CORRELATION BETWEEN OBSERVED SUPPORT PRESSURE AND ROCK MASS QUALITY

Authors: Bhawani Singh, J. L. Jethwa, A. K. Dube and B. Singh ABSTRACT The correlation between rock mass quality and support pressure proposed by Barton et al. (19 !" has pro#en use$ul% e&cept in cases o$ squee'in( rock (round conditions. )ield data collected systematically $rom *+ tunnel sections indicate a clear need $or correction $actors to account hei(ht o$ o#erburden and tunnel closure% which do not seem to be adequately accounted $or by the stress reduction $actor. As e&pected% the support pressure decrease rapidly with tunnel closure and then increases beyond a limitin( closure. That $act that the obser#ed wall support pressure were always close to 'ero e&cept in squee'in( (round conditions has taken care o$ by sli(htly modi$yin( wall $actors $or ,-wall. A criterion deri#ed $rom the $ield data shows that the squee'in( (round conditions would be encountered where the hei(ht o$ the o#erburden is (reater than ,1./. The data reported therein con$irm the earlier $indin(s o$ Barton et al. (19 !" that the support pressure is independent o$ the tunnel si'e. 01TR234CT021 The reliability o$ a realistic quantitati#e classi$ication system $or estimatin( tunnel support pressure has increased with the passa(e o$ time. 5#er since its de#elopment% the ,-system o$ Barton et al. (19 !" has attracted interest o$ tunnel en(ineer. $ield (eolo(ists and researchers. 0n spite o$ bein( o#erly comprehensi#e and complicated% this classi$ication method has now $ound acceptance. 6ethwa et al. (197*" measured the support pressure by load cells and contact pressure cells in se#eral steel-rib supported tunnel sections throu(h both squee'in( and elastic (round conditions and compared the measured #alues with those estimated a$ter ,system. The study brou(ht to li(ht si(ni$icant limitations o$ Barton8s methods $or application to tunnel sections under squee'in( (round conditions. )or e&ample% the support pressure is a $unction o$ tunnel closures% which% in turn% depend on the support sti$$ness. )urthermore% a tunnel at a (reater depth is likely to attract hi(her support pressure. The tunnel closure and there$ore the support pressure continues to built up $or a considerable time due to creep o$ the $ailed rock mass. 5mpirical correlation de#eloped in a e$$ort to eliminate the abo#e limitations o$ the ,-system are discussed herein. Because the proposed empirical correlation are based on only *! tunnel sections% there is scope $or re$inement.

R5C2R3019 2) )05:3 3ATA The $ollowin( $ield data were collected; a" Radius o$ tunnel e&ca#ation. b" 3epth o$ tunnel section $rom (round le#el. c" 4nit wei(ht o$ (round o#erlyin( the tunnel section. d" , o$ the rock mass around the tunnel section. e" R<R o$ the rock mass around the tunnel section. $" =oop load in steel ribs by compression load cells. (" Radial support pressure by contact pressure cells. h" Tunnel closure by the tape e&tensometers and closure meters. i" 3eep-seated radial displacement o$ the rock mass around the tunnel openin( by sin(le and multipoint borehole e&tensometers. , and R<R 0t is (eneral practice to di#ide a tunnel into se#eral rock mass units on the basis o$ the #ariation in the (eominin( conditions. 5ach rock mass is then assi(ned a , #alue% dependin( on the #alues o$the si'e parameters R,3% 6n% 6r% 6a% 6w% and SR). 0t has been e&perienced that a sin(le #alue o$some o$ these si& parameters is sometimes in$luenced by personal bias. There$ore% a ran(e o$ #alues is assi(ned to these si& parameters and a ran(e o$ , is obtained. The ran(e and a#era(e #alues o$ , obtained $rom tunnel sections are (i#en in Tables Al and A* in the Appendi&. The rock mass ratin(s R<R (a$terBieniawski 1971" were also obtained. The correlation o$ R<R and , pro#ided the necessary con$idence (6ethwa et al. 1971". >hene#er , #alue was doubt$ul% the doubt was re$lected in a wider ran(e o$ , and the absence o$ a close correlation with R<R. Support ?ressure Compression load cells o$ @+-1++ tonne capacity and contact pressure cells o$ @1@k(.cm* capacity were used to measure support pressure on steel ribs. The load cells were inserted into rib Aoints% a #ertical Aoint at the tunnel crown and two hori'ontal Aoints at the sprin( le#el. 1o load cell was installed at the bottom. The #ertical support pressure was obtained as a ratio o$ the arithmetic sum o$ the loads recorded by the two load cells installed at the sprin( le#el to the product o$ the e&ca#ation width and the rib spacin(. Since no load cell was installed at the bottom% the hori'ontal support pressure was taken as a ratio o$ the load recorded by the crown load cell to the product o$ hal$ the e&ca#ation hei(ht and the rib spacin(. The contact pressures cells were installed at the inter$ace o$ the steel ribs and the back-$ill. The load cell and the contact pressure cells were installed close to tunnel $ace. All o$ these instruments were protected a(ainst direct hit durin( blastin(. The blast #ibrations did not a$$ect these instruments.

Tunnel Closure 3iametrical de$ormations o$ the tunnel sections were measured by tape e&tensometers% closure meters% and sometimes e#en by simple in#ar tapes.The chan(e in tunnel diameter was hal#ed to obtain the radial tunnel closures. Type o$ Rack <asses The instrumented tunnel sections co#ered both hard rock masses such as quart'ites% metabasics% and dolomitesB and commonly occurrin( so$t rock masses such as shales% clays% slates% and phyllites. Criterion $or Squee'in( 9round Condition 0ncompetent or so$t rock masses characterised by low in-situ crushin( stren(th under(o plastic $ailure when o#er-stressed. Such a rock mass around a tunnel openin( $ails when the tan(ential stress e&ceeds its unia&ial stren(th. The $ailure o$ the rock mass is associated with #olumetric e&pansion% which is mani$ested in the $orm o$ radial inward displacement o$ the tunnel periphery called tunnel wall. These de$ormations are called tunnel closures. The tunnel closures can be #ery lar(e (measured closures ha#e been as lar(e as 1 C o$ the si'e o$ the tunnel openin(". This phenomenon is called 8squee'in(8 o$ the rock mass. The squee'e can occur not only $rom the roo$ and the sides% but also $rom the $loor. Tunnel closures resultin( $rom the elastic rela&ation o$ a tunnel openin(% on the other hand% are smaller than 1C o$ the tunnel si'e (see measured #alues in Tables Al and A*% Appendi&". Theoretical Criterion Theoretically% squee'in( conditions around a tunnel openin( would be encountered i$ ; (1" where is the tan(ential stress and qc is the unia&ial crushin( stren(th o$ the rock mass. 0n the case o$ a circular tunnel under hydrostatic stress $ield% 5q. 1 can be written as; *? (*" D qc D qc

in which ? is the primary stress #alue. 0t $ollows that a tunnel section e&periencin( elastic conditions in a (i#en so$t rock mass can encounter squee'in( conditions i$ the primary stress le#el increases due to increase in the tunnel depth or any other reason. This e&plains why phyllites and shales squee'e at one place and present elastic conditions at another% as shown in Tables Al and A* (in the Appendi&". 5quations 1 and * can thus be used to predict squee'in( conditions in a tunnel% pro#ided that ? and qc are known. 5mpirical Criterion <easurement o$ primary stress $ield and the in-situ crushin( stren(th o$ rock masses across a tunnel $or predictin( squee'in( conditions is both e&pensi#e and timeconsumin(. There$ore% an attempt was made to sock a simple criterion $or predictin( squee'in( conditions. An empirical criterion was de#eloped (as shown in )i(. 1" that (i#es a lo(-lo( plot between the tunnel depth = in metres and the lo(arithmic mean o$ the rock mass quality ,. Some o$ the case histories o$ Barton et al. (19 !" ha#e also been used in )i(.1. A clear line o$ demarcation between the elastic and the squee'in( conditions can be seen. The equation $or this line has been obtained as = (/" E /@+ ,8

Thus% a rock mass may under(o squee'in( when the depth o$ the tunnel section e&ceeds /@+ ,1./. Comparison o$ <easured and ?redicted Roo$ Support ?ressure Barton8s correlation% (i#en in 5q. ! below% was used to obtain predicted #alues o$ shortterm roo$ support pressure p%%. As discussed abo#e% measured roo$ support pressure #alues were obtained $rom instrumented tunnel sections. 2ut o$ a total o$ 19 case histories listed in Tables A1 and A*% 1F cases ha#e been included in this analysis. These 1F case histories in#ol#e 7 tunnel sections under non-squee'in( and 7 under squee'in( (round conditions. The measured roo$ support pressures ha#e been compared with the predicted #alues shown in )i(ure *. The comparison has not been shown $or the wall support pressure because the number o$ measurements is small. )or predicted roo$ support pressures% the classi$ication methods o$ Ter'a(hi (19!F"% 3ecre et al. (19F9"% ?rotodyakono#(19F/"%>ickham et al.(19 !"% Barton et al. (19 @"% and Bieniawski8s R<R method (supplemented by 4nal 197/" ha#e been used. 0t can be seen that the predictions are unreliable in all cases e&cept one. 0n the case o$ Barton8s ,-system% the predictions are reliable $or non-squee'in( conditions. The predictions turned out to be unsa$e $or squee'in( (round. )or e&ample% the measured support pressures were 1+.7 and 11.@ k(.cm* when compared to predicted

#alues o$ !.* and !.! k(.cm* $or tunnel sections * and !% respecti#ely. Such lar(e di$$erences in the measured and predicted support pressures prompted the authors to look $or possible reasons. )i(ure 1. Criteria $or predictin( squee'in( (round condition. )i(ure *. Comparison o$ predicted and obser#ed roo$ support pressures. Some o$ the data in Table A* that are related to these $our tunnel sections ha#e been shown in Table 1. 0t can be seen $rom Table 1 that in the cases o$ sections 1 and *% the di$$erence in support pressure could be the result o$ depth% tunnel closure% tunnel radius% and time o$ obser#ations. Similarly% in tunnel sections / and !% the di$$erence would be related to tunnel radius and tunnel closures. 0t $ollows that the $ollowin( $our $actors mi(ht ha#e in$luenced the measured support pressure; 1. Tunnel depth or thickness o$ the o#erburden% *. Tunnel closure% /. Time% and !. Tunnel si'e. 0$ other $actors are unchan(ed% the tunnel closures depend on the support sti$$ness. 0t is di$$icult to estimate the support sti$$ness in the present case% since the sti$$ness o$ back-$ill has to be taken into consideration while estimatin( the o#erall sti$$ness o$ a steel rib support system. There$ore% tunnel closure has been used to replace the sti$$ness o$ a support system (Table /". Table 1 3etails o$ tunnel sections under squeein( (round conditions ($rom Table A*" S.1o. Tye o$ rock , Tunnel Tunnel Support Radial 2bser#a mass radius depth pressure tunnel -tion (k(.sq.cm" (m" (m" ?redict <easu closur period -ed -red e (C" (months" 1 crushed red +.+*@ 1.@ *7+ /./ /.1 *.7 @ shales to +.1+ * -do-do!.@ F7+ !.* 1+.7 1.* 7 / So$t and plastic +.+1F 1.@ *7+ !.! /.* !.@ *F black clays to with in thrust +.+/ 'one ! -do-do!.@ -do!.! 11.* 1. *F

0n$luence o$ 2#erburden on Roo$ Support ?ressure Barton et al. (19 @" su((ested the $ollowin( correlation $or support pressures;

?ir (!" ?iw (@" where ?ir ?ir 6r ,ir ,iw

E E

*,ir-1./.6r *, iw -1./.6r

E short-term roo$ support pressure% E short-term wall support pressure% E BartonGs Aoint rou(hness coe$$icient% E short term roo$ rock mass quality% E short-term wall rock mass quality.

The #alues o$ ,ir and ,iw ha#e been taken as @ times ,r and ,w% where ,r% and ,w are BartonGs rock mass quality $or roo$ and wall rock% respecti#ely (#alues o$ , r% and ,w should be obtained separately $or the roo$ and the wall rock% respecti#ely". The short term roo$ and wall support pressures were estimated $rom 5qs. ! and @. These #alues were used to calculate correction $actor H$G $or o#erburden or tunnel depth. The correction $actor H$G is de$ined as a ratio o$ measured support pressure to the predicted support pressure. A relationship o$ $ to tunnel depth is shown in )i(ure /. Because the elasto-plastic theory su((ests a linear relationship between the o#erburden pressure and the support pressure% a linear relationship has been attempted in )i(ure /. Accordin( to )i(ure /% the correction $actor H$ H can be (i#en by $ (F" E 1 I (= - /*+"17++ 1

where = is the thickness o$ o#erburden or tunnel depth in metres. The data points $or squee'in( (round appear to su((est that the line in )i(ure / should be much steeper to represent a natural trend. 0n reality% the di$$erence between obser#ed support pressures and proposed line is mainly the result o$ e&cessi#e tunnel closures% which ha#e been taken into account by another $actor H$ 8 $or squee'in( (round condition. Some may doubt that the correlation proposed in 5q. ! can account $or the method o$ construction% the type o$ supports% the primiti#e stresses and tunnel closures. The instrumented tunnels were constructed by con#entional means% i.e.% drillin( and blastin( $ollowed by steel ribs. This practice resulted in si(ni$icant dama(e to the rock mass. There$ore% equation ! is on the sa$e side. 0n the case o$ machine tunnellin(% desi(ners should reduce the support pressures obtained $rom 5q. ! by perhaps *+C% as there will be reduced dama(e to the rock mass. Another #alid concern is that the $ield data are not su$$icient to pro#e the #alidity o$ the proposed correlation. 0n the opinion o$ the authors% the 0nternational Tunnellin(

Association should compile a data bank $or obser#ed support pressures $rom all parts o$ the world and should try to impro#e these correlation.

Ratio o$ >all Support ?ressure to Roo$ Support ?ressure Barton et al. (19 @" realised that the wall support pressure would be smaller than the roo$ support pressure and there$ore su((ested increasin( the obser#ed , #alues $or estimatin( the wall support pressure% as shown in Table *. The ratio o$ the wall support pressure ?w to the roo$ support pressure ?r% correspondin( to ,i-wall% ha#e also been shown in column / o$ Table *.The obser#ed wall suport pressures $rom some o$ the squee'in( and non squee'in( case histories ha#e been plotted in )i(. !. 0t can be seen that the recommendations o$ Barton et al. (19 @". )i( /; Correction $actor $or o#erburden in Barton8s correlation $or short-term roo$ support pressure under non-squee'in( (round conditions. )i(.!; Jariation o$ ratio between wall support pressure and roo$ support pressure with short term rock mass quality Table*. >all $actor ,i-wall . ,i % $or estimatin( wall support pressure. Recommendation o$ Barton et al.% 19 @ Ran(e o$ ,i 1 K+.1 +.1-1+ D1+ ,i-wall . ,i * 1.+ *.@ @ ?w . ?r / 1.+ +. +.F AuthorGs Recommendation Ran(e o$ ,i ! K+.1 +.1-@ D@ ,i-wall . ,i @ 1.+ *.@ D1@ ?w . ?r F 1.+ +. +.+-+.!

Correlation between Support ?ressure and Tunnel Closure in Squee'in( 9round Condition Jariation o$ the normalised roo$ support pressure with the tunnel closure at the crown is shown in )i(. @. The ordinate represents $ rH% which is the correction $actor which is the correction $actor $or tunnel closure at the crown. The correction $actor $ rH is (i#en as; $rH where ?robsd ?irobsd E ?robsd . ?ir . $ E measured roo$ support pressure% E predicted short term roo$ support pressure% and ( "

E correction $actor $or o#erburden (5q. F".

The data points in )i(. @ are taken $rom Table A* and represent ei(ht tunnel sections $rom $our di$$erent tunnels. The normalised roo$ support pressure are hi(her $or low roo$ tunnel closures. The roo$ support pressures decrease when tunnel closures increase and attain minimum #alues when the roo$ closures are appro&imately @C. Such a #ariation is in con$ormity with the (round reaction cur#e concept. )i(.@; Correction $actor $or roo$ closure under squee'in( (round condition (= >/@+,1./". )i(.F; Correction $actor $or wall closure under squee'in( (round condition (= >/@+,1./". This trend is repeated in )i(.F% which shows the #ariation o$ the normalised wall support pressure with the measured tunnel wall closures. The correction $actor $ wH $or tunnel wall closure is (i#en as; $wH where ?wobsd ?iwobsd $ E ?wobsd . ?iw E measured wall support pressure% E predicted short term wall support pressure% and E correction $actor $or o#erburden. (7"

Thus% the correction $actors H$rH and H$wH are the same as the normalised roo$ and wall support pressure% respecti#ely. The recommended #alues o$ these correction $actors are (i#en in Table /. The #alidity o$ Table / $or squee'in( (round has been questioned% particularly with re(ard to hi(hly squee'in( or $lowin( (round. 0t is su((ested that the application o$ Table / should be restricted to moderately squee'in( (round by limitin( closure to @C% by stren(thenin( the support system immediately. Serious construction problems may arise i$ this remedial measure is not $ollowed. 0t is recommended that all such tunnel sections be instrumented. 6ethwa (197!" concluded that the wall support pressure may be si(ni$icantly hi(her than the roo$ support pressure in the case o$ parallel tunnels i$ the clear spacin( is less than the sum o$ the tunnel widths. Jariation o$ Support ?ressure with Time A$ter studyin( the in$luence o$ the o#erburden and the tunnel closures and incorporatin( these in$luences in the correlation between , and support pressure% it has been possible to study the e$$ect o$ time on the support pressure. )i(ure shows the #ariation o$ the correction $actors H$G o#er time. )i(. ; Jariation in obser#ed support o#er time

The correction $actors $L $or time are (i#en as; $rM (9" $wM (1+" where $rM pressure% $wM pressure% ?robsd ?wobsd $ $GG E correction $actor $or the in$luence $or the time on the roo$ support E correction $actor $or the in$luence $or the time on the wall support E measured roo$ support pressure% E measured wall support pressure% E corection $actor $or o#erburden (5q. F"% and E correction $actor $or tunnel closures (Table /". E ?wobsd . ?iw . $G. $GG E ?robsd . ?ir . $G. $GG

Table /. Correction $actors $or tunnel closures in squee'in( (round conditions. S.1o. 9round Condition 1 * / ! @ F 1on-squee'in( (=K/@+,1./" Squee'in( (=K/@+,1./" <oderately squee'in( -do=i(hly squee'in( -doSupport system --------Jery sti$$ Sti$$ )le&ible Jery $le&ible 5&tremely $le&ible Tunnel Closure(C" K1 1-* *-! !-F F-7 D7 $Gw or $Gr 1.+ D1.7+ +.7@ +. + 1.1@ 1.7+

All the data points $rom Table A1 and A*% e&cept those $or the wall support pressure in the non squee'in( (round conditions (bein( ne(li(ible"% ha#e been plotted in )i(. . Accordin( to )i(. % the correction $actor $M can be (i#en by

$M

lo( 9.@ t+.*@

(11" where HtG is the time in months a$ter e&ca#ation o$ the tunnel linin(. Combinin( 5qs. 9%1+ and 11% the lon( term roo$ and wall support pressures can be (i#en as ?r (1*" ?w E ?iw.$.$G. lo( 9.@ t+.*@ (1/" where% ?r and ?w are lon( term roo$ and wall support pressures. Barton et al. (19 @" su((ested that the ratio o$ the ultimate to the short term support pressure is about 1. . 5quation 11% howe#er% su((ests the $ollowin( relationships; )i(.7; Comparison o$ obser#ed roo$ support pressure with predicted #alues $rom authorsG 5q. 1* (?r E ?ir.$.$G. lo( 9.@ t +.*@" ?r (1*" ?w (1/" where ?1 . ?+ are lon( term roo$ and wall support pressures. Barton et al. (19 @" su((ested that the ratio o$ the ultimate to the short term support pressure is about (@" 1./% i.e. 1. . 5quation 11% howe#er% su((ests the $ollowin( relationship; ? i.e. ?1 . ?+ E $H(lo( 9.@ t +.*@ $1H(lo( 9.@ t1 +.*@".( lo( 9.@ t+ +.*@" $+H E 1. @ E ?iw .$. $Hlo( 9.@ t +.*@ E ?ir .$ . $Hlo( 9.@ t +.*@ E ?ir.$.$G. lo( 9.@ t +.*@

where ?1 and ?+ are support pressures a$ter t1 and t+ month o$ e&ca#ation. 0n case o$ ri(id linin(% t1 and t+ so that the ratio o$ the ultimate support pressure a$ter 1++ years to that a$ter one month is (i#en by (t + E 1 month and t1 E1++ years 1**+ months". ?1 . ?+ E (lo( 9.@ t1 +.*@".( lo( 9.@ t+ +.*@" E 1. @

0n other words% the support pressure will increase in 1++ years to 1. @ times the support pressure obser#ed a$ter 1 month o$ e&ca#ation. The corrected support pressures compare well with the obser#ed #alues% as shown in )i(ure 7. This ratio o$ 1. between the ultimate and the short-term support pressure tallies with the (@"8 su((ested by Barton et al. (19 @". =owe#er% the ultimate support pressure $or

tunnels under squee'in( rock conditions may be *-/ times the short-term support pressure% accordin( to 6ethwa (197!". The ratio o$ the ultimate support pressure to the short-term support pressure% worked out here as 1. % is relati#ely small compared to the ratio o$ */ a$ter 6ethwa (197!"% probably because the period o$ obser#ations reported herein is relati#ely short and the number o$ squee'in( case histories is small. )urthermore% in special cases o$ soluble or erodible Aoint $illin(s and where seepa(e is a serious problem% the lon(-term support pressure may be as hi(h as the co#er pressure or F times the short-term support pressures% whiche#er is smaller. This trend has been indicated $rom a 1+-year per$ormance study o$ Chhibro-Nhodri under(round powerhouse comple& in 0ndia (<itra 1991". )or desi(nin( a temporary support system% one may assume a uni$orm distribution o$ the short-term support pressure% but a $actor o$ sa$ety o$ 1. @. =owe#er% the permanent support system should be desi(ned $or the net ultimate support pressures% with a $actor o$ sa$ety o$ 1.@. 5$$ect o$ 5&ca#ation Si'e on Support ?ressure Accordin( to Ter'a(hi (19!F" and 4nal (197/"% the support pressure is directly proportional to the si'e o$ a tunnel openin(. 2n the other hand% Barton et al. (19 !"% su((est that the support pressure is independent o$ the tunnel si'e. Columns / o$ Tables Al and A* Appendi&" hat support pressures obtained $rom Ter'a(hi8s method $or non squee'in( as well as squee'in( (round conditions. 0t may he noted that the estimated support pressure #alues (shown in column /" do not compare well with the obser#ed support pressures (in columns 1* and 1/". The support pressures corrected accordin( to the proposed correlation (columns 1+ and 11" are in better a(reement with the obser#ed support pressures. )i(ure 9 shows the #ariation o$ ?robsd . ?r . $G. $GG with the diameter o$ tunnel openin( (*a". The ordinate represents the obser#ed roo$ support pressure corrected $or o#erburden% the tunnel closure and the time o$ e&ca#ation. 0t may be seen that the corrected support pressure is independent o$ the tunnel si'e. The reason $or support loadin( in non-squee'in( conditions may be related to the dead wei(ht o$ the loosened rock blocks which become detached $rom the parent rock mass at the tunnel roo$ and rest on the support system. This type o$ support pressure is called the 8loosenin( pressureL. The loosenin( pressure has been attributed to poor blastin( practice% (ra#ity and delayed support in the $orm o$ steel ribs. 5&cessi#e tunnel closures under squee'in( (round conditions are also considered responsible $or mobilisin( lar(e loosenin( pressures% e#en in tunnel sections supported by shotcrete immediately on e&ca#ation. The loosenin( pressure (dead wei(ht or the loosened or a de-stressed 'one" is there$ore mobilised due to poor blastin( practice and is likely to be independent o$ the tunnel si'e.

0n a recent study at the under(round powerhouse comple& o$ :akhwar dam proAect in 0ndia (, E 7-9 and =E *@+ m"% the obser#ed roo$ support pressures were nearly the same (i.e.% about +.@ k(.cm*" $or the Fm wide approach adit% 1!-m-wide e&pansion sur(e tank and *1-m-wide powerhouse ca#ern% all e&ca#ated throu(h ti(htly Aointed traps. These obser#ations should erase all doubts about si'e e$$ect in under(round openin(s. 0t may be noted that rock mass quality , estimated $rom a lar(er tunnel would be smaller than that obtained $rom small dri$ts in a similar rock mass. This is due to the possibility o$ intersectin( (reater number o$ (eolo(ical discontinuities in a lar(er openin(. Thus% the si'e e$$ect is automatically accounted $or in the estimate o$ ,. The ad#erse e$$ects o$ deterioratin( hydro-(eolo(ical conditions (6 w" should also be determined% i$ possible% a$ter a water tunnel is commissioned. 0t would there$ore be unsa$e to obtain , $rom small dri$ts and use it to estimate support requirements $or lar(e e&ca#ations. )or under(round e&ca#ations in non dialatant rock masses (schists% slate% etc." with smooth planes o$ weakness% it is cautioned that Ter'a(hi8s concept may still be #alid. Conclusions The combined approach o$ $ield instrumentation and quantitati#e classi$ication o$ Barton et al. has pro#ed rewardin( at this sta(e o$ de#elopment o$ rock mechanics. 3espite limited $ield data% some practical trends showin( the in$luence o$ o#erburden% tunnel closures and time o$ e&ca#ation on the tunnellin( condition and the support pressures ha#e emer(ed. 0t would perhaps be hasty to draw any de$inite conclusions $rom these trendsB howe#er% some tentati#e correlation ha#e been possible. These correlation are subAect to re$inement as more $ield data is collected. The $ollowin( tentati#e conclusions are possible $rom these correlation; 1. Squee'in( is likely to occur in a tunnel section where the hei(ht o$ o#erburden in metres e&ceeds /@+ ,1./. *. The short-term roo$ support pressure is (i#en by the $ollowin( correlation; ?r E *.+(@,"-1./ .$. $G. 6r where H$ H is the correction $actor $or thickness o$ o#erburden (R" in metres% and $8 is the correction $actor $or tunnel closure (see Table /% equal to 1 in non squee'in( (round conditions". The-#alue o$ the correction $actor $ is (i#en as $ E 1 I (= - /*+"17++ D 1. )i(ure 9. Support pressure #irtually independent o$ tunnel si'e.

/. 0n squee'in( (round conditions% the support pressure is si(ni$icantly in$luenced by tunnel closures. The correction $actor $G $or tunnel closure #aries $rom +. to 1.7 in the case o$ a sin(le tunnel. The minimum support pressure occurs when the tunnel closure is about @C o$ the tunnel diameter. The support pressure increases rapidly beyond this limitin( closure. !. The short-term wall support pressure may be obtained $rom the abo#e correlation by substitutin( ,wall% $or ,. 0n (eneral% the actual wall support pressure $or the nonsquee'in( rock conditions is likely to be ne(li(ible. The short-term #alues o$ ? iw%. ?ir depend on ?r (i.e.% @,"% as (i#en below; ?iw%. ?ir ,i

1.+ @, K +.1 1.+ - +.+ @ K @, K +.1 +.+ @, @ @. The ultimate support pressure maybe 1. @ times the short-term support pressure $or tunnel sections under non-squee'in( (round conditions% e&cept $or cases o$ soluble and erodible Aoint $illin(s with seepa(e. F. The support pressure is independent o$ the tunnel si'e% pro#ided that , is obtained $rom a $ull si'ed openin(.

Re$erences Barton% 1.B :ien% R.B and :unde% 6. 19 !. 5n(ineerin( Classi$ication o$ Rock <asses $or the 3esi(n o$ Tunnel Jolume % 1umber 1% 199*% SupportRock<echanicsJol.F% 179*/F. Sprin(er-Jerla(. Barton% 1.B :ien% R.B and :unde% 6. 19 @. 5stimation o$ Support Requirements $or 4nder(round 5&ca#ations% ?roc. Si&teenth Symp. on Rock <echanics% 4ni#. o$ <innesota% <inneapolis% 4.S.A% 1F/-1 . Bieniawski% O. T. 1971. Case Studies- ?rediction o$ rock <asses Beha#iour by the 9eomechanical Classi$ication% Second Australia-1ew Oealand Con$erence on 9eomechanics% Brisbane% /F-!1. 3aemen% 6. 6. P 19 @. Tunnel support loadin( caused by rock $ailure. ?h.3. Thesis% 4ni#ersity o$ <innesota% 4.S-A.

3eere% 3. 4.B ?eck% R. B.B <onsees% 6. 5.B and Schmidt% B. 19F9. 3esi(n o$ Tunnel :iners and Support System. =i(hway Research Record 1o. //9% 4.S. 3epartment o$ Transportation% >ashin(ton% 3.C. 3ube% A. N. 19 9. 9eomechanical e#aluation o$ a tunnel stability under )ailin( rock conditions in a =imalayan Tunnel. ?h.3. Thesis% 4ni#. o$ Roorkee% 0ndia. 3ube% A. NB 6ethwa% 6. :.B Sin(h% B.B Sin(h% BhawaniB and <ithal% R. S. 197*. 9eoen(ineerin( e#aluation o$ problems o$ a lar(e under(round ca#ity $or Tehri 3am ?roAect (0ndia". 0SR< Symp. Rock <echanics; Ca#erns and ?ressure Sha$ts (ed. >. >ittke"% */9-*!!. Rotterdam; A-A. Balkema. 3ube% A. N.B Sin(h% B.B and Sin(h% Bhawani. 197F. Study o$ squee'in( pressure phenomenon in a tunnel-0 and 00. Tunnellin( and 4nder(round Space Technolo(y 1;/@-!7. 6ethwa% 6. :.B 3ube% A. N.B Sin(h% B.B Sin(h% BhawaniB and Jiladkar% <. 1. 19 9. 0nstrumentation and 3esi(n $or multiple openin(s in $ailin( rock-mass. 0nt. Symp. on 0n-Situ Testin( o$ Soils and Rocks and ?er$ormance o$ 4nder(round Structures% Roorkee% 0ndia% 3ec. 19-**%19 9% Jo: 1. 6ethwa% 6. :.B 3ube% A. NB Sin(h% B.B and Sin(h% Bhawani. 1971. Rock load estimation $or tunnels in squee'in( (round conditions. ?roc. o$ the Rapid 5&ca#ation Tunnellin( Con$erence% San )rancisco% Cali$% <ay /- % 1971% FF 7/. 1ew Pork; A0<5. 6ethwa% 6. :.B 3ube% A. N.B Sin(h% B.B Sin(h% BhawaniB and <ithal% R. S. 197*. 5#aluation o$ classi$ication system $or tunnels in non-squee'in( (round conditions% ?roc. o$ 0SR< Symp. Rock <echanics; Ca#erns and ?ressure Sha$ts% ed. >. >ittke% F+ -F1*. Rotterdam; A.A. Balkema. 6ethwa% 6. :.B Sin(h% B.B and Sin(h% Bhawani. 197!. 5stimation o$ ultimate rock pressure $or tunnel linin(s under squee'in( rock Conditions-a new approach. ?roc. 0SR< Symposium on 3esi(n and ?er$ormance o$ 4nder (round e&ca#ations% Cambrid(e% 4.N.% Sept. /-!% 197!. <tra%S.1991.Study o$ lon(-term beha#iour o$ under(round powerhouse ca#ities in so$t rocks. ?h.3. Thesis% 4ni#ersity o$ Roorkee (under submission". ?rotodyakono#% 1. <. 19F/. )irmness coe$$icient $or estimation o$ rock loads. ?ersonal communication to Beas 3esi(n 2r(anisation 1ew 3elhi% 0130A. Sharma% J. <. 197@. ?rediction o$ closure and rock loads $or tunnels in squee'in( (round%. ?h.3. Thesis (p. *@!"% 0.0.T.% 3elhi% 0ndia. Ter'a(hi% N 19!F. Rock de$ects and load on tunnel supports. 0n introduction to Rock Tunnellin( with Steel Support% R. J. ?roctor and T. C. >hite (Poun(sta#a% 2hio% 4.SA; Commercial Shearin( and Stampin( Co.".

4nal% 5. 197/. 3esi(n (uidelines and roo$ control standards $or coal mines roo$s. ?h.3. Thesis% ?ennsyl#ania State 4ni#ersity. Re$er to p. 11/ o$ Rock <echanics in <inin( and Tunnellin( (Bieniawski% O.T.% 197!% Rotterdam; A.A. Balkema". >ickham% 9. 5.B Tiedmann% =. R.B and Skinner% 5. =. 19 !. 9round support prediction model-RSR concept. ?roc. o$ 1orth American Rapid 5&ca#ation and Tunnellin( Con$erence% San )@.ancisco% Cali$ornia% Jol. 1% F91- +7. Table A1. Comparison o$ predicted and obser#ed support pressure $rom ,-system in non-squee'in( (round Table A*. Comparison o$ predicted and obser#ed support pressure in squee'in( (round conditions.

)actors $or;

Jertical Short-term Support Support ?ressure; ?ressure 9eolo$rom (ical Ter'a(hi8s 2bser3escrip- Classi$i2bser#ed Support S. 1o. tion cation Tunnel Closure Jert. Remarks wall ?% ?.

Corrected Short-term Support ?ressure Shortterm Rock <ass #ation ,uality =ori'. ?< ?ressure Correction

2#erburden ?eriod roo$

oi p

$ p ob t ?.Q ?>:000 > $rom $rom 5q.@ $rom 5q. F

$rl

f rom Table !

$ rom Table !

?rR $.

?.0B @, 5q.!

k(.sq. cm k(.sq. k(.sq.k(.sq. cm cm cm

k(.sq. k(.sqcm cm

k(.sq. month cm

1 9

2 1

3 11

! 12

5 13

F 1!

7 15

C"i#$o%K"o&$i Tunnel 1%7 to /.! +.1 *F to *.@ to !. * (*.7" (!./" /%1 1. (*. " +.@+ statae (0.25) a R,3 E1+*+ , +.+*@ +%1+ *. / uh E *.7 = E *7+ (6ethwa et aR. 197*" 1.7 to /.+ 1.+ (+.7F" @ Circular $ibs (/./" 1.7 (*.!"

ab!e A2 ("ontd.). * 1* / 1/ ! 1! @ 1@ 1+./ (1 1" to +.*@ (+.1*" squee'in(. :itde , E +.+1*-+.@ closure at crown. uhla E F% = E F7+ (6ethwa et a". 197*" /. +. compressmoderately squee'in(. aBP@% E *.F! 5nlar(ement o$ R23 E 1+ dri$t to 9 m si'e in + E +.+1 9A.+/ dose pro&imity uhla E !.1 delayed stabiu#la E !.@ lisation. = E *7+ (6ethwa etB>. 197*" (*.!" +.1@ (+.11" behind ribs. (!.!"(/.9" ible back$ill So$t plastic black1.@ to (/.1" (*. " /.* clays in thrust 'one% +.+7 *.F /./ !.+ to/.+ to 1.+ +. + *FCircular ribs were to !.7 !.7 stableB +.+F /./ to *.! to (/.7" 1+.7 /II 7 @.+ ribsB se#ere (!.*" (/. " buddin( due @.+ F 7 9 1+ 11

Chibro-Nhodri Tunnel (contd." *. Crushed red shale% 1.!@ 1.7 +. =ea#y circular hi(hly squee'in(to a E !8A% to R23 E 1+-*+ (1F.*" E *. /**.1

!. 1.7

So$t plastic blackdo ( .9" ( .+" 11.@ days within thrust 'one% moderately

do 1*.*

do do 1%+ 1.7 *FCircular ribs o$ #ery hi(h capacity were stable.

squee'in(. Consequently% y E *.F!% tunnel closures R,3 E 1+ were likely to be + E +.+1F-+.+/ low% appro&. =E *7+ . +-*C. (6ethwa et al. 197*" 9irl @. 1.1@ =ydro Tunnel Jery blocky and+. to (1.9" (1.*" *.+ seamy slates% *./ moderately closure as squee'in(. horseshoe ribs aE1.1% E*.@ with in#ert struts uhla E .F de$ormed% but not = E /7+ as se#ere as in , - +./*-+.7* case F. (6ethwa et al. 197*" (1.*" 1.@ to 1.* to+.7 to 1.2B 1.1@ *.! 1*Roo$ closure !.1 1.7 1./ considered equal (*.@" (1 R@" (1.+" to wall

ab!e A2 ("ontd.). 1 9 1+ * 11 / 1* ! 1/ 1! F is 1.7 to 1. 1.* to !.+

9irl =ydro Tunnel (contd." F. Crushed phyli$tes% *.1 to +.F* to 1.+ +. 1.7 (1.@" (*.9" * ?eak hi(hly squee'in(. #.$ (%.$) 1.9 (1.F" measured a E *@1% m *./ support R23 E 1+-*@ pressure o$ @ q E +. 1 *!S+./* k(.sq. cm uhla E 1*.! occurred at u#la E @ hal$ total wall = E *!+ closures when (6ethwa et ad 197*" horseshoe ribs with in#ert budded. :oictak =ydro Tunnel

1.F (1" *./ (*.1"

. Crushed shales% *.9 to +.+@@ to /.@ to *.@ to 1.+ 1.1@ 1.1@ (@.1" (!.F" @.!II @.!II / 1@-cm-thick moderately @.! (!.*" +.** @./ (!.!" @./ (!.+" shotcrete with squee'in(. !-rn-lon( rock a L *T% E *. bolts supple(Rcc mented with R23 m 1+-*+ circular ribs. , E +.+1 1-+.+!! Squee'in( uhla E occurred e#en = E /++ at = R 1F+ m. (6ethwa et al 197*" Roo$ closure is considered equal to wall closures. <aneri Bhati ?rosect 7. 1.1@ =i(hly $ractured *.F! 1.1@ (1.9" quart'ites. (*.@" (1./" 1.F *.+ 1.1 1.+! 1!Supports buckled. (+.1 1"

a E **!% E *.@ Jertcal and R23 E F+% + - +.@ hori'ontal uh E 19+ mm closures = E /@+ appeared (Sharma 197@" equal. See $ootnotes in Table Al $or notations.