United States
Environmental Protection
Agency
Municipal Environmental Research
Laboratory
Cincinnati OH 45268
Research and Development
EPA-600/S2-84-007 Feb. 1984
Project  Summary
Spill  Alert  Device for
Earth  Dam Failure Warning

Robert M. Koerner and Arthur E. Lord, Jr.
  A spill alert device based on the
monitoring of acoustic emissions (AE)
has been developed, field-tested, and
placed into an  operational mode at
several sites.  This  apparatus can be
useful in predicting and anticipating the
failure  of earthen structures such as
dams, waste storage lagoons, and spoil
piles. With sufficient advance warning,
repair of such structures becomes pos-
sible, thus avoiding possible catastroph-
ic discharges of their contents into the
environment.
  This report describes the fundamental
mechanisms that cause soils to generate
AE when placed  under strain and the
techniques and equipment necessary to
monitor such emissions. Results of
laboratory testing are shown to demon-
strate a relationship  between soil types
and characteristics  and the AE that
result when such soils are subjected to
applied stresses. Evidence is presented
to show that AE increase  as a soil
approaches failure due to imposed
stresses. Conversion of the  laboratory
apparatus to a portable system suitable
for field use is documented. This equip-
ment had an estimated cost of under
$2,000 in December 1978.
  Results are presented for field tests of
AE monitoring of 19 field sites. These
results  reveal potential weaknesses in
some  earthen dikes and stockpiles,
highway fill stockpiles, and embank-
ments and identify  sites of potential
failure so that corrective measures can
be undertaken.
  This project was a 1977 recipient of
one of the Industrial Research  Maga-
zine's  IR-100 Awards. A number of
companies are now  marketing AE de-
vices for earth structure monitoring.
  This Project Summary was developed
by EPA's  Municipal  Environmental
Research Laboratory, Cincinnati. OH.
to announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).


Introduction
  The problem of failures of earthen dams
and dikes, retaining walls, lagoon em-
bankments, etc., is ageless and has
continued to be a source of catastrophic
losses of life, property, and contained
materials through the years. In addition to
large, well-documented disasters (e.g.,
the Grand Teton and Taccoa Falls dam
failures), many smaller, less publicized
failures, also occur in privately owned
dams, storage piles, etc. Such failures
often have serious impacts on down-
stream  water quality  and  aquatic  life
when the hazardous (or foreign) materials
in these  ponds and  lagoons are dis-
charged in an uncontrolled manner.
  It has long been known that  certain
structures emit internally generated
sounds when placed under stress condi-
tions. In some cases  these sounds are
audible (e.g., "crying" of tin and cracking
of wood), while in others the sounds are
not in the audible range and can only be
detected  by  sophisticated  equipment.
Historically, AE monitoring began in the
mining industry to detect instabilities and
to predict when failures (rock  bursts)
might occur. Extensive research has now
been carried  out by numerous investi-
gators on the AE phenomena exhibited in
metals, metallic structures (e.g., pressure
vessels), ceramics, rocks, various mines,
plastics, soils, and earthen and other
structures under various conditions. In
most such programs, piezoelectric sen-
sors are used to detect the emissions. The

-------
very small electrical responses are then
amplified, filtered, counted, and recorded.
  An understanding of the AE emitted by
soils and an ability to relate them to the
characteristics of the structure allows the
information to be  used  in  protecting
structures such as dams, embankments,
etc., from unexpected and catastrophic
failures. Thus, a majorgoal of the current
project was the "translation" or conver-
sion of nonaudible AE from soil structures
to some measurable or recordable format
that could be used  as a  nondestructive
test of the safety of such containment
structures.


Laboratory Studies
  The laboratory work studying the  be-
havior of AE  in soil was focused  on
several granular soils (they are the most
emittive) and on different types of fine
grained soils (they are the most trouble-
some). The attempt in all cases was to
systematically vary  one parameter at a
time and, thereby, observe its influence
on  the subsequent response. The  re-
sponse that was generally monitored was
both stress/strain and stress/AE. Since
strain and AE are both cumulative phe-
nomena,  they should be capable of being
compared,  a feature  that was indeed
present and  was  brought out in  the
following studies.

Granular Soils
  Four types of sands were evaluated in
this phase  of  the  work. The  choices
provided  a broad range  of variation in
particle shape and uniformity; however,
the size  range was rather limited,  i.e.,
from 0.20 to 0.45 mm effective size.
  In the first series of tests, hydrostatic
pressure was applied to the specimen to
produce isostatic conditions. Cumulative
AE counts were recorded with time after
each pressure  increment was  applied.
Other than for the final level of AE counts,
the time for the AE to cease (i.e., to attain
equilibrium of particle  reorientation)
varied primarily  with particle shape.
Samples containing rounder particles
ceased emitting much before those with
angular particles.
  Using the same soil samples and experi-
mental test set up as with the isostatic
test results just covered, a series of triaxial
shear creep tests  was  performed.  The
deviator stress (or principal stress differ-
ence) versus strain and the deviator stress
(or principal stress difference) versus AE
behavior for the four soils were deter-
mined. Almost identical  behavioral  pat-
terns of  stress/strain  and  stress/AE
curves at all levels of confining pressure
were observed. This behavior indicates a
basic correlation between strain and AE,
the  determination of which  was  the
fundamental goal noted in the introduc-
tion to this section.
  On Particle Shape - The more angular
the soil particles contained within  the
total sample, the  most emittive is  the
sample under stress.
  On Coefficient of Uniformity - As coef-
ficient of uniformity increases, that is, as
the soil becomes well-graded, so does the
level of cumulative AE. This is a strong
conclusion for the triaxial test behavior
and is  in almost perfect agreement with
the isostatic test results. However, the
more angular soils also happen to have
the highest coefficient of uniformity. The
actual  cause of greater emissions may
therefore be a combined effect.
  On Effective Size - Little in the way of a
firm conclusion can be stated since the
range of effective size evaluated, 0.20 to
0.45 mm, is quite limited.


Fine Grained Soils
  Various aspects of  fine grained soils
were evaluated on a number of silts and
clays. Each is explained separately in the
following paragraphs.
  On Confining Pressure - The effect of
confining pressure on  the AE behavior of
cohesive soils was evaluated for two of
the four soils. The close parallel in the
behavior of stress/strain and stress/AE
curves was easily noted. Also, the fact
that the overall AE count levels are slightly
higher for the clayey silt with its silt-sized
particle component than for the kaolinite
clay is  in agreement with the AE ampli-
tude study described  in the report. The
analogous behavior  of strain and AE
indicates that the two parameters  are
related and that either or  both can be
used in conjunction with stress to char-
acterize or  monitor a given soil.
  On Water Content - The samples were
compacted at different water contents
and tested in unconfined compression.
There was a decrease in strength and AE
with increasing  water content. The ex-
tremely low number of emissions record-
ed at higher water contents emphasizes
the susceptibility of  the technique to
experimental error and external  noise
interference as water content approach-
es the liquid limit (i.e., loss of measurable
shear  strength)  of the soil  being moni-
tored. The low AE activity as the soil loses
its  shear strength because of moisture
inundation could possibly cause problems
in some monitoring situations.
  On Plasticity Index - The four cohesive
soils tested in this study had plasticity
indices of 10, 19,  19, and 512 percent.
Each soil was compacted to achieve a
void ratio of 0.89 and tested in consoli-
dated-drained triaxial creep at 34-kN/m2
(5-psi)  confining  pressure.  The most
emittive soil is the clayey silt, which has
the lowest plasticity index and the most
silt-sized material. The kaolinite clay and
siltyclay have the same plasticity indices
and similar AE response curves. Thus, a
strong correspondence  exists between
AE response and plasticity in fine grained
soils.
  On Sample Structure - All testing con-
ducted  up to this point  has been  on
remolded samples  prepared in the labor-
atory under closely controlled, thus nearly
ideal, conditions. Since one of the case
histories to be examined later provided
the opportunity for obtaining undisturbed
soil samples,  the soil (a  silty clay) was
tested in the as-received condition.  The
AE level was low, due  in part to  the
cohesive character of the predominantly
clay soil and  its relatively high  water
content. However, the AE response close-
ly resembles the stress/strain behavior
as has been the case for the remolded soil
samples examined previously.
  On Stress History - The "Kaiser effect"
is well-established in AE literature, in
which AE levels are low until a material is
stressed beyond that which it has experi-
enced in the past.  Thus, many materials
retain a record  of their  stress history,
which is evidenced by the AE response.
  In this phase of the study, stress history
testing was undertaken for AE monitoring
by fixing an accelerometer to the upper
load platen of a standard consolidation
odeometer. Tests were conducted in the
prescribed manner with deformation/
time and AE/time data sets being gener-
ated for  each pressure  increment.  The
soil tested was a sandy silty clay known
locally as a preconsolidated marl of  low
plasticity. The standard deflection  plot
was roughly reflected in the curve of AE
counts, i.e., during periods of low deflec-
tion rates, the AE count rates were  low
and, during periods of  high  deflection
rates, the AE count rates were high.  The
time for 50 percent consolidation, tso, for
each  pressure increment was  used to
obtain an AE count at 50 percent consoli-
dation. The AE data were normalized by
dividing the accumulated emission count
at tso for each  pressure increment by the
total emission count registered during all
pressure increments. A  graph of  the
response consists  of two nearly straight
lines intersecting  at about 810 kN/m2

-------
(8.0 tsf), which coincides with the begin-
ning of the straight line portion of the
virgin compression curve. Most important
is that the AE levels are generally lower at
stress levels below the preconsolidation
pressure than they are at stress levels
that exceed the preconsolidation pres-
sure. Thus, stress history seems to be
identifiable using the AE monitoring
technique.


Field Test Program
  Unlike the low attenuation and easier
detection of AE in some natural and man-
made structures, the high attenuation of
AE in soils requires that some mechanism
be  used  to transmit the acoustic emis-
sions generated within the mass of soil to
the  surface  and then to  convert the
transduced electrical signal  to  some
quantifiable format. To  overcome the
problem  of attenuation, "wave guides"
are used to transmit the emissions to the
surface;  these guides may simply be
lengths of steel rod, existing metal piping,
reinforcing bars,  etc.  Ideally, the wave
guides should be placed in the soil
structure during construction, but they
may also be driven into place when
needed. In general, the "design" of the
wave guides does not have a great effect
on the character of the AE, except that
increasing the length of the guide does
lower the frequency of the first observable
resonance.
  Other than requiring much longer wave
guides and incorporating minor changes
in the field unit to make it portable and
weather  resistant, the monitoring unit
used for field testing is similar to that
used in the laboratory. The system bas-
ically consists of an accelerometer, ampli-
fier, electronic counting  system, and
cables. In December  1978, the approxi-
mate cost of such a system was slightly
under $2,000. The project report includes
a more detailed description (with photo-
graphs) of suitable equipment for both
laboratory and  field  use and specifies
procedures for  installing and operating
an AE system in the field.
  At the  completion of the project work
period in June 1979, the apparatus had
been or was being installed at 19 field
sites. A listing of and a  few details
concerning these sites are presented in
Table 1. Complete data were not available
for a few sites at the time the report was
prepared. One particularly fruitful site is
described in detail below; more detail on
the others can be obtained from the full
report.
  Site  #14 consisted  of a 4.6-m (15-ft)
high stockpile of soil fill in southwest
Philadelphia to be used for future highway
construction. The contractor  agreed to
bring the  embankment to failure by
sequentially undermining the toe of the
slope.  Once preliminary arrangements
were made, the soil was sampled, tested,
and found to be a well-graded silty sand
with a  trace of clay (SW-ML). Its natural
water  content was  12 percent, and its
unit weight was approximately 1.92
g/cm3(120pcf).
  An 18-m (60-ft) length of the embank-
ment  was  excavated in a  series of
separate cuts beginning at the toe  and
extending into the  slope. To minimize
background noise, the front end loader
used for  the excavation actually left the
site after  each cut  until AE  ceased
completely, that is, until full stability was
Table  1.    Overview of Sites Being Monitored Using the Acoustic Emission Method
Site No. and
Location
1. PA
2. PA
3. NE
4. MD


5. PA

6. NE
7. PQ

8. DE
9. PA
10. NJ
11. VA
12. NY
13. PA
14. PA
15. PA
16. TX
17. KY
18. DE

Purpose
Flood control
Recreation
Flood control
Ore stockpile


Surcharge load

Flood Control
Tailings dam

Dredging spoil containment
Water supply
Chemical waste containment
Chemical waste containment
Petroleum waste containment
Stockpile for highway fill
Stockpile for highway fill
Seepage beneath earth dam
Gypsum dam
Sludge and wastewater lagoons
Water reservoir

Height
(ft)
30
66
67
40


6

68
95

15-40
120
8
4-15
8-20
15
15
12
150
13-28
25

Length
2.600 ft
2.500 ft
900ft
300ft


120ft

600ft
900ft

6 mi
600ft
4 mi
500ft
450ft
20ft
60ft
1.200 ft
2 mi
2 mi
1,000 ft

Embankment
Design and
Construction
Excellent
Excellent
Excellent
Good


Good

Excellent
Good

Poor
Excellent
Poor
Poor
Poor
Poor
Poor
Good
Poor
Good
Good

Foundation
Stability
Excellent
Excellent
Compressible
Poor


Poor

Compressible
Good

Good
Excellent
Very poor
Unknown
Unknown
Good
Good
Poor
Poor
Average
Good

Acoustic
Emission
Waveguides"
20 rods*
12 rods*
12 re -bar s+
2 pipes*
1 pipe*
1 re-bar*
1 pile*
3 rods*
6 rods*
3 rods*
3 pipes*
1 1 rods*
12 re -bars*
12 rods*
4 rods
6 rods*
1 rod*
4 rods*
8 rods*
ft
8 rods*
1 casing*
3 rods*
Flange of
Acoustic
Emission
Count Ftate
(counts/min)
0
0
0-200
0-20


2-750

ft
ft

2-10
0-5
0-40
0-3
2-1OO
10-190
2-7.7OO*
20-48O
ft
0-4
0-40

'Asterisk (*) - vertical; plus C) - horizontal.
"Monitoring in process.
"High count occasioned by intentional destabilization.
alnstallation in progress.

-------
reattained. Five separate cuts were re-
quired to bring the slope to failure, and
the process extended  over a 21-day
period. Figure lisa schematic diagram of
the approximate outline of the five cuts.
Acoustic emission readings were taken
from four 13-mm (1 /2-in.)diameter wave
guides driven vertically from the top of the
slope  down through the embankment to
within 1 m of the  relatively firm founda-
tion. For the first four cuts, the resulting
response curves of count rate versus time
are given in Figure 2. The data shown are
from the wave guide in the most actively
deforming region of the embankment.
From these curves the following observa-
tions can be made.
  The  general response from the first
four cuts indicated a high AE rate initially,
then an approximately exponential decay-
ing rate  with time  until  stability was
reached.  Overall AE rates generally in-
creased with each  successive cut.  An
exception occurs  during Cuts 2 and 3,
where it is seen that some AE levels are
greater after  Cut 2; however, AE is
detected for a much longer time after Cut
3.
  The  emission rate from the fifth and
last cut  initially  followed the general
trend; but, 30 min after the cut was made,
the AE rate began  to increase rapidly (see
Figure 3). When the count rate reached
its maximum (about 7700 counts/min), a
large secton of soil pulled away from the
intact mass and slid down the remaining
slope. Thereafter,  the count rate bega n to
subside and eventually came to equilib-
rium.  The post-failure count  rate curve
appears to be consistent with the original
curve.
  Not shown on these figures is the effect
of rain on the AE count rate. Approxi-
mately 8200 min (5.7 days) after Cut 3
was made, a heavy rainfall  caused the
count  rate to rapidly increase to 200
counts/min. Thirteen hundred minutes
(0.9 days) later, the count rate returned to
its former level of 2 to 5 counts/min. Rain
again  interrupted the testing  program
after  Cut 4 was  made. Approximately
3000 minutes (2.1 days) after the cut was
made, rainfall occurred  and the count
rate increased to 350 counts/min. An
additional 2400  min (1.7 days) were
required for the count rate to decrease to
"zero." The longer time period necessary
for readjustment of the slope to equilib-
rium  conditions after the rain of Cut 4
may be due to the gradual decrease in the
slope's factor of safety. From this infor-
mation, it can be concluded that the two
rainfalls had an  adverse effect on  the
                               Failure wedge
                        Tension crack
                        from Cut No.
                                            To acoustic emissions
                                            readout equipment
                                                      Waveguide
                                                               15ft
                     w"^    v '  *Z   !•••  ••
 •'•'"         Cut No.  12345

Figure 1.    Schematic diagram of embankment purposely brought to failure by successive
           excavation at toe of slope.
                                                            Car #2

.c
1
2
1
o
0
ki
^
5

4

3

2

1
Cut til

\
\
\

• ^^
* ^^^.^^
'
              10
          20
       Time (min.)
                             30
                                           500
20   30
 Time (min.)
                                                            2000
                                                                Cut #4
20  40 240 260 280   1680
          Time (min.)
                                     6000
                                                     40
                                                  80     120
                                               Time (min.)
                                                                           160
Figure 2,
Acoustic emission rate versus time response for Cuts 1, 2, 3, and 4 of embankment
shown in Figure 1.
slope's stability, at least on a temporary
basis.
  Additional data can be obtained from
this  particular site by plotting the AE
count rates of each cut as in  Figure 3.
Shown on this figure are curves for both
the maximum count rate and the average
count rate during  the 1-hr period after
monitoring began. The response curves
are somewhat linear for the first four cuts
but increase rapidly thereafter. This type
of behavior substantiates the  generally
acknowledged fact that loss of stability in
                              slopes is not a linear process, but one in
                              which instability progresses at an increas-
                              ing rate as failure is approached.

                               This field test was the most controlled
                              of all those listed in Table 1, and hence
                              allowed the  most  information to be ob-
                              tained. It shows quite  conclusively the
                              stability  predictive capability of the AE
                              method. The AE results from other field
                              sites have also affirmed the  potential
                              usefulness of the technique (details in
                              report).
                                    4

-------
Recommendations
  The  AE spill alert  device  has been
subjected to extensive laboratory and field
testing. It now should be subjected to
equally arduous tests  in the hands of
potential users such as hazardous site
owners,  engineering firrr\s, and others
involved in spill prevention and impound-
ment,  design, and construction work.
Extensive field testing in different situa-
tions and under various conditions  (in-
cluding controlled failure) must now be
carried out to "fine-tune" the apparatus
and its  use and broaden the  data base
needed for predictions.

Conclusions
  The AE generated by and in an earthen
structure such as a dam, embankment, or
storage pile can be correlated with  the
strain the structure is experiencing.
  By monitoring AE over time, changes in
the stability of the structures can  be
predicted and, where necessary, correc-
tive  action  can  be  taken to  prevent
catastrophic failure or, in the  most  ex-
treme  case, initiate  evacuation of  the
downstream area.
  By monitoring the AE of a dam or dike
over time, the current and expected safety
of such structures can be predicted. The
character of the soil in the structure and
the amount of moisture in the soil can
influence the  level of AE and make it
necessary to use such data with care.
Much  laboratory work has been per-
formed to determine the AE characteris-
tics of the various soil types (sands, silts,
and clays) under different conditions.
  A wide range of other potential uses
and applications exists for AE monitoring.
Such applications can supplement other
engineering techniques, identify problem
areas,  and help to avoid failures, which
could expose workers, inhabitants, and
aquatic species to potentially hazardous
conditions.
  The full report was submitted in fulfill-
ment of Grant No. R-802511 by Drexel
University under the sponsorship of  the
U.S. Environmental Protection Agency.
                                                                   to 7700
               20      40
                   Time (minj
                    60
3            4
  Cut Number
Figure 3.
Acoustic emission rate versus time response for Cut 5 of embankment shown in
Figure 1 and summary AE rate response from all five cuts.
  Robert M. Koerner and Arthur E. Lord. Jr.. are with Drexel University, Philadelphia,
    PA 19104.
  John E. Brugger is the EPA Project Officer (see below).
  The complete report, entitled "Spill Alert Device for Earth Dam Failure Warning,"
    (Order No. PB 84-138 189; Cost: $14.50. subject to change) will be available
    only from:
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
  The EPA Project Officer can be  contacted at:
          Oil and Hazardous Materials Spills Branch
          Municipal Environmental Research Laboratory—Cincinnati
          U.S. Environmental Protection Agency
          Edison, NJ 08837

-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
                0000329
                                                                                         U.S. GOVERNMENT PRINTING OFFICE. 1984-759-102/856

-------