Material and testing variability - Geotechnical Investigation and Design Tables - page 118

•

Experts consisted of 4 eminent engineers to predict the performance characteris-
tic, including height of fill required to predict the failure of an embankment on
soft clays.

•

30 participants also made a prediction.

•

Table shows the variation in this prediction process.

Table 10.12 Variations in prediction of height difference at failure (after Kay, 1993).

Standard of prediction No. of participants Coefficient of variation

Expert level 4 14%

Industry practice 30 32%

•

Amuch lower variation of experts also relates to the effort expended, which would
not normally occur in the design process.

•

The experts produced publications, detailed effective stress and finite element anal-
yses, including one carried out centrifuge testing. These may not be cost effective
in industry where many designs are cost driven.

10.13 Tolerable risk for new and existing slopes
•

The probabilities of failure are more understandable to other disciplines and clients
than factors of safety. A factor of safety of 1.3 does not necessarily mean that
system has a lower probability of failure than a factor of safety of 1.4.

•

Existing and new slopes must be assessed by different criteria.

Table 10.13 Tolerable risks for slopes (AGS, 2000).

Situation Tolerable risk probability of failure Loss of life

Existing slope 10 − 4 Person most at risk

10 − 5 Average of persons at risk

New slopes 10 − 5 Person most at risk

10 − 6 Average of persons at risk

10.14 Probability of failures of rock slopes
•

A guidance on catastrophic versus minor failures probabilities are provide in the
Table.

Table 10.14 Probability of failure in rock slope analysis (Skipp, 1992).

Failure category Annual probability Comment

Catastrophic 0.0001 (1 10 − 4 )

×

Major 0.0005 (5 10 − 4 )

×

Moderate 0.001 (1 10 − 3 )

×

Minor 0.005 (5 10 − 3 ) For unmonitored permanent urban slopes with free access

×