ML19308C829

Experimental Study of Combined Physical-Chemical Explosions in Aluminum/Water Sys

ML19308C829
Person / Time
Site:	Crane
Issue date:	04/30/1979
From:	Richard Anderson, Armstrong D, Nicole Parker ARGONNE NATIONAL LABORATORY
To:
References
TASK-TF, TASK-TMR ANL-RAS-LWR-79, ANL-RAS-LWR-79-2, NUDOCS 8002070579
Download: ML19308C829 (27)

Text

{{#Wiki_filter:. _ _. _ M ANL/ RAS / LWR 79-2 AN EXPERIMENTAL STUDY OF COMBINED PIIYSICAL-CIIDilCAL EXPLOSIONS IN AN ALUMINUM / WATER SYSTEM by D. R. Armstrong, R. P. Anderson, N. E. Parker, l R. L. McDaniel and W. M. Stevens April 1979 NOTICE: This information document contains preliminary information prepared primarily for interim use in light water reactor p rograms in the U.S. Since it does not constitute a final report, it should be cited as a reference only in special circumstances, such as requirements for regulatory needs. Reactor Analysis and Safety Division ARGONNE NATIONAL LABORATORY 9700 South Cass Avenue Argonne, Illinois 60439 s o os 070 5 77

The facilities of Argonne National Laboratory are owned by the United States Goverrt-me n t. Under the terms of a contract (W-31-109-Eng-38) among the U.S. Department of Energy. Argonne Universities Association and The University of Chicago, the University employs the staff and operates the Laboratory in accordance with policies and programs formulated, ap-proved and reviewed by the Association. MEMBERS OF ARGONNE UNIVERSITIES ASSOCIATION The University of Arizona The University of Kansas The Ohio State University Carnegie-Mellon University Kansas State University Ohio University Case Western Reserve University Loyola University of Chicago The Pennsylvania State University The University of Chicago Marquette University Purdue University University of Cincinnati The University of Michigan Saint Louis University 4 Illinois Institute of Technology Michigan State University Southern Illinois University University of Illinois University of Minne sota The University of Texas at Austin Indiana University University of Missouri Washington University The University of Iowa Northwestern University Wayne State University towa State University University of Notre Dame The University of Wisconsin-Madison NOTICE This report was prepared as an account of work sponsored by the United States Government. Neither the United States nor the United States Department of Energy, nor any of their employees, nor any of their contractors, subcontractors, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the ac-curacy, completeness or usefulness of any information, ap-paratus, product or process disclosed, or represents that its use would not infringe privately-owned rights. Mention of commercial products, their manufacturers, or their s* ppli-ers inthis publication does notimply or connote approvalor disapproval of the product by Argonne National Laboratory or the U. S. Department of Energy. I l i

) 4 AN EXPERIMENTAL STUDY OF COMBINED PHYSICAL-CHEMICAL EXPLOSIONS IN AN ALUMINUM / WATER SYSTEM by D. R. Armstrong, R. P. Anderson, N. E. Parker, R. L. McDaniel and W. M. Stevens 4 April 1979 NOTICE: This information document contains preliminary information prepared primarily for interim use in light water reactor programs in the U.S. Since it does not constitute a final report, it should be cited as a reference only in special circumstances, such as requirements for regulatory needs. Reactor Analysis and Safety Division ARGONNE NATIONAL LABORATORY 9700 South Cass Avenue Argonne, Illinois 60439

i ' TABLE OF CONTENTS Page 1 . ABSTRACT............................................................. 2 I. INTRODUCTION................................................... 3 II. WATER J ET INJ ECTION EXPERIMENTS................................ 8 - III. FALLING PISTON EXPERIMENTS..................................... 13 IV. EXPLODING WIRE EXPERIMENTS..................................... 26 V. CONCLUSIONS.................................................... 27 REFERENCES........................................................... I i t i

11 a LIST OF FIGURES No. Tit 19 Page 1. ' Equipment for Inj ecting Low-speed Jets........................... 4 2. High Veloc i ty Inj ec tor........................................... 6 3. Equipment f or - Inj ecting High-speed J ats.......................... 7 9 4. Impact Apparatus................................................. 5. Equipment f or Explosion-initiated Tosts.......................... 14 6. Aluminum-drop Apparatus.......................................... 17 7. Pressure Pulses from Aluminum-water Vapor Explosion.............. 19 8. Exploding-wire Pressures......................................... 20 9. Nonexplosion Aluminum Residue.................................... 22 10. Aluminum Residue after Explosion................................. 23 11. Small Water Tank................................................. 25 1 J f

_ _ ~ 1 . / iii-LIST OF TABLES -~ jlo_.. ~ Title Pay - ' 1. ; Results - of Falling Pis ton Tes ts................................. 12-t 2.:'Results'of Explosion-initiated Tests............................ - 16 i ' i ~ i s s i 4 4 f, i + i W. 1-4 9 5 f 1 i i 4 I 4 1 r 1 i i-f 1-- t 4 I A d w. d-a e fet =s t 9 u-T -'WM" -"'T"c-WP' =m%-w m--7-w e'9+-+"?T- %?E-+---4 W-e 77---++m?'--- e--1 4 N?+------'-9 M' N -W WWWN r

= AN EXPERIMENTAL' STUDY OF COMBINED PHYSICAL-CHEMICAL EXPLOSIONS IN AN ALITMINUM/ WATER SYSTEM by D. R. Armstrong, R. P. Anderson, N. E. Parker, R. L. McDaniel, and W. M. Stevens ABSTRACT An experimental program to study types of energy production in small-ccale (< 1 kg of reactants) aluminum water vapor explosions is described. A number of experimental techniques for forcing the contact of molten aluminum cnd water are described. These included high and low velocity water jets, cudden compression of a system of dispersed water and aluminum, and the use cf an exploding wire to generate a shock wave within an aluminum water system. A triggering threshold which is required to generate a small scale vapor ex-plosion was determined. On a few ocassions, a chemical contribution to the physical explosion was detected.

2 I. INTRODUCTION Explosions have been produced by sudden contact between molten aluminum and water in industrial accidents and laboratory experiments. The high-pres-sure gas in these events may have been generated by a sudden vaporization of the water (vapor explosion) or by a chemical reaction between the aluminum Both types of reactions probably occur in real metal / water sys-and water. l must be caused by vapori-Explosions in molten-silver / water systems tems: zation, since the chemical-reaction potential is small; while experimental studies in molten-aluminum / water systems have identified three sizes of explosions with the largest size generally assumed to be chemical in nature.2 It is important to the understanding and analysis of explosive reactions that This the two energy-releasing mechanisms be isolated and studied separately. is a difficult experimental task, since a vapor explosion and a partial oxi-dation reaction would produce similar pressurization rates. It may be possi-ble to experimentally identify the oxidation reaction by measuring the total kinetic-energy releane, sudden local temperature increases, or emission of light radiation. An experimental program to study types of energy generation in aluminum / water reactions has been undertaken..The first requirement for such a pro-gram is a system capabic of generating reproducible explosions while recording accurate measurements of pressure, temperature, and liquid velocities. An initial decision was made to try to generate explosions by contacting small quantities of water (less than 100 g) with modest quantities of aluminum (less than 1000 g). This geometry enjoys numerous advantages over larga-scale sys-tems; problems of peraonnel safety and equipment shielding are minimized, pressure and velocity variations are easier to measure, the geometry of the reacting water mass is easier to control, and the smaller dimensions make efficient reactions more likely, thus aiding the discrimination between chem-ical and vapor explosions. The geometry does have one outstanding disadvan-No lar,ge-scale aluminum / water reactions have been produced in small tage: systems of this type although explosions have been produced in small aluminum / water shock tubes.3 I 1

3 The difficulty in generating explosions in small-scale systems was tentatively assumed to be caused by a stable vapor film separating the two j liquids. The ease with which explosions have been produced in large-scale cxperiments,4 was attributed to the collapse of this vapor film in the pre-2 i E cence of large-scale mixing forces. An experimental program, set up on the basis of this physical picture, attempted to initiate explosions in small-reale systems by penetrating or collapsing the vapor film. Initially, two cxperimental techniques were tried--water jets were injected into molten i aluminum, hoping to penetrate the vapor film, and a molten aluminum mass con-l taining an entrapped water volume was subjected to a sudden pressurization, l hoping to collapse the vapor film. Several non-instrumented scoping tests were designed to try these techniques. II. WATER JET INJECTION EXPERIMENTS l Low velocity water jets were injected into molten aluminum using the apparatus shown in Fig. 1. The melting furnace used a 1000 watt resistance 1 l heater, surrounded by 8 cm of insulation enclosing a stainless steel crucible l 20 cm long with a 5 cm ID and 1/2 cm thick walls. It was capable of heating 1 kg of aluminum to 1000*C in three hours. The first three tests injected water jets of increasing velocity across a 10-cm air gap into the molten-aluminum mass. The water injector was simi-lar to that shown in Fig. 1, but it lacked a transport cylinder and was thus locked in place during each test. The first test used a pressure of 9 stm to inject through a 15-cm-long, 0.3-cm-ID tubt - the second test again used 9 atm, but removed the tube and injected directly out of the solenoid valve; and the third test retained this geometry while boosting the injection pressure to 31 atm. The results of all three tests were similar--a benign aplashing of aluminum with no indication of an explosive reaction. I The effect of a noncondensible gas layer formed around the water jet as it traversed the 10-cm air gap might deter jet penetration through the gas / vapor film. An injector transport system was installed (as shown in Fig.1). Immediately preceding the water injection, the injector was lowered until the needle tip was 8 cm below the aluminum surface. A test with an injector pres-E l l

4 TRANSPORT CYLINDER sg t ll l +

MMy t

/ HIGH PRESSURE GAS /, h mN p..; 'b [ ',i, i ",lr ,, - WATER v q, WATER CHAMBER p s PRESSURE GAUGE WS SOLENOID VALVE HIGH PRESSURE o GAS LINE WATER INJECTION LINE' MELTING FURNACE IN EbTION iM g -{,p$$ 5I I' ,. -INSULATION HEATING ColLS 4 _ \\ ~ ;.-y.e 5. q@-~ MOLTEN ALUMINUM [5'55 J ' hs(s-2N fE ~23 Fig. 1. Equipment for Injecting Low-speed Jets

5 cure of 25 atm again produced benign splashing with no apparent large pres-cures. Concurrently, another series of tests was being conducted in an attempt to produce a violent surface reaction between a molten-aluminum mass and ultrahigh-speed, small diameter water jets. The high-velocity water was ejected through an orifice plate in contact with a small water volume which w:s pressurized by an exploding wire (Fig. 2). The first test impacted seven 0.25-mm-dia jets on a molten-aluminum surface without effect. The jets were photographed with a 0.5-ps-duration flash and found to consist of a spray of fine droplets incapable of penetrating the oxide scum floating on top of the molten aluminum. A new plate with a single 0.5-mm-dia polished orifice was installed. The resultant water jet was capable of penetrating an 0.08-mm-thick brass shim and was shown by photographs to consist of a single jet with c velocity of at least 400 m/s. This injector was then incorporated into the apparatus shown in Fig. 3 in which a furnace was placed inside a safety cnclosure equipped with windows for photography and a transport cylinder was used to mcVe the injector into close proximity to the molten aluminum surface. The injector was activated by the contact of a sensor with the aluminum, al-lowing the injection distance to be varied from run to run. Three additional orifice plates were made with polished orifice holes of 0.6, 1.3, and 2.6-mm diameters. Photographs of the jets from each ori-fice plate showed that the water was ejecced in the form of a smooth jet, periodically surrounded by clouds of fir e particles. The diameter of the cmooth jet was the same as the diametr.r of the hole in the orifice plate. The cloud formation was assumed to ts due to periodic high-pressure spikes at the inlet to the orifice hole, with the spikes, in turn, due to cither sonic resonance in the water cavity or current oscillations through the exploded wire. By chance, the frequency of cloud formation, sonic-resonance frequency in the water avity and current-oscillation frequency were all equivalent, about 60 kHz. Teste proved the jets were capable of penetrating a 0.08-mm-thick brass shim. Thus they should be capable of pen-strating any scum on a molten-aluminum surface. High-speed jets from the orifice plates were injected into a crucible (28-mm ID and 76 mm long) con-l

6 A, u) '] ,v ,g [ sji \\ / Q x(', Q e -m lAs. k l . DEMOUNTABLE NOZZLE / EXPLODING WIRE Fig. 2. High Velocity Injector

7 i /b ,/ \\ TO s ')hc) e: ACTUATOR SENSOR HIGH VOLTAGE SOURCE C \\ c m Q. I ', CYLINDER REVERSING GAP I INJECTOR BOTTOM HE AT SHIELD TRANSPORT CYLINDER-REMOVAL MECHAMSM HIGH VOLTAGE /- C I SOURCE- / s ENCLOSURE - 7 N 'N 1 s h k 7-INJECTOR ~ /* l (,. g< ,9} / ,s x 1) LIGHTS "~ / V,; l1 ls} } 4 y REVERSING SgI. ~i l ' ARGON ; HOSE .r t / GA L i I THERMO-i c '"fecro"lW) ' # li I I /N COUPLE p' T UaWW JL !\\ .i x ~- $[7 / . :t 3Yi 1Grpu ii MELTNG FURNACE-]',.yM; N55 !f HEAING b / 4M) '/=$ l U'\\ si-l di N V"f' 1 Ili i 25-- MOLTE ALUMINUMn,,A ~ NV m/ NSULATION -Q f [A AC Nj f TO DIGITAL VOLT-f F LCAMERA METER SMALL' SCALE Al H,0 V Fig. 3. Equipment for Injecting liigh-speed Jets

8 taining 130 g of molten aluminum at 800-1000*C. The mass of the ejected aluminum varied between 4 and 60% of the total aluminum mass in the crucible; for comparison, when the same high-speed jets were injected into the same crucible filled with water, about 2% of the water mass was splashed out of the crucible. The increased ejection fraction indicates that some type of reaction occurred between the water jet and the molten aluminum. Ilowever, the reac-tion was relatively benign--no over-pressure was heard or measured and the fine dispersion of reactants, observed in other small-scale experimental vapor explosions, did not appear. This experimental approach to producing small efficient aluminum / water explosions was abandoned, as was the equally unproductive use of low speed jets. It was then decided to try to produce an explosion by collapsing the 4 vapor film surrounding an entrapped water droplet. III. FALLING PISTON EXPERIMENTS The equipment shown in Fig. 4 was designed to initiate a reaction by impulsively loading an aluminum mass, which contains an entrapped volume of water, by dropping a steel cylinder, which then impacts on a graphite toroid immediately above the moltan aluminum. The mechanical energy produced by any resultant subsurface reaction can be evaluated from the rebound velocity of the steel cylinder. A series of tests was conducted to assess the impulsive load seen by an entrapped water volume and the natural rebound velocity in the absence of any subsurface reaction. To calibrate the impulse loading due to the dropped steel cylinder the subsurface pressure produced by the impact of the falling cylinder on the graphite toroid was measured with a water filled crucible. The effects of cylinder mass and drop height, graphite-toroid clearances, and compressible-gas volumes within the liquid mass were investigated. With a tight toroid- (0- - 0.5-mm clearance), an 18.2-kg mass dropping through a 60-cm height before impact produced a pressure history with a symmetrical, linear rise and f all, a total period of 3.6 ms, and a peak pressure of s20 MPA. The cylinder re-bounded 10-15 cm af ter impact. Measured impulse I (where I is defined as the

9 NJ- // 1 (% (b h4 0# i /J \\~ 00' {0 N- .i W'a i 1 @ s v CONCENTRIC TUBES l I eWATER FILLED ^' s@ sY / GLASS SPHERE I 'f' e ~ m '8 N. xi - f fc i 7 a 1J [,- I. J mlj,N

s GRAPHITE l

l l l j-4 PISTONH l l f d-H;l h - l i l MOLTEN i s ~ ' ) ALUMINUM 4 / GAS u 8 l f s s s / GLASS SPHERE sN,A l l I ^ MELTING FURNACE - L l ,-,-,,,w,,,,,,,,,,,,- ,,/<-,,,,-,w,,,,,~-,,-- GLASS SPHERE OUTER TUBE BOTTOMING IMPACTING ON PISTON Fig. 4. Impact Apparatus

10 time _ integral of reaction force) was in reasonable agreement with the theore-tical impulse produced'by a liquid-tight, no-loss system with a 17% coeffi-cient of' restitution: Imeas/Icalculated = 0.91. As expected, the impulse varied directly as the first power of the cylinder mass and the square root of Small (0.5-cm ) compressible-gas volumes within the water 3 the drop height. mass had no measurable effect on the impulse. A second series of tests was run with a slightly looser toroid (0.8-mm ' radial clearance). The impulse dropped to 28% and the peak pressure to 18% of the_ equivalent values measured with a tight piston. Water was splashed up to the test-cell ceiling (6 m above the crucible). The drop cylinder did not visibly rebound. Several tests were run with this rig using the tightly fitting toroid and a drop height selected to give an impulsive pressure (25 NTa) larger than the critical pressure of water for a duration of about 1 msee based on the above tests with water. (Note: As described' later in this report, subsequent measurements showed that the pressures produced in an aluminum filled cruci- .ble were lower than pressures produced in an equivalent water filled crucible. The actual pressure produced in these tests was probably on the order of 40% i of the critical pressure of water.) With about 0.7 gm of water inserted into 1 kgm of aluminum at 900'C no efficient explosions were produced. A new series of tests was devised which used an explosive device (a pistol cartridge primer) inside the water-filled glass sphere to disperse the water into the molten aluminum before the impact of the impulse cylinder. Various schemes for assembling, waterproofing, and mounting primers and anvils ~ i inside water-filled glass spheres were tried. All schemes failed when the ( . primers failed to_ ignite'inside the glass sphere, although they readily ex-ploded, upon impact, outside the water-filled sphere. A small exploding wire was built to fit within a water-filled glass sphere; the resultant explosion.readily dispersed the water and the glass shell.- The amount of dispersion was tested by filling the sphere with dyed-L water and exploding the wire in a beaker filled with clear water. High-i. . speed: movies (2500 pps) showed.the initial explosion was followed immediately

11 (in the next frame) by a fine dispersion of cavitation bubbles throughout-the entire beaker. The cavitation bubbles collapsed behind slowly advancing fronts, so the beaker again appeared transparent af ter about 5 ms. The final collapse generated a second smaller shock with the attendant cavitation bub-bles. A series of expansions and collapses periodically generated cavitation bubbles while slowly dispersing the dyed water throughout the beaker. The impact apparatus was then modified by placing an exploding wire in-cide the glass sphere and solidly fixing the tube carrying the sphere to the -outer tube. The two concentric tubes were arranged so the sphere was midway b tween the top and bottom surfaces of the aluminum when the outer tube im-picted on the graphite piston. The exploding wire was triggered by a signal from a newly installed force transducer beneath the crucible so that the dispersion occurred while the system was compressed. This apparatus was first calibrated by dropping into a water filled crucible to determine the effect of the exploding wire energy on the measured impulse and the piston rebound. When dircharge energies (the energy stored in the capacitors) ranged from 300 to 3000 J, the effect on the measured data was within the normal reatter of the data. Several tests were then made dropping the piston, without water, onto molten aluminum. This system turned out to be much sof ter than the water system with a lower peak force and longer duration of the impulse. The peak force measured for the impact on molten aluminum was about one-third of that for a water-filled crucible. The piston rebounded was reduced by a similar amount. It is reasonable to assume that the reduction in peak pressure was of the same order, so that the initial impact experiments did not produce pressures of 25 MPa, but only about 8 MPa, well below the critical pressure of water. Table 1 lists the results of experiments made with this apparatus using an exploding wire for dispersing the water into the aluminum. After the first three runs the apparatus was modified by changing the crucible design. The original crucible was a length of 304 stainless steel pipe directly in contact with the molten aluminum. The stainless steel dissolution was so rapid ci,st the average lifetime was about the heat up time for an experiment. New cruci-bles were made of graphite with the same internal dimensions (5.8 cm ID x 28

12 TABLE'I. RESULTS OF FALLING PISTON TESTS Aluminum. Water Wire ' Output -Mass Temperature Mass . Temperature Energy Energy Efficiency gm 'C gm 'C J ~J 900-780 1 20 181 8.2 900 880 1 20 125 5.2 900-880 1 20 250 8.3 900 800 1 20 320 0 0 900 800. 1 20-320 0 0 900 800 1 20 380 0 0 900 800 1 20 380 357 13.8 900 800 1 20 720 110 4.4 900. 800 1 20 3400 202 7.8 900 -900 1 20 1500 .179 5.9 - 900 800 1 20 3420 540 900 1000 1 20 6000 100-3.6 900 1140-1. 20 380 47 1.23 i l I s. l i i-2_ ;

13 cml deep) placed on a tight fitting stainless steel sleeve. The output ener-gies-listed in the table are measured from the motion of the piston alone and the e'fficiencies are based upon the maximum work that could be done by the isothermal expansion of-.the water at the aluminum temperature from its initial liquid volume to the free volume of the crucible with the piston and graphite annulus removed. The very rapid dispersal of aluminum during the piston re-bound, as shown by the movies, indicated significant energy was being dissi-pated with the metal spray. The thrid run could be seen with enough clarity that a rough measure of the energy in the aluminum dispersed by the explosion was obtained by measuring the expansion velocity of the cloud of aluminum particles released when the impact weight cleared the end of the crucible. The mass of aluminum dispersed was estimated from the positions of the impact weight at the time of the explosion and when it reimpacted on the aluminum. The energy in the dispersed aluminum was 1100 J, and the total measured output energy in this. experiment was 1400 J, or 25% of the theoretical maximum energy that could be obtained from 1 g of water heated to 880*C and allowed to expand isothermally to 1 atm final pressure. These data indicate a threshold energy requirment for the exploding wire before any significant interaction occurs, however, once that level is reached, .there is no correlation between the input and output energies. No clear evi- .dence of a chemical reaction is shown as that would require an efficiency greater than 100% of the available physical interaction energy. IV. EXPLODING WIRE EXPERIMENTS A'second method of collapsing the vapor film between the aluminum and water was devised. For this system a small quantity of aluminum surrounded Eby water is shocked by an external explosion, forcing the two liquids into contact. Figure 5 shows the apparatus initially built to test this concept. 1 It. is designed : to spill molten aluminum into a beaker of water and then, fol-lowing a preset delay, explode a wire within the beaker by a massive electri-cal discharge. The shape of the aluminum within the beaker is partially con-trolled by the delay time. The size and shape of the wire-generated pressure -pulse is determined by the location of the wire and the characteristics of the discharge current.

14 TRIPPING SOLENOID ,.c, / \\ TILTING FURNACE ~' + /- , \\ .,/ [. ,'s ,N s s '-, t COUNTER WEIGHT t LITER BEAKER MOLTEN ALUMINUM i l SENSING GRID f TIME DELAY GENERATOR [ h/ i; w t l l ,A h EXPLODING WIRE l N% ,s 'N j WATER \\ 'xbj CAPACITOR BANK Fig. 5. Equipment for Explosion-initiated Tests

15 4 Seven tests were run in this equipment. Results are shown in Table 2. 'Vepor explosions were generated in three of the seven tests. The high-speed j movies showed that there was a short delay period between the wire explosion and the initiation of the aluminum-water reaction. These were rough, scoping tests without rigid control on the initial conditions (shape and location of the sluminum stream,.etc.) or much instrumentation, so that exact determina-tion of the explosion efficiency was difficult. Based on liquid velocities measured from the high-speed movies, the explosion in test 5 produced between one and 10% of the mechanical energy that could be produced by 2 vapor explo-sion with the injected quantity of molten aluminum. New apparatus as shown in Fig. 6 was developed. From 10 to 30 gms of aluminum were melted in a boron nitride crucible equipped with a plug valve in the bottom. The aluminum was dropped into a heavy wall stainless steel tank,15 cm long,10 cm wide and 13 cm deep, with 2.5 cm thick Lucite side windows. The tank was supported on a force transducer and pressure trans-ducers were mounted in each end. After a variable delay a wire was exploded in the water to initiate the vapor explosion. Several test photographs were taken at 200,000 frames /s using a 1.7-kJ ' flashlamp mounted in a parabolic reflector. This was used in a back-lighting mode, which silhouetted both the aluminum and the surrounding water vapor. Thus for these test films, the progress of the explosion could only be fol-loved by tracing the motion of the aluminum envelope, which may have been either the water / aluminum or the water / water-vapor interface. For a typical run, the film shows the aluminum pooled on the bottom of the water tank in a thin bubble 5-10 cm'in diameter. Several frames after the wire explosion, the portion of the puddle interface closest to the wire begins to move as if the vapor layer between the aluminum and water were collapsing. Wich suc-ceeding frames, portions of the puddle farther from the exploding wire begin to collapse and those closer to the wire show a motion reversal and start to l expand. This collapse motion propagates along the surface of the aluminum puddle with'a velocity of 0.5-1.5 km/s. Normally, runs were made while photographing at a much slower speed (5000 fps)..These slower but longer duration filsm showed different aspects I

TABLE 2. RESULTS OF EXPLOSION-INITIATED TESTS" Time Delay, First Aluminum Aluminum. Beaker Water Liquid Contact Run No. Mass, g -Temp. 'C Material Temp.

• C to Wire Explosion, s Results 1

.100 S1000 Stainless steel 20 .By Hand, NO.5 Possible explosion; fine. Al dust in beaker (steel beaker bottom bowed). No film. 2' 100 4900 Pyrex 20 0.5 No apparent result or explosion (glass beaker shattered). Pyrex 20 Wire calibration smaller 3 None (slightly) than run'No. 2 Feaker broken. 4 100 N900 Pyrex 20 0.1 Very small (about like run No. 2). Beaker shattered. 5 90 %950 Stainless steel 445 0.8 Large explosion. Belled bottom of beaker. De-stroyed entire furnace and crucible. 6 90 %900 Pyrex %55-60 0.8 More 'riolent than x wire alone. -7 83.5 4800 Pyrex %45 0.22 Seemed less violent than run No. 6. " Test conditions: Water mass, 900 g; exploding-wire dimensions, 25 mm x 0.25 mm, 380 J discharge energy

~ ~. ) 17. DROP SOLENOID s o ? N 4 t .y E' d DROPPING CRUCIBLE ~ ^ l . g... a~ H .,Y

, e js.i QUENCH TANK q K

\\ NyN! \\- [' - g, L -O N yy. .. io j kNhk h 'h>k '? ? p?d kN' 12' fik, 5 l PRESSURE TRANSDUCER 'Ik). jyjg$k [ t f (BOTH ENDS) III<'f. ,p l$I' R~;5 / f',$dg!g;#,- y,'id",od',h'<?,p,%; d,,,;g,, / / se ' p5 [ g, ,i ) f,j,j/,,/ j / CAPACITOR BANK h EXPLODING WIRE FORCE TRANSDUCER / Tig. 6. . Aluminum-drop' Apparatus-

i 18 explosive event. The direct effect of the wire explosion was the production

of -a shock wave which collapsed the vnpor film surrounding the aluminum.

For-Elow-pressure shocks, the film collapse produced a email disruption and expan-sion of the aluminum pool, which then collapsed and returned to its initial state. No effects significantly different from the exploding wire alone were observed. This expansion and return to a molten pool were seen even for weak shocks less than 1 MPa. High-pressure shocks initiated a different behavior of the aluminum. After the collapse of the vapor film the pool expanded rapidly (450 m/s) and fragmented, expelling much of the water from the tank. The pressure threshold dividing these two types of interaction was about 5 MPa. For one experiment with a shock wave of this magnitude (produced by an 80-J discharge), the aluminum pool first expanded to about twice its original diameter and then collapsed, triggering a vapor explosion in the pool of molten aluminum, and in a drop of aluminum that was falling through the water. This sequence of events provided an unambiguous separation of pressure pulses resulting from the exploding wire and from the vapor explosion. Figure 7 shows the pressure pulses resulting from the vapor explosion in this test. These pulses are short with widths of about 10 ps and amplitudes of a few MPa as measured 5-10 cm from the source. This is similar to the pulses generated by the exploding wire as seen in Fig. 8. Thus in the more energetic experiments f n which the vapor explosion occurred in one part of the aluminum pool while vapor collapse was still occurring in another part, the separation of pressure pulses causing and resulting from the vapor explosions is nearly impossible. However, two system parameters were more easily measured and show signif-icant differences between explosicn and nonexplosion cas s. First, the im-pulse produced by the wire explosion on the base of the _ank as measured by the supporting force link was less than 1 N s for the exploding wire alone. i l_ For a series of 10 explosive runs with discharge energies of 80-700 J, the im-pulse ranged from 5 to 10 N*s with no apparent correlation between initiating l energy and the resulting impulse. The discrepency between the measured pres-sures and impulses was due to the limited sensitivity of the pressure trans-l

19 Oo i I dI oo E o om oo N._ o o o o uo o u m 4 o 8 o a S 1 .3 o ~ SE o2 mC eo O w a+ e-18 8 ~g a 8s g o o 8u mo o c. w o EE 8 oc

a. -

a o e .~ o W o om a. o 5.e o o N g ao a a,c. o o -eg k i i l i o o m y odW '380ss38d

20 i I I I o - o e o - Oe P u a. 2 m. } m a. o - o R m a w ? E 3 w Eq n o .e _ o m a C l o - o m i I A o m m o DdW '3 BASS 3Bd l

21' du'ars which had been selected on the basis of surviving the peak pressure c Edeveloped by the exploding wire..Thus pressures somewhat less than 1 MPa could not be measured. Second, the extent of fragmentation of the. aluminum residue after an cxperiment varied markedly for the explosion and nonexplosion case. Figures 9 and 10 compare the aluminum residue after a vapor explosion run with one in which the aluminum was' fragmented by the triggering pulse but then collapsed End~ coalesced back into a single! aluminum puddle. Sieve analysis for two runs, initiated by shocks of different intensities and resulting in vapor cxplosions, gave similar results. For an initiating shock of 15 MPa and discharge energy of 700 J, the mean particle diameter was 2.8 mm. For a less violent-initiator, 5 MPa and 80 J, the nean diameter was 2.5 mm. These mean diameters are based on the mass average of spherical fragments, whereas the fragments are actually thin platelets. The energies produced by these explosions were determined from the measured impulse and the mass of water expelled from the tank: '2 E=I-- 2m These were rather small, ranging from 0.5 to 1.0 J/gm of aluminum, which is less than 1% of the maximum work that could be produced in an optimum situ-7 ction as determined by an analysis based on the method of Hicks and Menzies which gives' values ranging from 250 to 400 J/gm of aluminum for the maximum theoretical work that could be produced in a physical interaction with the initial conditions used in these experiments. A final set of experiments was run using the boron nitride dropping cru-cible and a small water tank (Fig. 9) and looking for evidence of a chemical interaction by other means than the measurement of efficiencies. The runs were made by dropping about 15 g of molten aluminum at tempera-tures of 800-1200*C into the small tank of water. The tank was a rectangular cavity (3.8 x 5.0 cm) in a 2.5-cm-thick stainless steel plate covered on each side by a 2.5-cm-thick glass window. A 1.9-cm-dia hole allowed aluminum to i e

22 M I [- g ~- I i g i L. $. ~. l .i %' ll e *i -?3 v-s g I ) 3 '?? . g d. i ? ~j '] s; y ~ , ;i < or :, sw I l Fig. 9. Nonexplosion Aluminum Residue -.<------.--v-e

23 l g ~ [ r + l'. .a, iN9 \\e l '~ .p. r s N' p V.. .^ Y. '% t y l ', i _.$' F ,p' ~'- ~ iI y 1 2 {f Q j $ ; ; ~- Y WY h\\ \\n . ~. IS-10. Aluminum nesidue after Explosion

24 be dropped into the tank; a pressure transducer centered in one short side ' faced an exploding wire in the opposite side. Two parameters were varied in the series: aluminum temperature and lighting method. The tank was always l filled with hot water (>90*C).- Three methods of lighting the experiment were used: no external light, i.e., only the light from the exploding wire trigger; a continuous 10-kW xenon are lamp; and a short-duration (41ps) flash lamp, synchronized to the camera shutter and arranged to operate on alternate frames. Two runs in which a chemical reaction occurred were illuminated by the arc lamp and had an initial aluminum temperature of about 1000*C; a third such event used 1200*C aluminum and no external light. All events had a similar appearance in photographs; several patches of aluminum that were expanding as part of the physical explosion had a ruddy glow that was obviously different from the specularly reflected arc light on the surface of the other aluminum. This light covered about 5% of the visible aluminum surface and had a color tem-perature between 2600 and 3000 K. The temperature was estimated by comparing it to a photograph on the same film emulsion of, a set of tungsten light bulbs at different measured temperatures. The intensity of the aluminum light was much less than that of the tungsten filaments, indicating a diffuse source. For the optimum case, in which a reaction occurs between stoichiometric amounts of steam and aluminum at the bulk aluminum temperature according to 2A1 + 3H O = Al 03 + 3H - 2 2 2 there is no net generation of gas, and the increase in available work is due to the thermal effects of the reaction. The adiabatic flame temperature is about 3000 K, a reasonable agreement with ta i measured color temperature. Only these three of a series of 30 runs show this clear evidence c' a chemi-cal interaction occurring during a physical explosion. The measured pressure and force from these runs were not significantly different from other runs with no chemical interaction. This would be expected from the small amount of aluminum interacting and the relatively small additional energy that would be produced by a chemical interaction. l.

as 1 PRESSURE TRANSDUCER WATER FILLED CAVITY hs (~ '/j/f! 'e, ' l' y,, $" Tj l Q 'A WINDOW i (fiQ>$ w q. Q; - Ng<d c s agNgg ^ pyg/lI,

• ldlll EXPLODING

%34 M WIRE N / / / WINDOW Fig. 11. Small 11ater Tank l

26 9 Since the work output of the system is proportional to the gas tempera-ture, the maximum work produced by the chemical interaction is less than twice that produced by the physical interaction. The run-to-run variation in energy ~ .for both the aluminum-drop and the falling-piston experiments is greater than -1300%. Thus chemical explosions could not be detected by energy measurements alone in these experiments. V. CONCLUSIONS Several conclusions can be drawn from these data. A definite-trigger threshold for.the initiation of a small scale vapor explosion does exist while the explosion energy appears to be independent of the trigger energy. A cherical interaction did occur occasionally as part of a vapor explosion although the chemical energy contribution did not add measurably to the over-all event. L t I i 'l

4 27 REFERENCES 1. j-R. W. Wright, A. F. Fustenberg, G. H. Hanberstone, L. G. Neal, L. B. Wentz, and S. M.' Zivi, Kinetic Studies of Heterogeneous Water Reactors-Annual Sununary Report-1966,, USAEC Report STL-372-50 (Dec 1966). ' 2.- P. D.1 Hess and K. J. Brondyke, Causes-of Molten Aluminum Explosions and Their Prevention, Metal Progress, 95,(4),-93-100 (1969).

3.. - R. W.-Wright.and G.'H.-Humberstone, Dispersal and Pressure Generation

- by Water Impact-upon Molten Aluminum, Trans. Am. Nucl. Soc. 9,, 305-306 (1966). G. Long, Explasions of Molten Aluminum in Water-Cause and Prevention, 5. Metal Progress, _7_1,(5), 107-112 (May - 1957). 1 5. ~J.'N. Guest, R. G. Turner, and N. J. M. Rees, Water Shock Tube Experi-ments-at A.W.R.E. Foulness Since January 1972, FPR-1-74 (Jan 1975). 16. A. J. Briggs, Experimental Studies of Thermal Interactions at AEE Winfrith, Third CSNI Specialist Meeting on Sodium Fuel Interactions in ' Fast Reactors, Tokyo, Japan ~(Mar 1976). 7. G. P. Hicks, and D. C. Menzies,' Theoretical Studies on the Fast Reactor Maximum. Accident, ANL-7120, Argonne National Laboratory, pp. 654-685 October 1965). s D}}