Design, specification and construction of filters - Geotechnical Engineering of Dams - page 360

These criteria were based on construction experience.
Kenney and Westland (1993) carried out laboratory tests and concluded that:

-All dry soils consisting of sands and gravels segregate in the same general way, inde-
pendent of grain size and grain size distribution;
-Dry soils containing fines 0.075 mm segregate to a smaller extent than soils not contain-

ing fines;
-Water in sandy soils (mean size finer than 3 mm to 4 mm) inhibits segregation but has
little influence on the segregation of gravels (mean size coarser than 10 mm to 12 mm).

The USDA-SCS (1994) and USBR (1987) have adopted a maximum size of 75 mm
and limits on the minimum D 10F and maximum D 90F to limit segregation. These are
described in Section 9.2.3. The authors believe that for narrow or thin filter zones, 75 mm
may be too large and recommend the use of maximum size of 37 mm or 50 mm in these
situations.

9.2.2.3 Permeability
The filter must be sufficiently permeable for the seepage flow to pass through it without
significant build up of pressure. This has been taken into account by using the criteria
D 15F /D 15B 4 or 5, which ensures that the permeability of the filter is 15 to 20 times that

of the soil. However just as important is to keep the fines content (silt and clay sized par-
ticles) to a minimum.
Figure 9.18 shows the influence on permeability of the type and amount of fines.
The authors' preference is to specify not greater than 2% fines and that the fines be non
plastic. Some, e.g. USDA-SCS (1994), USBR (1987), allow 5% (non plastic) fines, but as
is evident from Figure 9.18, this may reduce the permeability by one or two orders of
magnitude compared to clean filter materials. The cost in washing to achieve not more
than 2% fines is not high and generally worthwhile. The second advantage of low fines
content is that the filters are unlikely to hold a crack.
Note that the discharge capacity of a filter drain system is a separate issue, discussed in
Chapter 10.

9.2.3 Review of available methods for designing filters with flow parallel to the

filter

ICOLD (1994) summarize testing at Delft Hydraulic Laboratory (Bakker, 1987) to test
the condition where flow in the filter is along, or parallel to, the filter. Den-Adel et al.
(1994) provide details of a method for designing such filters.
It is apparent that in these situations, erosion is less likely and considerably coarser
filters than obtained from the criteria discussed in Sections 9.1 and 9.2.4 can be used.
Figure 9.19 summarizes some of the Delft Laboratory Testing.
In this figure the hydraulic gradient, i cr , e.g. equal to the slope of the filter under rip rap
under the upstream face of a dam, is plotted against D 15F /D 50B (D 15 /d 50 in Figure 9.19).
Curves for sandy base soils with D 50 0.15 mm and 0.82 mm are shown. Thus for a uni-

form sand base material with a d 50 of 0.15 mm on a slope of 2.5H:IV (i cr 0.4), the ratio

of D 15F /D 50B can be as high as about 8.
However great care should be taken in applying this approach, particularly to
situations where, if the filter fails and erosion occurs, it is not practical to repair the
damage. Readers should seek the latest literature and be willing to use the rather
daunting detail outlined in Adel et al. (1994) if they are to use other than normal filter
criteria.