In January of 2007, seven people in a São Paulo street, four of them in a small bus, were suddenly sucked into falling soil and saprolite, from a street elevation about 20 m above a metro station cavern of 19 m span and 40 m length. Despite the evidence of four surrounding and one central borehole, and six more boreholes around the adjacent station shaft, the assumed mean rock cover of just 3 m above the 20 m deep cavern arch, proved locally to be more than 10 m in error, due to a buried ridge of rock running high above the cavern arch, with one fateful low point exactly where drilled on the cavern centre-line. Seven lanes of adjacent highway and twin railway lines prevented viable seismic refraction, and a consistent drilling result had not proved this to be necessary. Due to the assumed low rock cover, heavy lattice girders, embedded in 40 cm of S(fr) were used as temporary support. The feet of the lattice girders were founded on broad ‘elephant’ footings. Due to the unknown adverse loading from a wedge-shaped, clay-bordered, giant ridge of rock and saprolite, weighing some 15,000 tons, all forms of temporary support would eventually have failed. Post-collapse, painstaking, police-supervised excavation of the entire 20 by 20 by 40 m of collapsed materials, taking some 15 months, finally revealed large remnants of the arch and wall support, crushed and folded beneath the fallen gneiss, amphibolite, saprolite, sand and soil.


Fortunately for the civil engineering tunneling profession, the dramatic case record to be described here, with tragic consequences for seven people, is extremely rare. The geometric peculiarities described in the abstract above, were compounded by opposed, steeply-dipping foliation, and by a steep, bounding discontinuity resembling a fault, and by the presumed cracking of a buried 1000 mm diameter storm drain, with change of cross-section to 700 mm immediately upstream of this major discontinuity.

A slight shearing of this feature, caused by the advance of the cavern many meters below, is presumed to have been the reason for the leaking, broken pipe that was filmed following collapse of the 16,000 m3 of soil, sand, saprolite, gneiss and amfibolite. To complete the list of adverse features, there was exceptionally heavy rainfall in a period three to four weeks before the sudden collapse, perhaps responsible for locally elevated water pressure and perhaps for softening of the clay and saprolite surrounding the 10 m high, sub-surface ridge-of- rock. In retrospect, this ridge was the result of a higher RMR-value, as consistently logged by the geologist, along the cavern centre, with lower RQD on either side.

(Figure in full paper)

The eighteen plastic containers seen in the above core box, contain the nominal recovery of 18 m of overlying sand, soil and saprolite. The rock head, encountered first at an average 18 m depth, at an average elevation of 706 m, and 3 m above the cavern arch, was weathered gneiss.

This content is only available via PDF.
You can access this article if you purchase or spend a download.