UNCERTAINTY AND RELIABILITY - Introduction to Soft Soil Geotechnique: Content, Context and Application - page 114

approached by 43 international experts using FEM, subgrade reaction and other
methods to predict deformations, moments and forces. The outcome showed a wide
spread: three quarter of the experts obtained an answer beyond 50% of the test
result. Those who disregarded capillary cohesion were completely out of range.
In 1999 an anchored sheetpiling test was organised in Rotterdam. 6.5 m of soft
clay and peat were retained at high groundwater table. 23 international experts
made a prediction for horizontal displacement, plastic hinge and oblique bending,
applying FEM and subgrade reaction methods. Results showed again a wide
spread: for the displacement, by FEM between 45 and 210 mm and by subgrade
reaction between 62 and 173 was calculated, while 107 mm was measured
Surprisingly, FEM was less accurate (more parameters had to be guessed). It was
observed that data interpretation for parameter values for strength and stiffness
varied significantly (19 o <

< 35 o , 2 < c < 10 kPa). Another striking outcome was
that different sets of parameter values could produce the same result.

Engineering factor
The examples mentioned above indicate that a significant part of the spread in
geotechnical prediction is due to lack of information, different interpretations and
subjective choices. If more specific information becomes available, the spread will
decrease, but uncertainty about soil stratification, initial state, soil behaviour and
boundary conditions remains. We could refer to that as the engineering factor ,
since it is related to individual knowledge and personal experience. For the
examples discussed in this chapter, the uncertainty (95% range) varies somewhere
between

0.70% which is quite large; it is 10 times larger than in steel
construction and 3 times more than in concrete design.

4 0.30% and 4

B PARTIAL SAFETY FACTOR (ISO 2394)
Structures and foundations on and in soft soils have to be reliable, meeting
certain standards of safety and functionality (serviceability and durability). The
general risk of failure or malfunctioning of a structure is related for a major part to
the soil, both during construction (temporary works) and during usage (permanent
works). Soil has a large heterogeneity and a wide variety of failure mechanisms.
Prediction models are in principle approximate. By risk analysis proper insight in
the possible failure (slip, uplift, squeeze, settling, deforming) of structural elements
(dam, embankment, wall, trench, pit, tunnel), and their interaction can be obtained.
One considers the ultimate limit state (ULS) and the serviceability limit state
(SLS). A collapse of a slope fits ULS and damage of a road surface because of soil
deformations SLS.
The reliability of a structure is expressed by 1 - P f , where P f is the probability of
failure. The safety is related to risk of malfunctioning, and risk is normally
expressed as Risk = Probability of Failure x Consequences. The consequences are a
matter that involves feeling of well-being and cannot be determined
unambiguously. In this respect, the term safety factor should be conceived as a
stability factor, when addressing the probability of failure.
Malfunction can be expressed in terms of a reliability function Z