Air Force Warehouses

Two warehouse roofs at Air Force Bases in Ohio and Georgia cracked and collapsed under combined load, shrinkage, and thermal effects in 1955 and 1956.  122 m (400 ft) lengths of reinforced concrete roof girders functioned as single units because of defective expansion joints.  Other warehouses, built to the same plans, survived because the separation between adjacent two-hundred-foot bays was maintained by functioning joints.  These failures led to more stringent shear reinforcing steel requirements in subsequent editions of the ACI Building Code.  In the warehouse structures, the concrete alone, with no stirrups, was expected to carry the shear forces, and the members had no shear capacity once they cracked (McKaig 1962, Feld and Carper, 1997).

Wilkins Air Force Depot, 1955

At the Wilkins Air Force Depot in Shelby, Ohio, about 370 m2 (4,000 ft2) of the roof collapsed suddenly on August 17, 1955.  At the time of the collapse, there were no loads other than the self-weight of the roof (Feld 1964, p. 25).

The Air Materiel Command (AMC) built warehouses to a common design at many Air Force bases and depots.  The original design was developed in April 1952, with a modification to reinforcement made in March 1954.  The Ohio warehouse had been built to the original 1952 design.  It was a six-span rigid frame building, 122 m (400 ft) wide and 610 m (2,000 ft) long.  The haunched rigid frames each had six 20 m (67 ft) spans and were spaced approximately 10 m (33 ft) on center.  The concrete for each frame was placed continuously in a single working day. Vertical steel plate construction joints were set at the center of each span before concrete placement, but they may not have been effective (Feld 1964, p. 25).

Severe cracking had been observed two weeks before the collapse, so the girder had been supported by temporary shoring.  The cracks occurred about 0.45 m (1 ft) past the end of the cutoff of the top negative reinforcement over the columns (Feld 1964, pp. 26- 27).  A typical AMC warehouse frame is shown below.

Robins Air Force Base, 1956

A second warehouse roof collapse took place at Robins Air Force Base near Macon, Georgia, early on the morning of September 5, 1956.  This warehouse had been built to the revised design.  The revision added top bars and nominal stirrups, at a volume of about 0.06 %, for the length of the frames.  This collapse included two adjacent girders and about 560 m2 (6,000 ft2) of the roof.  Before the collapse occurred, cracks in the concrete girders that reached 13 mm in width had been observed.  Feld (1964, p. 25) suggests It seems that the extent of shrinkage and resulting axial tensions may be somewhat related to the speed of concreting or to the extent of each separate placement.

In both cases, the design, materials, and workmanship were up to the codes and standards of the day.  However, the failures had still occurred.  Feld (1964, p 27) believed that failure took place by a combination of diagonal tension (shear) due to dead load and axial tension due to shrinkage and temperature change.  Circumstantial evidence suggested that high friction forces were developed in the expansion joint consisting of one steel plate sliding on another; some plates showed no indication of relative displacement since their installation.  In other words, the expansion joints locked and did not function to relieve stress.

Lessons Learned

Following these failures, construction of warehouse structures of this type was halted until a new design was developed.  Existing warehouse frames were strengthened by adding external shear reinforcement consisting of tensioned steel strapping and steel angles at the lower corners of the girders.  The ACI Building Code shear provisions were also revised (Feld 1964, pp. 27 -28).

Since these warehouse designs had been built in many locations, questions arose as to why these two warehouses had collapsed and not others.  At the time of the collapse, these two structures were subject to wider temperature variations, and were about a year to a year and a half old.  The temperature stresses, unrelieved by the locked expansion joints, combined with shrinkage and shear effects to cause high tensile stress.  At the Wilkins Air Force Base ( Ohio ) structure, the failure plane did not cross any shear stirrups.  The rapid, monolithic casting of the frames was thought to exacerbate shrinkage and to contribute to the problem (Feld and Carper 1997, pp. 255- 257).

Although it was not required by the code, many designers at that time used at least two continuous top bars and stirrups at a maximum spacing of 300 mm (12 in) along rectangular continuous beams.  Following the two failures, tests were carried out at the Portland Cement Association (PCA) laboratories in Skokie,Illinois.  Beams were tested with flexure only, and flexure plus axial tension.  In the latter test, the beams under tension and flexure failed in the same manner as the two warehouse frames (Feld and Carper 1997, pp. 258 -259).

These two cases illustrate the importance of providing at least some minimum amount of reinforcement, called temperature steel, to resist tension forces due to thermal, shrinkage, and other effects.  Real structures do not behave in the same way as our simplified models and develop forces and stresses where our analyses suggest there should be none.  The shear resistance of unreinforced, cracked concrete is virtually nil.  Feld (1964, p. 29) notes a similar warehouse slab collapse in which the crack through the slab did not cross any reinforcement, due to detailing error.


  • Chapter 5 of Beyond Failure: Forensic Case Studies for Civil Engineers
  • Feld, J., and Carper, K. (1997). Construction Failure. 2nd Ed., John Wiley & Sons, New York, N. Y.
  • McKaig, T. (1962). Building Failures: Case Studies in Construction and Design.  McGraw-Hill, New York, N. Y.