aos EE NCES Prediction of squeezing potential of rocks in tunnelling through a combination of an analytical method and rock mass classifications Omer Aydan*, Siileyman Dalgic Summary *°T, Kawamoto™* “This paper is concerned with the prediction of squeezing potential of rocks in tunnelling and of the deformation of tunnels in such rocks. ‘The authors present a procedure how (0 combine an analytical method proposed earlier by the au- thors (OA and TK) for predicting the deformation of tunnels in squee: proposed procedure is applied to Bolu tunnels (Ibrkey), which are currently under con pared with actual measurements and observations. L. Introduction ‘The deformational behaviour of tunnels, which underwent large deformations, so-called squec?- ing, have been recently receiving great attention in the field of rock mechanics and tunnelling. The first author proposed an analytical method to pre- dict squeezing potential of rocks in tunnelling, ‘This method is based on the utilisation of classic wave velocity of rock mass to estimate its uniaxial strength and related mechanical properties. Al- though the elastic wave velocity method is com- monly used in Japan, itis very rare to see the use of this method in other countries. Instead, rock Classifications are usually used in rock anass charac- terization. In this paper, the authors briefly present a procedure for combining the analytical method proposed by Avnan et al. [1993] with rock mass classifications. ‘Then an application of the proposed method to predict the squeezing poten- als of rocks along the alignement of Bolu Tunnels in Turkey, which are currently under construction, is given, and predictions are compared with obser- vations, 2. Assessing mechanical properties of squeezing rocks 2.1. Rock Mass Uniavial Strength Although there are several rock classificatio used in many countries, it seems that RMR and Q- "Tokai Universiy, Shimizu, Japan + Istanbul Univesity, tstanbul, Turkey + leh Institute of Technology, Toyota, Japan. ng rocks with rock mass classficagons, Then the suction, and predictions are eom- systems are the most widely known rock classifica tions [Bieniawski 1974; Barton ef al. 1974]. Hone and Brown [1980] tried to establish a relation be- ‘tween their empirical yield criterion and RMR. For uniaxial case, rock mass strength is related ts iva rock strength by Ww Ta Using the experimental results on granulated Panguna Andesite, they suggested the following for mala: 5 = ¢{R4R -100¥8 oy where B is a constant. Hoek and Brown used the fol: lowing relation between RMR and Q-value, propo- sed by Breniawstt [1978] to compute relation betwe- ens and Q-value. RMR = 9logQ + 44 @) Recently Barton [1995] suggested the following, relation between RMR and Q-Value RMR = 15 logQ + 50 “@ As discussed by Avpaw et al, [1997], the estima tions front these methods are generally conservative when one uses the values for constant B suggested by Hoxk and Brown. For example, when the values of RMR are less than 40 and Qovalues are less than 1, rock mass strength becomes very small. Since the value of RMR and Q-system for squeezing racks are sgenerally below these threshold values, the strength of squeezing rocks will be usually underestimated. Probably for these reasons, Sincu [1993] proposed the following direct relation between Quvalue and rock mass strength Ge 790" co) RIVISTA ITALIANA DI GEOTECNICA 1/2000 3eQDCeS2APSONM SCALARS RENO Si HNTB ALICDLAER LE ES Banton [1995] suggested the unit weight of rocks 26 kN/m* for engineering applications of the relation above. Thus the above equation is re-writ- ten in the following form Gam = 18.208 © The unit of uniaxial strength is MPa. Arpawet al. [1997] also recently proposed tke following relation between RMR and rock mass strength wsing in-situ test data on rock mass strength in Japan. Gem = 0.0016 RMR? oy “The unit of uniaxial strength is MPa. It should be also noted that this relation is derived for Basie RMR values. The relations given above are com- pared with the data of in-situ tests gathered from several constivction sites in Japan in Fig. 1 When rock mass is weak and jointing is negligi- ble, direct relations given above niay overestimate the strength of rock masses. Under such cireum- stances, it is recommended (0 use reduction coetti- cients related to rock classification values. Besides Ho#k-Brown’s proposal, Katawakss and BreNtawskt [1995] also suggested the following formula be- tween the uniaxial strength of intact rock and that of rock mass % ®) Reanalysing the data shovn is figure 1, the au- thors will propose the following retation for strength reduction coefficient = (Rue a Ga! (ae eu 9) ‘The relation given above is compared with that of Hoek-Baowns as well as with in-situ data in Fig. 2 2.2, Other related mechanical properties From an extensive survey of data, AvoaN ¢ al. 11993} developed a data-base system, called SQR- PROP and plotted various physical and mechanical properties of rocks as a function of the uniaxial strength of rocks and established relationships be- tween these properties and the uniaxial strength of squeezing rocks. The relationships between physical and mechanical properties and uniaxial strength of| squeezing rocks are summarised as follow: i.) Physical Properties: unit weight /kN/m’) porosity ‘m(9), elastic wave velocity Vp(kms) 01 + 0.8025), 60010", ao) Vp = 14 + 0,200? ft i eos AAPA ELLE SEGEMOER IE RIVISTA ITALIANA DI GEOTECNICA se Avan - Datorc - Kawamoto ne ,) Elastic constants: elastic modulus £ (MPa), Pois- son's ratiow B02", y=a25lte%) (11) iii-) Plastic constants: peak friction angle @(°), te sidual uniaxial strength op! (MPa), residual fri tion angle @(), peak plastic Poisson's ratio fy softening plastic Poisson's ratio f,, residual plas- tic Poisson's ratio fe Q-VALUE 10-2 10°" 102, am = 1590+ aon 188), + Sh 9 0 SK ‘Auxginn o-s8r04 rahe ts ROCK MASS RATING (RMA) Fig. 1 ~ Relation bewcen rock classifications and rock mass strength. Fig. 1 Relasione tra le clasificazion deltammasso roecosa ¢ la sua ressensa a vompressione monoassial. O-VALUE of ag? to! 10? 10! 10% 1 Fra 151290 +50 (cn 185) ne 9888) ‘Hos nett gry et vate) MASS STRENGTHINTACT STRENGTH ¢../0. MASS STRENGTHINTACT STRENGTH ¢ J/¢. oo 20 40~~«6)SO ROCK MASS RATING (RMA) Fig. 2~ Relation between rock classifications and strength eduction coefficient. Fig. 2 Relazione ta le clasificasion’ del'ammasso voccaso ed il coefficient di ridusione della resistenza a compressions mo snoassiale della rocia ita.PREDICTION OF SQUEEZING POTENTIAL OF ROCKS IN TUNNELLING THROUGH A COMBINATION. gpntoas®, Langton 3-036 — f,=0.5 + 0.25098, (12) f= 1040250, f, =0.5 4025004, It should be noted the unit of uniaxial strength is MPa in the above relations 3. An analytical method for predicting squeezing potential of rocks An analytical solution as been already developed by Avpan ¢ al [1993] for predicting strain and stress fields and the level of squeezing of eircular tunneks squeezing rocks under a hydrostatic state of stress. ‘Therefore, the final expressions for tangential strains for various states are only given herein: 1.) Elastic state Fangential stain and stress at tunnel wall can be obiained as: TE m-m) 08 -H, (13) If the tunnel is strained to its elastic limit, then 68 = o¢, with p= 0. Thus, we have the elastic strain limit as: ef = Lee & a4) Using the above relation in Eq, (18), one obtains the normalised wnnel wall strain as: Hot Beige hig % as & @ p p 2.) Perfectly plastic state ‘Tangential strain at cannel wall ean be obtained. jet » R, 8 = EXE, 04) Be 6) clastic limit is given as 7 (y- oy) an Using the above relation in Eq, (16), one obszizs the normalised aunnel wall strait as: pe erie a Da ye TH L GHD pra ] (18) GENNAIO- MARZO 2000 LPR AGRARIAN LS LAOS OPE AIRTIME 3-) Residual plastic state ‘Tangential strain at tunnel wall can be obtained as fat ¥ R, b= 0 Op )ny ag eee Using qs. (17) & (19), one obtains the normal ised tunnel wall strain as: ail (qQ-Dta | a al Gr) pte E=mny| (20) where cl oe a Py" 4, Applications to Bolu tunnels ‘Turkish Roadway Authority (Tirkiye Karayol- Jari Genel Miidirliigi) is now constructing the Ana- (olian Express Motorway which extends from the European borders of Tirkiye to the Asian borders. ‘This express motorway passes through the famous fault zones such the North Anatolian Fault Zone, East Anatolian Fault Zone, which are tectonically very active. ‘The Bolu tunnels have a cross-section of about 170-200 m#, which is approximately the half of typ- ical cross sections of underground power houses, and are 3250 m long; still under construction, the ‘excepted date of completion is year 2000. The Bolu tunnels pass through the North Ana- (olian Fault Zone (NAFZ) and the rock masses con- sist of various sedimentary rocks as well as igneous rocks (Fig. 3), Faulting along the tunnels is of strike- slip or thrust types. The excavation through igneous rocks was not problematic. Once the tunnels have entered the sedimentary rocks which are heavily Jaulted and folded, very excessive deformation of the tunnels was encountered. In some cross-sections deformation was more than 1500 mm. ‘The second author (SD) has been involved with these tunnels since preliminary site investigations, and has experienced the difficulty of assessing the characteristics of faulted sedimentary rock masses from outcrop surveying in heavily forested areas and vertical borings [Dancig, 1994; Davcig and Gex0801, 1996] In this article, che authors attempt to predict the squeezing potential of ocks and the measured defor- ration responses of the Asarsuyu section of the left SAR 43Nasa vy Fig. 3 Longitudinal geology of Bolu Tunnels Fig. 3~ Proflo geologio delle gallevie Bolu. tube of Bolu Tunnels, the excavation of which is al- ‘most completed. Rock masses have been assessed by the second author (SD) using both Bieniawski’s RMR and NG's Q-System and their strength has been es- timated from the formula of AvoaN et al. [1997] for RMR system and sinoit [1998] for Q-system. Fig. 4 shows the variation of overburden and of rock mass strength estimated feom each formula. Surprisingly, the strength estimated from both formula are very similar to each other. After estimating the strength of rock masses, other mechanical properties needed in computations are determined from the empirical for- mulas given in Subsection 2.2 and the deformation behaviour of the tunnels is predicted by using the method of AvoaN ef al. [1993]. Fig. 5 compares the prediction for Asarsuyu section of the left tube of Bo- Iu Tunnels for unsupported case. As expected, the «deformations predicted for unsupported case should be larger than those measured since the actual tunnel ‘BOLU TUNNEL - ASARSUYU SECTION ieee aa gg & (OVERBURDEN in) ‘tho “tt “ir i DISTANCE FROM ASARSUYD PORTAL (m) UNIAKIAL STRENGTH oP) Fig. 4 ~ Variation of overburden and estimated uniaxial strength of rock mass along cannel alignement Fig. 4~ Varazione dela coperrae stima dela esistensa a com- essione monoassile dell amasso roccioo lunge la sviluppo dela galeria ATES Avon - Darcig - Kawanoro is supported. The predictions show that when squeezing takes place, itwill range from fair squee7- ing (FS) to very heavy squeering (VHS). These pre dictions are consistent with both observations andl de- formation measurements. Although there exist some differences between predictions and observations, these may be solved through more detailed rock clas- sifications along the tannel alignement and the con- sideration of actual support system. 5. Conclusions In the present study, the authors combined ana- Iytical method proposed by Avpaw et al. (1993) and rock mass classifications. ‘The rock masses were as- Bs a] Erol weer 7 3 ol S 400) 8 soo ao 2 sof Fig. 5 - Variation of squeezing degree and comparison of predicted deformation of tunnel with measurements along. funnel alignemen. Fig. 5~ Variazione del ivello di comportamento spingente pre- visione dele deformazioni della galleria poste a confront con le smisure Lengo Uasse longitudinale ALTA LE PTTL OLEATE RIVISTA ITALIANA DI GEOTECNICAPREDICTION OF SQUEEZING POTENTIAL Of ROCKS IN TUNNELLING THROUGH A COMBINATION... I sessed using both Bieniawski's RMR and NGI's Q- System and their strength was estimated from the formulas given by AvpaN etal. [1997] for RMR sys- tem and Sino [1993] for Q-system. Once the strength of rock masses is determined, other me- chanical properties for computations are deter- mined from the empirical formula of Avoax et al 11993), and the behaviour of the tunnels is predict- ed by the method of Avnax etal, [1993]. The predic- tion for Asarsuyt section of the Bolu Tunnels are compared with observations. ‘The applicability and validity of the proposed procedure have been checked by comparing the predictions with actual observations, It is found that the predictions by the proposed procedure well agree with observations and it can be a usefull tool for engineers to predict not only the squeezing potential of rocks but also the magnitude of deformation of tunnels in squeez- ng rocks. 6, Acknowledgements ‘The authors thank Turkiye, Karayollari Genel Miidiirliigit (Tiirkish Roadway Authority) for infor ‘mation and the data related to Bolu tunnels. The second author (SD) thanks particularly Astaldi SPA, the constructor of Giimlisova-Gerede motorway for allowing to access the data on deformation measure- ments. References Avpan ©., Anat T., Kawaworo T. (1993) ~ The squeezing potential of rocks around tunnels; Theory ‘and prediction. Rock Mech. Rock Engng, 26, 2, pp. 137-163, Avpan O., ULusay R., Kawaworo T. (1997) — Asses tment of rock mass strength for underground excava- tions, 36th US Rock Mechanics Symposium, pp. 177-786. Baton N., Lit R., Lunve I. (1974) — Engineering classification of rock masses for the design of tunnel supports. Rock Mechanics, 6, 4, pp. 189-239. Baron N. (1995) ~ The influence of joint properties in ‘modelling. jonted rock masses. 8th Int. Rock Mech. Congress, Tokyo, 3, pp. 1023-1082. Brexwawskt ZT. (1974) ~ Geomechanics classification of rock masses and its application in tunnelling. Third GENNAIO- MARZO 2000 PEROERCHLLA AAD ER RRC Int. Congress on Rock Mechanics, ISRM, Denver, TIA, pp. 27-82. Brsaawskt Z.T, (1978) ~ Determining rock mass deform ability ~ experience from histories. Int. J. Rock Mech., Min. Sei., 15, pp. 237-247. Davcig S. (1994) ~ Engineering geology of the passage of the Anatolian motorway through the Bolu mountains. Ph.D. Thesis, University of Istanbul, Turkey, Dateic, §. and M. Gézitbol (1996) ~ Squeezing rocks ‘in Bolu tunnel. 3rd Nat. Rock Mech. Symp. of Tur~ key, TNSRM, Ankara, pp. 25-33. Hork E., Browns E:T. (1980) ~ Empirical strength cri- terion for rock masses. J. Geotech. Engng. Div., ASCE, 106, GT9, pp. 1013-1086. HOEK E., Buoww F-T. (1988) — The Hoek-Brown failure criterion ~ a 1988 update. 15th Can. Rock Mech: Symp., pp. 31-38 Hock E., Brown E:T. (1997) ~ Practical estimates of rock mass strength. V71,, J. Rock Mechs. and Min. Sci. 34, 8, pp. 1165-1186. Katananas G.S., Brentawskt 2:7 (1995) —A rock mass strength concept for coal seams incorporating the effect of time. 8th ISRM Congress, Tokyo, 1, pp. 295- 3002. SINGH B. (1993) - Indian case studics of squeezing grounds and experiences of application of Barton's Q- sgstem, Work-shop on Norwegian Method of Tun- nelling, CSMR,S, New Delhi. ‘Yorre E. and M. Varna (1983) ~ Rack Mechanics (in Turkish). Istanbul Technical University, Mining Faculty Pub, n. 4 Previsioni del potenziale comportamento spingente nello scavo di gallerie mediante metodi analitici e di classificazione dell’ammasso roccioso Sommario ‘La nota descrve la revision del potensiale camportamento spingente nell scavo di gallevee della conseguente deformazione indole. Gl tori presentana un procedimento che consite nel uso di un metodo analitico,propasto in recedenza dag autor: (OA TR), pe la prevision della daformazione di ‘gallere in condzion spingenti, congientaments ai meted di lassificazione degli ammassi soecias. I procedimento proposto viens applicato alle galleie Bolu (Ierkey),attualmente in castrucions, ele prevision’ formulate sono poste a canfvonta con i rinulat lle mise e delle sservazions PEELE LTTE ANON 45