Page 14 - jdrogers2017aa

World Environmental and Water Resources Congress 2017 392

appears to have been missed by the various investigative panels, who did not perform
independent verifications of the dam’s dimensions.
The testimony suggests that the only engineering design work was carried out for the
maximum cross section of the Hollywood Dam; and that this design was subsequently applied to
the topography of the St. Francis site in San Francisquito Canyon. Bayley’s calculations appear
to have been limited to a static analysis of the dam’s highest section, estimating the factors of
safety against 1) cantilever bending/overturning; and, 2) basal sliding.
The dam collapsed 10 days after the reservoir’s initial filling, to within three inches of the
spillway sill. This condition would have represented the maximum pore water pressure being
applied to the dam mass and its foundations.
The coefficients of friction assumed for the foundation materials were inappropriately
low for planes of foliation in mica schist, and for gypsiferous horizons in the arkosic
conglomerate, which was subject to slaking upon submersion.
Foundation exploration was minimal, consisting of 10 shot borings in the stream channel
and one exploratory adit 30 to 40 feet long into the Pelona Schist on the left abutment, just
downstream of the dam. Keyway excavations into the sloping abutments were also minimal, the
deepest being between 3.6 and 4.3 m.
No accommodation for uplift relief was installed beneath the sloping abutments, which
were comprised of contrasting materials (mica schist and arkosic conglomerate, separated by a
fault). The fault could also have served as a significant aquitard, restricting downward
percolation of seepage from the reservoir through the right abutment.
The dam was unknowingly constructed against an ancient bedrock landslide complex
developed in the Pelona Schist. This was identified by Prof. Bailey Willis of Stanford University
after the failure (Willis, 1928), and evaluated in some detail by Rogers (1992, 1993, 1995, and
1997). Elevated pore water pressure in the old landslide likely served to reactivate a small
portion of this mass, about six times greater mass than the dam.
William Mulholland’s decision to caulk and grout the transverse shrinkage cracks in
January and February 1928 (Fig. 7) likely triggered the dam’s untimely demise a few weeks
later. Mulholland’s goal was to save precious water being lost through the cracks, but caulking
the fissures with oakum on the downstream face served to trap reservoir water pressure within
the dam itself, a potentially catastrophic situation because it would have hastened internal
instability. This condition is born out to a noticeable degree in the downstream tilt of about one
degree, recorded by the Steven’s Gage on Block 1 of the dam, beginning around 8:30 PM on
th
March 12 , 3-1/2 hours before the failure (Rogers 1995; 2007). Few people at the time
understood the destabilizing impacts of pore water pressure beneath concrete arch dams, which
were altered radically by the failure of Malpasset Dam in France in 1959.

Import of Peer Review. Without any site-specific design input other than the site topography, it
must have been an awful embarrassment for William Mulholland and the City’s Bureau of
Waterworks & Supply to have the St. Francis Dam fail catastrophically, and for the first 62
victims to have been City employees and their dependents living by the dam and San
Francisquito Powerhouse No. 2.

In his testimony before the Coroner’s Inquest a sorrowful Mulholland said that he “only
envied those who were killed.” He went on to say Don’t blame anyone else, you just fasten it on
me. If there was an error in human judgment, I was the human.” No truer words were ever
spoken.

World Environmental and Water Resources Congress 2017