About a year ago there appeared in a magazine a short article on the self- L sealing of airplane fuel tanks which stated that such tanks had been developed in World War I and that their use in World War II was nothing new. That, undoubtedly, is entirely true and nobody ever intended to belittle the efforts, or the results of those efforts, of those pioneers who tried to change the old “flying coffins” into something a little more habitable for the aviators who were unfortunate enough to catch an enemy tracer or incendiary bullet in the gasoline tank. The self-sealing or, as it is commonly and erroneously called, the bullet-sealing gasoline tank, as we know it today is something as different from what the World War I model must have been as night is from day. Of course, the objective was the same in each case but that is about all that can be said for any similarity, or we would have heard more about the early models and, most certainly, we should have had more of a background at least to get a good start on the tanks of World War II. As it was, we had little or nothing. It is the purpose of this article to tell something of the surprises, troubles, and successes that were encountered in the development of the self-sealing gasoline tank of today; a development that is now as much a part of every American combat airplane as its engines or wings but has become so commonplace that it is taken for granted.
As far as the United States Navy is concerned the real development of the self-sealing gasoline tank started with the first gunfire tests conducted by personnel of the Naval Air Detail of the Naval Proving Ground at Dahlgren, Virginia, in the middle of 1940. Late in the year 1939 several tanks had been tested by Lieutenant Commander William S. Parsons, U. S. Navy, who at the time was Experimental Officer at the Naval Proving Ground. This was the same officer who later, as Captain in the Navy, is reputed to have designed the first atomic bomb dropped on Japan. He was also the Naval Observer in the Army 11-29 airplane which dropped this bomb on Hiroshimo. The Bureau of Aeronautics was of course integrating all of the design, manufacturing, and test work for the tank, since that was and still is the cognizant bureau, but as it had no adequate gunfire test facilities of its own at that time, that phase of the work was done by the Bureau of Ordnance for the Bureau of Aeronautics. Aviation personnel at Dahlgren conducted the gunfire tests. The war in Europe had already indicated that both the English and the Germans had some type of self-sealing tanks and armor in their combat airplanes and farsighted personnel in the Bureau of Aeronautics immediately saw the handwriting on the wall and initiated the development for our own service. They also reasoned correctly that since we were using .50-caliber machine guns, any potential enemy might, and probably would, do the same and they specified, therefore, practically from the start of the tests, that all of our self-sealing tanks be designed to withstand .50-caliber bullet impacts. Except for a few shots fired out of pure curiosity from a .30-caliber machine gun in order to compare it with the larger gun all tanks were tested with the .50-caliber gun. The wisdom of the Bureau of Aeronautics personnel in specifying this gun will be brought out later. There immediately comes to mind that the 20-mm. gun firing an explosive shell was already in use. So what were we going to do about that? The answer to that is that it was expected that an unfortunate 20-mm. hit was too much to protect against and that in such a case the tank would leak and the airplane might even be set on fire. That this was not always so will also be seen later.
Just how much research, calculating, burning of midnight oil, and garden variety of guessing on the part of the people in the U. S. rubber industry went into the construction of the first tanks no one will ever know. At that time none of the companies had their own gunfire test ranges, which they later set up to make preliminary tests and throw out ideas which obviously were no good, so that the first gunfire tests the self-sealing tank designers saw at Dahlgren were surprising and discouraging, to say the least. It was no less discouraging and surprising to the personnel at Dahlgren because it looked like a losing battle from the beginning.
To one who does not know about the devastating power of our .50-caliber machine gun (it is considered a cannon by European standards) the normal reaction is to think that a hole made in rubber by a .50-caliber bullet should not be difficult to seal. The hole is not actually that big because rubber tends to flow back into place after a hole is made in it. Besides, doesn’t rubber become “tacky” very quickly when exposed to gasoline? The answer to that question is that under normal conditions it does seal, and very quickly too. Strangely enough, we quickly realized that we did not have to worry too much about that; there was something far more difficult to control. This gremlin we called “water-hammer”; later, to be more precise and scientific, we called it hydraulic shock or ram. It is similar to the water-hammer that causes knocking in the water pipes of a house, but in our case, it had devastating effects. Prior to testing, the tanks were completely filled with gasoline. Pressure in a fluid normally acts the same amount in every direction, as any high- school boy knows after his first course in physics. In the case of our self-sealing tanks we discovered that although we had pressure exerted in all directions, sufficient in most cases to make a tank which was originally cubical in shape almost spherical at the end; of our test, the effect of gunfire was similar to what might be expected had a large metal rod been forced through the fuel at high velocity and against the rear side of the tank. The British made extensive tests to determine the value of these pressures and actually measured them by means of a modified crusher-gauge. We knew the pressure values were high but never actually measured them. We were more interested in developing materials that would hold up regardless of the pressures involved than we were in the pressures themselves. In our first tanks the bullet entrance hole would be the size of the bullet but the entire rear of the tank would be knocked out. Furthermore, the bullet, in passing through the liquid, would almost invariably tumble, and if the rear of the tank were not knocked out, an elongated hole of 2 1/2" to 3" in length would result from the tumbled bullet.
This was something nobody, apparently, had even remotely expected. Later, when we had acquired more experience with the effect of hydraulic ram, experiments were made to determine what would happen in tanks only partially filled with gasoline. It was found that there was no appreciable difference in the results obtained in completely filled tanks and partially filled tanks. Apparently the hydraulic ram occurred so quickly and carried through the liquid so rapidly that it reached the rear side of the tank before any gasoline could be forced up into the empty space above it and thereby relieve the pressure. Almost without exception every first tank submitted for gunfire test by the various manufacturers was made of slabs of self-sealing material cemented together at the corners and, also without exception, the tanks split open along the cemented corners at the first shot. Thus, we learned our first lesson. The remedy for the splitting corners was to avoid the use of slabs and instead build the tank as a unit carrying alternate layers of material that made up the tank around to the middle of adjacent sides and rounding the corners in the process in order to avoid concentration of stresses. This process immediately complicated the construction of the tank and increased its cost since it had to be built over a form which could be removed later, but it had to be done. Cemented slabs were easy to make and could be made better than the tanks built up over a form but they could not be made into a tank that would hold gasoline after being struck by a .50-caliber bullet.
It might be pertinent at this point to bring out the fact that the companies involved in the development of the self-sealing fuel tank at the start of the program were the big four: Firestone Tire and Rubber Company, B. F. Goodrich Company, Goodyear 'Fire and Rubber Company, and U. S. Rubber Company. The efforts of these four companies were soon supplemented by that of the Hewitt Rubber Company. These five companies were later to become the only acceptable sources of supply for self-sealing tanks. Another company, the Aircraft Protective Products Corporation, contributed its share to the extensive experimental program and to the general knowledge accumulated in this field. In time several chemical manufacturers such as Monsanto Chemical Company, Du Pont, and Dow Chemical Company were called upon to furnish special plastics or metals. At least one of the “Big Four” turned over the resources of its entire research department to the self-sealing tank program. Judging by the rate at which improved models of materials and tanks appeared for test at Dahlgren it became quite obvious that a veritable army had been turned loose on the job by these four companies. At times it seemed that the report on the test of one set of tanks had barely been finished before the next set with all the improvements indicated by the test were on hand and waiting to be shot to pieces. All this work they were doing for the Navy was in addition to any they might have been doing for the Army Air Forces. When those companies later established their own test ranges that type of speed meant that they had not only made improvements and built them into new tanks but that they had probably made a few others on the side and had tested them themselves. All of the rubber companies were working independently of each other. Although such a procedure might have caused a certain duplication of effort, in the long run more ground was probably covered and there was certainly no cause for biased opinions. We at Dahlgren attempted to make an impartial evaluation of each tank. From time to time we made recommendations to one company based on the tests of another company, such as when we found that cemented slab tanks were no good, and in that respect we were not impartial. The Navy needed self-sealing tanks in a hurry and our aim was to help get them developed as fast as it was humanly possible to do so.
It became evident early in the game that we weren’t going to be able to learn much about the self-sealing characteristics of the various materials if the supporting aluminum tanks fell apart too early in the gunfire test. It was also only too apparent that we hardly knew where to start in our efforts to build a sufficiently strong tank. The tanks made of self-sealing material were inserted into aluminum shells similar to the way it was expected to install them in airplanes. Since aluminum was becoming critical at that time due to the expansion of our aircraft industry and the early lack of aluminum plants, the tank manufacturers at first used the lightest gauge sheet aluminum they could find, which would still keep the self-sealing structure in its proper shape. Most of these early aluminum containers were in the neighborhood of 1/64" thick and they too were designed and assembled without regard for the hydraulic ram which was so totally unexpected. As the thickness of the aluminum going into the supporting structures was little by little increased as dictated by the results of our tests it also became evident that the type of aluminum as well as its thickness was going to affect the results of our tests. The 24ST aluminum used at first was relatively hard and brittle so that large pieces frequently were torn loose on bullet impact and tore out large pieces of self-sealing material as they were carried through it, thereby preventing sealing of the hole. This type of aluminum also tended to crack badly at the bullet exit side of the tanks and large pieces of it would be carried away so that the self-scaling material no longer had any support.
In addition to the troubles we had trying to find the right type and thickness of aluminum alloy to use, we were also bedeviled by hydraulic ram in the construction of these metal containers. The first ones tested, like the self-sealing liners, had square corners— in perfect condition for setting up large stress concentrations, as any engineer knows—and these corners were made very simply by bending the edge of the sheet forming one side and riveting it with a single row of small size, widely spaced rivets. When it was finally decided to use 3S 1/2 hard aluminum 1/8” thick, round the corners slightly, lap the metal on both sides of each corner, and use several staggered rows of larger size rivets or weld the seams, after many tests in which much time, money, effort, and material were expended most of our troubles which were attributed to the aluminum outer containers disappeared, or at least reached a point where we could account for them and where they no longer obscured the other results of the tests. This particular type of aluminum was softer and less brittle than the 24ST and, although it had a tendency to bend and stretch under bullet impact, it seldom tore much trouble holding our tanks together long enough to determine the sealing characteristics of our materials, one company submitted for test a sealing material in an aluminum shell which was reinforced on all six sides by large aluminum angles spaced about 4 or 5 inches apart in an effort to defeat the effects of hydraulic ram. As long as the bullets struck between the reinforcing angles, the sealing material performed reasonably well, but when the first one struck a reinforcing angle piece both it and the bullet broke up into innumerable small pieces and had the effect of a shotgun fired at short range. A large hole was blown in the self-sealing material and the test ended. But if we hadn’t learned what to do at least we learned what not to do. We received no more tanks reinforced like that one and we were always careful in succeeding tests to keep our bullets well clear of heavy metal parts. We were slowly but surely building up our knowledge of the self-sealing tank business—the hard way.
Some of the first self-sealing materials were made up of layers of material which were not cemented to each other. In testing these it was extremely difficult for the test personnel to determine the extent of any leakage after bullet impacts because any leakage tended to occur between the layers first and that from the hole in the outer layer occurred last if the inner layers did not act as a series of containers which held the remaining gasoline, which they usually did. Remember that it was sometimes difficult, if not impossible, to determine what was going on inside the aluminum shell. We soon discovered, however, that our old bogey, hydraulic ram, was entering the picture here, too. As the spaces between the various layers of sealing material filled with gasoline, usually after the first bullet hole was made, they assumed all the characteristics of a series of cells or small tanks and when succeeding bullet impacts occurred, hydraulic ram carried its destructive force through the individual layers of these materials and tore them badly.
One of the many variables that plagued the rubber people was that of thickness of the self-sealing material. Obviously, the thicker the material the easier it would be for it to absorb the force of hydraulic ram and, once a bullet hole was made in it, the longer would be the path for the gasoline to get through it to the outside, more rubber would be acted on by the gasoline as it traveled through the hole, and the better would be the sealing characteristics. However, thicker materials meant greater weight, always at a premium in a combat airplane, and what was just as bad, reduction in gasoline capacity, because our sealing materials are used inside the metal container and not outside as the early British materials were. The Bureau of Aeronautics, in its specifications for self-sealing materials, established upper limits for weight and thickness beyond which it was considered inadvisable to go because of weight and fuel capacity limitations in the aircraft in which it was expected to use these materials. In an effort to reduce weight some self-sealing constructions used a middle layer of sponge rubber. It was probably expected that, in addition to reducing weight, this type of rubber would affect quicker sealing as a result of having a greater surface area exposed to gasoline after bullet impact than would a solid piece of rubber. Both of these expectations were reasonable but in practice they failed to work out for the simple reason that the sponge rubber had relatively little strength and large quantities of it were torn loose and carried by the bullets into the interior of the tanks. If used in an aircraft tank, sealing of bullet holes might possibly be effected at the expense of clogging fuel strainers or carburetor jets with dissolved or partially dissolved rubber.
Early attempts of one of our Allies at self-sealing consisted of covering the regular airplane tank with a layer of rubber which in turn was held in place by a layer of doped or painted canvas. If they found these at all satisfactory it was because they had only .30-caliber guns used against them by the Germans. More than likely the protection was for the purpose of improving morale until a more satisfactory type could be developed. Two such tanks were tested at Dahlgren in the summer of 1940 and, poor as our own tanks were at that time, we considered our materials and methods of using them superior. Their tanks not only didn’t do well against .30-caliber gunfire; they were hopeless against the .50-calibcr gun.
No attempt will be made in this article to go into the chemistry of the component parts of the various self-sealing materials since that would require volumes in itself. Each rubber company bad its own ideas as to how the material should be made up, what should go into it and how much, consistent with having it meet the specifications laid down by the Bureau of Aeronautics. In at least one case the manner of making up the self-sealing material was governed by the type of machinery that was available. In general, however, all the self-sealing materials tested consisted of an inner layer which was impervious to and insoluble in gasoline, one or more middle layers of rubber, rubber compound, or synthetic which had a high rate of volume swell in the presence of gasoline— this was the part that did the actual sealing by its swelling action when exposed to gasoline—and an outer layer which was made up of three or more plies of tire cord fabric, impregnated on the outside with a gasoline resistant finish. Some engineers used one or more plies of fabric impregnated in the sealing material while others used none. The first self-sealing materials which were devised had the various component layers cemented together with rubber cement. When these materials were punctured by a bullet, gasoline came in contact with the cement, dissolved it, and separated the various layers making up the self-sealing material. Obviously the construction of the self-sealing material by cementing was a cheap and easy method of constructing it. It was quickly realized that it was necessary to treat the self-sealing materials so as to make them and keep them a solid mass even after damage by bullet or fragment penetration. Accordingly, the rubber industry developed a vulcanized self-sealing material which was built up by hand, layer by layer, and then vulcanized in much the same manner as a rubber tire. The inclusion of tire cord fabric in the self-sealing material gave it added strength and undoubtedly did more toward combating damage by hydraulic ram in the self-sealing material itself than any other single factor. It added stiffness to the self-sealing material which helped to prevent its sagging and to prevent holding open the bullet wound after the aluminum supporting structure was partially knocked away. After our source of Malayan rubber was lost early in 1942, the rubber companies were forced to reduce the rubber content of the self-sealing materials and use more and more reclaimed rubber. Very likely most of our old tires, overshoes, and articles of similar nature which were collected several years ago went into the making of these tanks. The first tanks which used reclaimed rubber did not perform nearly so well as those which used new rubber, but since new rubber was becoming extremely critical the rubber companies were forced to use it and, as usual, they came through with new ideas and new techniques and quickly produced self-sealing materials of reclaimed rubber which were fully as good as those made of new rubber. By blending the reclaimed rubber with various synthetics they soon reached the point where they could make self-sealing tanks perform equally well with our gasoline or the aromatic fuels of the East Indies and under conditions of both heat and cold.
At the beginning of the development of self-sealing fuel tanks all materials were tested at outside air temperatures and with 100 per cent octane aviation gasoline. No sooner had we gotten the material which looked promising under outside air temperatures than it was realized that our airplanes might have to operate out of such cold places as Alaska and that self-sealing materials probably would have difficulty in scaling under conditions of extreme cold since rubber gets harder and more brittle as it becomes colder. Accordingly, the Bureau of Aeronautics directed that these materials be tested at a temperature of —40° Fahrenheit, which temperatures might be expected in certain parts of Alaska, Siberia, and the high altitudes in which the modern airplanes are required to fly. In these days of elaborate equipment a temperature of —40° Fahrenheit can be reached quickly and easily. In 1940, while the Navy and the Proving Ground along with it were suffering the growing pains incident to the expected World War, adequate low temperature equipment was not available and it was necessary to have shipped from Washington, D. C., the nearest available source, large quantities of dry ice to be used as a cooling medium. To cool adequately the self-sealing materials prior to testing it was necessary to build a huge refrigerator. It may still seem a simple thing to reach that required temperature but, while we had our dry ice and refrigerator we had no way of adequately reaching the desired temperature and holding it. Because of the nature of the work at the Proving Ground, where many of the personnel must take cover many times during the day to avoid injury from shell and plate fragments, we were forced to perform our tests in an area safe from the fragments and this necessitated carrying our cold tanks by trucks a matter of several miles to the test area. In many cases, while on the way to our own test area, we were forced to wait some minutes before a large plate firing test was made. Naturally the self-sealing materials could scarcely be expected to have the same temperature during the test that they had when they left the refrigerator.
In time we overcame most of these difficulties. The methods, at first, were crude and anything but precise. We accepted such conditions at the time simply because we had to, but while we accepted them we were improving our methods and our equipment and somehow we got the answers which were needed to enable the rubber companies to go into production on tanks for the planes already in existence. There was one very amusing incident in connection with the early low temperature tests. We realized that carrying a tank from the refrigerator several miles to the point of test was most unsatisfactory and in an effort to improve conditions we decided to refrigerate the gasoline at the test point by dropping dry ice into our tank truck. The idea was fine. The gasoline went down to —85° Fahrenheit in a matter of a few minutes. We then proceeded to pump this gasoline into our test tank. A trickle came from the hose, then nothing—we had forgotten to drain the water from the water-and-sediment collector and what water happened to be in it froze solid and stopped up the entire system as soon as the cold gasoline hit it. We were then faced with the problem of removing this terrifically cold gasoline from the tank truck in order to perform our scheduled test. If anyone thinks it easy to handle liquids at that temperature by ordinary methods all you have to do is try it sometime.
One of the chief difficulties encountered in cooling gasoline by dropping dry ice into it lay in the fact that as the dry ice extracted the heat from the gasoline, gaseous carbon dioxide was liberated and in the process caused heavy boiling of the gasoline. While we could reach the desired temperatures very quickly by using adequate amounts of dry ice, we were limited to the amount which would not cause all of our gasoline to bubble away and be lost. We also had to determine how fast we were cooling our self-scaling material since, being a relatively poor conductor of heat, it cooled much more slowly than the gasoline. This same effervescing of the carbon dioxide from the cold gasoline caused us no end of trouble in keeping filling connection caps on the tanks during gun-firing tests. The heat generated by impact of the bullets on the aluminum containers was invariably sufficient to warm up the gasoline to the point where large quantities of gaseous carbon dioxide were set free inside the tank. The resulting pressure in turn invariably blew off the filling connection cap and permitted gas to spout from the tank with a geyser-like effect. Wherever it was possible to do so we cooled our self-sealing fuel tanks and the gasoline in them by keeping them in the refrigerator for several days preceding the test. This was possible in the case of the two-foot cubical preliminary test tank but it was not possible in tests which involved the complete airplane wing in which the self-sealing fuel tank was located. Not only was our refrigerator not big enough in the latter case but the very weight usually precluded any extensive handling by mere man power. In the latter case we were forced by the existing conditions to put dry ice into the gasoline through the relatively small filling connections and accept the difficulties which went with it. One of the biggest headaches we had when we first attempted to cool the gasoline in sample tanks by dropping dry ice into them was due to the fact that the tanks were constructed with various sizes of filling connections. At a time when materials were becoming more and more difficult to obtain it was natural for the engineers to use anything which would serve the purpose. Hence, they apparently used anything which was readily available, cheap, and easy to fabricate. It we received a tank with a filling connection of a normal size or larger we considered ourselves very lucky. If this connection was closed by a screwed on cap we considered ourselves even more fortunate. In most cases early in the game the tanks were closed by the simple expedient of using a wooden plug fitted into the filling connection. Naturally, these wooden plugs invariably were forced out with each shot. While this particular trouble may seem slight to the reader it was one of many little ones which caused no end of delay and headaches to the test officer, who was forced to work fairly close to a well-filled schedule and with manufacturer’s representatives who had come, in many cases many miles, to witness the testing of their products. Since these early days of testing more adequate equipment has been provided and now precise temperatures can be reached by means of properly controlled refrigeration systems, and tests can now be conducted under what might almost be called laboratory conditions.
One day early in the development of the self-sealing fuel tanks, while testing a number of 2-foot cubical self-sealing tanks, one of these tanks suddenly burst into flames when struck by a .50-caliber armor-piercing bullet. Here was a new problem which we had never encountered before among our many other worries and difficulties. We immediately began to watch the test tanks very carefully in an effort to discover what had caused this phenomenon. Remember that we were shooting copper jacketed projectiles against aluminum sheet metal and that is a condition which should have created no sparks. Yet there was no reason for gasoline to catch fire spontaneously under the conditions of our tests. Our close scrutiny finally revealed that occasionally we were getting large sparks or what might even be called small flashes of flame up to as much as 5 inches in diameter upon bullet impact. Sometimes these were isolated flashes. Sometimes they occurred with almost every bullet impact on the same tank. Some days they occurred in one tank and not in others. Sometimes they occurred in rainy weather and sometimes in dry weather. They occurred under all conditions of temperature, down to and including —40° Fahrenheit. They occurred with both hard and soft aluminum. Sometimes we got no sparks for weeks on end. At other times it seemed as though every tank we tested caught fire.
Here in itself was a problem of no little concern since it involved the safety of every combat airplane. Since any self-sealing tank will usually leak slightly while effecting a seal, the danger of fire from every bullet impact on a punctured tank was not to be considered lightly. Attempts were made to photograph these flash phenomena on high speed motion picture cameras, both during daylight and darkness, in an effort to determine exactly what was causing it and from that determine what could be done toward eliminating it. Various methods of eliminating the flashes were proposed but all involved difficulty of manufacturing and increasing cost of the aluminum sheets used. The exact nature of this fire producing flash was probably discovered one day when the Test Officer picked up a bullet which had spent itself just behind the test tank after passing through the gasoline therein which was at — 40° Fahrenheit. This projectile was so hot that it raised blisters on the skin even though it was held only a moment. In short, the flash was undoubtedly produced from the aluminum sheet which, in the vicinity of the bullet hole, had been pulverized and raised to such a temperature by friction and the work done on it by the bullet that it became molten metal.
The Bureau of Aeronautics and the rubber industry were fully cognizant of the fire hazard and everything possible was done to reduce it. Since it was the sparking caused by bullet impact on aluminum which created fires, it could be expected that bullet impact on nonmetallic materials would produce no sparks. This was found to be the case, and the Bureau of Aeronautics entered upon a long program of tests to determine the best type of nonmetallic materials to be used. As a result practically all of our combat aircraft have, for some time, been using fuel tanks containing no metallic material. By that it is meant that no metal has been used directly in contact with the self-sealing material for the purpose of supporting it and maintaining the shape of the tank. It was found that by designing the aircraft with a large air space between the structure of the aircraft and the fuel tank, sparking could occur in the structure of the aircraft and still not present a fire hazard because any leakage which might occur was some distance away. A typical example of such construction is in the F6F (Grumman Hellcat) in which the fuselage fuel tank is suspended in a hammock of nonmetallic material. The ultimate success of our method of protection against fire is adequately demonstrated by the fact that relatively few of our airplanes have been seen to go down in flames. The advent of fires in our self-sealing tanks under test necessitated the setting up of facilities for fighting fires. We found that by using a large asbestos blanket over a fire of even large proportions soon after the fire was initiated and by using carbon dioxide fire extinguishers under the asbestos blanket, we were able to save most of the tanks and even complete most of the tests. We were always bedeviled by low temperatures or high temperatures and sometimes we had both to worry about at the same time.
In an effort to conserve valuable rubber and sheet aluminum the Bureau of Aeronautics in 1940 designed a square steel test cell approximately 2 feet in each dimension. This cell, which was made of |-inch thick sheet steel with two opposite ends open, had one-inch angle irons around the open ends. In setting the cell up for test a slab of self-sealing material with a sheet of aluminum was placed over each end and a square frame of one-inch angle steel was then placed over the sheet aluminum and self-sealing material and held in place by “C” clamps spaced about 2 inches apart. This design was conceived after the first test on aluminum covered test tanks had indicated unexpected difficulties from hydraulic ram caused by .50-caliber bullet impacts. Since no actual values had been determined for the pressures being built up inside the tanks by hydraulic ram it was expected that a steel cell would certainly be sufficiently strong to withstand these pressures. Here again, in our general ignorance of the subject, strength values were “picked out of the hat” with the expectation that the steel cell would be sufficiently strong. The first .50-caliber bullet impact, however, promptly knocked off all of the “C” clamps, twisted half of them out of shape, and left the steel test cell more spherical in shape than cubical. We immediately went to work and designed a steel test cell which we felt would hold up under the greatest hydraulic ram that we could visualize. This again consisted of a cubical cell with two ends open, but this one was made with half-inch special treatment steel, known in the Navy as STS plate, with large flanges of half-inch plate. A sample of self-sealing material and aluminum sheet was placed over the flanges and over this was placed a large rectangular ring of half-inch steel. The rectangular retaining rings, sheet aluminum, and self-sealing material were all bolted to the steel flange of the steel test cell by half-inch diameter steel bolts spaced about 2 inches apart. The corners of the plates making up the steel test cell were welded together both inside and outside. This cell held up during all subsequent tests, but hydraulic ram encountered over a long period of time bulged slightly even the half-inch steel plates.
The description of this test cell may not be particularly interesting to the layman but it is included for the purpose of indicating what was happening to the Japanese airplane struck by our .50-caliber machine gunfire. In reports of air battles with the Japanese one continually encountered descriptions of their planes literally blowing up in flight, of the wings falling off, and of their becoming massive balls of fire. If one remembers the vaunted, wonderful maneuverability of the Japanese planes and realizes that it was gained by making their airplanes of relatively light construction as compared with ours and that they were without armor protection or self-sealing fuel tanks, he will realize that most of the damage to the Japanese was caused by hydraulic ram. Most modern airplanes have their fuel tanks in the wing roots, that is, in the part of the wing next to the fuselage. In many cases the upper and lower surfaces of the wing itself form a Part of the fuel tank. When the lightly constructed Japanese planes were struck in this area by our .50-caliber gunfire, hydraulic ram undoubtedly literally knocked the wings off the airplane. If tracer or incendiary bullets immediately followed, as they usually and struck in the released mass of gasoline, which in the meantime had mixed with air to form an inflammable vapor, the result was one flaming Japanese airplane. Our airplanes on the other hand were built to take A as well as hand it out. With armor which would effectively stop .50-caliber bullets and even 20-mm. projectiles, and self-sealing tanks which gave effective protection against .50-caliber bullets, the .30-caliber guns used almost exclusively by the Japanese early in the war would damage our planes but would not always bring them down or prevent them from getting back to their carrier or base. We have innumerable reports of our planes which, on returning to their base, were found to be so badly shot up that they were not considered worth repairing and so were immediately dumped overboard in order to get them out of the way and permit continued strikes against the Japanese.
When a new self-sealing material was developed slabs of it were first tested on the steel test jig mentioned above. If it showed promise it was then built into a two-foot cubical test tank, the outer shell of which was sheet aluminum. In arriving at the two- foot cubical tank the Bureau of Aeronautics personnel again picked a figure “out of the hat.” In this case, however, they inadvertently chose one which subjected our self-sealing materials to what were probably the most severe conditions that could be imagined. These particular dimensions were just sufficient to cause nearly every bullet which entered the tank in normal flight to tumble badly after entering and emerge from the tank while tumbling or “key-holing.” The result of this was that almost all of the energy of the bullet was expended in setting up hydraulic ram. In order to make the tests even more severe and to simulate conditions likely to be encountered in combat where a tumbling bullet would be likely to strike the fuel tank as a result of having passed through some heavy structure of the airplane first, we caused our bullets to tumble by the simple expedient of placing a soft pine box at an angle to the line of fire about 5 feet in front of the test tank. In this way about 90 per cent of our impacts tumbled or “key- holed” both on entry and exit. Every tank was tested under conditions as nearly identical as possible in order to arrive at conclusions for determining the relative self-sealing values of the various materials. Naturally, bullets did not always strike as desired, low temperature conditions were not always precisely controlled, sometimes the tanks caught fire, or any one of several dozen other things might happen. As experience was built up, however, test personnel were able to assess these many variables so that eventually the right answer was arrived at. If it was not precisely the right answer at least we had one which gave adequate results.
In 1940 and early 1941 self-scaling materials had been developed which were satisfactory with 100 per cent octane gasoline and temperatures as low as a minus 40° Fahrenheit. In 1941, when conditions were going from bad to worse in the Far East and more and more aircraft were operating from the Philippines area, we discovered that much of our work on self-sealing materials would have to be done over because of the fact that gasoline procured from the Netherlands East Indies area was an aromatic fuel and that the self-sealing materials already developed were not adequately resistant to them. The end of the year 1941 found our aeronautic personnel in a dilemma of serious proportions with regard to the aromatic fuel situation. The various manufacturers were called together late in November, 1941, at the Bureau of Aeronautics, were informed of the problems and told to go to work on it. Another meeting was held late in December at Wright Field where progress was reviewed and the seriousness of the problem was reemphasized. Early in January, the Naval Research Laboratory found the answer to the first phase of the problem, that of making existing tanks suitable for aromatic fuel. They recommended that existing tanks be sloshed with a fluid which left a thin coating of aromatic fuel resistant material on the inside of the tanks. The rubber industry soon came through with an answer to the second phase of the problem, that of providing new tanks capable of withstanding aromatic fuels without the use of any sloshing compound. One product called “Royalin” was developed which was suitable except for the fact that it cracked badly at low temperatures. For some time it looked as though we might have to use this material and accept its poor low temperature characteristics. New self-sealing materials were immediately made up, however, and, after a period of time during which we had to go through the whole series of preliminary testing of many new materials, self-sealing materials were found by February of 1942 which were good for gasoline found in any part of the world. It was for his outstanding work in the development of an aromatic gasoline resisting self-sealing material at a time when it was so urgently needed that Mr. James Merrill, a young chemical engineer with the Goodyear Tire and Rubber Company, was presented the War Production Board Meritorious Civilian Award by President Roosevelt.
Some of our most valuable information on self-sealing tanks came from the Germans prior to December 7, 1941. The British shot down so many German airplanes over England that it is reputed that they were in a position to supply the Turkish Government with spare parts for the airplanes which the latter had procured earlier from the Germans. A number of these shot-down German airplanes were sent to this country for various tests. At Dahlgren we had a Messerschmitt 110 which was to be subjected to various gunfire tests including those on the self-sealing fuel tanks with which the plane was equipped. It required only a casual glance to realize that the Germans must have done a very extensive amount of investigation before the war on the subject of self-sealing of aircraft fuel tanks. These tanks did not give adequate protection against our .50-caliber machine guns but they did protect against the .30-calibcr guns. The German tank, as built on the ME110 was a urge shallow affair which was given shape oy an inner piece of very hard fiber. Over the fiber was placed a formed piece of synthetic impervious to gasoline. Over this latter material was placed what appeared to be pure gum rubber about 3/16 inch thick. Then came a very thin layer of what appeared to be latex, which in turn was covered by a layer of chrome tanned hide. The outside cover was the same as the inner layer, namely about 1/8 inch of very tough synthetic. The lank itself was suspended from a rectangular ring of hollow hard fiber about 3 inches deep which passed completely around the tank. The tank was held in place within the fiber ring by web straps which passed over and under the tank. The fiber ring in turn was secured in the airplane by six suspension bolts which could be reached from the upper surface of the wing by removing small cover plates. The tank was placed in the wing from below through one large opening the cover of which in turn was held in place by a series of machine screws. The tank contained no fittings on the bottom or sides. A large manifold fitted into the top of the tank and contained connections for filling, for the fuel Sage, for overflow lines, and for the suction line to the fuel pump.
In the use of this manifold we discovered that the Germans knew the answer to the Difficulty we had encountered in testing full-scale airplane tanks of our own. In these tests we found that when bullets struck metal fittings in the bottom of the tanks the fittings Were invariably broken up and, in so doing, left a large hole which in every case was impossible to seal. By taking suction through the top of the tank the Germans had eliminated completely the type of leaks that we were encountering in the vicinity of our bottom and side fittings. With their type of installation it must have been possible to remove and replace a damaged tank in a matter of several hours. What they probably took years of non-hurried work to achieve we had to learn in a matter of months. Their airplanes were obviously engineered from the early design stage for self-sealing tanks whereas our airplanes were already in existence and had to be adapted to take self-sealing fuel tanks.
As new planes were designed the self-sealing fuel tanks were given more consideration but one has yet to be built by us which is as effective as the German system from a standpoint of ease of removal and replacement or in the complete absence of bottom fittings. The German type of tank suspension permitted a space between the fuel tank itself and the structure of the airplane, I his is important in taking care of hydraulic ram because the tank changes shape on bullet impact and damages the structure of the airplane where it comes in contact with it. Distortion of the tank as a result of hydraulic ram is impossible to avoid. By separating the tank from the structure of the airplane by several inches and using a tank of plastic material the tank is permitted to change shape without damaging the surrounding structure. This separation is valuable in another respect in that if the structure is struck by a 20-mm. projectile with an instantaneous fuze the self-sealing material in most cases has nothing more to do than seal the holes made by the 20-mm. fragments, whereas if the self-sealing material is in contact with the structure it, along with the structure of the airplane, will have a hole blown in it too large to seal. We discovered that our self-sealing materials would seal against 20-mm. projectiles even though the projectile detonated in the gasoline inside the tank. Pure gasoline will not explode and we found that such a detonation in gasoline was completely absorbed by it without serious damage to the tank itself. The hydraulic ram set up by the .50-caliber bullet impact did far more damage.
As the performance of our aircraft increased and started to reach higher altitudes, the ordinary fuel pump no longer was sufficient to carry fuel to the engine. It was decided, therefore, to apply pressure to the inside of the tanks to assist in forcing fuel to the carburetor. This, naturally, imposed another burden on the self-sealing material and on the people who made it. It also imposed a considerable amount of work on the test personnel who were required to set up conditions simulating those in aircraft in combat conditions. It is one thing to equip and operate an airplane with a pressurized tank but the actual testing of such a tank is a different breed of animal. Obviously any damage to the tank and self-sealing material by bullets is accentuated by the internal pressure. With the restrictions on the weight and thickness of self-sealing materials it was next to impossible to obtain self-sealing in a pressurized tank. However, we had managed to seal some holes after the pressure was relieved and in some cases the sealing held up when pressure was reapplied.
Just as a chain is no stronger than its weakest link, the fuel system in an airplane is no better protected than the part which will not seal when struck by bullets or fragments. Thus the fuel lines in the airplane represented a weak link and had to be strengthened. This necessarily led to the development of self-sealing hoses. Probably just as much effort, thought, and research went on in the development of the self-sealing hoses as went on the self-sealing tanks and many hundreds of feet of such hose were tested at the Naval Proving Ground. It was necessary to design and build fixtures which would accommodate hose from ^ inch inside diameter to inch inside diameter and to arrange for the pumping of gasoline through them at the proper pressures. Eventually the hoses, like the tanks, were tested at reduced temperatures. Under these latter conditions difficulties were encountered. Despite all that could be done moisture would be picked up by the system and, on encountering the cold gasoline, it would freeze and cause stoppages in the various lines. But here again, despite make-shift rigs and deficiency in equipment the proper answers were obtained and, for several years, combat aircraft have had adequately protected fuel lines. Naturally they have been unable to protect against large fragments or tumbling bullets which cut the hose in two but they have given protection against the small fragments which would ordinarily cause a metal tube to leak and quickly drain off the available source of gasoline.
An airplane engine will not run any better without lubricating oil than it will without gasoline to give it power. Therefore, it was necessary to protect the airplane’s oil tank and oil lines against leakage caused by bullet or fragment penetration just as it was necessary to protect the fuel tanks and fuel lines. In contrast to the fuel tank tests which were conducted at low temperatures, those for the self-sealing oil tanks and oil hoses had to be conducted, in the case of the oil tank, at a temperature of 215° Fahrenheit and, in the case of the oil hoses, at a temperature of 300° Fahrenheit. The tests of self-sealing materials for oil tanks and oil hoses were much the same as those for fuel tanks and fuel hoses and the same difficulties were encountered with hydraulic ram. Tests with gasoline tanks usually produced clean wounds in the self-sealing materials which were easy to classify but when it came to handling lubricating oil which was above the temperature of boiling water it was more difficult for the test personnel to find out exactly what was going on. Furthermore, the oil leakage which did occur usually made a slippery mess which became a hazard to life and limb. No really satisfactory self-sealing oil hose was developed because of the high pressures involved (80 pounds per square inch as compared to 4 or 5 pounds per square inch for self-sealing fuel hoses) and the very low rate of swelling of the self-sealing material in the presence of oil. The self-sealing oil tanks, on the other hand, being merely a reservoir for the oil and not subject to internal pressures, were a different matter and adequate self-sealing was finally achieved with these materials.
After the loss of the U.S.S. Lexington in the Battle of the Coral Sea in 1942 as a result of uncontrolled fires it looked for a while as though the testing of self-sealing hoses would really become big business. Gasoline lines for fueling the airplanes aboard that ship apparently became punctured by fragments from bombs and it was thought that the installation of self-sealing hoses in place of metallic pipe for gasoline in an aircraft carrier would reduce the probability of such calamities in other ships. Accordingly the Naval Proving Ground undertook a series of tests on self-sealing hoses of about 4 inches internal diameter. Here again the rubber companies were set to work and produced material which was reasonably effective, but before any very extensive work could be done the Bureau of Ships, which is responsible for such piping aboard ship, apparently evolved other plans and the project was discontinued.
The self-sealing fuel tank science for aircraft had a natural corollary in providing self-sealing tanks for gasoline powered PT boats and by the middle of 1943 the Naval Proving Ground began testing large sections of PT boat hulls with self-sealing fuel tanks installed. As in the case of aircraft these tanks have brought back many PT boats which had been shot up in battle, but here again, they are taken so much for granted that they are never mentioned in routine reports or in the newspapers.
Personnel of the rubber industry of the United States worked long and hard for the purpose of saving American lives and equipment and for the ultimate victory which superior personnel and equipment were bound to bring us. Too much credit cannot be given to them, from the engineers and chemists who designed the materials to the workmen who fabricated them into self-sealing hoses and tanks. If this article can bring to public attention some idea of the magnificent achievement of these people it is felt that it will in a small way have achieved its purpose.
Graduating from the Naval Academy in 1927, Captain Eckelmeyer won his wings in 1929, and later specialized in Aviation Ordnance Engineering. After duty as Aviation Ordnance Officer at the Naval Proving Ground at Dahlgren, Virginia, where he conducted the tests and experiments on the self-sealing gas tanks, he served as commanding officer of the Eighth Amphibious Force flagship in the Mediterranean. At present he is commanding officer of a CVE.