See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286383592

Failure of Swift 2 Dam

Conference Paper · May 2008

CITATIONS READS

0 598

1 author:

Patrick Regan
PJR Engineering
5 PUBLICATIONS   37 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Patrick Regan on 09 December 2015.

The user has requested enhancement of the downloaded file.

Failure of Swift No. 2 Dam

Patrick J. Regan, P.E.

Gary Huhta
William Lagnion, P.E.
On April 21, 2002, the Swift 2
canal embankment failed.
• Background Information
• Geologic Setting
• Construction History
• Operating History
• Surveillance and Monitoring Prior to
Failure
• Development of Failure
• Post Failure Investigations
• Failure Mode
• Lessons Learned
 Swift 2 Project Data
• Construction completed in 1958
• Power Canal
– Fed by discharge from the upstream
Swift 1 project
– Length - 16,700 feet
– Originally fully earth lined
– Maximum embankment height – 83 feet
– Maximum flow – 9,000 cubic feet per
second
– Hazard Potential - Low
The project is
located on the
Lewis River in
southwestern
Washington,
approximately
10 miles south
of Mount St.
Helens
 Geologic Setting
• The project is located on the western
slope of the Cascade Range
• The stratigraphy at the site is
dominated by volcanic activity in the
Cascades
• In ascending sequence the site
stratigraphy includes:
– Bedrock - a series of interstratified lava
flows, breccia, tuff and volcaniclastic rocks
45 to 31 million years old and up to 5,000
meters thick.
– An assemblage of generally unconsolidated
deposits of pyroclastic flows, lahar debris
and alluvium 36,000 to 12,000 years old
with a thickness of approximately 200
meters.
 Geologic Setting
– The Pre-Basalt Alluvium, a variable deposit of
sandy-gravel-cobble alluvium and sandy silt
alluvium associated with the Lewis River
– The Cave Basalt, interpreted as a single flow
from Mount St. Helens about 1900 years ago.
– Post Basalt alluvium/colluvium consisting of
mainly silty sandy gravels
Geologic X-Section
 Top of Cave Basalt

Impervious Fill

Post Basalt Alluvium/Colluvium

Cave Basalt

Pre-Basalt Alluvium
 Construction History
• The Project was constructed in 1957
• Approximately 2 million cubic yards of
material was placed to form the canal
and forebay embankments
• Embankment fill was placed directly on
the Cave Basalt and overburden.
• Cracks and crevices in the basalt were
filled with granular material using
compaction equipment or water sluicing.
 Construction History
• The canal bottom was lined with
impervious compacted Zone 1 material
varying from 12 to 54 inches thick.
• Near the embankment in the forebay
area the lining was up to 11 feet thick.
• Foundation grouting was limited to the
area near the intake structure and a
“cutoff” line extending 350 feet east
and west of the intake.
 Construction History
• Board of Consultant reports indicate
the grouting was mainly to improve the
foundation and not to provide a water
barrier
• A secondary purpose of providing a
partial cutoff was discontinued after
the Board observed that the first grout
holes had large and widely varying grout
takes.
 Operating History
• The canal was first filled in December
1958.
• Shortly thereafter leakage was observed
coming from the Cave Basalt on the north
bank of the Lewis River, in borrow areas
below the canal, and in some locations
along the toe of the embankment.
• Leakage measurements in mid-December
showed a loss of 118 cfs.
 Operating History
• The canal was dewatered for inspection
and repair.
• Between 50 and 75 sinkholes and cracks
were found in the bottom of the canal
with a significant crack noted in the
bottom of the forebay approximately
100 feet upstream of the intake
structure.
• The sinkholes and cracks were filled with
silty sand and gravel and the Project
returned to service.
 Operating History
• Subsequently the Project operated for
14 years with no serious problems.
• In 1973, “mud boils” were noted below
the canal near Station 151.
• The canal was dewatered on May 21, 1973
and inspections found 65 sinkholes 2 to
over 20 feet in diameter.
• A Board of Consultants, including Dr. A.
Casagrande, J. Hinds and E.R. de Luccia,
was convened to investigate the problem.
 Operating History
• The investigations determined that
fine-grained material in the blanket had
eroded into the underlying cavernous
basalt.
• Piping tunnels were discovered at the
embankment / soil-rock interface.
 Operating History
• The Board concluded that:
– flowing water had carried fines out of the
tunnels leaving behind coarse gravel-filled
tubes and
– the fines filled voids at the toe of the
embankment, increasing the phreatic
surface and ultimately resulting in water
exiting the embankment at elevation 570.
• A decision was made to remove a 140-
foot embankment section near Station
151 and to treat the foundation rock.
 Operating History
• The foundation was cleaned with high
pressure water and a D9 bulldozer was
used to break up the basalt surface to
locate additional tunnels.
• Two large caverns, 10 to15 feet across
and 8 feet deep were uncovered along
with other fissures and cracks.
• The fissures and cracks were filled with
sluiced sand and gravel and the caverns
were filled with lean concrete.
 Operating History
• Concern that other tunnels might exist
beneath the canal led to excavating
three additional trenches parallel to the
embankment along the upstream toe.
• Only a few cracks, fissures and small
holes were uncovered. The defects
were treated similarly to those
originally discovered.
• No further signs of distress were noted
until the failure in April, 2002.
Photo of excavation in canal
 Surveillance and Monitoring
Prior to Failure
• No instrumentation was incorporated into
the original Project to monitor
settlement, movement, seepage
pressures, or seepage quantities.
• After direction from the FERC, the
District conducted leakage tests of the
power canal from 1985 to 1995 to
estimate seepage losses downstream of
the Project wasteway structure.
 Surveillance and Monitoring
Prior to Failure
• Leakage calculations ranged from 11 to
24 cfs for 24- to 64-hour test periods.
• The District concluded from the results
that the seepage was gradually
decreasing over time and the tests were
discontinued in 1994.
 Surveillance and Monitoring
Prior to Failure
• In response to 1995 Independent
Consultant Report recommendations, six
seepage monitoring weirs, nine
piezometers, and five survey alignment
and settlement monitoring points were
established.
• Up to the time of the embankment
failure, the instrumentation data did
not indicate any adverse trends.
 Development of Failure
• 2:18 a.m. - A security guard inspected
the Swift No.2 PH and noted nothing
unusual.
• 3:00 a.m. - The guard drove past the
powerhouse again, nothing unusual was
noted.
• 3:50 a.m. - The Swift No.2
communication remote terminal unit
(RTU) failed.
 Development of Failure
• 4:00 a.m. - A high sump alarm was noted.
An operator was called and indicated he
would check on it in the morning.
• 5:25 a.m. - Telemetry indicated low
water level in the canal. The operator
was called again and he went to
investigate.
• 5:35 a.m. to 5:45 a.m. - A driver told the
guard that he had seen water flowing
across the highway about two hours
earlier.
Swift 2 Initial Flow Pic
 Development of Failure
• 5:40 a.m. – The security guard called the
Hydro Control Center (HCC) and reported a
lot of water flowing across the highway.
• 5:45 a.m. - The operator arrived at the site
and notified the HCC that the forebay
embankment had failed and water was flowing
across the highway.
• 6:55 a.m. - The operator reported a 200-
foot-long washout with water flowing 10 feet
deep across the highway.
Swift 2 Failure Pic
Taken from Report
Prepared by
CH2MHill
 Post Failure Investigations
• Initial inspections showed a large
sinkhole along the north side of the
forebay about 500 feet upstream of
the breach.
• A large crack led from the sinkhole in an
arc directly to the breach area.
• An enlarged crack or tunnel was noted
at a depth of 12 to 15 feet in some
areas.
Sinkhole Pic
 Post Failure Investigations
• Where the primary crack entered the
breach area an oval-shaped tunnel/void
was observed.
• The tunnel was about 16 inches in
diameter and located in a well-graded
sand and gravel layer in the native
material beneath the canal lining.
• The crack could be seen extending
below the tunnel into the native soils.
Crack

Void
 Post Failure Investigations
• Within the underlying basalt tree casts
up to 14 inches in diameter were found.
• Several lava tubes were also identified
in the basalt the largest of which was
about six feet wide and partially filled
with sand, gravel and cobbles.
 Post Failure Investigations
• Additional tunnels, sinkholes and
depressions were noted in other areas
of the canal.
• Although not associated with the
failure, some of these features were
fairly large and water could be heard
running in a few of them.
 Sinkholes

Water Observed
Entering River Limits of Basalt
Flow
 Post Failure Investigations
• The preliminary assessment of the
failure indicated that the breach was
associated with flow through previously
undetected lava tubes in the basalt
foundation leading to a piping/internal
erosion failure of the underlying native
material and, ultimately, the canal
embankment.
 Post Failure Investigations
• After the initial visual inspection and
failure mode assessment a more
detailed exploration plan was developed
including:
– Obtaining additional photo documentation;
– Preparing a detailed survey of the failure
area and other areas with sinkholes; and
– Conducting a subsurface exploration
program including trenches, test pits,
borings, and field mapping.
 Post Failure Investigations
• An initial trench was dug at the location
of the large sinkhole associated with the
breach.
• An east-west trending ridge of Cave
Basalt was exposed about 4 ½ feet below
the original bottom of the sinkhole. The
ridge generally followed the alignment of
the primary cracks leading from the
sinkhole to the breach area.
 Post Failure Investigations
• The Cave Basalt ended abruptly 10 to 20
feet north of and parallel to the north
side of the ridge.
• North of the ridge the basalt-soil
contact dipped underneath the basalt
leaving a rock overhang.
Taken from Report Prepared by CH2MHill
 Post Failure Investigations
• An opening into a large lava tube was
found on the north overhanging slope of
the Cave Basalt.
• The floor of the lava tube was about 12
feet below the top of the basalt ridge.
• The lava tube averaged about 10 feet in
diameter and was about half-filled with
sand and gravel.
• Observations indicated that water had
flowed through the tube in the direction
of the breach.
 Post Failure Investigations
• An additional trench was dug about 70
feet west of the initial trench to help
confirm the direction and boundary of
the ridge and edge of basalt
• The top of the Cave Basalt was found at
approximately the same elevation in this
trench and again no basalt was found to
the north helping to confirm that the
edge of the Cave Basalt flow was
located in the forebay area.
 Post Failure Investigations
• The breach area came to be called the
“amphitheater” as the failure left a
semi-circular basin.
• Test trenches were dug in the
amphitheater to expose the Basalt/Pre-
Basalt Alluvium contact.
• The test trenches exposed a large
cavity beneath the Cave Basalt.
Amphitheater Pic
 Post Failure Investigations
• The cavity varied from 12 to 36 inches
high and extended at least 75 feet back
under the basalt.
• The actual length was not determined
due to the limited available light and
undulations in the underlying alluvium.
• To determine connectivity between the
opening in the lava tube and the cavity
under the Cave Basalt, a smoke test was
performed.
 Post Failure Investigations
• Smoke was introduced to the opening in
the lava tube near the sinkhole using a
smoke generator and high powered fan.
• Increased airflow was felt in the
openings beneath the basalt in the
amphitheater within 6 seconds of
turning on the smoke generator; smoke
was noted after about 30 minutes.
 Post Failure Investigations
• The rapid response of airflow indicated a
large, direct, connection from the cavity
to the opening in the amphitheater.
• The delay of smoke arrival indicated a
large opening existed beneath the Cave
Basalt – on the order of 100,000 to
200,000 cubic feet suggesting a cavity
with an area of 1 to 2 acres existed
beneath the cave basalt.
 Failure Mode
• Failure of the Swift No. 2 embankment
was caused by a sequence of events that
began 1900 years ago with the formation
of the lava tube along the northwestern
margin of the basalt flow.
• Construction and operation of Swift
No. 2 raised the phreatic surface in the
post-basalt alluvium and colluvium
overlying the Cave Basalt.
 Failure Mode
• Piping slowly progressed in a near-
vertical direction from the opening into
the lava tube toward the floor of the
canal.
• Due to the buried entrance into the lava
tube and the large volume available for
piped material within the lava tube, it is
believed that piping could not have been
observed until it connected to the floor
of the forebay, at which point it would
have been too late for intervention.
 Failure Mode
• The direct connection of the canal to the
lava tube allowed delivery of high-
pressure water through the void beneath
the basalt to the downstream toe of the
embankment.
• Due to undermining of the Cave Basalt
beneath the toe of the embankment, or
by directly blowing out the toe, the
downstream shell slumped leading to
progressive failure of the structure.
 Lessons Learned
• 50-year old dams are not immune from
failure
– This is a lesson we seem to relearn with some
regularity.
– Past performance is not a perfect indicator of
future performance.
– Many failure modes are slowly developing
– It is situations just like Swift No. 2 that
makes critically re-evaluating all dams on a
periodic basis essential.
 Lessons Learned
• A thorough understanding of conditions
affecting the performance of the dam,
including past performance, is essential
– Several factors contributed to a lack of
thorough and detailed knowledge about the
Swift No. 2 Project among the owner,
regulator and consultants.
• The Project was constructed in the 1950s so few
people remained on the owner’s staff who were
familiar with the construction and early operation
history of the Project.
 Lessons Learned
• The Project was constructed and administered
under the direction of the FERC office in San
Francisco until 1985 when the Portland Regional
Office was opened. The change in FERC Regional
Office oversight resulted in the loss of
institutional knowledge related to the Project.
• A FERC Part 12D 5-year Independent Consultant
inspection of the Swift No. 2 Project was not
required until 1990 because the Project was
classified as Low Hazard Potential leaving a 35-
year gap between construction and the initial
report.
 Lessons Learned
– During preparation for a dam safety
inspection emphasis should be placed on:
• Review of the foundation geology; is it likely to
have open cracks; was the foundation properly
treated during construction; could voids exist in
the foundation materials that could allow piped
materials to be deposited undetected; is the
terrain such that underflow; especially
pressurized flow, could be present?
• Design of the water-retaining structure; is it
susceptible to piping; does it have proper
filters; is it susceptible to blow out?
 Lessons Learned
– During the inspection, talk with local staff.
They are the best source of information on
the area surrounding a project. They often
live in the area and have hunted, fished,
hiked, or otherwise explored much of the
surrounding area.
 Lessons Learned
• Emergency Action Plans work
– Although the Swift No. 2 Project is
classified as Low Hazard Potential, the size
of the impoundment and other factors
contributed to a decision to develop and
implement an Emergency Action Plan (EAP)
at the Project.
– The availability of an EAP provided a clear
notification procedure that resulted in
timely notification of all affected parties.
 Lessons Learned
• Foundations with open joints can provide
pathways for internal erosion of dams
– In situations where there is the possibility of
significant voids in the foundation, such as karstic
or volcanic areas, the possibility of having ongoing
piping that is not visible is very real.
– There are several mechanisms that can lead to
piping failures of dams including piping material
into openings in the foundations. Notable failures
and incidents include:
• Teton Dam
• Logan-Martin, and
• Baldwin Hills Dam
View publication stats

Lessons Learned
• BE VIGILANT
– A well trained staff with awareness of the
potential failure modes associated with the
dam is the best early warning system
possible. Encourage project staff to be
aware of changes in conditions that may be
precursors to a dam failure.