PUBLICATIONS Geotechnical Engineering in Research and Practice WBI-PRINT 5 Edited by Prof. Dr.-Ing. W. Wittke Beratende Ingenieure für Grundbau und Felsbau GmbH Consulting Engineers for Foundation Engineering and Construction in Rock Ltd. New Austrian Tunneling Method (NATM) Stability Analysis and Design Walter Wittke, Berndt Pierau, Claus Erichsen translated into English by: Jens Lüke and Johannes R. Kiehl Translated from the German edition: Statik und Konstruktion der Spritzbetonbauweise. Geotechnik in Forschung und Praxis, WBI-PRINT 5, VGE-Verlag Glückauf GmbH, Essen 2002, ISBN 3-7739-1305-2
Transcript
PUBLICATIONS
Geotechnical Engineering in Research and Practice
WBI-PRINT 5
Edited by Prof. Dr.-Ing. W. Wittke
Beratende Ingenieure für Grundbau und Felsbau GmbH
Consulting Engineers for FoundationEngineering and Construction in Rock Ltd.
New Austrian Tunneling
Method (NATM)
Stability Analysis and Design
Walter Wittke, Berndt Pierau, Claus Erichsen
translated into English by:Jens Lüke and Johannes R. Kiehl
Translated from the German edition: Statik und Konstruktion der Spritzbetonbauweise. Geotechnik inForschung und Praxis, WBI-PRINT 5, VGE-Verlag Glückauf GmbH, Essen 2002, ISBN 3-7739-1305-2
Within the series "WBI-PRINT, geotechnical engineering in researchand practice", volumes 4 to 7 are designed as a compendium of tun-nel statics. This compendium started with the volume WBI-PRINT 4"Stability analysis for tunnels, fundamentals", published in 1999in German and in 2000 in English.
The present volume WBI-PRINT 5 "New Austrian Tunneling Method sta-bility analysis and design" covers, beside fundamentals of the NewAustrian Tunneling Method (NATM), case histories of realized minedtunnels designed and constructed with participation of WBI.
The selected case histories from the years 1985 to 2001 includecrown headings with open and closed invert, sidewall adit head-ings, full-face headings and headings under the protection of pipeumbrellas and jet grouting columns.
Analyses according to the finite element method have proved to bean indispensable tool for the design of tunnels. The stabilityanalyses for all case histories presented were carried out usingthe program system FEST03. In order to enable this program systemto be used by our professional colleagues as well, we have beenoffering it for sale for some little time now.
WBI-PRINT 5 has been previously published 2002 in German as a pa-perback. Now the English translation is available online to pro-vide a worldwide access to those who are interested in tunneling.It is also available on CD-ROM via WBI company.
The next volume in the series WBI-Print is dedicated to the mecha-nized tunneling. This volume appears as WBI-PRINT 6 in German inDecember 2006. Special problems of tunnel statics will be coveredin WBI-PRINT 7.
I adress my special thanks to my two co-authors and directors atWBI, Dr.-Ing. B. Pierau and Dr.-Ing. C. Erichsen, who have beensupporting my work substantially for many years. I am also obligedto Dr.-Ing. J. R. Kiehl for his editorial work. The translationinto English was carried out by Dr.-Ing. J. Lüke as well as Dr.-Ing. J. R. Kiehl. I convey my sincere thanks to them. Furtherthanks are due to our secretary and design office.
The "New Austrian Tunneling Method" (NATM) was originally appliedfor tunnels in rock. Since the 1970ies, however, this tunnelingmethod was carried out more and more also in soft rock with lowoverburden and in urban areas. Because of the outstanding impor-tance of the shotcrete (sprayed concrete) for the application ofthis method the denotation "Sprayed Concrete Lining Method" orsimply "Shotcrete Method" is mainly used in Germany.
The NATM is a construction method, which is very adaptive accord-ing to changing subsoil conditions and changing shapes of cross-sections. Interacting with the subsoil the primary function of theshotcrete membrane is to form an arch around the tunnel, which iscapable to carry. With a favourable shape of the tunnel's cross-section and an adequate sequence of construction stages it is pos-sible to avoid or at least to minimize bending moments and shear-ing forces in the shotcrete membrane. Thus, large undergroundopenings can be supported by relatively thin shotcrete membranes.With an adequate design also the subsidence on the surface can belimited to relatively small values.
Stability analyses, in which the interaction of the subsoil withthe support are modeled in a realistic way, however, serve as aprerequisite for a successful tunnel heading using this method.The authors are convinced that this is possible only by numericalcomputation methods. Stability analyses, therefore, should be car-ried out generally using finite element codes. A powerful tool,which is suitable also for three-dimensional problems, is the fi-nite element code FEST03 developed by WBI and documented in thevolume WBI-PRINT 4 (Wittke, 2000). Since more than 20 years thisprogram, which in this period of time has been improved and en-larged several times, serves as an valuable device for a safe andeconomic design of tunnels.
The design of a tunnel according to the NATM is carried out step-wise with the following working steps, which are to be repeatedseveral time, if required:
- Geotechnical investigations of the ground and groundwaterconditions.
- Evaluation of the soil and rock mechanical parameters, basedon test results as well as experience.
- Stability analyses for both, the stability proof of the tun-nel and the design of the shotcrete membrane as well as theinterior concrete lining.
- Design and assessment of excavation methods and support meas-ures (excavation classes).
- Supervision of stability by geotechnical mapping and monitor-ing during construction.
- Back analysis of the results of measurements.
The authors of the given volume WBI-Print 5 since more than 20years are experienced with the stability analysis and design oftunnels carried out by the NATM. With this volume it will be at-tempted to transmit this experience by case histories.
In Chapter 2 an overview on the fundamentals of the NATM is given.Here also new developments such as non-alkaline shotcrete aretreated. Moreover in Chapter 2 geotechnical mapping and monitor-ing, which are essential parts of this tunneling method, are dealtwith.
Advancing crown headings with open and closed invert are treatedin the Chapters 3 and 4. In each chapter three case histories arepresented.
Advancing sidewall tunnel headings are subject of Chapter 5. Herealso three case histories are documented.
In Chapter 6 two more case histories are presented, in which afull-face excavation at least in sections were carried out.
Two case histories for headings under the protection of jet grout-ing columns are dealt with in Chapter 7.
The documentation of the case histories is, as a rule, arranged asfollows:
In order to improve workers' protection, minimize environmentalpollution (water and ground) and reduce the amount of eluates (al-kalis, calcium hydroxides), shotcrete mixes may only be applied ingeneral if they at least equivalent to conventional structuralconcrete mixes for support elements with respect to their physio-logical properties and their leaching behavior.
The following requirements, among others, have to be met by theshotcrete:
- Low water permeability,
- no use of alkali-containing additives,
- a minimum strength of the green shotcrete, termed earlystrength (see Chapter 2.1.3).
The required early strength for the shotcrete can be achieved byeither:
- The use of so-called spray bonding agents (SBM) or spray ce-ments, which allow to dispense with setting activators, or
- the use of alkali-free accelerating admixtures in powder orfluid form.
In special cases, e. g. with a high water discharge, spray bondingagents and alkali-free accelerating admixtures may also be appliedin combination (ÖBV, 1998).
Bonding agents
According to DIN 18551, the following bonding agents may be used:
- Standard cements according to DIN 1164 (Parts 1 and 100,1990),
- spray bonding agents or spray cements certified by the super-visory authorities.
If spray bonding agents or spray cements have not been certifiedby the supervisory authorities, the suitability of the bondingagent for the production of shotcrete must be proved before con-struction by a testing certificate from an approved institute formaterials testing. With respect to the leachability the amount ofeluate must not be greater for this bonding agent than for stan-dard cements. Proof of this must be provided by a testing certifi-cate from a public health institute.
On the basis of their rate of reaction, one distinguishes betweentwo types of spray bonding agents (ÖBV, 1998):
- Spray bonding agent SBM-T:With a maximum processing time of less than one minute, thistype of bonding agent can only be used for the production ofshotcrete with dry aggregates (water content w ≤ 0.2 M.-% andaccording to the manufacturer's specification, respectively).
- Spray bonding agent SBM-FT:With an admissible processing time of several minutes, thistype of bonding agent can also be used for the production ofshotcrete with wet aggregates (water content w generally2 M.-% to 4 M.-%).
Admixtures
With respect to the improvement of the shotcrete properties suchas workability, stickiness, formation of dust, rebound, strengthand tightness of the shotcrete fabric as well as reduction of theheat production, adding hydraulically active admixtures is useful(ÖBV, 1998).
Fly ash is a proven admixture, but the use of other admixtures isalso possible (e. g. silica dust, smelting sand, hydraulic lime).The total amount of added ground material and admixtures must notexceed 35 % of the bonding agent (ÖBV, 1998).
For shotcrete, concrete aggregates as specified in DIN 4226 (parts1 and 2, 1983) must be used (DIN 18551, 1992). The maximum grainsize must be selected between 4 and 16 mm (ÖBV, 1998).
Additives
Until a few years ago, alkali-containing accelerating admixtureswere used as additives for shotcrete. This way it was possible toachieve a favorable development of early strength (see Chapter2.1.3). These additives are strongly caustic and due to reasons ofenvironmental protection they are not used anymore. Furthermore,they have a negative effect on the leaching behavior of the shot-crete. This has lead e. g. to drainages being clogged by encrusta-tions and in some cases also to contaminations in the groundwatercaused by the eluates. In addition, the shotcrete became porousand permeable to water with the leaching. This results in decreas-ing strength with progressing age.
It is therefore state-of-the-art today to use spray bonding agentsor spray cements without accelerating admixtures or with alkali-free accelerating admixtures, added as powder or in fluid form andcertified by the supervising authorities.
The suitability of the planned shotcrete recipe including the usedadditive must be proved before construction by laboratory testingof the setting behavior, the early strength and the strength de-velopment. Laboratory tests yield reference values, but they can-not capture all influences from the construction site and there-fore cannot replace suitability testing on site (ÖBV, 1998). Fur-thermore, it has to be proven that the additives do not have anegative impact on the reinforcement and the remaining steelmounting parts.
Composition
According to ÖBV (1998), the mixes for dry-mix and wet-mix shot-crete are subdivided into:
These mixes are referred to as supply mixes. They differ in compo-sition from the sprayed concrete due to the rebound occurring dur-ing spraying. The rebound is the share of the shotcrete mix whichdoes not adhere to the surface of application during spraying andwhich must be disposed of.
2.1.2 Spraying methods
Dry-mix method
For the dry-mix method, TM, FM-L and FM-S mixes can be used. Themix is conveyed intermittently to the spray nozzle via compressedair using a piston or rotary engine (thin stream transport). Atthe nozzle, it is wetted with water and sprayed onto the surfaceof application at a speed of 20 m/s to 30 m/s.
Fig. 2.1: Placing of the shotcrete by a manually guided spraynozzle (DB, 1985)
Liquid accelerating admixture is added continuously to the supplywater using dosing pumps. Accelerated admixture in powder form isadded immediately before the mix transport using proportioners.
Nowadays, the shotcrete is only rarely applied to the tunnel sur-face by a manually guided spray nozzle (Fig. 2.1). It is standardpractice to apply the shotcrete using a spray vehicle with a re-mote-controlled spray arm (Fig. 2.2). Due to the high velocity ofthe shotcrete during placing, high rebound portions arise from thedry-mix method. A further problem is the resulting heavy formationof dust. The dry-mix method is therefore only permitted for spe-cial cases by the government safety organizations today.
An alternative transportation technique for dry-mix shotcrete wasdeveloped by the Rombold und Gfröhrer Co. Here, the dry-mix shot-crete is subjected to compressed air in a pressure tank (silo) andconveyed via dust-encapsulated dosing screws continuously anddust-free with the air stream to the spray nozzle. As in the con-ventional dry-mix method, the water is added only just before thenozzle (Balbach and Ernsperger, 1986). As a bonding agent, spraycement with a swift development of strength (fast cement) and ahigh final strength is used (see Chapters 2.1.3, 2.1.4 and 4.1.4).The addition of an accelerating admixture is therefore not neces-sary.
Advantages of the dry-mix method are the workability in smallamounts and the transportability over long distances.
Wet-mix method
With the wet-mix method, the wet mix (NM) is conveyed by thespraying machine to the spray nozzle either by compressed air(thin stream transport) or hydraulically using piston pumps (thickstream transport).
Like dry-mix shotcrete, wet-mix shotcrete is generally applied us-ing a spray vehicle with a remote-controlled spray arm (Fig. 2.2).Manually guiding the spray nozzle is problematic because of thehigh weight of the wet-mix shotcrete.
Less rebound, less formation of dust and a higher spraying per-formance are advantages of the wet-mix method over the dry-mixmethod.
Before the shotcrete is applied, loose rock must be removed fromthe excavation surface. The surfaces of application must be care-fully cleaned with compressed air in order to achieve the bestpossible adhesion of the shotcrete. This particularly applies ifthe shotcrete lining is constructed in layers or if longer inter-ruptions occur during the application of the shotcrete.
An immediate sealing of the exposed rock surfaces with shotcreteof at least 3 cm thickness is intended to provide early support tothe ground close to the excavation surface in order to largelyavoid loosening and the resulting decrease in rock strength.
The shotcrete must be applied in such a way that a homogeneous,dense shotcrete with a closed, even surface is achieved.
Thick shotcrete linings are applied in two or more layers in orderto avoid separation from the excavation surface. The shotcretemust be applied in such a way that spraying rebound and adherentspray dust into the shotcrete is avoided by all means. Rebound anddust must be removed before the next shotcrete layer is applied,and the shotcrete lining must always be constructed from the bot-tom to the top.
The distance between the nozzle and the surface of applicationmust be adapted to the delivery rate and the speed of application.It ranges between 0.5 and 2.0 m, depending on the air flow. Thenozzle should be oriented at right angles to the surface of appli-cation, if possible. Exceeding or falling below the recommendednozzle distance as well as an inclined orientation of the nozzlerelative to the surface of application generally lead to a reducedquality of the shotcrete and an increased amount of rebound. Incase of steel insertions such as steel arches, steel girders, lag-ging plates, pipes, etc., spray shadows cannot be totally avoided,but they can be considerably reduced by proper nozzle control(ÖBV, 1998).
Special care has to be taken when the connection is made to theexisting shotcrete lining in the crown invert, the bench invertand the permanent invert. Existing rebound must be removed first.The reinforcement and the support arches should be completelywrapped up in shotcrete. It is important that the visible surfaces
are constructed in a convex shape only, if possible, in order toachieve an arching effect.
2.1.3 Early strength
Shotcrete up to an age of 24 hours is termed green shotcrete.
With respect to the requirements on strength development, greenshotcrete is distinguished into the three early strength classesJ1, J2 and J3 determined on the basis of years of experience (Fig.2.3).
Fig. 2.3: Early strength classes of green shotcrete (ÖBV,1998)
Class J1 shotcrete is suitable for the application of thin layerson dry surfaces without specific statical requirements. It has theadvantage of little dust development and rebound.
If statical requirements exist with respect to the green shot-crete, e. g. for the exterior lining of a traffic tunnel, class J2shotcrete is generally used. Class J3 shotcrete is required if rap-
idly developing loads from water pressure and/or rock pressure areto be expected.
It is known from experience that dry-mix shotcretes allow toachieve the highest early strength values. Fig. 2.4 shows the com-parison of the strength development of two wet-mix shotcretes, onewith accelerating admixture containing alkali (1) and one with al-kali-free accelerating admixture (2), and one dry-mix shotcretewith alkali-free spray bonding agent (3). The latter type reachesclass J3. Using accelerating admixtures, either containing alkalior alkali-free, class J2 is reached. Here, the shotcrete with al-kali-free accelerating admixture shows greater early strength.
Fig. 2.4: Comparison of the strength development of shotcreteswith accelerating admixtures, either containing al-kali or alkali-free, and alkali-free spray bondingagent
Because of the disadvantages of the dry-mix method compared to thewet-mix method, which are discussed in Chapter 2.1.2, wet-mixshotcrete is often preferred over dry-mix shotcrete also when high
demands are made on the early strength, notwithstanding that ahigher early strength can be achieved with dry-mix shotcrete.
An example for this is the shotcrete for the Schulwald Tunnel ofthe new railway line (NBS) Cologne – Rhine/Main of German Rail.Due to the predicted poor geological conditions, dry-mix shotcretewith spray cement as bonding agent was selected at first. Aftercomprehensive preliminary testing, an early strength correspondingto class J3 was achieved with the recipe given in Fig. 2.5. Becauseof the high formation of dust during the application and the in-sufficient placement performance, it was later decided to changeto wet-mix shotcrete with a liquid, alkali-free accelerator (BEU22). With this shotcrete, the recipe and early strength develop-ment of which are given in Fig. 2.6, a class J2 early strength wasachieved which proved sufficient. A placement performance of up to25 m3/h was obtained with this wet-mix shotcrete, which clearly ex-ceeds the 14 m3/h achieved with the dry-mix method. Further, re-bound values of less than 10 % were reached (Brötz et al., 2000).
Fig. 2.5: Recipe and early strength development of the dry-mixshotcrete for the Schulwald Tunnel of the new rail-way line Cologne – Rhine/Main (Brötz et al., 2000)
Fig. 2.6: Recipe and early strength development of the wet-mixshotcrete for the Schulwald Tunnel of the new rail-way line Cologne – Rhine/Main (Brötz et al., 2000)
2.1.4 Final strength
Besides its rapid strength development, shotcrete with alkali-freeaccelerating admixtures or spray bonding agents also possesses ahigh final strength. While accelerators containing alkali impedethe hydration of the cement, this is not the case if alkali-freeaccelerating admixtures or spray bonding agents are used. A finalstrength of 30 to 40 N/mm2 can be obtained in practice (Brötz etal., 2000; Bauer, 2000; NATS, 1998). With shotcrete with accelera-tors containing alkali, even a final strength of 25 N/mm2 combinedwith a high early strength is difficult to achieve (NATS, 1998).
The deformation behavior of green shotcrete according to Hesser(2000) is essentially characterized by
- hardening with time and a creep ability decreasing with time,
- overproportional, non-linear creep with increasing load.
The Young's moduli determined by Hesser (2000) in laboratory testson dry-mix shotcrete test specimens of different ages show goodagreement with the relations by Weber (1979) and by the Comité Eu-ro-International du Béton (CEB, 1978) (Fig. 2.7a).
Fig. 2.7b shows the development of Young's modulus in the first 24hours. According to this, Young's modulus of the shotcrete amountsto approx. 15000 MN/m2 after 24 hours, with creep and shrinkage notbeing taken into account.
Investigations by Manns et al. (1987) showed that the creep andshrinkage deformations of wet-mix shotcrete are larger than thoseof dry-mix shotcrete. In comparison to standard concrete, thecreep and shrinkage deformations of shotcrete are generallyclearly larger.
In finite element stability analyses for tunnels, the developmentof Young's modulus of the shotcrete with time as well as creep andshrinkage are generally not taken into account. An equivalentmodulus is instead assigned to the shotcrete to account for hard-ening during application of the load as well as creep and shrink-age.
The interpretation of the displacements and stresses measured atdifferent tunneling projects by means of back analyses has shownthat a modulus of E = 15000 MN/m2 can be used as equivalent modulusof the shotcrete, corresponding to the 24-hour-value after CEB andWeber (see Fig. 2.7b). Specially in cases, where the shotcrete isalready loaded at a young age due to short round lengths and earlyclosing of the invert, values of 2000 to 7500 MN/m2 for the equiva-lent modulus have proven to be more realistic (see Chapters 6.1and 7.1).
The development of the deformability of shotcrete with time in-cluding shrinkage and creep and its representation in numericalanalyses are still a subject of further development.
2.1.6 Rebound
As mentioned above, the amount of rebound is higher for dry-mixshotcrete than for wet-mix shotcrete. Using dry-mix shotcrete withspray bonding agent, however, allows to reduce the rebound consid-erably as compared to conventional dry-mix shotcrete (Fig. 2.8).
Fig. 2.8: Comparison of rebound between conventional dry-mixshotcrete, dry-mix shotcrete with spray bondingagent and wet-mix shotcrete (NATS, 1998)
The rebound further depends, among other things, on the water ce-ment ratio, the aggregates and the cement type (Maidl, 1992).
2.2 Steel sets
2.2.1 Basic types
Steel sets are made with different profiles. Examples are shown inFig. 2.9. One distinguishes between plain girders and latticegirders.
Among the plain girders are e. g. GI profiles (mining I profiles),TH profiles, bell profiles and standard profiles. Their dimen-sions, weights, geometrical moments of inertia, section moduli andcharacteristic parameters are given in Fig. 2.10. Further plaingirders are star profiles, the specifications of which are shownin Fig. 2.11.
For the same bending stiffness, lattice girders have less weightper meter of girder length than plain girders. They are thereforeeasier to handle. Lattice girders are distinguished into 3-stringer girders (Fig. 2.15) and 4-stringer girders. Fig. 2.16shows the specifications for 3-stringer lattice girders in whichthe rods are welded to the stiffening elements from the inside.The rods of the Pantex 3-stringer and 4-stringer PS-girders arewelded to the stiffening elements from the outside (Fig. 2.17,2.18 and 2.19).
For the same height, plain girders like GI, TH and other profileshave a far greater normal and bending stiffness than lattice gird-ers (Fig. 2.20). The normal stiffness of lattice girders is inde-pendent of their height, if the cross sectional area of thestringer rods As remains constant. In Fig. 2.20, As = 12.4 cm2 wasassumed for the stringer rods (d1 = 20 mm, d2 = 28 mm, see Fig.2.16).
Steel sets only become fully effective as a support if they form aclosed ring. In this way they are often used as support in miningwithout additional support measures.
In tunneling according to the NATM, the steel sets have furthertasks as:
- Immediate support of the tunnel face area over the length ofthe foremost round,
- template of the tunnel profile for the application of theshotcrete and the excavation of the next round,
- support for spiles installed as advance support for the nextround (see Fig. 2.15 and Chapter 2.4.1).
In tunneling according to the NATM, steel sets are only rarely in-stalled as a closed ring after each round. Therefore, immediatelyafter their installation, they only have very little bearing ca-pacity.
For the NATM, the load-carrying behavior of the steel sets isbased on their bond with the shotcrete membrane. Immediately aftera steel set is installed and covered with shotcrete, when theshotcrete still has a very low Young's modulus, it is mostly thesteel set that carries the loads resulting from rock mass pres-sure. Since steel sets are usually installed closely behind thetunnel face, this loading at the beginning is generally small.With progressing hardening of the shotcrete with time, the normalstrength and thus the bearing capacity of the shotcrete membraneincreases. Finally, after the shotcrete has fully hardened, thenormal stiffness of the steel sets can be neglected compared tothe one of the shotcrete membrane.
Fig. 2.21 illustrates this for the example of a 30 cm thick shot-crete membrane. It shows the ratio of the normal stiffnesses ofthe shotcrete membrane and the steel sets vs. the Young's modulusand the age of the shotcrete, respectively for two GI profiles andone 3-stringer lattice girder. A spacing of the steel sets of 1 mis assumed for all cases. It becomes evident that already at theage of a few hours the normal stiffness of the shotcrete membrane
exceeds the one of the steel sets. After 24 hours, the normalstiffness of the shotcrete membrane amounts to 17 times of that ofthe lattice girders and 4 to 8 times the one of the GI profiles.
Fig. 2.21: Ratio of normal stiffnesses of shotcrete membraneand steel sets vs. Young's modulus and age of theshotcrete
The steel sets covered with shotcrete can also be accounted for aspart of the reinforcement for the dimensioning of the shotcretemembrane (e. g. stringer rods of lattice girders). This requireshowever that the steel sets are completely covered with shotcrete.Especially if plain girders are used, however, the bond is reduceddue to spray shadows. Therefore, in general steel sets are conser-vatively disregarded as reinforcement and in finite element analy-ses, the steel sets are generally not modeled.
In Guideline 853 of German Rail (DB, 1999), the following criteriafor the selection of steel sets are given:
- Plain girders yield more stable immediate support than latticegirders. This requires, however, friction-locked connectionsbetween the supports and the rock.
- Lattice girders bond better with the shotcrete and lead gener-ally to tighter shotcrete membranes than plain girders.
- Spray shadows are generally smaller for lattice girders thanfor plain girders.
Details regarding the better bonding effect of lattice girderswith shotcrete and the consequences for the strength and tightnessof the shotcrete membrane can also be found in Eber et al. (1985).
2.3 Anchors
2.3.1 Basic types
Except for special cases, in tunneling mainly non-prestressed (un-tensioned) anchors, termed rock bolts, are used in boreholes. Adetailed description of the terms and designations of rock boltsis given in the German standard DIN 21521 "Gebirgsanker für denBergbau und den Tunnelbau" ("Rock bolts for mining and tunnel-ing").
With respect to the load-carrying behavior (see Chapter 2.3.2),bond anchors, which are form-locked with the rock mass (Fig. 2.22ato c), are distinguished from anchors that are friction-lockedwith the rock mass (Fig. 2.22 d). In addition, there are anchorsthat are form-locked as well as friction-locked with the rock mass(Fig. 2.22e).
Among the form-locked anchors are the mortar anchors (e. g. SN-anchors, IBO-rods), the resin anchors and the friction pipe an-chors (e. g. Swellex anchors, split-set anchors). A friction-locked bond with the rock mass results for the expansion shellbolt.
In the case of the mortar anchor and the resin anchor, the bondbetween the anchor rod and the rock is effected by a setting agent(cement mortar or synthetic resin mortar) over a specific lengthof the borehole (Fig. 2.23). If the bond extends over the fulllength of the borehole, the terms full bond anchor or fully ce-mented anchor are used as well. Among the full bond anchors arealso the friction pipe anchors, the anchor rod of which consistsof a pipe folded in the longitudinal axis (Swellex-anchor) resp.slit open (split-set-anchor). In the borehole, this pipe is bracedagainst the rock mass by pressing it against the borehole wall.Expansion shell bolts are rock bolts in the case of which the bot-tom end of the anchor rod is braced against the borehole wall us-ing wedge-shaped or conical elements (expansion elements) (DIN21521, Fig. 2.22d).
The use of SN-anchors, resin anchors, friction pipe anchors andexpansion shell bolts requires that the boreholes for the instal-lation of the anchors are stable. Fig. 2.23 shows the workingsteps for the installation of a mortar anchor (SN-anchor). For un-stable boreholes so-called injection drill bolts (IBO-bolts) areused (Fig. 2.22c and 2.24). Injection drill bolts consist of ananchor pipe which is made of high-strength steel with a continu-ously rolled-on thread and constitutes the drill bar. A drill bitis screwed on to the bottom end of the anchor pipe. After theborehole has been drilled to the desired depth, it remains withthe anchor pipe in the borehole. The bond between the anchor pipeand the rock mass is accomplished by the injection of cement sus-pension via the anchor pipe through injection openings in thedrill bit and the anchor pipe. The loosened rock mass surroundingthe borehole is also injected and stabilized in the process (Fig.2.24).
Expansion anchors constructed as injection bolts (Fig. 2.22e) rep-resent a combination of a friction-locked and form-locked connec-tion between the anchor rod and the rock mass. The working stepsfor the installation of this anchor type are shown in Fig. 2.25.
In tunneling, non-prestressed (untensioned) anchors are generallyinstalled as systematic anchoring for the excavation support in araster determined on the basis of statical criteria. Anchorlengths between 3 and 6 m are common. Using sleeve connections,anchor lengths up to approx. 18 m are feasible. The anchors basi-cally consist of an anchor rod and an anchor head with an anchorplate.
Anchor manufacturers offer anchor rods made from steel and glassfiber reinforced synthetics with different strengths and cross-sections (plain section, pipe cross-section). The advantages ofglass fiber anchors over steel anchors lie mainly in the fact thatthey can be cut as well as bent. Glass fiber anchors are thereforeoften installed at locations where they have to be removed in thecourse of further excavation, e. g. for the support of the innerwalls during sidewall adit heading or for the support of the tun-nel face. Disadvantages of glass fiber anchors are the facts thatthey can carry only very small shear forces and that the transferof point loads into the anchor at the anchor head is difficult toenable with the anchor design.
The anchor heads should always be constructed in such a way thatthe anchor plates lie flat on the excavation surface or on theshotcrete and that the load transfer from the anchor plate to theanchor rod does not lead to bending or shear loading of the anchorrod. Therefore, mainly so-called sphere cap anchor plates are usedthat are spherically shaped around the hole. For the nuts screwedonto the anchor rods, so-called sphere cap nuts are used, whichhave a spherical surface in the contact area with the anchorplate. The anchor plates should have minimum dimensions of 150 x150 x 10 mm.
Special anchor head designs have been developed for the use ofnon-prestressed anchors in squeezing rock. If a certain anchorload is reached, these anchor heads yield, thus avoiding over-stressing of the anchors. Since these anchor types are special-purpose designs adjusted to the individual case, they will not bedealt with here in more detail.
The admissible anchor force is the maximum force the rock bolt ispermitted to be subjected to (maximum tensile force, maximum bondforce or limit creep force) divided by a factor of safety. Themaximum tensile force of the anchor rod is calculated according to
DIN 21521 from the relevant cross-section in connection with thedesign strain limit and the design tension yield stress of theused material, respectively. The maximum bond force between theanchor rod and the rock mass must be determined for the individualcase by pull tests according to DIN 21521. The limit creep forceis the force which leads to the chosen creep rate according to DIN4125 in the pull test.
For glass fiber anchors, attention must be paid to the fact thatthe failure load of the anchor rods is generally far higher thanthe failure load of the anchor head parts.
2.3.2 Load-carrying behavior
In tunneling, rock bolts are installed as single anchors to sup-port individual rock wedges susceptible to sliding, as surface an-choring to support the excavation surface (tunnel walls, tunnelface), and as systematic anchoring to improve the load-carryingcapacity of the rock mass. Surface and systematic anchorings fur-thermore serve the purposes of supporting regions prone to col-lapse and of improving the load transfer between the steel setsand the shotcrete on the one hand and the rock mass on the other.For systematic anchorings mostly fully cemented anchors (SN-anchors) or injection drill bolts (IBO-bolts) are used.
Rock bolts must carry tensile and possibly shear forces. Failureof the anchors due to overstressing may occur. The failure mode ofbond anchors depends on whether they are placed in rock or insoil. In rock, generally the anchor rod breaks before the bondfails. In soil, failure of the mortar over the bond length occursfirst.
Systematic investigations into the mechanism of operation and theload-carrying capacity of cemented steel anchors were first car-ried out by Bjurström (1974). These investigations led to formulaefor the shear resistance of inclined and not inclined anchors.Further relations for the shear resistance were provided by Azuarand Panet (1980). Empirical equations for the shear and tensileresistance were developed by Dight (1983). Spang and Egger (1989)provide formulae to determine the shear resistance contribution T0by which the shear resistance of a discontinuity increases as aconsequence of anchoring. This contribution can be converted intoan "anchor cohesion"
by which the cohesion of the rock mass increases due to a system-atic anchoring. In (2.1) A is the effective area per anchor of asystematic anchoring. Since glass fiber anchors only have verylittle shear resistance, they are not suited for a systematic an-choring to increase the rock mass strength.
Equation (2.1) makes it possible to model a systematic anchoringin FE-analyses by an increase in the cohesion of the rock mass. Asan alternative, individual anchors can be modeled by truss ele-ments, which, however, can only transfer axial forces. The com-bined tension and shear loading of cemented anchors can thereforenot be captured by the common truss elements.
Erichsen and Keddi (1990) and Keddi (1992) report on numericalanalyses of the influence of anchor design and bond between rockbolt and rock mass on the load-carrying behavior of cemented an-chors.
2.4 Advance support
2.4.1 Spiles
In loose rock, steel spiles are used as advance support of theworkspace at the tunnel face in order to limit overbreak and toprotect the miners against falling rock. The spiles are arrangedin the roof area of the tunnel approx. parallel to the tunnel axisin the form of a fan. They are installed before the underlyinground is excavated. The length of the spiles should be at leastthree times the round length in order to enable sufficient over-lap. The spacing between the spiles should not exceed 30 cm. Fig.2.26 shows exemplarily the advance support using spiles at thesidewall adit excavation of the Limburg Tunnel of the new railwayline Cologne – Rhine/Main in longitudinal section and cross-section.
Fig. 2.26: Advance support using spiles (spile umbrella LimburgTunnel, new railway line Cologne – Rhine/Main): a)Longitudinal section A-A; b) cross-section B-B
One distinguishes mortar spiles, driven spiles and injection drillspiles.
Mortar spiles made from rebars are installed with lengths of 3 to6 m. They are designed for a rock mass in which the drillholes arestable without a casing. The rebars (∅ 25 – 28 mm) are installedin mortar-filled drillholes. By the use of a accelerating admix-ture or by choosing an appropriate time of installation, respec-tively, it is ensured that the mortar achieves sufficient strengthby the beginning of the following excavation works. Fig. 2.27shows mortar spiles that were blasted free during excavation. Inthis case they do not fulfill their original purpose, which is tolimit overbreak and to protect the workspace against falling rock.The blasting was not carried out smoothly with respect to therock.
Fig. 2.27: Mortar spiles with lattice girder, blasted free
Driven spiles consist of steel pipes 3 to 6 m long, which aredriven into pre-drilled holes. The diameter of the steel pipes isslightly larger than the diameter of the drillholes. If the steelpipes are designed correspondingly and the rock mass is groutable,a later rock mass improvement using injections is possible. Fig.2.28 shows driven steel pipe spiles installed as advance supportin the Tunnel Deesener Wald of the new railway line Cologne –Rhine/Main.
Injection drill spiles (IBO-spiles) are installed with a length of4 to 12 m. They are used in rock mass conditions where stableholes cannot be drilled and/or the rock is to be strengthened bygrouting with cement suspension at the same time. Depending on thediameter of the anchor rod, the holes for the spiles are drilledwith a diameter ranging from 42 to 76 mm. The spiles consist ofsteel pipes with a lost drill bit, which are used for flushingduring drilling and for grouting the drillhole together with thesurrounding rock mass after the planned depth is reached. Instal-lation and grouting of injection drill spiles are carried out cor-responding to Fig. 2.24.
2.4.2 Pipe umbrellas
If the spiles described in section 2.4.1 do not offer sufficientsupport, pipe umbrellas are used. They are applied mostly in cohe-
sionless, loose ground if streets or structures are undercut withlittle cover. Just as spiles, pipe umbrellas are constructed overa certain part of the circumference of the excavation profile,preceding the excavation. Because of the dimensions of the pipes,pipe umbrellas are far stronger than spile umbrellas and extendfurther in advance of the excavation.
Instead of pipe umbrellas, composite pile and jet grouting umbrel-las are also constructed in connection with the NATM for the ad-vance support of the workspace at the tunnel face. In the follow-ing, only pipe umbrellas will be covered. Examples for compositepile umbrellas and jet grouting column are described in Chapters3.2 and 7. The jet grouting technique is covered in detail inChapter 7.
Two systems for the construction of pipe umbrellas are distin-guished:
- Pipe umbrella with niches (Fig. 2.29),- pipe umbrella without niches (Fig. 2.30).
If the pipe umbrella is constructed from niches, the excavationprofile of the tunnel is widened in the course of the excavation,so that the drill points for the steel pipes of an umbrella arelocated outside of the shotcrete membrane (Fig. 2.29). Thus, thegeometry of the steel sets must be adapted for each round and theniches must at least partially be filled with shotcrete before thesealing and the interior lining are installed.
If the pipe umbrella is constructed without niches, the drillpoints for the steel pipes are located in the tunnel face of thestandard excavation profile (Fig. 2.30). In the course of furtherexcavation, the pipe sections lying within the cross-section ofthe shotcrete membrane must be cut off after each round.
Pipe umbrellas are constructed with lengths of ca. 15 to 30 m witha heavy drill rig (e. g. drill carriage, Fig. 2.31). The outer di-ameter of the steel pipes used varies between 76 mm and 200 mm,while the wall thickness of the pipes ranges from 8 mm to 25 mm.The pipes serves as the casing while the boreholes are drilled,and remain in the soil or rock, respectively, as a structural ele-ment of the pipe umbrella. After the installation of the pipes,the annular gap between the pipes and the rock mass and as far as
possible also the surrounding rock mass itself are grouted withcement suspension from out of the pipes in order to improve thesupport effect of the pipe umbrella. To this end, the pipes areprovided with injection valves at a spacing of 0.5 to 1 m (Fig.2.32). The pipes are subsequently closed with a cover and filledwith suspension or mortar.
Fig. 2.29: Pipe umbrella with niche (Niedernhausen Tunnel, newrailway line Cologne – Rhine/Main): a) Longitudinalsection; b) Detail A
Fig. 2.30: Pipe umbrella without niche (Dernbach Tunnel, newrailway line Cologne – Rhine/Main): a) Longitudinalsection; b) Detail A; c) Detail section 1-1
The boreholes for the pipe umbrella are arranged over a certainportion of the circumference of the excavation profile at a spac-ing of 30 to 50 cm. They are drilled ascending at an angle of ap-prox. 5° with respect to the tunnel axis. The pipes of two pipeumbrellas should overlap by at least 3 m in the longitudinal tun-nel direction (Fig. 2.29 and 2.30).
In soil and in weathered, soft rock the boreholes are generallydrilled with an auger and air and/or water flushing (Fig. 2.33).In rock or for large pipe diameters, however, mostly down-holehammers are used. Here, the drill bit and the pipe are either con-nected by a bayonet joint so that the casing is continuouslypulled into the borehole with the advance of the drill pipe, orthe pipes are pushed hydraulically by the drill rig. After the fi-nal depth is reached, the drill pipe and the drill bit are re-leased from the pipe and pulled out.
Fig. 2.34 shows the construction of a pipe umbrella through thestarting wall of the Dernbach Tunnel of the new railway line Co-logne – Rhine/Main. The starting wall is supported by shotcreteand anchors. In this case, the shotcrete support of the startingwall was drilled through before the pipes for the pipe umbrellawere installed.
2.5 Geotechnical mapping and monitoring
2.5.1 Mapping
Before the excavation of a tunnel, in general only surface expo-sures and boreholes are available to assess the ground. Their re-sults must be extrapolated to the planned tunnel and its surround-ing area. Because of the ensuing uncertainties in assessing theground conditions, mapping of the temporary tunnel face duringconstruction is absolutely necessary. They form an essential basisfor adapting the support to the local conditions.
The extent of a geotechnical mapping of the tunnel face dependsessentially on the available time. A "detailed mapping" includingcomprehensive information on the intact rock, the fabric and thegroundwater can usually only be carried out if the tunnel excava-tion is halted. With routine mappings during the excavation, it isin general only possible to record these properties randomly. Inthis case, however, it is important to recognize and document de-viations or changes in the local conditions.
In a detailed mapping, the following properties should be recordedas far as possible (Wittke, 1990):
- Rock types and rock boundaries (boundaries of layers),
- degree of weathering of the intact rock,
- orientation of the discontinuities (strike and dip angle),
- location, spacing and trace lengths of discontinuities, i. e.of the intersections of the discontinuities with the excavatedrock surface,
- location and shape of visible discontinuities lying at therock surface,
- opening widths of open or filled discontinuities,
- fillings and coatings of discontinuities,
- location, dimensions and properties of large discontinuitiesand faults,
- location of seepage water and quantity of discharge.
Apart from these information, the field protocol should includefurther peculiarities, such as for example information on discon-tinuities which tracings are so closely spaced that they cannot berecorded individually for reasons of time alone. Half-quantitativedescriptions such as "strongly jointed" or "strongly fractured"are sufficient in this context. These descriptions should be com-plemented with information on the magnitude of the spacing in cmor dm. Further, recordings of the unevenness and roughness of thediscontinuities and of possible indications of weathering on thediscontinuity surfaces should be made. The discontinuity fillingsjust as bedding parallel and foliation parallel discontinuitiescan be labeled by special signatures (Wittke, 1990).
During mapping, samples for complementing laboratory tests must betaken, if indicated.
The standard mapping gear consists of
- a geological compass to measure the orientation of disconti-nuities,
- a measuring tape to record the location of the outcrops andthe tracings of discontinuities,
- a rock hammer to make discontinuities visible that are e. g.covered by a thin weathered layer, to loosen rock samples orto sound the rock (clear or dull sound),
- a note pad in which a sketch of the mapping is entered. It ismore appropriate, however, in most cases to use prepared formsfor the mapping (Fig. 2.35).
Fig. 2.36 shows an example of a detailed mapping of a tunnel face.For reasons of clarity, the information on the orientation and na-ture of the discontinuities is omitted here.
Fig. 2.36: Detail mapping of a tunnel face (Hasenberg Tunnel,Stuttgart urban railway)
Examples for mappings during excavation are given in Chapters 3.1,3.3 and 4.2 (see Fig. 3.20, 3.41 and 4.43).
Discontinuity orientations are measured using the geological com-pass (Fig. 2.37). The dip direction αD is read from the compasscircle, while the dip angle β is read from the vertical circle. Thedip direction αD is the angle between the projection of the dipline of the discontinuity onto the horizontal plane and north. αDis correlated with the strike angle α by the relation given inFig. 2.37 (Wittke, 1990).
Fig. 2.37: Measuring the orientation of a discontinuity usingthe geological compass (Wittke, 1990)
The orientation of a discontinuity, given by the angles αD and β orα and β, respectively, can be represented in a polar equal areanet by a point, the so-called pole. A polar equal area net is anequal-area hemispheric projection of the lower half of the refer-ence sphere. The term reference sphere denotes a sphere the equa-tor of which is located in the horizontal plane. The pole is theprojection of the point of intersection of the normal to the dis-continuity through the center of the reference sphere onto the po-lar equal-area net (Wittke, 1990).
Entering the discontinuity orientations measured during one orseveral mappings in the polar equal-area net leads to a so-calledpolar diagram. An example of a polar diagram prepared in this wayand for the grouping of the measured discontinuities into severalsets is shown in Fig. 3.42 (Chapter 3.3.6).
For a great number of measured discontinuity orientations, or ifthe measured discontinuity orientations scatter considerably, astatistical evaluation of the discontinuity orientations may beappropriate. To this end, instead of the poles, areas representingthe angular regions of equal relative frequency of the poles areentered in the polar diagram. These areas are assigned to poledensities. The pole density denotes the number of poles located on1 % of the area of the polar equal area net, relative to the totalnumber of measured discontinuity orientations, in percent (Wittke1990). In this way, the areas with the most frequent orientationscan be identified for the individual discontinuity sets. An exam-ple of the representation of discontinuity orientations by areasof equal pole density is shown in Fig. 5.39 (Chapter 5.2.6).
The mapping results should be summarized for homogeneous areas ina geometric structural model. To this end, first the parametersdescribing the geometry of the discontinuity fabric are determinedon the basis of a statistical evaluation of the measured fabricdata. For each discontinuity set, the mean values of the measureddip and strike angles, of the spacing and, if possible, of thetracings of the discontinuities must be specified. However, sinceespecially the spacing and tracings of discontinuities scatterconsiderably, in addition to the mean values the standard devia-tions of these parameters should be specified as well (Wittke,1990).
In addition to these data, a structural model should also includea description of the properties of individual discontinuities,which form the sets. This description should contain informationon the unevenness and the roughness as well as on the degree ofweathering of the discontinuity surfaces, which can be obtained e.g. from the mapping of profiles. Further, it should be notedwhether the discontinuities are closed or open, and whether and towhich degree they are filled.
Inhomogeneities such as discontinuities with an extent reachingthe magnitude of the dimensions of the tunnel must be describedindividually with respect to their location and orientation.
Finally, the model should be illustrated in a block figure con-taining as many as possible of the referenced parameters (Fig.2.38).
Fig. 2.38: Example of a structural model of a rock mass withthree discontinuity sets (Wittke, 1990)
2.5.2 Monitoring
General
Just as geotechnical mapping, geotechnical measurements representan essential element of the NATM. On the one hand, they serve tomonitor
- the stability of the tunnel and of adjacent structures,
- the deformations in the ground and the displacements on thesurface,
- the loading of the shotcrete membrane,
- the vibrations during heading.
On the other hand, just as the mapping, they form one of the foun-dations for adapting the support measures to the local ground con-ditions. Finally, the interpretation of already existing measure-ment results can and should be used to verify and, if indicated,optimize the dimensioning of the temporary and permanent lining.Stability analyses on the basis of measurement results furthermore
permit to reduce or remove the uncertainties associated with theassumption of the parameters (back analysis) and to capture theinfluences from actual construction.
Geotechnical monitoring comprises:
- Positional surveying,
- leveling,
- convergency measurements,
- extensometer measurements,
- inclinometer measurements,
- stress measurements,
- anchor force measurements,
- vibration measurements,
- water level and water pressure measurements.
Positional surveying and leveling
Displacements of points located at the ground surface and at thetunnel contour (Fig. 2.39) are measured geodetically by levelingand by using a tachymeter.
To measure subsidence, measuring cross-sections with several lev-eling points are installed perpendicular to the tunnel axis (Fig.2.39). In sloping locations the horizontal displacement componentsshould be measured as well. The zero reading should be carried outwhen the tunnel face is still several tunnel diameters away fromthe respective measuring cross-section to capture the subsidencedue to tunneling completely. If settlements are to be expected re-sulting from a lowering of the groundwater table due to tunneling,the zero reading should be carried out before the start of con-struction.
For the measurement of the displacement of points on the tunnelcontour, the measuring locations are installed closely (≤ 1 m) be-
hind the tunnel face. The measuring points are steel bolts ce-mented into the rock. Simple homing boards or triple prisms aremounted onto these bolts. The zero reading is carried out beforethe next round.
Fig. 2.39: Surface leveling perpendicular to the tunnel axisand positional survey of points on the tunnel con-tour
With the tachymeter, not only the vertical but also the horizontaldisplacement components parallel and perpendicular to the tunnelaxis are measured. The measuring points are surveyed three-dimensionally from bench marks by determining the direction, dis-tance and inclination from the bench mark to the measuring pointfor different points in time. The measurement data given in polarcoordinates are converted to Cartesian coordinates. From the dif-
ferences in coordinates from two measurements, the respective dis-placement vectors of the measuring points can be computed (Reikand Völter, 1996).
Fig. 2.40 shows as an example the vertical displacements of meas-uring points on the tunnel contour during the crown heading of thesouthern tube of the Gäubahn Tunnel in Stuttgart at chainage 113 m(see Chapter 3.2). The upper part shows the vertical displacementsof the measuring points versus time. In the lower part the advanceof the tunnel face is shown versus time.
Fig. 2.40: Geodetically determined vertical displacements ofmeasuring points on the tunnel contour during crownheading (Gäubahn Tunnel, southern tube, chainage113 m)
Convergency devices allow the measurement of changes in distanceof points on the tunnel contour.
Fig. 2.41 shows examples for the arrangement of convergency meas-urement sections for a crown heading and a sidewall adit heading.To measure the distance between two measuring points, the measur-ing points are connected by a tensioned measuring tape or a meas-uring wire and the convergency device (Fig. 2.42). The changes indistance can be determined as the difference of the measuredlengths of consecutive measurements (Reik and Völter, 1996).Changes in length in the order of 0.1 mm can be registered withthis technique.
Since convergency measurements interfere with tunnel heading,measurements with a convergency device are rarely carried out anymore in tunneling nowadays. Optical three-dimensional measurementsusing a tachymeter and triple prisms are preferred instead (Fig.2.39).
Fig. 2.41: Convergency measurement cross-sections: a) Crownheading; b) sidewall adit heading
Fig. 2.42: Schematic drawing of a convergency measuring system(Reik and Völter, 1996)
Extensometer and inclinometer measurements
Displacements and relative displacements (extension or compres-sion) in the ground are measured in boreholes with stationarygages or probes.
Displacements parallel to the axis of a borehole are generallymeasured using extensometers (stationary devices) or sliding mi-crometers (borehole probes).
Multiple extensometers are suitable for the determination of rela-tive displacements of measuring points with larger distances. Fig.2.43 shows the setup of a multiple-rod extensometer. Absolute dis-placements can be determined by tying the extensometer head (1 inFig. 2.43) in to a bench mark by leveling, or if the deepest an-chor is a fixed point. Details on displacement measurements with
extensometers are given e. g. in Recommendation No. 15 of the Ger-man Geotechnical Society (DGGT) (Paul and Gartung, 1991).
Fig. 2.43: Setup of a multiple-rod extensometer (Interfels,2000)
With a sliding micrometer, rock displacements can be measured atshort distances along the axis of a borehole. To this end, a plas-tic casing is cemented into a borehole ca. 100 mm in diameter.Measuring marks are fixed in this casing at a spacing of 1 m. Us-ing an inserted probe, the changes in distance between the measur-
ing marks are recorded one after the other (German Rail, Guideline853, DB 1999).
Displacements perpendicular to the axis of a borehole are usuallymeasured using inclinometer probes or stationary inclinometers.
In the case of the inclinometer probe, a plastic pipe with fourguiding grooves is installed in a borehole ca. 100 mm in diameter.With a measuring probe, the changes in inclination with respect tothe borehole axis and thus the displacements perpendicular to theborehole axis can be determined in two directions at right anglesto each other (Fig. 2.44).
Fig. 2.44: Measuring principle of an inclinometer probe:a) Measurement; b) evaluation
Fig. 2.45 shows an example of the arrangement of the measuringpoints in the case of a combined extensometer and inclinometermeasuring cross-section. Since during the undercrossing of themeasuring cross-section by the tunnel none of the measuring pointsconstitutes a fixed point, the measuring points on the ground sur-
face are tied in to bench marks by levelings (L) and positionalsurveys to determine absolute displacements.
Fig. 2.45: Measuring cross-section with extensometers and in-clinometers (constructed from the ground surface)
If the vertical as well as the horizontal displacements in longi-tudinal and transverse direction are to be determined in one bore-hole instead of separate boreholes, a device can be used that al-lows to measure the displacements in three directions perpendicu-lar to each other at the same time. Such a device is e. g. theTrivec probe, a combination of sliding micrometer and inclinometerprobe (German Rail, Guideline 853, DB 1999).
Just as for the positional surveys, the zero reading should becarried out as early as possible before the corresponding measur-ing cross-section is undercrossed.
Stress measurements
Stress measurements are carried out e. g. to assess the loading ofshotcrete membranes and interior linings. Typical arrangements of
pressure cells in a shotcrete membrane are sketched in Fig. 2.46.Here, one measuring cell is arranged tangentially on the outsideof the shotcrete to measure the radial stress (rock mass pres-sure). The other, radially arranged measuring cell, however,serves to measure the tangential stress resulting from the loading(concrete pressure).
Fig. 2.46: Arrangement of pressure cells in a shotcrete mem-brane for the measurement of concrete and rock masspressure (Wittke, 1990)
The stresses measured with the pressure cell arranged tangentiallyto the tunnel contours generally scatter significantly. On the onehand, this is due to the inhomogeneous stress distribution in theground and the comparatively small dimensions of the measuringcells. On the other hand, the excavation process leads to localloosening along the tunnel contour. This exaggerates the non-uniform distribution of the radial stress, which is also referredto as rock mass stress or contact stress. Most of the times,therefore, radial stress measurements yield relevant results onlyin special cases (e. g. swelling rock).
For radially arranged measuring cells, the scatter of the measure-ment results is generally significantly smaller. The tangentialstresses in the shotcrete measured with these cells scatter lessstrongly, since even an inhomogeneous radial loading of the liningresults in a relatively even distribution of the normal thrust andthus also of the tangential stresses. In addition, the dimensionsof the measuring cells are generally in the same order of magni-tude compared to the thickness of a shotcrete membrane. Finally,the absolute value of the tangential stresses is in general manytimes greater than that of the radial loading of a shotcrete mem-brane. Nevertheless, the measured tangential stresses may also benon-uniform, if for example the shotcrete membrane has a greatlyvarying thickness as a consequence of an uneven excavation profile(Wittke, 1990).
Measurements of the tangential stresses in the lining and the ra-dial stresses between lining and rock mass are carried out usinghydraulic valve gages or oscillating chord gages.
The valve gage of Glötzl Co. (Fig. 2.47) is a hydraulic pressurecell, in which a compressive stress develops due to the loading ofthe oil-filled flat jack. The measurement is effected by increas-ing the fluid pressure in a circuit separate from the flat jack,until the fluid circulates back due to a slight deformation of amembrane located between an inner and an outer chamber. The refluxleads to a pressure drop at a pressure gage positioned in the cir-cuit. The maximum stress measured before the pressure drop at thepressure gage can therefore be equated to the loading of the pres-sure cell (Wittke, 1990).
With oscillating chord gages, changes in length of a chord clampedfreely oscillating in the gage are measured. These changes lead toa change in the natural frequency of the measuring chord excitedto oscillate by a direct current impulse. The altered natural fre-quency measured while the oscillation of the chord fades outyields the chord strain. The change in stress is proportional tothe measured change in length (Schuck and Fecker, 1997). Fig. 2.48shows two oscillating chord gages of Geokon Co.
If high stresses must be expected in the ground, the in-situ orprimary stress state in the rock mass is determined. Aside fromthe already mentioned interpretation of displacement measurementsusing numerical analyses, which in general is only possible duringconstruction, stresses can be measured by stress relief overcoringtests, also referred to as stress measurements by the overcoringmethod. In this method, which can only be applied to rock, thestress state in the rock mass is derived from the deformations ofa drill core due to unloading. To this end, closed-form solutionsfor elastic isotropic and elastic anisotropic stress-strain behav-ior of the rock mass are applied (Kiehl and Pahl, 1990). Further,in-situ stresses can also be determined by the hydraulic fractur-ing method or the hard inclusion method in boreholes. Another pos-sibility to measure stresses or stress changes are compensationmeasurements using flat jacks inserted in sawed or drilled slots.
Anchor force measurements
The anchor forces of untensioned anchors are measured in specialcases only. In the case of tendons with a free anchor length,force measuring cells equipped with strain gages or oscillatingchord gages can be installed at the anchor head. Lately, the tech-nique of tension measurement with integrated optical fiber sensorshas been developed that can be used also with fully cemented an-chors.
Vibration measurements
In the case of a smooth blasting excavation, vibrations generallyneed not be considered with respect to the stability of tunnels.They may have an impact, however, on neighboring structures.Therefore, in these cases vibration velocity measurements are car-ried out to verify and ensure that the reference values accordingto DIN 4150 (Parts 2 and 3, 1999) are complied with. Compliancewith these values is controlled primarily by limiting the maximumcharge per ignition step in the case of a smooth blasting excava-tion (Wittke and Kiehl, 2001; DIN 4150, Part 1, 2001).
In the course of the improvement of the federal highway 42 (B 42)between Bendorf and Lahnstein, Germany, the Glockenberg Tunnel wasconstructed on the side of the Rhine river opposite to the city ofKoblenz.
This tunnel enables a non-intersecting access from the B 42 to thePfaffendorf Rhine bridge. The southern portal of the tunnel liesin the direct extension of the Pfaffendorf Bridge. For each direc-tion, the tunnel includes one driving lane, one stopping lane andan emergency sidewalk. In the mountain, the tunnel axis performs a180° turn with a radius of ca. 50 m and then branches out into twosingle-lane tunnel tubes. One of these tubes accesses the B 42 inthe direction of Bendorf, while the other one enables exiting theB 42 coming from Lahnstein into the direction of Koblenz (Fig.3.1).
Fig. 3.1: Glockenberg Tunnel, site plan with analysis cross-sections
The 315 m long Glockenberg Tunnel was constructed by undergroundexcavation starting at the southern portal 3 (Fig. 3.2). Up to lo-cation 220.5 m, the tunnel was excavated with the two-lane cross-section TC 3 shown in Fig. 3.3. From then on the cross-section waswidened (TC 3A) up to chainage 248 m. In this way, the tunnelcross-section TC 3 is transformed into the two smaller tunneltubes with the cross-sections TC 1 and TC 2. The smaller tunneltubes each have a length of approx. 40 m and end at the portals 1and 2 (Fig. 3.1).
Fig. 3.2: Glockenberg Tunnel, portal 3, crown
For the two-lane standard profile TC 3, a mouth-shaped profilewith a width of 17.5 m and a height of 12 m was carried out. Theexcavated cross-section amounts to 175 m2. The clearance is approx.15 m wide and approx. 4.5 m high. The thickness of the shotcretemembrane is 25 cm, while the interior lining made from watertightconcrete is 60 cm thick (Fig. 3.3).
In the vault area, a radius of curvature of R = 8.43 m was se-lected for the interior lining. The sidewalls were constructedwith a radius of R = 4.66 m. With a radius of R = 25.25 m, the in-
Fig. 3.4 shows the vertical section through the tunnel axis in thearea of tunnel cross-sections TC3 and TC3A as a development. Themaximum overburden height amounts to approx. 50 m.
The ground along the alignment of the Glockenberg tunnel was ex-plored using core drillings and test pits. One borehole wasequipped as an observation well (Fig. 3.4).
According to the exploration, an up to 9 m thick soil layer con-sisting of talus material, residual detritus, loess and loess loamis underlain by an alternating sequence of slate and fine- to me-dium-grained sandstones belonging stratigraphically to the LowerDevonian of the Emsian stage (Fig. 3.4).
In the course of the Variscan orogenesis phase, the originallyhorizontal strata series were narrowed in NW-SE direction and bentto folds of different sizes. Within the large-scale folding struc-ture, the alternating sequence of ribboned or flaser-like, locallysecondarily silicated slates with sandstone banks up to severalmeters thick is in parts specially folded and sheared.
The sandstones are mostly quartzitically, locally also calcare-ously and ferrugineously cemented. Quartz and calcite veins milli-meters to centimeters thick often occur in the rock zones rich insand.
According to the results of the petrographic microscopy and X-raydiffractometry investigations, quartz, feldspar, mica and carbon-ates are the main rock-forming components.
The tunnel area is characterized by a distinct tectonic deforma-tion in the form of a "special folding". Shallow as well as steepdipping of the strata to the northwest and southeast, respec-tively, occurs here with changes taking place over very short dis-tances.
The rock is fractured by bedding-parallel discontinuities, the fo-liation and joints. As a result of the folding structure, the ori-entation of the discontinuities varies within the tunnel area.Fig. 3.5 shows the mean discontinuity orientations measured onoriented drill cores.
Fig. 3.5: Glockenberg Tunnel, site plan showing the orienta-tion of the discontinuities and the location of theexpected fault zones
As far as it can be assessed from the surface exposures, the bed-ding-parallel discontinuities must be assumed completely sepa-rated. Due to the exceeding of the shear strength on the bedding-parallel discontinuities as a result of shear during folding,triturated material and slickenside lineations locally exist onthese planes. The same applies to quartz and calcite segregationswhich deposited in the cracks that developed due to dilatancy dur-ing the folding process.
Judging from surface exposures in the surroundings of the tunnel,the extension and the spacing of the foliation discontinuities de-pend distinctly on the appearance of the intact rock. In the caseof greater thickness of the banks and mostly in the clayey rock
zones, individual completely separated foliation parallel discon-tinuities exist. Locally also slickenside lineations and tritu-rated material were found on the foliation parallel discontinui-ties, indicating tectonic loading.
The joints contain many more rock bridges than the bedding- andfoliation parallel discontinuities. Their degree of separation iscorrespondingly smaller.
In some core drillings, locally an increased core fragmentation, amore frequent appearance of slickensides parallel to the beddingor to the foliation, numerous quartz and calcite veins as well asrather strongly disintegrated rock zones were encountered. Thisled to the assumption that a fault zone striking in NE-SW direc-tion in the area of tunnel cross-section TC 3 may be present (Fig.3.5). Another fault zone was assumed in the area of tunnel cross-section TC 2 on the basis of the exploration results (Fig. 3.5).
Beneath the soil cover, which was encountered down to a depth of 9m below ground surface during the exploration as mentioned above,the rock has only slightly been altered by weathering processes.An exception are more strongly fragmented rock zones, e. g. faultzones, in which weathering has clearly progressed.
To investigate the hydrological conditions, the groundwater levelin the observation well OW (see Fig. 3.4) was measured and perme-ability tests were carried out at different depths in the bore-holes.
The observation well was set up in the area in which the highestgroundwater level was expected due to the location on the slope.The measurements showed that the groundwater is encountered hereapprox. 1 m above the tunnel roof (see Fig. 3.4).
3.1.4 Excavation classes
Because of the size of the excavated cross-section of approx.175 m2 (see Fig. 3.3), it was necessary to carry out excavation andsupport separately for crown, bench and invert (Fig. 3.6).
alignment, the orientation of the discontinuities relative to thetunnel axis changed continuously with the heading. Further, faultzones locally had to be expected (see Fig. 3.5). Therefore fre-quently varying rock conditions were assumed for the tender designand the final design. Correspondingly, the excavation classes 4,5, 6A and 7A were specified in the final design in accordance withthe recommendations of the working group "Tunneling" of the GermanGeotechnical Society (DGGT, 1995: Table 1).
A crown heading with an open invert was planned. The excavationclass as well as the trailing distances of the bench excavationand the closing of the invert lining (E and F in Fig. 3.6) were tobe selected in the course of the crown heading on the basis of theresults of mapping and monitoring during construction (see Chapter3.1.6). In this context special emphasis was placed on the resultsof the stability analyses (see Chapter 3.1.5) and of the geotech-nical mapping.
The excavation methods, round lengths and support measures speci-fied for the excavation classes 4, 5 and 6A for tunnel cross-section TC 3 in the final design are listed in Fig. 3.6. The exca-vation class 7A including advance support and support of the tun-nel face was not carried out during construction.
Heading was planned by smooth blasting and/or mechanical excava-tion by a tunnel excavator and hydraulic chisel.
The unsupported round length during crown heading was selected as2.0 m for excavation class 4 and 1.5 m for excavation classes 5and 6A. In excavation class 6A an advance support of the crownwith 4 to 5 m long mortar spiles was planned for every secondround. Excavation class 5 included a sealing of the temporarycrown face with shotcrete (Fig. 3.6).
For the shotcrete membrane, shotcrete of grade B25 with a thick-ness of 25 cm and two layers of reinforcing mats Q295 were speci-fied. In the crown area, a systematic anchoring with 7 SN-anchors(excavation classes 4 and 5) resp. 9 SN-anchors (excavation class6A) per round was planned. In each round a lattice girder was tobe installed in the crown area (Fig. 3.6).
Due to reasons of construction management, a continuous crownheading was aimed for. If a trailing bench and invert should be-
come necessary for stability reasons, the maximum trailing dis-tance E resp. F should amount to two tunnel diameters D (Fig.3.6).
Unsupported round lengths for bench and invert of 4 m for excava-tion class 4 and 3 m for excavation classes 5 and 6A were planned.In the bench area, a systematic anchoring with 12 anchors (excava-tion classes 4 and 6A) resp. 10 anchors (excavation class 5) perround was specified. In each round one lattice girder was to beextended to the bench invert (Fig. 3.6).
The anchors were designed to be 4 to 6 m long in the crown as wellas in the bench for all excavation classes (Fig. 3.6). If faultzones had to be crossed, locally longer anchors were planned to beinstalled.
3.1.5 Stability analyses
In order to investigate the stability and to dimension the shot-crete membrane, two-dimensional FE-analyses were carried out usingthe program system FEST03 (Wittke, 2000). Fig. 3.1 shows theanalysis cross-sections AC 1, AC 2 and AC 3 investigated for thetunnel cross-section TC 3.
Because of the varying orientations of the discontinuities in thetunnel area and the changing direction of the tunnel axis due tothe circular plan of the tunnel, four idealized structural modelswere established. All of them assume that the discontinuitiesstrike parallel to the tunnel axis. They are thus on the conserva-tive side with respect to the stability and the support measuresensuing from the analyses (see Fig. 3.5 and 3.7). Further, the dipangles of the discontinuities were varied over wide ranges (Fig.3.7).
In model A, the bedding (B) and the foliation (F) are assumed dip-ping each at 45° perpendicular to the tunnel axis in opposite di-rections. In addition, a vertical joint set (J) striking parallelto the tunnel is taken into account. In model B the foliation isassumed dipping at 30° while the bedding dips at 60° in the oppo-site direction. Here as well an additional vertical joint set ex-ists. Model C includes a vertical bedding striking parallel to thetunnel and a horizontal foliation. Instead of the vertical beddingplanes, vertical joints are assumed in model D.
Aside from these different structural models, cases with a faultzone located in the area of the tunnel were investigated as well.Fig. 3.8 shows the assumed locations of these fault zones. Posi-tion L assumes a 5 m thick fault zone dipping at 70° close to theleft side of the tunnel. With position M a vertical, 5 m thickfault zone running through the tunnel cross-section is taken intoaccount. Position R assumes a 5 m thick fault zone dipping at 70°close to the right side of the tunnel.
Fig. 3.7: Considered structural models: a) Model A; b) modelB; c) model C; d) model D
In Fig. 3.9 and 3.10 the pseudo three-dimensional FE-mesh (slicewith thickness of 1 m) used for analysis cross-section AC 3 is
shown. The FE-meshes for analysis cross-sections AC 1 and AC 2have a corresponding setup, except for the overburden height.
Fig. 3.8: Accounting for fault zones
The selected computation section is 140 m wide in transverse di-rection of the tunnel (x-direction). The height amounts to 120 m.The FE-mesh consists of 3288 three-dimensional, isoparametric ele-ments with 7777 nodes (Fig. 3.9). The boundary conditions consistof vertically sliding supports for the nodes on the verticalboundary planes and of horizontally sliding supports for the nodeson the lower boundary plane (z = 0). All nodes are fixed in the y-direction.
Fig. 3.9: Analysis cross-section AC 3, FE-mesh, boundary con-ditions, ground profile and parameters (referencecase)
The setup of the ground layers is the same for all analysis cross-sections. The rock mass is encountered below a 5 m thick surfacesoil layer. In some analyses a fault zone in one of the positionsshown in Fig. 3.8 is modeled discretely.
The parameters assumed in the stability analyses for the soil andthe fault zone correspond to the specifications established during
the tender design. The rock mechanical parameters were specifiedin accordance with the parties concerned.
Fig. 3.10: Analysis cross-section AC 3, FE-mesh, detail
For the reference case, Young's modulus of the rock mass was as-sumed conservatively as E = 1000 MN/m2. In comparative analyses avalue of 1500 MN/m2 was specified. Poisson's ratio was assumed as0.33 (Fig. 3.9).
The angles of friction on the bedding-parallel and foliation par-allel discontinuities were assumed as ϕb = ϕf = 24°. For the fric-tion angle on the joints a value of ϕj = 24° was selected (Fig.3.9).
For the bedding parallel discontinuities no cohesion was assumedin all cases. The same applies to the foliation parallel disconti-nuities for the reference case (Fig. 3.9). In comparative analysesa cohesion of cf = 100 kN/m2 resp. cf = 200 kN/m2 was specified. Thecohesion on the joints was selected as cj = 150 kN/m2 in the refer-ence case (Fig. 3.9). In the comparative analyses it was assumedas cj = 200 kN/m2 and cj = 300 kN/m2, respectively.
The transfer of tensile stresses perpendicular to the discontinui-ties was excluded in all analysis cases.
A value of 15000 MN/m2 was specified as statically effectiveYoung's modulus of the shotcrete. This value is to include thehardening during the application of the load (see Chapter 2.1.5).As mentioned before, the thickness of the shotcrete membrane wasselected as t = 25 cm (Fig. 3.10, see also Fig. 3.3 and 3.6).
The analyses were based on elastic-viscoplastic stress-strain be-havior (Wittke, 2000) of the ground and elastic stress-strain be-havior of the shotcrete membrane.
According to the findings of the exploration, locally a ground wa-ter table lying above the tunnel roof had to be expected. With thedrainage during tunnel excavation, the ground water table is low-ered to the level of the tunnel invert. There was no water pres-sure therefore to be considered in the stability analyses for theshotcrete membrane.
Fig. 3.11 shows a schematic representation of the computationsteps for the simulation of the construction stages "crown excava-tion" and "bench excavation" and the excavation of the full cross-section.
The first computation step includes the computation of the stateof stress and deformation resulting from the dead weight of theground (in-situ state).
In the computation steps preceding the simulation of the individ-ual partial excavations (computation steps 2, 4 and 6), a preced-ing stress relief is accounted for in the respective cross-sectionarea to be excavated (Wittke, 2000).
To this end, Young's modulus in this area is reduced to the valueEred = av · E where E denotes Young's modulus of the ground andav < 1.0 is the so-called stress-relief factor. The precedingstress relief enables an approximate representation of the defor-mations and stress redistributions occurring in the rock precedingthe tunnel excavation and the installation of the support. Due tothe preceding stress relief, the shotcrete membrane installed inthe following computation step is subjected to a smaller loadingcompared to a simultaneous simulation of excavation and installa-tion of the shotcrete support without the intermediate step of thepreceding stress relief. The stress relief factor is specified onthe basis of experience gained from measurements on completedstructures and comparisons with the results of three-dimensionalanalyses. In the present case, the preceding stress relief factorwas chosen as av = 0.5 (Fig. 3.11).
In addition to the installation of the shotcrete membrane, the in-stallation of the SN-anchors in crown and bench for excavationclass 6A (see Fig. 3.6) is simulated in computation steps 2 (crownexcavation and support) and 4 (bench excavation and support). Theanchors are modeled by 4 m long truss elements (Wittke, 2000; Fig.3.11).
In the following, the results of stability analyses for analysiscross-section AC 3 (see Fig. 3.1, 3.9 and 3.10) are shown exem-plarily.
Analysis cross-section AC 3 is located in the area of the maximumoverburden of approx. 50 m (Fig. 3.9). In this area the occurringdiscontinuity orientations can be captured approximately withstructural models B and C (see Fig. 3.7).
The construction stage after the bench excavation (5th computationstep) proves to be critical with respect to the stability of thetunnel, if structural model B and the parameters given in Fig. 3.9are assumed. If the anchors are not taken into account, the dis-placements do not converge in the course of the viscoplastic it-erative calculation in the 5th computation step, i. e. the stabi-lity of the tunnel cannot be proven by the analysis for this con-struction stage. Even if the anchors are taken into account, con-siderable viscoplastic displacements still result in this computa-tion step, especially at the left bench lining toe. These dis-placements converge, however, in the course of the viscoplastic
iterative calculation. Fig. 3.12 and 3.13 show the principal nor-mal stresses, the areas where the shear strength on the disconti-nuities is exceeded and the displacements of the excavation pro-file computed for the bench excavation with the anchors taken intoaccount. The computed roof subsidence amounts to approx. 40 mm,the maximum heave of the invert to approx. 65 mm (Fig. 3.13).
Fig. 3.12: Analysis cross-section AC 3 (structural model B),principal normal stresses and exceeding of strength,bench excavation (5th computation step)
If structural model C is assumed, the stability of the tunnel canbe proven in the analysis for all construction stages even if thesystematic anchoring is not taken into account. Also, if a greater
shear strength is assumed on the discontinuities, the computeddisplacements become considerably smaller.
Fig. 3.13: Analysis cross-section AC 3 (structural model B),displacements, bench excavation (5th – 1st computa-tion step)
Fig. 3.14 shows the bending moments and normal thrust in the shot-crete membrane calculated for the 5th computation step. For asafety factor of 1.35, the design results in a statically requiredreinforcement of up to 4.0 cm2/m. If the lattice girders are takeninto account, this reinforcement is covered by the planned rein-forcement (see Fig. 3.6).
Fig. 3.14: Analysis cross-section AC 3 (structural model B),stress resultants in the shotcrete membrane, benchexcavation (5th computation step)
In Fig. 3.15 the tensile anchor forces computed for the truss ele-ments for the 7th computation step are shown. They reach a maximumvalue of approx. 130 kN.
Fig. 3.15: Analysis cross-section AC 3 (structural model B),tensile anchor forces, 7th computation step
The admissable anchor force of the installed anchors adm FA resultsfrom comparing the anchor force divided by the cross sectionalarea As with the tension yield point of the anchor steel βs takinginto account a factor of safety of η = 1.75. SN-anchors made fromBSt 500S (βs = 500 N/mm2) with a diameter of 25 mm were selected(see Fig. 3.6). This leads to the admissable anchor force:
kN140A1
Fadm SSA =⋅β⋅η
= (3.1)
This value is not exceeded in the analysis (see Fig. 3.15).
With the computation of the anchor forces it must be taken intoaccount that the placement of the anchors is simulated simultane-ously with the excavation and the installation of the shotcretemembrane (see Fig. 3.11). Thus the anchor loads are overestimatedin the analyses, since in reality the anchors are placed after theinstallation of the shotcrete membrane. To neutralize this effect,a Young's modulus smaller than the one of steel was assumed forthe anchors in the analyses.
Fig. 3.16: Analysis cross-section AC 3 (structural model B),fault zone to the left of the tunnel, principal nor-mal stresses and exceeding of strength, bench exca-vation (5th computation step)
Due to the uncertainties involved with this approach regarding thecomputed anchor forces and stress resultants of the shotcrete mem-brane, a comparative analysis was carried out, in which the shearresistance of the anchors was converted into an equivalent cohe-sion on the discontinuities. This cohesion was assumed within theanchored area (see Chapter 2.3.2). The results of this analysisconfirmed the reinforcement determined before and thus indirectlyalso the computed anchor forces.
Fig. 3.16 to 3.19 show the results of an analysis based on the pa-rameters given in Fig. 3.9 and 3.10, structural model B and in ad-dition a 5 m thick fault zone dipping at 70° close to the leftside of the tunnel (see Fig. 3.8, position L). For these unfavor-able assumptions the stability of the construction stage after thebench excavation (5th computation step, see Fig. 3.11) can only beproved in the analysis, if in the area of the left half of thetunnel 10 to 15 m long IBO-bolts (see Chapter 2.3.1) reaching be-hind the fault zone into the undisturbed rock mass are simulatedby truss elements.
The principal normal stresses, the exceeding of strength and thedisplacements computed for this case for the 5th computation stepare shown in Fig. 3.16 and 3.17. If the long anchors are simu-lated, the displacements of the excavation profile are of the samemagnitude as for the corresponding case without a fault zone (seeFig. 3.13 and 3.17).
Fig. 3.17: Analysis cross-section AC 3 (structural model B),fault zone to the left of the tunnel, displacements,bench excavation (5th – 1st computation step)
Fig. 3.18: Analysis cross-section AC 3 (structural model B),fault zone to the left of the tunnel, stress resul-tants of the shotcrete membrane, bench excavation(5th computation step)
Fig. 3.18 shows the bending moments and normal thrust in the shot-crete membrane determined for the 5th computation step. Compared tothe corresponding case without a fault zone, greater maximum bend-ing moments and smaller maximum normal thrusts occur in the shot-crete (see Fig. 3.14 and 3.18). For a safety factor of 1.35, thedesign results in a statically required reinforcement of up to6.2 cm2/m. If the lattice girders are taken into account, this re-inforcement is covered by the planned reinforcement as well.
Fig. 3.19: Analysis cross-section AC 3 (structural model B),fault zone to the left of the tunnel, tensile anchorforces, 7th computation step
In Fig. 3.19 the tensile anchor forces computed for the truss ele-ments in the 7th computation step are given. Only the anchors run-ning through the fault zone are loaded. The maximum computed an-chor force amounts to 150 kN, which is only slightly more than theadmissible anchor load according to (3.1) of the planned IBO-rods(∅ 25 mm) of 140 kN.
3.1.6 Crown heading and monitoring results
Geotechnical mappings of the crown face were carried out in regu-lar intervals during the heading. Special emphasis was placed onrecording the properties, extent and orientation of the disconti-nuities (Fig. 3.20). No fault zones were encountered in the tunnelcross-section.
Further, in the area of tunnel cross-section TC 3 ca. 20 measuringcross-sections with gage bolts were installed to monitor the dis-placements caused by the heading. In tunnel cross-sections TC 1and TC 2 further measuring cross-sections were installed.
As already mentioned, the excavation classes were determined onthe basis of the results of the stability analyses as well as themappings and the displacement measurements. Fig. 3.21 shows exem-plarily the maximum roof subsidence measured during crown headingin the area of tunnel cross-sections TC 3 and TC 3A. With the ex-ception of one measurement in the portal area (δR = 16 mm), themeasured values range from 1 to 7 mm.
In the area of the three analysis cross-sections AC 1, AC 2 and AC3, the measured roof subsidence can be compared with the resultsof the stability analyses. With this comparison it must be takeninto account, however, that the displacements that occurred in themeasuring cross-sections already before the zero reading cannot bemeasured. Since the elastic part of the displacements mostly oc-curred before the zero reading, the measured roof subsidence δR canapproximately be compared with the viscoplastic part of the dis-placements computed for the roof vp
Rδ . It becomes apparent that inthe area of analysis cross-sections AC 2 and AC 3 the measureddisplacements are smaller than the viscoplastic parts of the dis-placements computed with no cohesion assumed on the foliation-parallel discontinuities (cf = 0, see Fig. 3.9).
If a value of cf = 100 kN/m2 is chosen for the cohesion on the fo-liation parallel discontinuities, the measured roof subsidencecorresponds to the computed viscoplastic part of the displace-ments. For analysis cross-section AC 1, cf = 0 yields good agree-ment of the measured roof subsidence with the computed viscoplas-tic part of the displacements (Fig. 3.21).
The results of the geotechnical mapping of the tunnel face and thedisplacements measured during the crown heading allowed in consid-eration of the results of the stability analyses the tunnel to bedriven with excavation classes 4, 5 and 6A (see Fig. 3.6). Excava-tion class 7A was not carried out. Further, it was possible to ex-cavate the crown over the entire tunnel length without simultane-ously trailing bench.
3.1.7 Conclusions
The Glockenberg Tunnel is located in an alternating sequence ofslates and sandstones characterized by a tectonic deformation dueto a "special folding". The strength and deformability of the rockmass is predominantly determined by the discontinuities (beddingparallel discontinuities, foliation parallel discontinuities andjoints). As a consequence of the folding structure, the orienta-tion of the discontinuities is not uniform. In addition, thestrike directions of the discontinuities with respect to the tun-nel axis vary due to the almost circular tunnel alignment. Faultzones had to be expected locally as well.
Therefore, different structural models had to be assumed for thestability analyses. The comparison of the displacements measuredduring tunneling with the values computed for the different casesallowed to optimize the support measures and to apply economicalexcavation classes. It became apparent that the analyses repre-sented an essential contribution to an economical construction ofthis tunnel.
The Gäubahn Tunnel is part of the new construction of the federalhighway B 14 between the Schattenring intersection and the Süd-heimer Square providing a northern bypass of the city district ofVaihingen in Stuttgart, Germany. It undercrosses a railway line(Gäubahn) and buildings adjacent to the Rudolf-Sophien institu-tion. During the heading underneath the railway track railwaytraffic had to be maintained. Specific tunnel support measureswere therefore planned for this section.
3.2.2 Structure
The Gäubahn Tunnel consists of two parallel ca. 300 m long tunneltubes with two lanes each (Fig. 3.22). Approx. 270 m of each tun-nel tube were driven by underground construction ascending fromthe eastern portal. The precuts in the portal areas were con-structed by the cut-and-cover method (Fig. 3.22 and 3.23).
The maximum overburden amounts to approx. 20 m. In the area of theundercrossing of the Gäubahn an overburden of approx. 4 m exists(Fig. 3.23).
Fig. 3.24 shows the 11.5 m wide and 8.85 m high standard profileof a tunnel tube. The excavated cross-section amounts to approx.93 m2. The shotcrete membrane has a concrete grade of B25 and athickness of 25 to 30 cm. The interior lining consists of water-tight grade B35 concrete with a thickness of 45 cm.
The roof and the invert are shallowly rounded with radii of curva-ture of R = 8.15 m and R = 10.45 m, respectively. At the sidewallsthe radius of curvature amounts to R = 5.45 m. The transitionsfrom the sidewalls to the roof and to the invert, respectively,were designed with comparatively small radii of R = 3.06 m andR = 2.65 m, respectively (Fig. 3.24).
The two tunnel tubes were constructed by advancing crown headingwith trailing bench and invert excavation. The cross-section iscorrespondingly divided into crown and bench/invert (Fig. 3.24).
Fig. 3.25 is an aerial photograph of the eastern precut with thebuildings of the Rudolf-Sophien institution after the excavationand support of both tunnel tubes. The picture shows the open-cutarch sections (cut-and-cover construction) in the transition fromunderground excavation to the portal blocks not yet built in therepresented construction phase.
Fig. 3.25: Gäubahn Tunnel, eastern precut and Rudolf-Sophieninstitution
3.2.3 Ground and groundwater conditions
Below the ground surface a few meters thick overlying strata arepresent (Fig. 3.23). They consist mostly of rock weathered intosand and silt, respectively, and partially relocated.
The rock mass underneath belongs stratigraphically to the Stuben-sandstone formation consisting of a sequence of sandstones andsiltstones. The sandstone and siltstone layers are up to severalmeters thick. Locally, however, also narrowly spaced alternatingsequences occur (Fig. 3.23).
The sandstones as well as the siltstones are divided into banks bybedding planes. The thickness of these banks ranges between fewcentimeters and 1 to 2 m. The bedding planes are approximatelyhorizontal.
The joints are generally steeply dipping and deviate only littlefrom the direction perpendicular to the bedding (here: the verti-cal direction). In general they extend over many meters horizon-
tally. At least in some areas the joints are slightly opened ac-cording to the exploration results.
In the siltstone layers slickensides were found as well in addi-tion to the bedding planes and joints. These slickensides are ran-domly oriented and have mostly smooth surfaces.
In the middle tunnel section two boreholes were equipped as obser-vation wells. According to their readings, the groundwater tablewas located at the level of the two tunnel cross-sections (Fig.3.23). In two observation wells located close to the tunnel por-tals no groundwater was encountered.
3.2.4 Excavation classes
The excavation of the two tunnel tubes was planned mostly as anadvancing crown heading with open invert and trailing bench andinvert excavation following excavation class 4A.
The excavation sequence, excavation method, round lengths and sup-port measures for excavation classes 4A-1, 4A-2 and 4A-3 (seeDGGT, 1995: Table 1) are given in Fig. 3.26. These excavationclasses differ with regard to in the unsupported round length, thelattice girder spacing and the number of anchors per round. Theshotcrete membrane is reinforced inside and outside with steelfabric mats Q295. If necessary, the tunnel face was planned to besealed with plain shotcrete. The trailing distance of bench (D)and invert (E) was to be determined depending on the monitoringresults and the geotechnical conditions encountered during theheading as well as depending on the results of the stabilityanalyses.
The tunnel was excavated using a tunnel excavator and, in some ar-eas, also by smooth blasting. Measurements carried out on build-ings showed that the vibration velocities remained far under thereference values for the admissible structural vibrations given inDIN 4150, part 3 (DIN 4150-3, 1999).
In the area of the undercrossing of the Gäubahn a crown headingwith closed temporary invert with the excavation classes 6A-K and7A-K, respectively, was planned (see DGGT, 1995: Table 1) in orderto limit the tunneling-induced subsidence (Fig. 3.27). These exca-vation classes include an advancing support using so-called com-posite pile umbrellas. These consist of 8 m long piles with a di-ameter of 40 mm which were grouted in 116 mm boreholes using ce-ment based suspension and are slightly inclined against the hori-zontal in the longitudinal section (Fig. 3.23 and 3.27). In eachround one layer of composite piles (composite pile umbrella) wasconstructed. If required, spiles were planned to be installed inaddition. In excavation class 7A-K, the tunnel face was in addi-tion to be rounded out and supported with shotcrete (Fig. 3.27b).
3.2.5 Stability analyses for the design of the shotcretesupport
For the design of the shotcrete membrane, two-dimensional analyseswith the program system FEST03 (Wittke, 2000) were carried out.
The investigation comprised the analysis cross-sections AC 1 to AC4 shown in Fig. 3.23. Fig. 3.28 shows exemplarily the FE-mesh, theboundary conditions, the ground profile and the parameters usedfor analysis cross-section AC 2. Analysis cross-section AC 2 islocated in the area of the undercrossing of the buildings adjacentto the Rudolf-Sophien institution. The overburden in this areaamounts to approx. 20 m (see Fig. 3.23).
The 52 m wide, 66 m high and 1 m thick computation section in theform of a slice is discretized by an FE-mesh with 2703 iso-parametric elements and 13956 nodes. To simplify matters, only onetunnel tube is modeled. The plane of symmetry lies in the middleof the rock pillar between the two tunnel tubes. Thus the simulta-neous excavation of both tunnel tubes is simulated in the analy-ses. In reality, the crown excavation of one tunnel tube precedesthe other one. The simulation of a simultaneous heading of bothtubes however is on the safe side with regard to the loading ofthe rock mass and the stability of the openings. For the nodes ofthe vertical boundary planes vertically movable supports are in-troduced as boundary conditions, whereas for the nodes on thelower boundary plane (z = 0) horizontally movable supports arespecified (Fig. 3.28). All nodes are fixed in the y-direction.
Fig. 3.28: Analysis cross-section AC 2, FE-mesh, boundary con-ditions, ground profile and parameters
Four sandstone and siltstone horizons each of the alternating se-quence of the Stubensandstone formation were modeled. The posi-tioning of the siltstone horizon in the roof area is to be as-sessed unfavorable with respect to the stability of the tunnel,
since the strength of the discontinuities is considered smaller inthe siltstone than in the sandstone (see Fig. 3.28). At middletunnel level a sandstone layer is modeled which is followed by asiltstone layer in the area of the tunnel invert. The groundwatertable was assumed at tunnel invert level due to the drainage dur-ing construction. Below the groundwater table the rock mass issubjected to hydrostatic uplift.
In the tender documents ranges for the soil and rock mechanicalparameters are given. The stability analyses for the design of theshotcrete support are based on the most unfavorable values (Fig.3.28).
Fig. 3.29: Analysis cross-section AC 2, FE-mesh, detail
The intact rock strength of the sand- and siltstones was assumedinfinitely high as a simplification. The rock matrix was assignedelastic stress-deformation behavior. This simplification is justi-fied, because the strength on the discontinuities is markedlysmaller than the intact rock strength. Two discontinuity sets weretaken into account in the analyses. The horizontal bedding and avertical joint set striking parallel to the tunnel axis with anunfavorable effect on tunnel stability.
At and underneath the tunnel invert, the rock mass is unloadedduring the tunnel excavation. Previous experience has shown thatthe modulus relevant for the unloading of the rock mass is higherthan the loading modulus assumed as E = 300 MN/m2. Correspondingly,the elements in and underneath the invert area were assigned anunloading modulus of EU = 1000 MN/m2 (see Fig. 3.29).
The statically effective Young's modulus of the shotcrete wasspecified as 15000 MN/m2 taking into account the hardening duringthe application of the load (Fig. 3.29).
The excavation and support of the tunnel were simulated in fivecomputation steps (Fig. 3.30). The first computation step com-prises the determination of the state of stress and deformationresulting from the dead weight of the ground (in-situ state). Withcomputation steps 2 and 3 the crown excavation and its support us-ing shotcrete are simulated. To this end, Young's modulus of theelements in the crown was reduced in the 2nd computation step tothe value Ered = av · E with av = 0.5. av is the so-called stressrelief factor already mentioned in Chapter 3.1 (Wittke, 2000). Theexcavation and support of the crown follow in the 3rd computationstep. In the 4th and 5th computation steps the excavation and sup-port of the remaining cross-section are simulated correspondingly.
Fig. 3.31 shows the displacements due to the crown excavation (3rd
– 1st computation step) calculated for the excavation contour andthe ground surface. The roof subsidence amounts to approx. 18 mm,the invert heave is approx. 8 mm and the maximum subsidence at theground surface results to approx. 11 mm.
The computed bending moments and normal thrust in the shotcretemembrane are shown for the 3rd computation step in Fig. 3.32. Thedesign of the shotcrete support for this construction stage yieldsthat no reinforcement is required with respect to bending and nor-mal thrust for a factor of safety of 1.35. Steel fabric mats Q295were placed inside and outside as a minimum reinforcement (seeFig. 3.26).
After the bench and invert excavation in the 5th computation stepthe displacements increase only slightly. For this constructionstage as well, no reinforcement is statically required for a fac-tor of safety of 1.35.
In both tunnel tubes the crown heading was far ahead of the benchand invert excavation. Fig. 3.33 shows exemplarily the ground sur-face subsidence due to the crown heading in the area of the under-crossing of the Gäubahn and the buildings adjacent to the Rudolf-Sophien institution. In the area of the undercrossing of thebuildings adjacent to the Rudolf-Sophien institution the crownheading was carried out with an open invert and small roundlengths of 0.8 to 1.2 m following excavation class 4A-1 (see Fig.3.26). The subsidence measured here ranges from 7 to 19 mm. In thearea of the undercrossing of the Gäubahn the heading was changedover to a crown heading with closed invert under the protection ofan advancing support using composite piles and spiles (see Fig.3.27). The subsidence could thus be limited to 2 to 8 mm. Thissmall subsidence did at no point affect the railway traffic.
The measured ground surface subsidences are in good agreement withthe ones determined in the stability analyses. In the area of theundercrossing of the buildings adjacent to the Rudolf-Sophien in-stitution (analysis cross-section AC 2) a maximum surface subsi-dence of 11 mm was computed (see Fig. 3.31). The maximum surfacesubsidence computed for the undercrossing of the Gäubahn (analysiscross-section AC 3, see Fig. 3.23) amounts to 7 mm.
3.2.7 Conclusions
With a crown heading with open invert and small round lengths (ex-cavation class 4A-1, see Fig. 3.26) it was possible to achievesmall surface subsidence < 2 cm during tunneling in the alternat-ing sequence of sandstone and siltstone horizons of the Stuben-sandstone formation in the region of Stuttgart.
In order to avoid affecting the railway traffic during the under-crossing of the Gäubahn it was necessary in this area to limit thesubsidence to even smaller values. This was achieved mainly bysupporting the curved temporary crown invert with shotcrete. Forsafety reasons an additional advancing support with compositepiles was carried out. With a very small surface subsidence rang-ing from 2 to 8 mm it was possible to maintain the railway trafficwithout any interference.
In areas where the rock mass could not be excavated by a tunnelexcavator smooth blasting was carried out. It was possible toprove by measurements that the vibration velocities fell well be-
low the reference values for the allowable structural vibrationsgiven in DIN 4150, Part 3 (DIN 4150-3, 1999).
The results of the geotechnical monitoring during construction arein good agreement with the surface subsidence computed in the sta-bility analyses. It could thus be shown that FE-analyses are agood basis for estimating the subsidence due to tunneling, if theparameters are assessed appropriate and the excavation and the in-stallation of the support are simulated realistically.
3.3 Hellenberg Tunnel, Germany
3.3.1 Introduction
The alignment of the new railway line Cologne-Rhine/Main runsthrough the Rhine schist mountains (Rheinisches Schiefergebirge)in NW-SE direction tightly bundled with the freeway (Autobahn) A3.Coming from Cologne, it crosses in particular the Siebengebirge,Westerwald and Taunus mountains. The alignment comprises more than30 tunnels with a total length of approx. 47 km. The HellenbergTunnel is one of six tunnels driven in the Taunus mountains andlies to the south of Idstein.
3.3.2 Structure
The new railway line Cologne-Rhine/Main of German Rail (DeutscheBahn AG) undercrosses the Hellenberg mountain in a 552 m long tun-nel. The maximum overburden of the tunnel amounts to approx. 19 m.
The tunnel sections in the portal areas were constructed by thecut-and-cover method. The central tunnel segment with a length ofapprox. 470 m was driven by underground construction ascendingfrom southeast to northwest (Fig. 3.34).
A mouth-shaped profile with an excavated width of 15.4 m and aheight of 12.3 m was constructed for the double-tracked HellenbergTunnel (Fig. 3.35). This is the standard profile for double-Tracked tunnels of the new railway line Cologne-Rhine/Main. Theexcavated cross-section amounts to approx. 150 m2 (Fig. 3.35).
Fig. 3.35: New railway line Cologne-Rhine/Main, standard pro-file
In the vault area the radius of curvature of the shotcrete mem-brane amounts to R = 7.32 m. The invert is rounded with a radiusof R = 16.5 m. For the transition area from the sidewalls to the
invert radii of R = 4.38 m and R = 2.5 m, respectively, were se-lected (Fig. 3.35).
Fig. 3.36 shows the southern portal at the start of undergroundexcavation of the Hellenberg Tunnel.
Fig. 3.36: Hellenberg Tunnel, southern portal
3.3.3 Ground and groundwater conditions
The ground along the alignment of the Hellenberg Tunnel was ex-plored by test pits and core drillings. The boreholes wereequipped as observation wells (Fig. 3.34 and 3.37).
The Quaternary talus material, which is loamified in the upper re-gion, extends to a depth of approx. 2 m. In the portal areas it isencountered down to a maximum depth of 5 m.
Below the Quaternary cover Devonian rocks follow composed in thearea of the Hellenberg Tunnel of phyllitic slate with sandstoneand quartzite intercalations and embedded conglomerate lenses.These layers are referred to as Variegated Schist.
The rock mass in the area of the Hellenberg Tunnel is character-ized by a deep-reaching weathering extending to a depth of some35 m. Unweathered rock mass is thus only encountered below thetunnel's invert.
In the course of the exploration the drill cores and the test pitswere geotechnically mapped. Further, television probing and labo-ratory and in-situ tests were carried out.
From the northern tunnel portal the groundwater table rises to12 m above the tunnel roof. In the following, it drops towards thesouthern tunnel portal to below the tunnel's invert. The groundwa-ter table mapped in the tunnel's longitudinal section (Fig. 3.37)is based on the highest groundwater levels measured in the bore-holes equipped as observation wells.
3.3.4 Excavation classes
Because of the size of the excavation profile it was necessary tosubdivide excavation and support into crown, bench and invert(Fig. 3.35). An advancing crown excavation was carried out.
A prediction of the distribution of the excavation classes basedon the results of the exploration and of stability analyses and onexperience is shown in Fig. 3.37. According to the drilling re-sults, the strongly weathered rock extends in the portal areas toabout the tunnel's invert. Accordingly, excavation classed with atemporary crown invert support were planned here (6A-K1 and 7A-K1,see DGGT, 1995: Table 1). In the central tunnel section it was ex-pected that the crown heading can be carried out with an open in-vert. In parts excavation classes without advance support measures(4A-2 and 4A-3, see DGGT, 1995: Table 1) were predicted. For othersections it was assumed that the heading requires advance support(excavation classes 6A-1 and 6A-2, see DGGT, 1995: Table 1).
The excavation methods, round lengths and support measures plannedfor excavation classes 6A-1 and 6A-2 are shown in Fig. 3.38 and3.39.
It was planned to carry out the excavation mainly by tunnel exca-vators and hydraulic chisels. In areas in which the tunnel was lo-cated in slightly weathered and slightly jointed rock, local loos-ening blasting was planned.
The unsupported round lengths during the crown excavation werespecified in excavation class 6A-1 as 0.80 to 1.20 m and in exca-vation class 6A-2 as 1.21 to 1.60 m (Fig. 3.38 and 3.39). In bothexcavation classes an advance support of the crown with 3 to 4 mlong mortar spiles was provided for (Fig. 3.38 and 3.39). Shot-crete of grade B25 was selected for the shotcrete membrane at athickness of 25 cm and reinforced inside and outside by steel fab-ric mats Q221 (Fig. 3.38 and 3.39). In the sidewall area a system-atic anchoring with at least twelve 4 m long SN-anchors per tunnelmeter was planned (Fig. 3.38 and 3.39).
A possible installation of anchors in the roof area ought to bedecided on site. For each round one steel set (lattice girder) wasto be installed in the crown area (Fig. 3.38 and 3.39).
The trailing distance of the bench (section D in Fig. 3.38) andthe closing of the invert (section E in Fig. 3.38) was to bespecified depending on the results of monitoring during heading,the encountered geotechnical conditions and the results of thestability analyses.
The unsupported round lengths were specified in excavation class6A-1 for the bench and the invert as 1.60 to 2.40 m and ≤ 3.60 m,respectively. In excavation class 6A-2 the corresponding roundlengths amounted to 2.41 to 3.20 m and ≤ 4.80 m, respectively(Fig. 3.38).
To improve the stability, the shotcrete membrane was to be widenedto t = 60 cm in the area of the crown and bench support feet ifnecessary. It is shown calculational in Wittke et al. (1986), Wit-tke (1990) and Wittke (1998), however, that the stability of acrown heading with open invert can only slightly be improved bythese kind of measures.
3.3.5 Crown heading
The excavation classes were finally specified during the crownheading on the basis of the results of stability analyses, ofcrown face mappings (see Chapter 3.3.6) and of the monitoring re-sults.
The crown was excavated over the entire tunnel length without atrailing bench excavation (unlimited length of section D in Fig.3.38). In the area of the southern portal a crown heading with atemporary support of the invert was carried out over a length ofapprox. 80 m (excavation class 7A-K1, Fig. 3.40). At the northernportal a temporary support of the invert was only constructed inthe beginning of excavation (excavation class 6A-K1, Fig. 3.40).
In the remaining area a crown heading with open invert was carriedout (excavation classes 6A-1 and 6A-2, Fig. 3.40).
The evaluation of the crown face mapping resulted in a modifiedelevation of the boundary between strongly and slightly weatheredrock as compared to the exploration results (Fig. 3.37 and 3.40).Nevertheless, the predicted distribution of excavation classesagreed well with the construction (Fig. 3.37 and 3.40).
During the tunnel face mapping the appearance, the extent and theorientation of the discontinuities were determined as well.
Fig. 3.41 shows as an example the crown face mapping at chainage345.4 m. The strike and dip angles of the discontinuities weremeasured using a geological compass (see Chapter 2.5.1). For rea-sons of clarity of the representation only some of the measureddiscontinuity orientations are mapped in Fig. 3.41.
In Fig. 3.42 all discontinuity orientations measured from chainage300 to 350 m are shown in a polar diagram (see Chapter 2.5.1). Ac-cordingly the discontinuities can mainly be assigned to threesets.
The bedding planes and foliation discontinuities display the NE-SWstrike typical for the Rhine schist mountains and dip at a moder-ate steep to steep angle (50 to 80°) towards northwestern direc-tions. Two joint sets J1 and J2 further exist dipping mostlysteeply (60 to 90°) and striking parallel to the tunnel axis (Fig.3.42).
Fig. 3.43 shows the bench at the southern portal. The photographgives an impression of the strongly weathered rock in this areaand of the discontinuity fabric.
Within the tunnel sections driven according to excavation classes6A-1 and 6A-2 an advancing crown excavation with open invert andtrailing bench and invert excavation was planned (Fig. 3.38). Theevaluation of the mapping carried out during the crown heading re-vealed, however, that the steep joints J1 and J2 striking parallelto acute-angled to the tunnel axis (see Fig. 3.42) appear fre-quently in some areas and have a large extent. Before the start ofthe bench excavation the stability of the tunnel in this construc-tion stage was therefore investigated in FE-analyses with the pro-gram system FEST03 (Wittke, 2000). On the basis of the results ofthese analyses the length of the section E between bench and in-vert excavation (see Fig. 3.38) was to be specified.
Fig. 3.44 shows the computation section, the FE-mesh, the groundprofile and the parameters taken as a basis for the stabilityanalyses (Wittke et al., 1999). The tunnel cross-section is lo-cated in the strongly to slightly weathered Variegated Schist. Theoverburden is 18 m high. Below the tunnel's invert the rock is un-weathered.
Fig. 3.44: Computation section, FE-mesh, ground profile and pa-rameters
As mentioned above, the foliation or bedding, respectively,strikes approx. perpendicularly to the tunnel axis and dipssteeply (see Fig. 3.42). Since discontinuities of this orientationhardly influence the load transfer in transverse tunnel direction,the foliation/bedding was not taken into account in the analyses.The shear strength of the joints J1 and J2, which are relevant forthe stability, was, however, accounted for. The shear strength wasfurthermore varied because of the different appearance of thesediscontinuities. Three cases were investigated, in which the fric-tion angles on the joints ϕJ were assumed as 20°, 22.5° and 25° andthe cohesion as cJ = 0. The anchoring of the rock provided for inall excavation classes was not taken into account in the analysesas a conservative assumption.
The heading of the tunnel was simulated in four computation steps(Fig. 3.45). In the 1st computation step the in-situ state ofstress and deformation resulting from the dead weight of the rockmass was computed. After that, the excavation of the crown (2nd
computation step) and the bench (3rd computation step) was simu-lated, each time with simultaneous installation of the shotcretemembrane. In the 4th computation step the invert was excavated andsupported using shotcrete.
A thickness of t = 25 cm (see Fig. 3.35) and a Young's modulus ofE = 15,000 MN/m2 were specified for the shotcrete membrane.
Fig. 3.46 shows the development of the displacements computed forthe bench support feet in the course of the viscoplastic iterativeanalysis in the 3rd computation step.
Fig. 3.46: Viscoplastic displacements depending on the shearstrength of the joint sets J1 and J2, 3rd computa-tion step
For the case ϕJ = 25° the stability of the construction stage dueto the bench excavation can be proven in the analysis. The hori-zontal and vertical displacements computed at the bench supportfeet converge in the course of the viscoplastic iterative analysis(see Fig. 3.46). The horizontal and vertical components of theviscoplastic displacements of the bench support feet amount to
vpHδ = 15.6 mm and vp
Vδ = 9.2 mm.
With decreasing friction angle ϕJ markedly larger viscoplastic dis-placements are computed (see Fig. 3.46). In the case ϕJ = 20° thedisplacements do not converge in the analysis (see Fig. 3.46). The
stability of the construction stage as a result of the bench exca-vation can thus not be proven by the analysis for this case.
Summarizing it may be stated that the stability of the tunnel dur-ing bench excavation depends to a large degree on the friction an-gle ϕJ on the joints.
On the basis of the results of the stability analyses it wasagreed upon for those sections in which the crown was excavatedwithout invert support to set no limit for the length of the sec-tion E between bench excavation and the support of the invert (seeFig. 3.38). With the bench mapping during heading, however, greatimportance was attached to recording the appearance and extent ofthe joints. Whenever strongly jointed rock was encountered duringbench excavation, the anchoring of the sidewalls was intensified.
3.3.8 Construction and monitoring results
Fig. 3.47 shows the sequence of the heading. Following the crownheading (1), the bench was excavated up to chainage approx. 80 mat a round length of 2.0 m. The invert trailed with a round lengthof 3.6 m and was supported after each round (2). In this area thestrongly weathered rock mass extends into the tunnel cross-section(Fig. 3.47), and a crown invert support was installed.
After that, the bench was excavated from chainage 80 m to 462 mwith a round length of 2.4 m (3 in Fig. 3.47). As mentioned, theanchoring was locally intensified because of the heavy jointing ofthe rock mass.
In the northern portal area the bench was excavated again with im-mediately trailing invert over a short tunnel section (4 in Fig.3.47). The round lengths were the same as in the southern portalarea.
Finally, the invert was excavated backward and supported withround lengths of 3.6 m (5 in Fig. 3.47).
The heading was accompanied by a geotechnical monitoring programincluding surface leveling, extensometer and inclinometer measure-ments from the ground surface and leveling and convergency meas-urements in the tunnel.
Fig. 3.48 shows exemplarily the maximum subsidence at the benchsupport feet measured after bench excavation. Since the elasticpart of the displacements had mostly occurred already before thezero reading, it could not be recorded by the measurements. Themeasurement results were therefore compared to the computed visco-plastic displacement parts shown in Fig. 3.46. It becomes apparentthat the measured displacements are smaller than the viscoplasticvertical displacements computed assuming a friction angle ofϕJ = 25° (approx. 10 mm). It thus turns out that the constructionstage following the bench excavation was stable as predicted.
3.3.9 Conclusions
Crown headings are carried out in excavation classes with andwithout invert support, depending on the rock conditions. Sincethe different excavation classes vary strongly in cost, great im-portance is attached to the appropriate specification. The exampleof the Hellenberg Tunnel shows how the excavation classes can bespecified safely on the basis of the results of stability analy-ses, mapping during construction and monitoring. It becomes appar-ent that proofs of stability according to the FE-method can con-tribute essentially towards the prediction of and decision on theexcavation classes.
The Österfeld Tunnel undercrosses residential areas as well as therailway tracks of the Gäubahn and the S-Bahn (urban railway)partly with low overburden. During the heading of the tunnel un-derneath the railway lines the railway traffic had to be main-tained. In this area, therefore, special measures for the supportof the tunnel were required. Special attention was required, be-cause the road tunnel is located in mudstone layers of the Lias α,which are subjected to high horizontal stresses.
4.1.2 Structure
The Österfeld Tunnel, which has been excavated by means of theNATM is approx. 400 m long and part of the eastern by-pass aroundVaihingen, a suburb of the city of Stuttgart. The new road connec-tion with a total length of 1.9 km is joining the highway B 14 atthe Vaihinger triangle (Fig. 4.1). The Österfeld Tunnel under-crosses the Paradiesstraße and the railway tracks of the Gäubahnand the S-Bahn (Fig. 4.2). Furthermore the new road crosses theNesenbachtal by means of a 170 m long bridge. Following the south-ern end of this bridge, the road is running through the 780 m longHengstäcker Tunnel. The subsequent road section is joining theNord-Süd-Straße (Fig. 4.1). The new road connection was opened fortraffic in September 1999.
The northern portal of the Österfeld Tunnel is located at the endof the roadway Unterer Grund. Up to the Don-Carlos-Brücke the tun-nel runs parallel to the tracks of the Gäubahn and the S-Bahn(Fig. 4.2). Along the first 100 m of this section at the base ofthe adjacent railway trench a 9 m high angular retaining wall islocated. Moreover in the area of the tunnel some residentialbuildings are located. One of these houses is directly under-crossed by the tunnel. Behind the Don-Carlos-Brücke the tunnel un-dercrosses the four tracks of the above mentioned railway lines atan acute-angle. The tunnel ends at the northern flank of the Ne-senbachtal (Fig. 4.2).
For the tunnel a mouth-shaped cross-section was carried out with atotal internal width of 11.2 m and a total height of 9.7 m. Theexcavated cross-section amounts to approx. 98 m2. The thickness of
the shotcrete membrane is 25 to 30 cm and the thickness of the in-terior lining consisting of watertight reinforced concrete is40 cm. For the traffic in each direction a traffic lane of 3.75 mwidth is available (Fig. 4.3).
The radius of curvature at the area of the crown is R = 5.6 m. Theinvert (R = 10.4 m) as well as the temporary invert of the crown(R = 12.2 m) were carried out with larger radii. At the transi-tions from the crown to the temporary invert of the crown (R =1 m) and from the sidewalls to the invert (R = 2.4 m) small radiiwere selected (Fig. 4.3).
4.1.3 Ground and groundwater conditions
During the design phase for the new by-pass road an extensive pro-gram for the exploration of the subsoil and groundwater conditionswas carried out. This program includes core drillings, the inves-tigation of soil, rock and water samples, water level observa-tions, combined extensometer and inclinometer measurements as wellas in-situ stress measurements.
According to the results of this exploration program the ÖsterfeldTunnel is nearly completely located in the Psilonoten- and Angu-latenlayers of the Lower Jurassic (Lias α1 and Lias α2, Fig. 4.4),consisting of mudstones with single limestone and lime-sandstoneinterbeds. The mostly solid mudstones are transversed by beddingparallel discontinuities with small spacings and are distinctlyjointed. From the portal zones up to the central part of the tun-nel the degree of weathering decreases and the strength of the in-tact rock as well as of the rock mass, respectively, increases.The generally very hard layers of limestone and lime-sandstone arecharacterized by two sets of vertical joints J1 and J2, which areoriented perpendicular to the bedding planes and enable the exca-vation with an excavator. The almost horizontal bedding B dipsparallel to the tunnel axis towards the Nesenbachtal. The sets ofdiscontinuities B, J1 und J2 are also present in the mudstones,however, with a considerable less extent and frequency as in thelimestones and lime-sandstones. Concerning the rock mechanical pa-rameters the mudstones, the limestones and the lime-sandstones arecombined to one layer (Fig. 4.5). In the present case this is per-missible, because the thicknesses of the limestones and the lime-sandstones are small in comparison to the thicknesses of the mud-
The transition to the unweathered Knollenmergel, which is locatedunderneath the mudstones of the Rät, is formed by a disintegratedlayer, the so-called reduction zone of the Knollenmergel (Fig. 4.4and 4.5). In the reduction zone as well as in the unweatheredKnollenmergel slickensides are present, which dip with 20° up to40° and strike in all directions (Fig. 4.5).
Fig. 4.5: Lias α, Rät and Knollenmergel, structural model(Wittke, 1990)
On top of the above described layers of the Lias α up to theground overlying strata of reclaimed fill and weathered mudstone(clay) are located (Fig. 4.2). The thickness of these layers var-ies between 3 and 8 m. In the area of the railway trench the rocksurface is situated only a few decimeters below the ground sur-face.
The structural model for the Lias α, the Rät and the Knollenmer-gel, illustrated in Fig. 4.5, was developed within the scope of
other projects in the city of Stuttgart with comparable groundconditions (Wittke, 1990).
The soil mechanical parameters derived for the layers close to thesurface are based on the results of laboratory investigations onsamples taken from core drillings located in the area of the tun-nel. Due to the grain-size distributions and moisture contents aswell as data gained from experience a mean modulus of deformationof E = 25 MN/m2 as well as mean effective shear parameters ofϕ' = 20° and c' = 10 kN/m2 are assumed (Table 4.1).
The elastic behaviour of the Lias α can be approx. described by atransversally isotropic stress-strain-law. Based on the volumetricdistribution of the mudstones and the limestones and lime-sand-stones the mean elastic constants were calculated according toWittke (1990). In these calculations the mudstones were assumed tobe transversally isotropic and the limestones and lime-sandstoneswere treated as isotropic elastic rocks. The elastic constants ofthe mudstones, limestones and lime-sandstones were derived fromthe results of dilatometer tests and laboratory tests on rock sam-ples. Also the experiences gained from other projects in the areaof Stuttgart were utilized for the estimation of these parameters.The general relationships for the elastic constants of alternatingsequences consisting of transversally isotropic rocks are given inSalamon (1968).
layer deformability strength
Fill and clay E = 25 MN/m² ϕ' = 20°, c' = 10 kN/m²Mudstones with singlelayers of limestonesand lime-sandstones
than the shear strength of the intact rock. For the shear strengthof the bedding planes a friction angle of ϕB = 20° and a cohesionof cB = 0 are assumed. For the joints of sets J1 and J2 an angle offriction of ϕJ = 30° and a cohesion of cJ = 40 kN/m2 are estimated(Table 4.1). A tensile strength normal to the discontinuities isnot taken into account.
With regards to the rock mechanical properties the Rät and the re-duction zone of the Knollenmergel are combined to a uniform layer.On the basis of experience for the modulus of deformation a valueof E = 150 MN/m2 is assumed (Table 4.1). In laboratory tests onrock samples unconfined compressive strengths ranging from σu = 0.3MN/m² to 5.0 MN/m2 were determined. Though these values are quitelow, here also the strength on the discontinuities is decisive.The shear parameters of the discontinuities in the Rät and theslickensides at the reduction zone of the Knollenmergel, respec-tively, are estimated to be ϕD = 17.5° and cD = 10 kN/m2 (Table4.1). A tensile strength normal to the discontinuities here alsois not accounted for.
The unconfined compressive strength of the unweathered Knollenmer-gel is somewhat higher than that of the reduction zone. Decisivefor the strength of the unweathered Knollenmergel are however theslickensides with shear parameters of ϕs = 17.5° and cs = 10 kN/m2
(Table 4.1). The modulus of deformation of the unweathered Knol-lenmergel was not evaluated. From experience gained from otherprojects in the area of Stuttgart with comparable subsoil condi-tions (Wittke, 1990) for this parameter a value of E = 1000 MN/m2
is estimated (Table 4.1).
According to the results of investigations, carried out in thepast at different structures in the Lias α high horizontal in-situstresses are to be expected (Grüter, 1988; Wittke, 1990; Wittke1991). Thus in two exploratory boreholes in-situ stress measure-ments using the overcoring technique (Kiehl and Pahl, 1991) werecarried out. These measurements resulted in horizontal in-situstresses in an order of magnitude of ΔσH = 0.2 to 1.9 MN/m2 for themudstone layers of the Lias α. These stresses have to be accountedfor in addition to those resulting from the dead weight due tohorizontally confined in-situ conditions.
In the portal zones the ground-water level is located below theinvert of the tunnel. In the remaining area the water table is lo-
cated somewhat above the invert of the tunnel. In the central sec-tion of the tunnel it is levelled at the middle of the height ofthe cross-section of the tunnel (Fig. 4.2).
4.1.4 Fundamentals of the design
Because of the small overburden, the high horizontal stresses andthe low shear strength of the bedding parallel discontinuities inthe Lias α, in which the tunnel is located, special problems con-cerning the stability of the tunnel during construction arise. Af-ter the excavation of the tunnel the horizontal stresses have tobe transmitted around the tunnel's cross-section. As a consequencestress concentrations at the roof and the invert of the tunnel oc-cur, which may be considerably higher than in the horizontalstresses present in the undisturbed state of stress (in-situstate). If the tunnel remains unsupported over a greater sectionthe mudstones would be highly stressed in horizontal direction.Because of the above mentioned comparable low strength shear fail-ures on the bedding parallel discontinuities above the roof andbeneath the invert of the tunnel would occur. Moreover a bucklingof thin mudstone layers would be possible.
To avoid failures and collapses under these difficult conditions abolted shotcrete support has to be always installed immediatelyafter excavation. Thus the stresses can be transmitted around thetunnel mainly through the shotcrete. To achieve a stable stage ofconstruction the support must be adequately designed and must havean early bearing capacity. High requirements with regard to a highearly strength of the shotcrete are to be fulfilled.
Because of the large cross-section of the tunnel, which amounts to98 m2 (Fig. 4.3), as well as for reasons of the construction proc-ess and for stability it was decided to subdivide the cross-section in crown, bench and invert. Because of the same reason andin order to minimize surface subsidence the crown was designedwith a shotcrete membrane at the temporary invert. Also duringcrown excavation as well as bench and invert excavation, respec-tively, short round lengths and an early closing of the shotcretesupport were foreseen.
An alkali free shotcrete with a quick-setting spray cement wasused according to the dry shotcrete mixture transport technique ofthe Rombold and Gfröhrer company , which is described in Balbach
and Ernsperger (1996) (see Chapter 2.1.2). This shotcrete is char-acterized by a fast development of strength as well as a high ul-timate strength. Although it is a shotcrete with a concrete gradeof B25 corresponding to a C20/25 according to EUROCODE 2 (EC2), itwas well known from experience gained from other projects thatthis shotcrete develops a considerably higher strength comparableto a concrete grade of B45 corresponding to a C 35/45 (EC2). As aconservative assumption the design of the shotcrete membrane,therefore, was based on a concrete grade of B35 corresponding to aC30/37 (EC2). The strength of the shotcrete during constructionwas continuously checked by concrete tests.
4.1.5 Stability analysis for the stages of construction
The stability analyses for the stages of construction of theÖsterfeld Tunnel were carried out at vertical slices according tothe finite element method (Wittke, 2000). In the mudstone layersof the Lias α in addition to the stresses due to dead weight hori-zontal stresses ΔσH varying from 0.5 to 1.5 MPa - depending on thelocation of the considered computation section - were simulated.Analyses without consideration of increased horizontal stresseswere carried out too.
To account for the displacements which occur ahead of the tempo-rary tunnel face and before the shotcrete membrane is installed inthe two-dimensional analyses a stress relief was simulated by re-ducing the Young's modulus of the rock mass to be excavated (Wit-tke, 2000):
Ered = av ⋅ E with av ≤ 1 (4.1)
In the analyses the so-called stress relief factor av is varied be-tween 1.0 (no stress relief) and 0.5. In the next step of analysisthe excavation as well as the installation of the shotcrete mem-brane was simulated simultaneously.
In the scope of the review of the design two- and three-dimensional analyses using the computer code FEST03 (Wittke, 2000)were carried out. By means of these analyses the displacementsmonitored during excavation of the tunnel were back analyzed andthe assumptions and thus the parameters taken as a basis for thestability analyses were checked. In the following the steps as
well as the results of these analyses are explained by means of anexample.
In Fig. 4.6 the finite element mesh used for the three-dimensionalanalyses is illustrated (Hauck et al., 1998). The dimensions ofthe computation section are 100 m x 100 m x 40 m. The computationsection is subdivided into 9074 isoparametric elements with 12720nodes. The computation section is modeling the area of the Para-diesplatz, which is located approx. 120 m south of the portal "Un-terer Grund" (Fig. 4.1 and 4.2). The setup of the finite elementmesh enables the modeling of the stages of excavation, of theshotcrete membrane as well as of the subsoil profile and the rail-way trench. The analyses were carried out assuming an elastic-viscoplastic stress-strain behaviour for the ground. For the mud-stones of the Lias α with single layers of limestones and lime-sandstones as mentioned above a transversally isotropic stress-strain behaviour in the elastic domain as well as increased hori-zontal in-situ stresses were simulated. The soil and rock mechani-cal parameters as well as the three-dimensional finite elementmesh are shown in Table 4.1 and Fig. 4.6.
Fig. 4.6: Computation section, finite element mesh and ar-rangement of extensometers and inclinometers
In the three-dimensional analysis the stages of construction aris-ing during the heading of the tunnel were simulated in severalsteps, which approximately correspond to the driving and supportof the tunnel in reality (Fig. 4.7). The applied so-called "stepby step" method is explained in detail in Wittke (2000).
In the two first steps of the analysis the in-situ state of stressdue to dead weight and the additional horizontal stresses in theLias α were calculated. For the simulation of the increased hori-zontal stresses ΔσH the nodes located at the boundary plane withthe coordinate x = 100 m were displaced in x-direction. These dis-placements lead to horizontal stresses, which correspond to thestress ΔσH existing in the Lias α (Fig. 4.6). In the first step ofanalysis the whole computation section was horizontally loaded bythese displacements. The unit weight γ was accounted for, however,only for the Lias α. In the second step of the analysis the soillayer underneath the surface as well as the Rät and the Knollen-mergel, in which increased horizontal stresses are not existing,were substituted by materials, which have the same mechanical pa-rameters as before, which however are not weightless any longer (γ> 0). Since the new materials were installed stress-free into thealready deformed corresponding elements (Wittke, 2000) and becausein the second step of the analysis the horizontal displacements ofthe boundary x = 100 m were not changed, the soil underneath thesurface, the Rät and the Knollenmergel are loaded only by the deadweight and not subjected to increased horizontal stresses.
The excavation of the railway trench was simulated within thethird step of analysis. In the steps 4 up to 12 the crown excava-tion as well as the excavation of the bench and the invert follow-ing at a certain distance were simulated. In the computation case,which is illustrated in Fig. 4.7, unsupported round lengths of 3 mfor the crown excavation and of 6 m for the excavation of thebench and the invert were simulated. These are larger than thereal round lengths applied during construction which are mentionedbelow (Chapter 4.1.6). Hereby the development of strength of theshotcrete, which was not considered in the analysis, was roughlysimulated. By modeling greater round lengths it was taken into ac-count that the young shotcrete develops its complete bearing ca-pacity only after a number of days (see Chapter 2.1). Thereforethe distance between the load bearing shotcrete support and thetunnel face modeled in the analysis is larger as one round lengthin reality. The excavation of the bench and the invert was simu-
lated stepwise in a similar manner as the crown excavation (Fig.4.7).
By means of three-dimensional analyses it is possible to computethe stress redistributions in the area of the temporary tunnelface, which lead to stress concentrations in the rock mass aheadof the tunnel face, which is not yet excavated, as well as in therock mass adjacent to the excavated cross-section and in the sup-port already installed. To demonstrate the three-dimensional car-rying behaviour the development of the calculated vertical dis-placement of a point at the roof during the heading of the tunnelis illustrated in Fig. 4.8. Up to the 7th step of analysis, inwhich the tunnel face passes the considered point, already 50 % ofthe final displacement due to excavation are evaluated as advanc-ing displacement.
Fig. 4.8: Vertical displacements of a selected point at theroof in the course of analysis
Fig. 4.9 illustrates the calculated principal normal stresses inthe rock mass resulting from excavation and support of the crown(12th step of analysis) computed for a cross-section perpendicularto the axis of the tunnel, and located in the area of the unsup-ported crown (cross-section C-C in Fig. 4.7). For the consideredcomputation case an increased horizontal stress of ΔσH = 1 MN/m2
existing in the Lias α was assumed. As a result of computation anarch is formed in the rock mass around the crown. As a consequenceof the increased horizontal stresses beneath the crown's invertnear the contour of the excavation stresses up to 1.6 MN/m2 arecomputed. Above and underneath the unsupported cross-section ofthe crown as well as at the foot of the slope of the railwaytrench and also in the Knollenmergel the strengths on the discon-tinuities are exceeded (Fig. 4.9).
Fig. 4.9: Principal normal stresses and subsoil areas in whichstrength is exceeded resulting from the crown head-ing , 12th step of analysis (section C-C in Fig.4.7)
In a corresponding illustration Fig. 4.10 shows the displacementsdue to the crown heading. Fig. 4.10 represents the differences be-tween the nodal point displacements computed for steps 12 and 3 ofthe analysis for cross-section B-B (Fig. 4.7). The displacementsoriented towards the excavated opening amount to approx. 6 mm. Forthe foot of the slope of the railway trench displacements of ap-prox. 10 mm are computed (Fig. 4.10).
The comparison of the results of the two- and three-dimensionalanalyses shows that for the two-dimensional analyses a stress re-lief factor of av, which ranges from 0.5 to 0,8, is to be takeninto account to achieve displacements corresponding to the resultsof the three-dimensional analyses.
According to the results of the comparative analyses the shotcretemembrane can be designed with a thickness of 25 cm and a rein-forcement consisting of an inner and outer steel fabric met Q 295considering a safety factor of 1.35.
Fig. 4.10: Displacements due to the crown heading, 12th - 3rdstep of analysis (section B-B in Fig. 4.7)
4.1.6 Excavation and support
For the excavation of the tunnel with a total cross-section of98 m2 (Fig. 4.3) a tunnel excavator was used. The heading was sub-divided into a crown excavation and an excavation of bench and in-vert following the crown at some distance (Fig. 4.11).
The distance from the excavation of the bench and the invert tothe crown was chosen to at least 50 m. The round lengths for thecrown's excavation were chosen between 80 cm and 1.2 m. For theexcavation of the bench and the invert round lengths ranges from1.6 to 3.0 m (Fig. 4.11). Thus during the crown's excavation aswell as the excavation of the bench and the invert an early clo-sure of the shotcrete membrane was realized. This measure hasproven to limit the subsidence especially during the undercrossingof the buildings and the railway to a low level.
Fig. 4.11: Undercrossing of the Gäu- and S-Bahn (urban rail-way), longitudinal section
The regular support of the tunnel consists of a shotcrete membranereinforced by two layers of steel fabric mats and with thicknessesvarying from 25 cm to 30 cm. The temporary invert was supported bya 20 cm to 25 cm thick shotcrete membrane. Also steel sets and asystematic bolting around the crown and the bench are part of thesupport (Fig. 4.3 and 4.12). In Fig. 4.13 details of the supportat the foot of the crown are illustrated.
Due to the low overburden and the high frequency of the disconti-nuities near the ground surface in the area of the two portalsgrouted spiles were used as advancing support. The intensivelyjointed mudstones located immediately above the Oolithenbank (Fig.4.4) turned out to be caving to a major degree. In case of a unfa-vorable location of this rock layer at the tunnel roof, therefore,also in greater distance to the portals the installation ofgrouted spiles were required.
In the area of the railway trench the excavation of the crown wascarried out underneath the protection of a total of ten advancingpipe umbrellas (Fig. 4.11, 4.12 and 4.14). The steel pipes in eachcase were installed from niches and have a diameter of 140 mm anda length of 14.50 m. The overlap between two successive pipe um-brellas was selected to 3 m. The lengths of the niches, which wereexpanded continuously up to a maximum depth of 75 cm, is 6.5 m(Fig. 4.11). Before the excavation of the bench and the invert inthe area of the niches was carried out, the niches were filledwith shotcrete.
The pipes consisting of 2 m long pieces were installed using asolid eccentric bit with an overcut of approx. 1 cm. Using prefab-ricated openings for injection (Fig. 4.12) as well as a doublepacker the annulus was grouted by means of cement based suspen-sion. From the recorded grout volumes it could be concluded thatby this procedure only the annulus between the pipe and the bore-hole wall was filled with grout. An appreciable grouting of therock mass located between the pipes was not achieved. Finally, ina separate working operation, the steel pipes were filled withsuspension (Hauck et al., 1998).
The average completion time for a pipe umbrella was 5 days. Duringthe corresponding interruption of the crown heading the bench andthe invert excavation was carried out which was started after theinstallation of the first pipe umbrella because of their higherrate of advance. In this way an optimum rate of advance could beachieved (Hauck et al., 1998).
For the time interval between the excavation and the installationof the shotcrete membrane the bearing behaviour of the steel pipeumbrella in the longitudinal direction of the tunnel is activated.In other words the space between the tunnel face and the loadbearing shotcrete support in this stage is bridged by the pipes.As a consequence caving and loosening of the rock mass in thisarea as well as resulting subsidence are largely avoided.
4.1.7 Monitoring program and interpretation of the measuringresults
During the heading of the tunnel an extensive monitoring programwith special emphasis on the area of the undercrossing of therailway was carried out. Before the excavation started four main
measuring cross-sections (MC2, MC5, MC9 and MC11, Fig. 4.2) withvertical combined extensometer and inclinometer measuring equip-ments on both sides of the tunnel (Fig. 4.6) were installed.
On the surface above and aside of the tunnel, at the buildings, atthe "Don-Carlos-Brücke" as well as at the sleepers of the railwaytracks points for levelling were installed. The measuring programwas complemented by underground convergency and displacement meas-urements within various tunnel cross-sections, which were carriedout parallel to the tunnel driving.
Eventual subsidence of the railway tracks was monitored by opticalmeasurements carried out from fixed points located outside of thearea of the railway tracks, by means of installation of reflec-tors, which were fixed at the sleepers.
The measuring results show small subsidence of the ground surfacewith a maximum of 2.5 cm. During the undercrossing of the houseParadiesstraße no. 69 the maximum vertical displacement of thebuilding could be limited to 12 mm. The differential settlementsof the building were so small that the admissible angular rota-tions of the building were not reached and thus no visible damagesof the building occurred.
In the area of the railway tracks also no inadmissible subsidenceor differential settlements could be observed.
The results of the monitoring were evaluated and interpreted bymeans of finite element analyses (Hauck et al., 1998).
Exemplarily the measuring results of the measuring cross-section 5(MC5), which is situated in the area of the "Paradiesplatz" (Fig.4.2) will be discussed. It reflects the situation, in which thetunnel is running immediately adjacent to the slope of the railwaytrench and the roof of the tunnel is approx. located at the eleva-tion of the railway trench.
The horizontal displacements measured by inclinometers in fourboreholes during the crown's excavation above the tunnel are ori-ented towards the railway trench and amount up to 9 mm (Fig.4.15a). In the height of the tunnel's cross-section the measuredhorizontal displacements are oriented on both sides towards theexcavated opening and amount 6 to 7 mm. These displacements leadto horizontal convergencies of the side walls of the tunnel.
Above the roof of the tunnel a change of direction of the horizon-tal displacements occurs (drilling E7). This change can be ob-served also at the foot of the slope and was indicated by thetilting of the masts of the contact wire in the direction of therailway tracks.
The following excavation of the bench and the invert of the tunnelhas lead to an increase of the horizontal displacements from 50 upto 100 % (Fig. 4.15b).
The measurement of the vertical displacements resulting from thecrown excavation in the measuring cross-section 5 show a subsi-dence at the ground surface above the tunnel as well as in therock mass above and adjacent to the tunnel. On the other hand inthe area of the railway trench small heavings were measured (Fig.4.16a).
During the following excavation of the bench and the invert nosignificant changes of the vertical displacements were measured(Fig. 4.16b).
The interpretation of the results of the measurements by means offinite element analyses leads to the result that the measured dis-placements can only be understood, if increased horizontalstresses in the rock mass are taken into account. The rock me-chanical parameters on which the stability analyses are based onas well as the increased horizontal in-situ stresses in the orderof magnitude of ΔσH ≈ 1 MN/m2 could be veryfied by the comparisonof measured and calculated displacements (Fig. 4.15 and 4.16).
4.1.8 Conclusions
The design and construction of the eastern by-pass of Stuttgart-Vaihingen can be considered as a challenge. Based on the experi-ence gained in connection with large tunneling projects for roadas well as for urban railway traffic carried out in the city ofStuttgart the complex and partly new tasks, which are related tothe heading of the Österfeld tunnel, could be solved rather excel-lently.
With regard to the stability and the displacements resulting fromtunnel driving special attention was required because of the smalloverburden as well as the increased horizontal stresses and low
shear strengths on the bedding parallel discontinuities of themudstones of the Lias α, in which the tunnel is located.
These problems were solved by the following measures:
- Subdivision of the excavation of the tunnel's cross-sectionin crown, bench and invert,
- installation of a shotcrete membrane and a systematic boltingimmediately after excavation,
- short round lengths and an early closing of the shotcretesupport during the crown's excavation as well as the excava-tion of the bench and the invert,
- reinforced shotcrete support of the temporary invert of thecrown,
- use of a shotcrete with a quick-setting cement with a highearly and ultimate strength,
- carrying out of steel pipe umbrellas in the area of the un-dercrossing of the railway tracks with low overburden.
By these measures the excavation of the tunnel could be carriedout with very little subsidence at the ground surface. At no stageof construction the railway traffic was affected.
Moreover it has been found that the results of the three-dimensional analyses using the finite element method, which werecarried out within the scope of this project, have made an essen-tial contribution for prognosis as well as for the design of themeasures for excavation and support.
4.2 Road tunnel "Elite" in Ramat Gan, Israel
4.2.1 Introduction
The two-lane road tunnel "Elite" was headed in Ramat Gan, a cityin the Tel Aviv area. The tunnel was started from a undergroundparking lot located adjacent to the 260 m high Gate Tower (Fig.4.17), the highest building in the Middle East. The tunnel under-
The axis of the approx. 110 m long Elite Tunnel runs firstly alonga circular arc with a radius of approx. 120 m and then changesinto a straight line (Fig. 4.18).
The tunnel clearance is approx. 4.9 m high and approx. 10 m wide(Fig. 4.19). The tunnel is located in a calcareous sand with low
cohesion (Kurkar), which contains local lenses of cohesionlessfine sands (see Chapter 4.2.3). The overburden ranges from 3 to4.5 m.
The tender design included the advance installation of steel pipes90 to 110 cm in diameter above and adjacent to the tunnel over theentire tunnel length using a microtunneling machine (Fig. 4.19 and4.20). Under the protection of this umbrella of steel pipes filledwith concrete, the tunnel was subsequently to be excavated. Theapproximately rectangular reinforced concrete tunnel cross-section(Fig. 4.19) was to be constructed in blocks in the process withthe excavation being interrupted for each block (Fig. 4.20).
In cooperation with Walter construction company (Walter Bau AG),WBI prepared a contractor's design proposal described in Chapter4.2.4. This contractor's design proposal is based on the NATM andwas later carried out.
In the course of the exploration for other construction projects,eight boreholes were sunk in the tunnel area. Fig. 4.18 shows thedrill points of seven of these boreholes.
In Fig. 4.21 the drill logs of four boreholes are projected onto alongitudinal section through the tunnel axis. According to theseborehole logs, fill or clayey sands and clays exist down to adepth of 2 m. Below, medium dense to dense calcareously bonded,partially cemented sands with a fraction of gravel are encountered(Fig. 4.22). In these slightly cohesive sands, termed "Kurkar",locally cohesionless fine sands are intercalated, as mentionedabove.
Fig. 4.21: Elite Tunnel, longitudinal section with drill logs
To assess the relative density and the bulk modulus, StandardPenetration Tests (SPT) were carried out in each borehole. Fig.4.23 shows exemplarily the drill log and the SPT results for bore-hole B13 located closest to the tunnel alignment (see Fig. 4.18).
formly graded sand with fines. On the basis of the blow counts per30 cm of penetration depth determined with the SPTs ranging be-tween N30 = 16 and N30 = 46, the Kurkar and the uniformly gradedsand can be classified as medium dense to very dense (D = 0.5 to0.7) according to DIN 4094, appendix 1 (1990). An estimate of thebulk moduli according to DIN 4094, appendix 1 (1990) on the basisof the determined values for N30 leads to moduli ranging from ap-prox. 40 to approx. 100 MN/m2.
The groundwater table was found at 0 to 1 m a.s.l. in the explora-tion boreholes, which is equal to 6 to 7 m below the tunnel's in-vert (Fig. 4.23).
4.2.4 Design
Fig. 4.24 shows the tunnel cross-section according to the contrac-tor's design proposal prepared by WBI together with Walter Bau AG.The excavation contour circumscribes the clearance with a heightof approx. 8.2 m and a width of approx. 12.1 m. Because of the lowoverburden, the roof was designed shallow with a radius of curva-ture of 11.75 m. For the sidewalls and the invert large radii wereselected as well with 11.55 m and 14.06 m, respectively. At thetransitions from the roof to the sidewalls and from the sidewallsto the invert the selected radii are comparatively small with2.65 m and 2.12 m. The excavated cross-section amounts to approx.85 m2.
For construction management reasons and for reasons of the stabil-ity of the tunnel face, the cross-section is subdivided into thecrown with temporary invert support and the trailing bench and in-vert (Fig. 4.24 and 4.25). The height of the crown amounts to 5.6m. The transitions from the sidewalls to the temporary invert haveradii of 1.7 m. The temporary invert of the crown is rounded witha radius of 18.38 m (Fig. 4.24).
A thickness of 25 cm is selected for the shotcrete membrane in thevault. In the area of the bench, the invert and the temporarycrown invert the shotcrete membrane is planned with a thickness of20 cm. The thickness of the reinforced concrete interior liningamounts to 40 cm (Fig. 4.24).
To protect the work space and to relieve the area of the temporarytunnel face, the tunnel is planned to be excavated under the pro-tection of pipe umbrellas consisting of 12 m long steel pipes witha diameter of ca 17 cm (Fig. 4.25 and 4.26). The pipes are spacedat approx. 40 cm and have a wall thickness of 7 mm. Four rebarsare entered into each pipe to increase the section modulus. Fur-ther, the pipes are filled with B25 concrete (Fig. 4.26). Thesteel pipes overlap by 3 m (Fig. 4.25). As mentioned above, the
pipe umbrellas are designed to bridge the work space and shouldtherefore be able to carry the entire load resulting from overbur-den and traffic. Their design is therefore based on these loads(see Chapter 4.2.5).
If the sand has a cohesion or an apparent cohesion, the gaps be-tween the pipes can be bridged by the arching effect. If dry,loose sand or cohesionless fill appears in the roof area, the gapsbetween the pipes must be supported, e. g. by the installation ofadditional pipes (Fig. 4.26). Alternatively, the gaps may bebridged by steel plates welded to the pipes, or they may be sup-ported using spiles (see Chapter 4.2.6).
The contractor's design proposal includes a steep temporary tunnelface inclined at approx. 65° and supported by reinforced shotcreteand tunnel face anchors (SN-anchors) spaced at 2 to 3 m. The tun-nel face anchors shall be 12 m long, just as the pipes (Fig.4.25). By installing the anchors in parallel with the constructionof the pipe umbrellas, the interruptions of the heading necessarydue to the construction of the pipe umbrellas should be minimized.
Fig. 4.26: Elite Tunnel, contractor's design proposal, cross-section with pipe umbrella and detail
Round lengths of approx. 1 m were specified for the crown heading(Fig. 4.25). After each round, the shotcrete support including theclosing of the temporary invert are to be completely installed be-fore the excavation continues. This way the unsupported span inthe crown never amounts to more than 1 m. Round lengths of approx.
2 m were specified for the excavation of bench and invert (Fig.4.25). Here as well the shotcrete support are to be completely in-stalled after each round including its closing at the invert, sothat the unsupported length amounts to 2 m at the most.
The contractor's design proposal has essentially two advantagesover the tendering design (see Fig. 4.19 and 4.20). The headingand the installation of the reinforced concrete interior liningare carried out in two separate working steps, which leads to aconsiderable gain of time and thus to cost savings. Further, thecontractor's design proposal does not require the use of a micro-tunneling machine.
4.2.5 Stability analyses
Design of the shotcrete support
For the dimensioning of the shotcrete support two-dimensionalanalyses were carried out using the program system FEST03 (Wittke,2000).
Fig. 4.27 shows the computation section, the FE-mesh, the boundaryconditions, the ground profile and the parameters the analyseswere based on. The computation section consists of a 48 m wide, 45m high and 1 m thick slice. The FE-mesh was divided into 630 iso-parametric elements with a total of 3958 nodes.
For the nodes on the bottom boundary (z = 0) and on the lateralboundaries (x = 0 and x = 48 m) sliding supports were selected asboundary conditions. On the two planes perpendicular to the tunnelaxis, equal displacements were prescribed for the nodes with equalx- and z-coordinates (Wittke, 2000). All nodes were assumed fixedin y-direction. The traffic load acting on the ground surface wasaccounted for by a surface loading (pt = 23 kN/m2). The overburdenamounts to 4.5 m (Fig. 4.27).
The ground was subdivided into two soil layers. Down to a depth of8.5 m a medium dense sand was assumed with a Young's modulus ofE = 100 MN/m2 and a Poisson's ratio of ν = 0.35, corresponding to abulk modulus of Es = 160 MN/m2. Below that, a dense sand withE = 250 MN/m2 and ν = 0.35 corresponding to Es = 400 MN/m2 wasspecified. To be on the safe side, no cohesion was assumed forany of the two sands (c' = 0). An angle of friction of ϕ' = 30°
was specified. For the shotcrete membrane, a Young's modulus ofE = 15000 MN/m2 was assumed (Fig. 4.27).
Fig. 4.27: Computation section, FE-mesh, boundary conditions,ground profile and parameters for two-dimensionalanalyses
Since the ground profile and the cross-section of the tunnel aresymmetrical to the tunnel axis, only one half of the tunnel cross-section was modeled (Fig. 4.27).
Fig. 4.28 illustrates the computation steps. The 1st computationstep comprises the determination of the state of stress and defor-mation resulting from the dead weight of the ground and the traf-fic load pt. In the 2nd computation step the excavation and theshotcrete support of the crown are modeled. The 3rd computation
It should be pointed out that the simulation of excavation andsupport in one computation step leads to a overestimation of theloading of the shotcrete membrane, since the displacements preced-ing the excavation, which have already occurred before the supportis installed, are not taken into account in the analysis. This inturn results in the tunneling-induced ground surface subsidencebeing underestimated.
Fig. 4.29 shows the displacements of the ground surface and thetunnel contour (crown) computed for the 2nd computation step rela-tive to the 1st computation step. The largest displacements resultat the roof with 25 mm and at the ground surface with a maximum of22 mm. These values change only marginally in the 3rd computationstep.
Fig. 4.30 depicts the bending moments and normal thrust in theshotcrete membrane determined for the 2nd and 3rd computation step.Because of the small radius in the case of the crown heading the
bending loading of the shotcrete support is larger in the 2nd com-putation step than in the 3rd computation step. The largest momentsresult on the sidewalls. In the 3rd computation step the largestbending moment occurs at the transition from the roof to the side-walls. Compressive normal thrust are computed for the entire tun-nel circumference.
In Fig. 4.31 the statically required amounts of reinforcement aregiven for the design of the shotcrete membrane according to DIN1045 (1988) for a safety factor of η = 1.45. This factor of safetyincludes the safety factors of ηt = 1.6 for the traffic load (pt)and ηγ = 1.4 for the overburden weight (γ · Ho) which were predeter-mined by the constructor:
Fig. 4.31: Statically required reinforcement of the shotcretemembrane: a) 2nd computation step; b) 3rd computation
step
For a concrete grade of B25, a steel grade of BSt 500/550 and asurface distance of the reinforcement of t1 = 3 cm, steel cross-sections of ≤ 2.3 cm2/m on the inside and ≤ 5.2 cm2/m on the out-side are evaluated as maximum statically required reinforcement.
These can be covered by steel fabric mats Q295 on the inside andon the outside and by supplementary reinforcement within the uppersidewall area (Fig. 4.31). In the bench and invert area, the in-side steel fabric mat Q295 can be omitted. Fig. 4.32 shows the de-sign and the reinforcement of the shotcrete membrane in the areaof the crown's foot.
Fig. 4.32: Design and reinforcement of the shotcrete membranein the area of the crown's foot
To investigate the stability of the temporary tunnel face, three-dimensional analyses were carried out using the program systemFEST03 (Wittke, 2000).
Fig. 4.33 shows the computation section, the FE-mesh, the boundaryconditions, the ground profile and the parameters these analyseswere based upon. The computation section is 40 m wide, 40 m highand 62 m long. The FE-mesh was divided into 5848 isoparametricelements with a total of 14945 nodes.
Fig. 4.33: Computation section, FE-mesh, boundary conditions,ground profile and parameters for three-dimensionalanalyses
Sliding supports were selected as boundary conditions for thenodes on the bottom boundary (z = 0). The nodes on the planes x =0 and x = 40 m were assumed fixed in x-direction, and the nodes onthe planes y = 0 and y = 62 m were fixed in y-direction. The traf-fic load acting on the ground surface was increased by the safetyfactor ηt = 1.6 and applied as a uniform surface load. An overbur-den of 4 m was specified (Fig. 4.33).
The ground profile and the parameters corresponded to the assump-tions made for the two-dimensional analyses. The weight of thesoil (γ = 20 kN/m3) and thus also the overburden pressure were in-creased by the safety factor ηγ = 1.4.
The tunnel face was assumed inclined at 80°. The tunnel face an-chors were considered by a cohesion in the respective area (Fig.4.33). Five cases were investigated: c' = 0 (no tunnel faceanchors), c' = 25 kN/m2, c' = 50 kN/m2, c' = 75 kN/m2 andc' = 100 kN/m2, corresponding to an anchor arrangement with araster spacing between 1 m x 1 m and 2.1 m x 2.1 m.
The three-dimensional analyses were carried out in two computationsteps as a simplification (Fig. 4.34). Following the analysis ofthe primary state in the 1st computation step, the excavation andsimultaneous shotcrete support of the crown and the installationof the pipe umbrella were simulated in the 2nd computation step.The unsupported area adjacent to the tunnel face was assumed to be1 m deep. The shotcrete support of the tunnel face was not takeninto account.
It should be pointed out that with these analyses as well the tun-neling-induced displacements of the soil are underestimated, sincethe support exerted by the shotcrete membrane is overestimatedwith the computation sequence outlined in Fig. 4.34. For a realis-tic computation of the displacements, one of the two proceduresfor the simulation of a three-dimensional tunnel heading describedin Wittke (2000) (step-by-step method or iterative method) wouldbe necessary. The analyses, however, had the only purpose of as-sessing the stability of the temporary tunnel face.
Fig. 4.35: Principal normal stresses computed after completionof the viscoplastic iterative analysis and elementswith exceeded strength (c' = 25 kN/m2), 2nd computa-tion step (vertical section through the tunnel axis)
As a consequence of the excavation, the strength of the soil islocally exceeded in the area of the tunnel face. Fig. 4.35 showsthe principal normal stresses computed for the 2nd computation stepand the elements with exceeded strength, specifically marked, in avertical section through the tunnel axis for the case of a cohe-sion of the anchored area of c' = 25 kN/m2. A corresponding repre-sentation of the computed displacements is given in Fig. 4.36. Thedisplacements result from elastic and inelastic deformations ofthe soil. The latter are computed in a viscoplastic iterativeanalysis (Wittke, 2000). Fig. 4.35 and 4.36 show the results ofthe 2nd computation step after completion of the viscoplastic it-erative calculation.
The criterion for the proof of stability of the tunnel face is theconvergency of the nodal displacements in the course of the visco-plastic iterative analysis. Fig. 4.37 shows the development of the
displacements computed for a node on the tunnel face in the courseof the viscoplastic iterative analysis in the 2nd computation stepfor the five investigated cases. For the case without tunnelface anchors (c' = 0), the convergency of the displacement ofthis node cannot be proven by the analysis. In all other cases(c' ≥ 25 kN/m2) the displacement converges in the computations. Forc' = 25 kN/m2 a displacement of only 7 mm results (Fig. 4.36 and4.37). The cohesion of 25 kN/m2 corresponds to an anchor raster of2.1 x 2.1 m.
Fig. 4.37: Displacement of a node on the tunnel face in thecourse of the viscoplastic iterative analysis, 2nd
computation step
Design of the pipe umbrella
As mentioned above, the pipe umbrella must be able to carry theloads from traffic and from the weight of the overburden. As forthe proof of stability of the tunnel face, the traffic load wasincreased by the factor of safety ηt = 1.6 and the overburden pres-sure of the soil by the factor of safety ηγ = 1.4 for the design ofthe pipe umbrella, as requested by the client.
As explained in Chapter 4.2.4, the round length for the crownheading amounts to 1 m and the shotcrete support is to be in-stalled and temporarily closed at the invert after each round. Forthe design of the pipe umbrella it is conservatively assumed thatits maximum free span amounts to approx. 2 m. It is taken into ac-count here that the green shotcrete does not have any bearing ca-pacity yet directly after its application (see Chapter 2.1.3).
Fig. 4.38: Design of the pipe umbrella, statical systems andloading
At 80°, the inclination of the tunnel face is assumed somewhatsteeper than in the design (see Fig. 4.25). This results in a dis-tance between the tunnel face and the closed support of approx. 2m at the roof and approx. 1 m at the temporary crown invert (Fig.4.38).
Fig. 4.39: Design of the pipe umbrella, calculation of the sec-tion modulus of the pipes
The beam on two supports and the beam fixed at both ends are con-sidered as statical systems for the design of the pipe umbrella.For the reasons given above, the maximum span of the beam is as-sumed as 2 m. For the fixed beam, 0.5 m each on both ends of thebeam are added to the length and assumed to be fixed. From the su-perposition of the traffic load and the overburden pressure, tak-ing into account the spacing of the pipes (L = 418 mm), the load-ing of the beam results to q = 65.7 kN/m (Fig. 4.38).
In Fig. 4.39 the calculation of the section modulus W of thepipes, which is required for the stress proof is specified. Forthe proof of safety, the computed stresses are compared to theyield stress of the pipes made from steel of the grade St37(βy = σadm = 240 N/mm2). This is permissible, since the assumedloads were provided with factors of safety. The beam fixed at bothends is decisive for the design with a computed tension of192.2 N/mm2 (Fig. 4.40).
Fig. 4.40: Design of the pipe umbrella, stresses and deflection
The deflection of the pipe umbrella is estimated at approx. 3 mm(Fig. 4.40).
The Elite Tunnel was the first tunnel in Israel to be constructedby the NATM. Particularly high demands were therefore made on thetechnical construction supervision provided by WBI. The demandsfocused mainly on the works for the excavation and the installa-tion of the shotcrete membrane, which had to be continuously su-pervised.
Differing from the design the pipes only had a length of 10.4 m.Since the overlap of the pipe umbrellas amounts to 3 m, it waspossible to excavate for 7.4 m under one umbrella before the pipesfor the next umbrella had to be installed (Fig. 4.41).
Fig. 4.41: Elite Tunnel, excavation and support, longitudinalsection
At the roof and above locally fill or layers with cohesionless,fine grained, loose sands were encountered. Here, the gaps betweenthe steel pipes had to be supported (Fig. 4.42). In the area ofthe first two pipe umbrellas the gaps were supported by steel
plates welded to the pipes (Fig. 4.42a). In the area of pipe um-brellas 3 to 14 cemented rebars (spiles) were installed betweenthe steel pipes (Fig. 4.42b). The drillings for the spiles servedat the same time to explore the ground in advance. In the area ofthe last two pipe umbrellas, additional steel pipes, filled withB25 concrete but not reinforced, were installed (Fig. 4.42c).
Fig. 4.42: Support of the gaps between the pipes in case of lo-cally occurring dry, loose sand or fill: a) Pipe um-brellas 1 and 2; b) pipe umbrellas 3 to 14;c) pipe umbrellas 15 and 16
Due to reasons of construction, a vertical tunnel face was carriedout. For stability reasons the crown generally had to be excavatedin several steps and supported immediately with reinforced shot-crete (t = 5 to 10 cm) (Fig. 4.41).
Fig. 4.43 depicts a geotechnical mapping of the crown face atchainage 17. In Fig. 4.44, the supported crown is shown at chain-age 36. It can be seen here that in the middle of the crown a sup-
The excavation of bench and invert was only started after thecrown of the entire tunnel had been excavated (see Fig. 4.41).
4.2.7 Monitoring
To measure the subsidence due to the tunneling, 13 leveling pointswere installed at the ground surface above the tunnel roof, andeight measuring cross-sections with roof bolts were installed inthe tunnel. The measurements served to monitor the stability ofthe tunnel.
Fig. 4.48 shows exemplarily the subsidence of the ground surfaceand the tunnel roof measured during the crown heading with thetemporary face located at chainage 39. The subsidence of theground surface amounts to approx. 20 to 50 mm, the measurable sub-sidence of the roof accounts for approx. 15 to 20 mm.
Fig. 4.48 further shows the ground surface subsidence measured af-ter completion of the crown heading. It amounts to approx. 20 mmin the area of the northern portal and to approx. 40 to 60 mm inthe remaining tunnel section.
For the comparison of the measured subsidence with the results ofthe stability analyses it must be noted that the subsidence pre-ceding the excavation, which has already occurred before the shot-crete support is installed, was not captured in the analyses (seeChapter 4.2.5).
The subsidence that has occurred at the ground surface before theshotcrete support is installed amounts to approx. 20 mm in the ex-ample of Fig. 4.48. By adding this subsidence to the calculatedsubsidence of approx. 22 mm (see Fig. 4.29) and to the estimateddeflection of the pipe umbrella of approx. 3 mm (see Fig. 4.40), atotal subsidence of approx. 45 mm can be derived from the analysisresults which agrees well with the measured subsidence (see Fig.4.48).
4.2.8 Conclusions
The Elite Tunnel in Ramat Gan crosses through a medium dense todense, calcareously bonded, slightly cohesive sand termed"Kurkar". Loose, cohesionless sands are locally embedded. Theoverburden amounts to 3 to 4 m. Since the tunnel undercrosses aneight-lane road, the subsidence had to be kept small.
The tunnel was excavated by the NATM under the protection of apipe umbrella as a crown heading with a closed support at the tem-porary invert and trailing bench and invert excavation. The crownwas excavated in several parts with a vertical tunnel face, a sup-port core and short round lengths and supported by reinforcedshotcrete and tunnel face anchors. The shotcrete membrane was in-stalled and closed at the invert at a distance of approx. 1 m tothe tunnel face (see Fig. 4.41).
Following this procedure the tunnel was excavated in a stable way.The ground surface subsidence amounted to approx. 4 to 6 cm.
The results of the FE-analyses contributed essentially towards thedesign, the statics and the specification of the excavation andsupport measures.
4.3 City railway tunnel to Botnang, in Stuttgart, Germany
4.3.1 Introduction
In the early 1990's the Stuttgart city railway line U9 was im-proved up to Botnang terminal. As a part of this project the"Herder Street" and "Lindpaintner Street" stops were connected bya 550 m long, double-tracked tunnel (Fig. 4.49a). The tunnel un-dercrosses the Botnang saddle in a wide turn as well as theGäubahn and some built-up areas (Fig. 4.49).
4.3.2 Structure
Between the Botnang portal and chainage km 4+392 the two-trackedstandard profile was constructed over a length of 379 m (Fig.4.49b) with a height of approx. 8.5 m and a width of approx.10.5 m (Fig. 4.51). From chainage km 4+392 to the Herder Streetportal at chainage km 4+239.5 the height of the cross-section in-creases from approx. 8.5 to approx. 14 m (enlarged profile, Fig.4.49b). The maximum cross-section at the Herder Street portal hasa height of approx. 14 m and a width of approx. 12 m (Fig. 4.50).The maximum overburden amounts to some 47 m (Fig. 4.49b).
The maximum cross-section at chainage km 4 + 239.5 is shown inFig. 4.50. The shotcrete membrane is 30 to 35 cm thick, the thick-ness of the reinforced concrete interior lining amounts to 60 cm.In the area of the sidewalls and the invert, with R = 13.16 m andR = 14.5 m, respectively, comparatively large radii were selectedfor the rounding of both linings. In the roof area a smaller ra-dius of R = 4.6 m was designed for statical reasons. At thetransitions from the sidewalls to the invert smaller radii ofR = 1.1 m were selected. The excavated cross-section amounts toapprox. 142 m2.
4.3.3 Ground and groundwater conditions
To explore the ground and groundwater conditions core drillingswere sunk along the tunnel alignment and equipped as observationwells (Fig. 4.49b).
- Dunkle Mergel and Schilfsandstone layer (marlstone, sand-stones and claystones),
- Gypsum Keuper.
Fig. 4.51: Excavation and support, standard profile, excavationclass 7A, cross-section
Fill of larger thickness occurs at the ground surface and at theportal areas. It mainly consists of firm to stiff sandy silt withembedded solid and hard rock fragments and to a minor degree alsobuilding rubble. Fill with a thickness of up to 17 m was found atthe settlement-sensitive Gäubahn embankment, which the tunnel un-dercrosses over a tunnel length of 100 m (Fig. 4.49b).
The flanks of the Botnang saddle are covered with talus material,the thickness of which usually amounts to 1 to 2 m. Up to 4 m ofthickness are reached in shallowly dipping terrain where the talusmaterial changes into residual loam. The latter consists of softto stiff silts containing heavily varying amounts of solid andhard rock fragments. In the area of the Botnang Tunnel portal aninclinometer was installed already before the start of construc-tion. The measurements gave no indication of slope movements.
Layers of partially plastic leaching silts appear in the marls,which are belonging to the Keuper formation.
The Schilfsandstone layer is composed of a series of gray-green,gray and brown, mostly solid and hard marlstone and sandstonebanks that are weathered close to the surface. The upper part ofthe profile is dominated by marlstones with a varying however highfine sand content. Gray-black, sand-free claystones are interca-lated in some areas as well. The thickness of this layer amountsto approx. 25 m. The lower part of the Schilfsandstone layer, lo-cated within the tunnel's cross section, consists mainly of hardsandstone banks with clay flasers. The fine- to medium-grainedsandstones are clay-bonded or cemented by immediate siliceousgrain bonding. Quantitative mineralogical investigations yieldedquartz contents ranging from 30 to 40 %.
The thickness of the layers varies between 10 and 100 cm. The bed-ding parallel discontinuities are occasionally marked by soft orfirm clay layers, on which the banks tend to separation.
Sandstones of the Schilfsandstone layer have a medium to widejoint spacing, while joints in clay- and marlstones are narrow- tomedium-spaced. Two vertical joint sets exist, which intersect atan angle of 60 to 70°. Their acute angle includes the N-S-direction. Thin sandstone banks are fractured into plates, whilethick banks are fractured into columns rather. The joints aremostly rough, slightly undulating and closed. Open joints andclayey coatings occur mostly in the hillside area (slope dilata-tion) as well as close to the surface.
The Gypsum Keuper consists of an alternating sequence of soft tofirm silts, stiff claystone layers and hard claystone and dolomitelayers. The Gypsum Keuper layers encountered are almost completely
leached. In several boreholes single, strongly leached gypsum nod-ules as well as thin gypsum layers were found.
The soil and rock mechanical parameters listed in Table 4.2 arebased on the results of borehole expansion tests using the bore-hole jack model "Stuttgart", on laboratory tests on samples takenfrom the exploration boreholes as well as on experience gainedfrom other projects located in comparable ground conditions. Thestability analyses (see Chapter 4.3.5) were based on these parame-ters.
According to the results of the piezometer measurements, the un-disturbed groundwater table is located above or at the tunnel roofalmost over the entire tunnel length. In the area of the HerderStreet portal it falls to about the middle of the tunnel's cross-section (Fig. 4.49b).
4.3.4 Design
From the Herder Street portal to chainage km 4 + 310 the double-tracked standard profile was planned to be excavated first as atemporary stage. After that the final profile was to be excavatedfrom chainage km 4 + 310 to the Botnang portal. The final profileis equal to the enlarged profile from chainage km 4 + 310 tochainage km 4 + 392 and to the double-tracked standard profilestarting at chainage km 4 + 392. After the cut-through the cross-section was planned to be enlarged in the invert to the final pro-file from chainage km 4 + 310 backwards to Herder Street (Fig.4.49b).
Fig. 4.51 shows the standard profile divided into crown, bench andinvert and the excavation and support measures planned for thestandard excavation procedure. The excavation class was designatedas 7A according to the recommendations of the working group "Tun-neling" of the German Geotechnical Society (DGGT, 1995: Table 1).
Crown and bench were each excavated with an immediate installationof the support. The crown face may only be ahead of the bench faceby a maximum of approx. 3 m (Fig. 4.52, phase I in the left por-tion). Shotcrete with a thickness of 25 to 30 cm and reinforced bytwo layers of steel fabric mats Q257, support arches spaced at 0.8to 1.1 m and a systematic anchoring using SN-anchors were planned(Fig. 4.51). Mortar spiles should be used as advancing support.
The support should be closed at the invert no more than 2.4 m be-hind the invert excavation and no more than some 16 m behind theexcavation at the roof (Fig. 4.2, phase I in the left portion).The reinforced shotcrete should be placed at the invert with athickness of 20 to 25 cm (Fig. 4.51).
The tunnel face was planned to be excavated steeply and in steps,with a support core in the crown area. Immediately after excava-tion the tunnel face was to be supported using 5 to 10 cm thickshotcrete, possibly reinforced by a steel fabric mat Q257 (Fig.
4.52, left portion). During the excavation the shotcrete supportshould be opened in sections, and the exposed partial areas shouldimmediately be sealed again by shotcrete after the excavation.
Fig. 4.52: Excavation and support, longitudinal section
Fig. 4.52 (phase II, right portion) and 4.53 show the excavationand support measures planned for the later lowering of the invertin the area with enlarged cross-section.
In the upper part of the cross-section (crown, bench I and invertI), excavation and support was to be carried out as for the stan-dard heading, but with a thicker shotcrete membrane (t = 30 to 35cm) and systematic anchoring (see Fig. 4.51 and 4.53). A rein-forced shotcrete membrane with a thickness of 30 to 35 cm shouldalso be installed in the lower part of the cross-section (bench IIand invert II). The support should be closed at the invert no morethan 3.3 m behind the excavation of invert II and no more than 9.9m behind the excavation of bench II (Fig. 4.52, right portion).
The support of the tunnel face and the advancing support should beconstructed as for the standard heading.
Fig. 4.53: Excavation and support, maximum cross-section, low-ering of the invert, cross-section
4.3.5 Stability analyses for the design of the shotcretesupport
For the design of the shotcrete support, two-dimensional FE-analyses on vertical slices were carried out with the program sys-tem FEST03 (Wittke, 2000). Fig. 4.49b shows the locations of theanalysis cross-sections investigated in the final design analyses.The analyses were based on the parameters given in Table 4.2.
In the following, analysis cross-section 2 will be exemplarilytreated. It is located at chainage km 4+296 in the area of the un-dercrossing of the Gäubahn (Fig. 4.49b). In Fig. 4.54 the computa-tion section, the FE-mesh, the boundary conditions, the ground
profile and the parameters are shown. The computation section con-sists of a 63.9 m high, 45.7 m wide and 1 m thick slice of rockmass. The FE-mesh was divided into 1104 isoparametric elementswith a total of 1320 nodes.
For the nodes on the lower boundary plane (z = 0) and on the lat-eral boundary planes (x = 0 and x = 45.7 m), sliding supports wereintroduced as boundary conditions (Fig. 4.54). For the two planesperpendicular to the tunnel axis (y = 0, y = 1 m), equal displace-ments were assumed as boundary conditions for nodes with equal x-
and z-coordinates (Wittke, 2000). All nodes were assumed fixed iny-direction.
In the area of analysis cross-section 2, the tunnel has approx.23 m of overburden. The upper part of the cross-section is locatedin the Schilfsandstone, the lower in the weathered Gypsum Keuper.Above the Schilfsandstone, hillside loam and the fill of the em-bankment constructed for the Gäubahn are modeled, with thicknessesof approx. 3 m and 15 m, respectively. The weathered Gypsum Keuperis underlain several meters below the tunnel's invert by the un-weathered Gypsum Keuper (Fig. 4.54).
The vertical section through the tunnel axis constitutes a planeof symmetry with respect to the geometry of the tunnel cross-section and to the ground profile. Therefore, only one half of thetunnel was modeled in the analysis (Fig. 4.54).
Fig. 4.55 and 4.56 show the six computation steps used to simulatethe excavation and the support during tunneling according to thedesign (reference case). In the 1st computation step, the state ofstress and deformation resulting from the dead weight of theground (in-situ state) is determined. In the 2nd computation stepthe excavation of the crown and its support using shotcrete aresimulated. The 3rd computation step comprises the excavation andshotcrete support of bench I. In the 4th computation step excava-tion of the invert I and the closing of the temporary support atthe invert are simulated. Since the first two construction stages(computation steps 2 and 3) were simulated with an open invert,this analysis sequence accounts for a late closing of the invertas specified in the design with a distance of approx. 16 m to thetunnel face (see Fig. 4.52). The connection of the temporary in-vert support to the sidewall was simulated at first without anycurvature corresponding to the design (Fig. 4.55 and 4.56).
In the 5th and 6th computation step, the excavation of bench II andthe excavation of invert II with simultaneous installation of theshotcrete support also at the invert are simulated.
Fig. 4.57 presents the principal normal stresses determined forthe 6th computation step (complete excavation), as well as thoseareas in which the strength has been exceeded. It can be seen thatthe plastic zones are limited to the area of the tunnel contour.
Fig. 4.57: Analysis cross-section 2, reference case, principalnormal stresses and elements with exceeded strength,6th computation step
Fig. 4.58 depicts the displacements computed for the 6th computa-tion step and thus the displacements in the stage after the exca-vation of the total cross-section. The analysis results in com-paratively large vertical displacements of 12 cm at the roof and7.4 cm on the ground surface.
Fig. 4.59 shows the stress resultants computed for the stage afterthe excavation and support of invert I (4th computation step) andof the total cross-section (invert II, 6th computation step). Inthe 4th computation step (excavation of invert I) extremely highbending moments result in the area of the connections of the tem-porary invert support to the sidewalls (Fig. 4.59a). For thisloading, the shotcrete membrane cannot be reasonably designed. Inthe 6th computation step as well, comparatively large bending mo-ments occur in the area of the sidewalls (Fig. 4.59b). The designof the shotcrete membrane for a concrete grade of B25, a shotcrete
thickness of t = 35 cm and a safety factor during construction of1.35 yields that in the lower sidewall areas a higher amount ofreinforcement of the shotcrete membrane compared to the originaldesign is required (Fig. 4.60).
Subsidence of the computed magnitude (Fig. 4.58) could also not bepermitted for the undercrossing of the Gäubahn. On the basis ofthese analysis results and of the displacements measured duringthe heading (see Chapter 4.3.7), it was decided in agreement withall parties concerned to close the support at the invert earlierin order to keep the subsidence smaller. The computation steps forthe analyses simulating an early support closing at the invert areshown in Fig. 4.61 and 4.62. The excavation of bench I and invertI as well as the installation of the temporary invert support weresimulated in one computation step (3rd computation step, Fig.4.61). In order to reduce the loading of the shotcrete membrane inthe area of the connection of the temporary invert with the side-walls, a rounded connection was modeled (Fig. 4.61).
Fig. 4.60: Analysis cross-section 2, reference case, staticallyrequired inside reinforcement of the shotcrete mem-brane
To reduce the high bending loading of the shotcrete membrane inthe sidewall areas after the excavation of the complete cross-section (H ≅ 14 m, see Fig. 4.59 and 4.60), an anchoring of theshotcrete membrane in the sidewall areas by untensioned anchorswas accounted for in the 4th computation step in addition to theexcavation of bench II. The anchoring in the sidewall areas wasmodeled by truss elements. Five untensioned anchors per tunnel me-ter were assumed in the analysis with a cross sectional area of4.5 cm2 each. By the 5th computation step the excavation and sup-port of the invert was simulated (Fig. 4.62).
The results of this analysis show first, that the computed dis-placements decrease considerably due to the simulation of an earlyclosing of the invert and of the anchoring (see Fig. 4.58 and4.63). The subsidence of the ground surface amounts to only 27 mmas opposed to 74 mm in the reference case. The horizontal dis-placements of the sidewalls decrease from 70 mm in the referencecase to 18 mm.
The rounding of the cross-section in the area of the invert leadsto a marked reduction of the bending moment (see Fig. 4.59a and4.64). For a concrete grade of B25, a shotcrete support thicknessof t = 35 cm and a factor of safety of 1.35 the analysis yieldsthat supplementary reinforcement is required in the lower sidewallarea in addition to the planned steel fabric mats Q257. Further,in the area of the greatest change in bending moment correspondingto the greatest shear force a shear reinforcement is necessary. Itcan be covered by diagonally bent-up reinforcement.
Fig. 4.64: Analysis cross-section 2, case 2, stress resultantsin the shotcrete support, 3rd computation step
In Fig. 4.65a the tensile anchor forces determined in the 5th com-putation step (see Fig. 4.62) are given. Values of up to 100 kN(10 t) per anchor are computed.
The bending moments in the shotcrete membrane computed for thestage after the excavation of the complete cross-section (5th com-putation step) with the anchoring taken into account are given inFig. 4.65b. Compared to the reference case (without anchoring) astrong reduction of the bending loading becomes evident (see Fig.4.59b and 4.65b). The dimensioning yields that no supplementaryreinforcement is required in the sidewalls if steel fabric matsQ257 are installed. At the transition from the sidewalls to the
temporary invert, however, supplementary reinforcement in additionto the mats is required.
Fig. 4.65: Analysis cross-section 2, case 3, 5th computationstep: a) Tensile anchor forces; b) bending moments
4.3.6 Construction
Fig. 4.66 shows the excavation and support measures carried outwithin the area of the enlarged cross-section during the heading(Beiche and Kagerer, 1993).
The shotcrete membrane was constructed with a thickness of 35 cmand rounded in the transition zones from the sidewalls to the tem-porary crown invert. The reinforcement included inside and outsidereinforcement mats Q257 and supplementary reinforcement in thetransition zones. Further, according to the design a systematicanchoring of the vault and the sidewalls was carried out (Fig.4.66). In the central sidewall areas (invert I, bench II) the an-choring was intensified.
Fig. 4.66: Excavation and support within the area of enlargedcross-section, construction
In addition, advancing injections were carried out to improve theground ahead the tunnel face and above the crown, and injectiondrill spiles with mortar filling were installed.
The rounds were carried out with lengths of ≤ 80 cm. The temporarycrown invert was supported in the beginning at a distance of 9 to14 m to the tunnel face. In the area of the undercrossing of the
Gäubahn and in the immediately following part of the heading, thesupport was closed at the invert at a maximum of 6 m behind thetunnel face (Beiche and Kagerer, 1993).
4.3.7 Monitoring
To measure the tunneling-induced displacements, measuring cross-sections with extensometers and inclinometers were installed inthe area of the undercrossing of the Gäubahn and of ZamenhofStreet, among other locations. In the tunnel, measuring cross-sections for levelings as well as stress measurement cross-sections with rock mass pressure and concrete pressure measuringcells were installed. In addition, levelings were carried out onthe grout surface, on the tracks of the Gäubahn and on structures(Fig. 4.67).
The subsidence of the ground surface due to the crown headingamounted to approx. 25 mm at the beginning of the excavation andreached 51 mm at the edge of the Gäubahn embankment (Fig. 4.68).Up to this time the temporary crown invert was supported 9 to 14 mbehind the tunnel face. To reduce the subsidence, especially dur-ing the undercrossing of the Gäubahn, the support was closed atthe invert in the further course of the heading at the minimumpossible distance of 6 m to the tunnel face, as mentioned above.It was thus possible to reduce the subsidence, which then amountedto only 34 mm in the area of the tracks of the Gäubahn (Fig.4.68). The magnitude of this value is in reasonable agreement withthe subsidence of 27 mm computed under the assumption of an earlyclosing of the invert for the analysis cross-section 2 (see Fig.4.63).
Fig. 4.68: Measured ground surface subsidence, longitudinalsection through the tunnel axis
In the further course of the heading the tunneling-induced subsi-dence decreased to very low values due to the more favorable geo-technical conditions (Beiche and Kagerer, 1993).
During the construction of the Stuttgart city railway tunnel toBotnang, the tracks of the Gäubahn and built-up areas had to beundercrossed (see Fig. 4.49). Since in the area of the undercross-ing of the Gäubahn the ground had a high deformability and a lowstrength, special measures had to be taken to limit the groundsurface subsidence due to tunneling and to avoid interference withrailway operations and damage to the buildings.
A crown, bench and invert heading with closed support at the in-vert and a following lowering of the invert was chosen. With anearly closing of the invert the tunneling-induced subsidence couldbe limited to admissible values. It was possible to reduce theloading at the transitions from the sidewalls to the temporary in-vert decisively by rounding the temporary crown invert. The load-ing of the high sidewalls after the enlargement of the cross-section could be clearly reduced by a systematic anchoring (seeFig. 4.65).
With these measures it was possible to limit the ground surfacesubsidence during the undercrossing of the Gäubahn to an admissi-ble value of some 3 cm. Railway operations were not interferedwith and damage to the structures undercut and to the railway fa-cilities did not occur.
The FE-analyses contributed essentially to the specification ofthe excavation and support measures, such as the early closing ofthe invert, the curvature of the temporary crown invert as well asthe systematic anchoring of the sidewalls.
5.1 Road tunnel "Hahnerberger Straße" in Wuppertal, Germany
5.1.1 Introduction
To create a high-capacity and therefore non-intersecting east-westconnection between the freeways A46 and A1, it was planned tonewly construct and improve the state highway L418 running throughthe city of Wuppertal, Germany, as a four-lane road. Within thescope of this construction project, the Hahnerberger Straße, astreet running approximately in north-south direction, was to beundercrossed in the city district of Hahnerberg at an angle of 60°by underground construction (Fig. 5.1). This task proved to bequite difficult, since the tunnel has a low overburden and a largeexcavated cross-section. Also, a limitation to the tunneling-induced ground surface subsidence had to be kept due to the build-ings at the ground surface in the area of the tunnel alignment.
Fig. 5.1: Road tunnel Hahnerberger Straße, site plan and ex-ploration
The approx. 130 m long tunnel must provide room in each directionfor two main driving lanes and one exit or approach lane, respec-tively (Fig. 5.2 and 5.3). The widths of the required clearancesrange between 13.5 m and 16.2 m. The height of the clearancesamounts to 4.90 m. The tunnel has a total width of approx. 37 mand a total height of approx. 12 m (Fig. 5.3). To illustrate thesevery large dimensions, the cross-section of a double-tracked tun-nel for the new high-speed railway lines of German Rail (DeutscheBahn AG) is shown in Fig. 5.3.
The tunnel has an overburden height of approx. 10 m (Fig. 5.2). Toavoid damage to the buildings above the tunnel and to the waterand gas mains running along the Hahnerberger Straße, the subsi-dence due to tunneling was to be limited to a magnitude of 2 cm.
5.1.3 Exploration
Within the framework of route planning and in order to assess thefeasibility of underground construction, an investigation programwas carried out to explore the ground conditions and to determinethe rock mechanical parameters. Test pits were excavated and coredrillings were sunk with depths ranging between 10 and 34 m (Fig.5.1).
According to the exploration results, below an about 2 to 4 mthick layer of top soil and loam with cobbles, a narrowly beddedalternating sequence of sandstones and claystones of the MiddleDevonian Brandenberg layers is found. The sandstones andclaystones are weathered close to the surface and intenselyjointed. The orientation of the bedding planes and joints in therock mass was measured on oriented drill cores and in test pitsusing a geological compass. Fig. 5.4 shows the idealization of thediscontinuity fabric of the rock by a structural model (see Chap-ter 2.5.1). The rock mass is separated by bedding planes (B),which dip shallowly at approx. 30° and persist widely, and bythree steeply dipping main joint sets (J1 to J3). According to thedrilling results, the spacing of the bedding planes ranges betweenseveral centimeters and a few decimeters. The joint spacingamounts to a few decimeters on average.
The bedding planes are partially filled with clay and mixed-grained soils at a thickness of up to 10 to 40 cm. This leads to agreater deformability perpendicular to the bedding than parallelto it. After Wittke (1990), transversely isotropic deformation be-havior can be assumed in the elastic stress domain for an alter-nating sequence of this kind. This kind of anisotropy can be de-scribed by 5 independent elastic constants: Two Young's moduli E1and E2, one shear modulus G2 and two Poisson's ratios ν1 and ν2
(Fig. 5.4).
The modulus E1 relevant for loading parallel to the bedding was de-rived from the results of the pressuremeter tests and from experi-
ence gained from other projects with comparable ground conditionsas E1 = 1000 MN/m2 (Fig. 5.4).
Fig. 5.4: Structural model and rock mechanical parameters
The modulus E2, relevant for loading perpendicular to the beddingwhich is smaller than E1, was determined according to Wittke (1990)from the following relation:
2
BFIR
2 MN/m400
700.1
10000.9
1
EE
1E ≈
+
=β
+α
= (5.1)
The symbols in (5.1) denote:
α = 0.9: Fraction of the sandstone in the alternatingsequence.
β = 0.1: Fraction of the bedding plane filling in thealternating sequence.
EIR = 1000 MN/m²: Young's modulus of the jointed sandstone withthe bedding plane filling not taken into ac-count.
EBF = 70 MN/m2: Bulk modulus of the bedding plane filling.
α and β were derived from a statistical evaluation of the mappingresults of the drill cores and test pits. Taking the joints in thesandstone into account, the modulus E1 derived from the pressure-meter test results was chosen to EIR as represented above. The bulkmodulus of the bedding plane filling EBF was estimated on the basisof the results of soil mechanical laboratory tests.
The shear modulus G2, relevant for shear loading parallel to thebedding and thus strongly dependent on the mechanical propertiesof the bedding plane filling, was estimated at G2 = 325 MN/m2.
Poisson's ratio ν1 corresponds approximately to Poisson's ratio ofthe sandstone. In unconfined compression tests on intact rockspecimens a value of νIR = 0.25 resulted on average for the latter.This value was taken as a basis for the analyses.
According to Wittke (1990), Poisson's ratio ν2 can be computed asfollows:
ν2 = IRBF
IRBF
EEE
⋅β+⋅α
ν⋅ = 0.1
10000.1700.90.2570
≈⋅+⋅⋅
⋅(5.2)
Laboratory tests resulted in very high values for the shear pa-rameters of the intact rock. The failure behavior of the rock isthus essentially determined by the shear strength along the dis-continuities, which was modeled by the Mohr-Coulomb failure crite-rion. For the shear strength parallel to the bedding, the beddingplane filling is relevant. On the basis of the grain-size distri-bution and water content and of experience, a friction angle ofϕB = 25° and no cohesion were assumed (Fig. 5.4).
Unlike the bedding planes, the joints are mostly undulating andcontain sandy, rusty coatings. Close to the surface, however, theyare also partially filled with clayey, sandy silt. It is essentialfor the assessment of their shear strength that they mostly onlyextend through one layer and thus extend considerably less farthan the bedding. Therefore, a cohesion of cJ ≈ 0 to 0.02 MN/m2 anda friction angle of ϕJ ≈ 30 to 35° were assumed for the joints (J1to J3) (Fig. 5.4).
A tensile strength normal to the discontinuities was not accountedfor.
The water permeability of the rock is relatively low perpendicu-larly to the bedding, because the bedding plane filling has asealing effect. The groundwater table is located approx. 2 m abovethe tunnel roof.
To further explore the rock mass conditions and to test theplanned construction method, an adit was driven in the middle be-tween the two tunnel tubes as an advance construction measure(Fig. 5.5). In the course of construction, a reinforced concretebuttress was installed in the adit. The additional ground exposureby the adit confirmed the findings about the ground which had beengained in the first exploration phase.
Fig. 5.5: Central adit for additional exploration and fortesting of construction techniques
The adit was driven by means of smooth and true to profile blast-ing. To keep the blasting-induced vibrations low, the roundlengths and the charges per ignition step were limited. Especiallyclose to the buildings, the excavation profile had to be subdi-vided for the same reasons (I and II in Fig. 5.5). This experiencecould be used for the further planning and tendering of the tun-nel.
During the heading displacements were measured in the adit and atthe ground surface and interpreted using FE-analyses based on therock mechanical parameters given above (back analysis). The meas-ured and the computed displacements were in substantial agreement.The derived parameters were thus confirmed and remained unchangedin the further course of the design (Modemann and Wittke, 1988).
5.1.4 Design and construction
In the conditions present here, only the NATM is suitable for anunderground construction of the tunnel. In view of the width ofthe tunnel structure of approx. 37 m and of the overburden of ap-prox. 10 m, which is very low by comparison, it was necessary toprovide for one tunnel tube for each direction. For reasons of thealignment only a width of approx. 1 to 2 m was available for thebuttress between the two tunnel tubes. The option to leave a rockpillar between the tubes was thus ruled out. Therefore, as men-tioned above already, a 1.2 m thick reinforced concrete buttresswith a mushroom-shaped widening of the head and a base enlargementwas constructed in the exploration adit driven in advance (Fig.5.6 and 5.7). The two tunnel tubes thus formed still have a verylarge span compared to the overburden.
To enable at least a slight an arching effect in the remainingrock mass above the tunnel tubes, the tunnel profiles were de-signed with a very small camber above the prescribed clearance.The rounding of the inverts served to carry the water pressure onthe interior lining better. Only a slight rounding was necessaryhere in spite of the wide span, because due to the small height ofthe clearance sufficient space was available to strengthen the in-vert of the interior lining.
Since the groundwater table could be lowered during the heading,the shotcrete support did not have to be dimensioned for waterpressure. The interior lining, however, was to be constructed wa-
tertight. By request of the client, a PVC sealing was planned be-tween the shotcrete membrane and the interior lining (Fig. 5.6).As a consequence of the shape of the cross-section, each tunneltube was provided with a separate sealing, which did not includethe central buttress. The arrangement and installation of thesealing was considerably simplified this way.
Fig. 5.6: Design
The reinforced concrete buttress was constructed in segments of9 m each. Connecting reinforcement was provided for at the headand at the base for the shotcrete support of the two tunnel tubesto be constructed later (Fig. 5.7). Further, a sound load transferbetween the concrete buttress and the rock mass above was effectedby means of a contact injection. After that, the tunnel tubes wereexcavated in parts of the cross-section for reasons of stability,limitation of the heading-induced subsidence and vibrations.
The first step was the excavation and support of a sidewall aditin the northern tube. The cross-section of the sidewall adit wasagain subdivided by a crown heading with trailing invert (Fig. 5.8and 5.9). As illustrated in Fig. 5.8, the outside sidewall wassupported by a lattice girder and a 30 cm thick shotcrete membraneas well as by SN-anchors. The side of the sidewall adit facing thetunnel tube, on the other hand, was supported by only 15 cm ofshotcrete. Glass fiber anchors were further installed on thisside. This part of the support was removed again in the furthercourse of the works.
In the third step, the crown of the northern tube was excavatedand supported in sections (Fig. 5.10 and 5.11). The lattice gird-ers and the reinforcement of the shotcrete membrane were connected
with the corresponding support elements of the sidewall adit inthe process. The connection of the support to the central buttresswas already mentioned above (see Fig. 5.7). The length of the SN-anchors was increased in the vault area to improve the arching ef-fect in the rock above the tunnel roof.
The excavation of the bench and the invert constituted the laststeps of the excavation sequence of the northern tube (Fig. 5.12and 5.13). The invert was supported by 30 cm of shotcrete. Latticegirders and an invert anchoring could be dispensed with.
Fig. 5.12: Bench and invert excavation with closed invert sup-port
The stepwise excavation and installation of the support measuresand the short round lengths of approx. 1 m, specified for stabil-ity reasons and to limit the blasting vibrations, lead to a com-paratively low heading performance. This had the advantage thatthe shotcrete membrane was only fully loaded when the shotcretehad mostly reached its final strength. The same applies to the SN-anchors.
To keep the heading-induced subsidence as small as possible, theinterior lining of the northern tube was to be installed beforethe southern tube was excavated. For reasons of construction man-agement the contractor was allowed, however, to carry out the ex-cavation and support of the southern sidewall adit in parallelwith the concreting works in the northern tube. The sequence ofconstruction in the southern tube was analogous to the northerntube (Modemann and Wittke, 1988).
5.1.5 Stability analyses for the stages of construction
A rock slice with a width of 65 m and a height of 37 m was speci-fied as computation section for the stability analyses for theconstruction stages (Fig. 5.14). The FE-mesh was subdivided into440 three-dimensional isoparametric elements with 1140 nodes.
Sliding supports were specified as boundary conditions for thenodes of the lower boundary plane (z = 0).
Since the rock mass is anisotropic and the bedding is neither ori-ented perpendicularly nor parallel to the tunnel axis (see Fig.5.4 and 5.14), displacements in x- and y-direction already occurdue to the dead weight of the rock mass. Therefore the nodes onthe vertical boundary planes of the computation section must notbe fixed perpendicularly to the respective boundary plane.
The displacements in x- and y-direction resulting from the deadweight of the rock mass were determined by an advance analysis us-ing a column-like computation section. These displacements wereintroduced as boundary conditions for the nodes on the verticallateral boundary planes (x = 0 and x = 65 m) (Wittke, 2000).
For the two planes normal to the tunnel axis equal displacementswere assumed as boundary conditions for opposite nodes with equalx- and z-coordinates (Wittke, 2000).
The stability analyses were carried out with the program systemFEST03 (Wittke, 2000) in 4 computation steps: In the 1st computa-tion step, the stresses and deformations in the rock mass result-ing from the dead weight of the rock (in-situ state) are deter-mined. In the 2nd computation step, the central adit as well as thenorthern sidewall adit are simulated to be excavated and sup-ported. The 3rd computation step further accounts for the installa-tion of the reinforced concrete buttress in the central adit andthe crown excavation in the northern tunnel tube. Then, in the 4th
computation step, the stresses and deformations occurring afterthe complete excavation and support of the northern tunnel tubeare computed.
The principal normal stresses determined for the 2nd computationstep show clearly how the loads resulting from the overburdenweight are diverted around the central adit and the sidewall adit(Fig. 5.15a). A mutual influence of the two excavations is not ap-parent.
In the 4th computation step the loads resulting from the overburdenweight are diverted around the entire excavation. Above and belowthe tunnel unloaded areas develop, whereas stress concentrationsoccur in the rock mass beside the northern tunnel tube as well asabove and below the central buttress (Fig. 5.15b and 5.16).
Fig. 5.16: Normal stresses in a horizontal and a vertical sec-tion, 4th computation step
According to the results of the analyses, the dead weight of theoverburden is supported to a large degree by the rock mass besidethe tunnel and by the central buttress. The resulting stress dis-tribution in the concrete buttress is shown in Fig. 5.17. It be-comes apparent that a horizontal tension loading results for themushroom-shaped buttress head, whereas the shaft is loaded by nor-mal thrust and bending.
Fig. 5.17: Stresses in the central buttress, 4th computationstep
An appreciable bending loading of the shotcrete membrane only oc-curs at the connection to the central buttress (Fig. 5.18). A con-necting reinforcement is provided for to carry this bending load-ing (Fig. 5.18).
The maximum computed displacements due to the excavation of thenorthern tunnel tube amount to 22 mm at the tunnel's roof and to8 mm at the ground surface (Fig. 5.19).
An interpretation of the loading of the shotcrete membrane and thecentral buttress shows that due to the arching in the rock massabove the tunnel (see Fig. 5.15) only a portion of the overburdenweight is carried by the shotcrete membrane. To determine whetherthe shotcrete membrane and the central buttress can carry theoverburden weight alone, an analysis in which no arching can de-velop in the rock mass above the tunnel was carried out within theframework of safety considerations. For this purpose, above andbeside the tunnel tube a vertical discontinuity set was assumed inthe rock mass, striking parallel to the tunnel axis and having noshear strength (ϕ = 0, c = 0). Shear stresses thus cannot betransferred across these discontinuities. The principal normalstresses and stress trajectories determined for this case showclearly that the weight of the overburden must be completely car-ried by the shotcrete membrane and the central buttress (Fig.5.20).
Fig. 5.20: Principal normal stresses and stress trajectorieswithout arching in the rock mass, 4th computationstep
A comparison of the normal thrust in the shotcrete membrane com-puted for this case with the ones of the previously investigated
case, in which arching was possible in the rock mass above thetunnel, shows that the normal thrust increases considerably (Fig.5.21). Also for this case, however, it was possible to dimensionthe shotcrete membrane taking into account the inside and outsidesteel fabric mats Q221.
Fig. 5.21: Normal thrust in the shotcrete membrane with andwithout arching in the rock mass above the tunnel,4th computation step
5.1.6 Stability analyses for the design of the interior lining
A number of stability analyses with varying assumptions was car-ried out for the design of the interior lining of the two tunneltubes. Among other things it was assumed that the shotcrete mem-brane becomes ineffective over time.
A corresponding assumption was made in some analyses for the cen-tral buttress as well, since it is not protected from the ground-water by the PVC sealing. Furthermore, cases with and withoutarching in the rock mass were investigated.
The result of merely one analysis shall be presented here, inwhich the interior linings are only loaded by their dead weightand by the water pressure. This case is relevant as long as theshotcrete membrane remains intact and carries the rock mass pres-sure. The water pressure is applied to the linings in the form ofequivalent nodal forces (Fig. 5.22, detail I). The bedding of theinterior lining is simulated by truss elements arranged betweencorresponding nodes of the shotcrete membrane and the interiorlining and by the behavior of the shotcrete membrane and the rockmass, which is assumed elastic (Fig. 5.22, detail II). The trusselements possess a very high stiffness. They can only transfercompressive stresses, but not tensile or shear stresses. In thisway the lack of a shear bond between the interior lining and theshotcrete membrane due to the PVC sealing is simulated. Since therock mass pressure is not to be taken into account for the loadcase "dead weight and water pressure", the rock mass is assumedweightless.
The computed bending moment distribution represented in Fig. 5.23shows clearly the bending loading of the inverts and the centralbuttress resulting from the water pressure. The corresponding dis-tribution of the normal thrust is shown in Fig. 5.24.
Fig. 5.23: Bending moments in the interior lining, load case"dead weight and water pressure"
Fig. 5.24: Normal thrust in the interior lining, load case"dead weight and water pressure"
5.1.7 Monitoring
A measuring program was planned to monitor the stability as wellas the subsidence occurring at the ground surface and at thebuildings. It included leveling and convergency measurements inthe tunnel. In addition, extensometer measurements as well aslevelings at the ground surface and at buildings were carried outduring construction. Further, the groundwater levels in observa-tion wells located beside the tunnel (see Fig. 5.1) were read con-tinuously, and the blasting-induced vibrations were measured atthe buildings. Of course, evidence was perpetuated at the build-ings before the start and after the end of construction.
The subsidence measured at a building located right above the tun-nel is shown exemplarily in Fig. 5.25 for different stages of con-struction. It can be seen that the subsidence of the building atthe end of construction is distinctly less than 2 cm, and that thebuilding has subsided almost evenly. In the course of the differ-ent construction stages only small differential subsidence of thebuildings occurred as well. No damage was found on this or thesurrounding buildings (Modemann and Wittke, 1988).
The road tunnel Hahnerberger Straße is a tunnel with a very largespan and low overburden, for which the requirement was made tokeep the heading-induced subsidence at the ground surface and vi-brations as small as possible.
This task was solved with the following measures:
- Construction of two tunnel tubes with a reinforced concretebuttress in the middle providing additional support of therock mass,
- subdividing the cross-section of both tunnel tubes into sev-eral parts,
- limiting the round length to approx. 1 m,
- rounding the cross-sections of the two tunnel tubes in thevault area to enable the formation of a vault in the shotcretemembrane, the interior lining and the rock mass,
- designing the shotcrete membrane and the reinforced concretebuttress in such a way that they can carry the complete rockmass pressure,
- installation of the interior lining after the first tube wasexcavated and before the excavation of the second tubestarted,
- smooth blasting.
Essential for the success of the construction project were fur-thermore the appropriate characterization of the ground and theproofs of stability established using the FE-program developed bythe authors and their co-workers.
5.2 Limburg Tunnel, Germany
5.2.1 Introduction
The new railway line Cologne-Rhine/Main undercrosses a businessarea of the city of Limburg in a approx. 2.4 km long tunnel be-tween chainage km 109+680 and km 112+075.
In addition to business and industrial structures, the tunnel un-dercrosses among other things the freeway (Autobahn) A3 and thehighway B 49 (Fig. 5.26 and 5.27). With overburden heights from 10to 30 m it crosses through varying and geotechnically difficultground in the form of predominantly weathered to decomposed slateas well as Tertiary and Quaternary clay, silt and sand (Fig.5.27).
For reasons of stability the excavation profile had to be subdi-vided during tunnel heading. In most parts a sidewall adit excava-tion with additional tunnel face support measures was carried out.
5.2.2 Structure
In the portal areas and in a short middle section (intermediatestarting point), the Limburg Tunnel was constructed by the cut-and-cover method. The sections in between, approx. 1190 m and1090 m in length, were excavated by underground construction (Fig.5.27).
Seen from the direction of Cologne, the tunnel undercrosses firstthe sports grounds "Im Finken" with an overburden of approx. 16 m.Next, it undercrosses the freeway A3 and the traffic lanes of thefreeway exit Limburg North at an acute angle with an overburdenvarying between 10 and 20 m. Between chainage km 110+750 andkm 110+810, the highway B 49 and its underpass of freeway A3 areundercrossed with a small roof cover of approx. 4 m. Afterwards,the tunnel crosses under parts of the Massa shopping center andunder the county road K 472 with an overburden of approx. 15 to19 m. At chainage km 111+680 the tunnel undercrosses a high baywarehouse and between chainage km 111+800 and km 111+870 a ware-house of Tetra-Pak Co. (Fig. 5.26).
Fig. 5.28 shows the geometry of the standard profile and of thesidewall adits. The double-tracked tunnel was constructed with amouth-shaped profile with a width of 15.2 m and a height of 12.4m. In the vault area a radius of curvature of R = 7.28 m was se-lected. The transition from the sidewalls to the invert was con-structed with a radius of R = 4.43 m. For statical reasons the in-vert was deeply rounded with R = 11.63 m. The inside walls of thesidewall adits had a radius of curvature of R = 8.0 m. The roofand the invert of the sidewall adits were rounded with small radiiof R = 0.4 m (Fig. 5.28).
The excavated cross-section amounts to approx. 150 m2. The sidewalladits as well as the remaining cross-section (core) were subdi-vided for the excavation into crown, bench and invert (Fig. 5.28).
The shotcrete membrane has a thickness of 35 cm. The thickness ofthe inside shotcrete membranes of the sidewall adits amounts to30 cm. The interior lining is 40 cm thick (Fig. 5.28).
Fig. 5.29 shows the start of underground excavation at the north-ern portal.
The ground at the alignment of the Limburg Tunnel was explored bycore drillings, some of which were equipped as observation wells.
In the area of the northern portal the tunnel is located in theTertiary clays of the western sunken block of the Limburg rift.This formation consists predominantly of silty clay with interca-lated 1 to 9 m thick sand and gravel layers (Fig. 5.27).
At about chainage km 109+915 follows the middle horst block of theLimburg rift, consisting at first of slate decomposed to sandy andclayey silt. At the core of the uplifted block, solid Devonianslate is encountered. The slate has been strongly loaded in termsof fracture tectonics. It is strongly weathered in the upper zoneup to a depth of approx. 45 m.
At approx. chainage km 110+250, the tunnel leaves the upliftedblock and crosses through the central sunken block of the Limburg
rift up to about chainage km 111+640. As in the western sunkenblock, predominantly Tertiary clays are encountered here.
At about chainage km 111+640, in the area of the high bay ware-house of Tetra Pak Co., the tunnel reaches the uplifted block ofthe Greifenberg horst. On the first approx. 70 m, the tunnel islocated partially in decomposed keratophyre tuff and partially instrongly weathered, in some areas also completely decomposedslate, into which the tunnel enters completely in the followingsection. From about chainage km 111+910 to the southern portal,the tunnel crosses through slightly weathered to decomposed slate(Fig. 5.27).
The foliation represents the dominant discontinuity set in the De-vonian slate. It strikes from southwest to northeast and dipsmostly at 40 to 70° in southeastern direction. Because of thetight (isoclinal) folding, however, foliation planes dippingsteeply to the northwest occur as well. In connection with twosteeply dipping joint sets, an approximately orthogonal disconti-nuity fabric exists in most cases.
To determine the deformability of the ground, 9 dilatometer testsand 40 borehole jack tests were carried out in exploration bore-holes located in the area of the Limburg Tunnel. According tothese tests, a modulus of deformation for initial loading of 200to 1000 MN/m2 can be assumed for slightly weathered slate. In de-composed slate, however, comparatively small moduli of deformationfor initial loading of 5 to 80 MN/m2 were determined. Even smallervalues of 5 to 60 MN/m2 resulted for the soil.
For 63 samples from the clayey and silty surface layers as well asfrom the decomposed rock the unconfined compressive strength wasdetermined. The unconfined compressive strength of the clay andsilt of the surface layers scatters over the relatively wide rangeof 50 to 1900 kN/m2. The results of the tests on decomposed rock,however, are in the range of 100 to 500 kN/m2. The strength of theslightly to strongly weathered or decomposed slate is substan-tially determined by the low strengths on the discontinuities (seeChapter 5.2.6).
Between the northern portal and the eastern boundary of the middlehorst block, the groundwater table follows the course of theground surface at a depth of approx. 5 to 10 m (Fig. 5.27). In the
area of the central sunken block of the Limburg rift, the ground-water table is located at a depth of about 25 to 45 m at the levelof the two receiving streams, the Elbach and the Lahn. To the eastof the central sunken block, at the Greifenberg horst, the ground-water table rises again to approx. 5 to 10 m below the ground sur-face (Fig. 5.27). The seasonal variations in the water table canamount to several meters. The groundwater table shown in the lon-gitudinal section (Fig. 5.27) is based on the highest groundwaterlevels measured in the boreholes equipped as observation wells.The groundwater is assumed to flow roughly from northeast tosouthwest.
The water permeability of the Devonian rock and the Tertiary lay-ers is estimated at 10-7 to 10-5 m/s.
5.2.4 Excavation and support
In the areas of the Greifenberg horst and the middle horst block,in which the tunnel cross-section is located in slightly weatheredslate, a crown excavation with closed invert was planned accordingto excavation classes 5A-K, 6A-K or 7A-K, depending on the degreeof weathering. In the other much longer sections the tunnel is lo-cated in soil or in decomposed slate. Here, a sidewall adit exca-vation was planned according to excavation classes 4A-U, 5A-U, 6A-U or 7A-U (see DGGT, 1995: Table 1).
Fig. 5.30 shows the sequence of excavation and the support for ex-cavation class 7A-U-0, which was carried out for the most part.Excavation class 7A-U-0 is characterized by short round lengthsfor crown and bench (A = 0.6 m to 0.8 m), tunnel face support withplain shotcrete (t ≥ 7 cm), advance support with spiles and earlyclosing of the invert (C ≤ 3.2 m). The tunnel profile is supportedby a reinforced shotcrete membrane with two layers of steel fabricmats Q285 and by steel sets spaced at ea = 0.6 to 0.8 m. A system-atic anchoring of the sidewall adits with SN-anchors was plannedon the outside and as required also on the inside.
Just as the excavation of the advancing sidewall adits, the exca-vation of the core is characterized by a short round length at thecrown, tunnel face support with plain shotcrete and advance sup-port with spiles (Fig. 5.31). The round lengths at bench and in-vert amounted to B = 2.8 to 3.2 m. The support was closed at the
The sidewall adit excavation north (Fig. 5.32 to 5.35) started atfirst with the excavation class 7A-U-0 with the support beingclosed at the invert after 4 rounds (C = 4A ≤ 3.2 m, see Fig.5.30).
Fig. 5.33: Excavation of the crown of the sidewall adit
After 160 m, as the heading reached the decomposed slates, whichwere fractured to small sizes, water entered more intenselythrough joints and foliation parallel discontinuities within thearea of the tunnel face. This lead to stability problems at thevertical tunnel face and thus impeded the heading.
Starting at chainage 193 m the excavation sequence was thereforemodified. The step between crown and bench excavation was extendedand the bench face was steepened. The support was closed at theinvert at a distance of 6 rounds behind the crown excavation (Fig.
5.36). In this manner a tunnel face more shallowly inclined on av-erage was achieved. In addition, the safety of the staff duringclosing of the invert was increased.
Fig. 5.34: Excavation of the bench of the sidewall adit
As stability problems continued at the tunnel face, an inclinedcrown and bench face supported with plain shotcrete was con-structed starting at chainage 267 m. In addition, a reinforcedshotcrete support was installed at the invert of the temporarybench of the sidewall adits to achieve a rapid stabilization ofthe excavation. Furthermore, the final closing of the support atthe invert was carried out separately for the crown and benchheading. After the excavation and support of the crown and thebench of the sidewall adits had been completed ≤ 20 m in advance,the heading was interrupted and the tunnel face was sealed withplain shotcrete. Subsequently the excavation and support of theinvert was carried out (Fig. 5.37). To drain off the water inflowthrough joints and foliation parallel discontinuities in the tun-nel face area, a drainage was laid in the invert of the bench ofthe sidewall adits and a shaft sump reaching down to below thetunnel's invert was constructed every 10 to 20 m.
Fig. 5.37: Inclined tunnel face and support of the invert ofthe temporary bench of the sidewall adits, excava-tion north, chainage 267 to 336 m
At chainage 336 m the heading was changed back again to excavationclass 7A-U-0, since the strength of the rock mass increased andthe jointing as well as the water inflow decreased.
Fig. 5.38 is a photograph of the core excavation at excavationnorth.
5.2.6 Stability analyses for sidewall adit excavation north
The problems with the tunnel face stability that had occurred dur-ing the sidewall adit excavation north in the decomposed slate ofthe middle horst block (see Fig. 5.27) were attributed to the lowshear strengths on the discontinuities of the strongly weatheredto decomposed slate. The foliation and bedding parallel disconti-nuities F the orientations of which were measured during tunnelface mapping, strike approximately perpendicularly to the tunnelaxis and dip mostly steeply with dip angles between 40° and 70°towards the tunnel face. In addition, two joint sets J1 and J2 ex-ist, which strike in parallel and perpendicularly to the tunnelaxis and dip steeply as well (Fig. 5.39).
The water pressure acting in the foliation and bedding disconti-nuities, respectively, and in the joints had a further unfavorableeffect on the stability of the tunnel face.
Three-dimensional stability analyses were therefore carried outusing the program system FEST03 (Wittke, 2000) for the sidewalladit excavation in the area of the middle horst block to investi-
Fig. 5.40 shows the three-dimensional computation section, the FE-mesh and the parameters the analyses were based upon. One sidewalladit was modeled. The overburden amounts to 23 m.
In order to investigate the influence of the shear strengths onthe discontinuities on the tunnel face stability three three-
In case A discontinuities were not taken into account. In cases Band C the foliation/bedding F, dipping at 70° towards the tunnelface, was simulated, together with the joint set J1, vertical andstriking parallel to the tunnel axis, and the joint set J2, dip-ping at 60° opposite to the direction of heading (see Fig. 5.40).In case B the friction angle on the discontinuities was chosen asϕF/J = 20° and in case C as ϕF/J = 10° (Table 5.1). No cohesion wasassumed on the discontinuities. Young's modulus of the slate wasspecified as E = 50 MN/m2 in all cases (Fig. 5.40).
The heading of a sidewall adit with crown, bench and invert exca-vation was simulated in 10 computation steps, which are outlinedin Fig. 5.41. In the 11th computation step the seepage forces dueto the water seeping through the rock mass were applied (Fig.5.41), which had been determined in a three-dimensional seepageflow analysis using the program system HYD03 (Wittke 2000). Thisseepage flow analysis results in the distribution of piezometricheads h, from which the seepage forces can be calculated. Theanalysis was based on an undisturbed groundwater table located16 m above the roof of the sidewall adit. Fig. 5.42 shows the lo-cation of the groundwater table lowered due to the tunnel excava-tion, the computed equipotential lines (h = const.), as well as,qualitatively, the direction and magnitude of the seepage forces FSoriented perpendicularly to the equipotential lines.
The slate was simplificatively assigned a homogeneous andisotropic permeability with a permeability coefficient ofkf = 10-6 m/s. The distribution of the piezometric heads and theseepage forces is, however, independent of the selected permeabil-ity coefficient, since a homogeneous and isotropic ground was as-sumed for the analysis (Wittke, 2000).
Fig. 5.43 shows the development of the computed horizontal dis-placement δH of a point on the unsupported tunnel face in thecourse of the viscoplastic iterative analysis for computationsteps 10 and 11.
Case A results in a stable tunnel face with a maximum horizontaldisplacement of δH ≈ 3 cm. In case B the displacements are largerthan in case A, with max. δH ≈ 6 cm in the 10th computation step andmax δH ≈ 10 cm in the 11th computation step (with seepage pressure).The tunnel face is stable, however, since the displacements con-verge in the course of the viscoplastic iterative analysis. Case C
results in a maximum displacement of δH ≈ 22 cm for the 10th compu-tation step and of δH ≈ 35 cm for the 11th computation step. Al-though the displacement δH converges in this analysis as well, thetunnel face cannot be regarded stable anymore due to the magnitudeof δH. For a horizontal displacement of 35 cm it must be assumedthat the rock loosens up and disintegrates in the tunnel facearea.
Fig. 5.43: Horizontal displacement of the tunnel face dependingon the shear strength on the discontinuities
In summary it can be stated on the basis of the FE-analysis re-sults that the stability of the tunnel face of the sidewall aditis substantially influenced by the discontinuity fabric. For shearparameters of cF = cJ1 = cJ2 = 0 and ϕF = ϕJ1 = ϕJ2 = 10°, which hadbeen estimated from the results of the geotechnical mapping duringheading, the tunnel face must be considered unstable. The analysisresults further show that for low shear strengths on the disconti-nuities the seepage pressure acting on the tunnel face has astrong influence on the magnitude of the horizontal displacements.If follows that by an efficient drainage of the rock in advance ofthe sidewall adit excavation the problems during the heading couldhave been reduced decisively.
5.2.7 Monitoring results
The heading of the Limburg Tunnel was accompanied by a geotechni-cal monitoring program including surface leveling, extensometerand inclinometer measurements as well as leveling and convergencymeasurements in the tunnel.
Fig. 5.44 shows exemplarily the vertical displacements of thesidewall adit roofs δSL and δSR and of the tunnel roof δR measuredafter the excavation of the entire cross-section in excavationnorth from chainage 160 to 375 m (km 109+900 to km 110+115). Themeasured vertical displacements of the roofs of the two sidewalladits are larger than the one of the roof of the total cross-section, because with the latter only a part of the displacementsoccurring during the core excavation can be captured. The measureddisplacements of the sidewall adit roofs include a part of thesubsidence that occurred during the sidewall adit heading and inaddition the subsidence resulting from the core excavation.
The maximum subsidence of δSL = 45 mm was measured at chainage300 m (Fig. 5.44). In this area the sidewall adit was excavatedwith an inclined tunnel face and with a supported invert of thetemporary bench (see Fig. 5.37). From chainage approx. 340 m onthe measured subsidence decreased to a few millimeters. In thisarea the sidewall adit excavation was changed over to excavationclass 7A-U-0 (see Fig. 5.30). Starting with chainage approx. 360 mthe roof was located in slightly weathered to unweathered slate.Here, the measured roof subsidence decreased to ≤ 3 mm (Fig.5.44).
With the Limburg Tunnel of the new railway line Cologne/Rhine-Mainstability problems often occurred at the tunnel face during thesidewall adit excavation north in the area of the middle horstblock. These problems impeded the heading and required intensifiedsupport measures. In this section the tunnel cross-section was lo-cated in strongly weathered slate which is strongly jointed tofractured to small sizes. The groundwater table was encounteredapprox. 15 m above the roof.
By three-dimensional FE-analyses the influence of the shearstrength on the discontinuities of the slate and of the seepagepressure in the rock mass on the stability of the tunnel face wasinvestigated. The results show that the tunnel face stability issubstantially determined by the discontinuity fabric. It couldfurther be proven that the seepage pressure resulting from theseepage flow in connection with the low strength on the disconti-nuities caused the stability problems during the sidewall adit ex-cavation north. The impediments during the heading and the inten-sified support measures could have been considerably reduced by anefficient drainage of the rock mass in advance of the heading.
5.3 Niedernhausen Tunnel, Germany
5.3.1 Introduction
The Niedernhausen Tunnel lies in the southern part of the newrailway line Cologne-Rhine/Main in the area of the town ofNiedernhausen, south of Idstein, Germany.
Over a length of approx. 350 m following the northern portal thetunnel is located in the completely weathered and decomposedslates of the Schwall layers with an overburden of about 17 to50 m. Further, the groundwater table lies at roof level or abovein this area and the freeway A3 had to be undercrossed (Fig.5.45). Under these conditions the tunnel face was not stable with-out supplementary support measures. Moreover, the tunneling-induced subsidence of the ground surface in the area of the under-crossing of the freeway A3 had to be kept small.
The high demands and the difficult ground conditions made it nec-essary to drain the rock in advance and to construct the tunnel inthis area in partial excavations with additional tunnel face sup-port measures.
The two-tracked Niedernhausen Tunnel is 2765 m long (Fig. 5.45).From chainage km 140+867 to km 141+499 as well as in the southernportal area the tunnel was constructed by the cut-and-covermethod. The overburden in this area amounts to less than 10 m.Over the remaining 2101 m the tunnel was excavated by undergroundconstruction. The maximum overburden height is 95 m (Fig. 5.45b).
Fig. 5.46: Niedernhausen Tunnel, standard profile with sidewalladits
Between chainage km 141+600 and km 141+700 the tunnel undercrossesthe federal highway A3 at an acute angle with an overburden of 20to 30 m. The tunnel roof lies in the hillside loam in this area.The remainder of the cross-section is located in the already men-tioned deeply weathered and decomposed Schwall layers (Fig.5.45b). A sidewall adit heading was planned in this area.
Fig. 5.46 shows the 15.7 m wide and 12.6 m high standard profilewith the geometry of the sidewall adits. The excavated cross-section amounts to approx. 150 m2.
The vault area was constructed with a radius of curvature of R =7.25 m. In the sidewalls a radius of R = 9.05 m was selected. Thetransitions from the sidewalls to the invert were constructed witha radius of R = 2.8 m. The invert was rounded with R = 14.15 m.The inside walls of the sidewall adits had radii of curvature ofR = 6.375 m and R = 10.925 m, respectively. The roof and the in-vert of the sidewall adits were constructed with a small radius ofR = 1.0 m. The temporary crown inverts of the sidewall adits wererounded with R = 6.25 m (Fig. 5.46).
The shotcrete membrane was carried out unusually strong with athickness of 35 to 45 cm. The inside shotcrete membrane of thesidewall adits were constructed 30 cm thick. The thickness of theinterior lining amounts to 40 cm (Fig. 5.46).
The sidewall adits were subdivided into crown and bench for theheading. The remaining cross-section (core) was subdivided intocrown, bench and invert (Fig. 5.46).
Fig. 5.47 shows the starting wall of the northern heading at thenorthern portal.
Fig. 5.47: Niedernhausen Tunnel, starting wall of the northernheading
The ground at the alignment of the Niedernhausen Tunnel was ex-plored by core drillings. Some of these boreholes were equipped asobservation wells.
The Quaternary hillside loam and talus material extends to a depthof approx. 40 m in the northern tunnel section. In the southerntunnel area, however, the Quaternary surface layer is only 12 mthick at the most (Fig. 5.45b).
Below the Quaternary cover Devonian rock sequences follow (Fig.5.45b) consisting of silty and finely sandy slate (Schwall lay-ers), micaceous sandstone and quartzite with slate intercalations(Hermeskeil layers), quartzitic sandstone and quartzite with em-bedded sandy slate (Taunus quartzite) and phyllitic slate withsandstone and quartzite layers (Variegated Schist).
The slate of the Schwall layers is characterized by a far-reachingweathering which can extend to a depth of about 150 m. It corre-sponds in strength to a cohesive soil. Young's modulus, however,is roughly that of a well-compacted gravel sand. The discontinuityfabric has remained intact in spite of the weathering (Fig. 5.48).
Fig. 5.48: Deeply weathered, decomposed slate of the Schwalllayers
The bedding planes and foliation discontinuities strike in NE-SWdirection, which is typical for the Rhine schist mountains, anddip mostly steeply at 60 to 90° in varying directions. In combina-tion with two discontinuity sets usually an orthogonal discontinu-ity fabric exists.
To the south of the Taunus ridge upthrust, in the middle and thesouthern part of the tunnel, the rock mass is mostly unweathered.Individual steep bedding-parallel fault zones here exist (Fig.5.45b).
From the northern tunnel portal to about km 142+500 the groundwa-ter table follows roughly the course of the ground surface at adepth of 5 to 25 m. In the southern tunnel section it drops to ap-prox. 50 m below the ground surface due to a drinking water supplyfacility. This level is still up to approx. 40 m above the tunnelroof, however (Fig. 5.45b).
The seasonal variations in the groundwater table amount to up to10 m. The groundwater table shown in the longitudinal section(Fig. 5.45b) is based on the highest groundwater levels measuredin the boreholes equipped as observation wells.
5.3.4 Excavation and support
In the approx. 350 m long section in which the tunnel's cross-section is partially or completely located in the weatheredSchwall layers, a sidewall adit heading was planned according toexcavation classes 7A-U-0 and 7A-U-1, respectively (Fig. 5.49, seeDGGT, 1995: Table 1). The two excavation classes differ with re-gard to the unsupported round lengths, the spacing of the steelsets and the number of anchors per m2 (Fig. 5.49).
In both excavation classes the shotcrete membrane is reinforced bytwo layers of steel fabric mats Q295. As mentioned above, thethickness of the shotcrete membrane of 35 to 45 cm is unusuallystrong. The inside shotcrete membrane of the sidewall adits was tobe carried out with a thickness of at least 30 cm. Further, a tun-nel face support with plain shotcrete (t ≥ 7 cm) and an advancingsupport with spiles were planned (Fig. 5.49).
In the area of the undercrossing of the freeway the workspace forthe tunnel excavation was to be additionally supported by pipe um-brellas made of 14.5 m long pipes with an inclination of approx.10 % and an overlap of > 3 m. The pipe umbrellas were to be con-structed from niches (Fig. 5.50).
5.3.5 Three-dimensional stability analyses
Calibration of the analysis model (analysis cross-section AC 1)
In the course of the final design for the tunnel driven by side-wall adit excavation three-dimensional stability analyses werecarried out with the program system FEST03 (Wittke, 2000).
A total of four cases were investigated (cases A, B, C and D, seeTable 5.2).
To calibrate the analysis model and to verify the soil and rockmechanical parameters the analyses were to be based upon, the ver-tical and horizontal ground displacements measured at measuringcross-section MC 2 (Fig. 5.51) during the sidewall adit headingwere back-analyzed first. MC 2 (analysis cross-section AC 1) islocated at km 141+580 approx. 20 m in front of the undercrossingof freeway A3 (Fig. 5.51). To record the vertical and horizontaldisplacements resulting from the heading, leveling points were setup at the ground surface and 3 multiple extensometers as well as 2inclinometers were installed beside and above the tunnel. Zeroreadings were taken sufficiently ahead of the excavation. The ar-
rangement of these measuring devices is shown in Fig. 5.52 to-gether with the tunnel's cross-section.
Fig. 5.51: Longitudinal section, location of measuring cross-section MC 2 and analysis cross-sections AC 1 andAC 2
Fig. 5.53 shows the three-dimensional FE-mesh for the back-analysis of the measured displacements together with its main di-mensions, the boundary conditions and the modeled stratigraphy. Tokeep the computing time within justifiable bounds, symmetricground conditions as well as a horizontal ground surface and hori-zontal layers were assumed. It was thus sufficient to model onehalf of the tunnel's cross-section and the surrounding ground.
According to the assumed symmetry the two sidewall adits are simu-lated as being excavated parallel in time in the analyses. In re-ality the western sidewall adit heading ran ahead of the easternone by 10 to 20 m (g in Fig. 5.49). The influence of the simpli-fied simulation in the analysis on the final result is small, how-ever.
The overburden in the area of measuring cross-section MC 2 amountsapprox. 17 m. The boundary between the soil and the stronglyweathered rock mass was modeled in the FE-mesh at about half theheight of the tunnel's cross-section.
The characteristic parameters of the soil include a Young'smodulus of 10 MN/m2 and shear parameters of ϕ' = 25° andc' = 7.5 kN/m2. For the weathered rock mass, the corresponding val-ues are E = 20 MN/m2 to 40 MN/m2, ϕ = 25° to 27.5° and c = 5 kN/m2
to 10 kN/m2 (Fig. 5.53). Thus both layers differ essentially intheir deformability and only negligibly with respect to theirshear strength.
Fig. 5.53: Computation section, FE-mesh, boundary conditions,ground profile and parameters for the three-dimensional analyses (AC 1 and AC 2)
The outside shotcrete membrane was constructed 45 cm thick in thecourse of the sidewall adit heading, and it is modeled accordingly(Fig. 5.54). The shotcrete membrane of the inside walls of thesidewall adits is 30 cm thick. The shotcrete support of the tempo-rary crown invert of the sidewall adits shown in Fig. 5.54 was notinstalled in the area of MC 2. It was therefore not modeled inanalysis case A for the back-analysis of the measured displace-ments (see Table 5.2).
The computation sequence of the analyses comprising a total of 36computation steps is shown schematically in Fig. 5.55. The analy-sis of the in-situ state (1st computation step) is followed in the2nd computation step by the excavation of a 23 m long crown sectionof the sidewall adits together with an invert section trailing by2 m. The 2nd computation step is the starting stage for the actualheading simulation carried out in computation steps 4 to 36 ac-cording to the iteration technique described in detail in Wittke(2000).
The basic idea of this iterative analysis is that the computationsection moves with the heading or with the simulation of eachround, respectively. With increasing number of iterations the con-straints are reduced which develop in the starting stage due tothe restraint of the displacements in longitudinal tunnel direc-tion. In this way the computed displacements approach the actualdisplacements of the excavation surface at the time of the instal-lation of the shotcrete membrane. In this manner the loading ofthe shotcrete membrane and the deformations due to the heading canbe realistically determined.
According to excavation class 7A-U-1 applied in construction (seeFig. 5.49) a round length of 1 m is simulated. The distance be-tween the shotcrete support and the temporary tunnel face amountsto 1 m in the crown and 2 m in the invert. The closing of the in-vert of the sidewall adits is thus completed in each case at adistance of 4 m from the crown face (Fig. 5.55).
The tunnel face support using shotcrete is not taken into account.
The vertical and horizontal displacements in the rock mass and onthe ground surface computed in case A for the sidewall adit head-ing are compared with the measurement results in Fig. 5.56 and5.57. The computed displacements were taken from the plane located14 m behind the crown face in the 36th computation step. This dis-tance represents approximately the average of the distances ofMC 2 to the crown faces of the two sidewall adit headings at thetime of the measurement (see Fig. 5.56 and 5.57).
A comparatively good agreement results between the measured andthe computed displacements (see Fig. 5.56 and 5.57). It is only inthe computed vertical displacements that heave show at the levelof the tunnel's invert and below which was not measured. Accordingto experience this is because in the FE-analysis the same moduluswas specified for the areas unloaded by the excavation as for theloaded areas above and beside the tunnel. The modulus relevant forunloading is generally markedly higher than the one for loading.In the stability analyses for the sidewall adit heading in thearea of the undercrossing of the freeway (cases B, C and D, seeTable 5.2), which are described in the following, an unloadingmodulus EU equal to three times the loading modulus was thereforespecified below the tunnel's invert level (see Fig. 5.54).
Three-dimensional analyses of the sidewall adit heading (analysiscross-section AC 2)
In the area of the undercrossing of the freeway (analysis cross-section AC 2) the overburden height varies between 23 and 27 m(see Fig. 5.51). The stability analyses for this area were corre-spondingly based on an average overburden of 25 m (see Fig. 5.53).The investigation included cases B, C and D (see Table 5.2). Dif-fering from case A, an unloading modulus EU = 3 E was specified be-low the invert in these analyses, as already mentioned (see Fig.5.54), together with a crown face support with p = 0.04 MN/m2 (seeFig. 5.55 and Table 5.2). In this way the supporting effect ofhorizontal cemented anchors installed in advance of the crown facewas modeled.
In cases B and D a shotcrete support of the temporary crown invert(t = 10 cm, see Fig. 5.54) is accounted for. In case C the tempo-rary crown invert support is not taken into account. In case D thespecified shear strength of the decomposed slate is higher than incases A, B and C (see Table 5.2).
In Fig. 5.58 the stress resultants are shown as a function of thedistance from the tunnel face and from the boundary of the meshfor the example of the normal thrust and the moment in the roof.The broken lines in Fig. 5.58 show the extrapolated course of thestress resultants which would ensue without the influence of theboundaries. The stress resultants in the section located 16.5 mbehind the tunnel face (dimensioning section) can be regarded as
decisive and taken as a basis for the design of the shotcrete mem-brane.
Fig. 5.58: Sidewall adit, stress resultants vs. distance fromthe crown face, case B, 36th computation step
Fig. 5.59 shows the stress resultants M, N and S in the dimension-ing section for case B. Very large bending moments M occur in theareas of the small radii of curvature of 1.0 m at the roof and atthe transition from the inside wall of the sidewall adit to theinvert (see Fig. 5.46). It was assumed for the design of the shot-crete membrane that the computed bending moments would not developin reality to their full extent since the shotcrete membrane is
installed in several working steps. The construction joints in theareas mentioned above enable the formation of links which resultin a reduction of the bending moments but not of the normalthrust. In agreement with the parties concerned the calculatedbending moments were reduced for the design to 65 %.
Fig. 5.59: Sidewall adit, stress resultants in the shotcretemembrane, dimensioning section, case B, 36th compu-tation step
Fig. 5.60 shows the statically required reinforcement cross-sections for bending and normal thrust determined for cases B, Cand D. In addition to the reduction of the moments already men-tioned, the design was based on a B25 concrete grade, a BSt 500steel grade and a distance of the reinforcement from the edge oft1 = 4 cm. In the roof (sections 1 and 2) and at the transitionfrom the outside wall of the sidewall adit to the invert (section3) statically required reinforcement cross-sections are computedfor case B which are not covered by the planned reinforcement(Q295 inside and outside) and require supplementary reinforcement.In the remaining area (section 4), no reinforcement is requiredfor statical reasons. A minimum reinforcement of Q295 inside andoutside suffices here. For case C, in which a support of the tem-porary crown invert is not accounted for, even larger staticallyrequired reinforcement cross-sections result in sections 1 to 3than for case B (see Fig. 5.60). For case D, in which a higherYoung's modulus and a higher shear strength are specified for thedecomposed slate, statically required reinforcement cross-sections
larger than the minimum reinforcement only ensue in the roof (sec-tions 1 and 2, Fig. 5.60).
Fig. 5.60: Sidewall adit, required reinforcement cross-sectionsfor bending and normal thrust
As a consequence of the results of these analyses, during the un-dercrossing of the freeway A3 by sidewall adit excavation supple-mentary reinforcement was installed in the roof area and the tem-porary invert of the crown of the sidewall adits was supportedwith shotcrete. The tunnel face was additionally supported by tun-nel face anchors (see Chapter 5.3.6). Further, an advancing pipeumbrella was installed for the excavation of the core (see Fig.5.50).
Three-dimensional analyses of a crown heading with closed invertand tunnel face anchoring
Because only a low heading performance could be achieved with thesidewall adit heading due to the extensive support measures, WBIinvestigated in further analyses whether in the section followingthe undercrossing of freeway A3 also a crown heading with closedinvert and tunnel face anchoring could be carried out in a stableway for the same ground conditions and higher overburden (see Fig.
5.45b; Wittke and Pierau, 2000; Wittke and Sternath, 2000). Forthis purpose three-dimensional FE-analyses were carried out forthe computation section shown in Fig. 5.61 with the program systemFEST03 (Wittke, 2000).
Fig. 5.61: Computation section, FE-mesh, boundary conditionsand parameters for three-dimensional analyses of thecrown heading with closed invert and tunnel face an-choring
The analyses were based on an overburden of 50 m. Because of thehigher overburden compared to the undercrossing of freeway A3, ahigher Young's modulus and a higher shear strength were assumedfor the decomposed slate with E = 100 MN/m² and EU = 300 MN/m², re-spectively, ϕ = 27.5° and c = 20 kN/m2 (see Table 5.2 and Fig.5.61).
Fig. 5.62: Crown heading with closed invert and tunnel face an-choring: a) Stress resultants in the shotcrete mem-brane; b) required reinforcement
The computation sequence for the simulation of the crown headingwas analogous to the analyses simulating the sidewall adit heading(see Fig. 5.55).
The analysis results show that the stability of the crown headingcan be proved if the tunnel face is supported by advance anchoringand the support is closed soon at the invert. As in the analysesdescribed before, the tunnel face anchors were accounted for by asupport of the tunnel face with p = 0.04 MN/m2 (see Fig. 5.61). Itcan be proven that the subsidence due to the heading can be keptsmall in this way. The computed loading of the shotcrete membrane
of the crown due to bending and normal thrust (Fig. 5.62a) revealsa high bending compression loading at the transition to the tempo-rary invert. The reason for this is the small radius of curvatureof the shotcrete membrane of only 2 m. For reasons inherent to theconstruction process it is often attempted to keep the radius ofcurvature as small as possible. For statical reasons, on the otherhand, the radius of curvature should not be less than 2 m. Supple-mentary reinforcement cannot be dispensed with in this case, how-ever, although the membrane is rounded according to this require-ment (Fig. 5.62b).
5.3.6 Construction
The Niedernhausen Tunnel was excavated between the northern tunnelportal at chainage km 141+499 and chainage km 141+771.5 by side-wall adit heading. The mapping during tunneling showed that thelayer boundary talus material / slate is located in the crown areaof the tunnel's cross-section (Fig. 5.63).
The excavation work proved very difficult mainly because of thehigh groundwater level. As the closing of the invert was delayeddue to the inflow of water, large subsidence resulted at theground surface (Wittke and Sternath, 2000). To stabilize the tun-nel face it was necessary to drain the ground in advance and alsoto support the tunnel face using shotcrete and anchors (Fig. 5.64and 5.65). The heading performance was correspondingly low withapprox. 1 m/day in the sidewall adits.
Fig. 5.64: Tunnel face support of the sidewall adit, crown
Fig. 5.65: Tunnel face support of the sidewall adit, crown andbench
Fig. 5.66 depicts the support measures during the sidewall aditheading in cross- and longitudinal section. It shows that excava-tion class 7A-U-1 was modified as follows with respect to theoriginal design (see Fig. 5.49):
- Supplementary reinforcement in the sidewall adit roofs,
For the advance drainage of the ground, at first up to 50 m longhorizontal vacuum wells were constructed from the starting wall(see Fig. 5.47). In addition, vacuum lances were installed fromthe tunnel face during the heading. The effectiveness of the vac-uum lances remained limited, however. The outflow of water in thetunnel face area could only be reduced but not be prevented inthis way. The ensuing mud formation interfered considerably withthe excavation.
As a consequence, the groundwater was lowered using deep vacuumwells drilled from the ground surface in advance of the excavationdown to below the tunnel's invert. In the ground drained in ad-vance it was possible to increase the heading performance markedlyafterwards.
Fig. 5.67 is a photograph of the northern heading during the exca-vation of the core.
Fig. 5.67: Excavation of the core
From the south the tunnel was driven up to chainage km 141+771.5by crown heading. In the section where the tunnel cross-sectionwas located in the decomposed slate the crown was excavated up to
the cut-through with a closed invert and tunnel face anchoring(Fig. 5.63). The basis for this were the results of the FE-analyses described in Chapter 5.3.5.
Fig. 5.68 depicts the support measures carried out during thecrown heading in cross- and longitudinal section. The roundlengths and the excavation sequence are shown in the longitudinalsection as well.
Fig. 5.68: Construction, crown heading with closed invert andtunnel face anchoring: a) Cross-section;b) longitudinal section
The computed displacements were well confirmed by the results ofsurface leveling and convergency measurements in the tunnel. Theheading performance in the crown amounted to almost 2 m/day (Wit-tke and Pierau, 2000).
5.3.7 Conclusions
The Niedernhausen Tunnel was excavated over a length of approx.350 m in the completely weathered and decomposed slates of thegroundwater-bearing Schwall layers, which have a low strength anda high deformability. In this ground the freeway A3 had to be un-dercrossed with an overburden of 20 m to 30 m.
A sidewall adit heading with short round lengths was carried out.To stabilize the tunnel face the rock had to be drained in advanceand the tunnel face had to be supported by shotcrete and anchors.The results of three-dimensional FE-analyses showed that addi-tional measures were required to support the work space at thetunnel face, such as e. g. the support of the temporary crown in-vert of the two sidewall adits and the installation of spiles. Be-cause of the extensive support measures a very low heading per-formance could only be achieved.
It could be proven by further three-dimensional analyses that insections with higher overburden also a crown heading with closedinvert and systematic tunnel face anchoring could be carried outin a stable way in these unfavorable ground conditions. As a con-sequence, the Niedernhausen Tunnel was successfully excavated bycrown heading outside of the sphere of influence of freeway A3.Almost twice the heading performance of the sidewall adit headingwas achieved here.
The experience gained with the excavation of the NiedernhausenTunnel should also lead to economic solutions for future tunnelstructures in comparable ground conditions. It should be possiblein many cases to replace an expensive sidewall adit excavation bya crown heading with closed invert and tunnel face anchoring.
6.1 Urban railway tunnel underneath the Stuttgart airportrunway, Germany
6.1.1 Introduction
The urban railway of Stuttgart, Germany, was extended in 2001 by asection starting at the Airport station and running underneath theairport area to the city of Filderstadt-Bernhausen. The Airportstation (construction lot 72) and a continuation of limited extenttowards Bernhausen (lot 92) were constructed by the cut-and-covermethod. A 2.15 km long tunnel section driven by underground con-struction (lot 601) follows. It undercrosses among other areas theapron and the runway of the airport. The approx. 500 m long tunnelsection of the Filderstadt-Bernhausen station (lot 602) was con-structed by the cut-and-cover method (Fig. 6.1).
Fig. 6.1: Undercrossing of Stuttgart airport, site plan
The present Chapter describes the tunnel section of lot 601 drivenby underground construction, with an emphasis on the undercrossingof the runway of the airport. In this section the absolute and
differential ground surface subsidence due to the tunneling had tobe kept as small as possible. This task proved to be quite demand-ing, because settlement-sensitive, soft valley deposits and fillare locally encountered in the runway area and the groundwater ta-ble lies above the tunnel roof. Lowering the groundwater tableduring tunneling could therefore not be permitted, because of therisk of large settlements. The shotcrete support had thus to beconstructed with a low water permeability and designed to with-stand the water pressure. For the design of the shotcrete membraneit had further to be taken into account that high horizontalstresses exist especially in the mudstones of the Lias α forma-tion, in which the major part of the mined tunnel section is lo-cated (see Chapter 4.1).
6.1.2 Structure
The course of the alignment and the ground profile are shown inFig. 6.2. Following lot 92, the alignment descends in the direc-tion of Bernhausen up to the area in front of the runway.
Fig. 6.2: Undercrossing of Stuttgart airport, longitudinalsection with ground profile
The tunnel then runs horizontally over a length of approx. 1400 m(lot 601). The following gradient extends into the cut-and-coversection (lot 602). In the area of the station the alignment thenruns approximately horizontally again.
Fig. 6.3: Single-tracked tunnel tube, standard profile
Two single-tracked tunnel tubes are planned for the mined tunnelsection (lot 601), only one of which has been built for the timebeing, however. The tunnel tube was constructed with a circularprofile, among other reasons also in order to be able to designthe shotcrete membrane for the full water pressure. The 30 to 35cm thick shotcrete membrane was made from alkali-free dry-mixshotcrete with a low water permeability using spray cement as abonding agent (see Chapter 2.1.2). The excavated diameter is ap-
prox. 8.7 m, the inside diameter 7.0 m. The excavated cross-section of one tunnel tube amounts to approx. 60 m2 (Fig. 6.3).
In the area of the airport runway the tunnel roof is located ap-prox. 21 m below the ground surface (Fig. 6.4). Stuttgart AirportCo. (Flughafen Stuttgart GmbH, FSG) demanded that the tunneling-induced subsidence in this area has to be limited to 15 mm and thedifferential subsidence at the ground surface to 1 o/oo.
Fig. 6.4: Undercrossing of the runway of Stuttgart airport,longitudinal section with ground profile
6.1.3 Ground and groundwater conditions
The ground profile in the area of the mined tunnel of lot 601 issimilar to the one in the area of the Österfeld Tunnel (see Chap-ter 4.1). The stratigraphic sequence includes the layers of theKnollenmergel, the Rät, the Lias α and the Quarternary (see Fig.4.4).
To explore the ground and the groundwater conditions core drill-ings were sunk along the tunnel alignment. Some of these boreholeswere equipped as observation wells.
Fig. 6.5: Photograph of the tunnel face showing the transitionfrom the alternating sequence of mudstone and lime-sandstone to the layers consisting predominantly ofmudstone
formation (Fig. 6.2 and 6.4). In the area of the airport runway itlies completely within the layers consisting predominantly of mud-stone. Adjacent to the cut-and-cover tunnel sections, i. e. in theareas with descending or climbing alignment, the tunnel crossesthrough the alternating sequence of mudstone and banks of lime-sandstone (Fig. 6.2). Fig. 6.5 is a photograph of the temporarytunnel face, in which the lowest banks of lime-sandstone of thealternating sequence, which are also referred to as "main sand-stone", are clearly recognizable.
The runway of the airport is founded on cohesive layers (Fig. 6.4)consisting mainly of Filder loam and Lias α residual clay. To thewest of the urban railway alignment a lake was formerly located inthe area of today's runway, flowing out into a creek running to-wards the east. In the course of the construction of Stuttgartairport, the area of the lake and the creek was filled up. The ex-isting partially soft and settlement-sensitive valley deposits re-mained under the fill in the process. The fill and the valley de-posits are locally several meters thick (Fig. 6.6).
Fig. 6.7 shows the structural model (see Chapter 2.5.1) derivedfor the ground in the area of the tunnel section driven by under-ground construction. The discontinuity fabric is characterized byan orthogonal system of horizontal bedding parallel discontinui-ties and steep to vertically dipping joints. Unlike to the banksof lime-sandstone, the bedding parallel discontinuities and jointsin the mudstone layers are mostly closed or filled with clay andonly vaguely recognizable. In the mudstones of the Rät and theKnollenmergel slickensides exist, dipping at 20 to 40° and strik-ing in all directions (see Chapter 4.1.3).
The soil and rock mechanical parameters given in Table 6.1 werespecified on the basis of the results of laboratory and in-situtests as well as experience gained from projects in comparableground conditions (see Chapter 4.1). For the rock layers of theLias α formation encountered in the area of the mined tunnel, atransversely isotropic elastic stress-strain behavior, describedby 5 independent elastic constants (Wittke, 2000), was assumed forloading below the strength. A further characteristic of the Lias αlayers are the low shear strengths on the bedding parallel discon-tinuities and the joints.
Fig. 6.7: Structural model of the ground in the area of themined tunnel section (lot 601)
The parameters given in Table 6.1 should be interpreted as charac-teristic parameters according to DIN 4020 (1990) and were taken asa basis for the stability analyses of the mined tunnel.
Alternating se-quence of mud-stone and lime-sandstone
E1 = 1500 MN/m2
E2 = 750 MN/m2
G2 = 200 MN/m2
ν1 = 0.25
ν2 = 0.2
Bedding B:
ϕB = 20°
cB = 80 kN/m2
Joints J1, J2:
ϕJ = 35°
cJ = 40 kN/m2
Horizontally:
kfH = 5 ⋅ 10-5 m/s
Vertically:
kfV = 10-6 m/s
Mudstone withsingle layers oflime-sandstone
E1 = 1000 MN/m2
E2 = 500 MN/m2
G2 = 200 MN/m2
ν1 = 0.25
ν2 = 0.2
Bedding B:
ϕB = 20°, cB = 0
Joints J1, J2:
ϕJ = 35°, cJ = 0
kf = 10-7 m/s
Rät and leachedzone of theKnollenmergel
E = 150 MN/m2
ν = 0.3
Discontinuities:
ϕD = 17.5°
cD = 10 kN/m2kf = 5 ⋅ 10-7 m/s
Knollenmergel,unweathered
E = 1000 MN/m2
ν = 0.25
Slickensides:
ϕS = 17.5°
cS = 10 kN/m2kf ≤ 10-8 m/s
Table 6.1: Characteristic soil and rock mechanical parameters
The results of in-situ tests and measurements on different struc-tures in the area of Stuttgart have shown that increased horizon-tal in-situ stresses exist in the Lias α (Grüter, 1988; Wittke,1990; Wittke, 1991). According to these results, in addition tothe horizontal stresses resulting from the dead weight, horizontalstresses of ΔσH = 1 to 2 MN/m2 exist in the unweathered mudstonelayers and horizontal stresses of ΔσH = 0.5 to 1.0 MN/m2 exist inthe alternating sequence (Böttcher et al., 1998). The magnitude ofthese stresses was confirmed by the results of the stress measure-ments by the overcoring method (Kiehl and Pahl, 1990) carried outin the course of the project presented here. Horizontal in-situ
stresses of ΔσH = 0.5 to 1.7 MN/m2 were derived from the results ofthese stress measurements.
The permeability tests carried out as part of the explorationshowed that the mudstone layers have a much lower water permeabil-ity than the lime-sandstone banks. Accordingly, the alternatingsequence is inhomogeneous with respect to its permeability. How-ever, since the tunnel diameter is large compared to the thick-nesses and the spacing of the layers of the alternating sequence,the alternating sequence can be overall considered approximatelyhomogeneous, if the different permeabilities of the layers istaken into account by introducing an anisotropy (Wittke, 2000).The horizontal permeability of kfH = 5 · 10-5 m/s is determined hereby the banks of lime-sandstone, whereas the vertical permeabilityof kfV = 10-6 m/s is due to the mudstone (Table 6.1).
The overlying strata, the Lias α layers consisting predominantlyof mudstone, the Rät and the Knollenmergel have a much smaller wa-ter permeability than the alternating sequence (Table 6.1).
In the northern part of the airport the groundwater of the alter-nating sequence is mostly artesian. The water table is encounteredwithin the overlying strata (see Fig. 6.2 and 6.4). In the adja-cent section up to the southern limit of the airport the groundwa-ter table lies within the alternating sequence. A further sectionwith locally artesian groundwater follows (see Fig. 6.2).
The mined tunnel section of lot 601 is thus located almost overits entire length completely underneath the groundwater table. Themaximum height of the water table above the tunnel's invert isreached in the area of the airport runway with approx. 26 to 27 m(see Fig. 6.4).
6.1.4 Fundamentals of the design
During the cut-and-cover construction of the first 400 m of thetunnel (lot 92) the groundwater was lowered to below the construc-tion pit's invert using drawdown wells. Measurements of thegroundwater level in the airport and specially in the runway areashowed that the drawdown cone had a range of approx. 450 m (Fig.6.8). To avoid subsidence in the airport area, it became necessaryto recharge the groundwater using injection wells (Erichsen andTegelkamp, 1998).
Fig. 6.8: Advance construction (lot 92), site plan with draw-down cone
Especially the soft valley deposits in the runway area (see Fig.6.6) are very sensitive to settlements. Especially here, but alsoin other areas, a lowering of the groundwater table as a conse-quence of the tunnel heading would lead to subsidence due to lossof the hydrostatic uplift. The FSG therefore demanded that nogroundwater lowering must occur during the underground tunneling.
Three-dimensional, transient seepage flow analyses were thereforecarried out by WBI in the early stages of the project to investi-gate the influence of the tunnel heading on the groundwater condi-
tions. The program system used for this purpose was developed byWBI (Erichsen, 1994). It is described in detail in Wittke (2000).The assumptions made and the results of the analyses are describedand explained in detail in Wittke (2000) as well, and also in Wit-tke-Gattermann and Wittke (1997). The permeability of the layersconsisting predominantly of mudstone (see Chapter 6.1.3), in whichthe tunnel is located in the area of the undercrossing of the air-port runway, as well as the permeability of the shotcrete membraneof the tunnel were varied in the analyses.
Fig. 6.9: Analyzed drawdown curves of the groundwater tabledue to the heading, steady state, vertical sectionthrough the tunnel axis
Fig. 6.9 shows the drawdown of the groundwater table determinedfor the steady state in a vertical section through the tunnelaxis. For a low permeability of the mudstone layers of kf = 10-7
m/s and an impermeable shotcrete membrane, the groundwater draw-down of < 5 cm can be neglected. An increase in the permeabilityof the mudstone to kf = 10-6 m/s already leads to a groundwaterdrawdown of 0.5 m. An increased permeability of the shotcrete mem-brane yields a considerably increased groundwater drawdown evenfor a value of kf of 10-7 m/s for the mudstone. The steady state isalways reached within a period of time which is short in relationto the construction time (Wittke-Gattermann and Wittke, 1997; Wit-tke, 2000; Tegelkamp et al., 2000).
Corresponding to the results of the groundwater modeling analysesand to the demand by FSG that groundwater lowering must not occurduring the construction of the tunnel, the tunnel had to be sup-ported during the heading by a shotcrete membrane with a low waterpermeability. The membrane had to be designed to withstand the wa-ter pressure. An alkali-free dry-mix shotcrete with spray cementas bonding agent was used. A statically favorable circular profilewas selected for the tunnel's cross-section (see Fig. 6.3). Thewater pressure taken into account as well as the loads resultingfrom the rock mass pressure, which are essentially determined bythe increased horizontal in-situ stresses in the mudstone layers,lead to a required thickness of the reinforced shotcrete membraneof 30 to 35 cm.
An advancing crown excavation was not a reasonable option underthe given conditions, because on the one hand an open invert overgreat lengths could not be permitted since this would lead to agroundwater lowering, and on the other hand a watertight supportof the temporary crown invert could not be designed to withstandthe water pressure with economically justifiable expense. A full-face heading with a stepped tunnel face was therefore provided forin order to achieve as soon as possible a circular cross-sectionwith closed invert. This was important for statical reasons aswell as under the aspect of watertightness, since this design en-ables water to enter during the tunneling only through the tempo-rary tunnel face and thus only through a comparatively small crosssectional area. This design was also taken as a basis for thethree-dimensional seepage flow analyses.
Fig. 6.10 shows the sequence of excavation and means of supportspecified for the tunnel heading in the layers consisting predomi-nantly of mudstone. The full-face excavation was subdivided intocrown and bench/invert excavation.
Fig. 6.10: Standard heading in the mudstone, excavation andsupport
The tunnel cross-section was excavated using tunnel excavators,which were additionally equipped with a heavy hydraulic chisel.Thicker banks of lime-sandstone were loosened by blasting. Theround lengths ranged between 1 and 1.5 m in the crown and between2 and 3 m in the bench/invert (Fig. 6.10). The support was closedat the invert after 4 to 6 m.
In addition to the reinforced shotcrete support of the excavationprofile, a systematic anchoring with 3 m long SN-anchors was car-ried out locally. Steel sets were placed at a spacing of 1 m (Fig.6.10). The advancing support using spiles, included in the designfor more unfavorable rock mass sections, could be completely dis-pensed with.
In the deep tunnel section, in which the tunnel is located com-pletely within the low-permeability layers consisting predomi-nantly of mudstone, no further measures apart from the low-permeability shotcrete membrane were taken to maintain the ground-water table. In the northern tunnel section, in which the ascend-ing tunnel cuts into the strongly water-bearing layers of the al-ternating sequence, an advance sealing of water-bearing disconti-nuities was required to prevent a strong inflow of water throughthe tunnel face. To this end cement grouting was carried outthrough boreholes drilled from the tunnel face with an averagelength of 15 m (Fig. 6.11). Cement based suspensions with watercement ratios of 0.7 to 2 were used for the grouting.
Fig. 6.11: Sealing by advance cement grouting of the alternat-ing sequence
The advance grouting was carried out in sections in three workingsteps:
1. Sealing of the temporary tunnel face with shotcrete.
2. Construction of a grouting umbrella above the tunnel roof.
3. Construction of a transverse bulkhead in the area of the al-ternating sequence to seal off the area to be excavatedagainst groundwater flow in longitudinal tunnel direction.
The grouting boreholes were sealed against the borehole head withfabric packers (geotextiles) filled with cement based suspension.During the grouting works the heading had to be interrupted forapprox. 2 weeks each time.
By the advance sealing of the alternating sequence stronger inflowof water through the open tunnel face could be prevented to alarge extent. Only when the grouting boreholes were drilled (Fig.6.12), some inflow rates occurred for a short time due to the con-struction process (Tegelkamp et al., 2000).
6.1.6 Stability analyses for the design of the shotcretesupport
For the dimensioning of the shotcrete support, two- and three-dimensional FE-analyses were carried out using the program systemFEST03 (Wittke, 2000). These analyses were based on the character-istic parameters given in Table 6.1. In the Lias α increased hori-zontal in-situ stresses were taken into account. For the layersconsisting predominantly of mudstone ΔσH = 1.5 MN/m2 was specified.The alternating sequence was assigned a value of ΔσH = 0.5 MN/m2.Further analyses were carried out with no additional horizontalstresses assumed, as well as with horizontal stresses in the mud-stone increased to ΔσH = 2 MN/m2. In Fig. 6.13, the location of theanalysis cross-sections investigated in the design analyses isgiven in the geological longitudinal section.
Fig. 6.13: Location of the analysis cross-sections
Fig. 6.14 shows exemplarily the computation section, the FE-mesh,the boundary conditions and the ground profile of a three-dimensional analysis for the area of the airport runway (analysiscross-section 4A, see Fig. 6.13) with an overburden of 21 m.
Fig. 6.14: Analysis cross-section 4A, FE-mesh, boundary condi-tions and ground profile for three-dimensionalanalyses
In Fig. 6.15 the computation steps chosen for the simulation ofthe tunnel heading by the "step-by-step" method (Wittke, 2000) areillustrated.
In the first two computation steps the in-situ state of stress isdetermined, taking into account the dead weight of the rock massand the increased horizontal stresses in the Lias α. To simulatethe different horizontal stresses ΔσH in the alternating sequenceand in the mudstone, the nodes lying on the boundary planes
x = 40 m and y = 65 m (see Fig. 6.14) were assigned horizontaldisplacements in x- and y-direction corresponding to the respec-tive horizontal stresses ΔσH. To prevent shear stresses from beingtransferred across the boundary between the alternating sequenceand the mudstone due to the different horizontal displacements, aninterface layer is arranged between the two layers. In the 1st com-putation step, the dead weight is only taken into account for thealternating sequence and the mudstone. The overlying strata, theRät and the Knollenmergel are assumed to be weightless. In the 2nd
computation step the latter two layers are replaced by materialswith the same mechanical parameters, however with their deadweight. Since the new materials are installed stress-free in thealready deformed corresponding elements (Wittke, 2000) and thehorizontal displacements remain unchanged in the 2nd computationstep, only the stresses due to dead weight but not the increasedhorizontal stress ΔσH are effective in these layers.
In the 3rd computation step, the excavation and shotcrete supportof the crown and the bench and invert, trailing by 2 m, are simu-lated. Computation steps 4 to 16 include the simulation of theheading according to the "step-by-step" method. In each computa-tion step, the crown and the trailing bench and invert are bothadvanced by 2 m (Fig. 6.15).
Fig. 6.16 and 6.17 show the stress resultants in the shotcretemembrane determined in this analysis. The representation of themaximum stress resultants along the tunnel in Fig. 6.16 illus-trates that the loading of the shotcrete membrane continuously de-velops with increasing distance from the tunnel face, until themaximum values are reached at a distance of approx. twice the tun-nel's diameter. This cross-section is designated the dimensioningsection, because the stress resultants determined here have to beconsidered decisive for the design of the shotcrete membrane (seeChapter 5.3).
The stress resultants M, N and S in the dimensioning section areshown in Fig. 6.17. As expected, great compressive normal thrustexist. Because of the high horizontal stresses the maximum valuesresult in the roof and invert areas. The bending moments are rela-tively small due to the favorable geometry of the cross-section(circle), and the shear forces follow corresponding to the momentdistribution.
In addition to the three-dimensional analyses, two-dimensionalanalyses ones were carried out to investigate the influence of avariation of the parameters and of the horizontal in-situ stressesin the Lias α on the loading of the shotcrete membrane.
To model the influence of the displacements occurring in the tun-nel face area before the shotcrete membrane is installed, eitherthe computed stress resultants in the shotcrete membrane were re-duced in the two-dimensional analyses, or a preceding stress re-lief was simulated (Wittke, 2000). For the section in which thetunnel's cross-section is located entirely within the alternatingsequence, a calibration of the results of two-dimensional analyseson the basis of the results of a corresponding three-dimensionalanalysis resulted in a stress relief factor according to (4.1)(see Chapter 4.1) of av = 0.35.
The load case water pressure acting on the shotcrete membrane wasinvestigated in separate two-dimensional FE-analyses. Because thewater pressure builds up only at a certain distance from the tun-nel face, it suffices to superpose the stress resultants ensueingfrom the rock mass pressure and the water pressure and to designthe shotcrete membrane on the basis of the stress resultants thusobtained.
According to the results of the stability analyses, the shotcretemembrane of the standard tunnel sections could be designed with athickness of 35 cm for a factor of safety of 1.7 without requiringmore than the minimum reinforcement (inside and outside steel fab-ric mats Q285). The specified limits for the tunneling-inducedsubsidence and differential subsidence were not exceeded either.
6.1.7 Monitoring
The construction was accompanied by an extensive monitoring pro-gram aimed at compliance with all requirements as well as controland optimization of the heading works. This program was coordi-nated in mutual agreement with the FSG.
The monitoring program included the monitoring of the groundwaterlevel as well as displacement measurements at the ground surfaceand in the tunnel. Further, stress measurements were carried outin the shotcrete membrane. For the monitoring of the groundwaterlevel, the system of observation wells already existing on the
airport from previous construction and supplemented during the ex-ploration works for the urban railway tunnel could be used.
Fig. 6.18 shows the location of the measuring cross-sections andobservation wells as well as the heading location in November1999. From mid-February to mid-March 2000 the airport runway wasundercrossed by the excavation south coming from the northernstarting shaft. The excavation north emanating from the southernstarting shaft cut through to the excavation south on April 22,2000 (Fig. 6.18). The tunnel section excavated between November1999 and April 2000 is specifically marked in Fig. 6.18. The cut-through from lot 601 to the already existing lot 92, which havebeen constructed before by the cut-and-cover method, was carriedout on June 5, 2000.
In Fig. 6.19 the hydrographs of the observation wells in the areaof the airport runway are shown for the time between November 1999and June 2000. For the winter months a general rise of the ground-water level is apparent at all observation wells, which declinesagain in the spring of 2000. A reaction of the groundwater levelto the tunnel heading cannot be recognized. The measured varia-tions in water level can be attributed to the natural course ofthe groundwater flow and are not related to the tunneling.
Fig. 6.19: Observation well hydrographs during the undercross-ing of the runway of Stuttgart airport
A temporary drop of the groundwater table by a maximum of 2 to 3 mhowever occurred in the boreholes located in the area of thenorthern apron beside the rescue adit (Fig. 6.18). This decreasecorrelates in time with the grouting works carried out in the tun-nel section located within the alternating sequence (see Fig.6.11). With distances of up to 100 m to the observation wells andin view of the limited discharge in the tunnel through the grout-
ing boreholes of approx. 1 l/s, these observations demonstrate thesensitivity of the aquifer in the alternating sequence.
Overall it must be stated, however, that with the advance groutinga large-scale, long-term groundwater lowering could be avoided.Even in the northern apron area, subsidence due to interferencewith the groundwater did not occur.
Fig. 6.20: Results of the displacement measurements in MC 6
In Fig. 6.20, the results of the displacement measurements are il-lustrated exemplarily for measuring cross-section MC 6 (see Fig.6.18). Here, a tunneling-induced subsidence of 4 mm at the mostwas measured at the ground surface above the tunnel roof. Alongthe entire tunnel, predominantly a subsidence of between 4 and 7mm was determined above the tunnel axis. Because of these compara-tively low values, which were markedly smaller than the admissiblesubsidence of 15 mm, the airport facilities as well as the build-ings of Bernhausen, neighboring to the south, could be under-crossed without damage.
Comparatively large horizontal displacements in the ground of upto 9 mm were measured at the level of the tunnel (see Fig. 6.20).This displacement distribution is typical for this constructionproject and has to be attributed to the increased horizontal in-situ stresses in the Lias α.
6.1.8 Interpretation of the monitoring results
During the construction the displacements measured in the tunnelas well as the tangential stresses measured in the shotcrete mem-brane were compared to the values computed in the design analyses,which are based on the characteristic parameters (see Table 6.1).The resulting differences can be attributed to the conservativeassumptions made in the design analyses, mainly on the high hori-zontal stresses ΔσH assumed in the Lias α (see Fig. 6.14).
To be able to back-analyze the measured results, parameter studieswere carried out. The following parameters, which have a large in-fluence on the loading of the shotcrete membrane, were varied:
- Horizontal stresses in the mudstone (ΔσH),
- deformability of the mudstone layers (E1, E2),
- shear strength on the bedding parallel discontinuities in themudstone (ϕB),
- deformability of the shotcrete membrane (ESC).
The best agreement with the monitoring results was achieved in acomparative analysis, in which a Young's modulus of the shotcretesupport of ESC = 2000 MN/m2 was selected as opposed to the
Young's modulus taken as a basis for the design analyses(ESC = 15000 MN/m2), and in which the horizontal stress in the mud-stone was reduced from 1.5 MN/m2 (design analysis) to 1.0 MN/m2
(Table 6.2).
Design analyses Comparative analysis
Shotcrete ESC = 15000 MN/m2 ESC = 2000 MN/m2
Mudstone withsingle layers oflime-sandstone
ΔσH = 1.5 MN/m2 ΔσH = 1.0 MN/m2
Table 6.2: Comparison analysis for the interpretation of moni-toring results: Differences to the construction de-sign analyses
The low modulus of ESC = 2000 MN/m2 accounts for the deformabilitydevelopment as well as the creep properties of the shotcrete. Be-cause of the short round lengths and the early closing of the sup-port ring, the shotcrete is loaded in the present case at a veryyoung age, in which it still possesses a low strength and a highdeformability as well as a high creep potential (see Chapter 2.1).The value of ESC = 15000 MN/m2 used in the design analyses there-fore represents a conservative assumption, by which the loading ofthe shotcrete membrane is overestimated.
A good agreement with the measured values can however only beachieved if the water pressure and the seepage forces are ac-counted for that act on the ground and the shotcrete membrane dur-ing construction.
Since the water permeability of the alternating sequence is highcompared to the one of the mudstone, there is no significant de-crease of piezometric heads within the alternating sequence. Al-most the entire decrease of piezometric heads thus occurs in themudstone. Fig. 6.21 shows qualitatively the flow net with itsequipotential lines and streamlines which represents the groundwa-ter flow towards the tunnel. Thus, as an approximation, the entirewater pressure pw resulting from the difference in elevation be-tween the groundwater table and the top boundary of the mudstonelayer acts on the mudstone layer. The loading of the shotcretemembrane due to the groundwater flow towards the tunnel can there-
fore be replaced as an approximation by a surface load pW acting onthe mudstone layer (Fig. 6.21).
Fig. 6.21: Qualitative distribution of equipotential lines andstreamlines in the mudstone and replacement of thewater pressure by an equivalent load
Fig. 6.22 illustrates the determination of the stresses and defor-mations resulting from the flow towards the tunnel in two computa-tion steps. In the 1st computation step the surcharge from the wa-ter pressure pw is applied onto the mudstone layer. In the 2nd com-putation step, the excavation and the installation of the shot-crete membrane are simulated. The rock mass and the shotcrete areassumed weightless here. The deformations and the loading of theshotcrete membrane can therefore be determined approximately bysuperposing the deformations and stresses computed for the loadcases "rock mass pressure" and "water pressure on the mudstonelayer".
Fig. 6.22: Computation steps to determine the stresses and de-formations resulting from the flow towards the tun-nel
Fig. 6.23 shows the comparison between the measured and the com-puted values of the displacements of the tunnel contour and thetangential stresses in the shotcrete.
The comparison between measured and computed displacements isbased on so-called "representative displacements" determined asthe mean values of the displacements measured in different cross-sections. It can be seen that the representative displacements canbe captured by the analyses if the water pressure is taken intoaccount (Fig. 6.23).
Around the circumference of the shotcrete membrane differing tan-gential stresses were measured. The very low stresses measured inthe roof area (see Fig. 6.23, bottom left), are not consideredrepresentative. In the other areas the measured tangentialstresses are captured by the analysis.
During the heading of the rescue adit as well, geotechnical moni-toring was carried out (see Fig. 6.18). The representative dis-placements of the tunnel contour and the tangential stresses inthe shotcrete measured in the rescue adit are pictured in Fig.6.24 (left). The displacements are somewhat smaller than the rep-resentative displacements in the urban railway tunnel. The meas-
ured tangential stresses are markedly higher in the roof than thetangential stresses shown for in the shotcrete membrane of the ur-ban railway tunnel.
Fig. 6.23: Comparison of the measured displacements and tangen-tial stresses in the runway area with the corre-sponding values computed with and without considera-tion of the water pressure
The monitoring results obtained in the rescue adit can be back-analyzed with the same parameters as the monitoring results in theurban railway tunnel. In Fig. 6.24 the monitoring results are com-pared with the values computed with and without consideration ofthe water pressure. If the water pressure is taken into account,the displacements as well as the tangential stresses agree well.
Fig. 6.24: Comparison of the measured displacements and tangen-tial stresses in the rescue adit with the corre-sponding values computed with and without considera-tion of the water pressure
Stuttgart airport was to be undercrossed by a tunnel with low tomedium overburden. The water table is located above the tunnelroof, and settlement-sensitive layers are locally encountered atthe ground surface. Because of the demands with respect to the ad-missible subsidence, groundwater lowering had to be avoided duringtunneling. Accordingly, the shotcrete membrane had to be dimen-sioned for the water pressure. It had to be further taken into ac-count for the design of the shotcrete membrane that the Lias α, inwhich the tunnel cross-section is located, shows increased hori-zontal in-situ stresses and low strengths on the bedding paralleldiscontinuities.
This task was solved by the following measures:
- Full-face excavation with a stepped tunnel face and an earlyclosing of the support ring,
- construction of a low-permeability shotcrete membrane withhigh strength,
- specification of a circular profile which is statically favor-able for the design of the shotcrete membrane,
- sealing of water-bearing discontinuities in the alternatingsequence of mudstone and lime-sandstone by advance grouting.
By these measures, the tunnel could be excavated with very smallsubsidence of the ground surface. There was no interference withair traffic at any time.
It further showed that the transient seepage flow analyses andtwo- and three-dimensional stability analyses carried out in thecourse of this project with the program systems HYDOPO and FEST03(Wittke, 2000) represented an essential contribution towards thedesign, the statics and the specification of the excavation andsupport measures.
In the course of the new freeway (Autobahn) A71 connecting Erfurtand Schweinfurt (Kleffner, 2000), the "Berg Bock" freeway tunnelwas excavated between the exit Suhl/North and the intersectionSuhl (Fig. 6.25). The two tubes each 2700 m in length from northto south pass through the Suhl granite, the base sediments andporphyrite as well as, after passing the southern edge fault, thelayers of the Lower Triassic sandstone (Fig. 6.26).
Fig. 6.25: Tunnel Berg Bock, site plan
The two tunnel tubes were driven mainly by the full-face excava-tion method within a construction time of approx. 10 month. Withfour tunnel faces, maximum performances of approx. 30 m/d were at-tained.
6.2.2 Structure
Two tunnel tubes, eastern tube and western tube each 2700 m inlength were constructed (Fig. 6.27). The spacing of axis of bothtubes ranges between approx. 23 m at the northern portal and ap-prox. 30 m at the southern portal. Each tube comprises two trafficlanes each 3.75 m wide and two emergency sidewalks (Fig. 6.28).The excavated cross-section of the tunnel ranges between 80 m2 and100 m2 (Fig. 6.28 and 6.29). The alignment dips continuously to-wards the southern portal with 1.1 %. The maximum overburdenamounts to approx. 190 m (Fig. 6.26).
Fig. 6.29: Tunnel cross-section with open invert in stable rockmass
The safety conception is based on the German standard for theequipment and the operation of road tunnels (RABT, 1994). Thisstandard was currently revised on the basis of the evaluation ofseveral fire accidents in tunnels happened in recent years. Thus,additional, supplementary demands on the safety conception had tobe fulfilled at short notice in the final planning for the tunnelBerg Bock (Schmidtmann and Erichsen, 2001).
The safety conception comprises nine connection tunnels betweenthe tubes at distances of ≤ 300 m (Fig. 6.27). The connection tun-nels are equipped with emergency call niches and fire protectionlocks. Three breakdown bays are furthermore situated in each tun-nel tube with distances of ≤ 600 m. Three connection tunnels lo-cated close to the breakdown bays are passable for rescue vehi-cles. Besides that, emergency call niches and niches provided with
fire extinguishing devices are located in each tube at distancesof ≤ 150 m. Fire emergency lights are installed on one side of eachtube at distances of 24 m (Table 6.3).
Installation Distance Quantity
connection tunnels ≤ 300 m 9breakdown bays ≤ 600 m 2 x 3emergency call niches ≤ 150 m 2 x 19hydrant niches ≤ 150 m 2 x 19electrical niches approx. 180 m 2 x 14niches for drainage flushing shaftson both sides of the lanes
50 - 80 m 2 x 104
road sign displays approx. 300 m 2 x 9jet fans approx. 300 m 2 x 9fire emergency lighting 24 m 2 x 110
Table 6.3: Installations for operational safety
In the areas of the portals weathered rock of the Lower Triassicsandstone and completely weathered granite, respectively, were en-countered (Fig. 6.26). In these sections the tunnel tubes werecarried out with an approx. 11.6 m wide and approx. 9.9 m highmouth-shaped cross-section with a closed invert and an excavatedcross-section of approx. 100 m². The thickness of the shotcretemembrane (concrete grade B 25 corresponding to C 20/25) is 25 cm.The interior lining (concrete grade B 35 corresponding to C 30/37)is 40 cm thick (Fig. 6.28).
In the remaining sections, located in stable rock mass, a cross-section of approx. 11.4 m width and approx. 8.4 m height with anopen invert and an excavated cross-section of approx. 80 m² wascarried out. The tunnel walls mainly were supported by fiber rein-forced shotcrete with a concrete grade of B 45 corresponding toC 35/45 and a thickness of t = 15 cm. Locally reinforced shotcretewith a concrete grade of B 25 and a thickness of t = 25 cm was in-stalled. The shotcrete membrane was carried out with radii ofR = 5.9 m, R = 4.3 m and R = 8.4 m. The 30 cm thick interior lin-ing with a concrete grade of B 35 was founded on 50 cm high con-crete shoulders with the same grade (Fig. 6.29).
Between the interior lining and the shotcrete membrane a non-wovensynthetic and as a sealing a 2 mm thick foil were installed (Fig.
6.28 and 6.29). In the tunnel sections with closed invert betweenthe interior lining and the shotcrete membrane a separation foilwas installed at the invert (Fig. 6.28). Thus, in this area be-tween interior lining and shotcrete membrane no tensile and shearforces can be transferred.
Both tubes were carried out as fully drained road tunnels. Theseepage water was drained off by two lateral drainage pipes in thearea of the lower sidewalls and the gradient of the tunnel of1.1 % (Fig. 6.28 and 6.29). Flushing shafts for the washing of thedrainage ducts are provided at distances of 50 to 80 m on bothsides of the lanes.
Fig. 6.30 shows the northern portal of the Berg Bock tunnel.
Fig. 6.30: Tunnel Berg Bock, northern portal
6.2.3 Ground and groundwater conditions
The ground profile is shown in Fig. 6.26 in a geotechnical longi-tudinal section. The ground conditions and the overburden heightare approximately the same for both tunnel tubes.
In the starting area at the northern portal, the granite is pre-dominantly decomposed. The tunnel cross-section is alternatinglylocated here in hard, mostly strongly jointed granite and in com-pletely decomposed granite.
In the further course of the tunnel, unweathered, mostly very hardgranite was encountered. The rock is streaked with a multitude ofveins of different thickness. The rock mass is compact, with ajoint spacing of more than 1 m, to narrowly jointed, and in someareas it is traversed by joints with large extent.
Fig. 6.31: View of the working face in the Lower Triassic sand-stone
The granite is followed by the base sediments (sand-, mud- andsiltstone). In the area of the highest cover, the tunnel is lo-cated in the porphyrite, which is predominantly hard to very hardand slightly to narrowly jointed. After that, the tunnel crossesthe southern edge fault, which consists of water-bearing, stronglydecomposed and mylonized zones with a thickness of a few decime-ters. In the last tunnel section, the tunnel is located in theLower Triassic sandstone (Fig. 6.26).
The layers of the Lower Triassic sandstone and the base sedimentsconsist of an alternating sequence of sandstone and mudstone. Thewidely persistent bedding parallel discontinuities are mostlyhorizontal in the Lower Triassic sandstone (Fig. 6.31) and pre-dominantly steeply inclined in the base sediments. The joints nor-mal to the bedding usually end at the bedding parallel discontinu-ties.
The groundwater table is located up to 180 m above the tunnelroof.
6.2.4 Excavation and support
Heading in completely weathered granite
In the area of the northern portal (Fig. 6.26), the following sup-porting measures were carried out:
- Preceding drainage borings,
- preceding pipe umbrella,
- crown heading with closed invert,
- tunnel face support core.
The round lengths ranged between 0.75 and 1.0 m. A heading per-formance of approx. 1 m/d was achieved in the area of the com-pletely weathered granite.
Stable rock mass jointed to varying degrees was encountered in theunweathered granite and porphyrite. The rock mass conditions al-lowed a full-face blasting excavation of the tunnels in these ar-eas (Fig. 6.32). Mainly excavation class A0 was applied with around length of up to 3.5 m (Fig. 6.33). In part, an even greaterround length was chosen during construction. By the use of steelfiber shotcrete (t = 15 cm) with a fiber content of 40 kg/m3 forthe support of the excavation contour, the expenses for the sup-port were kept low (Fig. 6.33). Anchors were installed as requireddepending on the jointing.
Fig. 6.32: Loading of blastholes during the full-face excava-tion in granite
With this optimized heading scheme, approx. 2 to 3 rounds could beachieved per day and tunnel face. The heading performance thusamounted to up to 10 m/d per tunnel face. For the 4 tunnel faces,maximum performances of ≥ 30 m/d were reached.
Heading in Lower Triassic sandstone and base sediments
In these layers, a crown heading with closed invert and trailingbench excavation was carried out. In order to support the tunnelface and the working area against the dropping of so-called "cof-fin lids", if the bedding was approximately horizontal, precedingspiles were installed. In areas with the bedding dipping moder-ately steeply towards the tunnel, a support core was left standingto support the tunnel face.
A performance of approx. 4 to 5 m per day and tunnel face wasachieved with this heading technique.
Heading schedule
Both tunnel tubes were excavated within some 10 months each, cor-responding to an average heading performance of approx. 9 m perday and tube.
Fig. 6.34 shows exemplarily the heading progress over time for theeastern tube (see Fig. 6.27). The western tube was excavated inparallel with the eastern tube by so-called opposite heading. Theeastern tube was successfully cut through on January 19, 2001. Thecut-through of the western tube and thus the heading of the entiretunnel was celebrated on February 2, 2001.
6.2.5 Stability analyses for the stages of construction anddesign of the shotcrete support
To analyze the stability during construction and to design theshotcrete support, two-dimensional FE-analyses were carried outusing the program system FEST03 (Wittke, 2000). A total of eightanalysis cross-sections shown in Fig. 6.26 (AC 1 to AC 6, AC 2aand AC 6a) were investigated.
Fig. 6.35 shows exemplarily the FE-mesh, the boundary conditions,the ground profile and the parameters for analysis cross-sectionAC 2, which was used to analyze the stability of the tunnel tubesin the granite. The overburden amounts to 130 m (see Fig. 6.26).
Due to symmetry, only one tunnel tube is modeled as a simplifica-tion. The plane of symmetry lies at the center of the rock pillarbetween the two tunnel tubes. Such a model is to be consideredconservative with respect to the loading of the shotcrete mem-brane, because this way a simultaneous excavation of both tunneltubes is simulated.
The specified computation section consists of a 1 m thick slicewith a width of 75 m (x-direction). The height amounts to 157 m.The FE-mesh consists of 2135 isoparametric elements with 13004nodes. As boundary conditions, vertically sliding supports are in-troduced for the nodes of the vertical boundary planes (x = 0 andx = 75 m). For the nodes of the lower boundary plane (z = 0) hori-zontally sliding supports are specified (Fig. 6.35). All nodes areassumed fixed in y-direction.
The tunnel cross-section is entirely located in the granite. Belowthe ground surface, two 11 m thick layers of decomposed graniteand slightly weathered granite, respectively, are simulated. Theassumed mechanical parameters for these layers are also given inFig. 6.35. The loading modulus of the unweathered granite was as-sumed as EL = 15000 MN/m2. Underneath the tunnel's invert an un-
Fig. 6.36: Analysis cross-section AC 2 (shotcrete support), FE-mesh, detail
The orientations of the joints which are present in the granitewere not clearly determined. Thus randomly distributed joint ori-entations were assumed. The rock mass therefore was modeled with
Fig. 6.38: Analysis cross-section AC 2, principal normalstresses and rock mass areas in which strength isexceeded, 3rd computation step
For the shotcrete, a statically effective Young's modulus of15000 MN/m2 was assumed taking into account the hardening duringthe application of the load (Fig. 6.36).
In Fig. 6.37 the computation steps are outlined. In the 1st compu-tation step, the state of stress and deformation resulting from
the dead weight of the ground is determined (primary state). Incomputation steps 2 and 4, a preceding stress relief is each simu-lated in those areas of the cross-section, of which the excavationand the shotcrete support are simulated in computation steps 3 and5, respectively. The stress relief factor according to (4.1) isspecified as av = 0.5.
Fig. 6.38 shows the principal normal stresses in the rock massaround the excavation after the full-face excavation and the in-stallation of the shotcrete support at the end of the 3rd computa-tion step. The stress redistribution that occurred with the exca-vation as well as the areas of exceeded strength can be recog-nized. Although these areas extend around the entire circumferenceof the excavation, a pronounced arching is apparent. Due to thelow deformability of the rock mass (E = 15000 MN/m2, see Fig.6.35), the shotcrete membrane is only marginally loaded by therock mass and thus takes on a slightly stabilizing and assistingfunction only.
In Fig. 6.39 the heading-induced displacements computed for thefull-face excavation are shown. The total computed roof subsidenceamounts to 3.3 mm (3rd – 1st computation step, Fig. 6.39a). In the2nd computation step (preceding stress relief) the displacementspreceding the heading are determined. The shotcrete membrane istherefore only loaded in the 3rd computation step. The displacementof the excavation profile resulting from this loading (3rd – 2nd
computation step) is shown in Fig. 6.39b. Thus the computed dis-placement of the shotcrete membrane amounts to 2.1 mm at the roof.
Fig. 6.39: Analysis cross-section AC 2, displacements due tofull-face excavation: a) 3rd – 1st computation step;b) 3rd – 2nd computation step
Fig. 6.40 shows the stress resultants in the shotcrete membranefor the 3rd computation step. An approximate membrane state ofstress is computed. The normal thrust is ranging between 400 and1100 kN/m corresponds to only one tenth of the force resultingfrom the overburden. This confirms that the shotcrete membrane isonly subjected to slight loading because of the high Young'smodulus of the rock mass and the arching effect. The dimensioningyields that no reinforcement is statically required for the shot-crete membrane.
The excavation of the shoulders (4th and 5th computation step) doesnot lead to significant changes relative to the 3rd computationstep.
6.2.6 Stability analyses for the design of the interior lining
Investigated load cases and load combinations
Two-dimensional FE-analyses were carried out for the design of theinterior lining as well.
Fig. 6.41 shows exemplarily the FE-mesh, the boundary conditionsand the parameters specified for the design of the interior liningfor analysis cross-section AC 2 (see Fig. 6.26).
The computation section consists of a 1 m thick, 75 m wide and162 m high slice subdivided into 2397 isoparametric elements with14090 nodes.
Fig. 6.42: Analysis cross-section AC 2 (interior lining),FE-mesh, detail
Unlike the stability analysis for the design of the shotcrete sup-port, the decomposed granite and weathered granite (see Fig. 6.35)were not modeled here, since these layers are insignificant for
the loading of the interior lining. A loading modulus ofEL = 10000 MN/m2 was assumed for the unweathered granite (Fig.6.41 and 6.42). This value was derived from a comparison of thedisplacements measured during heading and the analysis results("back-analysis", see Chapter 6.2.7).
The remaining parameters, the overburden and the specified bound-ary conditions correspond to those of the stability analysis forthe design of the shotcrete support.
In Fig. 6.42, a detail of the FE-mesh is shown. The interior lin-ing is modeled with a thickness of 30 cm , the shoulders with athickness of 50 cm. The seepage water drainage is not discretized.
For the design of the interior lining it is assumed that the shot-crete will be decomposed in the course of time and lose its bear-ing capacity. The assumed parameters for the decomposed shotcreteare given in Fig. 6.42. Since the interior lining is only sub-jected to significant loads after having reached its finalstrength, the calculation value for Young's modulus of 34000 MN/m2
commonly used for concrete of grade B35 is assumed (DIN 1045,1988).
The following load cases and load combinations, respectively, wereinvestigated for the design of the interior lining:
- Dead weight of the interior lining (DW)
- dead weight as before and rock mass pressure (DW + RP)
The seepage water is to be lowered to the invert's level by thelateral seepage water drainages (see Fig. 6.28 and 6.29). Thusthere is not any water pressure acting on the interior lining, andno seepage pressure is applied to the rock mass above and closelybeside the tunnel's cross-section. Such loadings are therefore notaccounted for in the stability analyses for the interior lining.
The computation steps of the stability analyses for the design ofthe interior lining are shown in Fig. 6.43.
In the 1st computation step, the primary state in the undisturbedrock mass is computed. A preceding stress relief is simulated inthe 2nd computation step with av = 0.5 according to (4.1).
Fig. 6.43: Analysis cross-section AC 2 (interior lining), com-putation steps
The 3rd computation step includes the full-face excavation of thecross-section and the simultaneous installation of the shotcretesupport, which carries the rock mass pressure with a Young'smodulus of 15000 MN/m2.
In the 4th computation step, the installation of the interior lin-ing and thus load case DW is simulated. The shotcrete support isstill sustainable in this state and therefore able to continue tocarry the rock mass pressure. To account for the sealing betweenthe shotcrete membrane and the interior lining, no shear and ten-sile forces can be transferred in computation steps 4 and 5. Thisis simulated by insertion of a thin row of elements between theshotcrete membrane and interior lining elements. This row of ele-ments is assigned a stiffness of approx. zero in the 4th and 5th
computation step. The opposing nodes of this element row arelinked by truss elements (see Fig. 6.42), which can transfer com-pressive forces, but not tensile forces or shear.
In the 5th computation step, the shotcrete is assumed decomposed.As a result, the shotcrete membrane loses its bearing capacity andthe interior lining must carry the rock mass pressure in additionto its dead weight.
The stress resultants of the interior lining due to dead weight(4th computation step) are shown in Fig. 6.44. If a B35 concretegrade, a lining thickness of 30 cm in the vault and 50 cm at theshoulders, a cover of the reinforcement of d1 = 5.5 cm and safetyfactors according to DIN 1045 (1988) are assumed, it follows thatcircumferential reinforcement is not statically required. Due tothe small shear forces in load case DW also a shear reinforcementis not needed.
Fig. 6.44: Analysis cross-section AC 2, stress resultants inthe interior lining, load case DW (4th computationstep)
In the 5th computation step (load combination DW + RP) the shot-crete membrane is assumed decomposed. Stress redistributions re-sult for this computation step compared to the preceding one. Theground carries a portion of the load previously supported by the
intact shotcrete. On the other hand, a portion of the rock masspressure carried by the shotcrete membrane before is taken on bythe interior lining. This is apparent from the increase of thestress resultants from the 4th to the 5th computation step (Fig.6.44 and 6.45). Particularly the normal thrust in the vault andthe shear force in the shoulders increase markedly as compared tothe 4th computation step.
For the tunnel section, which is located in granite and porphy-rite, the design of the interior lining yields that reinforcementis not statically required (Fig. 6.46). The interior lining wastherefore constructed with plain concrete in this section. Only inthe area of the special cross-sections as the breakdown bays, thecross-connections, the niches and the blockouts, a constructivereinforcement was installed (Fig. 6.47).
In the portal zones, in the weathered rock mass, in the base sedi-ments and in the Lower Triassic sandstone a reinforcement wasstatically required, however.
In total, it was possible to construct the Berg Bock Tunnel overapprox. 50 % of its length with an interior lining made of plainconcrete.
For the shoulders, the load combination dead weight and rock masspressure results in a required shear reinforcement of 7.7 cm2/m2
for analysis cross-section AC 2 (Fig. 6.46). The proof of limita-tion of crack width according to DIN 1045 (1988) leads to a re-quired reinforcement for the shoulders of 12.88 cm2/m in both, lon-gitudinal and transverse direction. This amount of reinforcementis covered by top and bottom rebars Ø 10 mm spaced at s = 10 cm,to be placed in longitudinal and transverse direction. To coverthe required shear reinforcement, steel fabric mats were bent tostirrup cages (Fig. 6.48). The shoulders were reinforced over theentire tunnel length.
Fig. 6.48: Analysis cross-section AC 2, reinforcement of theshoulders
6.2.7 Monitoring
The heading of the Berg Bock Tunnel was accompanied by a geotech-nical monitoring program.
In the longitudinal section of Fig. 6.49 the range of the roofsubsidence measured after the heading of the entire tunnel wascompleted is exemplarily shown.
The largest subsidence was measured in the portal areas with val-ues between 10 and 50 mm. In those tunnel sections where thecross-section is located in granite or porphyrite, a roof subsi-dence between 2 and 5 mm was measured. In the base sediments andin the Lower Triassic sandstone the roof subsidence ranges from 5to 15 mm.
As an approximation, the measurements captured only those dis-placements that occurred after the installation of the shotcretemembrane. Therefore, the measurement results in the granite areamust be compared to the computed displacements shown in Fig. 6.39b(3rd – 2nd computation step). A roof subsidence of approx. 2 mm wascomputed. This analysis is based on a loading modulus of the un-weathered granite of EL = 15000 MN/m2 (see Fig. 6.35). A compara-tive analysis with EL = 10000 MN/m2 yields a roof subsidence of ap-prox. 4 mm. It is thus possible to reproduce the roof subsidencemeasured in the granite and the porphyrite well in the analysesusing a loading modulus of approx. 10000 MN/m2 (see Fig. 6.49).This value was therefore taken as a basis for the stability analy-ses of the interior lining.
6.2.8 Conclusions
The Berg Bock Tunnel is situated in granite and porphyrite over alength of 2 × 2000 m corresponding to 75 % of its total length. Inthese sections, the tunnel was headed by full-face excavation us-ing the drill and blast method with comparatively great roundlengths and limited support measures.
Due to the low deformability and the high strength of the rockmass, the ground was able to carry approx. 90 % of the overburdenload, and the means of support only had a slightly assisting func-tion. On this basis it was possible to optimize the heading con-ception and to excavate both tunnel tubes in a very short time.With the construction of the interior lining using plain concreteover approx. 50 % of the total tunnel length, the costs for theinterior lining could be kept low as well.
An important tool for the optimization of the heading and themeans of support were the FE-analyses. Their results were con-firmed by the experience made and the monitoring during the exca-vation.
7. Heading under the protection of jet grouting columns
7.1 Road tunnel for the federal highway B 9 in Bonn-BadGodesberg, Germany
7.1.1 Introduction
In the city of Bonn–Bad Godesberg, Germany, the federal highway B9 was relocated into a tunnel over a length of approx. 1200 m(Fig. 7.1). The tunnel undercrosses buildings as well as tracks ofthe German Rail (Deutsche Bahn AG) and the city railway.
The tunnel cross-section is located in gravel sand. The groundwa-ter table lies above the tunnel's invert. To guarantee the stabil-ity and to limit the subsidence due to tunneling, the tunnel washeaded by the NATM under the protection of advancing jet groutingcolumns (DIN EN 12716, 2001).
7.1.2 Structure
Two tunnel tubes with two lanes and a width of approx. 11 m eachwere excavated over a length of approx. 1000 m (Fig. 7.1 and 7.2).The overburden of the tunnel tubes amounts to approx. 6 to 8 m(Fig. 7.2). Approximately in the middle of this section the tun-nels undercross the ICE/IC (Intercity Express/Intercity) line Co-logne-Koblenz of German Rail as well as a city railway tunnel. Thelatter is located at the construction pit Moltke square (Fig.7.1).
This tunnel section is followed by the construction pit Van-Grootesquare (Fig. 7.1). In the approx. 200 m long tunnel section southof the construction pit Van-Groote square the four lanes runthrough a tunnel tube which is divided into three sections (Fig.7.1b and 7.3). This tunnel tube has a total width of approx. 30 mand an overburden of approx. 7 m (Fig. 7.3).
A mouth-shaped profile was selected for the cross-section of thetwo two-lane tunnel tubes. Fig. 7.4 shows the geometry of the11.2 m wide and 10.1 m high standard profile. The excavated cross-section amounts to approx. 92 m². The shotcrete membrane is 25 cmthick. The cross-section was excavated in two parts, as detailedin Chapter 7.1.4. The temporary invert of the partial excavation
was rounded with R = 13.3 m and supported by a 20 cm thick shot-crete membrane. The reinforced concrete interior lining was con-structed 40 cm thick (Fig. 7.4).
Fig. 7.1: Road tunnel Bonn–Bad Godesberg: a) Site plan;b) schematic representation of the structure
Relatively large radii of curvature were selected for the shot-crete support in the roof and sidewall areas with R = 4.726 m andR = 6.426 m. As a consequence the loading of the shotcrete mem-brane by bending moments and shear forces is small in these areas,as shown below.
The transitions from the sidewalls to the temporary invert and thepermanent invert, respectively, were constructed with relativelysmall radii (Fig. 7.4). This leads to bending moments and shearforces in the shotcrete membrane in these areas. As shown below,at a radius of 1.2 m supplementary reinforcement is required at
the transitions from the sidewalls to the temporary invert in ad-dition to the planned reinforcement with inside and outside steelfabric mats Q188. At the transitions from the sidewalls to thepermanent invert, however, no additional reinforcement is requiredat a radius of 1.8 m (see Chapter 7.1.5).
Fig. 7.3: Tunnel tube divided into three sections, cross-section
The invert was slightly rounded with a radius of curvature ofR = 15.4 m because of its loading by the water pressure (Fig.7.4). A greater curvature would lead to a smaller amount of rein-forcement in the interior lining, but also to additional excava-tion and would therefore not be economical.
Rhine valley in the area of Bonn. Below the tunnel invert siltlenses are sporadically embedded in the sand and gravel layerswhich will be termed gravel sand in the following (Fig. 7.5).
The gravel sand is covered by an up to 7 m thick silt layer ex-tending to the ground surface. Below the gravel sand is the Devo-nian base rock consisting of mudstone and sandstone layers (Fig.7.5).
Fig. 7.6 shows the grading ranges of the encountered gravel sandand the silt, together with the parameters derived from the sub-soil exploration results.
The gravel sand has a high porosity and permeability. It consti-tutes an aquifer connected to the Rhine river. The groundwater ta-ble is therefore influenced by the water levels of the Rhine. Onaverage the groundwater is encountered approx. 3 m above the tun-nel's invert (Fig. 7.5).
Fig. 7.6: Grain-size distribution of the soils and soil me-chanical parameters
7.1.4 Design and construction
The sequence of excavation and the support measures for the head-ing of the two two-lane tunnels are shown in Fig. 7.7 and 7.8 incross- and longitudinal section.
First the part of the tunnel cross-section located above thegroundwater table was excavated. For statical reasons the tempo-rary invert was rounded and supported by shotcrete ( in Fig.7.7). Regularly spaced gravity wells were drilled from the tempo-rary invert. With these wells, the groundwater table was loweredto the final tunnel invert ( in Fig. 7.7). Protected by thisgroundwater drawdown the tunnels were excavated in stages down tothe invert and supported ( in Fig. 7.7).
Fig. 7.7: Heading of the two-lane tunnels, excavation and sup-port, cross-section
The partial excavation above the groundwater table was subdividedinto crown, bench and invert and carried out with a stepped tunnelface. The round lengths of the partial excavations amounted to 1 meach. To limit the subsidence the temporary invert was closed 6 to8 m behind the roof excavation. This lead to an inclination of thetunnel face of 60° (Fig. 7.8). Because the tunnel face inclination
exceeds the angle of friction of the gravel sand (see Fig. 7.6),the tunnel face was not stable if the apparent cohesion was nottaken into account. Since the latter quickly vanishes with thesoil drying up, only small sections could be excavated in onestep. These sections were immediately sealed with reinforced shot-crete.
Fig. 7.8: Heading of the two-lane tunnels, excavation and sup-port, longitudinal section
The excavation contour was supported using reinforced shotcreteand steel sets (Fig. 7.7 to 7.9).
As already mentioned, the tunnel tubes were excavated under theprotection of advancing jet grouting columns forming a jet grout-ing vault (Fig. 7.7 and 7.8). This jet grouting vault transfersloads in transverse and longitudinal tunnel directions (Fig.7.10). Thus the green shotcrete close to the tunnel face as wellas the tunnel face area were less strongly loaded. Furthermore,with the jet grouting columns, the subsidence is limited, col-
lapses are avoided, and the safety of the tunneling staff is thusensured as well.
Fig. 7.9: View of the temporary tunnel face
The jet grouting columns are approximately horizontal, 15 m longand have a design diameter of 63 cm. Two successive jet groutingvaults overlap by 3 m (Fig. 7.8 and 7.11).
Fig. 7.11: Heading of the two-lane tunnels, excavation and sup-port, plan view (section I-I, see Fig. 7.8)
In order to comply with the tunnel clearance, the jet groutingcolumns were not constructed horizontally but rather at a slightoutward slant (Fig. 7.8 and 7.11). The excavation had thus to bewidened in a trumpet-shaped way (Fig. 7.8, 7.11 and 7.12).
An additional stabilization of the tunnel face was achieved by thesupport core shown in Fig. 7.11 and 7.12. Beside and above thissupport core was enough space for the jet grout drill carriage(Fig. 7.11 and 7.13). The tunnel face was supported in sectionsusing reinforced shotcrete (Fig. 7.8 and 7.11).
The jet grouting columns were constructed by the single-phasemethod (DIN EN 12716, 2001), according to which in a borehole ce-ment based suspension is injected under high pressure (approx.500 bar) into the soil via a rod and a nozzle at its end.
The nozzle rotates with the rod which is slowly pulled out of theborehole. In this way a column develops due to the mixing of thesuspension with the ground. After hardening of the cement, thiscolumn possesses a high strength in comparison with the undis-turbed soil. The surplus mixture of suspension and soil exits asbackflow through the annular gap between borehole and rod. Fig.7.13 shows the jet grout drill carriage in operation.
To optimize the production parameters six test columns were con-structed and dug out. Column diameters ranging from 60 to 90 cmwere obtained. The production parameters are listed in Table 7.1.The parameters in the lines marked with arrows in Table 7.1 wereselected for the construction of the jet grouting columns. Columndiameters between 60 and 70 cm were achieved with these parameters(Wittke et. al, 2000).
Parameters selected for the construction of the jet groutingcolumns
Table 7.1: Production parameters of six test columns
A comparatively small unconfined compressive strength ofσD = 0.75 MN/m2 was demanded for the jet grouting columns and ac-counted for in the stability analyses (see Chapter 7.1.5). Thisstrength was achieved already after a short time and the idle timecausing high cost and long construction times could be minimized.
Fig. 7.14 shows the unconfined compressive strength measured ondrill cores taken and on samples from the backflow as a functionof the sample age (Wittke et al., 2000). The diagram also includesthe unconfined compressive strength determined on drill cores andbackflow samples of jet grouting columns constructed during theheading of the city railway tunnel "Killesberg-Messe" in Stuttgart
(see Chapter 7.2). According to this diagram, values ofσD = 0.75 MN/m2 are achieved already after one day.
Fig. 7.14: Unconfined compressive strength measured on backflowsamples and drill cores versus sample age
The construction of the jet grouting columns included comprehen-sive quality management measures. The success of these measuresbecame apparent during the excavation. No defects were found inthe jet grouting vaults.
To determine the influence of the vibrations due to tunneling aswell as railway and road traffic on the stability of the tunnelface, the vibration velocity was measured during tunneling. Theoccurring vibrations turned out to be small. As an optional posi-tion it was planned to additionally stabilize the tunnel face us-
ing jet grouting columns (Fig. 7.15). To reduce the strength aportion of the cement in the suspension would have been replacedwith bentonite to facilitate the later demolition of these col-umns. The construction of tunnel face columns did not become nec-essary, however.
Fig. 7.15: Stabilization of the tunnel face using jet groutingcolumns (not carried out)
In the area where the tunnel tube is divided into three sections,the side tubes were excavated first under the protection of jetgrouting columns and supported. The central tube was only exca-vated after the interior lining had been installed in both sidetubes. A crown excavation with trailing bench and invert was car-ried out here. The shotcrete membrane of the central tube herebyis supported by the interior linings of both side tubes (Fig.7.16).
Fig. 7.16: Excavation sequence/construction stages for the tun-nel tube divided into three sections
7.1.5 Stability analyses for the design of the shotcretesupport
To design the shotcrete support two- and three-dimensional FE-analyses were carried out with the program system FEST03 (Wittke,2000). Fig. 7.17 shows the location of the 10 analysis cross-sections (AC 1 to AC 10) in the geological cross-section which areinvestigated in the design analyses.
Fig. 7.18 shows exemplarily the computation section, the FE-mesh,the boundary conditions, the ground profile and the parameters fora three-dimensional analysis.
Fig. 7.17: Location of the 10 analysis cross-sections
In the following the results of a two-dimensional analysis foranalysis cross-section 2 are presented. Analysis cross-section 2is located in the section of the two two-lane tunnel tubes (seeFig. 7.17).
Fig. 7.19 shows the computation section, the FE-mesh, the boundaryconditions, the ground profile and the parameters this analysiswas based upon. The computation section consists of a 25 m wide,30.5 m high and 1 m thick slice of the ground. The FE-mesh wassubdivided into 669 three-dimensional isoparametric elements witha total of 797 nodes.
For the nodes on the lower boundary (z = 0), horizontally slidingsupports were assumed as boundary conditions. Vertically slidingsupports were introduced for the nodes located on the verticallateral boundary planes (x = 0 and x = 25 m) (Fig. 7.19). Allnodes were assumed fixed in y-direction. The loading due to build-
In the stability analyses for analysis cross-section 2 the headingof only one tunnel tube was investigated. Because the distance ofthe two tunnel tubes amounts to more than one tunnel diameter inthis area, the two tubes influence each other only to a small de-gree. Since symmetry exists with respect to the tunnel axes, theground profile and the cross-sectional shape of the tunnel tubes,only one half of a tunnel tube was modeled. The vertical sectionthrough the tunnel axis constitutes the plane of symmetry (Fig.7.19).
The shotcrete membrane with a thickness of d = 25 cm was modeledby one element layer. Three element layers with a total thicknessof 60 cm were selected for the simulation of the jet groutingvault (Fig. 7.20).
The parameters chosen for the undisturbed soil and the soil stabi-lized with jet grouting columns were determined on the basis ofthe exploration results by the parties concerned during technicaldiscussions. They are shown in Fig. 7.19.
The deformability and the strength of the soil stabilized by thejet grouting columns develop with time (see Fig. 7.14). In agree-ment with the parties concerned, values of E = 500 MN/m2, ϕ' = 35°and c' = 200 kN/m2 were specified for Young's modulus and the shearstrength parameters of the jet grouting vault. These shearstrength parameters correspond to an unconfined compressivestrength of σD = 0.75 MN/m2. According to Fig. 7.14, these valuesare attained after a few days already.
Young's modulus assumed for the shotcrete was varied in the sta-bility analyses. In the following the computation sequence and theresults of an analysis are presented, in which a modulus ofE = 7500 MN/m2 was selected for the shotcrete (Fig. 7.19). Thisrelatively small value reflects the development of strength anddeformability and the creep properties of the shotcrete (see Chap-ter 2.1). In the case presented here the shotcrete is loaded at avery young age due to the early closing of the shotcrete supportapprox. 6 to 8 m behind the crown excavation (see Fig. 7.8).
Fig. 7.21 shows the eight computation steps applied to simulatethe excavation and support of the tunnel. In the 1st computationstep, the state of stress and deformation resulting from the deadweight of the soil and the loading due to buildings (in-situstate) is determined. In the 2nd computation step the installationof the jet grouting vault is simulated. A preceding stress reliefof the soil in the area of the crown excavation is modeled in the3rd computation step (Wittke, 2000). The reduced Young's modulus ofthe gravel sand of Ered = 45 MN/m2 corresponds to a stress relieffactor of aV = 0.6 according to (4.1). The 4th computation steprepresents the crown excavation and its support using shotcrete.In the 5th and 6th computation steps, the preceding stress relief ofthe soil in the area of the bench excavation as well as the benchexcavation with temporary invert support using shotcrete are simu-lated.
groundwater table to the invert level and the excavation and shot-crete support of the invert are simulated in the 8th computationstep.
In Fig. 7.22 the nodal displacements computed for the 8th computa-tion step related to the in-situ state (1st computation step) areshown in horizontal sections above the tunnel roof. The subsidenceof the ground surface above the tunnel roof amounts to 33 mm.
Fig. 7.23 shows a comparison of the bending moments in the shot-crete lining computed for the 6th and the 8th computation step. Dueto the small radius of curvature of R = 1.2 m in the area of theconnection of the temporary invert to the sidewalls (see Fig.7.4), comparatively large bending moments occur at this locationin the 6th computation step as previously mentioned (Fig. 7.23a).In addition to the planned steel fabric mats Q188 supplementaryoutside reinforcement becomes necessary in this area (Fig. 7.24a).In the 8th computation step the radius amounts to R = 1.8 m in thearea of the connection of the permanent invert to the sidewalls(see Fig. 7.4). Smaller bending moments are computed for this step(Fig. 7.23 a and b). They can be carried by the shotcrete membranewithout any reinforcement for an assumed safety factor of 1.35(Fig. 7.24b).
Fig. 7.25 shows the design of the support in the area of thebench's foot.
7.1.6 Monitoring
The subsidence due to the heading was measured by leveling using aclosely spaced raster of measurement cross-sections. In addition,subsidence and convergency measurements were carried out in thetunnel tubes.
Fig. 7.26 shows the maximum values of the ground surface subsi-dence measured during the heading between the two constructionpits Moltke square and Van-Groote square. The largest subsidenceof approx. 6.5 cm occurred close to the construction pit Moltkesquare in an area where the tunnel tubes were headed by sidewalladit excavation without jet grouting columns. In the other areasthe maximum subsidence of the ground surface ranged between ap-prox. 15 mm and approx. 45 mm (Fig. 7.26). The maximum ground sur-face subsidence of 33 mm computed for analysis cross-section 2 isin good agreement with the measured values (see Fig. 7.22 and7.26).
7.1.7 Conclusions
With the relocation of the federal highway B 9 in Bonn-Bad Godes-berg into a tunnel the whole tunnel cross-section was located incohesionless gravel sand. Buildings and railway facilities had tobe undercrossed with little overburden. Therefore, the subsidence
of the ground surface due to the heading had to be limited tosmall values.
Fig. 7.26: Maximum values of the measured ground surface subsi-dence, longitudinal section between the two con-struction pits Moltke square and Van-Groote square
The tunnel was excavated by the NATM under the protection of ad-vancing jet grouting columns. With this measure a part of theoverburden load could be transferred in lateral and longitudinaltunnel direction (see Fig. 7.10). Thus the green shotcrete in thetunnel face area and the tunnel face itself was less stronglyloaded. Furthermore, with the jet grouting columns, the subsidencewas limited, collapses were avoided, and the safety of the tunnel-ing staff was thus ensured as well. The tunnel was excavated witha steeply inclined stepped tunnel face, short round lengths and afast closing of the invert support (see Fig. 7.8). To guaranteethe stability of the tunnel face, an immediate tunnel face supportin sections using reinforced shotcrete was necessary.
With these measures, a stable excavation of the tunnel was feasi-ble, and the subsidence of the ground surface could be limited to2 to 4 cm (see Fig. 7.26). No damage occurred on buildings orrailway facilities.
The FE stability analyses carried out for this project representedan essential contribution towards the design, the statics and thedesign of the excavation and support measures.
7.2 City railway tunnel "Killesberg-Messe" in Stuttgart,Germany
7.2.1 Introduction
Between July 1990 and April 1991 the "Killesberg-Messe" city rail-way tunnel was constructed in Stuttgart, Germany. A 64 m long sec-tion of the tunnel alignment is located immediately adjacent tothe State Academy of Art and Design (Academy of Art, Fig. 7.27 and7.28). In addition, the tunnel was driven through a quarry fill inthis area. To keep the subsidence due to tunneling small, the tun-nel was headed in this section under the protection of an advancesupport constructed by jet grouting columns (EN 12716, 2001).
Fig. 7.27: Killesberg-Messe city railway tunnel, site plan
Fig. 7.28: Killesberg-Messe city railway tunnel, longitudinalsection with ground profile, excavation and supportmeasures
7.2.2 Structure
With the "Killesberg-Messe" city railway line in Stuttgart, theso-called Trade Fair Line, a fast and convenient rail transit con-nection was constructed between the central junction StuttgartCentral Station and the Killesberg heights with the hill park.Over a length of approx. 360 m up to the Killesberg-Messe stationthe city railway line runs in a tunnel (Fig. 7.27). The overburdenof the tunnel first rises to approx. 16 m, then decreases andamounts to approx. 6 m at the beginning of the underground Killes-berg-Messe station (Fig. 7.28).
From km 0+500 to km 0+710 the double-track standard profile wasexcavated. With a width of approx. 10.5 m, the excavated cross-section amounts to approx. 70 m² in this section. In the furthercourse up to the Killesberg-Messe station the tunnel widens to awidth of approx. 17.8 m. The excavated cross-section at the begin-ning of the station amounts to approx. 174 m2 (Fig. 7.27 to 7.29).
Fig. 7.29: Enlarged cross-section at the beginning of the Kil-lesberg-Messe station (km 0+860)
A mouth-shaped profile was selected for the tunnel cross-section.The geometry of the enlarged cross-section at the beginning of theKillesberg-Messe station (largest cross-section) is shown in Fig.7.29. The shotcrete membrane has a thickness of t = 35 cm, the in-side sidewalls of the two sidewall adits had a 25 cm thick shot-crete membrane. If required the shotcrete membrane of the sidewalladits was planned to be closed at the invert with t = 20 cm. Theinterior lining was constructed 80 cm thick with watertight con-crete of grade B35.
In the roof and sidewall areas of the largest cross-section radiiof curvature of R = 9.8 m and R = 5.4 m, respectively, were se-lected for the shotcrete support. The transitions from the side-walls to the invert were constructed with the comparatively small
radii R = 3.8 m and R = 1.9 m. Because of the water pressure load-ing of the interior lining the invert was slightly rounded with aradius of curvature of R = 15.8 m (Fig. 7.29).
7.2.3 Ground and groundwater conditions
The plateau of the Killesberg heights is formed by the Schilfsand-stone, which was mined in numerous quarries in the past. Thesequarries were later closed and backfilled 80 to 90 years ago withquarry fill (Fig. 7.30). Below the Schilfsandstone the layers ofthe Gypsum Keuper are encountered (Fig. 7.28).
At the portal (km 0+500) the city railway tunnel cuts into talusmaterial and the upper layers of the Gypsum Keuper (Estherien lay-ers). Starting at km 0+650 the Schilfsandstone enters into thetunnel profile from the roof. From km 0+700 to the end of the tun-nel at km 0+860 the roof and the upper part of the sidewalls arelocated in the quarry fill. The lower part of the sidewalls andthe tunnel invert cut into the Schilfsandstone. The boundary tothe Gypsum Keuper is located in this area at the level of the tun-nel invert or slightly below (Fig. 7.28).
The ground conditions were mainly derived from the results of coredrillings. In addition, test pits were excavated and dynamic prob-ing were carried out. The evaluation of old aerial photographs andseismic measurements served to localize the quarry edges. In frontof the Academy of Art a test shaft 4 m in diameter was sunk. Heresamples for soil mechanical laboratory tests were taken and plateloading tests were carried out to determine the deformability.
The quarry fill is very heterogeneously composed. It consists ofsandstone blocks of different sizes with edges up to 80 cm long.Partly the pieces of rock lie in a sandy silt matrix, partly theyconstitute a pure sandstone deposit without any filling material(Fig. 7.30). According to the geological report the porosityranges between 3 and 35 %. By means of density measurements onlarge-scale samples a content of 10 % of large voids or cavities,respectively, was estimated. The grain size distributions deter-mined for five samples are shown in Fig. 7.31. According to this,the fill consists of smoothly graded gravel/sand (GW), gravel/silt(GU) and sand/silt mixtures (SU).
between 10 and 50 cm thick. The sandstone is mostly verticallyjointed with a medium to wide joint spacing.
The Gypsum Keuper layers in the tunnel invert area belong to theWhite and Grey Estherien layers. Anhydrite or gypsum deposits werenot encountered in the course of the ground exploration down toapprox. one tunnel diameter below the invert. Swelling phenomenadue to the tunneling were therefore not to be expected in theground. No leaching cavities were drilled into either in the areaspecified above.
The soil and rock mechanical parameters of the different groundlayers determined or estimated from the exploration results arelisted in Table 7.2.
intact rock discontinuity sets
bedding jointing
layer
Young'smodulus
E[MN/m2]
Pois-son's -ratio
ν
unit
weightγ
[kN/m3]
ϕ
[°]c
[kN/m2] ϕB
[°]cB
[kN/m2]ϕJ
[°]cJ
[kN/m2]
talus
material7 0,40 21,5 25 25 - - - -
quarry
fill6 0,25 19 30 0 - - - -
Schilf-
sandstone2000 0,20 25 40 3000 40 30 40 0
Gypsum
Keuper100 0,33 23 30 50 30 20 30 20
Table 7.2: Soil and rock mechanical parameters
Five core drillings were equipped as observation wells. The meas-ured water levels show that the ground water table lies approxi-mately on the level of the quarry base and thus above the tunnel'sinvert. The boundary between the permeable Schilfsandstone and theGypsum Keuper with its low permeability forms a preferred ground-water horizon. Above the quarry edge seepage water has accumu-lated. An amount of 1 l/s of water was pumped out of the testshaft mentioned above over a period of several days.
Crown and bench were excavated with a stepped tunnel face and asupport core in the crown area and with round lengths of 2.4 to3.3 m. To ensure the stability of the temporary tunnel face, itwas supported in sections with reinforced shotcrete. The invertwas excavated 11 to 15 m behind the bench. The distance betweenthe crown face and the closing of the support at the invertamounted to between 16 and 22 m, depending on the ground condi-tions encountered (Fig. 7.32b).
Because of the large cross-section at the end of the widening seg-ment the tunnel was driven in the further course up to Killesberg-Messe station (km 0+808 to 0+860) by sidewall adit excavation(Fig. 7.28, 7.29 and 7.32c). The excavation profile was supportedwith reinforced shotcrete, steel sets and a systematic anchoringusing SN-anchors, the length and spacing of which were determinedas required.
The foundations of the Academy of Art are located at close dis-tance from the tunnel (Fig. 7.33a). In this section of the align-ment the tunnel roof and the upper sidewalls are located in thequarry fill. Because of the high deformability and low strength ofthe fill, the close distance of the foundations of the main build-ing of the Academy of Art to the tunnel and the large tunnelcross-section, it was feared that tunneling-induced subsidencewould lead to damages to the main building of the Academy of Art.
To limit the subsidence, the tunnel was excavated in the area ofthe quarry fill under the protection of an advance support by jetgrouting columns. To this end, from km 0+710 to km 0+785 seven jetgrouting vaults were constructed in sections by the single phasemethod described in Chapter 7.1 (Fig. 7.33b). The jet groutingcolumns were constructed with a length of 11.0 to 12.75 m, meas-ured from the tunnel face. At the end of each excavation sectionthe columns extended 3 m beyond the tunnel face. With 1 m of non-grouted borehole length, this results in an overlap of the columnsof 2 m (Fig. 7.34b).
The drillings were directed with a 6 degree outward slant with re-spect to the tunnel axis. Thus, at the end of the enlarged excava-tion section there was enough clearance to construct the columnsfor the following section (Fig. 7.34b).
ment based suspension from seeping into large voids or cavities inthe quarry fill, and in order to ensure that the columns could beconstructed according to plan with respect to diameter, continu-ity, strength and position, the quarry fill was grouted in advanceto fill existing cavities (Fig. 7.34).
To this end drillings 114 mm in diameter with air flushing weresunk, equipped with PVC sleeve pipes and grouted with packers insteps of 1 m from the bottom up with a cement-bentonite grout(250 kg cement and 40 kg bentonite for 1000 l) which is stablewith respect to sedimentation (Table 7.3). The purpose of this wasto achieve a void-free matrix filling potential cavities in thequarry fill which would not impede the construction of the jetgrouting columns and the tunnel excavation.
Water/cement (PZ 35 F) ratio of groutPump pressurePump capacityRetracting rateRotational speedGrout flow rateCement quantity
- bar
l/minm/minmin-1
l/minkg/m
0,840095
0,4120
233205
Table 7.4: Operational parameters of the jet grouting
Only a few hours after the completion of the grouting of thequarry fill the construction of the jet grouting columns started.With a planned minimum diameter of the columns of 0.5 m, thedrillings were placed at a distance of 0.35 m to the tunnel con-tour. To confine temporary decrease in strength in the soil lo-cally, the columns were constructed first with a spacing of 1.4 m,
then halfway between two columns each, and finally in the gaps be-tween two adjacent columns. This way a vault supported by theSchilfsandstone was constructed made up of intersecting or, in thearea more widely fanned out, touching columns. The operational pa-rameters of the jet grouting and the amount of work done are givenin Tables 7.4 and 7.5 (Beiche et al., 1991).
Fig. 7.14 shows the unconfined compressive strength measured onbackflow samples and drill cores as a function of the sample age(Wittke et al., 2000).
The heading generally recommenced 12 to 15 hours after the comple-tion of a jet grouting vault. At this time the columns were be-tween 12 hours and 2 days old.
An essential technical requirement for advance support by jetgrouting is the ability to construct the single structural ele-ments – the soil-concrete columns – in a self-contained continuousoperation. Suitable technical equipment must therefore above allpossess a feeding length at least equal to the length of a jetgrouting column.
For economic reasons it is important that the drill mount can bequickly positioned as desired over the entire excavation profile.
Fig. 7.35 shows the construction equipment SR 510 used for theKillesberg-Messe city railway tunnel in operation. The boreholesfor the grouting of the quarry fill were drilled with the sameequipment (Beiche et al., 1991).
Table 7.6: Construction of the jet grouting vaults
The grouting of the quarry fill and the construction of the jetgrouting columns had to be worked into the heading scheme. Sincethe columns could not be constructed in parallel with the heading,the established heading operation in ten-days periods was gener-
ally changed over for the area with jet grouting columns to a con-tinuous operation on demand for reasons of time (Beiche et al.,1991).
The length of the heading section protected by jet grouting col-umns was 64 m in total. This section was constructed in 68 days.24 days thereof account for the advance grouting of the quarryfill, 18.5 days for the construction of the columns and 25.5 daysfor the heading (Table 7.6).
7.2.5 Stability analyses
For the design of the shotcrete support and the interior liningtwo-dimensional FE-analyses were carried out using the programsystem FEST03 (Wittke, 2000). Eight analysis cross-sections wereinvestigated in total, differing with respect to
- the geometry of the tunnel cross-section,
- the ground conditions and overburden height,
- the construction stages and/or
- the support installations.
Two of these analysis cross-sections are located in the area ofthe Academy of Art. In the following a stability analysis for theanalysis cross-section km 0+785 (see Fig. 7.27 and 7.28) is exem-plarily presented (see Beiche et al., 1991).
Fig. 7.36 shows the computation section, the FE-mesh, the boundaryconditions, the ground profile and the parameters for this analy-sis cross-section.
The upper part of the tunnel cross-section lies in the quarryfill. The edge of the former quarry is located approx. 2 m awayfrom the tunnel sidewall opposite to the Academy of Art. The lowerpart of the tunnel's cross-section is situated in the Schilfsand-stone. The boundary to the Gypsum Keuper is encountered approx.1.2 m underneath the tunnel's invert (Fig. 7.36).
The computation section consists of a 1 m thick, 40 m high and66 m wide slice including the tunnel's cross-section as well asthe surrounding ground. It is divided into 934 isoparametric ele-ments with a total of 2206 nodes. Vertically sliding supports areselected as boundary conditions for the nodes on the vertical lat-eral boundary planes (x = 0 and x = 66 m), while horizontallysliding supports are chosen for the nodes on the lower boundaryplane (z = 0) (Fig. 7.36). All nodes are fixed in y-direction.
The loading resulting from the Academy of Art building to theright of the tunnel is simulated by a surface load. The surfaceload is applied to the corresponding nodes of the FE-mesh in theform of point loads (Fig. 7.36).
The shotcrete membrane (t = 25 cm) is simulated by one row of ele-ments, the interior lining (t = 50 cm) by three and the jet grout-ing vault (t = 50 cm) by two element rows. The effect of a sepa-rating non-woven material preventing the transfer of tensile andshear stresses between shotcrete support and interior lining issimulated by a thin row of elements with no stiffness assigned andby radially arranged truss elements of high stiffness (pendulumrods), which only allow the transfer of normal compressivestresses (Fig. 7.37).
The soil and rock mechanical parameters of the different layers aswell as the parameters assigned to the shotcrete, the reinforcedconcrete of the interior lining and the jet grouting vault aregiven in Fig. 7.36.
It was not possible to carry out advance tests to determine themechanical parameters of the quarry fill improved by jet grouting.These parameters therefore had to be estimated for the stabilityanalyses.
The deformability and the strength of the soil improved by jetgrouting develop with time (see Fig. 7.14). Since tunneling recom-menced already approx. 15 hours after the completion of each jetgrouting vault, Young's modulus and the strength of the improvedsoil are still low when the cross-section is excavated. Young'smodulus and the shear parameters of the jet grouting vault weretherefore assumed comparatively low and thus conservative in theanalyses with values of E = 500 MN/m2, ϕ = 35° and c = 100 kN/m2.These shear parameters correspond to an unconfined compressive
A three-dimensional analysis was not considered necessary in thiscase, since a three-dimensional arching cannot develop in thequarry fill due to the low strength.
Although the jet grouting vault constructed in advance reachesdown to the quarry base and is thus supported by the Schilfsand-stone, it cannot transfer any significant loading during crown ex-cavation. The reason is that the time span between the construc-tion of the jet grouting vault and the crown excavation is veryshort. The vault has therefore only reached a small fraction ofits final strength at this stage. It is only after the bench exca-vation that the shotcrete support reaches down to the Schilfsand-stone and a setting process has taken place in the vault. Bothmeans of support are then ready to carry loads. It was thereforedetermined that the crown heading must not be ahead of the benchby more than twice of the lattice girder spacing (2.4 to 3.3 m)(see Fig. 7.32b).
Fig. 7.38 shows the seven computation steps simulating the in-situstate and the construction stages, which are the installation ofthe jet grouting vault, crown, bench and invert excavation, in-stallation of the interior lining and rise of the groundwater ta-ble to roof level.
Fig. 7.39 to 7.41 show the analysis results for the 4th computationstep (crown and bench excavation).
Above the right half of the tunnel practically no arching developsin the fill due to its low strength and large deformability. Thiscan be recognized in Fig. 7.39a from the fact that the major prin-cipal normal stresses are almost vertically oriented.
Above the left half of the tunnel in the area of the verticalquarry wall the major principal normal stress deviates from thevertical. The vertical stresses are lower in the fill and higherin the Schilfsandstone than the overburden weight. This result isdue to the different Young's moduli of the Schilfsandstone and thefill. The quarry fill is thus "hung" on the edge already in thestage before the tunnel excavation and the construction of the jetgrouting vault. This effect should be less pronounced in realitythan in the analysis, since the fill was placed in layers ratherthan in one step as simulated in the analysis.
Fig. 7.39: Analysis cross-section km 0+785: a) Principal normalstresses, 4th computation step; b) displacements, 4th
– 1st computation step
Fig. 7.40: Analysis cross-section km 0+785, 4th computationstep: a) Vertical stresses in horizontal sections;b) horizontal stresses in vertical sections
The computed roof subsidence amounts to approx. 2.5 cm. Heave oc-curs at the bench base (Fig. 7.39b).
An arch develops in the jet grouting vault above the crown. Due toits bond with the shotcrete membrane the latter is strongly loadedas well and stress concentrations result in the area of the benchbase. To illustrate the load transfer described above, the hori-zontal and vertical stresses are shown in sections in Fig. 7.40.Stress concentrations in vertical as well as in horizontal direc-tion are apparent at the base of the bench and of the jet groutingvault. The loading of the shotcrete membrane exceeds the one ofthe jet grouting vault. This is due to Young's modulus of theshotcrete being markedly higher than the one of the jet groutingcolumns (see Fig. 7.36).
Fig. 7.41 shows the computed bending moments and normal thrust inthe shotcrete membrane. Large bending moments together with a com-paratively small normal compressive thrust occur in the roof andthe lower sidewall areas.
Fig. 7.41: Analysis cross-section km 0+785, bending moments andnormal thrust in the shotcrete membrane, 4th compu-tation step
These results change only slightly with the invert excavation andthe immediate closing of the support (5th computation step).
The design of the interior lining was based on the following loadcases:
- Dead weight of the interior lining,
- dead weight of the interior lining and rock mass pressure,
- dead weight of the interior lining, water pressure (groundwa-ter table at roof level) and rock mass pressure,
- dead weight of the interior lining and water pressure (ground-water table at roof level).
In the load cases with consideration of the rock mass pressure(Fig. 7.38, 6th and 7th computation step) it is assumed that the jetgrouting columns and the shotcrete membrane are decomposed and arethus not effective any more. The rock mass pressure generallyleads to a great normal compressive thrust in the interior liningwith favorable effects on the dimensioning for bending and normalthrust.
In the load cases without rock mass pressure it is assumed thatthe support effect of the jet grouting vault and the shotcretemembrane remains intact. The surrounding rock mass and the shot-crete membrane are assumed weightless. A bedding of the interiorlining is given, however, since for the shotcrete as well as forthe bedrock Young's moduli listed in Fig. 7.36 are effective. Theload case dead weight and water pressure is generally decisive forthe bending reinforcement of the interior lining at the invert andthe sidewalls.
The statically required cross sectional areas of reinforcement forthe interior lining range between 0 and 10.5 cm2/m.
7.2.6 Monitoring
To supplement and verify the results of the stability analyses,the tunnel stability and the ground surface subsidence were moni-tored by comprehensive measurements during construction.
One extensometer measuring cross-section each was positioned infront of and behind the Academy of Art building. Surface measuringpoints every 8 m on the tunnel alignment, bolts on the outside andinside of the buildings and elevation measuring points on theceilings facing the tunnel completed the measuring points on andabove ground surface. In addition, convergency measuring cross-sections were installed every 10 m in the tunnel, with roof boltsin the middle between them.
Fig. 7.42 shows the subsidence of a selected surface measuringpoint in the area of the 4th jet grouting vault as a function ofthe heading location. It can be seen that the total subsidence ofapprox. 30 mm can be divided into three parts. Approx. one thirdof the subsidence is preceding subsidence caused by the approach-ing heading. Another third occurs during the construction of thejet grouting vault while the heading is stopped. At this stage,the tunnel face is located 3 m away from the measuring point. Thelast third of the subsidence occurs as a trailing impact after theundercrossing of the measuring point (Beiche et al., 1991).
Fig. 7.42: Subsidence of surface measuring point 111 vs. head-ing location
In Fig. 7.43 measured and computed subsidences in the area of theAcademy of Art are compared. In the FE-analysis subsidences in theorder of up to 5 mm were computed for the area of the Academy ofArt. An expert's report on the Academy's actual state and sensi-tivity to settlements lead to the result that subsidences of thismagnitude would not cause any damages. The computed subsidence wasconfirmed by the monitoring results. A subsidence trough from thefront to the back side of the building was hardly recognizable.The subsidences measured in the tunnel and on the ground surface
above the tunnel could be reproduced by the FE-analysis as well(Fig. 7.43, Beiche et al., 1991).
Fig. 7.43: Comparison of measured and computed subsidence inthe area of the Academy of Art
7.2.7 Conclusions
The Killesberg-Messe city railway tunnel crosses partly through aquarry fill. In this area the tunnel alignment is located immedi-ately adjacent to the Academy of Art, which is founded on thequarry fill and sensitive to settlements. To limit the heading-induced subsidence, the tunnel was driven in this area under theprotection of an advance support, constructed by the jet groutingmethod. To do this it was necessary to fill the large voids andcavities existing in the quarry fill in advance by cement grout-ing. This way during jet grouting the cement based suspension wasprevented from seeping into the cavities.
The tunnel was driven by crown heading with trailing bench and in-vert excavation. It was constructed with a steep tunnel face,short round lengths and a fast invert support closing (see Fig.
7.32b). To guarantee its stability, the temporary tunnel face wassupported in sections using reinforced shotcrete.
With these measures it was possible to limit the tunneling-inducedsubsidence at the Academy of Art building to a maximum of 4 mm(see Fig. 7.43).
The FE-analyses carried out have been an important tool for thedesign, the statics and the specification of the excavation andsupport measures. The comparison of the FE-analysis results withthe measured displacements in the ground showed good agreement(see Fig. 7.43).
Azuar, J. J.; Panet, M.: Shear behaviour of passive steel in rockmass. Rock bolting, Revue Industrie Minérale, St. Etienne, 1980.
Balbach, W.; Ernsperger, H.: Entwicklung und Praxiserfahrungen miteiner neuen Verfahrenstechnik zur Herstellung und Verarbeitung vonumweltfreundlichem Spritzbeton. BMI/96, 43 - 47, 1996.
Beiche, H.; Erichsen, C.; Kagerer, W.; Schilcher, E.; Wooge, M.:Praktische Lösungen bei der vorauseilenden Schirminjektion unterEinsatz von Hochdruckinjektionen (HDI). Taschenbuch für den Tun-nelbau 1992. Verlag Glückauf, Essen, 173 - 211, 1991.
Beiche, H.; Kagerer, W.: Begrenzung der Setzungen beim innerstäd-tischen Tunnelbau. Proc. 10th Nat. Rock Mech. Symp., Aachen 1992,Special Edition Geotechnik, 49 - 56, 1993.
Bauer, M.: Liquid non-alkaline accelerators of the latest genera-tion. Tunnel 6/2000, 30 - 38, 2000.
Bjurström, S.: Shear strength of hard rock joints reinforced bygrouted untensioned bolts. Proc. 3rd ISRM Congress, Denver 1974,Vol. II B, 1194 - 1199, 1974.
Böttcher, K.-H.; Schmid, I.; Erichsen, C.: S-Bahn Stuttgart -Streckenverlängerung nach Filderstadt-Bernhausen. Die Untertunne-lung des Flughafens. Vorträge der Baugrundtagung 1998 in Stutt-gart, DGGT, 67 - 74, 1998.
Brötz, K.; Löschnig, P.; Müller, F.: Shotcrete in the SchulwaldTunnel. Tunnel 6/2000, 22 - 29, 2000.
CEB: Model code for concrete structures. Comité Euro-Internationaldu Béton, Paris, 1978.
Deuse, T.; Mann, K.; Rüßmann, F.: New spray cement for processingnaturally moist aggregates for the dry spraying method. Tunnel6/1998, 37 - 41, 1998.
DB (German Rail): Guideline 853, "Eisenbahntunnel planen, bauenund instandhalten", 1999.
DIN 21521, Part 2: Gebirgsanker für den Bergbau und den Tunnelbau;Allgemeine Anforderungen für Gebirgsanker aus Stahl, Prüfungen,Prüfverfahren, 1993.
DIN EN 12716: Ausführung von besonderen geotechnischen Arbeiten(Spezialtiefbau); Düsenstrahlverfahren, 2001.
DIN 18551: Spritzbeton; Herstellung und Güteüberwachung, 1992.
Eber, A.; Betzle, M.; Baumann, T.: Untersuchungen zum Einsatz vonGitterträgern im Tunnelbau. Bauingenieur 60, 137 - 141, 1985.
EC 2: Eurocode 2 – Design of concrete structures. Part 1: Generalrules and rules for buildings, 1991.
Erichsen, C.: Grundwassermodell für räumliche, instationäre Strö-mungen in doppelt porösen Medien. Veröffentlichungen des Institutsfür Grundbau, Bodenmechanik, Felsmechanik und Verkehrswasserbauder RWTH Aachen, Vol. 26, 247 - 261, 1994.
Erichsen, C.; Keddi, W.: Das Tragverhalten vermörtelter Anker undEntwicklung eines neuen Ankertyps. Proc. 9th Nat. Rock Mech.Symp., Aachen 1990, Special Edition Geotechnik, 10 - 21, 1991.
Erichsen, C.; Tegelkamp, M.: S-Bahn Stuttgart - Streckenverlänge-rung vom Flughafen nach Filderstadt-Bernhausen. Die Untertunnelungdes Flughafens. Taschenbuch für den Tunnelbau 1999, Verlag Glück-auf, Essen, 245 - 262, 1998.
Exostra International BV: Brochure, 2000.
Forschungsgesellschaft für Straßen- und Verkehrswesen, Arbeits-gruppe Verkehrsführung und Verkehrssicherung: Richtlinien für dieAusstattung und den Betrieb von Straßentunneln (RABT), Guidelines,1994.
Grüter, R.: Erkundung des Spannungszustandes in den Schichten desSchwarzjura durch einen Großversuch bei der S-Bahn zum FlughafenStuttgart/Echterdingen. Proc. 8th Nat. Rock Mech. Symp., Aachen1988, Special Edition Geotechnik, 99 - 110, 1988.
Hauck, C.-D.; Erichsen, C.; Tegelkamp, M.: Die Ostumfahrung vonStuttgart-Vaihingen: Städtischer Tunnelbau unter komplexen Bedin-gungen der Technik und Umwelt. Vorträge der STUVA-Tagung '97 inBerlin, Forschung + Praxis, Vol. 37, 113 - 120, 1998.
Hesser, J.: Zum Einfluß unterschiedlicher Spritzbetonqualitätenauf das Tragverhalten tiefliegender Gesteinsstrecken. LaborativeUntersuchungen und numerische Analysen. Institut für Aufbereitungund Deponietechnik der TU Clausthal, Vol. 10, 2000.
Interfels GmbH: Brochure, 2000.
Keddi, W.: Numerische Untersuchungen zum Tragverhalten vermörtel-ter Felsdübel in klüftigem Fels. Veröffentlichungen des Institutsfür Grundbau, Bodenmechanik, Felsmechanik und Verkehrswasserbauder RWTH Aachen, Vol. 23, 1992.
Kiehl, J. R.; Pahl, A.: Empfehlung Nr. 14 des Arbeitskreises "Ver-suchstechnik Fels" der DGEG (German Geotechnical Society): Über-bohr-Entlastungsversuche zur Bestimmung von Gebirgsspannungen.Bautechnik 67, No. 9, 308 - 314, 1990.
Kleffner, H.-J.: Die Tunnel zur Querung des Thüringer Waldes.Proc. 14. Symp. für Felsmechanik und Tunnelbau/Eurock 2000. Aa-chen, Verlag Glückauf, Essen, 13 - 19, 2000.
Maidl, B.: Handbuch des Tunnel- und Stollenbaus. Band I: Konstruk-tionen und Verfahren. Verlag Glückauf, Essen, 1984.
Maidl, B.: Handbuch für Spritzbeton. Verlag Ernst & Sohn, Berlin,1992.
Manns, W.; Neubert, B.; Zimbelmann, R.: Spritzbeton im Test, Fe-stigkeitsentwicklung und Verformungsverhalten. Beton 8/87, 1987.
Modemann, H.-J.; Wittke, W.: Probleme bei einem flachliegendenStraßentunnel mit großen Querschnittsabmessungen in Spritzbeton-bauweise. Vorträge der STUVA-Tagung '87 in Essen, Forschung + Pra-xis, Vol. 32, 79 - 88, 1988.
Schuck, W.; Fecker, E.: Geotechnische Messungen in bestehenden Ei-senbahntunneln. Taschenbuch für den Tunnelbau 1998, Verlag Glück-auf, Essen, 44 - 84, 1997.
Spang, K.; Egger, P.: Tragverhalten und Bemessung voll vermörtel-ter, schlaffer Anker im Diskontinuum. Felsbau 7, No. 4, 181 - 189,1989.
Tegelkamp, M.; Wittke-Gattermann, P.; Züchner, F.: S-Bahn Stutt-gart - Tunnelvortrieb im wasserführenden Gebirge unter dem Stutt-garter Flughafen ohne Grundwasserabsenkung. Taschenbuch für denTunnelbau 2001, Verlag Glückauf, Essen, 21 - 40, 2000.
Weber, J. W.: Empirische Formeln zur Beschreibung der Festigkeits-entwicklung und der Entwicklung des E-Moduls von Beton. Betonwerkund Fertigteil-Technik, No. 12, 1979.
Wittke, W.: Rock Mechanics - Theory and Applications with Case Hi-stories. Springer Verlag, Berlin, Heidelberg, New York, Tokyo,1990.
Wittke, W.: Hohe Horizontalspannungen im Jura und ihre bautechni-schen Konsequenzen. Proc. 9th Nat. Rock Mech. Symp., Aachen 1990,Special Edition Geotechnik, 174 - 184, 1991.
Wittke, W.: Stability Analysis for Tunnels, Fundamentals. Geotech-nical engineering in research and practice, WBI-PRINT 4, VerlagGlückauf, Essen, 2000.
Wittke, W.; Feiser, J.; Krieger, J.; Rechtern, J.: Standsicher-heitsnachweis für einen Kalottenvortrieb in einem horizontal ge-schichteten und vertikal geklüfteten Sedimentgestein. Vorträge derSTUVA-Tagung '85 in Hannover, Forschung und Praxis, Vol. 30, 132-144, 1986.
Wittke, W.; Kiehl, J. R.: Ausbreitung sprengbedingter Erschütte-rungen und deren Auswirkungen auf Bauwerke. Taschenbuch für denTunnelbau 2001, 41 - 64, 2000.
Wittke, W.; Pierau, B.: Neubaustrecke Köln-Rhein/Main. Die TunnelNiedernhausen und Limburg. ETR Bahnreport 2000, 20 - 24, 2000.
Wittke, W.; Pierau, B.; Erichsen, C.: Anwendungsbereiche der Vor-triebsklassen der Spritzbetonbauweise. Geotechnik 22, No. 2, 124 -133, 1999.
Wittke, W.; Pierau, B.; Erichsen, C.: Der Einsatz von Hochdruckin-jektionen zur Baugrundverbesserung und für den Tunnelbau im Lok-kergestein. 15. Veder Kolloquium. Graz, 155 - 182, 2000.
Wittke, W.; Sternath, R.: 10 Autobahnunterfahrungen im Zuge derNBS Köln-Rhein/Main. Vorträge der Baugrundtagung 2000 in Hannover,DGGT, 357 - 364, 2000.
Wittke-Gattermann, P.; Wittke, W.: Simulation der Absenkung desGrundwasserspiegels beim Tunnelvortrieb im Sedimentgestein mitHilfe instationärer, räumlicher Berechnungen. Taschenbuch für denTunnelbau 1998, Verlag Glückauf, Essen, 21 - 43, 1997.
