Environmental Engineering Reference
In-Depth Information
installed from the dam crest at depths of up to 16 metres. Calculations showed that fail-
ure surfaces encompassing this wedge had a limit equilibrium factor of safety of about 1.3.
Numerical analysis was consistent with this. What appears to have happened is that the rela-
tively stiff core has sheared locally particularly during drawdown events at a number of
depths within the section due to lack of support from the poorly compacted rockfill.
11.6.3
The lessons learnt
The lessons learnt can be summarised as:
(a) Conventional effective stress analysis ignores pore pressures generated during shearing.
This is conservative for over consolidated, dilative soils. As discussed in Section 6.1.3
Ladd (1991) shows that effective stress analysis of normally consolidated (and slightly
over-consolidated) clays can very significantly over-estimate the factor of safety. The error
can be as much as 2 times.
(b) What must be recognised is that even though the soil in modern dams is given a degree
of over consolidation by rolling and by partial saturation suction effects, this is often
only equivalent in over-consolidation to the stress due to
20 m to 30 m of dam height
so, in the lower part of larger dams, the dam core is likely to be normally consolidated.
If the dam is affected by transient loading, such as drawdown or high reservoir levels
or high pore water pressures in cracks in the crest (see below), the loading is of short
duration compared to the drainage path, so undrained loading occurs and to model
this correctly, either the pore pressures generated in shearing should be included in the
effective stress analysis or undrained strengths used.
(c) For poorly compacted fills such as those in Hume No.1 Embankment, which were wet-
ted by seepage through the dam or foundation, the soils behave as normally consoli-
dated clays even at shallow depths. In fact, the CPT data from adjacent to walls at
Hume No.1 Embankment and at O'Shannassy Dam shows that the fill may “hang up”
on the walls, so the effective vertical stress at depth is lower than that calculated by unit
weight x depth to the surface and strengths are even lower than expected.
(d) For cases of marginal stability it is important to model the actual conditions carefully
including using undrained strengths, cracks and softened zones in the partially satu-
rated zone and pore pressures in cracks or softened zones during rain events. This of
course means that site investigation methods capable of locating such cracks and tran-
sient pore pressures are needed, e.g. using trenching across the dam crest and using
continuous reading or maximum pressure recording piezometers.
(e) One should be conscious of the possibility that large deformations of the core may
occur where rockfill is poorly compacted, even though the factor of safety is well
above 1.0.
11.7
ANALYSIS OF DEFORMATIONS
The analysis of deformations of embankment dams, either as an aid to the assessment of
stability or to assess the likelihood of cracking or hydraulic fracture, has become possible
within most dam engineering organisations, with the advent of powerful personal com-
puters and relatively inexpensive finite element and finite difference analysis programs.
We do not propose to attempt to review these methods, and instead recommend that
several overview and case-specific papers listed below be read. These include:
(a)
Overview - State of the art papers
:
- Eisenstein and Naylor in ICOLD (1986a) Static analysis of embankment dams;
