Dams Without FiltersMichael J. Gobla, P.E.
Presentation OverviewReclamation ExperienceEvaluation using Potential Failure ModesSeepage Incidents – Is it an Emergency?Filters and DiffusersCase HistoriesConclusions
Presentation ObjectivesProvide an overview of how to use potentialfailure modes to evaluate a dam without filtersUnderstand the varied nature of seepageincidents and lessons learnedBe able to better communicate essentialinformation about a potential incidentUnderstand how to build a filter in response to aseepage incident
Filters at ReclamationMost Department of the Interior dams (BOR, BIA,BLM, FWS) do not have a filter.Almost every Interior Dam has been evaluated forpotential failure modes and risk.The decision to add a toe drain, filter, and/orcutoff wall is risk based and considers both thesite specific case and other dams in need toprioritize limited funds. In many cases thestatus is no action at this time (subject tomonitoring, periodic inspection, and review).
Reclamation ExperienceBelle Fouche Dam, SD – No filters, clay dam, highfoundation seepage in 1910 during first fill.Reclamation built its first toe drain, no additionalincidents in 100+ years of monitoring.Fontenelle Dam, WY – Outer filter zone (silt, sand, gravel)not designed. Silt core on open bedrock (somegrouting). Near failure on first fill. More grouting, didnot understand need for engineered filter & drain.Seepage incident in 1980 upon second fill, built cutoffwall in 1985-87. No more incidents.Teton Dam, ID – (Improper filter like Fontenelle). Failedon first filling in 1976. Conclusion – Embankmentdams need properly engineered filters and drains!
The previous slide in cross section:Belle Fouche DamOpen jointed clay pipesurrounded by screened gravel,surrounded by unscreenedgravel. Monitor seepage.Fontenelle DamConcrete cutoff wall 2 ft thickpenetrates 20 feet into thestress relief open bedrock toreach intact bedrock.Cutoff WallTeton DamFailure changed Reclamation CultureEstablish minimum design standards.Internal and external peer reviews.Designers must field approve foundation.
Current UnderstandingSeepage forces are always at work.Internal erosion can go on for years but even withseepage monitoring it is often difficult to detect.Progression towards failure may have begun. Very lowhorizontal gradients can cause internal erosion.Failure takes time but could be minutes, to hours to daysfrom the time of detection of a problem, one mustquickly get an understanding of what is happening andwhy.Problems can develop at any time despite years of goodperformance.
Number of Internal Erosion Incidentsat Reclamation EmbankmentsType ofInternal Erosion# of Incidents withParticle Transport# of All IncidentsEmbankment only 4 8Foundation only 30 69Embankment intoFoundation 4 7Into/AlongConduit 5 5Into Drain 12 12TOTAL 55 101
Years to IncidentYears to IncidentInternal Erosion HistoryInternal Erosion Incident History1009080706050Series14030201001900 1920 1940 1960 1980 2000 2020Year Completed1/3 of incidents - Less than 5 yearsYear Completed or Modified
Years to IncidentInternal Erosion Incident History1009080706050Series14030201001900 1920 1940 1960 1980 2000 2020Year Completed or ModifiedTeton Dam
Failure of dam attributed to failure of leadcaulking at valve-pipe connection.A robust filter and drain may haveprevented failure.
How do you decide what is needed?Risk based evaluation of potential failure modesImperviousblanketTransition zoneRiprap and beddingUpstreamshellCutoff wallImperviouscoreCutofftrenchChimney drainChimney filterDownstream shellSeepage stability bermTrenchfilterBlanketToedrainDraintrenchDrainage ditchRelief wellDam Components for Seepage Control
Why Risk Analysis• No single person or small group can compare thesafety issues of so many dams.• We need some consistent quantitative measure bywhich to judge priorities.• There must be a balance between public safety andthe cost to taxpayers and water users.• Risk analysis lets us focus our efforts to achieve thegreatest risk reduction for a given amount offunding.
How is it Accomplished?• Step 1. Understand the dam geology, construction,and performance history. Review records, talk withoperators, inspect the dam.• Step 2. Brainstorm potential failure modes. Usecritical thinking, do not just copy what others havedone. Each dam is unique. Sketch the seepagepathways, write out detailed descriptions of theprocess.• Step 3. Evaluate the potential failure modes andconsequences. Use of event trees is helpful.• Step 4. Get independent peer review and revise.• Step 5. Make a decision based on potential for failureand risk.
Potential Failure ModesAre Related to Loadings:• Seepage• Earthquake• Flood
Some Seepage RelatedPotential Failure Modes• Internal erosion along the outlet conduit.• Internal erosion through the embankment.• Internal erosion through the foundation.• Flow erosion through a stress crack.• Blowout of the toe due to high pressure flowthrough the foundation.• Internal erosion into a flaw in the conduit.• Internal erosion along the spillway wall.• Internal erosion of embankment into foundation.• Erosion due to water exiting a flaw in theconduit.
Some Earthquake RelatedPotential Failure Modes• Flow erosion through embankment cracks• Flow erosion through fault offset.• Foundation liquefaction causes slumpingand crest overtopping.• Structural Damage to Walls, Gates etc.leads to flow erosion failure.
Some Flood RelatedPotential Failure Modes• All of the Seepage Issues are possible andshould be evaluated for a flood elevatedreservoir. Also:• Seepage can be aggravated by a floodelevated reservoir (May not be first fill)• Consider Dam Overtopping• Consider Spillway Failure
ComponentEvents• What loadingmakes failurestart?• How and wheredoes it start?• And then whathappens?• And then what?
1. Assign probabilities to eachcomponent. Except for load, eachcomponent probability assumesprevious events/conditions haveoccurred. Load probability is theprobability of occurrence in a randomyear.2. Calculate probability for all branchesleading to failure of the dam, and theoverall probability of failure.
An Event Tree Is...• A logical and/or chronological depictionof all events and conditions required tocause failure of the dam by one or moremechanisms, decomposed into smallevents that are easy to understand andestimate probabilities for.Earthquake∩ High Res.]YNYSlopeInstability?NoFailureN21Dam isovertopped?NoFailureYNFailureNoFailure
Event Tree for Internal ErosionYFailureYReservoirFills?NErosionBegins?NoFailureYNUnfilteredExit?NoFailureYNRoof canform?NoFailureYNYFlow NotLimited?NoFailureNInterventionFails?NoFailureNNoFailure
Verbal Probability Scale• Virtually Certain• Very Likely• Likely• Neutral• Unlikely• Very Unlikely• Virtually Impossible• 0.999• 0.99• 0.9• 0.5• 0.1• 0.01• 0.001
0.4ReservoirFills?NNow, With the Numbers0.001ErosionBegins?NoFailureNUnfilteredExit?NoFailure0.3NRoof canform?NoFailure0.9N0.1Flow NotLimited?NoFailureNInterventionFails?NoFailure0.5NFailureNoFailure0.4 x 0.001 x 0.3 x 0.9 x 0.1 x 0.5 = 5 x 10 -6(Therefore, below guideline level for APF.)
WorldwideStatisticsMode of Failure% Total Failures(where mode offailure known)% Failures pre1950% Failures post1950Overtopping 34.2 % 36.2 % 32.2 %Spillway/gate (appurtenant works) 12.8 % 17.2 % 8.5 %Piping through embankment 32.5 % 29.3 % 35.5 %Piping from embankment intofoundation1.7 % 0 % 3.4 %Piping through foundation 15.4 % 15.5 % 15.3 %Downstream slide 3.4 % 6.9 % 0 %Upstream slide 0.9 % 0 % 1.7 %Earthquake 1.7 % 0 % 3.4 %Totals (3) 102.6 % 105.1 % 100 %Total overtopping and appurtenantworks 48.4 % 53.4 % 40.7 %Total piping 46.9 % 43.1 % 54.2 %Total slides 5.5 % 6.9 % 1.6 %Total no. of embankment damfailures (exc. During construction)Total embankment dam yearsoperation (up to 1986)124 61 63300,400 71,000 229,400Annual probability of failure 4.1 x 10 -4 8.6 x 10 -4 2.7 x 10 -4
Proposed “Best Estimate” Values ofAnnual Probability of Initiation• Applies only to Reclamation embankments• Values serve only as “starting points” – each dam isunique and must be evaluated separately, looking at sitespecific features/vulnerabilitiesType of Internal ErosionRange of Initiation Probability(Best Estimate)Embankment only 3x10 -4 to 1x10 -3Foundation only 2x10 -3 to 1x10 -2Embankment into foundation 3x10 -4 to 7x10 -4Into/along conduit 4x10 -4 to 1x10 -3Into drain 1x10 -4 to 1x10 -3
Annual Failure Probability, fReportingthe Results1.E-011.E-021.E-031.E-041.E-05IncreasedJustificationfor Action toReduce RiskGreatlyIncreasedJustificationfor Action toReduce Risk0.11.E-011.E-021.E-031.E-041.E-050.01f-N Chart1.E-061.E-071.E-08IncreasedJustificationfor Action toReduce Risk0.0010.00010.000011.E-061.E-071.E-081.E-091.E-090.1 1 10 100 1000 10000 100000Loss of Life, NStatic-*Piping Through EmbankmentNotes:Static-*Piping Embankment into Alluvium* Indicates risks are estimatedStatic-*Piping Embankment at Bedrock Contactto be de minimis, and are notHydrologic-*Overtoppingshown.Hydrologic-*Piping due to SurchargeSeismic-OvertopppingSeismic-*Cracking and ErosionSeismic-*CoseismicSeismic-Spillway GatesSeismic-Spillway WallsTotal Static Risk EstimateTotal Hydrologic Risk EstimateTotal Seismic Risk EstimateTotal Probability of Failure - All Loadings
We do not always have time to do anevaluation!Most calls come in on a Friday mid-day to lateafternoon or before a holiday weekend.Why: They may have known about it for a few days butare afraid to let it go unaddressed over the weekend.If they call it in it becomes someone else'sresponsibility, they are off the hook. (or so theythink)
The CallThe call can be from anyone, usually it’s the dam operatorA. V. Watkins Dam – Rancher noticed water in his field.Red Willow - A person from the inspection team fell into a sinkhole.All American Canal – My boss called me at home 9 days after theearthquake, need to go look at some seepage.Weber Siphon – Construction Inspector - we have a flood, is it aproblem?Lake Capote – BIA Dam Safety Officer relayed a message from theTribal caretaker about seepage changes.Box Butte – Dam Tender checked yes on the inspection checklist
Take Written Notes:Key QuestionsFirst - Get the callers name and telephone number.What is the problem?Can they describe it in terms of amount of flow – donot rely on gallons per minute estimates. Ask inrelation to common things like is it more or less thanthe flow from a garden hose or a faucet. If more howmuch more? (three times, ten times). If less howmuch less (1/2, ¼)Get pictures, close up, far away, from different angles.Give them your contact information – email, cell.
Key Questions• Where is it?• Is the water cloudy or muddy?• Are piles of sediment forming?• Is the water flow increasing?Caution: soil erosion issometimes imperceptible• Are the seepage areas expanding in extent?• What is the Reservoir level and recent history?• What’s the weather forecast?
Key Questions• Do they have equipment available?• What kind?• How long will it take to get it to the problem area?• Are operators available?• Can we get people there at night if needed?• Can they get a light plant?
Key Questions• Do they have sand and gravel to build a filter?• Where is it? Exactly what is it, are gradation plotsavailable, how much is available?• If no, where can we get material from?• Is there a local gravel pit or concrete supplier? Howfar away, what is the name and telephone number ofthe company?• Who else is available to work on this? What is theircontact information?
Key Questions• What instruments are available at the dam? (flowmonitors, piezometers, observation wells, etc.)• What do they show, when were they last read?• How are we monitoring the problem?• Do we need additional monitoring?• Round the clock observation?• Should we take action now?
Common Factors in Many Incidents• Erodible embankment/foundation soils. Low PIsilts, lean clays, fine sands, dispersive soils,collapsing soils• Backward erosion piping mechanism is common.• Often along a penetrating structure such as outletworks conduit, spillway wall, or floor slab.• Internal erosion can initiate under low hydraulicgradients (less than 0.08)
Response Team• Begin assembling a project team and set a specifictime to have a conference call (depends on urgency)• Assemble Key Information – EAP, risk evaluations,current instrumentation data, records of pastinspections, design documentation.• Identify other team members that may be neededsuch as Contracting Officer for emergency spending• Document discussions and decisions with email tothe entire team. Include action items and assign aresponsible person and schedule deadline for each.
Actions to Consider• Measure flow rate• Determine aerial extent (changing with time)• Measure for sediment transport• Increased inspections (up to 24hr)• Bring in equipment and filter materials• Build a filter• Lower the reservoir• Maybe inform authorities
FiltersConsist of a filter material, a drain, and a coverFilter – Its function is to stop the loss of soil particles(silt, clay, fine sand size material) but allow seepage flowto continue. If you block the seepage it will find a newroute to take or may build up and cause slope instability.Drain – Its function is to quickly remove the filtered waterso the seepage does not cause a rise in the ground waterlevels inside the embankment. Can also be a cover.Cover – Function is to protect the filter and drain from theelements and to provide weight so a surge in seepageflow will not displace the filter. (Washakie Dam)
When the flow is too swiftIf the leak is so great that it just pushes the filter sand asidewhen applied then you need to take other actions to slowthe seepage and allow a filter to be placed.Diffuser – Gravel or riprap used to spread out (diffuse) theflow and provide weight to support the remainingembankment.Back Pressure – A berm or coffer dam placed to create apond of water at the exit point of the seepage to putsome hydraulic back pressure on the seepage.Plug – Material placed in the upstream side of a seepagepath to block the flow. Sand, riprap, etc.
FiltersPrefer C-33 Sand with not more than 5% fines (minusno. 200 sieve size material) This will filter most soils(not the dispersive soil at Many Farms)Sand from wind deposited dunes or weathering ofsandstone is usually too fine to be an effective filter.A non-woven geotextile filter cloth covered with gravelmay also work in an emergency, but the C-33 sandworks better.Stockpiles at LakeCapote
Fontenelle Dam Wyoming1965 Fontenelle Dam WYnearly failed during firstfill due to seepage throughcracks in the abutmentrock causing the siltyembankment soil to erode.The reservoir was in floodstage with the spillwayflowing when the muddyseepage started.
Fontenelle DamThe dam was saved by:1. Riprap from the dam was thrown into the growing downstreamcavern, this providing support so the dam crest did not collapse.2. All outlets were opened full to quickly lower the reservoir.
Teton Dam – Seepage FailureJune 5, 1976Dozing fill and riprap into theupstream whirlpool had no effect.
A typical incidentSeepage incidents at a dike. In 2005 a 150 ft. longcutoff wall was installed. Additional seepageincidents In March 2011 and July 2011
Placing a sand filterRemoving topsoil July 1 2007Installing sand filter
Drain and CoverPlacing gravel drain.Completed filter withgravel drain as thecover material.
Earthquakes can cause seepage
How do you know when to take action?• Increase Monitoring?• Build a Filter?• Lower the Reservoir?• Evacuate Downstream Population At Risk?
A.V. Watkins Dam 2006• A “roof” of hardpan or caliche existed aboveerodible sands – allowed piping and masked earlydetection• Likely occurred at very low head to seepage pathlength evidenced by the horizontal flow pathway.• Occurred after many yearsof safe operation.
A.V. Watkins DamNOT TO SCALEInternal erosion through the foundation led to extensive erosionof material into a downstream ditch, with backwards erosionunder the dam leading to sinkholes and uncontrolled erosion
Conclusions• Seepage forces never stop acting as long as areservoir is present.• Dams lacking engineered cutoffs, filters, anddrains should be considered safety deficient.• Acceleration of seepage flow and/or materialremoval is a dangerous sign, you need to act now.• Regular Inspection and seepage monitoring areessential requirements for safe dam operations.
Conclusions• Factors that can prevent a complete failure includefiltering, self-healing, collapse of a developingpathway, inability to form a complete dam breach,and successful human intervention.• HOWEVER, each and every incidence of initiation ofinternal erosion needs to be taken seriously.• The high rates of internal erosion incidents atReclamation and BIA facilities is due to the lack offilters and the presence of easily erodible soils in ourWestern dams. Gradients as low as 0.08 can erode.