18G Ultimate Test
Jack E. Barth (11/17/19 - 10/10/10) was a Douglas Aircraft Structural Engineer during the 1940s, 1950s, and 1960s.
Edited June 15, 1992 by David Barth - David's comments are in square brackets.
By Jack Barth
February 15, 1992
The D558-1 was one of the research aircraft produced after the World War II. It was designed to gather aero data up to and past Mach 1.
[Mach 1 is the speed of sound, 343 meters per second, 1,125 feet per second, 1,236 kilometres per hour,
or 768 miles per hour at sea level. At Mach 1 at sea level, a vehicle would travel approximately one kilometer in three seconds and about one mile in five seconds. Note that there is an inverse relationship between the speed of sound and altitude. The speed of sound decreases as altitude increases.]
[The story begins when Jack was hired at Douglas Aircraft after the end of WWII in the late 1940s.]
The aeronautical engineers had not fully digested the German data and didn't appreciate the advantages of swept wings. The D558 was a sleek, single engine jet with unswept wings. Since the aircraft's purpose was to explore a flight envelope not previously encountered, the design vertical load factor was a very conservative 12G limit with an 18G ultimate load
factor. [1G is the effect of gravity. A 100 lb. person in a 2G aircraft maneuver would weigh 200 lbs. G forces in excess of 1G can be felt in elevators, automobiles, roller coasters, and moving devices that change direction.]
[Aeronautical engineers typically design aircraft with an ultimate load limit of 1 1/2 times the maximum load limit. At speeds that exceed the ultimate load limit, the aircraft could experience airframe damage and, possibly, disintegrate. At the lower, maximum load limit, the aircraft should not experience any damage or distortion to the airframe or skin of the aircraft].
First, a short explanation of terminology. Those conditions expected to be encountered in service are known as limit conditions. The aircraft is designed to survive loads fifty percent higher than limit loads. Those loads, and the
resulting stresses are designated as ultimate loads and stresses.
I should point out that the stresses calculated during analysis are compared with material allowables furnished by the U. S. Government. These are designated as "minimum guaranteed" and "typical." The "typical" material is ten percent stronger
than the "minimum guaranteed" material. In most cases, we were required to show positive margins in material strength when comparing the operating stresses with the minimum guaranteed.
In order to achieve surface smoothness, the fuselage was built of thick (0.25 inch) magnesium sheet. The structure type was pure "monocoque" rather than "semi monocoque" with thin skin supported by stringers. Complicating this simple structure
was a large door in the upper quadrant of the center fuselage needed for access to the engine.
In the critical structural design conditions, the down inertia loads and tail loads are resisted by up loads from the wing. The fuselage acts like a beam. The bending causes high tension loads in the top quadrant where the door was located. The
door had to be load carrying, yet be easily removable. The screws attaching the door were clearance (sloppy) fit and could not be depended on to carry the longitudinal loads across the joint. This load was carried by tongue and groove
fittings.
Now to the joint. The analysis was not easy and required some assumptions. After much head scratching, I concluded that the tongue and groove joint was under-strength, and I rejected the design. I informed my boss, Pete Shaw, and we discussed it with the designers. It soon became apparent that increasing the size of the parts in the joint snow balled into even more changes. Within our available space, a strength increase was difficult. Pete knew that the analysis was difficult and believed that the design might be O.K. It was finally agreed that I would approve the design and let the static test be the final proof. I had the prerogative to refuse, and Pete would have had to sign for strength. Pete and I had a good relationship, and I knew that he would take responsibility for any premature failure.
Now came the proof of the pudding: the static test in the critical flight condition, 18G ultimate load [12G multiplied by 1.5]. The lab test was typical with hydraulic jack loads representing the aerodynamic and inertia loads calculated to balance the aircraft. At that time, in 1947, the energy needed to load the jacks was provided by a mechanic manually pumping a jack rather than using an electric pump. The test was unusual in that, as the load increased to, and past, the limit load, the wing showed typical bending deflections, but the fuselage showed no sign of stress. Usually, a fuselage would show significant shear wrinkles at limit load and deep wrinkles at ultimate load. Now, as the mechanics sweated at the pumps to increase the loads to ultimate, I became nervous, knowing of the questionable joint strength.
Finally, the we reached ultimate load (150 percent of limit load), and still, the fuselage showed no sign of stress. This, of course, was because of the heavy magnesium monocoque construction. Everyone was jubilant. Some, including the chief engineer, Ed Heineman, inspected the test specimen, even walking under the highly loaded fuselage but not me! Satisfied, Heineman directed that we should increase the load until we had a failure or the load reached 200 percent of limit load, whichever came first. With this order, the mechanic at the pump started his first stroke. Before he completed that first stroke, the fuselage broke completely in two and crashed to the floor where the "big shots" had been casually looking at the structure. The source of the failure was the questionable joint.
When we returned to the office, I pointed out to Pete that I had been correct in my concern with the joint, since the material was probably "typical" rather than "minimum." He only smiled and responded, "It took ultimate loads, didn't it?" I did not mind because I knew that he agreed with me, and his confidence in my work was only increased.
Technically, after a failure like this, we should have taken a small sample of the failed part, determined the actual material strength, and calculated a failing percentage of limit load, based on guaranteed vs. actual. We seldom went through this process since the failing sequence was usually complex.
While not pertinent to the account of the static test, it is worth noting that in service, the two (or three?) D558 airframes performed well in collecting data in the high subsonic and transonic range. It established a maximum speed record of 653 mph in 1947. Its speed records were soon broken by the Bell XS-1.