CP620, Shock Compression of Condensed Matter - 2001 edited by M. D. Furnish, N. N. Thadhani, and Y. Horie 2002 American Institute of Physics 0-7354-0068-7 For special copyright notice, see page 844. REACTION OF SHOCKED BUT UNDETONATED HMX-BASED EXPLOSIVE P. Taylor, D.A. Salisbury, L.S. Markland, R.E. Winter and M.I. Andrew Hydrodynamics Department, AWE, Aldermaston, Reading, Berkshire, RG7 4PR, UK Cylindrical samples of the pressed plastic bonded HMX based explosive EDC37, backed by metal discs, were shocked through a stainless steel attenuator by an explosive donor. Reaction of the EDC37 sample was diagnosed with embedded PVDF pressure gauges and a distance to detonation for the geometry was determined. Sample length was then reduced to less than the observed detonation distance and laser interferometry was used to record the free surface velocity of the metal backing disc. The results provide data on the metal driving energy liberated by explosive which is shocked and reacting but not detonated. The results are compared with 2-D Eulerian calculations incorporating a 3term ignition and growth reactive burn model with desensitisation. It is found that a parameter set for the reaction model which replicates the PVDF pressure profiles before reflection also gives good agreement to the metal disc velocity history at early times. The results show that an appreciable fraction of the metal driving potential of an explosive can be released without detonation being established. experiments in a similar geometry have been performed with the aim of simultaneously recording the reaction profiles in the shocked explosive and the resultant velocity history of adjacent metal reflector discs. Measurement of both the reaction history and metal driving ability of the shocked explosive allows comparison with the energy release prediction of a reactive burn algorithm such as the Lee and Tarver ignition and growth model3 to be made. INTRODUCTION Prediction of the response of explosive containing systems to external shock stimuli is an increasingly important area of study for application to system safety assessments. A key question for any predictive tool is whether the explosive is detonated by the insult, as this is likely to lead to the highest energy responses. However, the work reported by Craig and Marshall1 on planar shocks in explosive samples shorter than their natural run to detonation distance shows that absence of detonation conditions does not preclude significant energy release from the shocked and reacting explosive material. Previous work by the authors of this paper2 on the reaction of the HMX based explosive EDC37 suggested that normal reflection of strongly reacting shocks from a high impedance barrier did not lead to detonation, but did produce significant energy release as observed from indentation depths in the stainless steel reflector blocks. In order to better quantify the amount of energy liberated from the shocked but undetonated explosive, further EXPERIMENTAL The basic SI2D experimental geometry used in this study is depicted in Figure 1. An EDC29 donor charge initiated by a detonator drives a divergent shock wave through a 45 mm thick stainless steel barrier into an acceptor charge of EDC37 explosive. This sample usually consisted of a set of machined discs with PVDF pressure gauges mounted between them. The sample was terminated with a reflector disc or discs. For the reactive drive experiments, the free surface velocity of this disc was 841 EDC37 Sample 1Discs F-P Velocity Probes V. Perspex >^^ Rings \^ Stainless Steel ^^^ Barrier ^^~- | 1 H X, U /" K^-————————————\ \ _ _————————— N ^ (50 |-c PVDF ^ Gauges s^ - 80 EDC29 Donor Charge ~~~~ consisting of a pulsed dye laser output via fibre optic cable, with Fabry Perot analysis of the returned light. The system has a velocity sensitivity of 0.9 km/sec per fringe order (etalon spacing of 50 mm wavelength 602 nm). The dynamic fringe patterns were recorded on Thomson 506 N electrooptical streak cameras. These records were calibrated by the use of a pulsed laser diode time marker system. eflector isc(s) r—-~ ^ , Perspex ^--~-T——————— Ji / Detonator en -o RESULTS AND DISCUSSION 150 Dimensions mm The first shock or detonation wave arrival time at each on axis gauge location is recorded in Table 2, where T NN.N is the arrival time (JLIS) of a wave at gauge location NN.N (mm), relative to the shock entering the sample. The time from detonator current zero (lo) to shock arrival in the sample was recorded as 15.6 ± 0.05 jus over the series. SI2D/10 and SI2D/14 were control experiments to establish the experimental distance to detonation for the 45 mm barrier geometry with and without embedded gauges. From the total charge transit time alone a simple 1 -D equivalent run distance can be calculated with the unreacted EDC37 Hugoniot and the Pop plot. This procedure leads to a run distance discrepancy of 5 mm between the two experiments. This suggests the effect of the embedded gauges is to reduce the effective 1-D equivalent input pressure from about 24 kb to 21 kb. The 2-D divergent nature of the geometry leads us to expect that the 1-D Pop plot will actually be FIGURE 1. SI2D geometry with 3mm reflector disc shown. diagnosed with Fabry -Perot velocity interferometry at locations on and off the axis. Control experiments without embedded gauges, with an inert impedance match sample and with a long sample length to determine the run to detonation distance were included in the experimental series which is detailed in Table 1. The PVDF gauges were commercially available Dynasen gauges insulated with Kapton and having a thickness of less than 90 jam. Outside the gauge area the void between the explosive discs was filled with Sylgardl84 potting compound after assembly. One experiment incorporated strain gauges in place of PVDF pressure gauges, and the strain data from this trial was used to provide a strain compensation correction for the pressure data. The Fabry-Perot measurement was performed with a BMI Doppler Laser interferometer system, TABLE 1.45 mm barrier SOD Experiments Experiment Number SI2D/10 SI2D/1 1 SI2D/12 Sample Material EDC37 EDC37 EDC37 Derived 1-D Reflector Reflector Gauge Planes Run Distance*1 Material(s) mm into sample Thickness 31.5mm 40mm St.Steel 0,10,17.5,22.5,42.5,62.5 40mm St.Steel None Ob,10,17.5,22.5b 3mm Oh,10,17.5,22.5h None Al.Alloy 3mma St.Steel KEL-F 22.5mm Al.Alloy 3mm Oh,10,17.5,22.5h None SI2D/13 a 3mm St.Steel EDC37 40mm SI2D/14 60mm 26.5mm 0,60 StSteel SI2D/15 EDC37 22.5mm Copper 3mma 0,10,17.5,22.5° None a The free surface velocity of these plates was monitored with F-P interferometry, on axis and 20 mm off axis. h These gauge positions contained two gauges, one on axis the other 20 mm off axis. c SI2D/15 contained strain gauges in identical positions to the PVDF gauges in SI2D/11-13. d Derived from transit time through the complete sample length the unreacted Hugoniot and the Pop plot. Sample Thickness 62.5mm 22.5mm 22.5mm 842 TABLE 2. PVDF Pressure gauge timing data. T10 Expt. T17.5 T22.5 T42.5 T60 US MS MS MS 4.84 6.24 10.86 4.88 6.30 6.48 8.67 MS SI2D/10 2.73 SI2D/1 1 2.77 SI2D/12 SI2D/13 3.62 SI2D/14 - - - returning to the input face of the sample, at a typical shock velocity, well below the normal detonation propagation speed in the explosive. T62.5 MS 13.17 . 11.75 - Gauge 1 S-Steel/EDC37 Gauge 3 10mm into EDC37 Gauge 4 17.5mm into EDC37 EDC37/S- displaced to higher run distances for a given input pressure. This is confirmed by the embedded gauge shock and detonation wave arrival times for SI2D/10 shown in Fig. 2. A simple run distance of 37 mm is suggested which corresponds to a 1-D Pop plot pressure of 19 kb. However the observed average shock velocity of 3.615 mm/us equates to a shock pressure of 34 kb. 2 PVDF Gauge Arrival limes f t+ Shock fj — Detonation 8 10 Figure 4 shows the pressure gauge data from SI2D/12 where the reflector consisted of 3 mm thick Aluminium Alloy and stainless steel discs. The impedance mismatch between the two discs leads to a double shock reflection, and there appears to be increased reaction behind both of the reflected shocks. The short duration release wave between the two reflected shocks seen on the 22.5 mm gauge indicates a small assembly gap between the two discs. The double shock nature of the reflected pulse is still evident when the shocks return to the input face of the sample. ^ • 6 FIGURE 3. Comparison of PVDF gauge data from SI2D/11 22.5 mm sample (black) and SI2D/10 - 62.5 mm sample (grey). ^* ** S 4 Time from main shock into samplers - Distance/mm into sample FIGURE 2. SI2D/10 x-t plot from gauge times. The results from the two run distance experiments suggest that most of the offset from the 1-D Pop plot run distance observed in SI2D/10 is due to the 2-D divergent shock flow in the experiment, with a small proportion of the increased run distance due to the presence of the embedded gauges. As the main study involves reflection of the reactive shock well before the experimental run distance of 37 mm, the small degradation of the shock pressure and reactive growth profiles due to the embedded gauges was judged to be acceptable. Comparison of the pressure gauge data from SI2D/10 and SI2D/11 shows the effect of reflecting the reacting shock wave from a thick high impedance barrier. Figure 3 shows that the reaction profiles before reflection are reasonably reproducible, and the reflection of the shock seems to produce an increase in reaction rate behind the reflected shock. The reflected shock is observed Gauge 3 - 10mm Gauge 4 - 17.5mm Gauge 5 - 22.5mm 2 4 6 8 10 Time from main shock into sample/us FIGURE 4. SI2D/12 pressure data against calculation in grey. Also plotted in the figure, in grey, are pressure profiles calculated using an Eulerian hydrocode with a 0.5 mm mesh. The reactive burn model used is a three-term ignition and growth model with a shock desensitisation model. The desensitisation model is required in the calculation to prevent the 843 good to around 5 jus, but at late times the hydrocode overestimates the drive and terminal velocity. explosive detonating when the shock pressure increases on reflection. The parameters for the model have been adjusted to give a reasonable fit to the experimental reaction profiles close to the reflection plane. Figure 5 compares the experimental free surface velocity measurements from SI2D/12 with calculated histories from the same calculation shown in Fig.4. Agreement is good for the first 3-5 jas, but the ignition and growth calculation appears to be over-predicting the late time drive, giving significantly higher terminal velocities than the experimental observations. —SI2D/15 data - on axis —Calculation - on axis —SI2D/15 data - 20mm off axis —Calculation - 20mm off axis FIGURE 6. Free surface velocity measurements from 3 mm Copper reflector experiment SI2D/15 compared to calculation. CONCLUSIONS SI2D/12-onaxis Calculation - on axis Detonation calculation - on axis SI2D/13 data - 20mm off axis Calculation - 20mm off axis SI2D/13 data (Kel-F samole) - on axis The data presented clearly shows significant energy release from the shocked but undetonated explosive. A 2-D Eulerian hydrocode with an ignition and growth reactive burn model and a desensitisation model provides a reasonable fit to the recorded reaction profiles recorded with embedded pressure gauges. This calculational modelling provides a good fit to the early time reflector plate velocity but tends to provide too much late time drive. A possible explanation of this discrepancy is that the calculated reaction rate remains finite at late times whereas experimentally the presence of 2-D divergence and release waves from the free surface of the reflector quench the reaction. 35 Time/Ms from lo FIGURE 5. Free surface velocity measurement from 3 mm Aluminium Alloy / 3mm stainless steel reflector experiments SI2D/12 and SI2D/13 compared to calculations of SI2D/12. A calculation was also performed on the same geometry where the EDC37 was modelled with programmed burn and a detonation point, rather than ignition and growth reactive burn treatment, and the resultant velocity profile is also plotted. It shows a terminal velocity of 1.75 mm/jas compared to 1.2 mm/us for the experiment and 1.6 mm/us for the reactive burn calculation. The free surface velocity trace on axis for the inert impedance match Kel-F sample experiment is also included in the figure, showing the drive given to the reflector discs from the donor charge alone. A simple velocity squared analysis suggests almost 50 % of the total detonation drive is released in the SI2D/12 experiment without detonation being established. SI2D/15 incorporated a single 3 mm Copper reflector disc. The free surface velocity measurements for this experiment are shown in Fig. 6 for comparison with a hydrocode calculation using the same reactive burn model used to calculate SI2D/12. The early time agreement is REFERENCES 1. Craig, B. G. and Marshall, E. F., "Decomposition of a Shocked Solid Explosive", in 5th Symposium (Int.) on Detonation, 1970, pp. 321-329. 2. Winter, R. E., Taylor, P. and Salisbury, D. A., "Reaction of HMX Based Explosive Caused by Regular Reflection of Shocks", in 11th International Detonation Symposium, 1998, pp. 649-656. 3. Lee, E. L. and Tarver, C. M, "Phenomenological model of shock initiation in heterogeneous explosives", Phys. Fluids 23(12), pp.2362-2372 (1980). © British Crown Copyright 2001/MOD Published with the permission of the Controller of Her Britannic Majesty's Stationery Office. 844
© Copyright 2025 Paperzz