Memoirs 5. Barrel Shellsby Milo Ketchum
The first barrel shells we designed were small, in particular I am reminded of a shell for the First National Bank in Boulder, Colorado, that was used as for the drive up window area. It had a span of about 50 feet and a width of 30 feet. In a few years, however, it was covered up by a remodeling of the bank facilities.
In the analysis of barrel shells, I took naturally to the beam method first described in Cylindrical Shells, Volume I, by H. Lundgren, published by the Danish Technical Press. The methods advocated, meshed with the methods for the calculation of folded plates, and described the physical action of barrel shell much better than the more mathematical methods. Later, in a paper "Cylindrical Shell Analysis Simplified by the Beam Method" by James Chinn in Journal of the ACI, May 1959, equations were developed, but we always used incremental methods because the equations never seemed to fit the structures we were designing. In a discussion of this paper, I pointed out that it was not necessary to use a straight line stress distribution, any shape can be used by combining two possible patterns to satisfy statics.
The classic Design of Cylindrical Concrete Shell Roofs, Manual 31 of the ASCE, authored by Alfred L. Parme, was also available at that time, but I never had real confidence in my ability to successfully navigate through the multitudinous interpolation of constants, and the cook book approach to calculations. I preferred something closer to statics; besides, these methods never gave me any real clue on how a barrel shell actually performed. I am sure that part of my problem was the lack of any real mathematical foundation. When I was a member of the shell committee of the ACI, it was interesting that other peoples conception of barrel shell action was different from mine, they based theirs on the concepts of the elasticity method which does not really describe how a barrel shell works. Later I made attempts to solve barrels by these methods and had a shell program written on an IBM 650 computer, but I never really made any personal progress until I had my own personal computer.
The only large scale, long barrel, shell we designed was a factory for the Continental Baking Company in Denver, Fig. 5.1 and 5.2. It had continuous spans of about 75 feet and bay spacings of 40 feet. The shells had an inverted curve at the valley, and there were no internal ties at the columns; forces from the ribs were carried by thrust abutments at the outside columns. Also there were intermediate structural columns at the outside walls which picked up some of the trusts as was done for the high school folded plate roof described in the previous chapter. This building covered a large area, so an efficient forming system was devised for construction. The inverted cylindrical valleys used permanent forms made from commercial scaffolding, were cast first. Then a movable form was used for the center section, and were cast a span at a time. The shoring on the column strip greatly reduced the thrusts from the shells that were cast before an adjacent unit was available to resist them, and the rigid frame stiffener ribs carried the remainder. Additional temporary timber struts in some cases were placed to hold the forces as they were constructed. The structural costs were well within those of similar bakeries belonging to the same chain.
Another bakery for the Rainbo Bread Company, Fig. 5.3, and 5.4, was a short barrel shell. Our client, T. J. Moore, Jr., was enamored of the barrel shells for the open markets in Mexico City built by Felix Candela, which both he and I had seen. The spans were to be much larger, and the exposed horizontal ties that Candela used above the roof were not suitable for the Denver climate. Instead we used rigid frames even though the thrusts become very high. The distance between frames was 35 feet, the span of the frames was 80 feet, and there were three frames in each series. The shell thickness was 2.5 inches over the top area, and 4 inches adjacent to the valleys, which were horizontal for a distance of 4 feet. It was possible to move the forms continuously through the structure with only minimum decentering because there were no internal ties. To provide support for the structure during construction, a center strip of roof was formed separately and left in place until the structure had cured sufficiently to spring the arch elements in the frames. The forms were slid on rails laid on the ground. As in the previous example, the costs of the structure were even with or less than those of the same chain of bakeries.
The short shell elements were designed, essentially, as an arch for the upper 2.5 inch thickness, and as a folded plate for the lower 4 inch thickness to carry the arch thrusts. Equipment loads were to be carried directly, even on the 2.5 inch thickness, so some areas were thickened, and had extra reinforcing. The design of the frames took the most time, and moment distribution was used for the analysis. No distress was ever reported, and several years ago I was asked to check the design for additional equipment loads. This time, with a space frame program, I was able to make a rational analysis of stresses. The entire structure was in excellent condition, and testified to the quality of construction, the maintenance, and the structural properties of shell structures.
One of the most visible shells we designed, was a banquet hall for the Chase Hotel in St. Louis. It was a short shell with a span of about 150 feet. In the space available, on one side there was a hotel bedroom unit with windows that could not be cut off by an adjacent building. On the other side there was an alley way. The solution was a short shell supported on a hybrid rigid frame-arch structure with a vertical on the alley side, and an arch on the hotel side. The vertical leg was prestressed to reduce the size and contain the moments. In order to carry the horizontal thrusts at the floor, prestressing was used for the entire floor; this was the first example we had designed of a complete prestressed structural floor. When we were deciding the shape to be used, having had prior experience, I alerted the owner, who was also the architect to the acoustical problems with curved roofs. He engaged an expert who suggested a system using holes through the slab at frequent intervals, with a sound deadening material above. This system was used, and several years afterward, I heard the Boston Pops under Arthur Fiedler play, and then I thought the acoustics were excellent.
A unique barrel shell is the hangar for the Ideal Cement Company, Fig. 5.6. At that time they produced a lightweight aggregate and wished to demonstrate the possibilities for construction of shells. We used a structure, square in plan, with abutments at two corners with a rear wall supporting the shell on two of the sides, and an arch, broken at the middle, to carry the structure above the doors. Also there was a stiffening rib from abutment to abutment. The front arch was not continuous, and created a tensile force on the shell which was carried by prestressing cables across the top. The structure acts essentially as a short barrel shell, and is in compression, the only difference is that there is no beam element at the base. We considered using a hyperbolic paraboloid, but rejected it because of the difficulty of forming, and inability of the structure to take the tensile forces caused by the broken arch.
The hangar fitted the airplane like a glove, and was a handsome structure when the hangar doors were open, but was rather box-like when they were closed. When I showed this picture later to a Turkish student in one of my classes, he dated the building precisely by the age of the cars parked in front.
In 1949, after only four years of practice, I decided that I should build an office of my own. It would be a good way to establish my firm as a leader in structural design. I choose a flat plate roof, made from natural light weight aggregate, with two spans of 15 feet, crosswise and four spans of 15 feet length wise. I designed the architectural features, and made all the drawings, with the exception of the radiant heating in the floor.
In another four years, we had outgrown this building, so I decided to build a larger barrel shell structure. This time we hired an architect. The span was 40 feet, and the width of the barrels was 20 feet, Fig. 5.7. At the front and back ends, the shell was turned up, and was supported by a beam and intermediate columns to frame the entrances. The north side was completely glazed to provide even light for drafting, and the low walls on that side were fixed at a height so that there was no glare from outside reflections. This created a dark space under the window, but served as a space for reference tables. On the south side, there was a larger overhang to keep off the sun in the summer, and to provide some warm light to mix with the cool north light. The structure was heated by warm air from a complete plenum under the floor, with the exhaust at the windows to prevent cold air from spilling on the occupants. There was some trouble from the whispering gallery affect, but the acoustics were otherwise satisfactory. These details were all very radical in their day.