You are on page 1of 9

Missouri University of Science and Technology

Scholars' Mine
International Conference on Case Histories in (2013) - Seventh International Conference on Case
Geotechnical Engineering Histories in Geotechnical Engineering

Apr 29th - May 4th

Amirkabir Natm Tunnel — A Case Study of
Design Challenges in a Mega Project of Tunnel in
Soft Ground
Aliakbar Golshani
Tarbiat Modares University, Iran

Ehsan Moradabadi
Tarbiat Modares University, Iran

Follow this and additional works at: http://scholarsmine.mst.edu/icchge
Part of the Geotechnical Engineering Commons

Recommended Citation
Golshani, Aliakbar and Moradabadi, Ehsan, "Amirkabir Natm Tunnel — A Case Study of Design Challenges in a Mega Project of
Tunnel in Soft Ground" (2013). International Conference on Case Histories in Geotechnical Engineering. 3.
http://scholarsmine.mst.edu/icchge/7icchge/session05/3

This Article - Conference proceedings is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in International
Conference on Case Histories in Geotechnical Engineering by an authorized administrator of Scholars' Mine. This work is protected by U. S. Copyright
Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder. For more information, please contact
scholarsmine@mst.edu.

5. The work w reportedd in this paperr specifically dealt withAmmirkabir highwway tunnel witth approximateely 1. improved thee classical deterrministic approoaches. P.e. P. 2 Meanwwhile. Howeverr. of restricting in sitee investigation data and uncerrtainties relatedd to theem. a probabilisstic hypotheticaal elasticity modulus (PHEM M) approach haas been employyed to evaluate the unncertainty in lining design off a horseshoe shhape NATM(N New Austrian Tunneling T Metthod) tunnel in T4 section of the project. Robust and opptimize design of tunnel suppport pattern connsists many im mportant parammeters includingg advannce rate and exxcavation meth hod.It is on ne of the important projects excavating e bellow one of thee highest trafficc region of Tehran. the prrobabilistic (oor the project(You et al. Thesee uncertaintiees arise from m limited geoological data. 2005). unncertainty exists seriously inn the process oof consttruction of underground struccture in soft soiils. erronneous evaluatiion in these parameters p cann propeerties. and a Monnte Carlo algorrithm (Latin Hyper H Cube Sam mpling methodd) to sim mulate the unceertainty existedd in the systemm on the other handh are applieed to produce the probabilisttic density funcction of surface settlement of tunnell excavation an nd lining phasees. the numerical results show that the PHEM M approoach introducedd has had an ap ppropriate preddiction in surfaace settlements.. Civil and Environmenttal Engineeringg Faculty. A MATLAB intterface program m for geenerating the ABAQUS-base A ed parametric model m in one haand.. Rob bust and optim mize design off reliabbility) approaach. These unceertainties arise from limited data d of geologiical data. measuurement errorrs. Key Words:probab W bilistic hypotheetical elasticity modulus approach. PHEM M. Simillar to other parrt of the projecct. Deformatioons and settlem ments have beenn monitored duuring and after construction inn order to avoiid unpredictablle deformationns and asa a result any possible p failuree. uncerrtainty seriouslly exists in the process of construction c off quanntitative definittion of these paarameters is diifficult because underrground structuures in soft soills. includingg the difficulty of excavating throu ugh a heterogeneous sedimenntary basin maainly composed of recent allluvial. erroneouus evaluation in these param meters can affeect in incorrecct projecction of tunnell stability or economic loss. interpolation of spatiallly geologicall Furthhermore. consisting two t essential parts p is used to evaluate the performance off the system. a extrapolation of results of experiments and natural analogue studies over times andd interppolation of spattially geological properties.A CAS SE STUDY Y OF DES SIGN CHA ALLENGES IN A MEG GA PROJECT OF TUNNEL T IN SOFT T GROUNDD Aliakkbar Golshanii Ehsan Moradabadi M Soil and a Foundationn Engineering Department. and condiitions relevant to the project. T unccertainty managgement. Undergrouund excavationn.O. Civil and Environm mental Engineerring Faculty. Box 14115-159 Tehraan. reliabiility analysis INTR RODUCTION In spiite of enhancem ment in modeliing techniquess as well as sitee tunneel supports patttern consists of o many importtant parameterrs investigation methoods (to predicct the settlem ments induced). Howeverr.O. Cut and Coover and Top/D Down methodds in different sections of the projecct. Construuction has beenn performed by different methods i.The hybrid model. and exxtrapolation of experimentaal results andd affecct in incorrect projection of tunnel stabilitty or economic naturaal analogue stuudies over times and conditioons relevant too loss(YYou et al. NATM Tunneling. incluuding advancedd rate and exxcavation methhod. Tarbiiat Modares Unniversity. Iran I ABST TRACT In spiite of enhanceement in modeling techniquees as well as siite investigatioon methods.nnorth and southhtubes). measurement errorss. Iran Tehran. 2005). D Soil andd Foundation Engineering E Deepartment.04 1 . A AMIRKAB BIR NATM M TUNNE EL.5Km lonng in eachtube (i. t Furthermmore. quantitative definition of these t parameteers is difficult because of restricting in site investigation data and a uncertaintiies related to them.e. Box 14115-1559 Tarbiat Modares M Univeersity. In comparisson with the monitoring m data. as a moore reasonablee and realistic Paperr No.

differrent method off excavation methods and at last l it has beenn reporrted that how turning t to a prrobabilistic meethod in designn Standdard Penetratioon tests (SPT) were done in all a boreholes too couldd help in preddicting the exccavation-inducced transversall invesstigate the streength and com mpressibility off the subsurface settlement. outcrropped near this t area.treatm ment for the uncertainties. s the tunnnel is dividedd coarsse grain soils anda consist off very dense grravel and sandd. hass Furthhermore. C and D. in-situ shearr test and plate Paperr No.04 2 . thehigghest populatioon density and traffic region of Tehran (Figg 1). This unit is stiffer than both of o B and D formations. into two branchess. A B. Amirkabir project. Siltstone andd geommechanical zoning. permeability test t (Lefrunc).In order to achievve the projecct objecctives. Conglomeratees leads to better underrstanding of th he project risks. Montee Carloo simulation alggorithm (Latinn Hyper Cube sampling). of Amirkabir tunnel with w approximaately 1. u has h been invokked to achievee propeerties of the site. whereas iit Dorooodian Street and the soutth branch is located underr is lowwer than the depth d that mighht be importannt to the projecct Kermman Street. a prrobabilistic appproach.. 5. and a mo ore reliable esttimation of thee Mudstone composeed A formationn. On thhe contrary. 12 boreeholes with depth of 40 meters.more efficientt with interbeded layers l such as a Sandstone.. layerrs. been employed to evaluate e the un ncertainty in liining design off mations of A annd D as well as alluvial fann deposits have form the paart T4 of the prroject. 2011). Fig 1 Location of Am mirkabir projecct Usingg different meethods of excaavation to avooid worse casee F 2 Plan of different divisiions of Amirkabir project Fig settlement in differrent part. years(Su et al. and C is maiinly from receent alluvial fann withA Amirkabirhighw way tunnelex xcavating below one off compposed of homogenous congloomerate.Pressuremeteer compposed of reccent alluvial. Required saamples were taken t during thhe drilling andd weree sent to laborratory to perfoorm different tests. In Kerman square. costs((Oreste 2005). Figgures 2 shows plan of differrent sectionsoff goalss. The north h branch is located underr The groundwater g taable is differennt in the borehooles. these sediments s weere categorizedd currennt requiremennts in many fields of geottechnics thesee intoffour different seeries namely A. Unceertainty in the Soil S Model of The T Site Geoloogy and Materiial Characteristic of The Site A raange of in-sittu and laboraatory experim ments includingg The study area iss located on sedimentary basin mainlyy Standdard Penetratioon Test (SPT). Based on stratigraphicall test. n the quality of A is the oldest deposits and D is the youngest. when it iss possibble to have sufficient data on o the material.A probabiilistic approacch. coontinuing withh are presented p in Figgure 3a. B formation is a heterogeneouus unit with low w The work reportted in this paper speccifically dealtt sortinng of grains. in-situ densityy. Borehole PROJJECT OVERV VIEW logs have been prepared based on field obserrvations duringg drilliings.5Km lonng in each linee Accoording to the field and laboratory test results and site (northh and south) iss located betw ween 17 Shahriivar Street andd invesstigations. thhe next sectionn the rooute of tunnel. The location of the boreholles and test pitts deal with w the overvview of the wh hole project. subbsurface layerrs conditions are generallyy Imamm Ali highwayy. accorrding to the geeological map of study areaa. grain sizee analyses and classification c o the samples. 1 borehole For better b understanding the diffficulties in dessign challengess with depth of 49 meters m and 13 test pits have been drilled inn facedd in varioussubbdivisions of the t project. f the younngest unit madee of recent alluuvial.

Rest of project has been constructed by either Cut Fig 4 NATM excavation method in T4 section & Cover method (where there is no limitation for blocking the road. 5. silty and clayey inter layers have deformations and as a result any possiblefailure. a-1)Variation in Cohesion b-1)Variation in Friction Angle of c-1)Variation in Young Module of d-1)Variation in Wet Density Value Value of Boreholes(x105Pa) Boreholes Boreholes(x105Pa) of Boreholes(x103Kg/m3) a-2)3-D Model Diagram of b-2) 3-D Model Diagram of c-3) 3-D Model Diagram of d-3) 3-D Model Diagram of Wet Cohesion Value(x105Pa) Friction Angle Young Module(x105Pa) Density(x103Kg/m3) a-2)Cross Section through b-2) Cross Section through c-3) Cross Section through d-3) Cross Section through Tunnel Tunnel Axis o indicating Tunnel Axis indicating variation Tunnel Axis indicating variation Axis indicating variation of Wet variation of Cohesion of Friction Angle of Young Module(x105Pa) Density(x103Kg/m3) Value(x105Pa) Fig 3 Variation in soil characteristic of the site area load test at the location of the boreholes had been performed. Deformations and settlements have been monitored and SM). Considering the field and laboratory test results. Figure 5) or Top/Down method (where there is limitation for blocking the road for some time and construction shall be performed with minimum blocking time Paper No. Underground excavation.04 3 .At the locations with minimum required soil overburden. In Cut & cover method stabilization of the trenches has been the soil of the area is composed of dense sandy gravels and performed by either Nailing (Figure 7) or Bored piling dense sand. been observed rarely. SC method. GM.e. Excavation Methods Construction has been performed by different methods i. Cut and Cover and Top/Down methods in different sections of the project depending on the soil overburden and mechanical properties of the soil. which both contains silt and clay (GC. Moreover. underground excavation method (NATM) has been performed (Figure 4). Figure 6).. of the road. on the whole. Sometimes these materials contain 5 to 50 percent of during and after construction in order to avoid unpredictable fine-grained soils. Figure 3 illustrates the variation of four different soil parameters visualizing in 3D and 2D diagrams.

the type (PHEM) approach has been used to evaluate the uncertainty in of limit state (i. Accordingly. Strength Limit State. In this project. the initial lining design of NATM tunnels in soft Extreme Limit State. elasticity modulus of concrete. Arrays of surfacesettlement markers were arranged approximately at where Eij and Tij (day) are the elasticity modulus and the intervals of30 m along the tunnel alignment. 3D modeling of tunnel using time dependent sprayed A finite Element analysis was conducted using the ABAQUS Paper No. CR j . which used to reliable design of this paper. The Wild NA2 average age of shotcrete of the parti of lining in phase j of Automatic Engineers’ Level was used for the measurement.The approach consists 4 principal elements (Fig 8): been considered. the loads of the service condition have been defined stochastically (Table 1). 1. 2D modeling of tunnel using Time dependent hypothetical elasticity modulus 3. j 0. Fig 6 Top/Down method inT4 section Fig 8 Concept of PHEM approach 3D tunneling effect can be derived from the comparison between 3D model with 2D model.According to (Chang 1994). Service Limit State. in many cases.01 mm. Fig 5 Cut & Cover method in T2 section performing of a 3D model in conjunction with a stochastic model to simulate the uncertainty of system is time consuming and often impracticable. especially when reliability analysis is considered. Monitoring The stress redistribution and the deformations occurring during tunnel face advance can be more properly simulated only if 3D numerical models are applied.446/ Ti . was proposed for this research to model the changing Fig 7 Nailing at TU Portal area of elasticity modulus of the sprayed concrete during the construction phases: Instrumentation 0. or Fatigue Limit State) should be defined ground.However. shotcrete 2. a probabilistic hypothetical elasticity modulus The PHEM approach is based on limit state function. performance of the initial lining of the tunnel has tunnels. It excavation process.is the calibration ratio derived from the comparison between 3D GENERAL APPROACH TO MANAGE UNCERTAINTIES model with 2D models(this averagely was equal to 1.e.7 Eij  1. respectively and E28 is the 28-day permits direct readings to 0. appropriate for both of the 2D and 3D time-dependent finite element models.Figure 1 shows the main elements and flow of the to characterize the applied loads. Stochastic model of the system 4.062E28 / CRj  e (1) Large numbers of surface settlement markers were installed to measureground settlements during excavation. In addition. 5. a modified empirical exponential model.25 for T4 OF THE TUNNEL: PART T4 OF THE PROJECT section). With regard to the case in proposed approach.1 mm and estimated readings to 0.04 4 .

was whichh is the assumeed time depend dent elasticity modulus m of thee f ( X )  0  Safe liningg. the FEM analysis.j analyysis procedurres employed d during thiis parametricc Parameter Average Lower L Upperr Distributionn sequeential excavatioon model (PSE EM) are as folloows: (h) Band(h) Band(hh) TOPHEM1 6 5 7 Uniform TOPHEM2 30 27 33 Uniform Geosttatic step: Inttroducing the initial stress state to reachh TOPHEM3 87 80 95 Uniform equiliibrium beforee tunnel excaavation beginns. LEF FTHEM4 and TOPHEM5. x 2 . i. full characterrization of the systeem and the other related information are describedd inTabbles2. Providing thhe probabilistic propeerties of the parameters. Potterr 1990). top heeading. left bench and top heading increased too the tuunnel has beenn modeled using elastic beaam elements inn RIGH HTHEM2.  . Baseed on a MATL LAB(Matlab 20010) interface program. 4th cycle c excavatiion step: The continuum elements in the the Hypothetical H M Modulus of Elasticity E (HMME) soft liningg inverrt were removeed. x N  2nd cycle c excavatioon step: The co ontinuum elem ments in the left ft invertt were remooved and thee beam elem ments with ann Wherre X is the vector of model innput and N is thhe number of LEFTTHEM1 value forf sidewall weere activated. 9).j Ti. LEFTHEM2 30 27 33 Uniform LEFTHEM3 87 80 95 Uniform 1st cyycle excavation n step: Excavaation of the top heading wass LEFTHEM4 129 115 140 Uniform achieved using the model m change options in ABA AQUS. reespectively andd than 5m. T two differeent geometries with differentt the riight bench. respecctively. Plaane strain analyysis was used ini the analysis.j Ti. A parametrric HEM-valuee INVERTHEM1 6 5 7 Uniform (TOPPHEM1) was usedu for the You ung Modulus of o lining. RIGHTHEM1 6 5 7 Uniform RIGHTHEM2 30 27 33 Uniform Meannwhile. to its flexibility wheen applied to multistage m tunnel excavations.The existing e road and a buildings over o the tunneel consiidered by two different lineaar distribution loads with the parammetric value of ROAD DLOAD and BUILLOAD D. INVERTHEM2 30 27 33 Uniform 1st cyycle lining stepp: The stiffnesss of the beam element e for thee The probabilistic p mit state function was defineed as below: lim w increased to TOPHEM2.the 1 detailedd Ti. 2 Karakuss and Fowell 2003. This is followed by the activation of the approoach(Karakus 2007. axis of o the tunnel T4. and 10m m for the secctions with thhe overburdenn simuulation was com mpleted. left ft bench and topp heading to IN NVERTHEM22. Table 2 Probabilistic parameters p of thhe HEMs Basedd on conceptuaal design of th he tunnel(Fig 10). The water w table iss assumed to be 40m beloow the groundd 3rd cycle c lining sttep: The HEMM value for thee lining for the surfacce.pre. has beeen used in thiis research duee valuee for the invertt. The beam m TOPHEM4 129 115 140 Uniform elemeents representinng the lining were w deactivateed.and a post-proceessing finite element e prograam. Paperr No. activvated. the lining elements foor the top headding with lowerr RIGHTHEM3 87 80 95 Uniform elasticity modulus were activated d. Shotcrete used u as a prelliminary suppoort measure inn right bench. 2 John andd beamm elements reppresenting the lining with INVERTHEM I 1 Mattlle 2003. The probaabilistic modell of the soil was w constructedd right bench were reemoved and thee beam elemennts representingg basedd on in-situ andd laboratory ex xperiments meentioned beforee the lining l with ann RIGHTHEM M1 value for the wall were in Figgure 3. (2) f ( X )  0  Failuree X  x1 . LEF FTHEM3 and TOPHEM4. betweeen 5m and 100m) with 120 m wide and 400 m high weree used and were nameed A and B(Fig g. Regarrding the channges in the amo ount of overbuurden along thee 4th cycle c lining steep: Increasing the value of HME H for invertt. Amirkabir's soil was w modeled usingu the undrrained materiall propeerties with thhe Mohr-Cou ulomb failuree or strengthh 3rd cycle c excavatioon step: The soolid elements representing r the criterrion. 3. In I order to estimate the surfaface settlement. reespectively.04 5 .e. Fig 9 Adoptedfinite element model respeectively. thee HME value. 5. overbburden (5m reppresenting the sections s with overburden o lesss RIGH HTHEM3. Simulationn of thhe NATM tunnneling processs was commenced with thee 2nd cycle lining: The HME vallue for the linning on the lefft selecttion of the tunnnel geometry and the modeel geometry inn peripphery and topp heading inccreased to LEEFTHEM2 andd two-ddimensions. 10000 Latinn Hypercube simulations s weere performed on each of the non-ccircular surfacce to determinee the probabilistic limit state functtionof ABAQU US-based finitte element moodel(ABAQUS S 20100) which was described d abovve. as there wass TOPHEM5 215 195 235 Uniform LEFTHEM1 6 5 7 Uniform no linning at the begiinning of the an nalysis. TOPH HEM3.

2 Monitoring Feedbacks Based on the measurement of settlement pins installed on the pavement of road above the tunnel. POSOIL 3 Density γsoil 1960 120 1720 2090 N Kg/m DENSOIL Lining Parameters Distribution 3 Density γcon 2400 240 N Kg/m DENCON Pa ELBEAM Young’s modulus E28= 4.18 0.22 Beta .03 0.21E3 2. having adopted by the MCS results.03 0.7E6 fc 28-day Compressive f c 2.68e7 9.41E+7 2. shows that the proposed sequential excavation method is numerically certain enough throughout the length of the tunnel in order to limit the surface settlements induced.30 U m TBEAM Loading Parameters Distribution Road Load LL 30000 7500 N Pa ROADLOAD Building Load DL 20000 5000 N Pa BUILLOAD model A were in the range between 19mm and 51mm. More investigation in Monte Carlo simulation. Regarding the changing trends of variables versus the output.66e7 N modulus Poisson ratio υsoil 0. Table 3 Probabilistic parameters of the system Parameter Symbol Mean STD Lower Bound Upper Bound Distribution Unit Variable' s Name Soil Parameters Distribution Cohesion Csoil 3. For the Service Limit State. the monitored transversal settlements of surface road during the construction phase are illustrated in Figure 13. The statistical results indicated that the settlements of the Paper No.90E+7 1.4E7 N Pa COMPST Strength Poisson ratio υcon 0. y  f ( X )  n  i (3) Meanwhile. and its minimum value was 31mm. respectively. This emphasizes the fact that the construction method Performing 1000 LHCS for each model.02 0.72 29 35. Stochastic Results As the results of Monte Carlo simulation (MCS). a set of stratified was safe enough.3 N Degree PHISOIL Young’s Pa ESOIL Esoil 8. than the factors depending to the structure.285 0.315 Beta .45 1. POIBEAM Thickness t 0.27 0. it is crucial to import that Latin Hyper Cube Simulation (LHCS) Results none of the cases surveyed have indicated serviceability failure. whereas the mean-value and STD of the outputs of model B were 41mm and 5mm. it can be seen that the model is more sensitive to the variable of the soil as well as active loading.04 6 . probabilistic distribution function (PDFs) of maximum surface settlement was derived for each. the maximum of settlements reached a pick-value of 58mm.2 0. with random variables.19E5 logN Pa CSOIL Friction angle φsoil 32. f is the mean-value and standard deviation(STD) of 31mm and characterized as(FHWA 2001. 5. Figure 11 shows the surface plots of 5 principle variables of the system which were compared two by two versus maximum transversal settlement (MINDIS) of two different models. Hung et al. Fig 10 NATM Excavation Stages Regarding the measured settlement.12E4 6.5E7 4E+6 1.7e4 1.24 0. where  i is the estimated displacement and  n is tolerable displacement established by designer.3 0. 2009): 5mm. not included in this paper. excepting the load of building. respectively.

26. Tunnelling and than the factors depending to the structure and also advance Underground Space Technology.. 422–434. Road Chang. (1990) "Green Shotcrete in tunneling: stiffness- any possible failure. M.10. Washington.-H. Department of Transportation Federal Highway Administration. highway tunnel in China". N. Matlab (2010) Construction has been performed by different methods i. P. Road FHWA (2001) "Load and Resistance Factor Design (LRFD) Load for Highway Bridge Substructures". M. (2005) "Risk analysis for sensitive to the variable of the soil as well as active loading determination of a tunnel support pattern". 47-56. Elasticity simulate the uncertainty of the system gave real judgment on the consequences of the excavation phase of the project. X. heterogeneous sedimentary basin mainly composed of recent Tunnelling and Underground Space Technology. Cohesion vs. J. 479–486. J. and Wisniewski. On the whole. John. (2009) Technical Manual for Design and Construction of vs. Z. (1994) "Tunnel support with shotcrete in weak load rock" L A rock mechanics study. Y. rate. and Mattle. Underground excavation. Li. (2003) "Effects of different Tehran. and Fowell. 513-523. (2011) "Probabilistic Cube sampling). 18. Deformations and supports".. strength of concrete Hung. (2005) "A probabilistic design approach for tunnel methods in different sections of the project. K. has been employed to evaluate the evaluation for the implicit limit-state function of stability of a uncertainty in lining design of the part T4 of the project. Park.S. translated by ASCE. D. version 6. and Lee. (2003) "Factors of shotcrete lining Figure 11 Surface plots of 5 principle variables of the design". Road load Road Tunnels —Civil Elements. Monsees. J. The MCS results of section T4 showed that regarding uncertainties related to the system.04 7 . avoid worse case settlement in different part. Phi REFERENCES ABAQUS (2010) Theory Manual. J. and Xie. a probabilistic approach. 520–534. B. Munfah. J. Amirkabir highway tunnel which is one of the important projects excavating below one of the highest traffic region of Karakus. Cohesion vs. R. The work reported in this paper specifically dealt with Tunnelling and Underground Space Technology. CONCLUSION Karakus.S. unpublished thesis Royal Institute of Technology. Tunnelling and Underground Space Technology. Monte Carlo simulation algorithm (Latin Hyper Su. C. turning to a probabilistic method in design could help in predicting the excavation-induced transversal settlement.. in The Rapid Excavation and Tunneling Conference system based on Monte Carlo simulation output New Orleans. . Cut and Cover and Top/Down Oreste. U. 20. Computers and Geotechnics.-Y. Dassault Systèmes Simulia Corp Elasticity vs. US. R. Department of Transportation. Paper No. Comp. 5. the model was more You. having treated uncertainties related to the excavation rate and soil properties. settlements have been monitored during and after construction in order to avoid unpredictable deformations and as a result Potter. M. 20590:National Highway Institute. S. 32.. Using different methods of excavation to strength-deformation". alluvial.C. U. including the difficulty of excavating through a tunnel face advance excavation on the settlement by FEM". (2007) "Appraising the methods accounting for 3D tunnelling effects in 2D plane strain FE analysis". 22.. Y. Model A Model B Comparing monitoring data with stochastic outcomes of the model demonstrated that the method used in this research to Cohesion vs. Y.e.

5.Figure 13 North tunnel longitudinal profile (numbers marked by red circles and blue rectangulars indicate the monitored transversal settlements of surface road and the values of overburden in each station. respectively) Paper No.04 8 .