“What Really Sank the Maine?”
(See T. B. Allen, pp. 30-39, March/April 1998 Naval History)
Ib S. Hansen
Mr. Allen has provided a well-worded article, but it obfuscates some key technical problems and gives the false impression that they have been solved correctly. The presented analysis indicates what has been suspected for a century—both a coal bunker fire and a mine were possible initiation sources for the magazine explosion—but provides no insight into the question of which one it actually was. The conclusion that the probable cause was a mine is based simply on an opinion suffering from some of the same mistakes made by the 1911 investigation. The analysis skips over the question of what mechanisms would be involved if a mine were to ignite the stowed gunpowder. It ignores the damage a mine capable of penetrating the double bottom and igniting the magazine would have made on the outer bottom. It misinterprets some definitive and conclusive features seen on the wreck. And it ignores a large amount of historical data, which for all practical purposes, eliminate the possibility of a mine.
The heat-transfer study is quite conservative. The article does mention the more severe cases of the fire being a hot spot directly against the bulkhead, and powder canisters stowed against or very close to the other side of it. The hot spot could in this case be considerably hotter, perhaps as hot as 2,600° Fahrenheit and the heat transfer could have been by direct conduction or radiation to the nearest canister. It appears that this would have been the most likely case for the start of the magazine explosion. However, it would not have changed the main conclusion that a coal fire would have been able to ignite the magazine.
The analysis of mines penetrating the ship bottom could have been confined to gunpowder mines alone. The hole produced by a contact or near-contact high- explosive mine would have been petalled, ragged, and unmistakable, even in a riveted structure. Such evidence was absent from the wreck completely. Even the 1911 board realized this, which is why it postulated a mine with a “low form” of explosive. The only worthwhile subject for analysis is a contact or near contact gunpowder mine.
The described analysis of mine explosions penetrating the ship bottom appears to be a scaling procedure using factors derived from existing experimental data. Nearly all such data were obtained from tests of high-explosive charges against welded structures. These data are not applicable to the case of gunpowder charges against a riveted structure. Gunpowder is a propellant, not a high explosive. It conflagrates rather than detonates—very slowly in comparison with the detonation of high explosives. The maximum explosion velocity in confined black powder is 1,400 feet per second, while the detonation velocity in TNT is about 21,000 feet per second. Gunpowder mines do not produce a shock wave in the water, but a wave of relatively low powder mines. The effects of gunpowder mines on a target cannot be derived directly through TNT equivalency factors, and the article does not explain how this was done. Similarly, the analysis method employs a strain energy factor for the bottom plating when it fails, but does not explain how this was derived for riveted plating.
The article mentions a hole or rupture in the bottom to pass energy through, but is silent on the size and shape of the rupture or deformation. This is one of the key issues in determining whether a mine was the cause of the explosion. The article appears to be of two minds concerning the bottom deformation or rupture inflicted by a mine. In the conclusion, it appears that the penetration analysis is ignored, since there it is postulated that a “non-shock wave” loading pushed that famous plate section inward, “unzipping” the rivet seams without deforming the plates. This could not have happened. Even though the rivet seams are weaker than the plate base material, they would not become “unzipped” unless they were loaded to failure, and they could have been loaded only through stresses in the plates. A closer examination of the bottom response is needed to prove that a gunpowder mine could have ignited the magazine without any significant visible effects on the outer bottom plating.
The use of the drop-test sensitivity data for black powder to describe the sensitivity of the powder in the 6-inch reserve magazine is entirely wrong, at least by an order of magnitude. The test consists of dropping a 2-kilogram hammer on a small amount of powder on an anvil. The brown gunpowder in the magazine was in canvas bags surrounded by wood shavings inside copper canisters stowed on wooden racks. The smaller amount of black powder, used for saluting guns and gun shells, was stored in roughly 200-pound copper tanks, which also were on wooden racks. There are no test data to indicate the sensitivity of gunpowder stowed in this manner. It is known from tests with high explosives, however, that the impact sensitivity for cased explosives is far less than that observed in a drop test. For example, for H-6 filled bombs the critical impact velocity is 17 times greater than the critical impact velocity obtained in a drop test of bare H-6 explosive. Gunpowder is ignited easily by heat and friction, but it is quite resistant to impact and stress. During the manufacture of black powder, the raw cakes are compressed in rollers under a pressure of 6,000 pounds per square inch. The article assumes that the magazine would have exploded in all cases where any penetration of the inner bottom occurred. Such an assumption cannot be made. The most likely mechanism for a gunpowder mine to have ignited the magazine would have been with “kinetic trauma.” This means that the powder canisters would have had to be mangled severely so that enough frictional heat was produced in the stowed powder to ignite it—before the water entered. Another potential mechanism would have been the injection of flame and heat through the double bottom. This would have required a mine in close contact with the ship bottom and with a charge burning long enough to ignite the magazine before the water entered. If a high-explosive mine had been under the ship, the mechanism most likely would have been by penetration into the magazine of ragged plate edges or hot fragments. The fact remains that the mechanism and its intensity by which a gunpowder mine could have ignited that particular magazine is unknown, and this is one of the major problems for those who want to prove that a mine was the initiator. The magazine would not have been ignited by a mine at all without a major impact on it, that is, by a mine of such size that the damage inflicted on the ship bottom would have been obvious over a substantial area. The entire presented analysis of the penetration of the ship bottom and ignition of the 6-inch reserve magazine is not applicable to gunpowder mines and gunpowder magazines—the only case worth considering.
Estimating the efficiency of the riveted plate joints on the basis of a finite element analysis alone is worthless unless it is keyed to experimental data. After all, riveted joints in steel structures were used for more than 100 years, and design rules for them have been available from at least the 1850s. The joint efficiency is entirely dependent on rivet size, rivet spacing, and the hole production method (cold punched only or punched and reamed). What actually was used on the ship is not clear from the photographs alone. If the rivets were of 7/8 inch diameter, the joint design rules, which are conservative, give efficiencies of 67% and 55% for 3d and 4d rivet spacing, respectively (d is the rivet diameter), which is quite different from the 45% assumed. It should also be noted that for the 4d spacing of these rivets, the joint failure is by shear on the rivets, not by plate yielding at the rivets.
Basing the internal explosion analysis on TNT equivalence of gunpowder is erroneous. The dynamics of gunpowder exploding in a closed volume is different from that of high explosives. The failure and venting times will be radically different. The detonation velocity in black powder, previously mentioned, is only about l/20th of that for high explosives. The brown powder in the magazines, which was used for the big guns, was prismatic powder composed of 1-inch by 1 1/8-inch molded cylinders. This form was used to make it bum slower than black powder, thus reducing the peak pressure in the gun barrels and still obtaining the same or greater muzzle velocity. In addition to the grain size, the explosion velocity in gunpowder will depend on temperature and pressure, and it can be less than the maximum. The explosion velocity of the stowed brown powder is not known. If a maximum velocity of 1,300 feet per second were assumed, then the explosion of the contents of the 6-inch reserve magazine would have taken a minimum of 9-18 milliseconds, depending on the initiation point. Most likely, this time would have been considerably longer, since most of the contents were brown powder stowed in individual canisters. Although the bounding bulkheads most likely had failed before all of the contents had exploded, the failure times indicated in the article are nevertheless way off. Also, with the bounding bulkheads failed before completion of the explosion, the venting would have affected the explosion event itself and reduced the developed pressures. The interaction between boundary failures and the explosion event is the reason why TNT equivalency factors cannot be used—they would not be constants. That boundaries failed before completion of the explosion is verified by the fact that unexploded powder canisters were found around the wreck. That they did not explode also bears witness to the fact that powder canisters were not ignited easily by impact—from a mine or anything else. The entire analysis of the internal explosion of the magazines is invalid, as is the estimated amount of powder that exploded. It is quite clear that most of the forward 6-inch and the 10-inch magazine contents did not explode.
At present, no theory or finite element method can simulate correctly the explosion of gunpowder mines, explosion of black and brown powder in magazines, the associated gas flow dynamics, and the resulting target response, especially not for riveted steel structures. The best approach to predict the effects is to rely on available test data. Of course, for a magazine explosion no data available can correlate the amount of powder exploding with the extent of damage, but this is of no significance for determining the source of the initiation. For prediction of the effect of a mine on the ship bottom some data applies from tests against a hull somewhat similar to the Maine hull, namely, the British ironclad Hercules, which was tested in both England and Scandinavia in the 1860s and 1870s. By extrapolating this data to the Maine, it can be shown that a powder mine big enough to breach the double bottom also would damage the outer bottom plating over an area that could not have been missed on the Maine.
How did the Section 1 plating end up bent into the ship? It was a result of the dynamic effects from the internal explosion. The magazine explosion, which lasted a relatively long time, created a huge high- pressure cloud—its weight was the same as the weight of the powder exploding. The cloud attempted to expand in all directions, including downward, destroying the inner bottom and pushing the outer bottom down into the water. This put the outer bottom into membrane tension, which is the failure mode clearly visible below the 6-inch reserve magazine. The four bottom plate sections mentioned in the 1911 report were neatly separated along rivet lines, as would be expected, except one of the strakes where the plate was ruptured across and through the plate material at a 45° angle. In addition, the plates of the four sections are amazingly smooth and could have been fitted together roughly again. In the course of the dynamic explosion event, the cloud eventually vented upward, rupturing the overhead decks in the process. It did not vent downward where it was retarded and stopped by the water. Thus, toward the end of the event the gas and water flow at the bottom was reversed, and it pushed the Section 1 plating with it. That the four bottom sections were not all pushed in is one of those vagaries of the dynamics involved. It was a result of the sequence of events. The explosion was not symmetrical or uniform, and the structures above the bottom did not all fail at the same time. Unfortunately, no theory currently available could simulate the sequence of the dynamic events involved, even if we had known the exact location of the start of the explosion. But just because we cannot model the sequence does not prove that it was a mine that pushed in the plate. A mine would have brought about its own sequence of events, including the infliction of damage to the plate that it was supposed to have pushed inward. That reversal of gas flows can cause unexpected effects has been seen in more recent explosion tests.
A feature visible in one of the photographs of the wreck, reproduced on page 37 of the article, verifies that the inward bending of the plate occurred late in the event. It shows that some of the extremely damaged inner bottom and framing resides within the fold of the smooth outer bottom. A mine could not have crumpled the interior structure while leaving the outer bottom smooth. If a mine had ignited the magazine by bending in the Section 1 plating, the following magazine explosion would have crumpled the outer bottom as badly as it did the inner bottom and framing. The only possible explanation is that the inner bottom and framing was crumpled first, while the outer bottom was stabilized by its water backing, allowing it to be bent smoothly inward at the later stage of the dynamic event when the gas and water flow at the bottom was reversed. The article blames the difference in appearance between outer and inner bottom plating on the difference in thickness. While it is true that the inner plate was 37.5% thinner than the outer plate, the difference is irrelevant, considering the loadings.
The article tries to make a point out of the depression or hole in the harbor bottom under the ship. A hole caused by a mine would have been more or less round. It was not. A bottom mine would have dug a round hole with a berm around it, and it also would have whipped the ship to the extent that it could have been broken in two without exploding the magazines. This did not occur, nor did the survivors report whipping motions of the stern section, which would have been proof of a mine. Whatever caused the hole, it was not a mine alone, and if there had been a mine, all distinct depressions it might have made on the harbor bottom were erased. Thus, the reported hole can provide no evidence either for or against the presence of a mine.
Mr. Allen’s article does not delve into some of the historical facts having a bearing on the likely source for the explosion. These have to do with the mistakes committed by the original official investigations, with questions concerning where the postulated mine could have been obtained, how it could have been planted so it could have produced the disaster, and who the perpetrators possibly could have been. A discussion of these problems would make this letter too long. One can only conclude that even when ignoring the complete lack of mine damage on the wreck, the likelihood of a mine being the possible cause of the Maine disaster is extremely remote. There was no mine under the Maine.