Raul Garza

Ignition & Growth and JWL++ Detonation Models in Coarse Zones

“Ignition & Growth” (I&G) and JWL++ models are compared for a variety of problems. The detonation velocity becomes nearly constant with zoning at the edge of convergence, which for TATB, is 8 zones/mm for I&G and 4 for JWL++. The use of pressure in the rate for I&G makes the detonation velocity rapidly decrease as the zones are coarsened. Using pressure plus artificial viscosity to some power in the rate for JWL++ allows the correction for coarsening zones. In coarse zones, the pressure and the burn fraction turn on independently and this feature dominates model behavior. If pressure lags burn fraction, then the maximum pressure will be lower than expected. An unexpected phenomenon is saturation, i.e. the slowing down of the detonation velocity as a function of the fast rate constant. This slowing can be weak and produce a plateau, or it can be strong and cause the detonation velocity to approach an asymptote. The saturation effect comes from a combination of the 1−F term and declining pressures. Failure (critical diameter effect) occurs in reactive flow but optimizing for this undoes the settings for other results. In JWL++ , the fast reaction pressure exponent is near −1 for the best fit for the size (diameter) effect, 2 for the Pop plot and near −3 to fit failure. The Pop plot deflagration rate is derived, although it needs not to be the same as the detonation rate. The use of additive pressures is compared with the pressure equilibrator and no difference is found. Increased zoning by a factor of 5 and improved code structure will be needed for future improvement.

Measurement of Low Level Explosives Reaction in Gauged Multi-Dimensional Steven Impact Tests

The Steven Test was developed to determine relative impact sensitivity of metal encased solid high explosives and also be amenable to two‐dimensional modeling. Low level reaction thresholds occur at impact velocities below those required for shock initiation. To assist in understanding this test, multi‐dimensional gauge techniques utilizing carbon foil and carbon resistor gauges were used to measure pressure and event times. Carbon resistor gauges indicated late time low level reactions 200–540 μs after projectile impact, creating 0.39–2.00 kb peak shocks centered in PBX 9501 explosives discs and a 0.60 kb peak shock in a LX‐04 disk. Steven Test modeling results, based on ignition and growth criteria, are presented for two PBX 9501 scenarios: one with projectile impact velocity just under threshold (51 m/s) and one with projectile impact velocity just over threshold (55 m/s). Modeling results are presented and compared to experimental data.

TIME-SEQUENCED X-RAY OBSERVATION OF A THERMAL EXPLOSION

The evolution of a thermally‐initiated explosion is studied using a multiple‐image x‐ray system. HMX‐based PBX 9501 is used in this work, enabling direct comparison to recently‐published data obtained with proton radiography [1]. Multiple x‐ray images of the explosion are obtained with image spacing of ten microseconds or more. The explosion is simultaneously characterized with a high‐speed camera using an interframe spacing of 11 μs. X‐ray and camera images were both initiated passively by signals from an embedded thermocouple array, as opposed to being actively triggered by a laser pulse or other external source. X‐ray images show an accelerating reacting front within the explosive, and also show unreacted explosive at the time the containment vessel bursts. High‐speed camera images show debris ejected from the vessel expanding at 800–2100 m/s in the first tens of μs after the container wall failure. The effective center of the initiation volume is about 6 mm from the geometric center of the explosive.

Size effect and cylinder test on several commercial explosives

Some size (diameter) effect and the Cylinder test results for Kinepak (ammonium nitrate/nitromethane), Semtex 1, Semtex H and urea nitrate are presented. Cylinder test data appears normal despite faster sound speeds in the copper wall. Most explosives come to steady state in the Cylinder test as expected, but Kinepak shows a steadily increasing wall velocity with distance down the cylinder. Some data on powder densities as a function of loading procedure are also given.

THE ENERGY DIAMETER EFFECT

We explore various relations for the detonation energy and velocity as they relate to the inverse radius of the cylinder. The effective detonation rate‐inverse slope relation seen in reactive flow models can be used to derive the familiar Eyring equation. Generalized inverse radii can be shown to fit large quantities of cylinder results. A rough relation between detonation energy and detonation velocity is found from collected JWL values. Cylinder test data for ammonium nitrate mixes down to 6.35 mm radii are presented, and a size energy effect is shown to exist in the Cylinder test data. The relation that detonation energy is roughly proportional to the square of the detonation velocity is shown by data and calculation.

DETAILED COMPARISON OF BLAST EFFECTS IN AIR AND VACUUM

The role of air as an energy transfer medium was examined experimentally by subjecting identical large‐area rectangular witness plates to short‐range blast effects in air and vacuum (∼50 mtorr) at 25 °C. The expanding reactant front of 3 kg C4 charges was observed by fast camera to be cylindrically symmetric in both air and vacuum. The horizontal component of the reactant cloud velocity (perpendicular to the witness plates) was constant in both cases, with values of 3.0 and 5.9 km/s for air and vacuum, respectively. As a result of the blast, witness plates were plastically deformed into a shallow dish geometry, with local maxima 30 and 20 mm deep for air and vacuum, respectively. The average plate deflection from the air blast was 11 mm, ∼10% deeper than the average vacuum plate deflection. Shock pressure estimates were made with a simple impedance‐matching model, and indicate peak values in the 30–50 MPa range are consistent with the reactant cloud density and velocity. However, more detailed analysis is...

Optical probes for continuous Fabry-Perot velocimetry inside materials

We have used velocimetry for many years at LLNL to measure velocity-time histories of surfaces in dynamic experiments. We have developed and now use special instrumentation to make continuous shock-velocity measurements inside of materials. The goal is to extend the field of velocimetry into a new field of application in shock physics. At the last Congress we reported the successful use of our new filter system for selectively eliminating most of the non-Doppler-shifted light. We showed one record of a fiber embedded inside an explosive making a continuous detonation velocity-time history. At that time it was difficult to obtain complete records. We have now carried out over 50 inexpensive experiments usually using small cylinders or rectangular blocks of explosives or metals. Most were started by detonating a 25 mm diam. by 25 mm long cylinder of Comp B explosive to drive a shock into an adjacent material of similar dimensions, using our embedded fiber probes. In contrast to surface velocimetry, embedded measurements involve detailed hydrodynamic considerations in order to result in a successful record. Calculations have guided us in understanding of various failed and successful experiments. The homogeneity of the explosive, poor contact, the materials used in the cladding and core of the fiber optic probes, and the shock speeds to be covered all greatly affect the success of an experiment. For example, a poor contact between the optical fiber and its environment causes severe loss of data. Non-symmetric air gaps on one side of the fiber cause 3 dimensional hydrodynamic effects, which cause the shock wave in the fiber core to be too steeply angled to reflect light. We have recently developed and successfully used a special probe to usually overcome this limitation. We have custom designed several unique types of fiber-optic probes for specialty applications, using both solid and liquid core materials, to extend the usable shock-velocity range.

Mix and instability growth from oblique shock

We have studied the formation and evolution of shock-induced mix resulting from interface features in a divergent cylindrical geometry. In this research a cylindrical core of high-explosive was detonated to create an oblique shock wave and accelerate the interface. The interfaces studied were between high-explosive/aluminum, aluminum/plastic, and plastic/air. Surface features added to the aluminum were used to modify this interface. Time sequence radiographic imaging quantified the resulting instability formation from the growth phase to over 60 μs post-detonation, thus allowing the study of the onset of mix and evolution to turbulence. The plastic used here was porous polyethylene. Radiographic image data are compared with numerical simulations of the experiment.

Experiences with decontaminating tritium-handling apparatus

Tritium-handling apparatus has been decontaminated as part of the shutdown of the LLNL Tritium Facility. Two stainless-steel gloveboxes that had been used to process lithium deuteride-tritide (LiDT) salt were decontaminated using the Portable Cleanup System so that they could be flushed with room air through the facility ventilation system. Further surface decontamination was performed by scrubbing the interior with paper towels and ethyl alcohol or Swish{trademark}. The surface contamination, as shown by swipe surveys, was reduced from 4{times}10{sup 4}--10{sup 6} disintegrations per minute (dpm)/cm{sup 2} to 2{times}10{sup 2}--4{times}10{sup 4} dpm/cm{sup 2}. Details on the decontamination operation are provided. A series of metal (palladium and vanadium) hydride storage beds have been drained of tritium and flushed with deuterium in order to remove as much tritium as possible. The bed draining and flushing procedure is described, and a calculational method is presented which allows estimation of the tritium remaining in a bed after it has been drained and flushed. Data on specific bed draining and flushing are given.