Actes du colloque - Volume 4 - page 655

3315
Technical Committee 210 + 201 / Comité technique 210 + 201
…”. He identified that the cracking was mainly caused by
differential settlement of homogenous clay dams or by
hydraulic fracturing of the core material due to the water
pressure after impounding of the reservoir.
Numerous filter tests were performed (Sherard et al. 1984a),
and based on the slot test data (Sherard at el 1984b) four soil
categories with four individual filter criteria were identified:
1.) Sandy silts and clays (d85b: 0.1-0.5 mm): D15f/d85b ≤ 5
2.) Fine-grained clays (d85b: 0.03-0.1 mm): D15f ≤ 0.5 mm
3.) Fine-grained silts (d85b: 0.03-0.1 mm): D15f ≤ 0.3 mm
4.) Exceptionally fine soils (d85b < 0.02 mm): D15f ≤ 0.2 mm
With the non-erosion filter test the filter criteria were further
developed and termed criteria for “critical filter” (Sherard &
Dunnigan 1985, 1989) as distinct from the “perfect filter”
discussed above. For the critical filters four categories were
defined based on the fines content (<0.075 mm, sieve 200) of
the base soil (or core material). The fines content was determined
on a gradation curve with a maximum grain diameter of 4.75 mm
(sieve 4). For base soils with a maximum grain size exceeding
4.75 mm, the gradation curve was regraded to ≤4.75 mm in order
to determine whether the base soil falls into category 1, 2 or 4.
Whether the base soil falls into category 3 was determined on
the original, non-regraded curve. For each of the 4 categories a
filter criterion was defined (Tab. 1). These criteria still apply
today. The current design approach is to use the conservative
values of these criteria, as given in the right column of Tab. 1.
Table 1. Filter criteria.
Soil
group
Fines
content
<0.075mm
Filter criterion
determined by tests
after Sherard &
Dunnigan (1989)
State-of-the-Art
criteria in dam
engineering
1 85-100 D15f = 7d85b to 12d85b D15f ≤ 9d 85b
2 40-80 D15f = 0.7 to 1.5 mm D15f ≤ 0.7 mm
3 0-15 D15f = 7d85b to 10d85b* D15f ≤ 4 to 5 d85b
‡
4 15-40 Intermediate between
group 2 and 3
Intermediate between
group 2 and 3
*For subrounded grain shape 7 and for angular grains 10.
‡ Incorporates a factor of safety of two.
2.2 Internal stability
For filter materials to be internally stable means that within
the soil skeleton the small particles do not move due to water
flow forces. All soil particles should remain at their position
even for water flow at high (>>1) hydraulic gradients such as
occur at a fracture in the sealing zone of an embankment. A
good definition of internal stability is given e.g. by Kenney &
Lau (1985): “Internal stability of granular material results from
its ability to prevent loss of its own small particles due to
disturbing forces such as seepage and vibration.” In more
recent literature, the term internal stability is used in a much
broader sense 3. However, in this paper the term will be used for
the filter material design, and the internal stability of natural
soils (in the foundation or dam fill) will be discussed at the end
of this chapter.
Concerning the formation of sinkholes at the crest of zoned
embankment dams, James Sherard (1979) studied the
phenomenon and recommended use of a method proposed by
3 Fell and his co-workers in Australia (e.g. Foster and Fell 1999)
discussed internal erosion by investigating the erosion process within
the soil structure. They divided the erosion process into four steps: (i)
initiation (ii) continuation (iii) progression (iv) breaching/ failure. The
term internal erosion was divided into four sub-categories: (a)
Concentrated leak erosion (b) Backward erosion (c) Contact erosion (d)
Suffusion. These four sub-divisions were taken over by the latest
ICOLD Bulletin “Internal Erosion of existing Dams and their
Foundations.” Hence, what was previously termed internal stability of
filters now falls into the sub-category (d) Suffusion.
Prof. Victor de Mello (1975) for the investigation of gap-graded
soils, in order to assess the internal stability of filter materials.
In this method, which is also called “retention ratio
criterion”, the gradation curve of the filter material is divided
into two curves at a selected grain diameter (dS ), gradation
curves for the portions finer and coarser than dS, respectively.
For the two gradation curves the retention ratio (RR) is
calculated from the Terzaghi filter criterion: RR = D15f/d85b. This
is repeated for different values of dS . All grains are considered
to be stable if they satisfy the criterion RR ≤ 7÷8 for subrounded
grains or RR ≤ 9÷10 for angular grains. The grain diameters (dS )
for which the retention ratio exceeds the given limits are
potentially unstable and can be eroded by the water flow. Using
this criterion to identify stable materials shows that gradation
curves with a more or less straight line in the semi-logarithmic
plot are stable.
Experimental investigations performed by Kenney & Lau
(1985 and 1986) lead to a strict criteria in which the gradation
curve of the fine part of the filter material (0<M%<30) should
be on the more uniform side of the Fuller curve and the
gradation curve of the coarser part of the filter material
(30<M%<100) should be on the more uniform side of a straight
line in the semi-logarithmic plot with a uniformity coefficient of
Cu ≤ 12 (Kenney & Lau 1986).
With this criterion, rather uniform filter materials are defined
as internally stable. Such materials can be produced for man-
made structures but they are rare in nature e.g. in soils present in
the foundation of dams. Hence, for the assessment of natural
soils with respect to internal stability, the approach is not to
define the gradation but the critical hydraulic gradient. These
studies were first done in Russia with the start of the
construction of large run-of-river power plants in the 1920s (e.g.
Pavlovsky 1922). Patrashev (1965) proposed a suffusion
criterion and Pravedny (1976) a criterion for contact erosion.
These criteria are not further discussed in the present paper as
they are not applied for the design of man-made filter materials.
2.3 Self healing
Self-healing means that cracks which can form in the filter zone
due to e.g. differential settlement, etc. do not stay open but close
in case of water flow. Hence, the filter material must not have
cohesion. This is assured by limiting the content of non-plastic
(IP<5%) fines to less than 5% (the latest ICOLD Bulletin on
CFRD’s, No. 141, allows 7% of fines). The sand-castle test
(Vaughan & Soares 1982), confirms that the selected filter
material meets the self-healing requirements.
2.4 Material segregation
When the filter material segregates, meaning that the coarser
particles separate from the finer particles, the filter zone can no
longer fulfill its purpose of preventing fine particles moving
from the core to the filter zone or within the filter zone, because
the segregated coarse grained components do not form a filter to
the adjacent materials. Hence, the segregation of filter materials
has to be avoided. Whether a material segregates depends on the
handling and placement methods and on the gradation of the
material. In the 50’s and 60’s of the last century, segregated
material zones were improved manually. Later, the focus was
put on the selection of appropriate gradation curves. One of the
first discussions on segregation criteria is given in Sherard et al.
(1984b) where they proposed a coarse boundary for filter
materials (see also Fig. 2). It was generally agreed that a high
content of sand and a small maximum grain size reduces the
segregation. Based on observations and laboratory investigations
(e.g. Sutherland 2002) a stricter criterion was presented by
Milligan (2003), which specifies that wetted filter material with
a gradation finer than the limit curve given in Figure 2 should
be selected. The latter criterion is nowadays commonly applied.