbances induced near a slightly wavy contact surface, or ame front, traversed by a shock wave. J Aerosol Science 24: 238 Meshkov YY (1970) Instability of the interface of two gases accelerated by a shock wave. NASA Technical Translation F 13 074 Moir RW (1996) Liquid wall inertial fusion energy power plants. Fusion Engineering and Design 32-33: 93104

Fig. 15ac. Shadowgraph images of the shock diraction around a 6.35 cm diameter cylinder at a Mach number of 1.85. The shock was formed from high pressure air into atmospheric air. The successive images show the diraction and reection of the shock as it contacts the cylinder

response of the tube to shock-loading needed for the ICF cooling tube design.

4 Conclusion
The physics involved with attaining useful energy from inertial connement fusion reactions is quite complex and requires continued experimental investigation. A shock tube provides an ideal environment to study the hydrodynamic issues related to ICF reactions. The Wisconsin shock tube is designed with the purpose of studying these issues in detail and extending the existing experimental database into strongly shocked ow regimes. Preliminary experiments

M.H. Anderson et al.: Shock tube investigation of hydrodynamic issues related to inertial connement fusion Richtmyer RD (1961) Taylor instablity in shock acceleration of compressible uids. Comm on Pure and Appl Math 8: 297319 Rightley PM, Vorobie P, Martin R, Benjamin RF (1998) Experimental observations of the mixing transition in a shockaccelerated gas curtain. Phys Fluids 11(1): 186200 Samtaney R, Zabusky N (1994) Circulation deposition on shock-accelerated planar and curved density-stratied interfaces; models and scaling laws. J Fluid Mech 269: 4578

387

Syshchicova MP, Beryozkina MK, Semenov AN (1967) The ow formation around a body in a shock tube (in Russian). In: Aerophysical Investigation of Supersonic Flows, Collection of Papers, Science Publishers, Moscow, pp 713 Zhang Q, Sohn S (1996) Nonlinear theory of unstable uid mixing driven by shock wave. Phys Fluids 9(4): 11061124