Geotechnical Instrumentation News

Geotechnical Instrumentation News

John Dunnicliff

IntroductionThis is the forty-seventh episode ofGIN. The episode includes discussionsof an article on embankment dam in-strumentation, which was in the previ-ous episode of GIN, together with theauthor’s reply. These are followed bythree articles by regular contributors toGIN: design of pile foundations, byBengt Fellenius; cost of geotechnicalinstrumentation, by Gord McKenna;and guidelines on selecting electricalcables, by Barrie Sellers. Thank you toall three for your willingness to helpkeep GIN alive.

Some Disappointing TotalStress and Pore Water PressureDataThe previous episode of GIN includedan article by Ali Mirghasemi, describ-ing measurements of total stress andpore water pressure at Karkheh Dam inIran. He told us about his experienceswith a large number of earth pressurecells installed in the core of the embank-ment dam, and concluded that “no con-sistent data were achieved”. He also de-scr ibed pore water pressuremeasurements with both openstandpipe and vibrat ing wirepiezometers which were inconsistent,and said, “The author will welcomeany comments and discussion thatmay help to explain the differences”.

I was both puzzled and fascinated bythe contradictions, so sent Ali’s articleto several colleagues prior to its publi-cation, asking them whether they couldshed any light on the issues. I’ve re-ceived discussions from ElmoDiBiagio, Erik Mikkelsen, Arthur Pen-man and Barrie Sellers, and one from

Louis Marci l of Roctest , themanufacturer of the instruments thatwere installed at Karkheh Dam. I’vealso received an unsolicited discussionfrom Donald Babbitt. All six discus-sions are published here, together withthe author’s reply. This has been, forme, an interesting and rewarding expe-rience, as all the discussions are rele-vant and helpful. Thank you to all thediscussers for making the effort to help.

I’m left with two ‘umbrella’ ques-tions:1. What is the best way for manufac-

turers to help users select the mostappropriate instruments for theirapplication? Manufacturers ofgeotechnical instruments are a re-source that should not be over-looked when selecting appropriateinstruments for any particular pro-ject. However, users can’t expectthat manufacturers will put them-selves in the same position asgeotechnical project designers. Itseems to me that if users have anyuncertainties about which instru-ments are appropriate, there needsto be interaction with manufacturersso that the decision can be made mu-tually. Is this practicable?

2. How can manufacturers help us-ers to learn about the most appro-priate installation methods? Forexample, in the case of KarkhehDam, earth pressure cells andpiezometers. I’ve always contendedthat manufacturers can provide ex-plicit details about such installationsas strain gages on steel, becausethere are no geotechnical variables.However, manufacturers cannot

provide explicit details on how to in-stal l their instruments ingeotechnical surroundings, becausethey can’t possibly know all the de-tails that are required to do so—thisis the job of the user. But of coursemanufacturers can provide generalguidelines. In the experience ofmanufacturers and users, is there aneed to improve interaction to en-sure the most appropriate installa-tion methods? If yes, how?I’m not intending to close the op-

portunity for discussions of AliMirghasemi’s article with this epi-sode of GIN. If readers have more in-put that might help to explain thecontradictions, or if any feel like putt-ing forward some answers to theabove two ‘umbrella’ questions, Ivery much hope that you’ll do so.Watch this space!

Design of Pile FoundationsBengt Fellenius, a regular contributor toGIN, has sent me an article that resolveslack of understanding relating to the de-sign of piles subjected to drag loads.

Cost of GeotechnicalInstrumentationAnother regular contributor, GordMcKenna, gives us some pragmaticrules of thumb that are applicable to thecost of geotechnical instrumentation.

Help with Selecting ElectricalCablesThe article by Barrie Sellers, yet an-other regular contributor to GIN, givesnuts-and-bolts guidelines to help us se-lect which electrical cable might be

Geotechnical News, June 2006 33

GEOTECHNICAL INSTRUMENTATION NEWS

most suitable for each particular appli-cation.

Two Reminders for Your DateBookThe next instrumentation course inFlorida will be on March 18-20, 2007 atSt . Petersburg Hil ton (www.stpetehilton.com). Details of the coursewill be on www.doce-conferences.ufl.edu/geotech as soon as they areavailable.

The next international symposium,Field Measurements in Geomechanics(FMGM), will be held in Boston, Mas-sachusetts on September 24-28, 2007,at the Park Plaza Hotel. W. Allen Marrof Geocomp Corporat ion([email protected]) and Prof.Charles Dowding of Northwestern Uni-versity ([email protected]) are serv-ing as co-chairs of the organizingcommittee. More information will soonbe available on www.geoinstitute.org,under the “Upcoming Events” tab. Any-one interested in helping to organizeand carry out this event is encouraged tocontact Allen or Chuck directly.

For more information about FMGMsymposia, please visit www.fmgm.no.

‘Meaningful’ BooksThis has nothing at all to do withgeotechnical instrumentation (but nei-ther has cricket!). I’ll give my excusefor including it later.

The most recent topic for discussion

at a local ‘discussion group’ was, “Thethree most meaningful books that we’veread” - the word “meaningful” to be in-terpreted in whatever way a personwished. My books were:• A Fortune-Teller Told Me, by

Tiziano Terzani. Non-fiction.Terzani, a middle-aged Italian, wasthe Middle East correspondent for aGerman publication, and had beenso for many years. His life wasplane-taxi-hotel-taxi-plane. On awhim, he went to a Chinese for-tune-teller in Hong Kong who, hav-ing established credibility byrecounting a strange and unique ex-perience that Terzani had had previ-ously, told him that under nocircumstance should he fly during awhole year 16 years hence – “noteven once”. Despite the huge impacton his normal life, Terzani eventu-ally accepted the prediction ofdoom, and spent that year travelingand reporting throughout the MiddleEast, without flying. His tales aboutpeople and places are extraordinary.And he visited fortune-tellers when-ever he could! I won’t spoil it for youby telling about any air accidentsduring that later year! Now to my ex-cuse – it was Giorgio Pezzetti, whotook the lead role in organizingFMGM-1995 in Bergamo, Italy,who recommended the book to me,so there is an ‘instrumentation con-

nection’!• A Prayer for Owen Meany, by John

Irving, who is probably best knownfor “The World According to Garp”and “The Cider House Rules”, bothof which were made into films. Fic-tion. Impossible in a few words tosay why I think this is so very mean-ingful. One reviewer wrote, “Whatbetter entertainment is there than aserious book which makes youlaugh?” Religion, precognition, andan unforgettable and endearing leadcharacter. John Irving at his best.

• Ishmael, by Daniel Quinn. Fiction.A gorilla is released by his owner,and sets himself up in rented space asa ‘teacher’ – in essence a psycholo-gist. Perhaps it can be belittled as abit of pop-psychology, but a mean-ingful reversal of human/animal re-lationships nevertheless.

ClosurePlease send contributions to this col-umn, or an article for GIN, to me as ane-mail attachment in MSWord, [email protected], or byfax or mail: Little Leat, Whisselwell,Bovey Tracey, Devon TQ13 9LA, Eng-land. Tel. and fax +44-1626-832919.

Here’s mud in your eye! (USA). Thanksto Charles Daugherty for this. But itsounds a bit Irish to me!

Discussions of“Karkheh Dam Instrumentation System –Some Experiences”

Ali Asghar Mirghasemi

Geotechnical News, Vol. 24 No. 1,March 2006, pp 32-36

Donald H. Babbitt

In the mid 1960s, 27 - 18-inch diameterCarlson type electrical soil-stress me-ters were installed in core and transitionzones of 770-foot high Oroville Dam

and 15 - 30-inch diameter custom builtstatic/dynamic electrical stress meterswere installed in the downstream shellzone (O’Rourke 1974, Cal DWR 1974).

The installation procedures were simi-lar to those used in Karkheh Dam andthe stress readings obtained were alsoquestionable.


34 Geotechnical News, June 2006

Kulhawy and Duncan (1972) per-formed a finite element analysis of theembankment and compared the resultswith the instrument readings. Theyfound good correlations with the mea-sured movements, but not the stresses.The extremes in their Table 2 are a ma-jor principle stress in the downstreammatch within 10 percent and a factor of7.5 variance in minor principal stress inthe upstream transition zone. They pro-vide compelling arguments of why thestress measurements could not becorrect.

Two of Mirghasemi’s three issues fornot being able measure stresses inKarkheh Dam, the inclusion effect anddifferent compaction, likely apply toOroville Dam. His third issue, possiblerotation of the stress meters, is unlikely,because Oroville is a coarse, stiff em-bankment. The core contains cobbles

up to 3 inches in diameter and cobblesin the transition and shell zones aregreater than 6 inches in diameter. Thiscoarseness led to a third possible sourceof error; the stress meters were sur-rounded by 2 inches of selected fine ma-terial to protect them and to precludecoarse particles producing unrepresen-tative stress concentrations at the facesof the meters.

The questionable stress readings thatultimately developed were a concernduring the very thorough design pro-cess, that included discussions with Dr.Carlson; and meticulous care was usedin installing all the instruments in thedam.

ReferencesCal DWR (1974). California State Wa-

ter Project, Bulletin 200, Volume III,State of California, Department of

Water Resources, Sacramento, No-vember

Kulhawy, F.H. and Duncan, J.M.(1972). Stresses and Movements inOroville Dam, Journal of the SoilMechanics and Foundation Divi-sion, ASCE, July

O’Rourke J.E.(1974). Performance In-strumentation Installed at OrovilleDam, Journal of the Soil Mechanicsand Foundation Division, ASCE,February

Donald H. Babbit, Consulting Civil En-gineer, 3860 West Land Park Drive,Sacramento, CA 95822 USA.Tel. ( 916) 442-0990,email: [email protected]

Elmo DiBiagio

The article by Ali Mirghasemi is a vividreminder that implementation ofgeotechnical instrumentation projectsis not always an easy task. This is espe-cially true for what I consider to be oneof the most difficult of all measure-ments, i.e., the measurement of totalstress in an embankment dam. My com-ments are, therefore, restricted primar-ily to the earth pressure measurementsreported by the author.

The article does not contain suffi-cient measurement data to evaluate thecorrectness of the stress measurementsor to pin-point the reasons for the re-ported inconsistent data. It particular, itwould be helpful to study the perfor-mance of the pressure cells in the lowerpart of the dam during placement of,say,10 to 20 meters of fill over the in-struments. In the lower portion of anembankment dam, the stress field dur-ing construction is essentially 2-dimen-sional if the valley is wide and notpredominately V-shaped. In this case, itis possible to calculate from the densityand thickness of the fill what the mea-sured stresses should be. This proce-dure is commonly used to check the

performance of pressure cells installedin high embankments.

The author lists three reasons for theinconsistent earth pressure data, distur-bance of the stress field caused by theinstrument itself, problems associatedwith compaction of the surrounding filland rotation of the instruments duringsubsequent compaction of layers overthe instruments. I agree with these con-clusions. However, in my opinion theprincipal reason is the manner in whichthe instruments were placed in the em-bankment. My conclusion is basedsolely on the photograph, Figure 4,which shows the pressure cells being in-stalled in small rectangular holes orslots with near vertical sides excavatedin the embankment. Differences incompaction (and compressibility) ofbackfill placed over the instruments rel-ative to the surrounding soil will lead toarching in the overlying fill resulting ineither measured pressures that are toohigh or too low. A better installationprocedure would be to form a wide bot-tomed excavation with flat sloping sidesin all directions and then install the in-struments in the bottom of the excava-tion, not in small holes as shown in

Figure 4. The size of the excavation de-pends on the number of instruments tobe installed and a significant amount ofearthwork may be required. For exam-ple, for installation of five pressurecells, an excavation of the order of 10 mx 10 m and 1 m deep would, in myopinion, be adequate.

The potential for rotation of the pres-sure cells in the core of Karkheh Dam isvery likely because the core material (amixture of 60% high plastic clay and40% gravel) is indeed plastic and wouldbe subject to large shear deformationsduring compaction or as a result of thekneading action in the fill caused byheavy construction traffic. On one NGIproject we fitted miniature inclinom-eters to earth pressure cells installed inthe core of a dam with a moraine core.In one zone where the moraine had ahigh water content because of a rainshower during installation of the instru-ments one of the pressure cells rotated20o.

In summary, measurement of totalstress in an embankment is extremelydifficult and should not be done unlessabsolutely necessary and, if necessary,particular care must be given to installa-



tion details. For a large project likeKarkheh Dam (510 earth pressure cellsinstalled) the cost of a full scale trial em-bankment to study installation detailswould be justified.

I wonder if the author misunderstoodthe drawing in Figure 3 when installingthe instruments. It appears that he mayhave had a large excavation with slop-ing sides but dug holes in the bottom forinstallation of the instruments. Thedrawing can be interpreted in two ways,i.e., instruments placed on the floor ofthe excavation or set in cubical socketsbeneath the floor.

I would like to make one brief com-ment on the reported discrepancy be-tween pore pressures measured by openstandpipe piezometers and vibratingwire piezometers. Again, I don’t feelcompetent to comment in depth on this

without seeing all the measurementdata, in particular the initial measure-ments prior to filling of the reservoir.However, I would like to make one im-portant observation. The vibrating wirepiezometer used at Karkheh Dam,Model PWS, has a pressure sensor witha circular membrane mounted inside acylindrical housing. The housing is rel-atively thin-walled and in the manufac-turer’s catalog the housing is describedas a “slim housing” model. In a highembankment the earth pressure actingon embedded instruments can be verylarge, so large in fact that deformationof the housing of an instrument like theModel PWS piezometer may actuallydeform the pressure sensing element in-side the instrument as well. The end re-sult could be a shift in zero-point and/ora change in the calibration characteris-

tics for the sensor. I illustrated this pointonce at an instrumentation course. Iplaced a thin-walled vibrating wirepiezometer on the floor and put my footon it. The zero point changed by a verysignificant amount. The lesson to learnfrom this is simply: Install robust instru-ments in high embankments.

I wish to express my thanks to AliMirghasemi for reminding us thatgeotechnical instrumentation is not al-ways straightforward.

Elmo DiBiagio, NorwegianGeotechnical Institute (NGI), P.O. Box3930 Ullevaal Station, N-0806 Oslo,Norway. Tel.: (+47) 22 02 30 00,Fax: (+47) 22 23 04 48,email: [email protected]

Louis Marcil

In his article, Mr. Mirghasemi reportsdiscrepancies in results between twopairs of neighbouring piezometers in-stalled at the Karkheh dam. At the re-quest of Mr. Dunnicliff, we submithereafter a few comments derived fromour instrumentation experience. It ishoped that they will help evaluate possi-ble reasons for these discrepancies.

In Figures 5 and 6 of his article, Mr.Mirghasemi presents readings obtainedfrom two pairs of neighbouringpiezometers, each comprising an openstandpipe and a vibrat ing wirepiezometer. In both cases, readingsfrom the vibrating wire piezometers areof lesser value by approximately fivemetres than the ones from the standpipepiezometers. This difference is constantover a period of four years. Assumingthat the standpipe piezometers provideaccurate readings, one would betempted to conclude that the divergenceis due to a problem of defective func-tioning of the vibrating wire instru-ments. Our experience shows that this ismost improbable. From a point of viewof design, vibrating wire sensors cannotproduce such a response with a constantnegative offset over such a long periodof time. This mode of instrumentation

provides either very good results, read-ings that are unstable, or no readings atall. In fact, the reliability and long-termdurability of the vibrating wire princi-ple are the reasons for the widespreaduse of this mode of instrumentation.

The accuracy of the piezometers atKarkheh dam is ± 0.5 % for a full scaleof 1700 kPa, which means that this fac-tor could contribute to no more than ±0.85 metre of the actual difference of 5metres.

Amongst causes of error in readingsoccasionally mentioned to us, such asvoltage surge due to lightning, water in-filtration, accidental overpressure afterinstallation of the piezometer, and zeroreferencing, only the last of these canhave a significant effect on the actualdifference in readings. It must be re-membered here that vibrating wirepiezometers are instruments that pro-vide relative measurements - not abso-lute - so that each reading must bereferenced to an initial reading consid-ered as the zero reading. This zero read-ing must be taken after completedissipation of the overpressure momen-tarily induced by the installation of thehigh air entry filter on the piezometer.This may take over 24 hours. In the

present case, we recommend referringto the initial readings - the readingstaken before and after installation of thefilter - and to quantify accurately the un-dissipated overpressure in order to de-termine the importance of this cause oferror.

Other causes of error should also beconsidered. For instance, measure-ments done in standpipe piezometers,though generally considered reliable,must be verified. This should includemeasuring the tape of the water level in-dicator as well as the stickup of thestandpipe, and making sure that thesealing around the standpipe is efficientin order to make sure that the standpipepiezometers measure the pore pressurein a given point, as the vibrating wirepiezometers do, and not the level of thefree water surface. The peculiar fact thatthe difference in readings between bothpairs of piezometers are equivalentraises the point that further investiga-tion is required as to what these pairshave specifically in common, e.g.where and how they were installed, andhow they are read.

From our experience, it is most im-probable that the discrepancies be es-sentially explained by the vibrating



wire piezometers themselves. All ma-nipulations related to both types ofpiezometers should be considered. Ide-ally, potential effects of each cause

mentioned in this discussion should beverified and quantified.

Louis Marcil, Product Line Managerand Technical Services, Roctest Ltd,

665 Pine Street, Saint-Lambert, QC,Canada, J4P 2P4. Tel: (+1450)465-1113, Fax: (+1450) 465-1938,email: [email protected]

P. Erik Mikkelsen

I thank the author for sharing his experi-ences, especially the disappointing andpuzzling observations. My experienceswith free field total pressure cells havealso been disappointing and the cost oftheir installation in the clay core of adam is not justified, in my opinion. Theresults are rarely reliable.

The reliability of piezometer datahowever is another matter. Piezometerdata are generally reliable and I wouldlike to comment on why, however puz-zling, open standpipe piezometers(OSP) and vibrating wire piezometers(VWP) or electrical pressure sensorsare not measuring the exact same porepressure. Your article indicates thefollowing:• Figures 5 and 6 show quite consis-

tently, regardless of the difference inmeasured pressures by OSP andVWP, that the upstream pore pres-sures are lower than the downstreampore pressures.

• The measured pressures are alsogenerally higher than the reservoirlevel.

• The difference between OSP andVWP readings are greater at lowerreservoir levels.From this I conclude that there are

considerable amounts of air in thepores and that an anticipated seepagegradient (saturated) through the corehas not been established at elevation165 m. The pore pressure is being influ-enced by three factors:• Vertical embankment stress and re-

sulting consolidation• Variable stress from reservoir pool

causing consolidation• Migration of pore water and air due

to above stresses and hydraulic gra-dientThere are several types of moisture

or water in the compacted clay in addi-tion to the air, the most significant here

being the adhered water and the free(capillary) water containing air asshown schematically in Figure 7. Theadhered water is a film covering the soilparticles, and forms strong menisciforces in the corners where particlesmeet and contain no air. Most literatureon the subject seems to explain this interms of natural conditions in the capil-lary zone of low permeability soils

where suction is present. I have not seendiscussions on this subject for higherpore pressures (45 m), but it seems thatin view of the results for Karkheh Damthere still is a significant difference athigher pressure, and that is what isbeing measured.

Saturated permeability is signifi-cantly lowered by the presence of air inthe clay. The air has decreased the per-



Figure 7. Schematic of pore air/water measurements with open standpipe and vibrat-ing wire piezometer (VWP).

meability and slowed the consolidationprocess. In this case the standpipe is noteffective as a vent for the pore air as onemight have assumed. In the past, I havethought of the standpipe as being a natu-ral vent for the pore air, but neverthought very much about how the poreair would get there. Based on these mea-surements the pore air is not readily es-caping, at least not yet. The situationwill change when saturation eventuallyoccurs, increasing permeability andboosting consolidation rates.

Therefore the difference of 3 to 5meters of head (30 to 50 kPa) at eleva-tion 165 m may be caused by the differ-

ence in the way the two instrumentinstallations include or exclude the poreair pressure as shown in Figure 7. Thewater in the standpipe is connected tothe free water in the clay. The pressurein the free water is controlled by thepore air pressure as long as airremains.

The VWP on the other hand was in-stalled with a high air entry filter, care-fully under fully saturated conditions.This would set up the potential for beingable to measure just adhered water pres-sure. The adhered water pressure andthe tension in the corner menisci be-

tween particles support the free poreair/water pressure. This condition con-tinues through the high air entry filter,excludes a portion of the pore air/waterpressure and allows only the adheredwater pressure to be measured by theVWP.

P. Erik Mikkelsen, Consultant inGeoengineering & Instrumentation,Geometron Inc. PS, 16483 SE 57th

Place, Bellevue, WA 98006 USA.Tel: (425) 746-9577,email: [email protected]

Arthur D. M. Penman

This large dam in Iran has been well in-strumented. It is valuable to hear aboutsuch detailed instrumentation in a largedam. Progress in dam design is greatlyhelped by results from such instrumen-tation and we are grateful to Ali AsgharMirghasemi for his article describingthe construction and instrumentation ofthe dam. He has a problem, he says, be-cause a valuable comparison he madebetween a standpipe piezometer and avibrating wire instrument has shown thestandpipe piezometer reading almost5m head of water greater than thatshown by the vibrat ing wirepiezometers and he seeks an explana-tion for this unexpected difference.

In comparing piezometers one needsto compare like with like. Compactedfill is a non-saturated material whichinitially must have negative pore pres-sures to give it the strength to supportthe heavy placing and compacting ma-chinery. Pore pressures only becomepositive when the weight of overburdenbecomes great enough. The fill usuallycomes from a borrow pit and lumps ofthe original material become incorpo-rated in the fill. At Karkheh the core ma-terial is a mixture of 60% clay and 40%gravel. The gravel was a crushed con-glomerate rock and was added to theclay. This is an important point. The ad-dition of stone to a clay for dam fill isvery difficult to accomplish. It mighthave been passed through a brickmaker’s type of pug mill, but I think the

volumes would be too great for that.The other way would be to spread thestone over the fill during constructionand mix it in with agricultural equip-ment but this would lead to layers ofmore stony fill amongst the clayey part.That aspect combined with the chanceof lumps of original material being in-corporated in the fill means that porepressures at individual points may notbe equal everywhere. Vibrating wirepiezometers measure the pore pressurethat exists around them locally.Standpipe piezometers, on the otherhand, measure an average pore pressureover a considerable volume of fill. AtKarkheh the piezometers had pockets ofunsaturated sand of 15 cm diameter by150 cm long connected to continuouslengths of 15 cm diameter rigid pipe.They were ‘sealed’ by bentonite pelletsdropped around them followed by theaddition of bentonite grout. It is difficultto see how the 15 cm rigid pipe contain-ing a 2 cm pipe rising from the intakefilter could be sealed to prevent porewater from getting along the length ofthe pipes.

Figure 2 indicates that the string ofpipes went down to a bottom intake fil-ter, past the mid height position wherethe comparisons were being made.Were the outer 15 cm rigid pipes com-mon to both piezometer? Exactly howwere they installed? Figure 2 suggeststhat they might have been installed in aborehole made after the fill had reached

full height. In that case the measuredpore pressures could have originatedfrom any height along the length of thepipes. If the added stone was in layers,then higher pore pressures could be col-lected by these drainage layers.

It is most interesting that this valu-able comparison has been made and wemust encourage others to repeat this sortof experiment in a new high dam. Theerrors indicated above can producestrange results but comparisons be-tween different types of piezometers,with intake filters of high or low air en-try, pressed directly into the fill will bevery welcomed by the profession.Standpipes of 2.06 cm can be connecteddirectly to the intake filters and sealedby placing in the fill during construc-tion. Careful hand work is needed toplace and compact the clayey fillaround the pipes and there is always adanger that the hand compaction cannotequal that produced by the heavy ma-chines. But other interesting compari-sons can be made between differenttypes of remote reading piezometers,placed close together. My comparisonmade with high and low air entry intakefilters that I made at Chelmarsh dammany years ago has never been re-peated. [See Penman, A.D.M., Mea-surement of Pore Water Pressures inEmbankment Dams, GeotechnicalNews, Vol. 20 No. 4, December 2002,pp 43-49]. I should have placed thepiezometers much lower in the dam,



when I would expect the two measuredpressures to become the same when theweight of overburden compressed mostof the air out of the fill making an almostsaturated material.

May I add support for the detailed in-strumentation made at Karkheh, the re-sults from which add greatly to ourunderstanding of dam behaviour bothduring construction and subsequently?

Arthur D. M. Penman, Chartered Engi-neer, Sladeleye, Chamberlaines,Harpenden, Herts AL5 3PW. England.Tel and fax: +44-1582-715479, email:[email protected]

J. Barrie Sellers

It is not unusual for earth pressure cellsto give results which do not meet expec-tations.

According to Oosthuizen et al(2003), Høeg (2000) remarked at theICOLD meeting in Beijing that:

“...you should have awfully goodreasons for putting in earth and rockpressure cells in any dam of anykind. We spend so much time andmoney on putting in the cells and wespend much time on not believingthe data we get, because, if we do getdata, we do not understand them...Forget about pressure cells in gen-eral and there is a very good reasonfor doing so.”Quite commonly the measured earth

pressures are less than expected and inmost cases the discrepancy can betraced to poor compaction around cells.It seems likely that this was the casehere, although, it is not clear from Ta-bles 1 and 2 whether the cells un-der-registered or over-registered whencompared with the expected values.

Oosthuizen et al (2003) devised amethod to minimize this problem inearth dams: Installation of the cells be-gins when the fill has reached a heightof 800mm above the instrument level.The instrument location and the cabletrenches are excavated 500mm deep. Apocket, with 45° sloping sides, of only afurther 300mm depth is required to beexcavated at each instrument location.The cells, complete with pinch tubes(see later), are positioned on a thin layerof non-shrink sand-cement grout andnailed in position using the lugs on thecells provided for this purpose. The ex-cavated pocket is then backfilled with aweak concrete (19mm aggregate), in100mm layers, vibrated with a poker vi-

brator. The concrete is in all probabilitystiffer than the surrounding soil so thatit acts as a rigid inclusion and attractsadditional load from the softer sur-rounding. This would tend to make theearth pressure cell over-register,possibly by 5%, maybe up to 15%maximum.

The purpose of the pinch tubes is sothat the cells can be pressurized to en-sure that they are in intimate contactwith the concrete. This is standard pro-cedure when installing cells in concrete,because as the concrete cures heat isgenerated, the cell expands, and then re-turns to its initial shape after the con-crete cools, with the potential ofbecoming disconnected from the con-crete. For this application the pinchtubes may be a bit of overkill, but it ischeap insurance and a good thing to in-clude. After 24 hours the cells are pres-surized, by pinching the pinch tubesuntil the pressure in the cell, displayedon a connected readout box, starts tochange. The instrument location, con-taining the grouted cells and the cabletrench, is then backfilled in 100mm lay-ers. Each layer is compacted by a vibra-tory trench roller. After this, standardconstruction filling and compactionpractices can continue.

Other possible reasons for the unex-pected performance could be (a) failureto extract all the air from the cells whilefilling them with de-aired liquid, and(b) a mismatch between the modulus ofthe cell and the modulus of the sur-rounding material.

Regarding Tables 1 and 2, there is noindication, either here or in the text,whether the measured stresses weremore or less than the theoretical. Itwould have been better, I think, if the

measured values had been presented asa percentage of the theoretical.

The discrepancy between the vibrat-ing wire (vw) piezometer readings andthe open standpipe readings looks like aconstant offset—the data curves arepretty much parallel. Maybe due to afaulty base-line zero readings.

Maybe there is a naturally occurringpressure gradient that would cause thepiezometers to read consistently lowerthan the standpipes. (I assume that thestandpipes are being read with a dipmeter)

In one case there is a two month timelag between reservoir level and vwpiezometer and open standpipe read-ings; in the other case there is nodiscernable time lag. But the time lagbetween the vw piezometer and theopen standpipe cannot be seen on thistime scale. There may have been a timelag amounting to a day or two.

ReferencesOosthuizen, C. , Naude, P.A. &

Hattingh, L.C. 2003. Total and PorePressure Cells: Method in the Mad-ness. Proc. 6th International Sympo-sium on Field Measurements inGeomechanics, Oslo, Norway,Balkema, pp 273-278.

Høeg, K. 2000. Question 78 Theme BDiscussion. Proc. 20th ICOLD Con-gress 19-22, Sept. Vol 5: 336.Beijing, China.

J. Barrie Sellers, President, GeokonInc., 48 Spencer Street, Lebanon, NH03766 USA. Tel. (603) 448-1562,Fax (603) 448-3216,email: [email protected]



John Dunnicliff

In contrast to the other discussers, I wasable to read all these discussions beforewriting mine!

Because neither Barrie Sellers nor Ihad a complete copy of Høeg (2000), Iasked Kaare Høeg (NorwegianGeotechnical Institute) whether thequotation in Oosthuizen et al (2003)was correct and complete. With his ap-proval to publish his reply, here it is:

“The quote is valid. I did not writedown the oral discussion and did notget a chance to review what was re-corded by the secretary at the discus-sion session. I am not sure I usedexactly those words, but I gave thatopinion after yet another paper pre-sentation, describing the use of somehundred earth pressure cells, and theauthor could not figure out what thereadings told him. The essence ofwhat I said was:

My experience is, and that ofmany authors at ICOLD Congressesand other Conferences seems to be,that it is very difficult to interpret andrely on the readings from earth pres-sure cells installed in embankmentdams, especially rockfill dams.Many investigators have spent timeand money installing such cells, buthave, in general, found the measure-ments of little value (refer toHvorslev’s classical work). Thereare examples of pressure cells givingvaluable readings when the cells areinstalled on a structural interface, butnot inside the dam body. I would dis-courage the use of pressure cells inembankment dams, but encouragemeasurements of seepage/leakage,pore pressures, movements and ac-celerations in seismic regions. Youshould provide very good reasons

and convincing arguments beforegoing to the installation of earthpressure cells in the dam body.”Again—having had the luxury or

reading the other discussions beforewriting this—I can respond to ElmoDiBiagio’s comments about Figure 3 inMirghasemi’s article, which is titled“Layout of embankment earth pressurecells (Dunnicliff, 1993)”. In the 1993reference this same figure was titled“Typical layout of embedment earthpressure cells (courtesy of Soil Instru-ments Ltd., Uckfield, England)”. AsElmo correctly says, “The drawing canbe interpreted in two ways, i.e., instru-ments placed on the floor of the excava-tion or set in cubical sockets beneath thefloor”. The second interpretation wouldcertainly lead to large errors. Iapologize for perpetuating confusion.

Author’s Reply

I thank all discussers Donald H. Bab-bitt, Elmo DiBiagio, Louis Marcil, P.Erik Mikkelsen, Arthur D. M. Penman,J. Barrie Sellers, and John Dunniclifffor their useful comments to clarify theproblems. And a special thank you toJohn Dunnicliff, who encouraged me toprepare this article, which was first pre-sented at the 2005 instrumentationcourse in the Netherlands, and waswhere we met. Also I would like ac-knowledge him for all of his sugges-tions and editing made throughoutpreparation of my article and this reply.

The number and variety of discus-sions express the interest in my article,and I am very pleased with their contri-butions. This might persuade others toreport not only the consistent data butalso the observed uncertainties. Or asJohn Dunnicliff says, “it’s refreshingwhen someone is willing to publishsomething that didn’t work out well”.

Before continuing with my reply, Iwant to emphasize that in this articlethe concentration was on instrumentswhere no consis tent data wereachieved. As I said in the article, valu-

able and consistent information wasalso gained from the other instruments,indicating satisfactory performance ofthe dam. Because of length limitations,this information was not reported in thearticle.

Uncertainties in Earth PressureMeasurementsMost discussers agree that earth pres-sure measurement in an embankmentdam is not an easy or straightforwardtask. Therefore, being faced with a setof inconsistent data seems to be normal.However to reply to the discussers, thefollowing explanations are given.

Installation of the pressure cells wascarried out according to the manufac-turer’s instructions:

“To prevent damage to the cells, thelens and cells are installed in an excava-tion made to accommodate them. Thecells are installed in a lens forming amound on the base of the excavation orwithin a lens in a pocket excavated inthe base of the excavation. The cellpockets should be excavated with ex-treme care to avoid disturbance to the

soil. The pockets should be located 1meter away from adjacent pockets orthe excavation walls. The width of thepocket should be equal to a minimumof 3 pad diameters to avoid load bridg-ing. The length of the pocket (axis ofthe pressure transducer) should be 6pad diameters to accommodate thelength of the pressure cell and to pro-vide 1 pad diameter clearance at eitherextremity of the pressure cell.”As may be noticed, the manufacturer

suggests either way of installation (refer toJohn Dunnicliff and Elmo DiBiagiocomments). However, the compaction ofsurrounding soil when the instrumentare placed on the floor, without diggingthe pockets, is much more difficult (es-pecially for vertical and 45-degreecells). Also, when the cell is not placedin a pocket, the possibility of cell rota-tion increases during the compaction. Ifthe cells are placed on the excavationfloor vertically or 45 degrees inclined,the compaction around them should beperformed with great care. In this situa-tion the compacted fill around the cellswill not have the same quality as the



main embankment, and arching aroundthe cell will occur. The procedure ex-plained by Barrie Sellers (Oosthuizen etal, 2003) is a good idea to reduce theproblem of stress arching and rotationof the installed cells. However the selec-tion of concrete specification is acritical issue in this method.

In Tables 1 and 2 in the article, I triedto compare the calculated and measuredstresses. As mentioned, the calculatedstresses acting on a cell are based on themeasured stresses on the other cells. Inother words, using Mohr circles, thestresses measured by the other cells areused to calculate the stresses in a certaincell and then the results are comparedwith the stress measured by that cell.Thus, no comparison is made betweenmeasured and theoretical stresses aspointed out by Barrie Sellers. There isno certain trend for the differences in

positive and negative values as reportedin revised Tables 1 and 2

The measured and calculated (basedon embankment height) stresses as sug-gested by Elmo DiBiagio, are comparedin Figures 8 and 9 for cells PC5-5 andPC6-5 respectively. Again some uncer-tainties can be found from the figures:• In Figure 9 the measured pressure by

a 45 degree cell is higher than thatmeasured by horizontal cell (verticalpressure).

• Since at the dam the valley is wideand cell PC6-5 is located at the cen-ter of the core, the stress field is es-sentially two-dimensional, asdescribed by Elmo DiBiagio. Also,at the centre of the clay core the max-imum principal stress direction isnear vertical. Therefore measuredstresses by 45 degree downstreamand 45 degree upstream cells should

be equal. Figure 9 shows a signifi-cant difference between two sets ofmeasured stresses.Based on the experiences described

in the article and by the discussers, itcould be recommended that if there are“very good reasons and convincing ar-guments” (as mentioned by KaareHøeg), at the most only horizontal earthpressure cells may be considered for in-stallation in embankment dams. In thiscase the installation of the cell at thefloor of excavation will not be difficult,and the possibility of rotation of cellsduring compaction is minimized. Theinstallation of vertical and inclined cellsin the embankment may not be justified.

Comparison of Vibrating Wireand Open StandpipePiezometersA wide variety of ideas about the differ-ences observed between vibrating wireand open standpipe piezometer data arepresented by the discussers. The varietyof the excellent comments on this issueis very helpful to explore the real rea-sons for these differences. That is oneside of the story. From the other point ofview, this makes it very difficult to cometo a certain conclusion.

To respond to the comments bydiscussers, I would like to start with theinitial reading procedure for the vibrat-ing wire piezometers, which was a mainsubject among the comments. In the fol-lowing, the steps carried out for initialreading of Karkheh VW piezometersare presented.1. At least 48 hours before piezometer

installation, the filter was installedon the piezometer and the readingwas carried out and then the instru-ment was immersed in water. Thereadings were not recorded becausethe objective of making readings wasto make sure that the device was re-sponding.

2. 24 hours after filter installation (atleast 24 hours before piezometer in-stallation), readings was made andrecorded as L0 & T0. After this read-ing the piezometer was used tomeasure a known water level in anobservation well. In the calculationof water level L0 & T0 were used. Ifthe difference between the real



Table 2, Revised. Comparison between computed and measuredhorizontal stresses

Pressure cell number Differences between measured andcomputed stresses for the vertical cells

(in percent)

PC5-4 +266

PC5-5 -33

PC6-4 +120

PC6-5 +41

PC7-3 -70

PC7-4 -64

Table 1, Revised. Comparison between computed and measuredstresses at 45 degree planes.

Pressure cell number Differences between mea-sured and computed

stresses for the cell ori-ented 45 degrees down-

stream (in percent)

Differences between mea-sured and computed

stresses for the cell ori-ented 45 degrees up-stream (in percent)

PC5-4 -17 no measurement

PC5-5 +23 no measurement

PC6-4 -21 +11

PC6-5 +17 +115

PC7-3 +48 +70

PC7-4 +23 +22

water level and the calculated levelwas acceptable, the next steps of in-s tal la t ion procedure werecompleted. This step ensured the ac-curacy of L0 and T0.

3. After at least 24 hours of initial read-ing and piezometer check-up, thepiezometer was installed.

4. One and 24 hours after installation,readings were made and recorded asL1 & T1 and L24 & T24, respectively.Table 4 presents the recorded values

for EP5-15 and EP5-17 VW

piezometers. The installation procedureincluded the four points identifiedabove, together with the following, toensure correct initial readings:• The time between filter installation

and the initial reading was 24 hours

to allow “complete dissipation of theoverpressure momentarily inducedby the installation of the high air en-try filter on the piezometer” as ex-plained by Louis Marcil.

• The L0 and L24 are close to eachother; therefore there is no sign ofcontinuation of pore pressure dissi-pation between a time 24 hours afterfilter installation (L0) and at least 48hours after filter installation (L24).This means that during this time in-terval the piezometer had came to asteady state condition.The next subject that I have to ex-

plain is the procedure employed forpreparation of the mixed clay, pointedout by Arthur D. M. Penman. TheKarkheh conglomerate unit at the bor-row area is weak to moderately ce-mented, and it breaks down duringexcavation to the sand and gravel grada-tion without any special effort. For thefield preparation of the mixed clay, al-ternate layers of clay and gravel weredeposited at their optimum moisturecontent. These sandwiched layers ofclay and gravel were then cut and mixedby appropriate excavating machines.The mixed clay materials were thenloaded and carried to the core zone ofthe dam by special trucks, where theywere compacted in layers of limitedthickness to achieve the required den-sity. An appropriate test program wasimplemented to ensure the homogene-ity of the fill.

As recommended by Arthur D. M.Penman, the open standpipepiezometers were “connected directlyto the intake filters and sealed by plac-ing in the fill during construction”.They were not placed in boreholes aftercompletion of the embankment. Thismethod facilitates a better and easiersealing procedure. A single 15-cm(6-in.) PVC pipe was used for the pro-tection of two piezometer riser pipes(each with internal diameter of 20.6 mm



Table 4. Recorded values for EP5-15 and EP5-17 VW piezometers

PiezometerName

L0 T0 L1 T1 L24 T24

EP5-15 3409.16 20.3 3405.61 13.3 3408.25 13.3

EP5-17 3097.99 19.9 3093.96 15.2 3095.39 13.7

Figure 8. Comparison of calculated vertical pressure based on fill height and mea-sured stresses for PC5-5.

Figure 9. Comparison of calculated vertical pressure based on fill height and mea-sured stresses for PC6-5.

(0.8 in.)), located at the same stationsand distances but different elevations.

Based on above explanations it is noteasy to be convinced that the main rea-son for the difference between SP andVW is either the existence of the morepervious layer in the fill or the breakageof the SP sealing.

In summary, it is not a straightfor-ward and easy job to make a convincingconclusion about this issue. All thenoteworthy opinions raised by thediscussers such as deformation of the

VW housing under high pressure of em-bankment (Elmo DiBiagio), inaccuracyin zero reading (Louis Marcil & BarrieSellers), breakage of the SP sealing(Louis Marcil & Arthur D. M. Penman),existence of pervious layer in the fillconnected to SP filter (Arthur D. M.Penman) and effect of pore air pressureon measured water pressure (ErikMikkelsen) may contribute to describethe problem with different levels of in-fluences. But for me it is still very hardto say which one contributes more, and

is the key role player.

Ali Asghar Mirghasemi, Associate Pro-fessor, School of Civil Engineering,College of Engineering, University ofTehran, Enghelab Ave. P.O. Box11365-4563, Tehran, Iran. Also seniorgeotechnical engineer for KarkhehDam project, Mahab Ghodss Consult-ing Engineer, Tel: (+9821) 61112273,Fax: (+9821) 66461024,email: [email protected]

Piled Foundation Design –Clarification of a Confusion

Bengt H. Fellenius

AbstractA frequent confusion and lack of under-standing exists with regard to the designof piles subjected to drag loads. Somewill lump the drag load in with the deadand live loads when assessing pile bear-ing capacity. Also common is to disre-gard the root of the problem: settlementof the piled foundation. It must be real-ized that: dead and live loads appliesto bearing capacity, dead load anddrag load applies to structuralstrength, and downdrag is settle-ment.

A few weeks ago, I was once againasked if the allowable load for a pileshould be reduced when consideringdrag load. Shortly thereafter, when Itook a look at the discussions atwww.Geoforum.com, I noticed a verysimilar question. Perhaps I should notbe that taken aback by the lack ofknowledge displayed by the questions.The persons asking may not have beentaught better. The following is a quotefrom a textbook published in 2001 andassigned to 4th Year Civil Engineeringstudents at several North AmericanUniversities:

Piles located in settling soil layersare subjected to negative skin fric-tion called downdrag. The settle-ment of the soil layer causes the fric-

tion forces to act in the samedirection as the loading on the pile.Rather than providing resistance,the negative skin friction imposesadditional loads on the pile. The neteffect is that the pile load capacity isreduced and pile settlement in-creases. The allowable load capac-ity is given as:

QQ

FQallow

ult

s

neg= −

where Qallow = Allowable load capacity

Qult = Load capacity

Fs = Factor of safety

Qneg = Downdrag

First, “negative skin friction” is not“downdrag” but defines a downward di-rected shear force along the pile, whiledowndrag is the term for settlement of apile (caused by the settling soil ‘drag-ging a pile along’). Second, the term“load capacity” means different thingsto different people and “allowable loadcapacity” is an abominable concoctionof words. Third, and very important, thephrasing in the quoted paragraph con-fuses cause and effect. Drag load is notdowndrag, and it does not cause settle-ment, but is caused by settlement of thesurrounding soil and is mobilized when

the pile resists this settlement. Theworst boo-boo, however, lies in thequoted formula, which does not recog-nize that the factor of safety and the dragload are interconnected, i.e., changingthe factor of safety changes the dragload. As this may not be immediatelyclear to all, the following example willtry to clarify the interaction between thepile, the factor of safety, and the dragload.

ExampleConsider the case of a 300 mm diameterpile installed to a depth 25 m through asurficial 2 m thick fill placed on a 20 mthick layer of soft clay deposited on athick sand layer. The case is from a re-cent project in the real world. Let’s as-sume that a static loading test has beenperformed and the evaluation of the testdata has established that the pile capac-ity is 1,400 KN. As is visually presentedin Fig. 1, applying a factor of safetyof 2.0 results in an allowable loadof 700 KN (dead load 600 KN and liveload 100 KN). Moreover, due to the filland a lowering of the groundwater ta-ble, an almost 200 mm settlement of theground surface will develop after theconstruction. How should the designerassess this case? Incidentally, as severalfull-scale case histories have shown,



whether or not the soil at the site settles200 mm or 2 mm, or for that matter2,000 mm, the magnitude of the dragload will stay the same.

Comments and QuestionsSome practitioners believe that all iswell because the foundations includepiles with a capacity of twice the de-sired allowable load. Then, there arethose who understand that, for the num-bers indicated above, the pile will be af-fected by a drag load of about 300 KN,acting at a neutral plane located at 17 mdepth (Fig. 1). A few of these practitio-ners will subtract the drag load from thepile capacity before applying the factorof safety and arrive at an amended al-lowable load of 550 KN (which is a vio-lation of principles as this approach ineffect has reduced the drag load by afactor of 2.0). Others will apply thequoted formula, and arrive at an allow-able load of 400 KN. Yet others will re-alize that the last approach means thatthe drag load is applied without a factorof safety, preferring to apply the for-mula with the drag load increased by a

factor of safety, say 2.0. This results inan allowable load of (1,400/2 - 2x300) =100 KN — don’t laugh, I have seen itdone several times and it was proposedfor this project! So, which allowableload is right? Is it 100 KN, 400 KN,550 KN, or 700 KN? (Similar divergingapproaches abound in the load-and-re-sistance-factor-design, LRFD).

Suppose the structure supported onthe piled foundation was built beforethe drag load conditions were recog-nized (no signs of distress are notice-able). Then, what factor of safety woulda back-analysis show the piles to have?Would it be 1,400/700 = 2.0, or(1 ,400 - 300)/700 = 1.6, or1,400/(700+300) = 1.4? And, I wonderhow the fellows advocating the laugh-able approach would react when realiz-ing that the piles are supporting seventimes more load than the maximum loadtheir approach would allow as safe.

Before answering, consider that themagnitude of the drag load depends onthe magnitude of the dead load on thepiles. Reduce the dead load and the dragload increases, and vice versa. For ex-

ample, after reducing the allowable loadby 150 KN to arrive at a 550 KN value(made up of a dead load of, say, 475 KNand a live load of 75 KN), the drag loadis no longer 300 KN, it is 400 KN! If theallowable load is reduced by an addi-tional 150 KN, say to 400 KN (made upof a dead load of, say, 325 KN and a liveload of 75 KN), the drag load increases500 KN!

Note, for the three values of deadload — 600 KN, 475 KN, and 325 KN— the neutral plane location changesfrom depths of 17.0 m to 18.0 mto 19.5 m, respectively. To simplify theexample, no change of the toe resistanceis included. In reality, however, thedeeper down the neutral plane lies, thesmaller the enforced penetration of thepile toe into the sand and the smaller themobilized toe resistance, and when thetoe resistance is reduced, the location ofthe neutral plane moves upward and thedrag load changes. Altogether, the loadat the neutral plane, that is, the maxi-mum load in the pile, is essentially un-changed for the three alternative valuesof allowable load. In stark and impor-tant contrast, for each reduction of al-lowable load, the project foundationcosts increase.

ClarificationBewildering, ain’t it? Many select oneof the four approaches as the one to becorrect, ignoring the others, thus avoid-ing having to make the small leap of un-derstanding of what a proper designneeds to include, as follows:

First, the drag load does not affectthe pile bearing capacity — the ultimateresistance. That is, the pile capacity isthe same whatever the magnitude of theload from the structure. The factor ofsafety is applied to ensure that, shouldthe load on the pile be inadvertentlylarger than intended and should the pilecapacity be inadvertently smaller thanthought, the pile might be close to fail-ure, but it would not fail. No negativeskin friction—no drag load—is presentclose to failure. Therefore, only the firstapproach, that with the 700 KNallowable load, is correct.

Second, the drag load has to be con-sidered, of course, but not in the contextof bearing capacity. The concern for the



Figure 1. Distribution of load in the pile. “A” is the long-termload-distribution and “B” is the resistance distribution measuredat ultimate resistance (capacity) in a static loading test

drag load only affects the pile structuralstrength at the location of the maximumload, i.e., at the neutral plane. For theexample case, if the structural integrityof the pile is safe considering the sum ofdead load and drag load, that is, 900 KNfor the example case, the design for dragload is complete.

Third, with (A) the dead load pluslive load safe considering the pile ca-pacity, and (B) the dead load plus drag

load safe considering pile structuralstrength, it remains to show that (C) thepile will not settle more than acceptable,that is, that downdrag is kept in check.

In checking the pile for downdrag, itmust be realized that there are two differ-ent definitions of the neutral plane. Bothgive the same result, or location, rather.One defines the neutral plane as locatedat the force equilibrium in the pile, whichis where the shaft resistance changesfrom negative to positive direction andwhere the sum of the dead load plus dragload is in equilibrium with the positiveforces in the pile. The second defines thelocation to be where the pile and the soilmove equally. (Note, the toe resistance is

only as large as is needed to establish theequilibrium between forces and move-ments. Moreover, whatever the factor ofsafety chosen in the design, if the soil issettling at the neutral plane, the pile willsettle too and as much as the soil settles atthat location). The influence of varyingdead load and toe resistance is illustratedin Fig. 2. The diagram to the left showsthe load distributions and locations of theneutral plane for the dead loads associ-

ated with the mentioned different allow-able loads on the pile (the 100 KN case isexcluded). The diagram to the rightshows the distribution of soil settlementand location of neutral planes for the threeapproaches. (More explanation and dis-cussion is available in Fellenius, 2004).

The Biggest ProblemThe foregoing appears to be news tomany. It shouldn’t. The long-term re-sponse of piles in settling soil was madeknown in several very accessible publi-cations as early as some 40 years ago.How does one get the textbook writersand the teachers of foundation design tobecome aware of the knowledge andconvey it in the teaching of future prac-

titioners? How does one get practitio-ners to fill the voids in their professionaleducation and to keep abreast with ad-vances in the profession? The latter, un-fortunately, may be the biggest problemof all, but discussing it lies outside thiscontribution.

ReferenceFellenius, B.H., 2004. Unified design of

piled foundations with emphasis on

settlement analysis. “HonoringGeorge G. Goble—Current Practiceand Future Trends in Deep Founda-tions” Geo-Institute Geo-TRANSConference, Los Angeles, July 27 -30, 2004, Edited by J.A. DiMaggioand M.H. Hussein. ASCEGeotechnical Special Publication,GSP 125, pp. 253 - 275.

Bengt H. Fellenius, Bengt FelleniusConsultants Inc., 1905 Alexander StreetSE, Calgary, Alberta. T2G 4J3.Tel.: (403) 920-0752,email: [email protected]



Figure 2. Distribution of load in the pile and interaction with soil settlement. The dashed curves represent the gradualchange in a transition zone from negative direction of shaft shear to positive direction.

Rules of Thumb for GeotechnicalInstrumentation Costs

Gord McKenna

IntroductionHow much does a piezometer cost? Mostof us would guess it equated to the cost ofthe transducer and a certain amount ofsignal wire – often about $1000. But thelife-cycle cost of many geotechnical in-struments is closer to $15,000 when drill-ing, technician time, future readings, datamanagement, and geotechnical analysisand reporting are factored in. This articleprovides a few rules of thumb and somesuggestions on managing this cost.

In writing this article, I had in mind ahypothetical example of instrumentinga small embankment dam (about a halfday away from a major centre) with afew inclinometers and a dozenpiezometers to confirm design assump-tions by monitoring construction per-formance. Readers are encouraged tolook at their typical projects and adaptthese rules of thumb to their own pro-jects. Readers are also directed to a fewgood references at the end of this articlethat provide additional informationwith similar themes.

Instrument Life-cyclePurchasing and installing an instrument isjust the start. It can be useful to think of in-strumentation costs in terms of thelife-cycle of the instrument – from its ini-tial geotechnical design and procurement,through drilling and installation, includ-ing reading and maintenance, and datamanagement and geotechnical analysisback in the office. In addition, we must re-member decommissioning.

Rules of Thumb for GeotechnicalInstrumentation CostsFigure 1 shows the costs of a typicalgeotechnical instrument, in this case aninstrument installed in a borehole, readquarterly, with annual data analyses overa ten-year period. Below are some rules ofthumb.• Rule of thumb #1: Most geotechnical

instruments cost about $1000 to pur-

chase (some more, some less)• Rule of thumb #2: The cost of drill-

ing for an instrument is often about$1000 and the cost of the installationand the first couple of readings is an-other $1000.

• Rule of thumb #3: It costs about$1000 to datalog an instrument(datalogger + cables + installation)

• Rule of thumb #4: It costs about$5000 to read each instrument for itslife and another $5000 doinggeotechnical analysis and reporting onthe results.

• Rule of thumb #5: A typicalgeotechnical instrument costs about$15,000 for its life-cycle cost, about15x the initial instrument hardwarecost.

• Rule of thumb #6: Expect to replaceabout 20% of the instruments every 10years

• Rule of thumb #7: Geotechnical in-strumentation costs are often 2 to 10%of the total earthwork costs for thestructure being monitored (less forlarge structures)Some lessons learned from the above

rules:• Lesson #1: The best way to save

money is to install only the instru-ments that are really needed.Dunnicliff and Powderham (2001)point out “The purpose ofgeotechnical instrumentation is to as-sist with answering specific questionsabout soil/structure interaction. Ifthere are no questions, there should beno instrumentation.” Challenge your-self to have the minimum amount ofinstrumentation that will provide an-swers to clearly defined questions.

• Lesson #2: The actual cost of the in-strument is small compared with thecosts of reading and analyzing thedata. Purchase the instrument that hasthe lowest life-cycle cost while keep-ing in mind reliability – having to re-place a faulty instrument, or spending



Friesens

Do not print key lines

Please centre the attached originaldrawing within this frame.

John

The final hand-drawn copy whichwas delivered to me today, was slightlysmaller than the one I received electron-ically. This is the reason for the entirecolumn not being used.

Figure 1. Rule of thumb costs for the life ofa single geotechnical instrument. Sketchby Derrill Shuttleworth, Studio Two,Edmonton, Alberta, Canada

hours trying to salvage “bad” data aretwo ways to increase the life-cyclecosts greatly. American Society ofCivil Engineers (1999) states: “Whenconsidering costs, the lowest initialcost of an instrument should never dic-tate the selection of an instrument. In-stead, a comparison of the overall costof procurement, calibration, installa-tion, life-cycle maintenance, reading,and data process of the available in-struments (plus a consideration of thereplacement cost if there is some riskthe instrument should go bad) shouldbe made... the cost of instrumentsthemselves is usually a minor part ofthe total cost.” For example, vibratingwire piezometers have a higher initialhardware cost, but generally arecheaper and more reliable to read thanpneumatic piezometers.

• Lesson #3: Be fanatical about ensur-ing that every instrument is properlyinstalled and maintained and that allare read properly with good QA/QC.Ensure the technician or engineer whoinstalls the instrument is top notch,well trained, and diligent. Have a for-mal QA/QC program that involveschecking the data in the field when it isread, that it fits historic data when it isuploaded into the database, and that itfits seasonal trends or other patterns onan annual basis. One bad reading cancost as much as 10 good readings.

More Good Practices• The biggest cost of instrumentation is

getting the readings over the years andmanaging the data. One of the greatestrisks is the expense (and sometimesembarrassment) caused by poor data.

• Your instrumentation suppliers arepartners in your project – you will berelying of their advice and service.Most geotechnical people have de-veloped long-term relations withpreferred suppliers.

• Ensure there are proper proceduresin place, good training, and that theequipment (including readoutboxes) are kept in good working or-der and calibrated regularly. Alwaysstore the raw data – not just the cal-culated data.

• Try to use the same person and sameequipment for each reading set. The

data are often more operator-de-pendent than you might think.

• Protect your instruments – markthem well in the field, ensure they arewell labelled and marked with a con-tact phone number. Keep wirestucked away where they won’t beshot or chewed.

• Where it would be difficult or impos-sible to replace instruments, con-sider duplicate installations.

• Consider low-impact access – youdon’t need a gravel road into everyinstrument – consider using an ATVor hiking in where practicable (thismakes it cheaper when it comes todecommissioning too).

• Look for opportunities for continu-ous improvement of your instrumen-tation practices and programs withan eye to reliability and life-cyclecosts. Seek out opportunities to gainand share experience with col-leagues at conferences andtradeshows, continuing educationcourses, and chatting with your in-strumentation suppliers and drillers.

In ClosingThe cost of geotechnical instrumenta-tion is high, so it deserves care in attend-ing to all aspects of design, procure-ment , ins tal la t ion, reading,maintenance, data management,QA/QC, and analysis. Some fanaticism

balanced with practicality helps. Don’tskimp on the instrument costs only topay the piper down the road. Critical is-sues include the design of the program,the installation of the instruments, en-suring that the data are processed, andproperly and efficiently managing thedata.

ReferencesAmerican Society of Civil Engineers,

1999. Instrumentation of embank-ment dams and levees. Technical en-gineering and design guides asadapted from the US Army Corps ofEngineers. no. 26. American Societyof Civil Engineers, Reston, VA, 88phttp://www.usace.army.mil/inet/usac e - d o c s / e n g - m a n u-als/em1110-2-1908.

Dunnicliff, J. and Powderham, A.J.,2001. Recommendations for pro-curement of geotechnical instru-ments and field instrumentationservices. Geotechnical News, Vol.19, No. 3, September, pp 30-35. Alsodiscussion by Klinger and closure byauthors in the same issue.

Gord McKenna, Senior GeotechnicalEngineer, Norwest Corporation,#830-1066 West Hastings Street, Van-couver, BC Canada V6E 3X2. Phone604-602-8992, Fax 604-602-8951,email: [email protected]



For Cyril’s Canadian Geot. Jnl

frame 27,6 x 21 for ad

Electrical Cables for GeotechnicalInstrumentation Applications

J. Barrie Sellers

INTRODUCTIONThe choice of cable for geotechnical in-strumentation applications must takeinto consideration many factors, someof them not obvious, others often con-flicting. The dictates of environmentalfactors may need to be reconciled withquestions of physical constraints andcost. The purpose of this article is tosuggest which cable might be most suit-able for each particular application.

CABLE DESIGNCables are most often made from indi-vidual copper conductors, which maybe solid or stranded, and encased in aninsulation material. Individual conduc-tors are twisted into pairs or bundled in-side a conductive shielding material,and then covered by an outer jacketmade from the most suitable material.Cables may, in addition, be wa-ter-blocked, armored, or may containsteel or Kevlar® strands for additionalstrength, or plastic tubes for circulatingfluids, or for venting to atmosphere.

The Number, Type and Size ofConductorsThe number of conductors in a cable isdetermined by the number of sensors tobe connected to the cable, and the num-ber of conductors required by each sen-sor. Sometimes economies can be madeby ‘commoning’ one of the leads fromeach sensor. But to do so runs the risk oflosing all the sensors should the com-mon lead fail. (Note: if the leads must becommoned it is better to common theground leads connected to the negativeterminal of the power source rather thanthe positive power lead). Also, the con-nection of numerous sensors to onecommon conductor increases the possi-bility of electrical noise on that conduc-tor giving rise to unstable readings onall the sensors (this is discussed morefully in a later section under the head-ing, “Electrical Noise”). Thus, as a gen-

eral principle it is advisable to avoid us-ing common lead wires, particularlywhere a datalogger is being used.

Also in this context, there may be atemptation to use the shield as the returnor ground conductor. While technicallyfeasible, this is open to the same objec-tions as before, in that it would allowany electrical noise to contaminate theoutput signal.

The type of conductor normally usedis the stranded tinned copper type.Stranded conductors are more flexiblethen solid conductors, and this makesthe cables easier to handle during instal-lation.

The size of the conductors should bechosen to be as small and inexpensive aspossible, consistent with the need toavoid line losses and voltage drops.Physical strength is another consider-ation. For electrical resistance type sen-sors, 16 or 18 AWG wire is preferred,while for vibrating wire types 22 AWGis sufficient (the larger the AWG num-ber, the smaller is the diameter of theconductor).

ShieldingShielded cables should always be usedwhere dataloggers are in use. Shieldingprovides protection from electrostaticradiation coming from nearby electricalequipment, from lightning strikes, andfrom electromagnetic fields surround-ing power lines, transformers, etc. Indi-vidually shielded, twisted pairs are thebest construction for multi-conductorcables. Drain wires connected electri-cally to Mylar-type shields provide asimple means of connecting all theshields to a common ground. Individu-ally-shielded, twisted -pairs are prefera-ble for the rejection of common-modeinterference. In a Wheatstone bridge,one shielded pair should be the powerleads, another shielded pair the outputleads, and a third shielded pair the re-mote sensor leads. For vibrating wire

sensors, one shielded pair goes to the vi-brating wire sensor, the other goes to thethermistor.

Mylar tape type shielding is usuallysufficient, but in severe electrical noisecases cables with braided copper shield-ing offer slightly better protection.Shielding may also be provided by putt-ing the cable inside a magnetic metallicconduit. Cables should be located as faras possible from the potential source ofnoise. They should never be placedalongside power cables.

Shields should be grounded only atthe readout location. They should neverbe connected directly to the externalsensor housings, Connection at bothends allows the formation of groundloops and defeats the purpose of theshield.

Where additional shielding of thesensor itself is required, it can some-times be achieved by placing the sensorinside a mild steel (magnetic) enclo-sure. This is often required wherepiezometers are located in the sameborehole as an electrical pump.

Insulation of ConductorsIndividual copper conductors are nor-mally insulated by a layer of rubber orplastic insulation. In general, polyethyl-ene or polypropylene insulation is usedat normal temperatures, since PVC hastoo high a dielectric absorption. Forhigh temperature Teflon is most oftenused.

Outer JacketsAn outer jacket which is round and firmis easier to grip and seal at the pointwhere it enters the body of the sensor.For this reason it is preferable to use ex-truded outer jackets.

Several material formulations areavailable:1) Neoprene – A synthetic rubber

compound commonly used for out-door applications. It has good resis-



tance to gasoline, oils etc. Ordinaryrubber should never be used.

2) PVC – A common choice for itsgood electrical properties and forbeing waterproof. It should not beused at low temperatures, when itbecomes brittle.

3) Polyurethane – This material isvery resistant to cuts and abrasions,making it useful for cables that aresubject to repeated rough handling.It is not as water-resistant as PVC,but has better low temperature capa-bilities.

4) High Density Polyethylene – Thishas many desirable properties. It isthe most resistant to environmentalattack, and has excellent low tem-perature characteristics. Unfortu-nately, like Teflon, the material is soslippery that splicing and pottingcompounds will not stick to it. Thismakes it difficult, without the aid of“O” rings, to build waterproof con-nections to sensors, cable splicesand cable connections.

5) Teflon – This material is essentialwherever sensors and cables aresubject to high temperature. It hasoutstanding resistance to environ-mental attack and also has excellentlow temperature properties. It is ex-pensive, and potting compoundswill not adhere to it. Therefore seal-ing around connections and splicesis difficult.

6) Other compounds such as Kevlar®

or Silicon Rubber etc. may be re-quired where there is a need for lowsmoke emissions, flame retardant,or resistance to nuclear radiation.

ArmorArmor may take the form of a helicallylaid layer of steel wires for heavy-dutyprotection, or a corrugated aluminum orsteel layer for light duty protection, e.g.for protect ion against rodents .Unarmored cables may also be confinedinside flexible conduit or rigid conduit.Armored cables are very stiff and diffi-cult to work with. Cable splicing be-comes a problem and can be expensiveif continuity of mechanical strength isrequired.

Armored cables are most oftenneeded for sensors installed in earth em-bankments, where large forces are ex-erted on the cable by compactingequipment and earth moving vehicles,and by settlement, differential settle-ment, “weaving”, and sideways spread-ing of the embankment as it is built.They should not be connected directlyto strain gages, crackmeters, etc., be-cause the stiffness of the cable would al-low it to pull on the gage and alter thereadings. They are not necessary inconcrete.

FillingsCables should be of compact construc-tion, using fillers and extruded outerjackets to fill the spaces between theconductors, and to keep the cablesround and firm. Water-blocked typesfor underwater use are available with agel, powder or grease to prevent the pas-sage of water along the cable.

Cables filled with gel, powder orgrease are difficult and messy to dealwith and should be considered applica-ble only to underwater situations, e.g. inearth dams, and in marine offshore en-vironments. They are more necessarywhen used with sensors outputting lowvoltages, such as electrical resistancestrain gages, where waterlogged cablescould affect the sensor output. With vi-brating wire sensors, these cable effectsare much less pronounced, and filledcables are not required.

I.D. TagsIt is sometimes useful to tag long cablesat intervals along their length, so theycan be more easily identified at any lo-cation.

Strain ReliefWhere a cable is required to have extratensile strength, it should include eithera Kevlar® or a braided stainless-steelaircraft cable running down its axis.This construction is typical for incli-nometer probe cables, where stretchingof the cable is undesirable, and wherethe cable may have to be used to pull ajammed probe out of a deformed cas-ing.

Another typical application of strainrelief is where a sensor hangs vertically

inside a deep borehole, and the weightof the suspended cable and sensor ex-ceeds the tensile strength of the cable. Inthis situation there is an alternative - totape the cable alongside the aircraft ca-ble, which is then used to take theweight.

Vented CablesSpecial cables are available which con-tain plastic tubes inside them, as well asthe usual conductors. These tubes canbe used to transport air or other fluids.This kind of cable is required for ventedpiezometers, where a single vent tubeallows the inside of the pressure sensorto be connected to the ambient atmo-sphere, so as to provide automatic baro-metric pressure compensation.

Cable SplicesCable splicing is best done using com-mercially available splicing kits con-taining mechanical butt-splice connec-tors and epoxy potting compounds.These provide a good waterproof andmechanically strong splice. Typical kitsare the Scotchcast 3M, Models 82A1and 72N1. It is advisable also to solderthe mechanical butt splices since propercrimping of the butt splices can be diffi-cult to achieve without the propercrimping tools. Armored cables are dif-ficult to splice if the mechanicalstrength is to be maintained - specialmechanical connections need to be fab-ricated which will grip the armor firmly.

ENVIRONMENTAL FACTORS

WaterCables should be waterproof, especiallywhere the sensors have a low voltageoutput. (e.g. electrical resistance types).Vibrating wire types are less susceptibleto water effects. Most plastic and neo-prene formulations are waterproof.Polyurethane jackets are less water-proof, making them less suitable forprolonged use under water.

Pressure-extruded jackets that fitsnugly around the inner conductors andfill the intervening spaces are less likelyto act as conduits for water if the cable iscut and water enters. Water-blocked ca-bles, filled with gel, grease or powderare most effective.



Cable splices ought not to be locatedunder water. Whenever possible theyshould be kept above ground in a dry ac-cessible location. Where underwatersplices are unavoidable they should bemade using commercially availablesplice kits, designed specifically for un-der-water applications, and supple-mented where necessary by additionalwaterproofing.

Water & Sun ResistanceMost plastic jackets and neoprene cablejackets have excellent resistance toweathering. Teflon cables have out-standing resistance.

Chemical ResistanceThe resistance of different cable materi-als to different types of chemicals canbe found in the literature of cable manu-facturers (e.g. www.belden.com). Tef-lon cables have outstanding resistanceto almost all chemicals.

In very aggressive environmentssuch as landfills (garbage heaps), thatcontain many aggressive chemicals andleachates, it may be necessary to encasethe cable inside stainless steel tubing.This is expensive but effective, and alsoprovides good mechanical protection.

Abrasion, Physical DamagePolyurethane cables have outstandingresistance to abrasion and cutting. Thisis important in high-traffic areas, or

where cables are being dragged alongthe ground, or being used repeatedlywith portable sensors.

Cables that are buried in the groundshould be protected from sharp objectsby use of conduits, armor, or by sur-rounding them with fine-grained mate-rial. Particular attention should be paidto zones which will be prone to highshearing forces during the life of the ca-ble. Here it may be advisable to leaveenough slack in the cable to accommo-date these movements withoutbreakage.

TemperatureThe normal operating temperaturerange of various types of cable is as fol-lows (all temperatures in degrees centi-grade):• Neoprene -20 to +60• PVC -20 to +80• Polyethylene -60 to +80• TFE Teflon -70 to +260• FEP Teflon -70 to +260• Polyurethane -40 to +80

For low temperatures, high-densitypolyethylene or polyurethane is prefer-able.

For high temperatures, Teflon ispreferable.

Electrical NoiseElectrical noise in the ambient environ-ment, at any location along the electri-

cal cable, can affect the output signal.Electrical noise has many sources, forexample adjacent power lines and trans-formers can give rise to powerful mag-netic fields, which can induce unwantedalternating current and voltage onnearby sensor cables. Electrostatic radi-ation from circuit breakers, motor gen-erators, nearby lightning strikes, radiobeacons, vehicle ignition systems etc,all can create individual voltage spikesin the sensor cable, and unsteady read-ings where voltage (electrical resistancegages) or frequency (vibrating wiregages) are being read. Most of these un-wanted effects can be filtered out usingproper shielding techniques and appro-priate electronic circuitry.

Shielded cables are essential wheredataloggers are in use and where cablesare long.

SUMMARYIt is hoped that the above remarks willbe helpful in choosing the right cablefor any application. If in doubt alwaysconsult the cable manufacturer or theinstrument suppliers.

J. Barrie Sellers, President, GeokonInc., 48 Spencer Street, Lebanon, NH03766. Tel. (603) 448-1562, Fax. (603)448-3216, email: [email protected]



do not print key lines

space is 42 x 20,6

Geotechnical Instrumentation News

Documents

Transcript of Geotechnical Instrumentation News