4. Stability and water inflow in the machine area 4.1 ...

- 138 -

WBI-PRINT 6 WBI GmbH, Henricistr. 50, 52072 Aachen, Germany www.wbionline.de

4. Stability and water inflow in the machine area

4.1 Risks and selection of TBM

4.1.1 New Austrian Tunneling Method (NATM)

Applying the NATM the rock mass is usually excavated by means ofblasting or with the aid of a tunnel excavator. Compared to mecha-nized tunneling, here, we are talking about conventional tunnel-ing. According to DIN 18312, the excavation and support works,which are decisive for performance rates and costs, are defined byso-called excavation classes.

The excavation classes 1 and 2 are assigned to a heading, whicheither requires no support or only a support which does not hinderthe excavation process. Excavation class 3 characterizes a headingwith a support for which the excavation has to be interrupted andwhich follows the temporary face of the tunnel in a small dis-tance.

In traffic tunneling in sedimentary rock quite frequently excava-tion classes 4 to 7 are applied, which require the support of thetunnel contour and if necessary also of the temporary face aftereach round. The support of the tunnel contour is carried out withshotcrete, steel fabric mats, steel sets and rock bolts, if neces-sary.

Excavation class 4 requires the support of the tunnel contour onlyafter each round (Fig. 4.1).

Fig. 4.1: Excavation class 4A according to DIN 18312 (Wittkeand Wittke-Gattermann, 2006)

- 139 -


Excavation class 5 in addition requires a support of the temporaryface, e. g. with a shotcrete membrane and temporary face anchors(Fig. 4.2). In tunneling, rock bolts are installed as single an-chors to support individual rock wedges susceptible to sliding, assurface anchoring to support the tunnel contour, and as systematicanchoring to improve the load bearing capacity of the rock mass.



In case the rock mass is not stable, an advance support may becomenecessary (excavation class 6). Applying the NATM, spiles, pipeumbrellas and jet grouting columns are utilized as advance support

- 140 -


(Fig. 4.3 to 4.5). In fractured rock, spiles are installed at thetunnel face as advance support for the working area. In casesspiles are not safe enough, pipe umbrellas are installed (Fig.4.4). They are often used in fractured rock with low cohesionand/or in case of the undercrossing of buildings and roads.

Excavation class 7 includes a tunnel face support as well as anadvance support (Fig. 4.4 and 4.5).

Fig. 4.4: Excavation class 7A according to DIN 18312 with pipeumbrella (Wittke and Wittke-Gattermann, 2006)

To increase the stability of the temporary face area, the cross-section can further be subdivided for excavation classes 4 to 7(excavation classes 4A to 7A). In the case of a crown heading, thecross-section, as an example, is subdivided into crown, bench andinvert (Fig. 4.1 to 4.3). Further possibilities to subdivide thetunnel's cross-section are for example a sidewall heading withsubsequent core excavation (Fig. 4.6) or a heading with advancingroof adit. In difficult ground conditions, the described parts ofthe cross-section may be further subdivided in order to minimizethe temporarily unsupported areas.

- 141 -


Fig. 4.5: Excavation class 7 according to DIN 18312, stabili-zation with jet grouting columns

As a result, in excavation classes 4 to 6, the tunnel contourand/or the temporary face is unsupported over not more than oneround length which in most cases amounts to approx. 1 to 2 m. Ifthe tunnel's cross-section is subdivided into parts, the unsup-ported area is accordingly smaller (e. g. excavation class 4A,Fig. 4.7). In excavation class 7, only parts of the cross-sectionremain temporarily unsupported.

Thus, the conventional tunneling method is characterized by a highflexibility with regard to alternating ground conditions andshapes of cross-section. The conventional tunneling method nor-mally enables a change of excavation class, i. e. a modifiedmethod of excavation and support or a subdivision of cross-sectionas well as an adjustment of the type and amount of support, atshort notice and without problems.

- 142 -


Fig. 4.6: Sidewall adit heading (Tunnel Heslach II, Stuttgart,Germany)

Fig. 4.7: Unsupported areas at the tunnel face and the tunnelcontour, excavation class 4A according to DIN 18312

- 143 -


In DIN 18312, excavation classes are also specified for machinetunneling. Here, a change of excavation class or an adjustment tochanged ground conditions, respectively, is however only possibleto a minor degree, since the support measures to a large extentare determined by the type of TBM. This will be explained in de-tail in the following.

4.1.2 Gripper-TBM

The support measures described before for the NATM in principlecan also be applied to a mechanized tunneling with a Gripper TBM(see Fig. 2.2). However, an economical heading is only possible,if the rock mass does not require a regular support with rockbolts, steel sets or shotcrete. This is the case, if the temporaryface and the tunnel contour are stable (Fig. 4.8a).

Fig. 4.8: Stability of the unsupported opening in the workingarea of a TBM: a) temporary face and tunnel contourstable; b) temporary face stable, tunnel contour notstable; c) temporary face and tunnel contour notstable

In case of a heading with a Gripper TBM, steel sets and meshes canbe installed over long distances. In fractured rock, the installa-

- 144 -


tion must be carried out immediately behind the cutterhead or thecutterhead shield, respectively.

An advance support of the temporary face using spiles or a shot-crete support of the tunnel's contour immediately behind the cut-terhead, however, lead to a dramatical decrease of performance.Such support measures, therefore, can only be applied over shortdistances, e. g. during driving through a fault zone. If duringthe heading with a Gripper TBM extensive support measures becomenecessary, the wrong type of TBM was selected. The influence ofsupport measures on the performance rates are exemplarily illus-trated for two headings with Gripper TBMs in Sections 9.3 and 9.4.

In case of a heading with a Gripper TBM, also the bracing forcesof the grippers have to be carried by the rock mass. If the shearstrength of the rock mass is exceeded, next to the sidewalls simi-lar phenomena occur as in the case of a failure under a footing(Fig. 4.9).

Fig. 4.9: Plastic zones underneath the grippers and looseningof rock wedges above the roof because of the openingof joints due to transversal tension

Furthermore, the reduction of the horizontal normal stresses inthe rock mass above the roof due to the gripper forces, which re-

- 145 -


sults in a decrease of the shear strength on the discontinuities,may lead to a sliding or loosening, respectively, of rock wedges(Fig. 4.9). The analyses carried out during the design stage,therefore, must account for the state of stress in the rock mass,for the strength of the intact rock as well as for the influenceof discontinuities and fabric planes with reduced tensile andshear strength. If the described phenomena appear to a large ex-tent, this may lead to increased support measures and a drop ofrates of performance so that another type of TBM should have beenselected. In Section 5.4.1, it is shown by means of an example,how the stability of the excavated cross-section can be proven un-der consideration of the gripper forces.

4.1.3 TBM-S, open mode

During the heading with a TBM-S with open mode (Fig. 4.10), thetunnel face is not supported, if the minor and punctually actingsupport of the discs is neglected.

For a mechanized tunneling using a TBM-S, the unsupported area ofthe tunnel contour is considerably larger than for the conven-tional tunneling method (Fig. 4.10). A several meters long gap,depending on size and design of the cutterhead and of the shieldskin, is reaching from the cutterhead to the tail-skin. In case ofloosening at the roof (Fig. 4.8b), rock wedges are supported bythe shield skin. In this way, large overbreaks in the roof areacan be avoided. The loosened rock mass can only be supported bythe shield's skin, if the latter is sufficiently dimensioned forthis load. Suggestions with regard to corresponding load assump-tions are included in the recommendations for static analyses forshield tunneling machines of the German Committee for UndergroundConstruction (DAUB, 2005). An example for the design of a shieldskin is given in Section 9.6.

During the heading with a TBM-S with open mode, problems arise inparticular when unstable rock wedges are loosened from the tempo-rary face or when the temporary face is ripped and individual rockwedges fall into the extraction chamber (Fig. 4.8c). The orienta-tion of the discontinuities can have a favorable or unfavorableeffect on the stability of the temporary face (Fig. 4.11).

- 146 -


Fig. 4.10: Unsupported areas of a TBM at the temporary face andin the area of the shield

Fig. 4.11: Influence of the orientation of the discontinuitieson the stability of the temporary face: a) dippingtowards the temporary face; b) dipping away from thetemporary face

Ripping of the temporary face by the TBM may increase the insta-bility of the temporary face. Corresponding problems may arise, if

- 147 -


the rock mass excavated by the discs cannot sufficiently be re-moved by the conveyor openings which are arranged at the cutter-head. In this case, great quantities of excavated material can beaccumulated between the temporary face and the cutterhead. Thismaterial is rubbed over the tunnel face and thus rock blocks maybe torn out of the face. Also the scrapers at the conveyor open-ings at the cutterhead can cause a similar effect. This is thecase, if these scrapers reach too far beyond the cutterhead andscratch over the tunnel face. Since the discs are slightly pushedinto the rock mass, the scrapers in no case are allowed to stickfurther out of the cutterhead than the discs. The arrangement ofadjustable scrapers may be helpful with respect to these problems.

If the above described phenomena lead to a loosening of rockwedges also in the areas above and aside of the cutterhead, so-called excess excavation above and aside of the shield results(Fig. 4.12). The corresponding holes and disturbed rock zones, re-spectively, cannot completely be backfilled with the mortar whichis used for annular gap grouting.

Fig. 4.12: Excess excavation at the tunnel's roof caused by in-stabilities of the temporary face

As a result, the bedding of the segmental lining, which is neces-sary in particular for the single lining method, cannot beachieved and the segmental ring may be deformed due to self-weightand a potential load resulting from the disturbed rock. As a con-sequence, the longitudinal joints of the segmental lining areopened and thus the compression of the gaskets and the watertight-ness do not correspond any longer to the requirements of the de-sign (Fig. 4.13). In the case of a high groundwater level, e. g.if the tunnel is completely located underneath the groundwater ta-ble and is not to be drained, there is also the risk that the

- 148 -


loads resulting from water pressure cannot be carried by the seg-mental lining which is already deformed due to self-weight causedby an insufficient bedding.

Fig. 4.13: Deformation of the segmental lining caused by an in-sufficient bedding

If holes or disturbed rock zones, respectively, are formed abovethe roof, there is a risk that mortar penetrates into the steeringgap between shield skin and excavation contour and also into thearea of the cutterhead (Fig. 4.14). If the mortar hardens withinthe steering gap, this can lead to a clogging of the shield skinand thus to impediments of the heading. In case of a longer haltof the TBM, there is even the risk that the shield skin may be"set in concrete". At least the shear resistance existing betweenthe shield skin and the ground can be considerably increased insuch cases.

In case of a heading with a TBM-S, the aforementioned advancingsupport measures, such as tunnel face anchors and spiles, in prin-ciple can also be applied. A continuous application of such sup-port measures, however, leads to a distinct reduction of perform-ance and thus to a non-economical heading.

- 149 -


Fig. 4.14: Lateral infiltration of a shielded TBM caused by ex-cess excavation above the roof

In case of a heading with a TBM-S, the tunnel face can also besupported using glass fiber anchors. For this purpose, openings inthe cutterhead must be provided (see Section 5.5). When the drill-ings for the glass fiber anchors are carried out, there is a riskof a failure of the drill pipe. Since the recovery of the brokendrill pipe leads to impediments of the further heading, the in-stallation of glass fiber anchors in the tunnel face should belimited to exceptional cases only.

If in the scope of site investigations it was not realized thatapplying a TBM-S in open mode would lead to the above mentionedproblems or would require a temporary face support over long sec-tions, a considerable overrun of the planned construction time andcost may occur, which could have been avoided by selecting anothertype of TBM or another tunneling method.

4.1.4 Earth pressure balanced TBM-S, closed mode (EPB)

Applying an EPB TBM (TBM-S5) with closed mode is one possibilityto support or stabilize the tunnel face (Fig. 2.18). Using aTBM-S5, loosening of rock at the roof or at the tunnel contournear the cutterhead can be avoided to a large extent. On the otherhand, a driving with closed EPB mode is considerably more expen-sive than a heading with open mode.

- 150 -


In case of a widely stable tunnel face, the heading with a TBM-S5is often carried out with open mode or with only partly filled ex-traction chamber (transition mode). Conveying the excavated mate-rial with a screw conveyor, which is necessary for the closed EPBmode, the risk of tearing out rock blocks from the temporary face,associated with the problems described in Section 4.1.3, arises.This is because the conveyance with a screw conveyer requires acertain amount of material at the invert of the extraction chamberwhich is permanently rubbed against the face. This can be avoidedby using a belt conveyor when driving in the open mode, which iscommonly used for a TBM heading with open mode. The combination ofa TBM without face support and an EPB shield, however, due to thedifferences in construction, is always a compromise. In the caseof classical hard rock TBMs, the excavated material is scooped upby conveyor openings into individual chambers mounted at the cut-terhead. With the rotation of the cutterhead, the material istransported upwards and then falls down on a centrically arrangedbelt conveyor. The cutting wheel of a EPB shield, however, to alarge extent is open and constructed as a simple star or slice.

The evaluation of the required supporting pressure is of great im-portance and can be obtained with the aid of stability analyses inwhich the supporting pressure can be varied, such as in finiteelement analyses (see Section 3.5.1). For the evaluation of thecharacteristic parameters of the ground, which are necessary forsuch investigations, extensive site investigations and experienceare required. In Section 4.3, an example for the estimation of therequired supporting pressure is presented. The estimation of therequired supporting pressure using simplified analysis models (seeSection 3.5.2) may result in too large supporting pressures, whichamong other things would lead to higher cost as a result of ahigher wear of the cutting tools.

The stability analyses must also account for the influence of po-tentially present groundwater. In the case of TBM headings carriedout underneath the groundwater table with open mode or with EPBmode, when the water pressure cannot be completely balanced by theearth pressure, a seepage flow arises which is directed towardsthe tunnel and which leads to a drainage of the ground. This isassociated with a lowering of the groundwater table and with aloss of uplift in the area above the phreatic surface and belowthe undisturbed groundwater table (Fig. 4.15). In case of tunnelslocated in highly compressible ground, this results in additional

- 151 -


subsidence at the ground surface. Furthermore, the seepage forcesresulting from the seepage flow directed towards the tunnel mayaffect the stability of the unsupported area of the tunnel (Fig.4.15). The influence of seepage flow, therefore, must be consid-ered in the stability analyses (see Section 3.5.1).

Fig. 4.15: Influence of seepage on the stability of the tempo-rary face

- 152 -


One precondition for the application of a TBM-S5 with an EPB-modein unstable rock is the capability to transform the excavated rockinto an earth mud with soft to slurry consistency, high water ab-sorption capacity and low water permeability (see Sections 2.1.6and 5.3.4).

In order to provide the earth mud with an adequate and uniformconsistency, if necessary, conditioning agents are injected intothe extraction chamber and in front of the cutterhead. These con-ditioning agents cause a homogeneization and furthermore lead to areduction of the shear strength of the earth mud. Conditioningagents are also applied to avoid adhesion of the earth mud on thecutterhead and in the extraction chamber.

If the excavated intact rock is not completely transformed into anearth mud, extremely large torques of the cutterhead may becomenecessary to overcome the resistance, which results from the rota-tion of the cutter discs on the tunnel face and from the rotationof the cutterhead in the earth mud. If these torques cannot becarried, a rotation of the TBM on its shield may occur (see Sec-tion 5.4).

During a heading with closed mode, i. e. with earth pressure act-ing in the extraction chamber and on the tunnel face, the thrustforces must overcome the resistance resulting from the shield'sfriction, from the supporting pressure and from the contact pres-sure of the discs (see Section 5.3.3). If the friction force ofthe shield is assumed to be known, then the required thrust forcepredominantly is determined by the supporting pressure and by thecontact pressure of the discs. In practice, however, the separa-tion of the components "supporting pressure" and "contact pressureof the discs" has proven to be difficult. In particular in thecase of mixed face conditions, the number of discs which are indirect contact with the tunnel face are unknown. If for exampleonly one half of the discs are in contact with the temporary face,then the contact pressure force of an individual disc is twice ashigh as in the case that all discs are in contact with the face.This may lead to an exceeding of the critical contact pressure ofthe discs and thus to an increased wear (Fig. 4.16, see also Sec-tion 5.1.3).

- 153 -


Fig. 4.16: Overstressing of individual discs in the case ofmixed face conditions

If the above mentioned problems occur to a large extent and overlonger distances, the type of TBM or tunneling method selected wasnot adequate for the respective ground conditions.

4.1.5 Slurry TBM-S, closed mode

When carrying out mechanized tunneling in rock, the tunnel faceand the unsupported excavated cross-section, besides by the meth-ods already described, can also be supported with a slurry (Fig.2.14). The problems associated with the conditioning of intactrock are avoided, if a TBM-S with slurry supported face (TBM-S4)is used instead of a TBM-S5.

When applying shield machines with slurry supported face in soil,the slurry, as a rule a suspension of water and bentonite, pene-trates into the grain skeleton and forms a filter cake. The filtercake seals the tunnel face and forms a membrane via which theslurry pressure can be transferred to the tunnel face. In the caseof an extremely coarse-grained soil, no filter cake can be formedanymore and as a consequence no supporting pressure on the tunnel

- 154 -


face can be achieved. The applicability of a slurry support inrock is limited by the discontinuity fabric and by the aperturesof the discontinuities.

Shield machines with slurry supported face predominantly are usedin sands and fine gravels (see Fig. 2.17). When applying such ma-chines in clayey soils, there is a risk of plugging the conveyoropenings and of adhesion at the cutting wheel and in the extrac-tion chamber. Furthermore, with increasing fraction of clay, theeffort for separation of the bentonite-soil mix also increases(see Section 2.1.5).

In case of a heading in mudstone or siltstone, the effort forseparation can also be extensive, because these rocks, at leastpartially, can be decomposed into particles of clay and silt dur-ing excavation and conveyance (see Section 5.3.4). In such rocks,the efficiency of the application of a hydro shield must be accu-rately checked, since the costs which are associated with theseparation are not negligible.

4.1.6 Squeezing rock mass

In case of a tunnel with high overburden and low strength of theintact rock, not only the strength on the discontinuities but alsothe strength of the intact rock may be exceeded. In such case, weare talking about "squeezing rock". For a tunnel in squeezing rockat great depths, the loads which are acting on the lining normallyare too large for an economic design of a stiff lining. Tunnels insqueezing rock, therefore, are planned with a yielding support.The latter allows for a limited radial displacement of the tunnelcontour, which leads to a redistribution of stresses in the rockmass and thus to a considerable lower loading of the support.

For tunnels in squeezing rock with large diameters, sometimes ra-dial displacements of several decimeters are required to enable aneconomic support.

A machine-driven tunnel with a yielding support designed for ra-dial displacements with the aforementioned order of magnitude, asfar as known to the authors, was not carried out up to now. A po-tential solution is represented in Fig. 4.17.

- 155 -


Fig. 4.17: Mechanized tunneling in squeezing rock

In the annular gap between the tunnel contour and the segmentallining, a yielding material such as expanded clay is installedwhich allows for the required displacement of the tunnel contour

- 156 -


at reasonable radial stresses. For a yielding support in heavilysqueezing rock, a radial displacement of for example approx. 1 mmay be required. The layer of yielding material must be carriedout with a correspondingly larger thickness. Taking into accountthe deformability of an expanded clay, for example, a thickness ofapprox. 2 m would be necessary for that layer (Fig. 4.17, see alsoSection 9.5).

The concept described above, however, requires a number of new de-velopments. An adaptation of the TBM to different ground condi-tions and corresponding radial displacements would be desirable.

In such a concept, a potential deadlock of the TBM is particularlycritical. In this case, the steering gap between tunnel contourand shield skin would close due to radial displacements and theshield skin would be loaded by almost the entire rock mass pres-sure. On one hand, the shield skin would have to be designed forthis pressure. On the other hand, by means of additional designfeatures, it would have to be assured that the friction forcesacting along the shield skin which result from the radial loadingof the TBM can be overcome by the thrust forces.

Currently, a mechanized tunneling in heavily squeezing rock mustbe regarded as a great risk which can hardly be assessed duringthe design phase.

4.1.7 Conclusions

The problems mentioned above, which may arise in connection withmechanized tunneling in rock, demonstrate that the choice of a TBMwhich is suitable for the geotechnical and operational boundaryconditions, is essential for the success of the heading. If therange of application of the TBM, which is given by the design fea-tures, is not in agreement with the existing ground conditions, anoverrun of construction time and cost is usually unavoidable.

A remodelling of the TBM during construction requires a greatamount of time and cost and can only be regarded as an emergencysolution. For machine tunneling, therefore, an extensive explora-tion of the ground is even more important as for conventional tun-neling.

- 157 -


The evaluation of the load bearing behavior of the ground and ofthe effects resulting from the interaction of TBM and rock mass isessential for the selection of a suitable type of TBM. In thisconnection, an expert who is familiar with the ground under con-sideration should be involved, and tunnel headings carried out inthe same or in similar rock formations should be evaluated. Thegeotechnical parameters, which are necessary for the evaluation ofthe load bearing behavior, must be determined using suitable ex-ploration methods and tests. In each individual case also influ-ences due to seepage flow as well as risks resulting from regionalgeology and hydrogeology have to be assessed by ground experts andplanners in an early stage of the project.

In the following, it will be demonstrated by means of two exampleshow the stability in the machine area can be proven in the case ofstable and unstable rock mass conditions.

4.2 Stability proof for a TBM heading in a stable rock mass

4.2.1 General

As discussed above, a vital question regarding the selection ofthe type of TBM in case of a mechanized tunneling in rock is theassessment of the stability of the tunnel in the area of the ma-chine and of the shield respectively. It must be investigated ifand to which extent a support of the tunnel contour and of thetunnel face is necessary.

A mechanized tunneling in rock in principle can be carried outwith a Gripper TBM and a shielded TBM (TBM-S). The possibilitiesof support in the machine area of these types of TBM are describedin Sections 2.1.2, 2.1.3, 4.1.2 and 4.1.3.

According to the authors' opinion, stability analyses according tothe FEM are mandatory for the choice of an adequate type of TBMand for the estimation of the support measures required, respec-tively. Such analyses have to be based on an appropriate rock me-chanics model and on reliable parameters (see Section 3.5.1). Inmany cases, it is very difficult to gather this information, be-cause the tunnel may be located at great depth and the potentialsof exploration are limited. However, the corresponding data areindispensable for reduction of the risks of a machine tunneling.For evaluation of models and parameters, it may often be helpful

- 158 -


to use experiences gained from tunnel projects in the same orsimilar ground conditions, for calibration of the models.

In the following, a stability proof for a TBM heading in stablerock mass and the selection of an adequate type of TBM based uponit, is demonstrated by means of an example. A single-tracked rail-road tunnel with an excavation diameter of 10.1 m is considered.The entire tunnel cross-section is assumed to be located withinthe unleached Gypsum Keuper (Fig. 4.18). The maximum overburden isassumed to be approx. 225 m.

Fig. 4.18: Tunnel located within unleached Gypsum Keuper, exam-ple of analysis

- 159 -


In the past, experience has been gained from tunnels, which havebeen excavated according to the conventional method and are lo-cated in the unleached Gypsum Keuper, especially in the area ofthe city of Stuttgart, Germany. In particular, it is referred toconstruction lots 12 and 13 (turning loop and Hasenberg Tunnel),which were completed during construction of the urban railway inStuttgart (see Wittke and Rißler, 1976; Wittke, 1978; Wittke,1979; Wittke and Pierau, 1979; Beiche et al., 1995; Wittke, 2004).

This experience is valuable with regard to the assessment of a TBMheading of a tunnel in unleached Gypsum Keuper. Fig. 4.19 shows aphotograph of the temporary face taken during construction of theHasenberg Tunnel in unleached Gypsum Keuper.

Fig. 4.19: Unleached Gypsum Keuper (Mittlerer Gipshorizont),temporary face in the Hasenberg Tunnel

4.2.2 Rock mechanics model of the unleached Gypsum Keuper (Mit-tlerer Gipshorizont)

The "Mittlerer Gipshorizont" forms part of the unleached GypsumKeuper formation. Further stratigraphic members of this formationare the "Bleiglanzschichten", the "Dunkelrote Mergel", the"Berchinger Horizont" and the "Grundgipsschichten".

The unleached Gypsum Keuper is composed of thin to massive layersof sulfate rock which alternate with siltstone layers. Sulfate

- 160 -


rock also appears in the form of nodules. The unleached GypsumKeuper contains horizontal bedding parallel discontinuities andvertical joints, which are widely closed or healed with fibrousgypsum, respectively. Besides these discontinuities, also sporadi-cally slickensides occur with angles of dip of approx. 45° andrandomly distributed angles of strike. Underneath the anhydritelevel, the sulfate rock predominantly consists of anhydrite. Abovethis level, the anhydrite as a consequence of the access of waterwas transformed into gypsum. Above the gypsum containing rock, theleached Gypsum Keuper is located. This is a residual rock, inwhich the gypsum was dissolved and removed over geological periodsof time. The boundary between the leached and the unleached GypsumKeuper is referred to as leaching horizon or gypsum level, respec-tively.

Fig. 4.20 shows a structural model of the "Mittlerer Gipshorizont"as part of the unleached Gypsum Keuper. In Fig. 4.21, drill coresfrom the same formation are depicted. The approximately horizontalbedding as well as bedding parallel discontinuities healed withfibrous gypsum are visible.

Fig. 4.20: Structural model of the "Mittlerer Gipshorizont" aspart of the unleached Gypsum Keuper

- 161 -


Fig. 4.21: "Mittlerer Gipshorizont", drill cores

According to experience gained from Stuttgart tunnel projects,the intact rock strength of the unleached Gypsum Keuper ranges ap-prox. from 5 to 40 MN/m². Locally, maximum values of up to130 MN/m² occur. The modulus of deformation ranges from 2000 to

- 162 -


8000 MN/m². The permeability is very low. In numerous permeabilitytests, kf values from 10-10 to 10-8 m/s were determined. In the areaof the leaching horizon, the coefficient of permeability increasesto values of 10-6 to 10-4 m/s.

In case of access of water, the unleached Gypsum Keuper can de-velop swelling strains or, if these are inhibited, swelling pres-sures (see e. g. Wittke, 2000). If the tunnel is driven below andwith a considerable distance to the gypsum level and to the anhy-drite level, water access is hardly to be expected, since the per-meability of unleached Gypsum Keuper is very low, as mentionedabove. If the access of operative waters or similar during con-struction is also prevented, swelling deformation during construc-tion can practically be avoided in such cases.

4.2.3 Calibration of the rock mechanics model

For calibration of the rock mechanics model for the unleached Gyp-sum Keuper, the experiences gained during the construction of theturning loop for the urban railway in Stuttgart are suitable.

The approx. 1500 m long underground turning loop (Fig. 4.22 and4.23) belongs to the tunnels for the urban railway in Stuttgart,which were excavated between the main station and the suburbStuttgart-Vaihingen until 1985. With a radius of 190 m it forms anunderground loop west of the station Schwabstraße, which serves asa reverser for half of the trains coming from the main station.

The first 1100 m of the turning loop were carried out single-tracked with a diameter of the tunnel tube of approx. 8 m. An ap-prox. 300 m long double-tracked section of the turning loop with adiameter of the tunnel tube of approx. 12 m follows (Fig. 4.22).

The height of the overburden of the turning loop varies between22 m and the maximum of 87 m (Fig. 4.23).

The underground turning loop is completely located in the "Mit-tlerer Gipshorizont". The gypsum level cuts the tunnel at chain-ages 0+535 and 1+312 so that the front part of the turning loop islocated in the leached Gypsum Keuper and the rear part in the un-leached Gypsum Keuper. About 150 m of the double-tracked sectionare passing through the unleached Gypsum Keuper (Fig. 4.23).

- 163 -


Fig. 4.22: Urban railway Stuttgart, turning loop, site plan

Fig. 4.23: Urban railway Stuttgart, turning loop, geometry andgeology

- 164 -


During construction of the turning loop, the tunnel proved to bestable in the unleached Gypsum Keuper with support measures notworth mentioning (Fig. 4.24). For the tunnel with an overburdenheight of 75 m and an excavation diameter of 12.1 m, only a shot-crete sealing with a thickness of only 3 to 5 cm was applied.

Fig. 4.24: Urban railway Stuttgart, turning loop, lot 12, un-leached Gypsum Keuper

Until the installation of the internal lining, which lasted untilmax. one year after the excavation of the tunnel, the tunnel wasstable with this sealing.

The excavation of the tunnel was carried out by means of blasting.To avoid loosening of the rock mass and access of moisture to thearea near the tunnel contour, a 0.5 to 1.0 m thick rock zonearound the tunnel contour was not blasted but removed with a roadheader in a separate working step.

With the aid of the finite element analyses described below, thoserock mechanics parameters of the unleached Gypsum Keuper are to befound, for which the unsupported tunnel is on the transition fromstability to instability.

- 165 -


FE-mesh

The two-dimensional analyses were carried out with the aid of thecomputation section marked in Fig. 4.22 and 4.23. The overburdenheight of the double-tracked tunnel tube located in the unleachedGypsum Keuper is assumed to be 75 m. The gypsum level is assumedto be situated approx. 10 m above the tunnel's roof (Fig. 4.25).

Fig. 4.25: FE-mesh, boundary conditions and ground profile

- 166 -


A 56 m wide, 107 m high and 1 m thick slice of the ground is se-lected as computation section (Fig. 4.25). The vertical planethrough the tunnel axis represents a plane of symmetry, because ofthe circular cross-section of the tunnel, the horizontal beddingand the vertical jointing which is assumed to strike in parallelwith and perpendicularly respectively to the tunnel axis. There-fore, only one half of the tunnel tube with an excavation diameterof 12.1 m is modeled.

The analyses are carried out with two computation steps. In thefirst step, the stresses and deformations resulting from the self-weight of the ground are calculated (primary state). In the secondstep, the excavation of the unsupported tunnel is simulated, as-suming an elastic-viscoplastic stress-strain behavior (see Section3.3). The analyses are carried out using the program system FEST03(Wittke, 2000).

Parameters and analysis results

The selected rock mechanics parameters are summarized in Table4.1. The essential parameters with regard to the stability of thetunnel are the strengths on the discontinuities. The joints areassumed to dip vertically and to strike in parallel with and per-pendicularly to the tunnel axis. The bedding parallel discontinui-ties are assumed to be horizontal. The slickensides are neglectedbecause of their small extent and low frequency of occurrence.

The angle of friction of the discontinuities can be assumed tobe 30° for the bedding parallel discontinuities and 35° for thejoints. The cohesion on the discontinuities is varied in theanalyses between cB = 35 kN/m² and cB = 57.5 kN/m² for thebedding parallel discontinuities and between cJ = 50 kN/m² andcJ = 500 kN/m² for the joints. The intact rock is assumed elasticin the analyses, which is reasonable in view of the existing over-burden.

The parameters of the leached Gypsum Keuper and of the Schilfsand-stone are also given in Table 4.1, but not further discussed here,since the analysis results are scarcely influenced by these pa-rameters.

- 167 -


parametersunleached

Gypsum Keuperleached

Gypsum KeuperSchilfsand-

stone

unit weight � [kN/m³] 27 22 25

modulus ofdeformation

E [MN/m²] 4000 150 2000

Poisson'sratio

� 0.25 0.35 0.25

angle offriction

��

bedding:30

joints:35

30

cohesion c [kN/m²]

bedding:35 - 57.5joints:50 - 500

0

tensilestrength

t [kN/m²] 0 0

elastic

Table 4.1: Rock mechanics parameters

The criterion for stability of the unsupported tunnel is the con-vergency of the displacements of the tunnel contour in the courseof the elastic-viscoplastic iterative analysis. For each investi-gated combination of cB and cJ, this criterion was checked. The re-sult (stable, plastic limit state, not stable) is marked in Fig.4.26 using different signatures.

With cJ = 50 kN/m² and cB = 40 kN/m², for example, the stability ofthe tunnel cannot be proven (Fig. 4.26).

With values of cB ranging from 42.5 kN/m² to 55 kN/m² andcJ = 50 kN/m² the tunnel contour is on the transition from stableto unstable. Convergency is reached, however, the time-displacement lines computed for these combinations of parametersindicate that a state of equilibrium is obtained only after alarge number of iterations. Also for other combinations of parame-ters with cJ > 50 kN/m² a corresponding behavior is observed. Thesecombinations of parameters are included in the shaded area in Fig.4.26 which is referred to as "plastic limit area".

- 168 -


Fig. 4.26: Required cohesion on the bedding parallel disconti-nuities as a function of the cohesion on the joints

The drawn through, light-red line plotted in Fig. 4.26 character-izes the transition from unstable to stable conditions. For pa-rameter combinations located above this line, i. e. for cB 40 kN/m² and cJ 100 kN/m², stability of the tunnel can be proven.

4.2.4 Stability proof by means of two- and three-dimensionalfinite element analyses

Two-dimensional finite element analyses

As an example, a machine-driven single-tracked railroad tunnel lo-cated in the "Mittlerer Gipshorizont" with an overburden height of225 m is considered (Fig. 4.18). The stability of the unsupportedtunnel contour is at first considered in a simplified manner withthe aid of two-dimensional analyses using the program system

- 169 -


FEST03 (Wittke, 2000). The tunnel contour is assumed to be unsup-ported.

The computation section, the FE-mesh, the boundary conditions andthe rock mechanics parameters are represented in Fig. 4.27. Thelatter are selected on the basis of the analysis results for thecalibration of the rock mechanics model. The vertical planethrough the tunnel axis again represents a plane of symmetry.Therefore, only one half of the tunnel's cross-section is modeledin the FE-mesh.

The analyses are carried out with two computation steps. In thefirst step, the stresses and deformations resulting from the self-weight of the ground are calculated (primary state). In the secondstep the excavation of the tunnel is simulated.

The strengths on the discontinuities, which are decisive for thestability, as mentioned are given in Fig. 4.27. The angles offriction on the bedding parallel discontinuities (�B = 30°) and onthe joints (�J = 35°) are assumed to be constant. The values forcohesion cB and cJ are varied within the ranges given in Fig. 4.27.The bedding parallel discontinuities are assumed to be horizontaland the joints are assumed to dip vertically and strike in paral-lel with and perpendicularly to the tunnel axis.

Fig. 4.28 shows the principal normal stresses and the areas inwhich the strength is exceeded, resulting for the analysis withcB = 40 kN/m² and cJ = 100 kN/m². This parameter combination is lo-cated on the limit of stability line derived from the results ofthe calibration analyses (see Fig. 4.26). As a result, thestrength on the bedding parallel discontinuities is exceeded aboveand underneath the tunnel, while next to the sidewalls thestrength on the joints is exceeded. At the transitions from theroof to the sidewalls and from the invert to the sidewalls, re-spectively, both, the shear strength on the bedding parallel dis-continuities and the shear strength on the joints are exceeded.

- 170 -


Fig. 4.27: Stability of the unsupported tunnel contour in themachine area, FE-mesh, boundary conditions and rockmechanics parameters

- 171 -


Fig. 4.28: Principal normal stresses and areas in whichthe strength on the discontinuities is exceeded(cB = 40 kN/m², cJ = 100 kN/m²)

The displacements, however, converge in the course of the visco-plastic iterative analysis so that the stability of the unsup-ported tunnel contour can be proven calculationally.

Assuming cB = 0 to 30 kN/m², no convergency of the tunnel's roofsubsidence can be achieved, even though a comparatively large co-hesion on the joints of cJ = 500 kN/m² is assumed. With values forthe cohesion of cB = 35 kN/m² and cJ = 500 kN/m², however, the sub-

- 172 -


sidence of the tunnel's roof converges and the stability of theunsupported tunnel contour can be proven (Fig. 4.29).

Fig. 4.29: Subsidence of the tunnel's roof in the course of theviscoplastic iterative analysis as a function of thecohesion on the bedding parallel discontinuities

Fig. 4.30 shows the cohesion on the bedding parallel discontinui-ties required for the stability of the unsupported tunnel contouras a function of the cohesion on the joints. According to this,cohesion values of cB = 50 kN/m² and cJ = 50 kN/m² are necessaryfor stability. In case of a higher cJ value, the tunnel is stablealso with cB = 35 kN/m². In Fig. 4.30, the line which characterizesthe limit of stability for the turning loop is also plotted forcomparison. This line to a large extent lies above the line whichcharacterizes the limit of stability for the investigated example.This means that the strengths on the discontinuities which arenecessary to stabilize this tunnel are smaller than the corre-sponding strength parameters which result from the calibrationanalyses carried out for the turning loop. This is because of thesmaller diameter of the tunnel compared with the double-trackedtunnel tube of the turning loop. As a consequence, the contour ofthe tunnel investigated should be stable without support.

- 173 -


Fig. 4.30: Stability of the unsupported tunnel contour as afunction of the cohesion on the discontinuities

Three-dimensional finite element analyses

From the results of a two-dimensional analysis no information canbe gained regarding the stability of the temporary face. There-fore, additional three-dimensional analyses were carried out usingthe program system FEST03 (Wittke, 2000).

The computation section, the FE-mesh, the boundary conditions andthe rock mechanics parameters are given in Fig. 4.31. The verticalplane through the tunnel axis represents a plane of symmetry.Therefore, only one half of the tunnel's cross-section is modeled.The height of overburden and the rock mechanics parameters onwhich the analyses are based correspond to those of the two-dimensional analyses.

- 174 -


Fig. 4.31: Stability of the unsupported opening, FE-mesh,boundary conditions and rock mechanics parameters

- 175 -


The analyses are carried out with two computation steps. In thefirst step, the stresses and deformations resulting from the self-weight of the ground are calculated (primary state).

Fig. 4.32: Displacements on the tunnel contour:a) cB = 40 kN/m²/cJ = 100 kN/m²b) cB = 200 kN/m²/cJ = 600 kN/m²

- 176 -


In the second step, the excavation of the tunnel is simulated.Since the application of a TBM-S in a stable rock mass is expectedto be feasible, the temporary face and the tunnel contour in thearea of the cutterhead and of the shield skin, which extends overa length of 11 m, remain unsupported. The installation of the seg-mental lining is simulated in a distance of 11 m from the tempo-rary face.

With cohesions of cB = 40 kN/m² and cJ = 100 kN/m², the stabilityof the unsupported opening in the area of the temporary face andof the shield can be proven calculationally. At the temporaryface, horizontal displacements of approx. 3 mm and maximum roofsubsidence of approx. 18 mm are calculated (Fig. 4.32a).

The aforementioned displacements and the corresponding exceedingof strengths on the bedding parallel discontinuities and on thejoints to a certain amount may also lead to loosening in the areaof the tunnel's roof. Therefore, also analyses with other valuesof cB and cJ were carried out for the purpose of comparison.

In the following, the analysis results for the case describedabove with cB = 40 kN/m² and cJ = 100 kN/m² are compared with theresults for a case with increased cohesions of cB = 200 kN/m² andcJ = 600 kN/m².

As shown in Fig. 4.32, the maximum subsidence of the tunnel's roofdrops from 18 mm to 12 mm in the case with the higher cohesions.

In Fig. 4.33, the areas with exceeding of strength on the discon-tinuities, which will be associated with certain loosening of therock mass, are represented for both cases in two vertical sectionsparallel and perpendicular to the tunnel axis. With increasing co-hesion these areas become smaller.

In conclusion it can be noticed that, with increasing values forthe cohesion on the discontinuities, both, the displacements andthe areas with exceeded strength (disturbed rock zones) are re-duced, as it is shown with the aid of the three-dimensional finiteelement analyses.

- 177 -


Fig. 4.33: Areas in which the strength on the discontinuitiesis exceeded

4.2.5 Selection of driving method and type of TBM

In case the tunnel under consideration in the unleached GypsumKeuper is driven with a TBM, it can be assumed on the basis of thefinite element analysis results (see section 4.2.4) that the tun-nel contour and the temporary face are stable. Since a support ofthe temporary face is not required, either a Gripper TBM or ashielded TBM may be used, as already mentioned in Section 4.2.1.

If a Gripper TBM is applied, the bracing of the grippers may leadto tensile stresses in the rock mass at the roof and at the invertof the tunnel. As a result, frequently cracks are generated whichcan lead to a loosening of rock wedges at the tunnel's roof. It isshown in Section 5.4.1 for a tunnel bored in the unleached GypsumKeuper that, in case the gripper forces are taken into account,

- 178 -


the tunnel is only stable if a systematic anchoring is foreseen.Moreover, the disturbed rock zones around the tunnel contour,which result from the gripper forces, lead to an increased perme-ability of the rock mass. This should be avoided with regard tothe problems, which are associated with swelling rock. Therefore,the application of a Gripper TBM in the unleaded Gypsum Keuperdoes not seem to be appropriate.

In case of the tunnel under consideration, a shielded TBM withoutface support (TBM-S) could be applied. The possibility to installa light advance anchoring (spiles) in the area of the roof shouldbe foreseen for potential, locally appearing weak zones in therock mass. Generally, the application of anchors and spiles is,however, not necessary for statical reasons.

4.3 Stability proof for a TBM heading in unstable rock massabove the groundwater table

4.3.1 General

With the following example, the stability of the temporary faceand of the tunnel contour in the area of the cutterhead is inves-tigated for a TBM heading in an unstable rock mass.

A tunnel with an excavation diameter of 9.4 m and a maximum over-burden height of HO = 75 m is considered (Fig. 4.34). The tunnel islocated in the leached Gypsum Keuper. Similar to the unleachedGypsum Keuper, also for this type of rock mass, which is encoun-tered in the area of Stuttgart, experiences with NATM tunnelingare available (Grüter and Liening 1976; Wittke and Rißler, 1976).The gypsum level, which in Fig. 4.34 is designated with "y" andforms the boundary between the leached and the unleached GypsumKeuper, is assumed to be located 10 m below the invert of the tun-nel. The groundwater table is also assumed to be situated belowthe tunnel's invert.

The tunnels which are located within the leached Gypsum Keuper andwhich were excavated according to the conventional method had tobe supported with a reinforced shotcrete membrane and rock boltsafter each round. Furthermore, the cross-section had to be subdi-vided in parts (crown, bench and invert) and an immediate supportof the temporary face was required. As an advance support, spileswere installed, when necessary.

- 179 -


Fig. 4.34: Tunnel located in the leached Gypsum Keuper, examplefor analysis

- 180 -


With these support measures, which correspond to the excavationclasses 5 to 7 (see Section 4.1.1), the temporary face resulted tobe stable.

Up to now, no experience is available with regard to machine tun-neling in the leached Gypsum Keuper. It is therefore to be inves-tigated, if and to what extent a support of the temporary face maybe necessary when a TBM heading is carried out. When applying aTBM-S in open mode, the tunnel contour and the temporary face to alarge extent should be stable without support. If this is not thecase, a TBM-S with slurry or earth pressure support (TBM-S4 orTBM-S5) should be taken into account.

The evaluation of the stability of the rock mass in the area ofthe temporary face in case of a mechanized tunneling in theleached Gypsum Keuper, and on this basis the selection of a suit-able TBM, is carried out in the following with the aid of three-dimensional finite element analyses. These are based on a rock me-chanics model which was elaborated in the scope of work for com-pleted structures located in the leached Gypsum Keuper in theStuttgart area (urban railway station Schwabstraße, tunnels forthe turning loop and Hasenberg Tunnel).

4.3.2 Rock mechanics model of the leached Gypsum Keuper

The leached Gypsum Keuper, as mentioned above, is a residual rockwhich was transformed from undisturbed original Gypsum Keuper bydissolution of sulfate. The unleached Gypsum Keuper above the Gyp-sum level has been completely alternated due to the influence ofthe groundwater. The sulfate rock contained in the unleached Gyp-sum Keuper was dissolved and removed with the groundwater, whilethe clay and silt fraction of the rock were decomposed due toleaching, however, remained in place.

According to the structural model represented in Fig. 4.35, therock mass consists of an alternating sequence of residual silts,weak siltstones and sound siltstone benches, corresponding to thevarying distribution of sulfate rock and to the varying sulfatecontent of the individual layers in the original condition. Therock can be subdivided in four classes with respect to its degreeof leaching and its strength (Fig. 4.35):

- 181 -


Class I: Sound siltstone.

Class II: Siltstone with moderate to low strength, mostly ap-pearing in thin layers, closely jointed, crumblydisintegrating.

Class III: Weak siltstone, intensely jointed, alternating withthin layers of silt.

Class IV: Residual silt.

Fig. 4.35: Structural model of the leached Gypsum Keuper

The water content is an essential criterion for the allocation ofthe rock to the classes mentioned above. The water content in ten-dency increases from the sound siltstone (class I) to the residualsilt (class IV).

- 182 -


Fig. 4.36: Leached Gypsum Keuper, drilled cores

- 183 -


Fig. 4.37: Leached Gypsum Keuper, temporary face in the Hasen-berg Tunnel

- 184 -


The following ranges of the mean water content wm were determinedon a great number of rock samples taken from the Stuttgart areafor different tunneling projects:

Class I: wm = 0.05 – 0.06

Class II: wm = 0.08 – 0.09

Class III: wm = 0.13 – 0.17

Class IV: wm = 0.18 – 0.23

The siltstones which are not yet completely decomposed, are inter-sected by steeply dipping joints J with a spacing of approx. onedecimeter or less. These joints mostly end at the approximatelyhorizontal bedding parallel discontinuities B (Fig. 4.35). Thelatter frequently are filled with fine-sandy silts.

Immediately above the Gypsum level, the rock mass is stronglyloosened and in many cases softened due to seepage water. At theboundary between the leached and unleached Gypsum Keuper, smallholes and cavities respectively and locally also larger cavitiesin an order of magnitude of a cubicmeter may appear (Fig. 4.35).These may be filled with leached gypsum Keuper from above.

Fig. 4.36 shows a photograph of drilled cores from the leachedGypsum Keuper. These can predominantly be allocated to classes IIIand IV, apparently are weak and consist to some extent of silt.

Fig. 4.37 shows a section of the temporary face in the HasenbergTunnel. The intact rocks which are assigned to classes I, III andIV as well as gypsum residues labeled with "y" are identified.

The classes II to IV, which consist of weak siltstone to residualsilt, can be summarized with regard to the rock mechanics parame-ters describing the deformability and strength of the leached Gyp-sum Keuper, which are decisive for the stability of the tunnel.These parameters are compiled in Table 4.2.

- 185 -


parameters class I class II - IV

unit weight � [kN/m³] 24 - 26 21 - 23

modulus ofdeformability

E [MN/m²] 300 - 100080 - 200

(40 - 802))

poisson's ratio � 0.30 0.35

unconfined com-pressive strengthof the intact rock

C [MN/m²] 0.3 - 5 (151)) -

angle of friction ��

bedding parallel

discontinuities:

25

joints:

30

30

cohesion c [kN/m²]

bedding parallel

discontinuities:

0 - 50

joints:

20 - 100

0 - 40

1) locally appearing peak values2) in case of high fractions of class IV

Table 4.2: Rock mechanics parameters of the leached GypsumKeuper

4.3.3 Assessment of the stability by means of three-dimensionalfinite element analyses

The selected three-dimensional computation section, the FE-meshand the boundary conditions for the tunnel represented in Fig.4.34 are given in Fig. 4.38.

A 40 m wide, approx. 80 m high and 103 m long section of theground is selected for the analyses. As for the analyses carriedout for the unleached Gypsum Keuper (see Section 4.2.4), only onehalf of the tunnel's cross-section is modeled, because also forthis example the vertical plane through the tunnel axis representsa plane of symmetry. The upper boundary of the computation sectionis loaded by a surface load corresponding to the weight of therock mass above the computation section. The total overburden ofthe tunnel as before is assumed to amount to 75 m.

- 186 -


Fig. 4.39 shows a detail representing the machine area and thetemporary face in a vertical section through the tunnel axis.Since in these analyses primarily the stability of the temporaryface is to be investigated, it is assumed for simplification thatthe ground is abutting directly on the shield. Thus, the tunnelcontour in this area is supported by the shield skin which is mod-eled with a thickness of 160 mm (Fig. 4.39).

Fig. 4.38: Stability of the area at the temporary face, compu-tation section, FE-mesh, boundary conditions, groundprofile and parameters

- 187 -


In order to support the temporary face and the tunnel contour inthe area of the cutterhead, a calculative support pressure pS isapplied. The latter can be produced by means of an earth or slurrypressure or by a mechanical support with the cutterhead or thediscs, respectively. Simplifying, the support pressure is simu-lated by means of a constant surface load in the analyses (Fig.4.39).

Fig. 4.39: Leached Gypsum Keuper, FE-mesh (detail). Assumptionfor supporting pressure ps

The analyses were carried out with the program system FEST03 (Wit-tke, 2000). In the first computation step, the stresses and dis-placements resulting from the self-weight of the rock mass arecalculated (primary state). In the second step, the excavation ofthe tunnel is simulated. The modulus of deformation E and thestrength on the discontinuities of the leached Gypsum Keuper as-sumed in the analyses, are selected according to the parametersderived for classes II to IV (see Table 4.2).

- 188 -


Three cases are investigated, in which the supporting pressure isvaried between 0 and 100 kN/m².

The calculated displacements at the face and at the roof duringthe second computation step are represented in Fig. 4.40 for thecases investigated. In the case without consideration of a sup-porting pressure (pS = 0), no convergency of the displacements isachieved after approx. 5000 iterations. Thus, the stability of thetemporary face cannot be proven in this case. When applying a sup-porting pressure of pS = 40 kN/m², a convergency of the viscoplas-tic displacements is reached (Fig. 4.40). The resulting displace-ments amount to approx. 14 cm at the temporary face. By means of asupporting pressure of 100 kN/m², the displacements at the tempo-rary face are reduced to approx. 8 cm.

Fig. 4.40: Displacements at the face and the roof of the tunnelin the course of the viscoplastic iterative analysisas a function of the mean supporting pressure ps

4.3.4 Selection of type of TBM

The three-dimensional stability analyses lead to the result thatfor a machine-driven tunnel with an overburden height of 75 m lo-cated in the leached Gypsum Keuper of classes II to IV, a support-

- 189 -


ing pressure of only 40 to 100 kN/m² is required to stabilize thetemporary face. In case of a lower overburden or in case of aleached Gypsum Keuper with a large position of class I, the tempo-rary face also should be stable without a supporting pressure.

Based on the soil mechanics properties of the leached Gypsum Keu-per of classes I to IV, such as grain size distribution, plastic-ity and consistency index, a support of the temporary face using aslurry or EPB shield should be possible. When applying a slurryshield, however, a considerable effort for separating the bento-nite-soil mix may be required, because the leached Gypsum Keupercontains a high amount of fines. When applying an EPB shield, itmust be investigated if the required conditioning of the leachedGypsum Keuper is feasible (see Section 5.3.4).

The application of a TBM-S with open mode can also be considered,since the required pressure to stabilize the temporary face iscomparatively small. The required supporting pressure approxi-mately corresponds to the pressing forces of the discs during aTBM heading. The supporting pressure induced by the discs, how-ever, would have to be maintained also during interruptions of theheading. Moreover, in the area of the cutterhead there is locallythe risk of outbreaks at the roof and the face leading to a lackof support by the cutterhead (see Section 4.1.3).

The selection of the type of TBM, which should be applied for theexample under consideration, is essentially dependent on the ex-pected ground conditions, i. e. the distribution of classes I toIV of the leached Gypsum Keuper. The application of a TBM-S withopen mode is advantageous when only short tunnel sections are lo-cated in classes II to IV. If the leached Gypsum Keuper of classesII to IV is encountered over long tunnel sections, the use of aconvertible TBM, which can be driven with open mode as well aswith closed mode, may have advantages.

The example demonstrates that for a mechanized tunneling, particu-larly if to be carried out in unstable ground conditions, site in-vestigation is considerably more important than for a more adapt-able tunneling method such as the NATM, since the type of the TBMhas to be selected in the stage of planning and practically cannotbe changed after construction of the machine has been started.

- 190 -


4.4 Mechanized tunneling in the presence of high water pres-sure, lowering of the groundwater table and inflow ofseepage water

4.4.1 General

Tunnel headings in rock with high overburden and high groundwatertable normally lead to a lowering of the groundwater table, atleast during construction, and thus to a drainage. This is associ-ated with a seepage flow directed towards the unsupported area ofthe tunnel. This seepage flow has an unfavorable effect on thestability of the tunnel and in addition leads to an inflow of wa-ter, the quantity of which is dependent on the permeability of therock mass and may lead to an impediment of the heading.

On the other hand, requirements demanded from the authorities withregard to admissible lowering of the groundwater table and to themaximum permitted inflow of seepage water into the tunnel must beobserved. In such cases, it is necessary to evaluate the seepagewater quantities, the drawdown curves as well as the seepageforces acting on the ground and resulting from the seepage flow.On this basis, measures to stabilize the temporary face and to re-duce the seepage water quantities can be planned.

In case of conventional tunneling as well as of mechanized tunnel-ing there is basically the possibility of sealing the rock masssurrounding the tunnel by means of advancing grouting. This, how-ever, requires the choice of a suitable grouting technique andgrouting agent, respectively, which is dependent on the groundconditions (see e. g. Kutzner, 1991; Wittke and Breder, 1985).

In case of mechanized tunneling also TBMs with slurry mode(TBM-S4) or EPB mode (TBM-S5) can be used. When applying a TBM-S4,besides the support of the temporary face it can be achieved toavoid the inflow of seepage water to a large extent by adaptingthe slurry pressure to the water pressure. Since the supportingpressure, which can be applied, is limited due to technical andeconomical reasons, a lowering of the groundwater table and theinflow of seepage water into the machine area cannot be avoided inany case by the supporting pressure alone.

With the aid of the example of a mechanized tunneling in a rockmass with a high groundwater table it will be demonstrated in the

- 191 -


following how seepage water quantities, drawdown and seepageforces can be evaluated by means of corresponding analyses. Fur-thermore, it will be shown how the inflow of seepage water intothe machine area can be influenced and limited, respectively, bymeans of measures which reduce the permeability of the ground.

4.4.2 Three-dimensional seepage flow analyses

A tunnel with an overburden height of 150 m is considered. The un-disturbed groundwater table is assumed to be located at the groundsurface. A heading with a shielded TBM is planned. The boring di-ameter of the TBM is assumed to be 10.6 m, the length of theshield is supposed to be 12 m (Fig. 4.41).

Fig. 4.41: Mechanized tunneling in the presence of high waterpressure

The applied TBM is convertible from open mode to closed slurrymode. The slurry pressure, which can be applied for the support ofthe temporary face, is assumed to be limited to a maximum value of8 bar (Fig. 4.41).

- 192 -


Since the maximum slurry pressure is smaller than the water pres-sure resulting from 150 m water column, a seepage flow directedtowards the temporary face and the shield skin results (see Fig.4.15). In the following, the results of three-dimensional seepageflow analyses are described which are carried out with the programsystem HYD03 (Wittke, 2000).

The 4000 m long, 2000 m wide and 215 m high computation section isrepresented in Fig. 4.42. The tunnel tube is modeled over a lengthof 2000 m. The vertical plane through the tunnel axis (x = 0) isassumed to be a plane of symmetry. Accordingly, only one half ofthe tunnel needs to be modeled. The planes x = 0, y = 0 and z = 0are assumed to be impermeable boundaries. Constant piezometricheads of h = 215 m are assigned to the nodal points which are lo-cated on the planes x = 2000 m and y = 4000 m. This head corre-sponds to the z coodinate at the ground surface (Fig. 4.42). Thelocation of the groundwater table lowered as a result of the head-ing (phreatic surface) is determined by means of the iterativeanalysis (Wittke, 2000).

Fig. 4.42: Computation section, FE-mesh and boundary conditions

The boundary conditions in the area of the tunnel are representedin Fig. 4.43 and 4.44. The segmental lining is assumed to be im-permeable. The inflow of water into the tunnel is possible only

- 193 -


through the temporary face and in the area of the shield up to 12m behind the face (Fig. 4.41 and 4.44). As boundary condition apiezometric head is assigned to all nodes located in this area,which corresponds to the nodle's geometric head z plus the pres-sure head of the slurry pS/�W. In open mode pS = 0 is valid (Fig.4.44).

Both, isotropic and anisotropic permeability of the ground, whichcan be described with a permeability tensor (see Section 3.4.1),were investigated to evaluate the water quantity Q seeping intothe machine area depending on the slurry pressure pS. The principaldirections of permeability for simplification are assumed to coin-cide with the coordinate axes (kfxx, kfyy and kfzz, see Fig. 4.43).

Fig. 4.43: FE-mesh, detail, boundary conditions in the area ofthe tunnel

- 194 -


Fig. 4.44: Boundary conditions in the area of the tunnel: a)Open mode; b) closed mode

In Fig. 4.45, the calculated drawdown water table and the distri-bution of piezometric heads in form of equipotential lines arerepresented in a vertical section through the tunnel axis (planex = 0) for the case of a heading with open mode (pS = 0) and steadystate seepage flow. The permeability of the rock mass is assumedto be homogeneous and isotropic (kfxx = kfyy = kfzz = kf, see Fig.4.43). According to Section 3.4.2, equation (3.61c), in this casethe equation of steady state seepage flow and consequently alsothe distribution of piezometric heads and the drawdown are inde-pendent of the magnitude of the permeability coefficient:

� � � � 0 h T �� . (3.61c)

The maximum drawdown in this case results to �h � 30 m (Fig.4.45).

- 195 -


Fig. 4.45: Equipotential lines and drawdown water table, steadystate, vertical section through the tunnel axis(open mode, homogeneous and isotropic rock mass per-meability kfxx = kfyy = kfzz = kf)

If the permeability of the rock mass in horizontal direction inparallel with the tunnel axis (kfyy) is reduced by two or four or-ders of magnitude respectively, compared to the permeability coef-ficients kfxx and kfzz, very steep, narrow and deep drawdown conesresult for the case of steady state flow with a maximum drawdownof �h � 109 m (Fig. 4.46a) and �h � 161 m (Fig. 4.46b), respec-tively. Also here, the distribution of the piezometric heads andthe drawdown are only dependent on the ratio but not on the abso-lute magnitude of the permeability coefficients. Also in theseanalyses, a heading with open mode (pS = 0) is simulated.

In Fig. 4.46, also the calculated hydraulic gradients at the tem-porary face of I = 45 and I = 65 respectively, are represented.According to Section 3.4.3, the seepage forces {FS} can be calcu-lated from the gradients {I} as follows:

{I}}{F wS � � . (3.69)

In equation (3.69), �w is the unit weight of the water. The seepageforces have to be accounted for in the stability analyses (seeSection 3.5.1). An example is given in Section 9.6.

- 196 -


Fig. 4.46: Equipotential lines and drawdown water table, steadystate, vertical section through the tunnel axis(open mode, homogeneous and anisotropic rock masspermeability, kfxx = kfzz = kf): a) kfyy = kf � 10-2 ;b) kfyy = kf � 10-4

4.4.3 Seepage water quantities

From the results of the finite element analyses carried out for arock mass with homogeneous and isotropic permeability, a relation-ship for the calculation of the water quantity Q seeping into themachine area was derived, taking into account the influence of thepermeability coefficient kf, the magnitude of the slurry pressure

- 197 -


pS0 at the level of the tunnel's invert and the boundary conditionsrepresented in Fig. 4.42 to 4.44:

[m/s] kba[l/s] Q f�� (4.1)

with a = 1 – 0,055 � pS0 [bar]c = 1,04 � 107.

Accordingly, Q increases proportionally with kf and decreases withincreasing slurry pressure pS0.

Fig. 4.47: Rock mass with isotropic permeability kf, water in-flow Q into the machine area as a function of kf fora mean supporting pressure pSO = 8 bar

- 198 -


In Fig. 4.47, the water inflow Q, evaluated on the basis of thisrelationship and for a slurry pressure of pS0 = 8 bar, is plottedversus the water permeability coefficient kf in a double logarith-mic scale.

It is assumed that the water quantity seeping into the tunnel isnot allowed to exceed the following limit values according to therequirements of a supervising authority:

- moving average over four weeks: adm. Qm = 65 l/s,

- moving average over one week: adm. Qw = 200 l/s,

- moving average over 24 hours: adm. Qd = 300 l/s.

In Fig. 4.47, these admissible inflow values are represented ashorizontal lines. Accordingly, in a homogenous rock mass with iso-tropic permeability, these values can be kept with permeabilitycoefficients of kf � 10-5 m/s, if a slurry pressure of pS0 = 8 baris applied.

In the following the influence of the rate of advance and ofgrouting measures on the inflow of seepage water is examplarilyinvestigated. For reasons of simplification, an isotropic perme-ability of the rock mass is assumed. Furthermore, the rock masspermeability in the area of the tunnel is supposed to range be-tween 5 � 10-7 m/s and 5 � 10-5 m/s. Along the considered tunnel sec-tion from chainage 350 to 800, the permeability distribution whichis represented in Fig. 4.48a is assumed.

As a basis for the evaluation of the moving averages over differ-ent time intervals performance data are required. In Fig. 4.48,the assumed rates of advance along the considered tunnel sectionare given. Because of the water pressure and to reduce the seepagewater quantity, the TBM must be driven with closed mode. Thereforecomparatively low values are, as a start, selected for the ratesof advance.

In the following, the evaluation of the moving averages for theinflow of seepage water will be explained. In the first step, theaverage coefficients of permeability of the rock mass in the ma-chine area are calculated for all positions of the temporary facein the considered tunnel section, assuming a spacing of one meterbetween the positions. This is done on the basis of the distribu-tion of permeabilities along the tunnel represented in Fig. 4.48a.

- 199 -


In the second step, with the aid of relationship (4.1), the quan-tities of water Q seeping into the machine area are evaluated forall positions of the temporary face. Thus, Q as a function of theposition of the TBM is obtained.

Fig. 4.48: Moving averages of water inflow into the machinearea in case of low rates of advance v, pSO = 8 bar:a) Permeability distribution; b) rates of advance;c) moving averages of seepage water quantities

- 200 -


The positions of the TBM then can be converted into points in timewith the aid of the assumed rates of advance. In this way, Q as afunction of time is evaluated. Hence, the moving averages over 24hours Qd, over one week Qw and over four weeks Qm can be evaluated.In Fig. 4.48c, these data are plotted as a function of the chain-age on the basis of the permeability distribution represented inFig. 4.48a and the rates of advance given in Fig. 4.48b. As a re-sult, the limit value adm. Qd is kept in the complete tunnel sec-tion under consideration. The limit values adm. Qw and adm. Qm,however, are exceeded in some areas (Fig. 4.48c).

The water inflow could be reduced by means of an increase of theslurry pressure pS0. A complete support of the 150 m high watercolumn, however, would require a slurry pressure of 15 bar, whichcannot be applied by the TBM.

Water inflow can also be influenced by the rate of advance v. InFig. 4.49c, the moving averages Qm for two different distributionsof rates of advance represented in Fig. 4.49b are given. They arereferred to as "slow tunneling" (see also Fig. 4.48b) and "fasttunneling". In the case of the "fast tunneling", the rates of ad-vance are assumed to be twice as high as in the case of the "slowtunneling". The "fast tunneling" leads to a considerable drop ofthe evaluated Qm values compared to the "slow tunneling". The limitvalue adm. Qm, however, is still exceeded over a length of approx.200 m even when a higher rate of advance is assumed.

A further reduction of the water inflow can be obtained by meansof a sealing of the rock mass with advancing grouting. A reductionof the coefficient of permeability between chainages 600 and 700to a value of 10-6 m/s in the case of "fast tunneling" for exampleleads to the Qm values represented in Fig. 4.49c by the red line.For this case, the limit value adm. Qm is only marginally exceededover short tunnel sections. The limit values adm. Qd and adm. Qware not exceeded over the complete length of the considered tunnelsection.

Thus, for this example the given limit values for the inflow ofseepage water into the tunnel can be kept to a large extent by asupport of the temporary face in connection with a high rate ofadvance and advancing grouting.

- 201 -


Fig. 4.49: Moving averages Qm of water inflow into the machinearea as a function of the rate of advance and ofgrouting measures, pSO = 8 bar:a) Permeability distribution; b) rates of advance;c) moving averages of seepage water quantities

4. Stability and water inflow in the machine area 4.1 ...

Documents