Difficult rock comminution andassociated geological conditions

Peter J. Tarkoy B .Sc ., M .Sc. . Ph .D., CEng ., M.A .S .C .E ., M .A .E .G ., M .i .A .E .G .GeoConSol, Inc ., Medfield, Massachusetts, USAMassimo Marconi Dou.lng.

Building Division, Lodigiani SpA, Milan, Italy

SYNOPSIS

Conditions encountered by the authors onseveral projects (tunnel boring, blast hole drill-ing, and large diameter pile socket drilling)were unanticipated, undetectable, and evendifficult to identify even after they had beenencountered. These conditions have causedfailure of mechanical components, low penetra-tion rates, and remarkably high cutter wear .Coincidentally, all projects had one or moregeotechnical conditions in common . The diffi-culties encountered, however, could not beattributed to geotechnical factors measurableby classical laboratory or field methods . Thedifficulties were not discernible or predictableby conventional exploration, rock mechanicstesting, or analysis, prior to tender .

A study of common elements and an analysisof these conditions have shed some light onhow to avoid similar difficulties. These prob-lems may be avoided using exploration andtesting specifically tailored to mechanicalexcavation, and by investigation for conditionswhich have been associated with such difficul-ties in the past .

What can be learned from these widely undoc-umented and unpublished experiences? Howwere these conditions missed? What are theunderlying causes and motive mechanisms?How can such problems be avoided in thefuture?

In all case histories presented here, problems ordifficulties were encountered . It is the intentof this paper to use these negative experiencesto produce new possibilities for avoiding themin the future .

INTRODUCTION

In 20 years of consulting practice and con-struction management, we have come toaccept the unexpected and the unexplainable,and still complete the project, sometimeswithout delay. Years after agonizing experi-ences, we learn from new experiences andponder possible explanations . This paper issuch a speculation, based sometimes onmeasurable geotechnical conditions, occasional-

195

Tarkoy, P .J . and Marconi, M . (1991), Difficult rockcomminution and associated geological conditions, In :Tunnelling '91, Sixth International Symposium, 14-18 April1991, Hovotel, London, England, Instutute of Mining andMetallurgy, Elsevier Applied Science, London, pp . 195-207,

ly on quantitative performance and machineparameters, and, now and then, on reports andsecond hand descriptions .

Not intended as a purely scientific endeavor,this paper is a simple pragmatic approach thatmerely documents common effects, somegeotechnical conditions common to severalprojects, and possible causes . There may bemore than one explanation or other explana-tions which are not considered here . Extensivereferences and sources of information areprovided to permit independent evaluation ofour analyses .

Our purpose is to establish a new distinctionthat will make it possible to measure TBM rockcomminution efficiency. The measure of effi-ciency must take into account geotechnicalcharacteristics and mechanical variables of theequipment .

Our ambition is to sow the seed here for fur-ther investigation and a more rigorous solutionto the problem . It is essential to present pre-liminary findings in order that they may per-haps prevent others from such folly.

When projects experience difficulties such asunanticipated geological conditions, urgencygenerally dictates that all resources be directedto the solution of the problem and the recoveryof unanticipated costs . It is unlikely that agreat deal of effort will be expended to estab-lish an understanding of the underlying causefor future benefit. This has been true in manyof our projects (Token, 1985; Tarkoy, 1981a,1986a, 1988a, 1989, 1990; Thompson, 1988 ;Montacie, 1988 ; Marconi, 1989; and VanRyswyk, 1990) .

With increasing frequency, we have notedcommon patterns which have provided insightinto the problem. Our own experience withrock comminution (rock disintegration), primari-ly in tunnel boring, have led us to reviewunexplained similar case histories to find acommon thread . These experiences are offeredhere for the benefit of others, should they findit useful in avoiding similar events .

The appearance of common patterns are thefoundation of this paper . The fundamental and

most obvious common thread is difficult tunnelboring or difficult rock drilling, hence the refer-ence to difficult rock comminution .

THE COMMON THREAD

The investigation at Selkirk railroad tunnel(Figure 1) in British Columbia, Canada, was thefirst case where detailed operating data wasavailable and suggested that the encounteredrock appeared harder to the machine than anyof the rock mechanics tests indicated . Rockcharacteristics, intact or mass properties(uniaxial strength ; Total Hardness, Tarkoy,1973, 1975, & 1986b ; jointing ; stresses ;), andthe mechanical variables of the Tunnel BoringMachine (TBM), could not account for the verypoor TBM or drilling performance. The loss ofperformance could not be explained by meas-urable rock properties or machine variables .

It was general known that in-situ stresses werepresent at Selkirk from the excavation ofnearby and parallel tunnels . Speculation aboutthe impact of in-situ stresses on rock commi-nution by mechanical boring brought aboutdiscussions with The Robbins Company.Robbins brought to our attention a similarcondition encountered many years ago withone of their machines . The investigation of theunusual experience with an earlier RobbinsTBM, other tunnel boring projects, drillingexperience, experience with rock crushing, andindependent research on available project datahas provided insight into rock comminutionunder certain unique conditions . This underly-ing case history data will form the basis of thispaper.

FIGURE 1 : ROGERS PASS RR TUNNEL

THE FIELD PENETRATION INDEX

In an attempt to establish a quantitative

1 96

measure of phenomena not represented byconventional rock testing, a field indicator,which would reflect how the machine "saw" orinteracted with the rock, was developed andutilized . This field indicator had to take intoaccount the mechanical variables required toproduce efficient rock comminution .

The most effective energy expended in tunnelboring and drilling is the force applied normal tothe face. The measure of rock comminutionefficiency should take into account the forcenormal to the face and the depth of penetrationproduced. If measured for each revolution, itallows the scale to be independent of the toolor machine diameter and rotational rate . As aresult, the Field Penetration Index (FPI) wasdefined as follows :

F (kg)FPI, kg/mm =

Penetration (mm/rev)

Where :

F = gross cutter loadnormal the facein kilograms

gross cutter load =

total applied machinethrust or bit load /number of cuttingedges or bit contacts

Analyses of TBM performance based on de-tailed TBM records (machine variables andperformance) and comparison with TotalHardness, revealed consistency as well assome extraordinary variation . The resultsappeared to be particularly sensitive to TBMcutter load, tool condition, and rock "behavior"as seen by the TBM, and reflected by thecutting efficiency (height above the regressionline) .

Since the Total Hardness is essential to theapplication of the FPI, it bears defining . TheTotal Hardness was originally established as asimple, practical, and timely test for predictingboreability (penetration rate and cutter costs)by Tarkoy (1973, 1975, & 1986b) . It consistsof measurements made with the Shore Sclero-scope (D-type), Schmidt (L-type) Hammer, anda Modified Taber Abrasion Apparatus . Thetests require that specific and precise proce-dures be followed according to Tarkoy (1973) .The Total Hardness test procedures have beensubmitted to the American Society for Testingand Materials to assure standardization ofequipment and procedures . The Total Hard-ness is calculated as follows :

HT = HR1HA

The Total Hardness has been a reliable indica-tor of boreability for nearly a quarter of acentury. Total Hardness for earlier case histo-ries discussed in this report were approximatedfrom typical rock hardness for similar materials .Typical values are illustrated in Figure 2 .

FIGURE 2: RANGE OF TOTAL HARDNESS (After Tarkoy, 1975)

1 97

The FPI serves as a gage of cutting efficiencyas a function of cutter load and rock hardnessbased on penetration per revolution . It meas-ures the efficiency of cutting or amount of loadrequired to obtain a unit of penetration .

The FPI had been determined for a number ofprojects which encountered no difficulties withboreability and are designated as "normal" inFigure 3 . These sites include several subwaytunnels in Buffalo, NY, sewer tunnels in Chica-go, IL, New York City, and Toronto, Canada .These cases were used to develop the basicregression equation for successful or "normal"machine boring . The correlation between TotalHardness (Tarkoy, 1973, 1975, 1986a) andthe Field Penetration Index (FPI) for more thana dozen projects represented as "normal" wasextraordinary, with a coefficient of 85% .

The correlation between Total Hardness (repre-senting a multitude of intact and some rockmass characteristics) and mechanical variables(cutter condition, machine stiffness, etc .) isexcellent (> 75%) . This has confirmed thevalue of Total Hardness in predicting penetra-tion for efficient excavation . Points represent-ing inefficient machinboring, fall above the line .Large departures from the regression appear toindicate mechanical difficulties or unusualinteraction between equipment and rock mass .

One would conclude that inefficient excavationcannot be predicted by conventional rockproperties alone or even with known mechani-cal variables, in each and every case.

Even though some of the case histories used inthis analysis may be over twenty years old, thepossible interpretations have only been maderecently (Tarkoy, 1988a) . These conclusionswill be discussed under the section dealingwith individual case histories .

Significance Qf .4 Low FPI

An FPI lower than the regression line indicateswhat appears to be an unusually high efficien-cy in rock excavation . This has been related toa variety of geological or tunnel conditionssuch as highly fractured rock which may actu-ally break out with little contribution from themachine or machine conditions (such as allnew cutters) which are superior to the operat-ing average condition .

There is evidence for only one project in Figure3 where unusually high cutting efficiency (lowFPI) was sustained. This occurred during test-ing, when a completely new cutter dressinghad been placed on the cutterhead of theKerckhoff TBM .

200

Specific Rocks and TheirAverage Hardness Values

Rock Types and Their Known Range of Hardness

Igneous Rocks

I Metomorph .c Rocks I Sedirnenlory Rocks--BoraDoo Guortzite

C I190 1 I

,soII O I

I

no 1 II

I

,Go II I

( I II I

SOI

0I

140

average vawe toeI3o Rock Type

P0h500,5 O~aoase(NYC) F - Common Range

~Pegmatile

./----Uncommon Rang,0 INYC)

-Lower Granite 605011 EI

a o -

IS

loo

It I

I -

'a I Ia 90 -- Manhotton SanstINYC) I

I

I

I

I

I

iO So WissOhl<kOn Schist t

I

It I I -~(PndadelpnIa )Solennotsn LuMstone

I

II

NI

ITO_

I

I

I-iISolsnnolsn, Gsrmony )

glI

I

IHackensock Slltstone

t

ISO ~INew JerseyI

I Iro

50 ~Dolom t o Ls ICmcago,LllI

E0

I a40 IS -a

w $I

IE

20 ILewes Snob ISomh Central,

31Colo Norm Coma, N M .1

I

0

,t t

000 rn,nto Fm ISandslone

0- and S.1, Fo,mmglon,N .M.I

FIGURE 3 : FIELD PENETRATION INDEX(After Tarkoy, 1988a)

50

100

150

200

250

Jolal Hardness

Significance Q $ Hi ah FPI

Efficient cutting requires higher and highercutter loads per unit of penetration as rockcomminution becomes more difficult . Conse-quently, for harder rock, small changes in therock properties may cause a disproportionatedecrease in penetration as one crosses belowthe threshold of efficient cutting . In effect,TBMs operating successfully in hard rock (closeto their limitations), can come to a grinding haltwith only a small increase in rock resistance tocomminution . Similarly, TBMs operating belowtheir capabilities can sustain an increase in rockhardness without suffering an inordinate loss ofpenetration .

An FPI above the regression line indicates aninsufficient cutter load to deal with the encoun-tered rock properties or behaviour . Suchconditions have been encountered on severalprojects and are illustrated in Figure 3 .

Conclusions About the FPI

x

S smos

0

1 98

The Field Penetration Index has been useful inidentifying and evaluating difficulties of rockcomminution . It has become apparent that theField Penetration Index can be used as a gaugeof a TBM's interaction with the rock . HighFPIs appear to have been associated with aconspicuous absence of jointing, high in-situstresses, and microscopic features of the rockmass .

Often, the FPI is sensitive enough to expressthe loss of cutting efficiency between cutterchanges . In such a capacity, the FPI can beused to determine the most cost effective timeto change cutters .

These conditions commonly exist in "young"mountain chains, such as the Selkirk Mountainsin Canada, the Sierra Nevada in California,USA, in the French Alps on the border betweenFrance and Italy, and the Andes in SouthAmerica .

The Field Penetration Index is a tool which canbe used only after the fact . As such, it hasbeen useful to identify, define, and study theeffect of difficult rock comminution conditions .A separate means of predicting these condi-tions will be necessary.

CASE HISTORIES

Case histories to be discussed in this paper ispresented in Table 1 in chronological order ofconstruction. Table 1 summarizes projectlocations and elements held in common . Table2 provides a summary of ground conditionsand Table 3 summarizes relevant mechanicalcharacteristics of boring equipment used . Datafrom older projects was limited and less defini-tive, more general, and utilized average condi-tions which would not reflect the extremesfound in the greater detail of latter projects .

Mont ni

Experience in the Mont Cenis Tunnel in theFrench Alps during the mid-1960s first illus-trated the effect of tectonic and residualstresses on the rock behaviour as "seen" by aTBM. This experience has never been ex-plained or published (Tarkoy, 1988) . Commongeotechnical elements, geotechnical conditions,and machine characteristics are summarized inTables 1 through 3 .

In a section of tunnel about 100 meters long,high tectonic stresses were encountered andthe rock was reported to be as hard as"steel ."9 At the beginning of the machinestroke (0.33 m), the TBM bored at rates of 1 .8m/hr (upper Mt . Cenis point) . Beyond the firsthalf meter the Robbins TBM ground to a halt .After a pause of 5-10 minutes (or after re-gripping), during which the stresses wereallowed to relieve, it was possible to resumeboring at the normal rate. Behind the machine,the rock around the perimeter "exploded" orpopped resulting in overbreak of 8-10 cm(Montacie, 1988) .

TABLE 2: GROUND CONDITIONS

1 99

PROJECT

TOTAL

UNIAXIAL OVER- GROUND CONDITIONSHARD- STRENGTH BURDENNESS

MPa

meters

Mt Cenis

60

70- 124

1,200 schist & gneiss ; tunnel alignment parallel to foliation ; boring tectonicstresses much higher than overburden ; encountered popping rock ;

Star Mine

128-194 69 -351

2,225 Revette quartzite ; hard, brittle, intensely fractured ; encounteredpopping rock;

Libanon

83-225 125-275

2,100 hard quartzitic rock; high stresses ; popping or slabbing rock ;Tunjita

160

68-241

N/A

schist and very hard quartzite (98 .8 % silica) ;Carhuaquero 160-200 169-210 1,100 very hard and tough igneous lithologies consisting of andesite, da-

cite, rhyolite, tuff, & agglomerates ; very hard aphanitic andesites andwelded tuffs exceeded H T > 200; rock above tunnel alignment wasmassive, hard, glistened in the sun, and formed vertical valley walls ;Besides the high strength, the rocks were also tough and absorbed agreat deal during testing and boring .

Kerckhoff

175

84-140

N/A

massive granodiorite;Selkirk

111-136 56-211

2,000 metasediments, quartzites, phyllites ; high in-situ stresses (document-ed in nearby tunnels) were anticipated and "popping rock" was infact encountered .

Paute

130-147 68-313

1,000 very massive granodiorite;Esquimalt

N/A

43

0

jointed granodiorite with meta-andesite and quartz-plagioclase-porph-yry ;

Quarry

N/A

N/A

0

granodiorite ;

TABLE 3 : MECHANICAL DATA OF EQUIPMENT USED

TABLE 1 : COMMON ELEMENTS OF DIFFICULT ROCK COMMINUTION

PROJECT LOCATION LITHOLOGY STRESSES QUARTZ INTRU- RELIA-OR STRESS SUTURING SIVES BILITYFEATURES

Mont Cenis France MetaSediments Yes Unknown Unknown FairStar Mine Idaho Quartzite Yes Unknown Unknown FairLibanon Africa Quartzite Yes Unknown Unknown FairTunjita Columbia Quartzite Yes Unknown No FairCarhuaquero Peru Igneous Unknown Unknown No FairKerckhoff California Granodiorite&lntr Yes Yes Yes GoodSelkirk Canada MetaSediments Yes Yes No ExcellentPaute Ecuador Granodiorite & Intr Yes Yes Yes ExcellentEsquimalt Vanc ., Is . Granodiorite, Etc. Unlikely Yes Yes ExcellentQuarry Vancouver Granodiorite & Intr Local Yes Yes Excellent

PROJECT Year TBM MachineNumber

NEWorUSED

Dia-meterm

TunnelLength

m

CuttingEdgeLoad kg

Cutter-Heedrpm

PenetrationRatem/hr

Mont Corns 1965 Robbins 074-115 N 2 .2 300 5,500 (avg) 14 . 0.18-1 .8Star Mine 1969 Jarve Mk8 .900 U 2 .74 125 1,000 (avg) 15 .5 0.12Libanon 1974 Robbins 114-163 N 3 .35 N/A 14,500 (avg) 6 .9 0.80-2 .1Tunjita 1978 Robbins 145.168 N 4.3 10,500 20,000 6 .1 0.7Carhuaquero 1981 Jerva Mk-12 N 3 .8 8,500 22,700 10 .2 1 .1-9 .0Kerckhott 1981 Robbins 243-217 N 7 .3 6,700 16,000(avg) 5 .8 .78-2 .39Selkirk 1985 Robbins 222-183-1 U 6 .8 8,350 17,600 (max) 8 .2 2 .2-2 .6Paute 1985 Demag TV8OHA N 7 .8 6,000 26,000 (avg) 5 .12 .32- .90Esquimalt 1988 N/A N/A N/A 0 .5 N/A N/A N/A N/AQuarry 1989 N/A N/A N/A 0 .2 N/A N/A N/A N/A

The phenomenon, experienced in terms of TBMpenetration and popping rock, illustrated theeffect of high tectonic stresses on a relativelylow strength rock when bored by a TBM . Ineffect, the tectonic stresses made the 70-124MPa rock appear much harder to the TBM thancould possibly have been indicated by theuniaxial strength . Some of the difficulty wasalso attributed to having to break the foliation(parallel to the tunnel) on "edge ."

Similar experience was sustained by a WirthTBM on another section of the project, be-tween Mont Cenis and Grenoble .

The FPI, after the stresses were relieved duringre-gripping, designated as Mt. Cenis (N) inFigure 3, was consistent with the regressioncurve. The FPI would increase to infinity as itwould grind to a complete halt ; however, theFPI at a rate of 0 .18 m/hr was over 25,000kg/mm, higher than any of the other casehistories .

The problems associated with this project initi-ated the development of the Union IndustrielleBlanzy-Ouest (Unibo) TBM . The Unibo hadthree arms, each of which had one cuttermounted on it . The arms would traverse theface from the smallest diameter to the fulldiameter to cut the entire face . This allowedstresses to be relieved during boring . ThisTBM is currently known as the "BouyguesTunnel Boring Machine ."

Star Min Idaho . USA

Common geotechnical elements, geotechnicalconditions, and machine characteristics aresummarized in Tables 1 through 3 .

At the Star Mine, Idaho, a TBM was chosen fordrilling exploratory and development drifts in agold mine (Hendricks, 1969) . High in-situstresses and popping rock were encounteredwhile penetration rates were far below whatwas reasonable with available cutter loads .The inefficiency of the TBM is reflected by acalculated FPI > 8,000 kg/mm . Popping rockwas reported and in some sections where thepenetration was less than 0 .12 m/hr (Tarkoy,1975) .

The fact that the TBM could apply a cutter loadwhich was barely adequate or at times inade-quate for the encountered rock, is now consid-ered the primary cause of the difficulties ofboring . Although the quartzite was hard, itcontained many micro- and macro- fractureswhich weakened the rock mass and shouldhave made the rock comminution less difficult .

The stresses, however, held the fractured rocktogether and the face was relatively impenetra-ble to the TBM, where the stresses had notbeen relaxed. However, when the face had notbeen advanced for a period of time, the stress-es caused slabbing and fallout . The slabbingcaused problems at the face and the cutter-head. This project illustrates two extremeproblems and their effect on the TBM .

200

The use of the TBM in the Star Mine wasconsidered a failure and it was removed afterboring only 125 meters .

Li n n gold Mine, South Africa

Common geotechnical elements, geotechnicalconditions, and machine characteristics aresummarized in Tables 1 through 3 .

The Libanon Gold Mine in South Africa utilizeda TBM to excavate access drifts . Highlystressed rock was encountered and broke loosebehind grippers as they were released . Oursuspicion that the stresses also affected TBMpenetration were confirmed by an analysis ofthe average penetration rates and typical cutterloads utilized during tunnel boring .

Calculations using average boring conditions(Thompson, 1988) produced an FPI > 9,000kg/mm, high enough to indicate a considerabledeparture from the normal values (regressionline in Figure 3) of about 4,500 - 5,000kg/mm .

T n i Hydroelectric, Colombia

The Tunjita Hydroelectric Project was excavat-ed between 1978 and 1985 (Marconi, 1990) .Two tunnels were driven as follows :

1 . a 14.5 km long tunnel with anexcavated diameter of 4 .3-4.5 m bythe 144-151-1 Double ShieldedRobbins TBM and

2 . a 10.5 km long tunnel, excavateddiameter of 4 .3-6.0 m, by the 145-168 Robbins TBM .

Robbins TBM (144-151-1)

The Double Shielded Robbins TBM excavated 5km in shale and siltstone and 2 .5 km in lime-stone. There were no major difficulties encoun-tered in this tunnel . The remainder of thetunnel was excavated by drill and blast .

Robbins TBM (145-168)

An open design Robbins TBM excavated 1 .5km in very hard quartzite with a cutter life ofonly 10 cubic meters (0.7 m of tunnel) percutter . When drilling 76 mm exploratoryboreholes the diamond bit life was only 0 .7meters. The problem of hard abrasive rock wascompounded by the inflow of thermal water ( >42°C) in the quartzitic portion of the tunnelwhich caused problems with machine compo-nents, washing away muck fines, and causingbreakdown of lubrication components .

The very hard quartzite resulted in low penetra-tion rates and high FPIs (higher than the re-gression line in Figure 3) in spite of very highcutter loads . An analysis of the TBM excava-tion revealed that the 145-168 Robbins TBM

sustained an FPI > 10,000 kg/mm . Since verylittle detail of this project is known, additionalopinions cannot be formulated .

Carhuaauero Hydroelectric, PM

TBM performance was generally satisfactory,except in the harder rhyolites . Analysis of TBMperformance data confirms that some of therock behaviour produced very low penetrationrates (Tokes, 1985 ; Ivan-Smith, 1988; Tarkoy,1988), requiring an FPI > 12,000 kg/mm . Theaverage encountered FPI, based on averageconditions, conceals a far greater inefficiencythat could have been identified with more de-tailed performance data and analysis .

The Carhuaquero project illustrates generallysatisfactory performance, some marginal per-formance, some inefficient performance, andalso serves to confirm difficulties encounteredelsewhere in young mountain chains such asthe Andes of Colombia and Ecuador .

Kerckhoff Hydroelectric . California

Common geotechnical elements, geotechnicalconditions, and machine characteristics aresummarized in Tables 1 through 3 (Kennedy,1982; Woodward, 1983 ; Tarkoy, 1985) .

The Kerckhoff 2 Underground HydroelectricPower Plant Project was bored by a new state-of-the-art Robbins TBM . The rock turned outto be consistently more difficult to cut (harder)than anticipated by anyone, including themanufacturers of the TBM, The RobbinsCompany.

It was reported that the quartz grains in thediorite were stress sutured, suggesting tectonicstresses had affected the petrofabric of therock .

The relationship between cutter load and cutterpenetration for the granodiorite is illustrated inFigure 4 . The curves suggest that a thresholdof efficient cutting exists . Below this thresh-old, the cutting is less efficient and above it,more efficient . The differences in efficiencycan be attributed to the differences in the useof energy to produce crushing or chipping .

FIGURE 4: CUTTER LOAD & SHARPNESS

201

Data in Figure 4 illustrate the :

1 . sensitivity of penetration rates tocutter load,

2. sensitivity of penetration rates to thecutter profile (state of wear), and

3. reduction of penetration as cutterswear.

The rock hardness encountered at Kerckhoffappears at times to have been close or beyondthe efficient cutting limit of the TBM . In ef-fect, the TBM running tests, plotted on Figure3, illustrate that cutting efficiency decreases(increasing FPI) as the :

1 . cutter wears and

2 . as the rock hardness increases (illus-trated conversely by Figure 3 as thecutter load decreases) .

In other words, harder rock causes lowerpenetration and higher cutter wear. Conse-quently, lower penetration also increases cutterwear per unit volume of rock cut . Therefore,the cutters are in a duller condition for longerperiods of time between cutter changes, evenwith more frequent cutter changes . As a re-sult, the penetration is decreased even more asa consequence of the dull cutters .

The owner compensated the contractor forextra costs sustained as a consequence oflower penetration rates, higher cutter wear,and delays associated with having encounteredharder than anticipated rock (marginal to theefficiency of the TBM) .

Selkirk Railroad Tunnel, BC Canada,

High in-situ stresses encountered in varioustunnels in the Selkirk Mountains of BritishColumbia, Canada, suggested that stressesshould be anticipated in the nearby SelkirkRailroad Tunnel . Subsequently, "popping rock"was encountered in sections of the SelkirkTunnel . Common geotechnical elements,geotechnical conditions, and machine charac-teristics are summarized in Tables 1 through 3(Knight, 1987 ; Tanaka, 1987 ; Tarkoy, 1988) .

The portion of the Selkirk tunnel excavatedwith a Robbins TBM sustained less than antici-pated penetration rates, penetration rates lowerthan normal with the available TBM, exceeding-ly high cutter wear, and a broken main bearing .Penetration rates dropped and cutter wearskyrocketed when crossing into the garnetisograds of metamorphism (a degree ofmetamorphism typically producing an abun-dance of garnets) . Conversely, penetrationrates rose and cutter wear dropped whencrossing out of the garnet isograd . No clue,from intact or rock mass properties, was avail-able to indicate that this behaviour should beanticipated within the garnet isograds .

A 30 point moving average, to attenuate

extreme variation of the encountered FPI, isillustrated in Figure 5 . It is immediately evidentthat the encountered FPI was significantlyhigher than the average anticipated value of2,500 and the highest values occurred be-tween the garnet isograds. The variation inthe FPI, even with a 30 point moving average,is still notable, The variation reflects a sensitiv-ity to the condition of the cutters, therebyproviding a measure of cutter condition . Thecurve illustrates the unusually high energy perunit penetration required to excavate the rock .

FIGURE 5 : FPI AT SELKIRK

ui

F0

30pt Moving Average

E

Ywoo

Anticipated

Garnet Isograd

024300

:6500

76500

50500

'Choinoge. meters

Figures 5 & 6 illustrate differences betweenanticipated and encountered (encountered -anticipated) penetration rates and cutter costs .

These figures represent performance above orbelow expectations . Performance as expected,corresponds to zero. As tougher rock wasencountered, higher cutter loads were applied(increased from 18,000 to 22,000 kgf) tocompensate and maintain penetration . In spiteof an increase in cutter load, the difference inpenetration increased (dropped) dramaticallybetween chainage 260+00 and 290+00 asillustrated in Figure 6 .

FIGURE 6 : PENETRATION AT SELKIRK(note the location of the garnet isograd)

' .6

E

FormationAverage

, /J

202

Differences between anticipated and encoun-tered cutter costs also skyrocketed betweenchainage 260+00 and 290+00 as illustratedin Figure 7. Cutter costs are normally infinitelyhigh at the start of boring since the originalcutter dressing cost has bored no rock (dress-ing cost, $/0 meters) . As the TBM begins tobore, cutter costs typically continue to drop,approach, and remain near to anticipated costs( - 0 on the ordinate is equal to "no cost over-run") . Normal saw tooth movements of thecurve represent cutter changes, generally in theouter positions first since these sustain thelongest rolling paths in the shortest time . Withcontinued boring, fluctuations of the curvebecome more subdued since the curve iscumulative. Consequently, the increasingoverrun of cutter costs at the garnet isograd(Chainage 260+00) is considered precipitous .

FIGURE 7 : CUTTER OVERRUNS(note the location of the garnet isograd)

600

74500 26500

7000

50500Choinoge. meters

32500

The severe reaction of the FPI, penetrationrates, and cutter costs at the garnet isogradsreflects a prominent change in the nature ofthe behavior of rock between chainage260+00 and 290+00. This is consistent withthe theory that the formation of garnets inmetamorphism represent radical changes ofmineralogy and mechanical properties of therock. Garnets are difficult to nucleate andgenerally represent conditions of formation farbeyond conditions required for stability.Consequently, these changes tend to be radi-cal, as exhibited in Figures 4 through 6 . Inter-estingly, these changes do not appear to havebeen identified by conventional exploration ortesting techniques . Further investigation willbe required to determine the precise rockcharacteristics which cause the TBM to reactand which were not discovered prior to con-struction .

In any case, the contractor was compensatedfor unanticipated conditions ; consequently,future investigations no longer have the sup-port of financially interested parties .

e

Garnet0

Isoctrado: -.sC

a .t .s

uc V 7-a

b xsoo xeeoo

32!00Choinoge . meters

Paute Hydroelectric Project, Ecuador

The Paute Hydroelectric Project, Phase A wasconstructed between 1977 and 1983 andconsisted of a concrete dam (1,200,000 cubicmeters of concrete, 170 meter high gravityarch, and crest length 400 meters), spillwaycapacity 7720 cubic meters/sec, reservoir 120million cubic meters, five 100 MW units with667 meters of head, and a power tunnelapproximately 6 km long and 5 meters indiameter, excavated by D&B (Marconi, 1990) .

During excavation popping rock was encoun-tered in a few sections of tunnel . Very littlerock support, primarily rock bolts, were re-quired . Water inflow was minimal . The tunneltraversed schists, hornfels, and granodiorite .

The second stage, Paute Phase C, was con-structed between 1985 and 1991 . It includeda total of 9 km of tunnel and shaft excavationand the doubling of the installed generatingpower in a new powerhouse excavated andbuilt adjacent to the first stage one . The 6 kmlong, 7 .8 meter diameter power tunnel and the300 meter long access tunnel were excavatedby TBM . The alignment of this tunnel wassubparallel to the one driven for Phase A at adistance of 80 meters further into the moun-tain. Common geotechnical elements, geo-technical conditions, and machine characteris-tics are summarized in Tables 1 through 3 .

Excavation was started at the downstreamend, beginning in schist, traversing hornfels,and ending at the upstream portal in a verytough fine to coarse grained granodiorite. TheFPI (Figure 8) and cutter costs increased andpenetration rates decreased (boring towarddecreasing chainage) with increasing distancefrom the valley walls .

FIGURE 8 : FPI AT PAUTE

zoooo .EEaY

. 17000400

0 '0000aN

soon

awo

20pt Average-+

uoo

eioo

sQ0

Choinoge, metersr3W

A very high Field Penetration Index, sustainedbetween chainage 60+00 and 65+00 as illus-trated in Figure 8, reflects damage sustainedby the TBM. This damage was repaired andthe FPI in the hornfels again became consistentwith earlier values in the schist . The FPI first

203

began to rise in the hornfels and continued torise in the granodiorite . Ignoring the earlier risein the FPI due to mechanical problems, anincreasing FPI becomes evident in the hornfelsand granodiorite, especially when examiningthe lithological averages in Figure 8 (Tarkoy,1989) .

As a result of encountering the increasing FPI,the contractor carried out extensive site andlaboratory investigation . The investigationsdetermined that a high residual horizontalstress field was originated at the time of intru-sion of the granitic rock mass and duringsubsequent tectonism . This state of stresswas not relieved in spite of the deep valleyerosion .

Even though the first tunnel provided an indica-tion of anticipated conditions, the drill and blastexcavation could not foretell the behaviour ofthe rock as "seen" by the TBM . The FPI antic-ipated for Paute "C", was expected to be FPI< 4,000 kg/mm . The average encountered FPIfor schist, hornfels, and granodiorite, were7,720, 7,450, and 9,200, kg/mm, respective-ly.

It is generally known that as the rock hardnessand/or the strength increase, a higher cutterload is necessary to maintain efficient cuttingabove threshold . This phenomenon has beenillustrated for rocks of various strengths andthresholds of efficient excavation by Robbins(1972). Consistently high cutter loads, inexcess of 23,500 kgf, had to be applied,especially in the granodiorite sections of thetunnel .

Recorded cutter loads confirm that every avail-able bit of TBM capability (over what wasanticipated or required) was applied to therock . The IBM had capabilities far in excess ofthose required by anticipated rock conditionsas indicated by the high cutter load capabilitiesactually used . However, even this excesscapability appears to have been less thanadequate to deal with the increased resistanceof the rock mass to boring . The TBM was ableto deal with the softer and more brittle schistnear the valley walls, where stresses are likelyto have been relieved by erosion of the valley .

Observations in the tunnel suggested a possiblerelationship between FPI, joint frequency,increase in cover, stress phenomena, andpenetration rates . Consequently, quantitativemeasurements of joint frequency, stressphenomena, were made for comparison withpenetration rates. The raw data showed atrend, however, expected scatter reflectedmany variables that influenced penetration,including variations in cutter load, cutter condi-tion, jointing, stresses, etc .

It was found that as joint frequency increased,it appeared initially to improve and later de-crease penetration rate (Tarkoy, 1989) . Pene-tration rate tends to suffer as jointing becomesmore frequent, as in urban tunnel environ-ments, where the rock mass is in an advancedstate of degradation compared to conditions

found in young mountain chains . A similar plot(Tarkoy, 1989) illustrated that a decrease inthe penetration rate was associated with anincrease in stress related phenomena (poppingrock, stress slabbing, and stress fracturing) .

Based on detailed petrographic examinationsand unconfined and confined compressionstests, Buechi (1990) concluded that "tectonicstresses produce an immediate micro-crackingin the crushed zone area due to relaxation" and"This immediate micro-cracking enlarges thecrushed zone which means that more thrustwill be necessary to build up the pressure levelto create further median cracks for penetrationand lateral cracks for chipping of the rock ."

As a result of the high in-situ stresses andassociated conditions present in a youngmountain chain, it is not surprising that thelack of joints, increasing cover, and tectonicstresses appear to have had an impact onpenetration rates and comminution efficiencyas measured by the FPI .

Coincidentally, the Carhuaquero, Tunjita, andPaute projects are both located in the geologi-cally young Andes mountain chain .

Esauimalt. Vancouver Island

During the drilling of shallow pile sockets, aproblem of very slow drilling rates and high bitwear were encountered at Esquimalt, Vancou-ver Island, BC, Canada. Common geotechnicalelements, geotechnical conditions, and ma-chine characteristics are summarized in Tables1 through 3 (Tarkoy, 1990) .

The differences in performance could, in part,be explained by differences between anticipat-ed and encountered uniaxial strengths .However, some of the drilling difficulty andgeotechnical conditions had an uncannyresemblance to previous experience. Thesimilarities included unpredictability of commi-nution behaviour, very low and unanticipatedpenetration rates, presence of granodiorite, andstress sutured quartz grains .

Since the site geology was in a comparativelyadvanced state of degradation with well de-veloped jointing and no apparent residualstresses, this project's contribution is limited .However, it serves to illustrate that effects ofprevious tectonic and stress history reflected inlithology and in the sutured quartz grains mayno longer apply when altered .

The discovery of sutured quartz did, however,identify at least one feature, in addition to thepresence of granodiorite, which might explainsome of the unanticipated drilling difficulty .

Vancouver Quarry

At about the same time as the work at Esqui-malt, a contractor supplying railroad ballastencountered unpredicted and unusually slowdrilling rates, high bit wear, and high crusher

204

wear in one part of his quarry. Commongeotechnical elements, geotechnical conditions,and machine characteristics are summarized inTables 1 through 3 .

This project's contribution is that it confirmsexperience at Esquimalt. The conditions wereunanticipated and occurred only in one sectionof the quarry. However, a geologist (VanRyswyk, 1990) investigating the unanticipatedconditions found stress sutured quartz crystalsin the area responsible for the difficulty . Thestress suturing of quartz in granodiorite resem-bled the conditions found at Kerckhoff, Paute,and Esquimalt .

A number of similarities between the localgeology and the foregoing projects becameevident .

Inadeauate TBM Capabilities

While developing the Field Penetration Index toidentify unusual geotechnical phenomena intunnel boring, it became apparent that it wasalso useful as a measure of :

1 . TBM capabilities (particularly usedTBMs),

2. boring efficiency, and

3. condition of the cutters .

In two cases illustrated in Figure 3, unantici-pated geotechnical conditions (harder thananticipated rock) combined with limited TBMcapacity resulted in exceeding the mechanicallimits of the machine and efficient tunnelboring . The original investigations of the inef-ficient behaviour of this TBM on two separateprojects provided adequate data to calculatethe FPI . On the first project, the FPI confirmedthat TBM capabilities were adequate in theanticipated geotechnical conditions . Moreimportantly, the calculated FPI verified that theTBM was operating far beyond its capabilitiesin the unanticipated conditions .

A number of case histories are therefore pre-sented herein to contrast with the foregoinganomalous geotechnical conditions .

The WSSC-80 project in Bethesda, Maryland, acontract required TBM excavation of only 1000meters of 2 .2 meter diameter tunnel (Tarkoy,1981b). A relatively old machine, capable ofcutting rock having an average Total Hardnessof 95, was used. When the machine encoun-tered consistently harder rock on "edge" (atright angles to foliation) averaging a TotalHardness of 125 with a maximum of 160, itwas unable to perform efficiently and penetra-tion dropped below 0.3 m/hr.

At penetration rates of 0 .24 m/hr, failure of thecutterhead (cracking and bending), a high rateof cutter failure, and failure of hydraulic sys-tems, there was the danger that the TBM couldnot complete the project . The FPI exceeded9,000 kg/mm .

Subsequently, the contractor was awardedcompensation by a board of arbitrators fordecreased penetration rates, increased cutterwear, and delays resulting from encounteringunanticipated hardness of rock that wasbeyond the efficient capabilities of the TBM .

The same TBM was subsequently used at theKiena Mine, Val D'Or, Canada, to excavateexploratory drifts in a greenstone belt . TheJarva TBM was refurbished and fitted with anew cutterhead (Vanin, 1988) .

When encountering hard rock, the FPI in-creased to nearly 6,000 kg1mm (Tarkoy,1986b). The available cutting edge loads of9,000 kgf on the older machine were inade-quate to cut hard rock efficiently. Maintainingthe cutter loads as high as possible to preservethe penetration rate, the main bearing deterio-rated (Vanin, 1988) .

Problems encountered on this project weresimilar to problems experienced on the WSSC-80 project and without the benefit of differingsite conditions .

A rigorous evaluation of the TBM thrust,torque, cutter loads, and anticipated geologicalconditions could have prevented the problem .

SUMMARY AND CONCLUSIONS

The phenomena of difficult and unexplainablerock comminution have, in several recentcases, been detrimental to project performanceand caused considerable cost overruns . Recentcase histories with substantial geotechnical andconsiderable mechanical performance datamade it possible to investigate the unanticipat-ed and unusual manifestation of rock comminu-tion phenomena . Investigation into similarbehaviour on earlier projects allowed the identi-fication of a number of common elements .These elements have been identified and theireffect on rock comminution examined . Theinvestigation was necessarily limited to practi-cal resolution of the problems on a project levelrather than on a purely scientific basis . Conse-quently, studies were limited to the client'simmediate concerns .

Detailed studies of two projects and generalknowledge of a number of others, cannot leadto definitive or ultimate scientific conclusions .However, our knowledge has been advancedenough to have identified problems and tomake recommendations to them in the future,even without fully understanding the underlyingcause.

Summary

The foregoing discussions have documented anumber of features that have been associatedwith unanticipated rock comminution difficul-ties . The difficulties have not been predictableby classical methods of geotechnical explora-tion (core borings ; D&B tunnel experience ;

205

survey of structural geological ;) and testing(uniaxial strength and Total Hardness) .

The features in common to several projectsmay be categorized as follows :

1 . Granodiorite Batholith

a. with stress deformations (undu-lous extinction, prismatic wallswith preferred orientation,deformation twins, and pressursolution, sutured grain bound-aries; Buechi, 1990),

b. high tectonic (non-hydrostatic)stresses

c. scarcity of jointing, and

d. high cover,

e. sites found in young mountainchains,

2. Metamorphic Rock

a . within a high grade of metamor-phism (garnet isograd) and

b. sites found in young mountainchains .

Since the phenomena that caused difficult rockcomminution are difficult to detect directly,sites in the foregoing two geological domainsserve as adequate justification for additionalinvestigations .

Conclusions

The advance of technology has allowed theboring of increasingly harder rock . With tunnelboring in new domains such as hard rock(granodiorites) and young mountain chains newproblems have been encountered . Experienceindicates that anomalous geological phenomenamay be encountered that are not predicted bycurrently used typical exploration and testingmethods, capable of having a substantialimpact (an order of magnitude) on mechanicalrock comminution, exceedingly costly and haveenormous consequences on project completionand costs .

Reported difficulties with rock comminutioncaused by geotechnical conditions or mechani-cal inadequacies have been recognized andquantified with the Field Penetration Index .The Field Penetration Index is an effective andquantitative distinction able to quantify rockcomminution efficiency. The FPI is equallyeffective in anomalous geotechnical conditionsand as with substandard performance due toTBM deficiencies .

Since neither classical exploration techniquesnor a single test can predict the types of prob-lems and conditions discussed in this paper, analternate technique must be found. We con-clude that the only possibility of detecting

difficult rock comminution is to make furtherinvestigations when the conditions common tothe discussed projects are found . It is ouropinion that sites anticipating granodiorite, highstresses, tunnelling within garnet isograds ofmetamorphism, or young mountain chains,warrant more intensive investigations . Theseinvestigations should include petrographicanalyses seeking stress phenomena, micro-cracking, and triaxial compression tests todetermine effects of stress relief on testsamples .

REFERENCES

Blackburn, W. H . (1990) . Personal Communica-tion, University of Windsor, Ontario, Canada .

Buechi, E . (1990) . Rock Mechanic and MicroFabric Analyses, Geotest, Report No. 89947-1&2, Switzerland .

Deere, D .U. (1968) . Geologic Considerations,in Rock Mechanics in Engineering Practice,edited by K .G . Stagg and O.C. Zienkiewicz,John Wiley & Sons ., ppl-20 .

Diet), B. and Tarkoy, P. J . (1974) . Etude de laDurete des Roches et des Vitesses d'Avance-ment d'Une Machine Foreuse dans les Schistesde Manhattan . [A Study of Rock Hardness andTunnel Boring Machine Advance in ManhattanSchist .] Tunnels et Ouverages Souterrains2:97-99 (March-April) .

Golder Associates (1981) . Carhuaquero Hydro-electric Project, Draft Report on AnticipatedGeotechnical Conditions for the Proposed LowPressure Tunnel, February.

Hendricks, R. S. (1969). Hecla Mining Compa-ny Raise Boring and Tunnel Boring MachineExperience, Proceedings, Second Symposiumon Rapid Excavation, Sacramento, October 16-17 .

Ivan-Smith, A. (1988) . Personal Communica-tion, Projecto Carhuaquero.

Kennedy, E . R . (1982) . The Kerckhoff 2Underground Hydroelectric Power Plant Project,Tunneling Technology Newsletter, June.

Knight, G . B. (1987) . TBM Excavation ofRogers Pass Railroad Tunnel, ProceedingsRETC, pp 673-683, New Orleans, June .

Marconi, M . (1989-90) . Personal Communica-tion, Projecto Tunjita and Paute .

Montacie, M . (1988). Personal Communica-tion, the Mt . Cenis tunnel .

Rast, Nicholas (1990) . Personal Communica-tion, University of Kentucky.

Robbins, R. J . (1972) . Economic Factors inTunnel Boring, Proceedings, Symposium onRaise and Tunnel Boring, Australian Geome-chanics Society, 13p .

206

Sugden, D. (1988) . Tunnel Boring Consultant,Personal Communication .

Tanaka, R . S . (1987) . CP Rail Tunnels in theRogers Pass, British Columbia, Canada, Pro-ceedings RETC, pp 1174-1184, New Orleans,June.

Tarkoy, P. J . (1973) . Predicting TBM Penetra-tion Rates in Selected Rock Types, Proceed-ings. 9th Canadian Symposium. on RockMechanics, Montreal, Dec. 13-15 .

Tarkoy, P. J . (1975) . Rock Hardness IndexProperties and Geotechnical Parameters forPredicting TBM Performance, UnpublishedPh.D. Thesis, University of Illinois, September,327p .

Tarkoy, P. J. and Hendron, A .J ., Jr . (1975) .Rock Hardness Index Properties and Geotech-nical Parameters for Predicting TBM Perform-ance, 325p, Final Report of NSF RANN Re-search Grant GI-36468, available from NTIS,Springfield, VA 22151, Document number PB246293 or RA-T-75-030 .

Tarkoy, P. J . (1979). Predicting Raise andTunnel Boring Machine Performance : State-of-the-Art, Proceedings Fourth RETC, Atlanta,GA, June.

Tarkoy, P. J. (1981a) . Carhuaquero ProjectFiles .

Tarkoy, P. J . (1981b) . WSSC-80 Project Files .

Tarkoy, P. J . (1981c) . Tunneling MachinePerformance as a Function of Local Geology,Bulletin AEG, 18 :2, May, pp 169-186 .

Tarkoy, P. J . (1985) . Kerckhoff 2 Project Files .

Tarkoy, P. J . (1986a) . Practical Geotechnicaland Engineering Properties for Tunneling-BoringMachine Performance Analysis and Prediction,Transportation Research Record 1087, Trans-portation Research Board, Washington, DC .

Tarkoy, P. J . (1986b) . Kiena Mine ProjectFiles .

Tarkoy, P. J . (1988) . Analysis of Anticipatedand Encountered Geological Conditions andTBM Performance, Rogers Pass Tunnel - CPRail, for Selkirk Tunnel Constructors JV, 63p, 6June.

Tarkoy, P. J . and Wagner, J.R. (1988) . Back-ing up a TBM, Tunnels and Tunnelling, 20:10,pp. 27-32, October.

Tarkoy, P. J. (1989) . Analysis of DifficultBoring at Project Paute Phase "C", Prepared forImpregilo SpA, 67p .

Tarkoy, P. J . (1990) . CFB Esquimalt ProjectFiles .

Thompson, D. (1988) . Personal Communica-tion, Libanon Gold Mine .

Tokes, 1 . (1985) . Personal Communication,Projecto Carhuaquero.

Vanin . D. (1988). TBM at Kiena Gold Mines,World Tunnelling, pp. 145-149, June.

Van Ryswyk, R . (1990) . Personal Communica-tion, Vancouver Quarry .

Woodward, E . M . (1983) . Construction ofPacific Gas and Electric Company's Kerckhoff-2Underground Hydro-Electric Project 1980-1983, Proceedings RETC, Chicago, pp. 962-978 .

207