You are on page 1of 19

Geotech Geol Eng

https://doi.org/10.1007/s10706-018-0453-7

STATE-OF-THE-ART REVIEW

A Critical Review on Filter Design Criteria for Dispersive


Base Soils
Amir Hossein Vakili . Mohamad Razip bin Selamat . Parsa Mohajeri .
Hossein Moayedi

Received: 19 January 2016 / Accepted: 8 January 2018


Ó Springer International Publishing AG, part of Springer Nature 2018

Abstract Dispersive soils have become common dispersive soils are discussed and significant findings
materials for the construction industry. Highly sus- from previous studies on protective filters are sum-
ceptible to internal erosion and piping, dispersive soils marized. It is worthy to note that the current review
must only be used with specific engineering measure considers both, the conventional, rather empirical
in order to avoid failures that were often catastrophic. filter design criteria based on particle sizes and the
In an earth dam, clayey soils are used for the core and current, quite theoretical state-of-the-art filter design
sandy materials are used for the filter to retain the criteria based on constriction sizes, with discussion
eroded core soils and prevent their migration. In the given on the advantages and disadvantages of both.
absence of first-rate core material, dispersive soils The information provided by this review should be
have been used instead. This paper provides a review handy for the study, design, construction, and opera-
of the current knowledge and experiences regarding tion of related geotechnical and geo-environmental
filtration of core soils, particularly the dispersive ones. projects.
The engineering problems associated with the use of
Keywords Dispersive base soils  Earth dams 
Erosion  Piping  Filter design criteria
A. H. Vakili (&)
Department of Civil Engineering, Faculty of Engineering,
Zand Institute of Higher Education, Shiraz, Iran
e-mail: amirhoseinvakili@gmail.com
List of symbols
PSD Particle size distribution
A. H. Vakili Dif Diameter of filter particle at which i%
Young Researcher and Elite Club, Estahban Branch, of the filter material is smaller than in
Islamic Azad University, Estahban, Iran
mm
M. R. Selamat  P. Mohajeri di Diameter of base soil particle at which
School of Civil Engineering, Universiti Sains Malaysia, i% of the material is smaller than in mm
14300 Nibong Tebal, Penang, Malaysia i The percentage passing diameter di
e-mail: cemrs@eng.usm.my
d85regard d85 size of base soil, in mm, recalculated
P. Mohajeri by excluding the fraction retained by
e-mail: parsa.mohajeri@yahoo.com
the 4.75 mm sieve
H. Moayedi d85fine d85 size of the fine portion of base soil
Department of Civil Engineering, Kermanshah University materials
of Technology, Kermanshah, Iran
e-mail: h.moayedi@kut.ac.ir

123
Geotech Geol Eng

d15coarse d15 size of the coarse portion of base soil Due to shortage in first-rate clayey soils, the
materials dispersive ones often made their way to construction
d85reduced The d85 size, in mm, calculated using sites. Of particular concern is when these soils are used
reduced particle size distribution curve for the core of an earth dam where serious geotech-
d85m d85 size of base soil, in mm, recalculated nical problems could take place such as the internal
by excluding the fraction retained by tunneling erosion or piping. When the soils are used
the 0.075 mm sieve for an embankment, problems such as suffusion that
dR Size of the smallest particles of the base leads to foundation failure and general subsidence
materials in lm could also occur if the necessary technical procedures
PI Plasticity index have not been carried out (Indraratna et al. 2012a, b;
pp% Percent passing through the 0.075 mm Vinod et al. 2010; Paige-Green 2008; Burns and
0.075 mm size Ghataora 2007; Richards and Reddy 2007).
Fines pp% 0.075 mm Protective filter layers are used to minimize the
content occurrence of piping failure in embankment dams
F Fine content (Indraratna and Locke 1999). In a core-filter system,
NEF No erosion filter the potentially erodible core particles are retained
kf Permeability coefficient of filter among the filter materials (Indraratna and Raut 2006).
materials in cm/s However, designing a suitable filter for the given core
kbase soil Permeability coefficient of base soil requires knowledge of the interaction between the
materials in cm/s base clayey soil making up the core and the corre-
Dc35 Constriction size, in mm, whereby 35% sponding sandy material making up the filter. Thus
of the filter constriction are finer than complex analyses have been employed in assessing
Dc95 Constriction size, in mm, whereby 95% the behavior of the appointed granular filter materials
of the filter constriction are finer than for the given base soils due to the conflicting
Dm Mean constriction size in mm requirements expected of the filter zone (Lone et al.
c
d85 The d85 size, in mm, calculated using 2005).
modified particle size distribution curve The main matter discussed in this paper is on the
by neglecting the base soil particles development of various filter design criteria for
larger than Dc95 controlling base soils with dispersivity problems.
The paper mainly intends to comprehensively review
published literature on the required filtration of
dispersive soils in the pursuit of enhancing the design
1 Introduction criteria for the filters. The wider aim is to mitigate the
destructive circumstances surrounding the use of
Deposits of dispersive soils can be found in various dispersive soils in geotechnical engineering.
climates and locations, such as in Iran, South Africa,
Thailand, India, United States, Australia, and many
other places (Goodarzi and Salimi 2015; Vakili et al. 2 Internal Erosion and Piping Processes
2013a, b, c; Abbasi and Nazifi 2013; Umesh et al.
2011; Ouhadi and Goodarzi 2003, 2006; Sherard et al. Internal erosion and piping can be differentiated in the
1976a, b). Dispersive soils are clayey soils that rapidly sense that the latter is a large scale and more serious
deflocculates in waters with low salt concentration, version of the former. Internal erosion can lead to the
even when the waters are still (Vakili et al. 2017; piping phenomenon and consequently the failure of
Paige-Green 2008). These soils are significantly rich levees and dams. There are four dominant mechanisms
in sodium ions and as a consequence the internal by which a water-retaining structure or its foundation
structures have become quite unstable (Umesh et al. can get affected by the internal erosion hazard:
2011; Zorluer et al. 2010). suffusion, backward erosion, contact erosion, and
concentrated leak erosion. These different means
initiate the internal erosion, although by different

123
Geotech Geol Eng

procedures, that lead to the eventual breach of were the main reasons for 48% of all dam failures with
structural safety (Fell et al. 2003). Internal erosion approximately 42% of these occurred during the early
and piping phenomena are widespread in many or first fillings. Gutierrez et al. (2003) also have
different localities, climates, and geological settings attributed piping process to the presence of dispersive
throughout the world and the mechanisms involved clays in the cores of earth dams.
can be quite complex (Romero Diaz et al. 2007; Foster and Fell (1999, 2000) have identified four
Hagerty 1991). The catastrophic damage and irre- sequential phases involving erosion that lead to the
versible failure of such events could be traced back to failure of an earth dam, which are as follows:
the dispersive behavior of the construction soils (Savas
(a) Initiation of erosion: in this stage, erosion is
2015; Richards and Reddy 2007; Romero Diaz et al.
initiated by the concentrated leaks running
2007; Gutierrez et al. 2003).
through cracks in the core. Cracks in cores
As given in Fig. 1, internal erosion and piping have
and embankments are formed because of sev-
been the main causes of failure in earth dams, while
eral factors, such as shrinkages, differential
the other major cause being the overtopping (ICOLD
settlements, seismic movements, hydraulic
2013; Regazzoni and Marot 2011; Flores-Berrones
fractures, construction deficiencies, and tension
et al. 2010; Fell and Fry 2007; Minguez et al. 2006;
stresses (Indraratna et al. 1996; Indraratna and
Foster et al. 1998). Foster et al. (2000b) stated that 9
Vafai 1997; Arulanandan and Perry 1983). The
out of 51 piping failures that occurred in the embank-
likelihood of having an initiation of erosion is
ments of large dams that they studied have involved
increased with increasing presence of dispersive
dispersive clays that were used as the base soils. Foster
clays and erodible silts (Gutierrez et al. 2003;
et al. (2000a) also said that piping and internal erosion
Indraratna and Vafai 1997).

Fig. 1 Possible locations of initiation of internal erosion desiccation on top of core; (5) embankment to foundation; (6)
(ICOLD 2013). Note (1) spillway wall interface; (2) adjacent foundation (if the foundation is soil or erodible rock); (7)
to conduit; (3) crack associated with steep abutment profile; (4) embankment through poorly compacted layer and or crack

123
Geotech Geol Eng

(b) Continuation of erosion: this phase could pro- 3 Engineering Properties of Dispersive Clays
long in case of an inappropriate filter material
being used. The continuation of the erosion The dispersive behavior of soil depends on its clay
phase depends on the core–filter zone condi- mineralogy, sodium absorption ratio, and the total
tions. According to Foster and Fell (2001), the dissolved salts concentration in the pore water of the
continuing erosion could also be influenced by soil or in the eroding water (Goodarzi and Salimi
the particle size distributions (PSDs) of the base 2015; Abbasi and Nazifi 2013; Ouhadi et al. 2012;
soil and the filter material. Penner and Lagaly 2001; Sherard and Decker 1977).
(c) Progression of piping: in this phase, the con- Dispersive clay minerals have significant amounts
centrated leak is progressively enlarged, and the of dissolved sodium (Zorluer et al. 2010; Arulanandan
pipe is gradually formed. and Heinzen 1977) and the dispersion and cracking
(d) Formation of breach mechanism: the breach potentials enhance in a Na-rich environment (Gutier-
mechanism and failure are the ultimate culmi- rez et al. 2003). The exchangeable sodium percentage
nation of piping and the related progression. A (ESP) is the most important chemical factor affecting
piping failure could rapidly develop into failure the dispersive behavior of clays (Premkumar et al.
even for as short as 6–12 h after the concen- 2015; Zorluer et al. 2010; Elges 1985). In the general
trated leaks first appear (Fell et al. 2003; Foster literature, soils with ESP values of more than 15 have
et al. 2000a, b; Sherard et al. 1972). The failure been called dispersive and were classified as highly
could be in the form of downstream slope erodible (Mcintyre 1979). However, in the Australian
collapse and crest settlement (Fell et al. 2003). literature, any soil with ESP greater than 6 is deemed
susceptible to dispersion and thus is known as sodic
The scenarios described above are given in Fig. 2. soil (Premkumar et al. 2015; Raine and Loch 2003;
For a new dam, the potential development of Isbell 2002).
internal erosion and piping can be avoided through a Clays containing sodium montmorillonite have the
good design and close supervision during construction highest dispersivity when compared with other soils
especially of the core, and the provision of an containing kaolinite or halloysite (Fell et al. 2010;
appropriate filter to intercept the formation of any Nevels Jr 1993). Dispersive soils were found to have
concentrated leak (Benahmed and Bonelli 2012; Fell thicker diffused double layers relative to non-disper-
and Fry 2007). Thus suitable filters are designed and sive ones thus describing the montmorillonite miner-
emplaced to control internal erosions and to minimize als makeup. Under saturated condition, for dispersive
the likelihood of piping failures (Indraratna and Locke soils, the repulsive forces between particles are more
1999). The appointment of an appropriate filter in than the attractive forces, causing separation and
protecting the core has become a critical aspect of suspension of the particles (Premkumar et al. 2015;
design because insufficient filters were found to be the Umesh et al. 2011; Paige-Green 2008; Burns and
main cause of 26% of all piping failures (Richards and Ghataora 2007; Mitchell 1993; Nevels Jr 1993;
Reddy 2007). Arulanandan and Heinzen 1977).

Fig. 2 Sequential phases of erosion that lead to piping failure in an earth dam (Wan and Fell 2004)

123
Geotech Geol Eng

4 Laboratory Test Procedures Soil permeability and soil porosity decrease with
increasing compaction. Thus with increasing com-
Standardized laboratory procedures are available for paction, the likelihood of having salts leaching from
the assessment of soil dispersivity. Dispersive soils are the soil matrix decreases. With increasing compaction,
recognizable from the results of Emerson crumb tests, water movement within the soil matrix decreases,
double hydrometer tests, pinhole tests, and the various leading to reduction of dispersivity related hazard
chemical tests (Premkumar et al. 2015; Umesh et al. (Raine and Loch 2003).
2011; Burns and Ghataora 2007). The dispersivity of soils can be reduced by mixing
The natural water content of the specimens in a test dispersive soils with chemical additives such as lime,
procedure should be maintained because the charac- cement, and pozzolan (Vakili et al. 2013a, 2015a,
teristics of dispersive soils may change due to the 2017; Savas 2015; Indraratna et al. 1991). However,
drying process (Paige-Green 2008; Knodel 1991). the dispersivity of soils could also be controlled by
Dispersivity verification tests should be carried out well-designed and well-constructed filter structures
using distilled water because the results could be out of select materials which are positioned either
different if other types of water are used instead, such downstream or both upstream and downstream of the
as tap or river water (Vakili et al. 2016). Using water core. Nevertheless this paper dwells only on the
with a high concentration of dissolved salts would protection of the dispersive core against internal
cause misleading results because dispersivity could erosion by having a downstream filter zone.
significantly be reduced by the increasing dissolved
salts in the pore water (Paige-Green 2008). In the 5.1 Design of Suitable Filters
dispersivity identification tests, dispersive soils could
be wrongly identified as non-dispersive if the water Probable concentrated leaks within the impervious
used, in terms of its TDS, is changed from distilled core of dams can be successfully and effectively
water to any water with high dissolved salt concen- sealed by designing an appropriate filter zone (Foster
tration, such as river water (Fernando 2010). and Fell 2001; Khor and Woo 1989; Sherard 1985).
The downstream filter zone is known as an essential
defensive part of a dam, whether the core is con-
5 Precautions and Considerations structed out of dispersive or normal clay soils.
Therefore, preventing dam failure by providing ade-
Attentive supervision should be paid during construc- quate filter materials is possible, even if the core is
tion of a geotechnical structure, particularly if the constructed using erodible, dispersive clay (Arulanan-
project is involved with dispersive soils. Dispersive dan and Perry 1983).
soil should be compacted well, especially around rigid A filter must be sufficiently fine to perform the
abutments and conduits as poorly compacted ones are retention function and control the migration of
prone to rapid erosion, regardless of the types of soil impervious core particles caused by internal erosion.
(Fell et al. 2003). At the same time, the filter must be sufficiently coarse
To reduce the risk of piping, dispersive soils should and having adequate permeability to drain seepage
be compacted at moisture content between optimum water (Indraratna et al. 2004; Indraratna and Locke
and 2% above optimum to achieve a dry unit weight of 1999; Arulanandan and Perry 1983).
at least 98% of the maximum dry unit weight by
standard Proctor method (Bell and Bruyn 1997; Elges 5.2 Concepts in Filter Designing
1985). Such measures are also recommended to avoid
formation of cracks due to brittleness (Fell et al. 2010). Designing a suitable filter is a complex process. The
To achieve sufficient compaction in the field, the use interaction between filter and base soil could be
of sheep foot roller is recommended. The compaction influenced by geometric factors, physical factors,
requirement for dispersive soils is normally greater chemical factors, and hydraulic factors among others
than those of other soils (Mcdonald et al. 1981; (Indraratna et al. 2008; Indraratna and Locke 1999).
Sorensen 1995). Relative PSD of filter material and base soil, base soil

123
Geotech Geol Eng

cohesion, base soil dispersion, pore water and eroding can perform its expected responsibility in terms of
water properties, and pressure gradient of seepage sealing the concentrated leaks of the core. Previous
water are all considered in a filter design (Indraratna studies have put D15f as the acceptable specification
and Locke 1999). However, to simplify a design, that defines the void size (Foster 1999; Honjo and
certain aspects may have been ignored in the modeling Veneziano 1989; Sherard et al. 1984a). However, the
of filter behavior. With the consideration of the above constriction size of a performing filter seems to be
factors and circumstances, a filter is usually simply governed by filter particles in the range of D5 to D15
designed in terms of the PSD in fulfilling the required (Indraratna et al. 2008; Kenney et al. 1985). Never-
criterion. theless, the controlling constriction size of Dc35 and
Filter retention criteria are usually expressed in two the self-filtering constriction size of Dc95 have been
forms—the geometric grading criterion and the taken as filter representative sizes in the new method
hydraulic criterion. The design of a filter based on (Raut and Indraratna 2008; Indraratna et al. 2007).
geometric grading criterion basically involves PSDs of For the base soil, the acceptable representative is
the base soil and the filter, and is presented in the form the d85 (Indraratna et al. 2008; Kwang 1990; Vaughan
of filter-base soil grain size ratios (Indraratna et al. and Soares 1982). Several studies have shown that
2008; Locke et al. 2001; Indraratna and Locke 1999; considerable particle loss takes place when the filter
Indraratna and Vafai 1997; Reddi and Bonala 1997). opening size is greater than d90 while very little
In response to the usage of geometric grading particle loss occurs when the filter opening size is less
criteria, some researchers such as Vaughan and Soares than d80. This phenomenon is the reason for d85 being
(1982) and Delgado et al. (2006) stated that these used as the base soil representative in the conventional
grading criteria alone cannot adequately express the filter design (Indraratna and Locke 1999; Honjo and
retention capacity of a filter material. Instead, the Veneziano 1989). The experimental results obtained
behavior of the filter material could only be assessed by Sherard et al. (1984a, b) and the statistical analyses
more truly through the hydraulic criteria by which the performed by Honjo and Veneziano (1989) showed
main concern would be filter permeability and not that the grain size ratios involving other representa-
particle sizes of the filter. Thus in this regard, the tives, i.e., D15f/d15 and D50f/d50, must not be used in
general relationships between filter permeability and filter design because these ratios do not show any
PSD of base materials were created (Delgado and suitable correlation with filter performance. These
Poyatos 2008; Indraratna et al. 1996). findings have also been corroborated by Indraratna and
However, in the state of the art filter design criteria, Vafai (1997). However, not all researchers have
the constriction size of filter materials are used instead agreed to the stereotype of appointing a filter criterion
of either the PSD or the permeability of the filter based on a given d85 of a base soil (Raut and Indraratna
material. In other words, the constriction size criterion 2008; Indraratna et al. 2007; Khor and Woo 1989). To
would already incorporates the effects of filter PSD, cite an example, the current design practice by NRCS
filter compaction, filter grading, filter surface area, (1994) recommends the use of d85 of base soil, but
filter porosity, and uniformity coefficient of the filter recalculated it by excluding the coarse fraction
material, thus most limitations associated with the retained by the 4.75 mm sieve size, termed as the
conventional design criteria could be removed by this regarded d85 size. By using the same design practice,
new design method (Raut and Indraratna 2008; Lafleur (1984) has introduced some filters with
Indraratna et al. 2008). Thus, the conventional way retention ratios even lower than 4, i.e. D15f/(regarded
of relating the PSD’s of filter and base soil materials or d85) \ 4, for the broadly graded base soils that had
between filter permeability and base soil PSD is now become ineffective. Hence, a modification of the
changed to the new way of relating the filter constric- retention ratio or an adjustment of the PSD of base soil
tion size distribution (CSD) to base soil PSD which is materials has been called for in cases of the broadly or
the more inherently correct approach to designing the well graded base materials. Thus for the broadly
filter-base system. graded base soils, as the regarded d85 would not be as
In the most conventional filter design criteria, the representative as for normal ones, it has also been
filter size is represented by D15f where voids in the considered unsuitable for use in filter designing.
filter must be adequately small until the filter system Researchers then suggested an even lower size

123
Geotech Geol Eng

because the regarded d85 size is still larger than the kf [ 20 kbasesoil ð1Þ
actual self-filtering size (Raut and Indraratna 2008;
Locke and Indraratna 2002; Khor and Woo 1989). D15f [ 4d15 ð2Þ
Accordingly, the filter materials would be effective
Furthermore, Vaughan and Soares (1982) proposed
if at least 15% of the base soil particles are retained
the ‘‘perfect filter’’ approach on the basis of the
(Indraratna et al. 2007). But if the filter materials of the
relationship between the floc sizes of the base soil and
broadly graded base soils are designed based on actual
the permeability of the filter material. Here, filter
d85, the confidence would be lesser. Even if the filter
effectiveness is related to filter permeability and size
materials are designed based on regarded d85, the
of flocculated clay particles (Indraratna et al. 1996;
confidence would still be insufficient.
Vaughan 1978). Vaughan and Soares (1982) held to
the notion that the behavior of the filter materials
5.3 Widely Accepted Filter Design Criteria
would be more truly assessed through hydraulic
criteria by which the main concern would be filter
Some filter design criteria as introduced by researchers
permeability and not the particle sizes of the filters and
such as Sherard and Dunnigan (1989), Foster and Fell
thus in the perfect filter approach the filter materials
(2001), Raut and Indraratna (2008), Shourijeh and
are designed for the required permeability and not for
Soroush (2009) and Vakili et al. (2015b) are given in
the required PSD. With evidences, Sherard (1983)
Table 1. The majority of these design criteria is based
pointed out that this approach might be too conserva-
on the relationships between the PSDs of base soil and
tive, particularly for general use, because the criterion
filter materials. However, there are also established
requires that the designed filter must be able to retain
criteria based on the filter permeability and PSD of
the smallest clay size. In addition, in the case of base
base soil.
soil made of fine-grained clay, laboratory studies have
Terzaghi (1926) suggested a retention ratio of D15f/
indicated that filters with D15f = 1 mm are able to seal
d85 B 4–5 for non-cohesive uniform base soil and
concentrated leaks, whereas filters with
filter materials. Although the veracity of criterion has
D15f = 0.5 mm would have a considerable factor of
been repeatedly verified for cohesionless base soils
safety and therefore could be considered conservative
and the factor of safety was deemed sufficient (Honjo
(Sherard 1984). Bases on the results of a large scale
and Veneziano 1989; Townsend et al. 1987; Bertram
filter test carried out by Hillis and Truscott (1982), the
1940), it could be too conservative in the case of
‘‘perfect filtering’’ method was judged to be realistic
cohesive base soil because the cohesion factor has not
by Truscott (1983). In addition, Vakili and Selamat
been considered in the initial derivation (Indraratna
(2014) have carried out tests on some dispersive base
and Vafai 1997). The retention ratio should therefore
soils and found that the perfect filter approach is
be increased when the base soil is having considerable
credible and is not unduly conservative for use in
cohesion (Indraratna and Vafai 1997; Sherard and
designing filters for such materials. Hence Vakili and
Dunnigan 1985). In fact Sherard and Dunnigan (1985)
Selamat (2014) recommended the perfect filter
and Indraratna and Vafai (1997) stated that a criterion
approach for base samples with a high degree of
must be intentionally derived to protect all clays,
dispersion instead of the other traditional indirect filter
including the dispersive ones. Note that no failure was
design criteria.
recorded in the filter testing performed on cohesive
Several dam failures were thought to have occurred
base materials when the criterion applied was D15f/
due to using filter design criteria without considering
d85 B 4 (Pinto 1989). It is worthy to note that
erodibility of the base soil making up the core. This
according to Bertram (1940), the D15f/d85 ratio is not
aspect was studied by Arulanandan and Perry (1983)
related to the shape of soil particles.
resulting in the consideration of base soil erodibility in
In addition, according to Terzaghi (1926), as
every filter designing. The same authors introduced a
reported by Fannin (2008), the following equations
classification system for base soils based on erodibility
on filter coarseness should be applied at the same time
related properties. Critical shear stress, which is the
to cater for the permeability function:
minimum shear stress required to initiate erosion, is
featured prominently in the system as described here:

123
Geotech Geol Eng

Table 1 Some widely accepted filter design criteria


Filter design criteria Type of base soil Criteria for non dispersive base soil Criteria for highly dispersive base
proposed by soil

Terzaghi (1926) Non cohesive uniform D15f B 4.0–5.0 d85 Not recommended
soil
Bertram (1940) Silt and fine sand D15f B 6.0 d85 Not recommended
Vaughan and Soares (1982) Clay base soils kf = 6.7 * 10-6 * d1.52
R Not recommended
Sherard and Dunnigan F C 85 D15f B 9 d85 D15f B 9 d85
(1989) F = 40–85 D15f B 0.7 mm D15f B 0.7 mm
Khor and Woo (1989) F = 40–85 D15f B 12 d85m Not recommended
Lafleur et al. (1993) Cohesive base soils D15f B 0.4 mm D15f B 0.2 mm
Indraratna et al. (1996) d85s = 50–60 lm D15f B 5–5.5 d85 Not recommended
d85s = 60–80 lm D15f B 4–5 d85
Indraratna et al. (1996) d85s = 40–100 lm kf = 6.3 * 10-4 * d1.25
85 Not recommended
Foster and Fell (2001) F C 85 D15f B 9 d85 D15f B 6.4 d85
F = 35–85 D15f B 0.7 mm D15f B 0.5 mm
Locke and Indraratna (2002) F C 85 D15f B 12 d85reduced Not recommended
F = 40–85, PI [ 10 D15f B 9 d85reduced
F = 40–85, PI \ 10 D15f B 4 d85reduced
Fell et al. (2005) F C 85 D15f B 9 d85 D15f B 6.0 d85
F = 35–85 D15f B 0.7 mm D15f B 0.5 mm
Delgado et al. (2006) F C 40 kf = 59.738 9 exp (- 0.102 9 F) Not recommended
Raut and Indraratna (2008) All base soils Dc35/d85 B 1 Dc35/d85 B 1
Shourijeh and Soroush F C 85 D15f B 9 d85 D15f B 7.5 d85
(2009) F = 80–85 D15f B minimum of (0.7 mm and D15f B 0.5 mm
6.4 d85)
F = 35–80 D15f B 0.7 mm D15f B 0.5 mm
Vakili et al. (2015b) F C 85 – D15f/d85 B 5.5
F = 40–85 – D15f B 0.28 mm
Vakili et al. (2015b) F C 85 D15f/d85 = 10.0 exp (- 0.009 D15f/d85 = 10.0 exp (- 0.009
D) C 4.0. D) C 4.0.
Vakili et al. (2015b) F C 85 D15f/d85 = 0.6037 (d85)-0.66 D15f/d85 = 0.6037 (d85)-0.66
Vakili et al. (2015b) F = 40–85, D \ 20% D15f B 0.6 mm D15f B 0.6 mm
F = 40–85, D15f B 0.5 mm D15f B 0.5 mm
20% B D \ 65%
F = 40–85, D C 65% D15f B 0.28 mm D15f B 0.28 mm
Vakili et al. (2015b) F C 85 Dc35/d85 B 1.25 Dc35/d85 B 1.25
F = 40–85 Dc35/d85 B 1.0 Dc35/d85 B 0.5

1—Soils with pinhole classification D1 or D2 or Emerson class 1 or 2 are known as highly dispersive soils; 2—F = fine content (%);
3—kf = permeability coefficient of filter materials; 4—dR = size of the smallest particles of the base materials in lm; 5—
PI = plasticity index; D = percent dispersion

1. Erodible base soils are defined as soils with erodibility property. Note that 1 dyne equals to
critical shear stress less than or equal to 4 dynes/ 10-5 N.
cm2. Dams constructed using these soils have 2. Base soils with moderate erodibility are defined as
experienced piping failure. Other than dispersive soils with critical shear stress between 4 and
soils, some non-dispersive soils have shown high 9 dynes/cm2.

123
Geotech Geol Eng

3. Erosion resistant base soils are defined as those satisfying the filter design criteria as proposed by
with critical shear stress of more than or equal to Sherard and Dunnigan (1989) would not necessarily
9 dynes/cm2. mean success in laboratory filter testing.
Khor and Woo (1989) studied protective filters
By and large, the filter design criteria could be used required to seal the concentrated leak through the dam
only in designing filter for protecting erosion resis- core constructed using tropical residual soils from
tance core. Filter testing should be performed for the Malaysia. The results from their investigations show
other aforementioned base soils. The capability of the that, in addition to the D15f size, the grading below
filter materials to prevent probable erosion through a 15% size also has significant influence on filter
dam core constructed using highly to moderately performance. Different performances were found
erodible materials must be separately inspected. from filter materials with similar D15f sizes and
One of the widely accepted filter design criteria was grading above the 15% size, but with different grading
that proposed by Sherard and Dunnigan (1989). As below the 15% size. Generally, the flow increased
given in Table 1, the criteria do not differentiate substantially with increasing coarseness of grains
between cases of dispersive and non-dispersive clays. below the D15f size. Thus, the permeability of the filter
Out of using only two dispersive samples, the authors materials, which is influenced by the fine portion of the
suggested that the dispersivity of the base soil does not grading, has remarkable effect on filter performance.
have significant influence on the no-erosion filter Furthermore, based on Khor’s and Woo’s (1989)
boundary (Sherard and Dunnigan 1985, 1989). The observation, a similar filter boundary was recorded for
adequacy of the criteria was later challenged by broadly graded base soils with similar fine content but
Lafleur et al. (1993) and Foster and Fell (2001) who with different sand or gravel contents. Therefore, the
put forth a finer no erosion filter boundary for the extent of the fine content of base soils has a significant
dispersive base soils. In agreement with latter effect on the behavior of base soil–filter systems. No
researchers, Vakili and Selamat (2014) stated that marked correlation has been determined between the
the dispersivity property of the base soils decreases the D15f size of filter materials and the d85 size of base soil
no erosion filter boundary and noted that although the materials. Thus, the d85 size is not an appropriate
criteria by Foster and Fell (2001) has a reasonable parameter to represent a base soil with broadly graded
factor of safety for soils with a moderate degree of characteristics in the design of a filter. Designing filter
dispersion, the widely used filter design criteria are not using d85 size derived from the actual gradation of
reliable in the case of base soils with a high degree of broadly graded base soils would result in a filter with a
dispersion. grading too coarse.
The adequacy of the filter design criteria as Material would be lost if the actual d85 size of the
proposed by Sherard and Dunnigan (1989), or the broadly graded base soil is used in filter designing
lack of it, became subject of further researches. (Lafleur et al. 1989). Thus, the actual or the regarded
According to Fell et al. (2005), the filter design d85 size is larger than the one required for formation of
criteria are applicable for base soils of Groups 1 and self-filtering in which the finer particles would be
Group 2. The criteria were further challenged when retained by the retained coarser particles (Raut and
Locke and Indraratna (2002) introduced some broadly Indraratna 2008; Locke et al. 2001). A stable self-
graded base soils that had finer no erosion filter filtration zone would be formed by an effective filter
boundaries in terms of D15f. Note that dispersive soils by retaining the eroded base soil particles within the
could have broad gradation. filter (Indraratna and Raut 2006).
Thus, certain laboratory investigations, such as by Knowing that the coarse portion of a base soil does
Locke and Indraratna (2002) and Soroush et al. (2006) not affect much on the filtration mechanism, splitting
have resulted in outcomes suggesting the inadequacy the PSD curve in designing appropriate filter for the
of the widely used filter design criteria. By performing broadly graded base soil has been recommended by
36 experiments, the widely used filter design criteria Sherard (1979). In this splitting method, the PSD
proposed by Sherard and Dunnigan (1989) were curve is divided into two separate portions: fine
evaluated by Vakili and Selamat (2014) to be only portion and coarse portion, as shown in Fig. 3. The
50% compatible with the NEF test results. Thus main difference between the various criteria is on the

123
Geotech Geol Eng

broadly graded base soils, Lafleur et al. (1989) and


Sherard and Dunnigan (1989) have proposed the
relatively fine downstream filter to allow self-filtering
circumstances to prevail through the retention of finer
particles by the coarser particles.
In the reduced PSD method, Locke and Indraratna
(2002) proposed the following filter criteria for base
soils of Groups 1 and 2:
(a) D15bdy/d85reduced B 12 for base soils of Group 1,
when PI [ 10 and d98/d85 [ 2
(b) D15bdy/d85reduced B 9 for base soils of Group 2,
when PI [ 10
Fig. 3 Coarser and finer frictions of PSD curve (Indraratna
et al. 2011) (c) D15bdy/d85reduced B 4 for base soils of Group 2,
when PI \ 10.
position of the separation point. The d15 of the coarse Here, instead of the regarded d85 size, the d85reduced
portion of the PSD is denoted as d15coarse, and the d85 is considered as the new representative base size for
of the fine portion of the PSD is denoted as d85fine. self-filtering circumstance and is calculated from the
Consequently, the modified PSD of base soil materials reduced gradation curve, the definition of which is
by ignoring all particles larger than d15coarse has described by Locke and Indraratna (2002). Basically,
become the basis of the filter design. Thus, base soil the reduced gradation curve is a redrawn gradation
particles coarser than d15coarse are excluded because created by disregarding the coarse portion of the
the particles are deemed as not affecting the filtration material, i.e., anything coarser than d15coarse is
mechanism (Indraratna and Raut 2006). d15coarse is the excluded in the new gradation. The remaining amount
base soil size by which base soil particles coarser than is all that is effectively considered of the material.
this size would not enter the filter or base soil particles The procedure of determining d15coarse is as
finer than this size would participate in the self- follows:
filtration mechanism by entering the self-filtration
zone (Raut and Indraratna 2008). The splitting or 1. Determine the gradation curve of the base soil.
modification of base soil PSD is strongly recom- The curve of Fig. 4 is given for illustration
mended in the industry particularly in the case of purposes and as an example of a general situation.
broadly or well-graded material (Li and Fannin 2008). 2. Determine a point on the curve which more or less
Investigations by Khor and Woo (1989) have marks the boundary between fine and coarse
determined that the d85m size is a proper representative materials. In Fig. 4, this point corresponds to n
of the base soil in a filter design criteria. The d85m size and Dn. The curve to the right of the boundary
is the d85determined from the new gradation curve represents the coarse portion of the soil, whereas
(modified PSD) using mostly particles finer than 75 the curve to the left represents the fine portion of
micron. Khor and Woo (1989) recommended the filter the soil.
criterion of D15/d85m B 12 for broadly graded base 3. The final position of the point is obtained by
soil with 40–85% particles finer than 0.075 mm. applying Eqs. 3–5, where n is the percentage of
Sherard and Dunnigan (1989) proposed the criterion of the overall material passing diameter dn. If
D15f B 0.7 mm for this group of soils. The study necessary, trial and error is employed to achieve
carried out by Khor and Woo (1989) uncovered some the largest diameter of d15coarse that agrees with
ambiguities in designing appropriate filter materials Eq. 5. An example of PSD curve of a given base
for broadly graded base soils. soil and its corresponding reduced PSD curve is
Locke and Indraratna (2002) presented the reduced shown in Fig. 5.
PSD method for designing sandy filters for broadly d15coarse ¼ dnþ0:15ð100nÞ ð3Þ
graded, fine base soils though not fully addressing the
dispersivity problems. Nevertheless, in the case of

123
Geotech Geol Eng

Fig. 4 Method to assess


self-filtration of a soil
(Locke and Indraratna 2002)

PSD

Percent passing (%)


n +0.15(100-n)
n
0.85n

d85fine dn d15coarse

Particle size (mm)

1. D15f/d85reduced \ 9 for dispersive soils of Group 1,


without significant silt content, when PI [ 10, and
when d98/d85 [ 2.
2. D15f/d85reduced \ 7 for dispersive soils of Group 2,
without significant silt content, and when PI [ 12.
3. D15f/d85reduced \ 2 for dispersive soils of Group 2,
with significant silt content, and when PI \ 12.
A D15f B 0.2 mm was recommended by Lafleur et al.
(1993) for dispersive cohesive soils. This recommen-
dation is not in agreement with the conclusion
Fig. 5 An example of PSD curve of a given base soil sample obtained by Sherard and Dunnigan (1985, 1989). Note
and its reduced PSD curve (Vakili and Selamat 2014) that soils with different dispersion degrees are con-
sidered just the same by Lafleur et al.’s (1993)
d85fine ¼ d0:85n ð4Þ approach while for the other cohesive soils, the
D15f B 0.4 mm was recommended. Sherard and Dun-
ðd15coarse =d85fine Þ  5 ð5Þ nigan (1989) suggested D15f B 9 d85 and D15f -
The study carried out by Vakili and Selamat (2014) B 0.7 mm for soil of Groups 1 and 2, respectively.
revealed that the reduced PSD method could be The design criteria as proposed by Lafleur et al. (1993)
considered credible, as long as the broadly graded base apparently have suggested finer filter materials in
materials do not have any considerable degree of comparison to Sherard’s and Dunnigan’s (1985, 1989)
dispersion. In other words, the reliability of the and therefore could be considered relatively conser-
reduced PSD method could become questionable if vative, particularly for soils with low dispersivity. No
the base soils involved are of high dispersivity type. differences exist between the recommended filters for
The applicability of the reduced PSD method was soils of Groups 1 and 2 in the design approach of
further studied by Vakili et al. (2015b). They reported Lafleur et al. (1993).
that the reduced PSD method were found to be A filtration mechanism for the cohesive lateritic
compatible with those coming from the NEF tests for base soil of Southeast Asia was tested with high
base soils with percent dispersion less than 40%. pressured slurry by Indraratna et al. (1996). From
Consequently, the following reduced PSD criteria experimental results, the following filter criteria were
were recommended when dealing with dispersive suggested:
soils:

123
Geotech Geol Eng

(a) D15f/d85 B 5–5.5 for d85 = 0.050–0.060 mm (2016) stated that the NEF test results carried out by
(b) D15f/d85 B 4–5 for d85 = 0.060–0.080 mm. Vakili and Selamat (2014) and Vakili et al. (2015b) on
dispersive soils were completely compatible with the
Following Indraratna et al.’s (1996) criteria would filter design criterion of Delgado et al. (2006). Thus
mean filter materials finer in comparison to Sherard’s the criterion was found to be credible enough for base
and Dunnigan’s (1989). In addition, Indraratna et al. soils collected from tropical origin, even for samples
(1996) pointed out that the reservoir water chemistry with high dispersivity.
has significant effects on the re-flocculation of Delgado et al. (2012) found that the D15f size of the
dispersive clay particles thus should be considered in required filter decreases with increasing water content
a filter design. Reddi and Bonala (1997) agreed that of the base soil. Therefore, clay zones with high
the accumulation of fine particles in the filter zone moisture content are more susceptible to internal
could be affected by the pore fluid compositions. erosion as compared to those with low moisture
Delgado Ramos and Locke (2000) ruled that content. As a consequence, the clay core must be
distilled water must be used in any NEF test procedure compacted at the optimum moisture content (Delgado
whenever the base soils are dispersive. Note that et al. 2012).
Vakili et al. (2015b) have also used distilled water in Delgado et al. (2012) also stated that no significant
their NEF testing programs. The main reason for this is relationship exists between the D15f size of the
that the quality of water, in terms of total dissolved required filter and the clay mineralogy of base
salt, could determine dispersivity characteristics of materials. However, base soil plasticity was found to
soils (Fernando 2010). Sherard et al. (1984a, b) have a significant effect on the D15f of the required
showed that fine filter materials—about 0.1 mm filter. Delgado et al. (2012) further found that the D15f
diameter—are required to protect non-dispersive base size of the required filter tends to decrease with
soils of similar PSD as in the case of dispersive soils increasing base soil plasticity. Note that in the study on
when distilled water is used in filter test procedure. base soils by Delgado et al. (2012), there was a direct
Although the quality of water was found to have correlation produced relating base soil plasticity index
significant effects on the dispersivity characteristics of to base soil fines content. The further discussion by
base soils, the extent of its effects on no erosion filter Delgado et al. (2012) stated that a high base soil
boundary is not entirely clear, reflecting the need for plasticity index corresponds to high base soil fines
further investigation. content and thus the D15f size of the required filter
The filter design criteria proposed by Sherard and must be smaller. Delgado et al. (2012) concluded that
Dunnigan (1989) were slightly modified by Shourijeh the base soil plasticity has influence in determining the
and Soroush (2009) for dispersive clay base soils, as boundary filter, but not as much as the base soil
given in Table 1. This proposed filter reflects the particle size distribution has. However, studies per-
effects of dispersivity on the required filter materials. formed by a number of researchers showed that the
Based on the results of an extensive NEF testing, Atterberg limits of base soils have no apparent
Delgado et al. (2006) stated that the filter criteria influence on base soil resistance to internal erosion
derived from the permeability value of a filter and the and on D15f/d85 ratio (Leonards et al. 1991; Sherard
PSD of the corresponding base soil are more appro- and Dunnigan 1985; Sherard et al. 1984b). In addition,
priate than those derived from the PSDs of a filter and Zomorodian and Moghadam (2011) stated that filter
the corresponding base soil, the latter being the criteria behavior is not influenced by the plasticity of base
set by Sherard and Dunnigan (1989). Thus, Delgado soils.
et al. (2006) suggested the filter design criterion of Delgado et al. (2012) also studied the effects of
Kf \ 59.738 9 exp (- 0.102 9 pp% \ 0.075), but treating or improving base soils on the required D15f
Kf should not be less than 0.004 cm/s. size of the filter. In this case, aluminum sulfate was
The applicability of filter design criterion by used as a base soil stabilizer. The D15f size of the
Delgado et al. (2006) was revisited by Delgado Ramos required filter was found increased with increasing
et al. (2016) by using the results published by Vakili content of aluminum sulfate in the base soil. There-
and Selamat (2014) where the base soils involved were fore, the flocculation and aggregation of base soils
broadly graded and dispersive. Delgado Ramos et al. could change the required filter. In this case coarser

123
Geotech Geol Eng

filters could be used instead, due to improvement indirect method, conducting experimental filter tests is
among the clay particles of the base soil that inhibits strongly recommended by authors to protect the
dispersivity (Delgado et al. 2012). dispersive core against internal erosion. The actual
Vakili et al. (2015b) studied the filtration mecha- condition of core materials, such as dispersivity and
nisms of broadly graded, cohesive, dispersive base erodibility, could be simulated in the experimental
soils. They reported that the criteria of D15f/d85 B 5.5 equipment. Consequently, to reduce the risks of
and D15f B 0.28 mm, each with a reasonable factor of appointing inappropriate filter criteria for a project,
safety, were found applicable for the broadly graded, the predicted filter based on the indirect method should
highly dispersive soils from Group 1 and Group 2, always be verified against the outcomes of the NEF
respectively. tests (Vakili and Selamat 2014).
The study performed by Vakili et al. (2015b) also
revealed that the no erosion lower boundary could be 5.4 The Current State of the Art in Filter Design
estimated by simultaneously considering the d85 size
and the dispersion percent of base soils of Group 1 by In mitigating the deficits of design by size ratios,
using the relationship of D15f/d85 = 10.0 exp researchers have introduced the constriction-based
(- 0.009 D) C 4.0. The D15f/d85 derived from this design criteria with Dc35 and Dc95 respectively defin-
lower bound no erosion filter boundary varies from 4 ing the controlling constriction size and self-filtering
to 10 with dispersion changing from 100 to 0%. In constriction size (Raut and Indraratna 2008; Indrar-
addition, it was found that the correlation between D15f atna et al. 2007). The procedure however is more
size of filter and d85 size of base soils could be involved as measuring the constriction size is not as
predicted by a power function of D15f/d85 = 0.6037 easy as appointing D15f and d85. Nevertheless, the
(d85)-0.66. Note that almost all widely used filter criterion is considered an enhancement as it takes into
design criteria have proposed linear function between account the effects of compaction, PSD, and surface
D15f and d85. Because there was no apparent correla- area of filter material (Raut and Indraratna 2008).
tion between D15f size of the required filter and d85 size Eventually, based on the PSD and given relative
of base soils of Group 2, the designs of filter density, the curve of constriction size distribution
boundaries based on percent dispersion of base soils (CSD) of the filter is created from which Dc35 and Dc95
of Group 2 were recommended to be as follows: are determined. The computational procedures and
subroutines have been made available by Raut (2006).
1. D15f B 0.6 mm when D \ 20%;
An example of PSD curve of given filter and its CSD
2. D15f B 0.5 mm when 20% B D \ 65%;
curves for different relative density are shown in
3. D15f B 0.28 mm when D C 65%.
Fig. 6. As observed in Fig. 6, same filter materials in
Based on the results obtained by Vakili et al. terms of PSD could have different CSD due to having
(2015b), a filter material can also be designed by different degrees of compaction, demonstrating the
considering the percent dispersion of base soil by the
criterion of D15f = - 0.0039 D ? 0.5902 for base
soils of Group 1 and the criterion of D15f = - 0.005
D ? 0.7297 for base soils of Group 2.
Finally, filter design can also be achieved with the
direct method instead of the indirect methods. Accept-
ing the established criteria in designing constitutes the
indirect method, while carrying out actual tests, such
as the NEF test, constitutes the direct method (Vakili
and Selamat 2014). From Goldsworthy’s (1990) point
of view, when vigorous erosion of the core material is
anticipated in the field, it is essential that the designing
of filter material be based on the direct method. The
Fig. 6 an example of PSD curve of given filter and its CSD
indirectly predicted filter size is sometimes coarser curves for different relative density (RD) (Indraratna et al.
than the required size. In that case, instead of using an 2012a)

123
Geotech Geol Eng

significant effects of compaction level on voids sizes instead of filter coarser fraction (Indraratna et al. 1996;
and voids distributions. Honjo and Veneziano 1989; Kenney et al. 1984).
Consequently, the PSD of base soil materials is The permeability of a filter can be estimated based
modified in the constriction-based technique by on the sizes of its fines, i.e., D5, D10, or D15 as given by
ignoring all particles coarser than Dc95, resulting in Sherard et al. (1984a, b) and Indraratna et al. (1996), as
the determination of a new self filtration size for the given respectively in Eqs. 6 and 7:
base soil and denoted as d85 . Thus, the behavior of
kf ¼ 0:35ðD15 Þ2 ð6Þ
base soil–filter combination is considered successful if
Dc35/d85 B 1. This method has been suggested for a
kf ¼ 1:02ðD5 D10 Þ0:934 ð7Þ
large array of base soil–filter combinations (Raut and
Indraratna 2008). Nevertheless, further research was Delgado et al. (2006) however pointed out that, in
needed in determining the applicability of the criterion addition to the sizes or PSD, the particles shape and the
in controlling dispersive base soils with broad compaction degree of the filter materials must also be
gradation. considered in the correlations for predicting filter
The reliability of constriction-based filter design permeability. Thus the following equations for pre-
criterion of Raut and Indraratna (2008) were experi- dicting permeability of the granular filters were given,
mentally examined by Vakili et al. (2015b) for base which simultaneously incorporate the compaction
soils with high dispersion degree and broad gradation measure or degree and the PSD:
curve. The results showed that the constriction-based
criterion of Dc35/d85 B 1.25 was found to be accept- Tv ¼ 0s kf ¼ 0:349ðD15 Þ1:388 ð8Þ
able for base soils of Group 1 with high dispersivity,
with the corresponding filter having a relative density Tv ¼ 15s kf ¼ 0:179ðD15 Þ2 ð9Þ
of 75%. However, for base soils of Group 2, the
criterion of Dc35/d85 B 1.0 was found to be accept- Tv ¼ 60s kf ¼ 0:118ðD15 Þ2 ð10Þ
able if the broadly graded base soils had a percent
Tv ¼ 120s kf ¼ 0:0808ðD15 Þ2 ð11Þ
dispersion of less than 80% and for the highly
dispersive type and broadly graded soils, i.e. soils where Tv is compaction time in (s) and kf is
with dispersion degree more than 80% and with permeability coefficient in (cm/s). Note that the
Cu C 20, the required criterion was Dc35/d85 B 0.5. compaction times of 0, 60, and 120 s respectively
correspond to the relative densities of 63.2, 78.5, and
5.5 Filter Hydraulic Conductivity 85.2%.
However Indraratna et al. (2012a), in a later
Filter materials must be designed to fulfill both, the statement, stated that the hydraulic conductivity of
PSD and permeability coefficient requirements. In granular soils such as permeability of a filter material
other words, since measuring the filter permeability in would be better predicted using the constriction size
the laboratory or field would be time-consuming, it is technique. In other words, predicting the hydraulic
imperative to make a logical correlation between a conductivity of a filter material by solely considering
particle size and the corresponding permeability of a the particle size distribution would lead to a signifi-
filter material (Indraratna et al. 2012a; Boadu 2000). In cantly inaccurate value, as mentioned by Odong
fact, the PSD, porosity, particles shape, and relative (2007). The effects of density, gradation curve, and
density of a filter material are all indirectly implicated particle size distribution on filter hydraulic conduc-
when evaluating the filter permeability of any new tivity could actually be taken into account by the
filter design. Thus, the coefficient of permeability of a constriction size technique. Thus, as given in Eq. 12,
filter from compacted sand and gravel can be related to in predicting the hydraulic conductivity of granular
its PSD and porosity (Indraratna and Vafai 1997; soils, Indraratna et al. (2012a) provided a new semi
Indraratna et al. 1996). However, the porosity would empirical equation using the constriction size distri-
be mainly influenced by filter finer size fraction bution instead of using the particle size distribution,
where Dm c is the mean constriction size and is

123
Geotech Geol Eng

calculated from the constriction size distribution curve a filter. The textbook by Terzaghi (1996) put it that a
of filter material. filter should also not to be widely, broadly, or gap
1:644 graded as such filters would be prone to clogging
kf ¼ 36:91 Dm c ð12Þ (Indraratna and Locke 1999). Dams with poor filter
Finally, Delgado Ramos et al. (2016) noted that the performance have been constructed using broadly
permeability values of filter materials tested by Vakili graded core materials, generally (Foster and Fell
and Selamat (2014) and Vakili et al. (2015b) to be 2001). Filter materials should also not susceptible to
higher than those that had been estimated using the segregation or internal instability (Foster and Fell
existing empirical correlations as given in Eqs. 6–11. 2001). In comparison to the widely graded filters, the
Delgado Ramos et al. (2016) thus proposed that the uniformly graded filters have a lesser susceptibility for
filter design criteria by Vakili et al. (2015b) should be segregation and are also more economical (Hadj-
further analyzed by carrying out the NEF tests on the Hamou et al. 1990). A good filter must be durable
same base soil–filter material systems but with the under the wetting and drying processes, under the
filter materials compacted to a higher relative density. chemical action of water, and against mechanical
In other words, Delgado Ramos et al. (2016) suggested actions, such as occurring during placement and
a measurement of the filter permeability by laboratory compaction (Fell et al. 2005).
test and by simulating real dam condition, i.e. with a
higher compaction degree of the filter material.
6 Summary, Conclusions, and Future Directions
5.6 General Properties of the Filter
This review paper focuses on dispersive base soils and
Given that filter materials must not sustain cracks, the their filtration requirements. Due to the damaging
filter zone should generally be constructed out of non- properties of dispersive soils, using them in civil and
cohesive materials. Problems associated with the geotechnical structures without considerable precau-
adverse effects of fine particles on filter performance tions would be problematic and risky. The two
have been extensively reported by Cedergren (1977). foremost problems associated with using dispersive
According to Fell et al. (2005), the amount of fine soils in earthen structures are surface and internal
content in a filter should be limited to 7% in order to erosions. The dispersivity properties of base soils are
retain the self-healing and collapsibility properties. reason for piping in earth dams and embankments.
The self-healing and collapsibility of a filter There are filter design criteria already existing in
material can be evaluated using the sand castle test, the literature with the original concepts initiated by
which was recommended by Vaughan and Soares Terzaghi (1926) then evaluated and expanded by
(1982). Generally, carrying out the sand castle test others (Vakili et al. 2015b; Indraratna et al.
in situ during filter construction has been strongly 2004, 2008; Raut and Indraratna 2008; Delgado
proposed (Soroush et al. 2011). Any cohesive behavior et al. 2006; Indraratna and Vafai 1997). The indirect
within the filter material, which is unwanted, could be but handy filter criteria are largely empirical in nature
sensitively detected using the sand castle test (Fell thus practical and could be easily applied in designs.
et al. 2005). The complex factors affecting filtration process are
By comparing the existing dam filters against filters actually already taken care of by these criteria
that were considered good and poor, as given in (Indraratna and Locke 1999). In the presence of the
Table 2, Foster and Fell (1999) and Foster (1999) various criteria however, attention should be paid as to
found that dams with D15f [ 1.0 mm have generally which of these are suitable to be used for any given
displayed poor filter performance, whereas dams with base soil and the circumstances in the field. The
an average D15f B 0.5 mm have shown good filter constriction–based criteria on the other hand can be
performance. considered as more theoretical, quantifiable, and
However, the literature has also reported that filters comprehensive where the limitations associated with
with low D15f size could be poor in performance that the particle-based criteria such as due to the effects of
the D15f should not be the only considered property of filter gradation, filter compaction, and filter PSD can
be removed.

123
Geotech Geol Eng

Table 2 Comparison of dam filter against good and poor filters (Foster and Fell 2001)
Characteristic of filter Dams with poor filter performance Dams with good filter performance

D15f Average D15f [ 1.00 mm Average D15f B 0.5 mm


Coarsest D15f = 7–40 mm Coarsest D15f = 0.5–1.5 mm
Maximum particle size C 100 mm B 75 mm
Percent finer than 4.75 mm \ 40% [ 40%
Susceptibility to segregation High Low

The overall knowledge of filtration undoubtedly References


has been enhanced by the developments made in both,
the conventional particle size-based and the state-of- Abbasi N, Nazifi MH (2013) Assessment and modification of
Sherard chemical method for evaluation of dispersion
the-art constriction size-based criteria nevertheless potential of soils. Geotech Geol Eng 31:337–346
there are still gaps and unknowns particularly when Arulanandan K, Heinzen R (1977) Factors influencing erosion
the dispersive, broadly graded base soils are consid- in dispersive clays and methods of identification. Int Assoc
ered. For example, the effects of particle uniformity of Sci Hydrol Publ 122:75–81
Arulanandan K, Perry EB (1983) Erosion in relation to filter
base soils have not been researched, either for the design criteria in earth dams. J Geotech Eng 109:682–698
particle-based criteria or the constriction-based crite- Bell F, Bruyn ID (1997) Sensitive, expansive, dispersive and
ria. In addition, the effects of water quality on the no collapsive soils. Bull Int Assoc Eng Geol 56:19–38
erosion filter boundary should be examined, especially Benahmed N, Bonelli S (2012) Investigating concentrated leak
erosion behaviour of cohesive soils by performing hole
when the base soils involved are dispersive. Consid- erosion tests. Eur J Environ Civ Eng 16:43–58
ering the fact that the veracity of some of the Bertram GE (1940) An experimental investigation of protective
established criteria has been compromised, as struc- filters. Soil mechanics series, vol 7. Graduate School of
tural safety has been breached in various cases, further Engineering, Harvard University, Cambridge
Boadu FK (2000) Hydraulic conductivity of soils from grain-
investigations have become necessary in order to size distribution: new models. J Geotech Geoenviron Eng
enhance the designers’ intuition and to achieve a more 126(8):739–746
reliable and economical filter design. The NEF test Burns B, Ghataora G (2007) Internal erosion of Kaolin. In:
nevertheless should be considered the conclusive test Puppala AJ, Hudyma N, Likos WJ (eds) Problematic soils
and rocks and in situ characterization. ASCE, Denver,
in demonstrating that a certain filter design is having pp 1–8
the required performance. Cedergren HR (1977) Evaluation of seepage stability of dams.
The information provided by this review should be In: Proceedings of the conference on evaluation of dam
applicable for the study, design, construction, and safety, Asilomar, California. ASCE
Delgado F, Poyatos JM (2008) Discussion of ‘‘A procedure for
operation of related geotechnical and geo-environ- the design of protective filters’’ appears in Canadian
mental projects. Nevertheless, this review may also Geotechnical Journal, 44: 490–495. Can Geotech J
require updating due to the progresses made in the 45:437–439
field. Delgado Ramos F, Locke M (2000) Design of granular filters:
guidelines and recommendations for laboratory testing.
Filters Drain Geotech Environ Eng Geofilters
Acknowledgements This study has been fully funded by
2000:115–122
Universiti Sains Malaysia through various research,
Delgado Ramos F, Escudero Merino D, Olalla C (2016) The
postgraduate, and postdoctoral grants.
importance of permeability in granular filter design and
control. In: Scour and Erosion—Harris, Whitehouse and
Compliance with Ethical Standards
Moxon (Eds) Ó 2016 Taylor & Francis Group, London,
pp 979–985. ISBN: 978-1-138-02979-8
Conflict of interest The authors declare that there is no con- Delgado F, Huber N, Escuder I, De Membrillera M (2006)
flict of interests regarding the publication of this article. Revised criteria for evaluating granular filters in earth and
rockfill dams. In: Transactions of the 22nd international
congress on large dams (ICOLD), Barcelona, Spain, 18–23
June 2006. International Commission on Large Dams
(ICOLD), Paris, vol 3, pp 445–456

123
Geotech Geol Eng

Delgado F, Poyatos M, Osorio F (2012) Internal erosion of Hillis SF, Truscott EG (1982). Magat dams-design of internal
clayey soils protected by granular filters. In: 6th interna- filter and drain zones. In: 35th Canadian geotechnical
tional conference on Scour and Erosion, Paris, pp 871–878 conference
Elges HFWK (1985) Dispersive soils: problem soils in South Honjo Y, Veneziano D (1989) Improved filter criterion for
Africa-state of the art. Civ Eng S Afr 27:347–353 cohesionless soils. J Geotech Eng 115:75–94
Fannin J (2008) Karl Terzaghi: from theory to practice in ICOLD (2013) Internal erosion of existing dams, levees and
geotechnical filter design. J Geotech Geoenviron Eng dikes and their foundations. In: Bulletin 164, International
134:267–276 Commission on Large Dams (ICOLD)
Fell R, Fry J (2007) Internal erosion of dams and their founda- Indraratna B, Locke M (1999) Design methods for granular
tions. Taylor and Francis, London filters—critical review. Proc ICE Geotech Eng
Fell R, Wan CF, Cyganiewicz J, Foster M (2003) Time for 137:137–147
development of internal erosion and piping in embankment Indraratna B, Raut AK (2006) Enhanced criterion for base soil
dams. J Geotech Geoenviron Eng 129:307–314 retention in embankment dam filters. J Geotech Geoenvi-
Fell R, Macgregor P, Stapledon D, Bell G (2005) Geotechnical ron Eng 132:1621–1627
engineering of dams. Balkema, Leiden Indraratna B, Vafai F (1997) Analytical model for particle
Fell R, Macgregor P, Stapledon D, Bell G (2010) Geotechnical migration within base soil–filter system. J Geotech
engineering of dams. CRC Press, Boca Raton Geoenviron Eng 123:100–109
Fernando J (2010) Effect of water quality on the dispersive Indraratna B, Nutalaya P, Kuganenthira N (1991) Stabilization
characteristics of soils found in the morwell area, Victoria, of a dispersive soil by blending with fly ash. Q J Eng Geol
Australia. Geotech Geol Eng 28:835–850 Hydrogeol 24:275–290
Flores-Berrones R, Ramirez-Reynaga M, Macari EJ (2010) Indraratna B, Dilema E, Vafai F (1996) An experimental study
Internal erosion and rehabilitation of an earth-rock dam. of the filtration of a lateritic clay slurry by sand filters. Proc
J Geotech Geoenviron Eng 137:150–160 ICE Geotech Eng 119:75–83
Foster MA (1999) The probability of failure of embankment Indraratna B, Raut A, Locke M (2004) Granular filters in
dams by internal erosion and piping. University of New embankment dams: a conceptual overview and experi-
South Wales, Sydney mental investigation. In: Geo Jordan conference 2004,
Foster MA, Fell R (1999) Assessing embankment dam filters Irbid, Jordan. ASCE, pp 15–34
which do not satisfy design criteria. University of New Indraratna B, Raut AK, Khabbaz H (2007) Constriction-based
South Wales, School of Civil Engineering, Sydney retention criterion for granular filter design. J Geotech
Foster M, Fell R (2000). Use of event trees to estimate the Geoenviron Eng 133:266–276
probability of failure of embankment dams by internal Indraratna B, Trani LDO, Khabbaz H (2008) A critical review
erosion and piping. In: Proceeding of the 20th international on granular dam filter behaviour—from particle sizes to
congress on large dams, Beijing, International Commission constriction-based design criteria. Geomech Geoeng Int J
on Large Dams (ICOLD), Paris, Question 76, vol 1, 3:279–290
pp 237–260 Indraratna B, Nguyen, VT, Rujikiatkamjorn C (2011) Assessing
Foster M, Fell R (2001) Assessing embankment dam filters that the potential of internal erosion and suffusion of granular
do not satisfy design criteria. J Geotech Geoenviron Eng soils. J Geotech Geoenviron Eng 137:550–554
127:398–407 Indraratna B, Nguyen VT, Rujikiatkamjorn C (2012a) Hydraulic
Foster MA, Spannagle M, Fell R (1998). Analysis of embank- conductivity of saturated granular soils determined using a
ment dam incidents. UNICIV Rep. No. Re—347, Univer- constriction-based technique. Can Geotech J 49:607–613
sity of New South Wales, Sydney, Australia Indraratna B, Athukorala R, Vinod J (2012b) Estimating the rate
Foster M, Fell R, Spannagle M (2000a) A method for assessing of erosion of a silty sand treated with lignosulfonate.
the relative likelihood of failure of embankment dams by J Geotech Geoenviron Eng 139(5):701–714
piping. Can Geotech J 37:1025–1061 Isbell R (2002) The Australian soil classification. CSIRO Pub-
Foster M, Fell R, Spannagle M (2000b) The statistics of lishing, Clayton
embankment dam failures and accidents. Can Geotech J Kenney T, Lau D, Ofoegbu G (1984) Permeability of compacted
37:1000–1024 granular materials. Can Geotech J 21:726–729
Goldsworthy M (1990) Filter tests—direct or indirect? Kenney T, Chahal R, Chiu E, Ofoegbu G, Omange G, Ume C
Geotechnique 40:281–284 (1985) Controlling constriction sizes of granular filters.
Goodarzi A, Salimi M (2015) Stabilization treatment of a dis- Can Geotech J 22:32–43
persive clayey soil using granulated blast furnace slag and Khor CH, Woo HK (1989) Investigation of crushed rock filters
basic oxygen furnace slag. Appl Clay Sci 108:61–69 for dam embankment. J Geotech Eng 115:399–412
Gutierrez F, Desir G, Gutierrez M (2003) Causes of the catas- Knodel PC (1991) Characteristics and problems of dispersive
trophic failure of an earth dam built on gypsiferous allu- clay soils. US Department of the Interior, Bureau of
vium and dispersive clays (Altorricón, Huesca Province, Reclamation, Denver Office, Research and Laboratory
NE Spain). Environ Geol 43:842–851 Services Division, Materials Engineering Branch, Denver
Hadj-Hamou T, Tavassoli MR, Sherman WC (1990) Laboratory Kwang T (1990) Improvement of dam filter criterion for cohe-
testing of filters and slot sizes for relief wells. J Geotech sionless base soil. MEng thesis, Asian Institute of Tech-
Eng 116:1325–1346 nology, Bangkok, Thailand
Hagerty D (1991) Piping/sapping erosion. I: basic considera- Lafleur J (1984) Filter testing of broadly graded cohesionless
tions. J Hydraul Eng 117:991–1008 tills. Can Geotech J 21:634–643

123
Geotech Geol Eng

Lafleur J, Mlynarek J, Rollin AL (1989) Filtration of broadly pavement embankment with dispersive clay. J Mater Civ
graded cohesionless soils. J Geotech Eng 115:1747–1768 Eng 28:04015179
Lafleur J, Mlynarek J, Rollin AL (1993) Filter criteria for well Raine SR, Loch RJ (2003) What is a sodic soil? Identification
graded cohesionless soils. In: Braun J, Helbaum M, Schuler and management options for construction sites and dis-
U (eds) Filters in geotechnical and hydraulic engineering. turbed lands. In: Workshop on soils in rural Queensland,
Balkema, Rotterdam, pp 97–106 2003, Toowoomba, Queensland, Australia
Leonards G, Huang A, Ramos J (1991) Piping and erosion tests Raut AK (2006) Mathematical modelling of granular filters and
at Conner Run Dam. J Geotech Eng 117:108–117 constriction-based filter design criteria. PhD thesis,
Li M, Fannin RJ (2008) Comparison of two criteria for internal University of Wollongong, Wollongong, Australia
stability of granular soil. Can Geotech J 45:1303–1309 Raut AK, Indraratna B (2008) Further advancement in filtration
Locke M, Indraratna B (2002) Filtration of broadly graded soils: criteria through constriction-based techniques. J Geotech
the reduced PSD method. Geotechnique 52:285–287 Geoenviron Eng 134:883–887
Locke M, Indraratna B, Adikari G (2001) Time-dependent Reddi LN, Bonala MV (1997) Analytical solution for fine par-
particle transport through granular filters. J Geotech ticle accumulation in soil filters. J Geotech Geoenviron
Geoenviron Eng 127:521–529 Eng 123:1143–1152
Lone M, Hussain B, Asawa G (2005) Filter design criteria for Regazzoni PL, Marot D (2011) Investigation of interface ero-
graded cohesionless bases. J Geotech Geoenviron Eng sion rate by Jet Erosion Test and statistical analysis. Eur J
131:251–259 Environ Civ Eng 15:1167–1185
Mcdonald L, Stone P, Ingles O (1981) Practical treatments for Richards KS, Reddy KR (2007) Critical appraisal of piping
dams in dispersive soils. In: Proceedings of the 10th phenomena in earth dams. Bull Eng Geol Environ
international conference on soil mechanics and foundation 66:381–402
engineering, pp 355–366 Romero Diaz A, Marin Sanleandro P, Sanchez Soriano A,
Mcintyre D (1979) Exchangeable sodium, subplasticity and Belmonte Serrato F, Faulkner H (2007) The causes of
hydraulic conductivity of some Australian soils. Soil Res piping in a set of abandoned agricultural terraces in
17:115–120 southeast Spain. Catena 69:282–293
Minguez R, Delgado F, Escuder I, Membrillera M (2006) Savas H (2015) Consolidation and swell characteristics of dis-
Reliability assessment of granular filters in embankment persive soils stabilized with lime and natural zeolite. Sci
dams. Int J Numer Anal Methods Geomech 30:1019–1037 Eng Compos Mater. https://doi.org/10.1515/secm-2014-
Mitchell J (1993) Fundamentals of soil behavior. Wiley, New 0202
York Sherard J (1979) Sinkholes in dams of coarse, broadly graded
Natural Resources Conservation Services (NRCS) (1994) Gra- soils. In: 13th international congress on large dams, New
dation design of sand and gravel filters. In: Mattinson M Delhi, Q47, R2, Paris, pp 325–334
(ed) National engineering handbook, chap 26, Part 633. Sherard J (1983) Discussion of ‘‘Design of filters for clay cores
USDA, Washington of dams’’ by Peter R. Vaughan and Hermusia F. Soares
Nevels Jr J (1993) Dispersive clay embankment erosion: a case (January, 1982). J Geotech Eng 109:1195–1197
history. Transportation Research Record Sherard J (1984) Trends and debatable aspects in embankment
Odong J (2007) Evaluation of empirical formulae for determi- dam engineering. Int Water Power Dam Constuction
nation of hydraulic conductivity based on grain-size anal- 36:26–32
ysis. J Am Sci 3(3):54–60 Sherard JL (1985) Hydraulic fracturing in embankment dams.
Ouhadi V, Goodarzi A (2003) Pore fluid characteristics effect on Seepage and leakage from dams and impoundments.
the dispersivity behaviour of soils from macro and micro ASCE, New York, pp 115–141
structure aspects. In: Proceedings of the 2nd international Sherard JL, Decker RS (1977) Dispersive clays, related piping,
symposium on contaminated sediments, 2003, pp 200–206 and erosion in geotechnical projects. ASTM Special
Ouhadi V, Goodarzi A (2006) Assessment of the stability of a Technical Publication, Philadelphia
dispersive soil treated by alum. Eng Geol 85:91–101 Sherard JL, Dunnigan LP (1985) Filters and leakage control in
Ouhadi V, Bayesteh H, Pasdarpour M (2012) Analysis of dis- embankment dams. Seepage and leakage from dams and
persivity behavior of clay minerals by fuzzy computational impoundments. ASCE, New York, pp 1–30
system and experimental methods. J Dispers Sci Technol Sherard J, Dunnigan LP (1989) Critical filters for impervious
33:420–428 soils. J Geotech Eng 115:927–947
Paige-Green P (2008) Dispersive and erodible soils—funda- Sherard JL, Decker RS, Ryker NL (1972) Hydraulic fracturing
mental differences. In: SAIEG/SAICE problem soils con- in low dams of dispersive clay. In: Performance of earth
ference. Midrand, pp 59–65 and earth-supported structures. ASCE, pp 653–689
Penner D, Lagaly G (2001) Influence of anions on the rheo- Sherard JL, Decker RS, Dunnigan LP (1976a) Identification and
logical properties of clay mineral dispersions. Appl Clay nature of dispersive soils. J Geotech Eng Div 102:287–301
Sci 19:131–142 Sherard JL, Steele EF, Decker RS, Dunnigan LP (1976b) Pin-
Pinto SE (1989) Filters for clay cores of embankment dams. In: hole test for identifying dispersive soils. J Geotech Eng Div
The 12th international conference on soil mechanics and 102:69–85
foundation engineering, Rio de Janeiro, pp 1689–1692 Sherard JL, Dunnigan LP, Talbot JR (1984a) Basic properties of
Premkumar S, Piratheepan J, Arulrajah A, Disfani M, Rajeev P sand and gravel filters. J Geotech Eng 110:684–700
(2015) Experimental study on contact erosion failure in Sherard JL, Dunnigan LP, Talbot JR (1984b) Filters for silts and
clays. J Geotech Eng 110:701–718

123
Geotech Geol Eng

Shourijeh PT, Soroush A (2009) Statistical study of no-erosion Vakili AH, Davoodi S, Arab A, Selamat MR (2015a) Use of
filter (NEF) test results. Proce ICE Geotech Eng artificial neural network in predicting permeability of dis-
162:165–174 persive clay treated with lime and pozzolan. Int J Sci Res
Sorensen S (1995) Dispersive soils: guide for use in farm dam Environ Sci 3(1):23–37
construction. DNRQ96017, Rural Water Advisory Ser- Vakili A, Selamat MR, Abdul Aziz H (2015b) Filtration of
vices, Department of Natural Resources Queensland, broadly graded cohesive dispersive base soils. J Geotech
Brisbane Geoenviron Eng 141(5):04015004
Soroush A, Shourijeh PT, Disfani MM (2006) Filter testing: an Vakili AH, Selamat MR, Mohajeri P (2016) Discussion of
essential aspect in designing filters for embankment dams. ‘‘Experimental study on contact erosion failure in pave-
In: Proceedings of the 7th international congress on civil ment embankment with dispersive clay’’ by S. Premkumar,
engineering, Tarbiat Modarres University, Tehran, Iran J. Piratheepan, A. Arulrajah, M. Disfani, and P. Rajeev.
Soroush A, Shourijeh P, Mohammadinia A (2011) Controlling J Mater Civ Eng. https://doi.org/10.1061/(ASCE)MT.
internal erosion in earth dams and their foundations: case 1943-5533.0001452
studies. In: Proceedings of the Indian geotechnical con- Vakili AH, Selamat MR, Abdul Aziz H, Mojiri A, Ahmad Z,
ference, pp 64–71 Safarzadeh M (2017) Treatment of dispersive clay soil by
Terzaghi K (1926) Soil physical basis of mechanics of earth ZELIAC. Geoderma 285(2017):270–279
structures. F. Deuticke, Wien (in German) Vaughan PR (1978) Design of filters for the protection of
Terzaghi K (1996) Soil mechanics in engineering practice. cracked dam cores against internal erosion. American
Wiley, New York Society of Civil Engineers, Reston
Townsend F, Shiau J-M, Pietrus T (1987) Piping susceptibility Vaughan P, Soares HF (1982) Design of filters for clay cores of
and filter criteria for sands. In: Symposium on engineering dams. J Geotech Geoenviron Eng 108:17–31
aspects of soil erosion, dispersive clays, and loess, Atlantic Vinod J, Indraratna B, Mahamud MA (2010) Internal erosional
City, New Jersey, USA, 29 Apr 1987. ASCE behaviour of lignosulfonate treated dispersive clay. In:
Truscott EG (1983) Discussion of ‘‘Design of filters for clay Geotechnical society of Singapore—international sympo-
cores of dams’’ by Peter R. Vaughan and Hermusia F. sium on ground improvement technologies and case his-
Soares (January, 1982). J Geotech Eng 109:1197–1198 tories, ISGI’09, pp 549–554
Umesh T, Dinesh S, Sivapullaiah PV (2011) Characterization of Wan CF, Fell R (2004) Investigation of rate of erosion of soils in
dispersive soils. Mater Sci Appl 2:629–633 embankment dams. J Geotech Geoenviron Eng
Vakili AH, Selamat MR (2014) An assessment of veracity of 130:373–380
filter criteria for earth dams. Proc ICE Geotech Eng Zomorodian SA, Moghadam MJ (2011) Investigation of effec-
167(6):574–584 tive parameters on the embankment dam filter behavior in
Vakili AH, Selamat MR, Moayedi H (2013a) Effects of using simultaneous cracking in the core and filter. Geotech Geol
pozzolan and Portland cement in the treatment of disper- Eng 29:637–644
sive clay. Sci World J 1–10 Zorluer I, Icaga Y, Yurtcu S, Tosun H (2010) Application of a
Vakili A, Selamat M, Moayedi H, Amani H (2013b) Stabiliza- fuzzy rule-based method for the determination of clay
tion of dispersive soils by pozzolan. In: Forensic engi- dispersibility. Geoderma 160:189–196
neering 2012. ASCE, pp 726–735
Vakili A, Selamat MR, Moayedi H (2013c) An assessment of
physical and mechanical properties of dispersive clay
treated with lime. Casp J Appl Sci Res 2:197–204

123

You might also like