You are on page 1of 19

SEEPAGE-INDUCED INTERNAL INSTABILITY TESTING FOR DAM SAFETY

ASSESSMENTS

Paul Slangen, Ph.D.1


Jonathan Fannin, Ph.D., P.Eng.2

ABSTRACT

Engineers commonly rely on empirical screening tools, and non-standardized laboratory


tests, to assess the internal erosion hazard in embankment dams. For erosion by seepage-
induced internal instability, such laboratory tests are recommended in dam safety
guidelines for important projects, albeit without guidance on how to conduct a test. The
rigid-wall permeameter is a relatively simple device, which is well-suited for state-of-
practice characterization of internal instability. We review details of select tests published
in the research literature, including features of the test device, test procedures, and
interpretation of the test results. We discuss the various test alternatives, and provide
broad recommendations for seepage-induced internal instability testing in the context of
dam safety assessments.

INTRODUCTION

Internal erosion of embankment and foundations soils has been identified as the cause of
many reported dam incidents and failures (Foster et al. 2000). It is a major concern to
dam owners (e.g. Hartford 2008), exacerbated by need to rely on subjective expert
judgements, in the absence of fully-developed mechanics-based models of internal
erosion, for dam safety assessments.

Assessing the risk of internal erosion in an embankment dam includes evaluating the
likelihood of seepage-induced internal instability of core and filter materials. For
important projects, ICOLD (2017) recommends conducting project-specific tests to
determine (i) the susceptibility of a material to seepage-induced internal instability, and
(ii) the hydraulic demand required to initiate internal instability. Soils potentially subject
to internal erosion are identified as: broadly-graded glacial tills, colluvium, and
weathered granite, alluvium with very broad gap-gradings, and broadly-graded drains in
dams containing excessive fines content.

In this paper, we will provide a brief overview of the mechanics of seepage-induced


internal instability, and review experimental setups, procedures, and reported results of
published laboratory programs for seepage-induced internal instability testing. The scope
of this review is limited to rigid-wall permeameters, as these are relatively easy to
assemble and operate. We will discuss limitations of the tests, including limitations of the
interpretation of results, and provide general recommendations for testing to determine
seepage-induced internal instability testing for engineering practice.

1
Geosyntec Consultants, Columbia, MD, pslangen@geosyntec.com.
2
University of British Columbia, Vancouver, BC, jonathan.fannin@ubc.ca.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 1


SEEPAGE-INDUCED INTERNAL INSTABILITY

Seepage-induced internal instability is the migration of fine particles through the pores of
a coarse-particle-dominated soil element by seepage flow. A distinction is typically made
between suffusion and suffosion (Fannin and Slangen 2014, USBR-USACE 2015), based
on mass loss, volume change, and a change in hydraulic conductivity (see Figure 1 and
Figure 2). Suffusion refers to erosion of non-load-bearing fine particles from an
undisturbed assembly of coarse particles, with typically little or no volume change and
hence an increase in hydraulic conductivity. In contrast, the term suffosion is used to
describe erosion of fine particles from a soil that exhibits a rearrangement, or partial
collapse, of the coarse-particle dominated soil structure, yielding a contractive volume
change. The combined effects of mass loss and contractive deformation can result in
either an increase or decrease of hydraulic conductivity.

Seepage-induced internal instability is governed by material susceptibility, and generally


conceived as the geometric condition of constrictions between coarser-fraction particles
relative to the size of the finer-fraction particles (e.g. Kenney et al. 1984), and the
hydraulic demand required to detach and transport fine particles.

Figure 1. Suffusion and suffosion (from Fannin and Slangen 2014).

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 2


Figure 2. Suffusion, suffosion, and fluidization (from Fannin and Slangen 2014).
Material susceptibility can be evaluated by one of several empirical methods (e.g. Kezdi
1979, Sherard 1979, Kenney and Lau 1985, 1986, Li and Fannin 2008, Wan and Fell
2008) based on inspection of the shape of the gradation curve as a proxy for the
geometric condition of a controlling constriction size. None of these methods distinguish
between potentially suffusive and suffosive soils. Noteworthy is both the Sherard and the
Wan and Fell methods are the only ones that have been developed for soils with
considerable silt and clay-size content (i.e. material passing the # 200 sieve), with the
notion that the Wan and Fell method has not been validated on a dataset different than the
one used for its development. The Kenney and Lau method has been widely used for
granular materials and found to perform satisfactory as a screening tool, based on an
assessment using laboratory test results from several investigations (Slangen and Fannin
2016).

No models exist that can predict the hydraulic demand required to initiate suffusion or
suffosion. For granular soils, suffusion in soils with a clast-type microstructure, typically
a gap-graded material comprising a coarse fraction and a fine fraction of less than 25 %
mass passing (e.g. Skempton and Brogan 1994, Crawford-Flett 2014), has been observed
at very small hydraulic gradients 0.1 < i < 0.4. These small hydraulic demands appear to
correspond to seepage velocities required to transport fine particles through the particle
assembly (Vardoulakis, 2004) and thus appear reasonable lower boundaries for the
initiation of suffusion in granular material. The hydraulic conditions at the onset of
suffosion, which appear to be governed by effective stress (e.g. Moffat and Fannin 2011,
Chang and Zhang 2013) and the specimen size (Li 2008), remain poorly understood.

RIGID WALL PERMEAMETER

Laboratory tests using a rigid wall permeameter have been commonly used in research
studies to determine the susceptibility to seepage-induced internal instability of a
material, and to explore the hydraulic conditions at which it initiates. A rigid wall
permeameter test set up comprises several components: permeameter cell, top and bottom
boundaries, a seepage control system, an optional stress control system, and
instrumentation (see Figure 3 and Table 1). The components are described in detail in the
following paragraphs.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 3


Table 1. Rigid Wall Permeameter Characteristics
Bottom Stress
Study Cell Top Boundary1 Seepage Control2 Instrumentation4
Boundary1 Control3
USACE d = 12.5 cm N/S N/S I-CHR N/S Flowmeter
(1953) h = N/S O-CHR Standpipes x 4
Thermometer
Wittmann d = 15 cm N/S N/S I-CHR N/S Flowmeter
(1977) h = 150 cm O-CHR Piezometers x 8
Weigh-scale
Kenney and #1: d = 24.5 cm, h
N/S PP I-CHR NSL Flow rate (type N/S)
Lau (1985) = 45 cm
# 2: d = 58 cm, h
CA O-A
= 86 cm
Rubber liner along
wall
Lafleur et al. d = 19.7 cm N/S WMS I-CHR N/S Standpipes x 4
(1989) h = 23.0 cm O-CHR Flow Rate (type N/S)

Skempton and d = 13.9 cm Unconfined WMS O-CHC None Standpipes x 4


Brogan h = N/S CA Inflow Control N/S Flow Rate (type N/S)
Honjo et al. d = 30 cm N/S WMS I-CHR NSL Piezometers
(1996) h = 51 cm I-CHP
Outflow Control N/S
Sterpi (2003) d = N/S Unconfined PS I-CHR None Manometer
h = 14 cm O-CHC
Wan and Fell d = 30 cm #1: WMS & CA #1: WMS & CA #1 & #2: I-CHR N/S Piezometers x 4
(2004) h = 30 cm #1: O-CHR Flow Rate (Manual)
#2: Unconfined #2: WMS & CA #2: O-CHC
Moffat and d = 29.7 cm PP PP I-VHT TPP Load Cells (Top and Bottom)
Fannin (2011) h = 45 cm WMS WMS O-VHT LVDT
NWGP NWGP TPT x 7
DPT x 6
Flow Rate (Manual)
Sail et al. d = 28.0 cm PP PP I-VHT TPP Load Cell (Top)
(2011) h = 45 cm CA WMS O-VHT LVDT
NWGP NWGP TPT x 14
Flow Meters x 2
Densometer
Ke and d = 10 cm Unconfined CA O-CHC None Standpipes x 4
Takahashi h =17 cm NWG I-CHR Flow Rate (Manual)

Notes
1. Perforated Plate (PP), Coarse Aggregate (CA), Wire Mesh Screen (WMS), NonWoven Geotextile Perimeter Seal (NWGP),
Nonwoven Geotextile (NWG), Porous Stone (PS)
2. Inflow Constant Head Reservoir (I-CHR), Outflow Constant Head Reservoir (O-CHR), Outflow Constant Head in Permeameter
Cell (O-CHC), Inflow Variable Pressure Tank (I-VPT), Outflow Variable Pressure Tank (O-VPT), Inflow Constant Head Pump (I-
CHP), Outflow to Atmospheric Pressure (O-A)
3. Nomina Surcharge Load (NSL), Top Plate and Piston (TPP)
4. Linear Variable Differential Transformer (LVDT), Total Pressure Transducer (TPT), Differential Pressure Transducer (DPT)

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 4


Figure 3. Typical Rigid Wall Permeameter for Upward Seepage Flow
Permeameter Cell

The size of the cell (see Table 1) varies from a diameter of about 10 cm (Ke and
Takahashi 2012) to 58 cm (Kenney and Lau 1985). An acrylic tube is commonly used as
permeameter cell (see Figure 4). A clear, transparent acrylic tube facilitates visual
observations, which have been found useful to characterize the erosion phenomenon. The
smooth surface of an acrylic cell wall raises the potential for preferential seepage flow
along the wall, which was mitigated by Kenney and Lau (1985) by installing a rubber
liner along the wall. Alternative measures to mitigate against preferential seepage flow
can be taken at the inflow and/or outflow boundary.

Figure 4. Left: Permeameter cell of Moffat and Fannin (2006). Right: Permeameter cell
from Skempton and Brogan (1994)

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 5


Top and Bottom Boundaries

At the bottom boundary, specimen support is typically achieved by means of a wire mesh
screen or an aggregate-layer (see Table 1), with each of sufficiently small opening sizes
to retain the test material. In the case of upward seepage flow, these components also
serve to evenly distribute the seepage flow, in what is sometimes termed a “rectifying
filter”. Notably, Sterpi (2003)used a porous stone as bottom boundary. Nonwoven
geotextile perimeter seals have been used to seal the space between the bottom boundary
assembly and the wall of the permeameter cell to reduce the possibility of preferential
seepage paths along the cell wall. An unconfined top boundary provides an unimpeded
exit for seepage flow and any eroded particles (e.g. Skempton and Brogan 1994), whereas
a semi-confined top exit, of for example a wire mesh screen and perforated plate (e.g.
Moffat and Fannin 2011), needs to be designed to allow passage of eroded particles
(Moffat and Fannin 2006).

For the case of downward seepage flow, the bottom boundary provides specimen support
and must also allow relatively unimpeded passage of eroded particles. Aggregates and
wire mesh screens have also been used to fulfil these functions (e.g. Kenney and Lau,
1985). Although the top boundary was not always clearly defined for the case of
downward seepage flow (e.g. Lafleur et al. 1989), aggregates and wire mesh screens have
been used as a rectifying filter at the top of the specimen (e.g. Wan and Fell 2004).

Seepage Control System

Seepage flow through the rigid-wall permeameter was typically governed by control of
hydraulic head difference between an inflow and an outflow reservoir (see Table 1). Only
Honjo et al. (1996) adopted control of the seepage by means of a pump. Upstream of the
specimen, the hydraulic head was controlled by means of (i) constant-head reservoirs
subject to atmospheric pressure (e.g. Wittmann 1977), or (ii) pressurized constant-head
tanks (e.g. Moffat and Fannin 2011). Downstream of the specimen, hydraulic control was
governed by either: (i) atmospheric pressure in case of downward seepage flow (e.g.
Kenney and Lau 1985); (ii) a constant-head reservoir (Wan and Fell 2004) or constant-
head tank (e.g. Sail et al. 2011); or (iii), a constant-head water level on top of the
specimen in the permeameter cell, in case of upward seepage flow (e.g. Ke and
Takahashi 2012).

Stress Control System

In a rigid-wall permeameter, stress control is limited to the application of a surcharge


load on the top of the specimen (see Table 1). Thus, the options are those of (i) no
surcharge load (e.g. Skempton and Brogan 1994), (ii) a constant surcharge pressure of
aggregates (e.g. Kenney and Lau 1985), or (iii) a variable surcharge pressure imposed by
loading piston and top plate (e.g. Moffat and Fannin 2011).

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 6


Instrumentation

The two most common types of instrumentation (see Table 1) have been devices to
measure flow, either flowmeters (e.g. Wittman 1977) or manual measurements (e.g.
Kenney and Lau 1985), and devices to measure hydraulic head, including standpipes (e.g.
USACE 1953), manometers (e.g. Sterpi 2003), or total or differential pressure
transducers (e.g. Sail et al. 2011; Moffat and Fannin 2011). Additional instrumentation
has included a thermometer (e.g. USACE 1953), a weigh-scale for mass loss (e.g.
Wittmann 1977), a Linear Variable Differential Transformer (LVDT) for deformations
(e.g. Moffat and Fannin 2011, Sail et al. 2011), and load cells (e.g. Moffat and Fannin
2011, Sail et al. 2011).

TEST PROCEDURE
Specimen Preparation

Most laboratory tests have been conducted on reconstituted specimens of sand and
gravel, else on glass beads of comparable size distribution (see Table 2). Only Sterpi
(2003), Wan and Fell (2004), and Moffat and Fannin (2011) tested mixtures of fine-
grained material, sand, and gravel. All but two studies adopted a moist-tamping
technique, or variation thereof, to reconstitute a specimen. Moist-tamping involves
placing moist material in discrete layers, and tamping each layer using a compaction
hammer. A method of under-compaction (Ladd 1978)was typically employed. Several
studies did not saturate the specimen prior to subjecting it to seepage flow (e.g. Kenney
and Lau 1985). In contrast, USACE (1953), Honjo et al. (1996), and Ke and Takahashi
(2012) used a vacuum to remove air from the specimen and slowly percolate water
through the pores. Skempton and Brogan (1994) saturated the specimen by percolation
with water but did not apply a vacuum.

Both Moffat and Fannin (2011) and Sail et al. (2011) used a modified slurry-deposition
technique to prepare the specimen. Slurry-deposition involves preparing batches of
saturated material in bowls, and de-airing each batch in a vacuum chamber. The saturated
slurry is placed under a thin film of standing water in the permeameter cell. The slurry-
deposited specimen can reasonably be assumed saturated prior to consolidation.

Phase 1: Consolidation

Most studies have imposed no, or very small, surcharge loads, and accordingly have not
reported any specimen consolidation. Only Moffat and Fannin (2011), and Sail et al.
(2011), imposed substantial vertical surcharge loading on the test specimens. The
surcharge loading was applied in undefined increments up to a target consolidation stress,
with the specimen being subject to double-ended drainage.

Phase 2: Seepage Flow

Seepage was imposed in either a downward or upward direction, and with either a
constant head or in a multi-stage head procedure. Only Sterpi (2003) provided a rationale
for the selection of the direction of seepage flow: “[t]he upward direction of the flow was

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 7


preferred to the downward direction due to the tendency, observed in the preliminary tests, of
the fine particles transported by a downward flow to clog the available filtering layer [..]”.
Only Wan and Fell (2004), and Moffat and Fannin (2011) conducted both upward flow tests
and downward flow tests; all other studies conducted all tests with either upward or
downward flow. Notable is that USACE (1953), Kenney and Lau (1985), Honjo et al.
(1996), and Sterpi (2003) all subjected the permeameter to some type of vibration loading
during the seepage stage.

Table 2. Characteristics of Test Procedure


Specimen Preparation Seepage Stage4,5
Study Material Cons.2
Recon.1 Saturation Mode3 i0 i max Duration
USACE Sand and MT Vacuum N/S MSH, DF, V N/S N/S N/S
(1953) gravel Percolation

Wittmann Sand and N/S N/S N/S MSH N/S N/S N/S
(1977) gravel
Kenney Sand and MT N/S N/S DF, V N/S N/S N/S
and Lau gravel
(1985)
Lafleur Glass beads N/S N/S N/S DF, CH N/S 2.5 to 6.5 T= 2.5 h
et al. (1989) (sand-size)
Skempton Sand and MT Percolation N/S UF, MSH 0.05 1 T = 1.5 hr
and Brogan gravel t = N/S
(1994)
Honjo Sand and MT Vacuum N/S CH, DF, V 15 to 50 15 to 50 T = 2 hr
et al. (1996) gravel Percolation

Sterpi Sand and MT N/S N/S UF, CH, V 0.18 to 0.75 0.18 to 0.75 T = 32 or 48 hr
(2003) gravel, trace
fines
Wan Silt, sand, MT N/S N/S #1: CH, DF 8 8 T ≤ 12 hr
and Fell and gravel #2: MSH, UF 0.2 4.8 T ≤ 10 hr
(2004) t ≥ 10 min

Moffat and Silt, sand, MDS N/S 2D #1: MSH, DF 1 45 T ≤ 13 hr


Fannin and gravel #2: MSH, UF t = 1.5 hr (typ.)
(2011) 1 28 T ≤ 193 hr
t = 1.5 hr (typ.)
Sail Glass beads MDS N/S 2D MSH, DF 1 4.9 t = 1 hr
et al. (2011) (sand-size)
Ke and Sand MT Vacuum N/S MSH, UF 0.05 1 t = 0.5 hr
Takahashi Percolation
(2012)

Notes
1. Moist Tamping (MT), Modified Slurry Deposition (MSD)
2. Double-Ended Drainage (2D)
3. Multi-Stage Head (MSH), Constant Head (CH), Downward Flow (DF), Upward Flow (UF), Vibration V
4. Initial Hydraulic Gradient i 0 , Maximum Hydraulic Gradient i max
5. Test Duration T , stage duration t

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 8


The seepage flow stage of early experiments was not always well documented: USACE
(1953), Wittmann (1977) and Kenney and Lau (1985) provided little information on the
seepage stage beyond a very general description; Lafleur et al. (1989) reported a duration
of 2.5 h for each test and a general range of applied hydraulic gradient.

Sterpi (2003) conducted constant head tests with only upward seepage flow. Specimens
were subjected to constant hydraulic gradients between 0.18 and 0.75 for up to 48 h.
Honjo et al. (1996) and Wan and Fell (2004) conducted constant head tests with
downward seepage flow and with hydraulic gradients up to 52, and 8, respectively, and
durations of 2 h and up to 12 h, respectively.

The multi-stage head procedure comprised imposing a small initial hydraulic head across
the seepage control system for a designated duration and, thereafter, incrementally
increasing the hydraulic head across the system by increasing the hydraulic head in an
inflow device while maintaining a constant head in an outflow device. The tests
conducted by Skempton and Brogan (1994), Ke and Takahashi (2012), and the upward
flow tests conducted on granular material by Wan and Fell (2004), involved multi-stage
seepage flow in an upward direction: hydraulic gradient was increased from an initial
value of about 0.05 (Skempton and Brogan 1994, Ke and Takahashi 2012), or 0.2 (Wan
and Fell 2004), to a final value of about 1 in the last stage. The upward flow tests
conducted by Wan and Fell (2004) on fine-grained material, imposed seepage flow in a
similar procedure, but reached much larger hydraulic gradients up to about 4.8. The
duration of each stage varied from several minutes (Skempton and Brogan 1994) up to 30
min (Ke and Takahashi 2012). The duration of each stage of seepage flow of Wan and
Fell (2004) was at least 15 min, but longer in case erosion of particles was observed.

Moffat and Fannin (2011), and Sail et al. (2011) imposed multi-stage seepage flows by
incrementally increasing the hydraulic gradient in steps of about 1. The duration of each
stage was typically 60 min (Sail et al. 2011) to 90 min (Moffat and Fannin 2011),
although the latter reported longer durations in case erosion of particles was observed.
Moffat and Fannin (2011) imposed seepage flow in similar increments of hydraulic
gradient and duration, for both the downward and the upward seepage flow tests.

Criteria to end a test varied substantially. Many tests appear to have been terminated
based on qualitative evidence of erosion (e.g. Wan and Fell 2004, Moffat and Fannin
2011). Some tests appear to have been terminated after a maximum hydraulic head
across the seepage control system was reached in the absence of erosion (e.g. Kenney and
Lau 1985, Wan and Fell 2004). Other tests were terminated at the end of a pre-defined
test duration (e.g. Lafleur et al. 1989). Multi-stage upward seepage flow tests on granular
material were terminated when boiling was observed at a hydraulic gradient of about 1,
or, in case erosion of particles was observed, at hydraulic gradients smaller than 1.

REPORTING RESULTS

The particle size distribution of the specimen prior to seepage flow was always reported
(see Table 3). Although reporting of the void ratio, or porosity, of the reconstituted and

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 9


consolidated specimen was common, only Lafleur et al. (1989) and Ke and Takahashi
(2012) reported the relative density of their specimens. It was often not clear if any mass
was lost from the bottom of the specimen during reconstitution. The specimen response
during consolidation was not reported by Moffat and Fannin (2011) and Sail et al. (2011),
which were the only studies to subject their specimens to any substantial surcharge loads.
These two studies were also the only ones to report axial strains during the seepage stage.

The specimen response during the seepage stage was generally reported in terms of time,
hydraulic gradient, discharge velocity or flow rate, and hydraulic conductivity. The
definition of hydraulic gradient, in particular whether its value depended on any change
in specimen length, was usually unclear. In addition, visual observations (see Figure 5)
were found useful in describing the specimen response during seepage flow (e.g. USACE
1953, Skempton and Brogan 1994, Moffat et al. 2011). Several studies reported mass loss
during the seepage stage (e.g. Wittmann 1979, Sterpi 2003, Sail et al. 2011), else at the
end of the test (e.g. Lafleur et al. 1989, Skempton and Brogan 1994, Honjo et al. 1996).
Forensic analysis of the specimen yielded the particle size distribution of the eroded
specimen (e.g. Kenney and Lau 1985). Notably Sail et al. (2011) reported the relative
density of the specimen during the seepage stage.

Figure 5. Visual observations from Moffat et al. (2011)

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 10


Table 3. Reporting Results
Initial Final State
Study Seepag Stage2
State1 (post test)3
USACE PSD Vis N/S
(1953) Vis k, i
Wittmann PSD k, i, m N/S
(1977)
Kenney PSD N/S PSD
and Lau
(1985)
Lafleur RD k, t PSD
et al. (1989) PSD m
Skempton PSD i, q m
and Brogan n Vis
(1994)
Honjo PSD N/S m
et al. (1996) s
PSD
Sterpi e m, i, t N/S
(2003) PSD Vis
Wan PSD i, q, t PSD
and Fell n Vis
(2004)
Moffat PSD i,  ' v ,  a , t N/S
and Fannin e Vis
(2011)
Sail PSD i, Q, k,  a , RD, m, t PSD
et al. (2011)  Vis Vis
Ke and PSD i, q, k, t Vis
Takahashi e PSD
(2012) RD CPT

Notes
1. Particle Size Distribution (PSD), Visual Observations (Vis),
Relative Density (RD), Porosity (n), Void Ratio (e), Density ()
2. Hydraulic Conductivity (k), Hydraulic Gradient (i), Time (t),
Mass Loss (m), Specific Discharge (q), Vertical Effective Stress
('v), Axial Strain (a), Flow Rate (Q)
3. Cone Penetration Test (CPT)

DISCUSSION

Permeameter Cell and Top and Bottom Boundaries

The size of the permeameter cell governs the largest particle size that can be tested.
Typically, a ratio of permeameter cell diameter over maximum particle greater than six is
recommended (ASTM D5856). The transparency of acrylic tube was found very useful in
characterizing the response of the specimen to seepage flow with visual observations (e.g.
Skempton and Brogan 1994, Moffat and Fannin 2011).

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 11


The primary function of the bottom boundary is to provide specimen support. In case of
upward seepage flow, the bottom boundary also serves to distribute seepage flow evenly
across the specimen. Wire mesh screens, in combination with coarse aggregates (e.g.
Skempton and Brogan 1994, Wan and Fell 2004), can fulfil both requirements. While the
use of a perforated plate in combination with wire mesh screens (e.g. Moffat and Fannin
2011, Sail et al. 2011) does provide adequate specimen support, its ability to evenly
distribute seepage flow appears not to have been demonstrated. An open top boundary
does not impede flow of water or eroded particles and can thus be used with upward
seepage flow.

For downward seepage flow, the bottom boundary provides specimen support and
functions as the outflow boundary. A wire mesh assembly, often in combination with a
coarse aggregate, is typically used as bottom boundary in this set up. The basal support
function requires relatively small typical opening sizes of the bottom boundary, but the
outflow function requires large opening sizes. Preliminary testing by Moffat and Fannin
(2006), Ke and Takahashi (2012), and Slangen and Fannin 2017, showed that the opening
size of the outflow boundary had a considerable effect on the specimen response. For
gap-graded materials, there appears to be an emerging consensus that the opening size of
the outflow boundary should be slightly smaller than the smallest coarse particle (Moffat
and Fannin 2006, Ke and Takahashi 2014). For the top boundary, which is also the inflow
boundary, similar considerations as for the inflow boundary in the upward seepage flow
case hold true.

Seepage Control System

It is now standard to use deaired water in permeameter tests (ASTM D5856). Garner and
Sobkowicz (2002) showed that the introduction of air into the water yielded considerable
adverse effects on specimen erosion. Test configurations with downward seepage flow
and a bottom boundary subject to atmospheric pressure (e.g. Kenney and Lau 1985),
appear particularly susceptible to de-saturation of the specimen. The dissolved air content
of water should be monitored, in particular if open seepage control systems are used or if
test durations are long.

Constant head tanks provide adequate means to control the hydraulic head across the
seepage system. The hydraulic head across the seepage system comprises head losses
associated with hydraulic friction of flowing water through all components of the seepage
control system, including inlet and outflow tanks, pipes, valves, wire meshes, perforated
plates, etc., in addition to the head loss across the specimen (see Error! Reference
source not found.). Accordingly, the notion of head control should not be misunderstood
as control of the hydraulic head, or gradient, across the specimen. This limitation must be
kept in mind when designing the test procedure and analyzing the test results.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 12


Stress Control

The work of Moffat and Fannin (2011) in particular shows the challenges of controlling
the effective stress in a specimen in a rigid wall permeameter. The stress control system
allowed control of effective vertical stress at the top of the specimen. The vertical
effective stress distribution along the specimen was governed by the effective vertical
stress at the top of the specimen, but also by friction between the specimen and the cell
wall, and seepage flow. While the vertical effective stress at the top and bottom of the
specimen were measured, values of vertical effective stress at intermediate elevations
were inferred from these measurements and an assumption of wall friction. The radial
effective stress was governed by the radial flex in the permeameter cell; it is possible the
specimen approached “at rest”, K0, conditions. The undetermined effective stress state in
a rigid-wall permeameter poses severe limitations on the interpretation of the influence of
stress on seepage-induced internal instability. Flexible-wall permeameters (e.g. Ke and
Takahashi 2014, Slangen and Fannin 2017) are much better suited for this purpose.

The “heave” theory of Terzaghi (1943) provides useful insights into the stress state of a
granular specimen subject to upward seepage flow. The theory was developed for
granular soil and defines a critical hydraulic gradient across a specimen for the onset of
fluidization (see Figure 2), which occurs when the pore water pressure at the bottom of
the specimen equals the initial total vertical effective stress at that elevation, yielding a
vertical effective stress equal to zero. The theory of fluidization applies to the tests
reported by Skempton and Brogan (1994) and Ke and Takahashi (2012). The theory
demonstrates that a critical hydraulic gradient of about unity is an upper limit for upward
seepage tests on granular soils not subject to a vertical surcharge load. The theory does
not provide an upper limit for seepage tests on fine grained material, as is evident from
the results reported by Wan and Fell (2004), presumably due to significant friction
between the permeameter wall and the specimen, and due to potential tensile strength of
fine-grained materials.

Test Procedure and Reporting Test Results

Preparing homogeneous specimens of internally unstable soils can be challenging, as


there is a potential for migration of fine particles during both the reconstitution and
saturation stages. In general, the moist tamping technique is appealing as this method of
reconstitution resembles a typical construction technique used to place and compact
embankment soils in the field. However, the action of pounding the specimen with a
compaction hammer may cause migration of fine particles prior to the seepage flow
stage. This may be exacerbated for specimens prepared for downward seepage flow tests
with a bottom boundary with relatively large openings. In addition, hydraulic loads
associated with specimen saturation and consolidation may also cause particle migration.
As a reference, Skempton and Brogan (1994) and Crawford-Flett (2014) observed
migration of fine particles at hydraulic gradients as low as 0.1. It is conceivable that
specimens are subject to comparable, or even greater, hydraulic loads during saturation.
The modified-slurry deposition technique (e.g. Moffat and Fannin 2006) is an appealing

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 13


alternative, as it does not subject the specimen to vibrations during reconstitution. The
method has been shown to yield saturated gap-graded specimens (Slangen and Fannin
2017), which eliminates the need to subject the specimen to a saturation stage with
potential effects for particle migration. The disadvantage of the slurry deposition
technique is that it may not resemble actual construction techniques.

Considering the potential for unwanted particle migration during specimen reconstitution
and saturation, assessing the homogeneity of the reconstituted and saturated specimens, is
highly desirable. The variation of the particle size distribution can be determined by
dissecting a dummy specimen reconstituted and saturated for this purpose. The initial
mass of the reconstituted specimen should be recorded, as well as mass loss that occurs
during saturation of the specimen. The initial void ratio of the specimen can then be
established, which can be used to estimate the critical hydraulic gradient in case of a
specimen comprising a granular material.

Upward seepage flow poses two main advantages over downward seepage flow, in
particular for granular materials: (1) upward seepage flow allows the separation of the
basal support function and the outflow function to the bottom and top boundaries of the
specimen, respectively; and (2) the upward seepage flow configuration yields relatively
small range of hydraulic loads that can be applied, as a granular specimen fluidizes at a
hydraulic gradient of about unity. Multi-stage seepage flow provides a relatively simple
means of incrementally increasing the hydraulic load on the specimen and recording its
response at each loading increment.

In order to determine if a change in hydraulic conductivity occurs, it is essential to


establish the hydraulic conductivity of the specimen prior to particle migration. The
discharge through the specimen can be determined using relatively simple manual
monitoring methods. Considering the hydraulic head losses in the pipes and valves and
inflow and outflow boundaries, it is pertinent that pore water pressures are measured at
the top and bottom of the specimen. Sufficiently accurate pore water pressure
measurements can usually be made using standpipes or transducers.

For the first stage of seepage flow, a small hydraulic gradient of 0.05 is recommended to
avoid particle migration. It is recommended that each stage of seepage flow is continued
until no erosion is observed and at least two discharge measurements have been made
that yields no significant difference in discharge. Subsequent seepage flow stages ideally
have the same duration and incremental increases in hydraulic gradient as the initial
stages, although time constraints may lead to shorter durations and larger hydraulic
gradient increases, or both. As a guideline a minimum duration per stage of ten minutes is
recommended, and a maximum increase in hydraulic gradient of 0.1. The flow and
pressure measurements should be supplemented by visual observations of migration of
fine particles, associated with suffusion, and possible rearrangement of coarse particles
associated with suffosion. Estimates of axial deformation may also be monitored, for

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 14


example using grid lines on the permeameter cell. It is recommended to plot a time series
of the imposed load (hydraulic gradient) and discharge on one plot, which may be
supplemented by measurements of mass loss and volume change or observations of
particle migration and rearrangement. Plotting the flow rate versus the hydraulic gradient
(e.g. Skempton and Brogan 1994) provides an easy way to determine any changes in
hydraulic conductivity. It should be noted that plotting the hydraulic conductivity versus
time is typically prone to relatively large inaccuracies of calculated hydraulic
conductivities during the first few stages of seepage flow, where, typically, small values
of discharge and pore water pressures are measured.

For upward seepage flow on a granular material, the test can be terminated when
substantial erosion has occurred, or when fluidization of the specimen occurs. At the end
of the test, measurement of particle size distribution of several layers of the specimen
(e.g. Kenney and Lau 1985) can be used to support the observations made during the
seepage stage. It is pertinent that the homogeneity of the specimen prior to the seepage
stage is established if a change of gradation is the only measurement, as the initial
variation in grading provides the base case from which migration of fine particles is
inferred.

The previous discussion is in general also applicable to fine grained soils, although the
critical hydraulic gradient theory is not applicable. Accordingly, fine grained specimen
may require hydraulic gradients much larger than unity for initiation of particle migration
to occur (e.g. Wan and Fell, 2004) and the seepage control system must be designed
accordingly. Relatively large test durations are also not uncommon. There is no well-
defined criterion to end a test on a fine-grained material.

CONCLUDING REMARKS AND RECOMMENDATIONS

Bulletin 164 (ICOLD, 2017) establishes the state of practice for internal erosion of
existing dams, dikes, and levees and their foundations. Informed by its development,
USBR-USACE (2015) had given recognition to four mechanisms of internal erosion,
including the mechanism of erosion by internal or grading instability that can occur with
little or no volume change (suffusion), else occur with volume change (suffosion). For
the seepage-induced internal instability phenomena of suffusion and suffosion, laboratory
tests are recommended in dam safety guidelines for important projects.

Our state-of-art understanding of internal instability has developed through laboratory


permeameter testing of soil specimens, and related fundamental studies of glass ballotini
specimens. Early research studies used a rigid-wall permeameter (e.g. Skempton and
Brogan, 1994) that provides little or no control of stress on the test specimen, and an
approximate measure of volume change. In contrast, the state-of-art testing now uses a
flexible-wall permeameter (e.g. Slangen and Fannin, 2017) that provides for precise
control of stress, flow in either the upward or downward direction, and a very accurate
measurement of volume change

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 15


The state-of-practice characterization of internal instability is well-suited to use of a
rigid-wall permeameter. In its simplest form, the rigid-wall permeameter is configured
for upward flow, and thus the design of the inflow (bottom) boundary is governed by soil
retention alone. Furthermore, in its simplest form, the test is performed with no
surcharge pressure on the outflow (top) boundary of the test specimen.

The soil specimen is reconstituted, typically by a method of moist-tamping, else a method


of slurry deposition. The method of reconstitution imparts a fabric to the grain assembly
of the test specimen. Moist-tamping replicates a fabric believed similar to that of field
compaction, and thus the fill material of an embankment dam. Slurry-deposition
replicates a fabric believed similar to that of an alluvial deposit, and thus is likely
representative of strata existent in the foundation of the dam.

A benefit of state-of-practice testing, in a rigid-wall permeameter with upward flow, and


no surcharge pressure, is that a relatively small range of hydraulic gradient need be
applied in order to induce an instability in the test specimen. Failure by the mechanism of
internal instability will take the form of suffusion or suffosion, at a hydraulic gradient
i ≤ 1.0, and with the eroded soil grains being deposited (and easily observed) on the
outflow (top) surface of the specimen. In the absence of any failure by internal instability,
a stable gradation will fail by the phenomenon of heave, at a hydraulic gradient i ≈ 1.0,
and without erosion of soil from the test specimen.

Internal instability is typically a concern in non-plastic, gap-graded, coarse-grained soils.


Suffusion occurs with little or no volume change in the soil element whilst, in contrast,
suffosion and heave are defined by the occurrence of a volume change. Thus, a rigid-wall
permeameter, configured to provide for easy observation of the outflow (top) boundary,
and an approximate measure of volume change in the test specimen, allows for a
characterization of the failure mode of the soil gradation examined in testing.

ACKNOWLEDGEMENTS

Funding for this work has been provided by BC Hydro and the National Sciences and
Engineering Research Council. The first author would also like to acknowledge the
support from Geosyntec Consultants for this publication.

REFERENCES

ASTM D5856-15. 2015. Standard Test Method for Measurement of Hydraulic


Conductivity of Porous Material Using a Rigid-Wall, Compaction-Mold Permeameter.
ASTM International, West Conshohocken, PA.

Chang, D.S., and Zhang, L.M. 2013. Critical Hydraulic Gradients of Internal Erosion
under Complex Stress States. Journal of Geotechnical and Geoenvironmental
Engineering, 139(9): 1454–1467.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 16


Crawford-Flett, K. 2014. An improved hydromechanical understanding of seepage-
induced instability phenomena in soil. The University of British Columbia, Vancouver,
Canada.

Fannin, R.J., and Slangen, P. 2014. On the distinct phenomena of suffusion and
suffosion. Geotechnique Letters, 4(4): 289–294.

Foster, M., Fell, R., and Spannagle, M. 2000. The statistics of embankment dam failures
and accidents. Canadian Geotechnical Journal, 37(5): 1000–1024.

Garner, S.J., and Sobkowicz, J.C. 2002. Internal instability in gap-graded cores and
filters. In Candadian Dam Association Annual Conference 2002. Victoria, BC, Canada.
pp. 6–10.

Hartford, D.N.D. 2008. Risk analysis in geotechnical and earthquake engineering: state-
of-the-art and practice for embankment dams. In 6th International Conference on Case
Histories in Geotechnical Engineering. Arlington, VA.

Honjo, Y., Haque, M.A., and Tsai, K.A. 1996. Self-filtration behaviour of broadly and
gap graded cohesionless soils. In Geofilters’96 , May 29-31, Montreal, Canada.

International Commission On Large Dams (ICOLD). 2017. Internal erosion of existing


dams, levees and dikes, and their foundations. Paris, France.

Ke, L., and Takahashi, A. 2012. Strength reduction of cohesionless soil due to internal
erosion induced by one-dimensional upward seepage flow. Soils and Foundations, 52(4):
698–711.

Ke, L., and Takahashi, A. 2014. Triaxial Erosion Test for Evaluation of Mechanical
Consequences of Internal Erosion. Geotechnical Testing Journal, 37(2): 1–18.

Kenney, T.C., Chahal, R., Chiu, E., Ofoegbu, G.I., Omange, G.N., and Ume, C.A. 1984.
Controlling constriction sizes of granular filters. Canadian Geotechnical Journal, 22(1):
32–43.

Kenney, T.C., and Lau, D. 1985. Internal stability of granular filters. Canadian
Geotechnical Journal, 22(2): 215–225.

Kenney, T.C., and Lau, D. 1986. Internal stability of granular filters - Reply. Canadian
Geotechnical Journal, 23(3): 420–423.

Kezdi, A. 1979. Soil physics. Elsevier, Amsterdam.

Ladd, R.S. 1978. Preparing test specimens using undercompaction. Geotechnical Testing
Journal, 1(1): 16–23.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 17


Lafleur, J., Mlynarek, J., and Rollin, A.L. 1989. Filtration of broadly graded cohesionless
soils. Journal of Geotechnical Engineering, 115(12): 1747–1768.

Li, M. 2008. Seepage induced instability in widely graded soils. The University of British
Columbia, Vancouver, BC, Canada.

Li, M., and Fannin, R.J. 2008. Comparison of two criteria for internal stability of granular
soil. Canadian Geotechnical Journal, 45(9): 1303–1309.

Moffat, R., and Fannin, R.J. 2011. A hydromechanical relation governing internal
stability of cohesionless soil. Canadian Geotechnical Journal, 48(3): 413–424.

Moffat, R., Fannin, R.J., and Garner, S.J. 2011. Spatial and temporal progression of
internal erosion in cohesionless soil. Canadian Geotechnical Journal, 48(3): 399–412.

Moffat, R.A., and Fannin, R.J. 2006. A large permeameter for study of internal stability
in cohesionless soils. Geotechnical Testing Journal, 29(4): 273–279.

Sail, Y., Marot, D., Sibille, L., and Alexis, A. 2011. Suffusion tests on cohesionless
granular matter Experimental study. European Journal of Environmental and Civil
Engineering, 15: 799–817.

Sherard, J.L. 1979. Sinkholes in dams of coarse, broadly graded soils. In Thirteenth
Congres of Large Dams, ICOLD. New Delhi.

Skempton, A.W., and Brogan, J.M. 1994. Experiments on piping in sandy gravels.
Geotechnique, 44(3): 449–460.

Slangen, P., and Fannin, R.J. 2016. A commentary on ICOLD Bulletin 164 on internal
erosion: The significance of volume change in internally unstable soils. In 69th Canadian
Geotechnical Conference. Vancouver, BC.

Slangen, P., and Fannin, R.J. 2017. A Flexible Wall Permeameter for Investigating
Suffusion and Suffosion. Geotechnical Testing Journal, 40(1): 1–14.

Sterpi, D. 2003. Effects of the Erosion and Transport of Fine Particles due to Seepage
Flow. International Journal of Geomechanics, 3(1): 111–122.

Terzaghi, K. 1943. Theoretical soil mechanics.

U.S. Army Corps of Engineers (USACE). 1953. Filter experiments and design criteria.
United States Army Corps of Engineers, Waterways Experiment Station, Vicksburg,
Missisippi.

U.S. Bureau of Reclamation - U.S. Army Corps of Engineers (USBR-USACE). 2015.


Best Practices in Dam and Levee Safety Risk Analysis.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 18


Wan, C.F., and Fell, R. 2004. Experimental investigation of internal instability of soils in
embankment dams and their foundations. In UNICIV Report No. R-429, the University
of New South Wales, Sydney, Australia.

Wan, C.F., and Fell, R. 2008. Assessing the potential of internal instability and suffusion
in embankment dams and their foundations. Journal of Geotechnical and
Geoenvironmental Engineering, 134(3): 401–407.

Wittmann, L. 1977. Some aspects of transport processes in porous media. In 6th


Australian Hydraulics and Fluid Mechanics Conference. Adelaide, Australia. pp. 121–
124.

Copyright © 2019 U.S. Society on Dams. All Rights Reserved. 19

You might also like