CEng 7033 - Design and Analysis

of Embankment Dams:
Design of Filters

GENERAL REQUIREMENTS FOR DESIGN AND

THE FUNCTION OF FILTERS

 Filters in embankment dams and their foundations are required to

perform two basic functions:
(a) Prevent erosion of soil particles from the soil they are protecting;
(b) Allow drainage of seepage water.
 Filters are usually specified in terms of their particle size distribution.
 They are required to be sufficiently fine, relative to the particle size of
the soil they are protecting (the “base soil”), to achieve function (a),
while being sufficiently coarse to achieve function (b).

1
 GENERAL REQUIREMENTS FOR DESIGN AND
THE FUNCTION OF FILTERS

 To achieve the desired two functions the ideal filter or filter zone will
(ICOLD 1994):
 Not segregate during processing, handling, placing, spreading or
compaction;
 Not change in gradation (by degradation or break down) during
processing, handling, placing and/or compaction, or degrade with time
e.g. by freeze-thaw or wetting and drying by seepage flow;
 Not have any apparent or real cohesion, or ability to cement as a result
of chemical, physical or biological action, so the filter will not allow a
crack in the soil it is protecting to persist through the filter;
 Be internally stable, that is the fines particles in the filter should not
erode from the filter under seepage flows;
 Have sufficient permeability (and, if a drain, thickness) to discharge the
seepage flows without excessive build-up of head;
 Have the ability to control and seal the erosion which may have initiated
by a concentrated

FLOW CONDITIONS ACTING ON FILTERS

 Figure 9.1 illustrates the basic flow conditions that can occur between a
filter and base soil. These are:
 N1 Flow normal to the base soil – filter interface, with potentially high
gradient conditions; e.g. at the downstream face of the earthfill core, the
contact between the horizontal drain and the foundations, and within the
foundation where seepage is across bedding.
 N2 Flow normal to the base soil – filter interface, with low gradient
conditions; e.g. at the upstream face of the earthfill core under reservoir
drawdown conditions, or into the upper Zone 2A filter for the horizontal
drain.
 P Flow parallel to the interface, e.g. at the base of rip-rap layers, or in the
foundation where seepage is along bedding.
 The erosive stresses are greatest for case N1, and less for N2 because for N2 the
flow is draining from the base soil under gravity, not under reservoir water head.
 The erosive action for case P is different and also less severe than for N1. As a
result less conservative (and therefore coarser) filter may be used for cases N2
and P than for N1.

2
 FLOW CONDITIONS ACTING ON FILTERS
 Flow Conditions

Critical and non critical filters

 Figure 9.2 and Table 9.1 show a number of applications of filters in
dams. Some filters are critical to the control of internal erosion in the
dam and, if they fail, give an increased likelihood of piping
progressing and potentially breaching the dam. These are termed
critical filters, and they should be designed and constructed to meet
stringent, no-erosion filter criteria. Examples are filters “g” in Figure
9.2.

3
 Critical and non critical filters
 Some filters are non-critical, in that, if some erosion occurs, it can be
repaired (e.g. beneath rip-rap, locations “a”, “b” and “c” in Figure
9.2) or it will cause problems only during construction (“d” and “e”, in
Figure 9.2).
 Most critical filters are in an N1 flow condition (Section 9.1.2) and
non-critical filters in an N2 or P flow condition.

Critical and non critical filters

 Filters “f” and “h” are critical to the performance of the dam, but are
in N2 flow conditions, so may be designed and constructed to lesser
standards.
 Filters “d” and “I” may be critical or non-critical, depending on
whether or not erosion of the embankment fill into the foundation
can occur. If it can, the filters are critical. As shown, “d” is a
construction drainage system, so would probably be non-critical.

4
 Critical and non critical filters
 1. Sources of materials, for the following purposes:

Critical and non critical filters

 1. Sources of materials, for the following purposes:

10

5
 Filter Design Notations
 D15F Particle size of the filter material for which 15% by weight is
finer.
 D85B Particle size of the base material for which 85% by weight is
finer.
 pp% 0.075mm Percent finer than a particle size of 0.075 mm.
 Fines content pp% 0.075 mm.

11

Filter Design Concepts

 Figures 9.4(a) & (b) show the interface b/n a filter and base soil.
 The basic concept of filter design is to design the particle size distribution of
the filter so that the voids in the filter are sufficiently small to prevent
erosion of the base soil.
 The void size in the filter is controlled by the finer particles and, for design
purposes, the D15F is usually used to define this void size.
 Sherard et al. (1984a) showed that for granular soils the void size between the
soil particles, known as the opening size, is given by OE D15F/9.

12

6
 Filter Design Concepts
 A further basic concept, inherent in filter design, is that the base soil will
generally provide a degree of “self-filtering”.
 Hence in Figure 9.4(a), in a well graded base soil, the coarser particles in
the base soil are prevented from eroding into the filter and they in turn
prevent the medium sized particles in the base soil from eroding and the
medium sized particles in the base soil prevent the fine particles in the base
soil from eroding.

13

Filter Design Concepts

 If the base soil is gap-graded or graded concave upwards (Figure 9.5), there
is a deficiency of medium sized particles, as shown in Figure 9.4(b). The self
filtering does not occur and the fine particles in the base soil will erode
through the coarse particles, a process called suffusion, or internal
instability. In these situations, for the filter to be successful in controlling
erosion, it must be able to control the erosion of these finer particles.
 In most filter design methods, the base soil particle size is characterised by the
D85B size, although some use D95B and D50B

14

7
 Filter Design Concepts
 ICOLD (1994) describes the evolution of filter design practice.
 These have basically evolved from the concepts of Terzaghi (1926) who
proposed that D15F/D85B ≤ 4 to control erosion and D15F/D15B ≥ 4 to ensure
the filter was sufficiently permeable.

15

Filter Design Concepts

 ICOLD (1994) describes the evolution of filter design practice. These have
basically evolved from the concepts of Terzaghi (1926) who proposed that
D15F/D85B 4 to control erosion and D15F/D15B 4 to ensure the filter was
sufficiently permeable.

16

8
 Filter Design Concepts
 ICOLD (1994) describes the evolution of filter design practice. These have
basically evolved from the concepts of Terzaghi (1926) who proposed that
D15F/D85B 4 to control erosion and D15F/D15B 4 to ensure the filter was
sufficiently permeable.

17

Filter Design Concepts

18

9
 Filter Design Concepts

19

Filter Design Concepts

 ICOLD (1994) describes the evolution of filter design practice. These have
basically evolved from the concepts of Terzaghi (1926) who proposed that
D15F/D85B 4 to control erosion and D15F/D15B 4 to ensure the filter was
sufficiently permeable.

20

10
 Filter Design Concepts
 ICOLD (1994) describes the evolution of filter design practice. These have
basically evolved from the concepts of Terzaghi (1926) who proposed that
D15F/D85B 4 to control erosion and D15F/D15B 4 to ensure the filter was
sufficiently permeable.

21

Filter Design Steps – Steps 1 to 3

 Step 1: Plot the gradation curve (grain-size distribution) of the base
soil material.
 Use enough samples to define the range of grain sizes for the base soil
or soils. Design the filter using the base soil that requires the smallest
D15 size for filtering purposes. Base the design for drainage purposes on
the base soil that has the largest D15 size.
 Step 2: Proceed to step 4 if the base soil contains no gravel (material
larger than No. 4 sieve).
 Step 3: Prepare adjusted gradation curves for base soils that have
particles larger than the No. 4 (4.75 mm) sieve.
 Obtain a correction factor by dividing 100 by the percent passing the No.
4 (4.75 mm) sieve.
 Multiply the percentage passing each sieve size of the base soil smaller
than No. 4 (4.75 mm) sieve by the correction factor determined above.
 Plot these adjusted percentages to obtain a new gradation curve.
 Use the adjusted curve to determine the percentage passing the No. 200
(0.075 mm) sieve in step 4.
22

11
 Filter Design Steps – Step 4

 Step 4: Place the base soil in a category determined by the percent

passing the No. 200 (0.075 mm) sieve from the regraded gradation
curve data according to table 26–1.

23

Filter Design Steps – Step 4

 Step 4: Place the base soil in a category determined by the percent
passing the No. 200 (0.075 mm) sieve from the regraded gradation
curve data according to table 26–1.

24

12
 Filter Design Steps – Step 5
 Step 5: To satisfy filtration
requirements, determine the
maximum allowable D15 size for
the filter in accordance with the
table 26–2.
 If desired, the maximum D15 may
be adjusted for certain noncritical
uses of filters where significant
hydraulic gradients are not
predicted, such as bedding
beneath riprap and concrete slabs.
For fine clay base soil that has d85
sizes between 0.03 and 0.1 mm, a
maximum D15 of ≤ 0.5 mm is still
conservative. For fine grained silt
that has low sand content, plotting
below the "A" line, a maximum
D15 of 0.3 mm may be used.
25

Filter Design Steps – Step 6

 Step 6: If permeability is a requirement, determine the
minimum allowable D15 in accordance with table 26–3.
 Note: The permeability requirement is determined from the d15 size of the base
soil gradation before regrading.

26

13
 Filter Design Steps
 Step 7: The width of the allowable filter design band must be kept
relatively narrow to prevent the use of possibly gap-graded filters.
Adjust the maximum and minimum D15 sizes for the filter band
determined in steps 5 and 6 so that the ratio is 5 or less at any given
percentage passing of 60 or less. Criteria are summarized in table
26–4. This step is required to avoid the use of gap-graded filters

27

Filter Design Steps – Step 7

 Step 7 Continued:
 This step is required to avoid the use of gap-graded filters. The use of a
broad range of particle sizes to specify a filter gradation could result in
allowing the use of gap-graded (skip-graded) materials. These materials
have a grain size distribution curve with sharp breaks or other undesirable
characteristics.
 Materials that have a broad range of particle sizes may also be susceptible
to segregation during placement.
 The requirements of step 9 should prevent segregation, but other steps are
needed to eliminate the use of any gap-graded filters.
 Gap-graded materials generally can be recognized by simply looking at their
grain size distribution curve.
 However, for specification purposes, more precise controls are needed. In
designing an acceptable filter band using the preliminary control points
obtained in steps 1 through 6, the following additional requirements should
be followed to decrease the probability of using a gap-graded filter.

28

14
 Filter Design Steps – Step 7
 Step 7 contiued:
 First, calculate the ratio of the maximum D15 to the minimum
D15 sizes determined in steps 5 and 6. If this ratio is greater
than 5, adjust the values of these control points so that the
ratio of the maximum D15 to the minimum D15 is no greater
than 5. If the ratio is 5 or less, no adjustments are necessary.
Label the maximum D15 size as Control point 1 and the
minimum D15 size as Control point 2. Proceed to step 8.
 The decision on where to locate the final D15 sizes within the
range established with previous criteria should be based on one
of the following considerations:
1. Locate the design filter band at the maximum D15 side of the range if
the filter will be required to transmit large quantities of water (serve as a
drain as well as a filter). With the maximum D15 size as the control point,
establish a new minimum D15 size by dividing the maximum D15 size by 5,
and locate a new minimum D15 size. Label the maximum D15 size Control
point 1 and the minimum D15 size Control point 2.

29

Filter Design Steps – Step 7

 Step 7 continued:
2. Locate the band at the minimum D15 side of the range if it is probable
there are finer base materials than those sampled and filtering is the most
important function of the zone. With the minimum D15 size as the control
point, establish a new maximum D15 size by multiplying the minimum D15
size by 5, and locate a new maximum D15 size. Label the maximum D15
size Control point 1 and the minimum D15 size Control point 2.
3. The most important consideration may be to locate the maximum and
minimum D15 sizes, within the acceptable range of sizes determined in
steps 5 and 6, so that a standard gradation available from a commercial
source or other gradations from a natural source near the site would fall
within the limits. Locate a new maximum D15 and minimum D15 within the
permissible range to coincide with the readily available material. Ensure
that the ratio of these sizes is 5 or less. Label the maximum D15 size
Control point 1 and the minimum D15 size Control point 2.

30

15
 Filter Design Steps – Step 8
 Step 8: The designed filter band must not have an extremely
broad range of particle sizes to prevent the use of possibly gap-
graded filters. Adjust the limits of the design filter band so that
the coarse and fine sides have a coefficient of uniformity of 6 or
less. The width of the filter band should be such that the ratio
of maximum to minimum diameters is less than or equal to 5
for all percent passing values of 60 or less.
 Other filter design criteria in step 8:
 To prevent gap-graded filters—Both sides of the design filter
band will have a coefficient of uniformity, defined as:
CU = D60/D10 ≤ 6
 Initial design filter bands by this step will have CU values of 6.
For final design, filter bands may be adjusted to a steeper
configuration, with CU values less than 6, if needed. This is
acceptable so long as other filter and permeability criteria are
satisfied.

31

Filter Design Steps – Step 8

 Step 8:
 Calculate a maximum D10 value equal to the maximum D15 size
divided by 1.2. (This factor of 1.2 is based on the assumption that
the slope of the line connecting D15 and D10 should be on a
coefficient of uniformity of about 6.) Calculate the maximum
permissible D60 size by multiplying the maximum D10 value by 6.
Label this Control point 3.
 Determine the minimum allowable D60 size for the fine side of the
band by dividing the determined maximum D60 size by 5. Label this
Control point 4.

32

16
 Filter Design Steps – Step 9
 Step 9:
 Determine the minimum D5 and maximum D100 sizes of the
filter according to table 26–5. Label as Control points 5 and 6,
respectively

33

Filter Design Steps – Step 10

 Step 10:
 To minimize segregation during construction, the relationship
between the maximum D90 and the minimum D10 of the filter is
important. Calculate a preliminary minimum D10 size by dividing the
minimum D15 size by 1.2. (This factor of 1.2 is based on the
assumption that the slope of the line connecting D15 and D10 should
be on a coefficient of uniformity of about 6.) Determine the maximum
D90 using table 26–6. Label this as Control point 7.
 Sand filters that have a D90 less than about 20 mm generally do not
require special adjustments for the broadness of the filter band. For
coarser filters and gravel zones that serve both as filters and drains,
the ratio of D90/D10 should decrease rapidly with increasing D10
sizes.

34

17
 Filter Design Steps – Step 11
 Step 11: Connect Control points 4, 2, and 5 to form a partial design
for the fine side of the filter band. Connect Control points 6, 7, 3, and
1 to form a design for the coarse side of the filter band. This results
in a preliminary design for a filter band. Complete the design by
extrapolating the coarse and fine curves to the 100 percentfiner
value. For purposes of writing specifications, select appropriate sieves
and corresponding percent finer values that best reconstruct the
design band and tabulate the values.

35

Filter Design Steps – Step 12

 Step 12: Design filters adjacent to perforated pipe to have a
D85 size no smaller than shown in table 26–7. For critical
structure drains where rapid gradient reversal (surging) is probable, it
is recommended that the D15 size of the material surrounding the
pipe be no smaller than the perforation size.
 Additional design considerations: Note that these steps provide
a filter band design that is as well graded as possible and still meets
criteria. This generally provides the most desirable filter
characteristics.
 However, in some cases a more poorly graded filter band may be
preferable; for example, if more readily available standard gradations
are needed or where onsite filters are used for economy.
 The design filter band obtained in steps 1 through 12 may be
adjusted to a steeper configuration in such cases. The width of the
filter band should be maintained so that the ratio of the maximum
diameters to the minimum diameters at a given percent finer is no
greater than 5 below the 60 percent finer value.

36

18
 Filter Design Steps – Step 12
 Only the portion of the design filter band above the previously
established minimum and maximum D15 sizes should be adjusted.
The design band may be adjusted so that the coefficients of
uniformity of both the coarse and fine sides of the design band are
less than 6, but not less than 2, to prevent use of very poorly graded
filters.
 Note that the requirements for coefficient of uniformity apply only to
the coarse and fine limits of the design filter band. It is possible that
an individual, acceptable filter whose gradation plots completely
within the specified limits could have a coefficient of uniformity
greater than 6 and still be perfectly acceptable.
 The design steps of this procedure will prevent acceptance of gap-
graded filters, which is the main concern associated with filters
having a high coefficient of uniformity, and it is not necessary to
closely examine the coefficient of uniformity of a particular filter as
long as it plots within the design filter band.

37

Filter Design Example

 Given the following grain size distribution for the base
soil, design a filter for it.

38

19
 Filter Design Example

39

Filter Design Example

 Step 2: Proceed to step 4 if the base soil contains no gravel
(material larger than the No. 4 sieve).
 The example base soil has 100 percent finer than the No. 4 sieve, and
the grain size distribution curve does not need to be regraded.
 Proceed to step 4.
 Step 3: Not applicable because the base soil contains no
particles larger than the No. 4 sieve
 Step 4: Place the base soil in a category determined by the
percent passing the No. 200 (0.075 mm) sieve from the
regraded gradation curve data according to table 26–1.
 The example soil has 90 percent finer than the No. 200 sieve. From
table 26–1, the soil is in category 1.

40

20
 Filter Design Example
 Step 5: To satisfy filtration requirements, determine the
maximum allowable D15 size for the filter according to table
26–2.
 The filtering criteria for base soil category 1 is (table 26–2): The
maximum D15 of the filter will be less than or equal to 9 times the d85
of the base soil, but not less than 0.2 mm
 The d85 size of the base soil is 0.06 mm. Thus, the maximum D15 of
the filter is ≤ 9 x 0.06 = 0.54 mm (not < 0.2 mm)
 This is labeled as Maximum D15 in figure 26–1.

41

Filter Design Example

 Step 6: Determine the minimum allowable D15 according to table
26–3 (permeability requirement). Note: The permeability
requirement is determined from the d15 size of the base soil
gradation before regrading.
 The permeability criterion for all categories of base soils is that the filter will
have a minimum D15 of no less than 4 times the d15 of the base soil (before
any regrading of the base soil), but will not be less than 0.1 mm in any case.
 The example base soil does not have a meaningful d15 size. The data show
that the base soil has 32 percent finer than 0.002 mm, the smallest commonly
determined particle size. Therefore, use the default value of 0.1 mm for the
minimum D15 of the filter.
 This value is the preliminary value for minimum D15. Proceed to step 7 for any
needed adjustments.

42

21
 Filter Design Example
 Step 7: The width of the allowable filter design band must be kept
relatively narrow to prevent the use of possibly gap-graded filters.
Adjust the maximum and minimum D15 sizes for the filter band
determined in previous steps 5 and 6 so that the ratio is 5 or less, at
any given percent passing of 60 or less.
 Adjustments may be required based on the following considerations.
 For the example, the ratio of the maximum D15 to the minimum D15 sizes is
equal to 0.54 / 0.1 = 5.4. Because the value is slightly greater than 5, a slight
adjustment is needed in this step.
 The minimum D15 is the control because filtering is stated as the most
important purpose. Label this as Control point 2.
 Determine an adjusted maximum D15 size for the final design filter band as
equal to the minimum D15 size, 0.10 x 5 = 0.50 mm. This is the final Control
point 1 labeled in figure 26–1.
 Go to step 8.

43

Filter Design Example

 Step 8: The designed filter band must not have an extremely broad range
of particle sizes to prevent using possibly gap-graded filters. Adjust the
limits of the design filter band so that coarse and fine sides of the filter
band have a coefficient of uniformity of 6 or less. Width of the filter band
should be such that the ratio of maximum to minimum diameters is less
than or equal to 5 for all percent passing values of 60 or less.
 For the example, calculate a value for maximum D10 by dividing the
maximum D15 size of 0.5 mm (determined in step 7) by 1.2 = 0.42 mm.
Determine the value for the maximum D60 size by multiplying the value of
D10 by 6 = 0.42 x 6 = 2.5 mm. Label this as Control point 3.
 Determine the minimum allowable D60 size for the fine side of the band by
dividing the determined maximum D60 size by 5:
 D60/ 5 = 2.5 / 5 = 0.50
 Label this Control point 4.

44

22
 Filter Design Example
 Step 9: Determine the minimum D5 and maximum D100 sizes of the filter
according to table 26–5. This table shows that filters must have a D5
greater than or equal to 0.075 mm, equal to the No. 200 sieve size. Label
this value as Control point 5 in figure 26–1. It also shows that filters must
have a D100 of less than or equal to 3 inches. Label this value as Control
point 6 in figure 26–1.
 Step 10: To minimize segregation during construction, the relationship
between the maximum D90 and the minimum D10 of the filter is important.
Calculate a preliminary minimum D10 size by dividing the minimum D15
size by 1.2. Determine the maximum D90 using table 26–6. Label this as
Control point 7. Calculate the minimum D10 size of the preliminary filter
band as equal to the minimum D15 value of 0.1 mm (obtained in step 6)
divided by 1.2:
 0.10 / 1.2 = 0.083 mm
 Table 26–6 lists maximum D90 sizes for filters for a range of D10 sizes.
Because the D10 value is less than 0.5 mm, the maximum D90 size is 20
mm. Label this value as Control point 7 in figure 26–1.

45

Filter Design Example

 Step 11: Connect Control points 4, 2, and 5 to form a partial design for
the fine side of the filter band. Connect Control points 6, 7, 3, and 1 to
form a partial design for the coarse side of the filter band.
 Complete the design of the filter band by extrapolating the coarse and fine
curves to the 100 percent finer value. For purposes of writing
specifications, select appropriate sieves and corresponding percent finer
values that best reconstruct the design band and tabulate the values.
 Refer to figure 26–1 for an illustration of the complete filter design. Note
that adjustments have been made in straight line portions of the design
band to intercept even values for percent passing at standard sieve sizes
and to prevent the use of very broadly graded filters.
 The final design specified gradation is shown in table 26–8.

46

23
 Filter Design Example
 Step 12: Design filters adjacent to perforated pipe to have a D85 size no
smaller than the perforation size. For critical structure drains where rapid
gradient reversal (surging) is probable, it is recommended that the D15 size
of the material surrounding the pipe be no smaller than the perforation
size.
 For this example, the filter will not be used around a perforated collector
pipe, so step 12 is not applicable.

47

24