Water Explosives

Research on water explosion inhibiting systems is providing an avenue of future protection possibilities against vapor cloud explosions. British Gas experimentation on the mitigation of explosions by water sprays, shows that flame speeds of an explosion may be reduced by this method. The British Gas research indicates that small droplet spray systems can act to reduce the rate of flame speed acceleration and therefore the consequential damage that could be produced. Normal water deluge systems appear to produce too large a droplet size to be effective in explosion flame speed retardation and may increase the air turbulence in the areas. 

Molten Metals and Water Explosions, HSE Rept., London, HMSO, 1979... 

Whereas one might classify the LNG-water studies as a response to a concern that industrially sized operations might result in a large-scale spill on water with subsequent RPTs, studies of molten salt-water explosions were carried out because industrial accidents had taken place. Emphasis has been placed on events occurring in the paper industry where molten smelt is produced in recovery boilers. This smelt is primarily a mixture of sodium chloride, sodium carbonate, and sodium sulfide. In normal operations, the molten smelt is tapped from the furnace, quenched, treated, and recycled to the wood digestors. Accidents have taken place, however, when water inadvertently contacted molten smelt with severe explosions resulting. The smelt temperature is much higher than the critical point of water 1100 K compared to 647 K (see Section IV). 

Laboratory investigations into the mechanism of smelt-water explosive boiling events have been primarily of value in delineating the effect of smelt composition on the sensitivity of the salt in producing RPTs. For example, pure molten sodium carbonate has never led to explosive boiling. Addition of either (or both) sodium chloride or sodium sulfide lead to smelts which are more prone to explosive boiling. Investigators experimented with many additives both to the smelt and to the water in an attempt to obtain less sensitivity. Most had little or no effect. 

The superheated-liquid model introduced earlier to explain LNG-water RPTs was not considered applicable for smelt-water explosions since the very large temperature difference between the smelt and water would, it... 

Experimental test results for molten aluminum-water RPTs are described in Section V. Also shown is a tabulation of most documented aluminum-water explosive boiling incidents (see Table XIV). In many accidents, the quantity of water was quite small, e.g., some resulted when wet aluminum ingots were loaded into melting furnaces containing molten aluminum. In contrast, one notes that few, if any, serious events have ever been obtained when small quantities of aluminum were contacted with a large mass of water. Since laboratory tests were often carried out in the latter fashion, most of these have produced negative results. 

As with the smelt-water case, if an RPT did take place, the event was localized and rarely was dam e severe far from the site of contact. Modeling molten aluminum-water incidents (and, in fact, other molten metal-water explosions such as in the steel industry) has not been partic-... 

In certain instances, when LNG contacts ambient water, explosive vaporization occurs with concomitant shock waves both in the air and water. While isolated instances of such events were recorded as early as 1956, it was during the 1968-1969 Bureau of Mines tests that the phenomena first attracted wide interest. In these experiments, three explosive... 

Liquid propane spills into 341 K water. Explosive vaporization always occurred and the delay between the start of the spill and the event was typically 0.2 sec. Strain-gauge pressure transducers were located 7.6 cm under the water interface and also in the air, 1.38 m from the spill container. The overpressure data are shown in Table I. The highest overpressure measured in the water was 410 kPa (60 psi) and the highest... 

Fig. 1. RPT compositions for an ethane-propane-n-butane system on 293-K water ( ) explosion (0) pop (O) boil.

Chemical reactions were shown to play a minor role in a smelt-water explosion, although gas samples from kraft smell-water incidents showed that hydrogen evolution could be correlated with Na2S content. There was general agreement that the explosion mechanism was physical in nature. 

Apart from the laboratory studies, statistical surveys of actual recovery bmler explosions have shown that such incidents are relatively rare and, in the United States occur, on the average, about once every 100 years of boiler operation. All explosions have been traced to events which allowed water to enter the furnace and contact the smelt, e.g., broken water tubes or dilute liquor feed. A listing of the presumed causes of all known smelt-water explosions is given in Table XII. 

Few in-depth studies have been made of actual furnace smelt-water explosions and, therefore, it is difficult to delineate expected overpressures and impulses. One case history is presented to indicate in a qualitative fashion the type of damage in a large explosion. 

As noted earlier, no viable theories of the smelt-water explosion had been widely accepted during the early period of investigation. Nelson in... 

The smelt-water explosion problem is of primary interest in kraft recovery furnaces where, from operational error or an equipment failure. 

The research results clearly indicated that the smelt-water explosion yielded a localized, high-energy shock wave which moved at velocities of over 700 m/sec. The maximum pressure rise was achieved in about 1 msec. These facts were related to the results found in real boiler explosions in descriptive terms as there were 3-4 inch depressions 3 to 6 feet... 

Lougher et al. (1968) reviewed the situation, presented some additional data, and developed recommendations for further work. They examined existing theories to explain smelt-water explosions and rejected all. They too noted that sodium aluminate in the smelt reduced the probability of an incident and also stated that CaCOj and Fe203 were effective. Most other additives (and over 90 were studied) either led to more violent explosions or were ineffective. No correlation of the smelt additive results was given. Also, water with various additives such as surfactants, starch, sucrose, glycerine, and hydroxymethyl cellulose still exploded when contacted with a sensitized smelt. 

Krause et al. (1973) carried out the last detailed U.S. experimental investigation of the smelt-water explosion phenomenon. A large number of experiments were conducted with variations in the smelt composition. The scale was quite small with 0.03-1 g quantities of water injected at high velocity (20-30 m/sec) onto the surface of the smelt. A few tests were also made with small drops (0.8-0.3 g) of water on the end of a ceramic tube that was dropped into the smelt. Some information concerning pres-... 

A model based on inertial restraints was developed, but it has not been widely accepted since it does not explain many of the experimental facts for smelt-water explosions. 

Taylor and Gardner (1974) and, more recently, Grace and Taylor (1979) have made detailed statistical analyses of the reported smelt-water explosions in modem recovery boilers operating in the United States and Canada. There have been 77 incidents in the period from January 1958 through July 1979. Of these 77 explosions, 55 occurred in the United States and 22 in Canada. The time period covered 5120 boiler-years of operation in the United States and 1070 boiler-years in Canada. 

The following was excerpted from a damage report of a smelt-water explosion. 

In a somewhat different type of furnace that suffered from a combustion-gas explosion, the floor beams, which were similar in size, were deflected at most 2-3 cm. A structural analysis of this explosion led to the conclusion that peak pressures were in the order of 30-40 kN/m (5-6 psi). Comparing the two damage descriptions, it is obvious that the smelt-water explosion generated pressures well in excess of 40 kN/m on the floor. 

While the mechanism proposed by Nelson explained many of the characteristics of a smelt-water explosion, it had one very serious drawback, i.e., the smelt temperature was significantly higher than the expected superheat-limit temperature of water (1100-1200 K compared to 577 K). For LNG-water, it was shown earlier in Section III that if the water temperature were much higher than the superheat-limit temperature of the LNG, explosions were then rarely noted. For such cases, the filmboiling mode was too stable and collapse of this vapor film was unlikely. 

In this article, we suggest that a modified superheated-liquid model could explain many facts, but the basic premise of the model has never been established in clearly delineated experiments. The simple superheated-liquid model, developed for LNG and water explosions (see Section III), assumes the cold liquid is prevented from boiling on the hot liquid surface and may heat to its limit-of-superheat temperature. At this temperature, homogeneous nucleation results with significant local vaporization in a few microseconds. Such a mechanism has been rejected for molten metal-water interactions since the temperatures of most molten metals studied are above the critical point of water. In such cases, it would be expected that a steam film would encapsulate the water to... 

Besides the aluminum industry, the nuclear power industry has been interested in molten aluminum-water explosions due to the presence of aluminum metal in some boding water reactors. Certain accident scenarios lead to a meltdown of the reactor core with concomitant contact of molten aluminum and water. 

Results from extensive test programs on molten aluminum-water explosions have been reported by Long (1957), by Hess and Brondyke (1969), and by Hess et al. (1980). In almost all experiments, molten aluminum, usuaUy 23 kg, was dropped into water from a crucible with a bottom tap (see Fig. 9). In only a few tests was there instrumentation to indicate temperatures, pressures, delay times, etc. The test results were normally reported as nonexplosive or explosive—and if the latter, qualitative comments were provided on the severity of the event. A large number of parameters were varied, and several preventative schemes were tested. Over 1500 experiments were conducted. Some of the key results are summarized below. ... 

It is clear that the Alcoa research teams have provided a valuable data base to examine the mechanism of aluminum-water explosions. However, before considering proposed theories, the research studies at Argonne National Laboratory and elsewhere are summarized. 

Higgins quotes unpublished work at both Aerojet and DuPont that aluminum-water explosions were difficult to obtain, but alloys of aluminum with small amounts of lithium, sodium, or uranium were quite reactive. 

In Table XV, some data are given for molten metal-water explosions not involving aluminum. In most of these cases, the quantity of metal was large. Because only the more serious explosions are reported, these tend to illustrate the most damaging type of event. Again, it is interesting to note that usually only small quantities of water were involved. 

Extending this concept, we now consider those experiments which led to molten aluminum-water explosions without the presence of a wet, solid surface. In all of these there was an external shock applied to the system—usually in the form of an exploding wire or a detonator. As presumed by the investigators, these artificial shocks could be very effective in collapsing steam films. 

Very few molten metal-water explosions are well documented essentially no data are available to estimate the overpressures or force-time relationships. The few incidents which have been described in any detail suggest that a two (or more)-step sequence is involved. First, contact is made between water and molten metal. Second, the mix is tamped or a shock wave occurs near the mix. The resulting explosion is sharp and has an associated blast wave. 

Dewing, E. W. (1980). The initiation of molten aluminum-water explosions. Memo., AIME Meet., 1980, Las Vegas, Nevada. 

Hess, P. D., Miller, R. E., Wahnsiedler, W. E., and Cochran, C. N. (1980). Molten alumi-num/water explosions. In Light Metals 1980 (C. McMinn, ed.), p. 837. (Proceedings of Technical Sessions Sponsored by TMS Light Metals Committee at 190th AIME Annual Meeting.)... 

---