Evaluation of new Austrian tunnelling method applied to Bolu tunnel’s weak rocks

Ebu Bekir Aygar

Fugro Sial Geosciences Consulting and Engineering Ltd., Cankaya, 06690, Ankara, Turkey

Abstract Since the development of the new Austrian tunnelling method (NATM) in the 1960s, this technique has been applied successfully in many tunnels.However,opinions of NATM principles emerged till 2000,i.e.NATM is not a tunnelling method, but an approach covering all general principles of tunnelling. To investigate the general principles of the NATM, this study focused on tunnelling practises in the Bolu tunnel,and evaluated the conditions under which the NATM practises could be effective.The Bolu tunnel project was designed following the NATM principles. It is evident that practises adopted in this tunnel are important with respect to the NATM. In addition, it shows that the solutions to the problems encountered in this tunnel are consistent with the NATM principles. Finally, the study determines the ground types of the NATM principles and proposes associated updates.

2020 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:Bolu tunnel New Austrian tunnelling method (NATM)Support types

1. Introduction

The new Austrian tunnelling method (NATM) was developed between 1957 and 1965.In 1944,Rabcewicz categorised rock masses into seven types (from clay to hard rock) and loosesing,squeezing and swelling.Moreover,the rock mass was classified into loosening,squeezing and swelling status (Goricki, 2003). Different support systems have been developed subsequently based on this classification(Goricki,2003).In 1957,rock mass behaviour,rock mass parameters, duration of failure, and tunnel supports were correlated with each other. The fundamental principles of the NATM were proposed by Rabcewicz(1964a,b;1965).In 1974,Pacher,Rabcewicz,and Golser (Goricki, 2003) correlated rock mass behaviour with different elements and classified rocks into four main groups:

(1) Stable, to potential for large overbreak;

(2) Friable, to heavily friable;

(3) Squeezing, to heavy squeezing; and

(4) Special classes.

From then on,NATM has been widely applied in many countries.Significant developments in the NATM were explained by Romano(2009). In the study conducted by Golser and Mussger (1979), the application and success of the NATM with the scenarios during tunnel excavation are evaluated.Jakoubek(1986)demonstrated the advantages and applications of NATM in a tunnel. Wagner (1986)compared the applications of the NATM in shallow and deep underground structures with other methods. Studies were also conducted on deep tunnels under difficult rock conditions,applications of NATM in different tunnels, and NATM principles (Braun, 1980;Brown, 1981). In addition, Müller (1978) emphasised that NATM has often been misunderstood“It is not so much,a way of excavating and supporting,but rather a concept. Success depends on following a set of principles,one of which is to utilise the surrounding rock mass to become the main load bearing component,with the lining establishing a load bearing ring”. The importance of ring closure distance and time and that of the invert section in weak soils are also emphasised in his study.

Romano(2009)suggested different nomenclatures that are used worldwide with respect to the NATM, such as Spritz Betonbauweisse in Germany, Méthode convergence-confinément in France,SCL (Sprayed Concrete Lined Tunnels in UK), NMT (Norwegian method of tunnelling) (Norway), several other names in Japan(CDM, UHVS), and SEM (Sequential Excavation Method) that is frequently used.

Debates and different opinions have been proposed for the NATM since its proposal. Some researchers hold different option against the integration of general tunnelling method and classical drilling and blasting (DB) method with a single name. In addition,Kovari (1994) criticised the NATM with the concept of “Erroneous concepts behind the new Austrian tunnelling method”.In 1994,the collapse of Heathrow tunnel caused a revision of NATM, owing mainly to the English publications regarding collapses using NATM(Romano, 2009).

In 1978, rock mass classification for tunnelling became part of the Austrian standards (Goricki, 2003). In 1983e1994, the standards were revised without changing the main concept by Rabcewicz and Golser (1973) on the basis of their principles, Rock mass was then divided into three main groups: A (stable), B(friable), and C (squeezing).

In 2001, the rock mass classification was revised again. A significant modification of ONORM B2203 was that the issues of rock mass characterisation and classification were excluded. Standards were re-visited in 2001 and 2004 (Goricki, 2003; P?tsch et al.,2004; Schubert, 2004). In addition, a flowchart illustrating the soil behaviour and support system behaviour as geotechnical design was proposed.P?schl and Kleberger(2004a,b)developed an approach in terms of rock mass classification of geotechnical risks.In a work by Palmstrom and Stille (2007), engineering tools for tunnels were proposed. The study of Marinos (2010) on tunnel behaviour mechanism was similar to that of Schubert(2004).In the present study,the works of Goricki(2003)and P?tsch et al.(2004)were considered as the basis and compared with other methods.The behaviours of approximately 62 tunnels were examined, and eventually, a tunnel behaviour chart was created.

In China, many highways and railway tunnels are under construction. However, some problems were encountered in these tunnels because of the high rate of deformations during tunnel excavations(e.g.Sun et al.,2019).The most popular tunnels among those in China are the Muzhailing railway tunnel and Muzhailing road tunnel.The length of the Muzhailing tunnel on the Lanzhoue Chongqing railway line is 19.1 km.Its cross-sectional area is 70 m2.The excavation process was completed in eight years.Deformations continued for three years after completion of the excavations. The tunnel has a highly complex geological structure, passing through 11 faults with a total length of 4500 m(Chen et al.,2017).Geological units along the tunnel route generally exhibit low strength,due to high content of carbonaceous slate, and high overburden stress.During tunnel excavation, failures in the steel shafts, cracks in the shotcrete, and swelling at the basement were observed. In certain parts of the tunnel,deformations reached up to 1.6 m.The supports were applied according to the principle of “strong support, weak release, support while release, rapid construction”. Rapid construction implies rapid excavation,rapid support,and rapid closure(Xi,2017).Moreover,the support system was selected with respect to pipe ahead,stringent grouting,short excavation,strong support,rapid closure,regular measurement,and tight lining principles(Xi,2017).

Furthermore,the Zhegushan,Laodongshan,Minyazi,Xiangshan,Yingfeng, and Yezhping tunnels in China represent similar problematic cases (Wang et al., 2019). Many problems were encountered similarly in these tunnels. During excavations, many fault zones, weak soils, and shear zones were identified. While excavating these critical sections,large deformations were encountered,failures occurred in the support systems,and the deformation rates in the Xiangshan tunnel archetypally reached up to 5.4 cm per day.High-risk scenarios in these sections were successfully overcome by modifying the tunnel support systems appropriately (Wang et al., 2019).

In this study,applications of NATM principles in Bolu tunnel will be discussed.The conditions under which NATM principles fail will be discussed and possible solution suggestion will be provided.The NATM principles which failed in weak soils and fault zones will be determined and the revisions to be made will be demonstrated.

Fig.1. Bolu tunnel location map.

Fig. 2. Cross-section of Bolu tunnel (unit: m) (Geoconsult,1997b).

2. Principles of NATM

The NATM is based on the principle of maximising the capacity of the ground to sustain its own weight by precisely and rationally balancing the pressures that affect the surrounding rock and support. This is achieved by forming a cavity in the rock or ground through which the tunnel will pass and reinforcing this cavity with support elements(Rabcewicz and Golser,1973).

Fig. 3. Geological profile of left tunnel of Bolu tunnel (Simsek, 2004).

Table 1Rock behaviour according to Austrian standard ONORM B2203 (Dalg??, 2002).

There are two types of support systems. The first system involves a flexible outer arch or protective support designed to balance the structure.The system is reinforced by additional steel ribs,bolts, and shotcrete. The ring is then closed using invert concrete.During the rearrangement of forces, the behaviours of the protective support system and the ground surrounding the tunnel are controlled by a highly developed measurement system (Vardar,1979,1985).

The second support system involves an inner concrete lining.However,this lining is not fabricated until the outer lining reaches the balance. The purpose of this concrete arch is to maintain or increase the safety factor as necessary (Rabcewicz and Golser,1973). The fundamental principles of NATM, which were proposed by Müller (1978), are listed below:

Table 2Rock class designation according to Austrian standard ONORM B2203(Dalg??,2002).

(1) The main element of the tunnel support is the surrounding rock.The main function of the artificial supports is to aid the rock around the tunnel in supporting itself.

Fig. 4. Details of rocks of classes A2, B1, and B2 (unit: m) (Aygar, 2000).

(2) The initial robustness of the rock should be preserved. The main principle is preserving the original strength of the rock mass to the feasible extent.

(3) The loosening of the surrounding rock should be prevented to the feasible extent. This is because loosening of the rock reduces its bearing resistance and increases dead loads.

(4) The protective zone should be formed without reducing the bearing resistance of the rock. The deformations that occur after excavation will be sufficient for forming the protective zone. However, the deformations should be controlled so that they will not lead to loosening of the rock, which will reduce the bearing resistance. If this is achieved, the safety factor and cost-efficiency of the works will increase.

(5) Reinforcement should be carried out in a timely manner(neither too early nor too late) and with the necessary flexibility to ensure that the structure providing the lining resistance is neither too rigid nor too weak.

(6) The reinforcement forces must be using the binding type. If substantial deformations and loosening are likely to occur after excavation, the reinforcement resistance should be in the form of distributed loads, whereas the reinforcement measure should cover the cavity surface. This is most effectively achieved by using shotcrete, which binds within a short period of time.

(7) Both temporary and permanent reinforcement linings should be in the form of ‘thin shells’, i.e. thin, shell-shaped,and bendable. The bending moments in the shell, as well as the pull and shear fractures resulting from these, can be prevented in this way.

(8) Reinforcement should be carried out with wire meshes,bolts, shotcrete, and steel ribs. Wire meshes and ribs can provide the necessary flexibility instead of thickening the shell.In order to enable the rock to carry its own weight,the effective stresses should be conveyed into the rock mass through anchorages.

Fig. 5. Details of support systems for rocks of classes C1, C2, L1, and L2 (unit: m) (Aygar, 2000).

Table 3Details of support class at design stage of Bolu tunnel.

Table 4Anticipated support class distribution of Bolu tunnel.

Table 5Predicted support class distribution of Bolu tunnel.

Fig. 6. Predicted support class percentage.

Fig. 7. Standard CM class, Option 1 (Geoconsult,1998b).

Fig. 8. CM 35 support class details, stretch 2-Bolu tunnel by-pass class CM for metasediment refinement initial support (unit: m) (Geoconsult, 2001).

Fig. 9. CM 45 support class, longitudinal section (Option 2) (Geocosult,1998b).

The reinforcement shell should have a closed ring-shape.From a static perspective, the rings must be closed so that an opening could be completely load-bearing. A ring with joints, springs, or notches is unstable compared with a circle.With this consideration,excavation and reinforcement lining of the top heading and the right/left sides should adhere to the cavity wall in an annular manner. Evidently, this can be achieved well by reinforcing shells that follow the complete cross-section of the excavation. The ring should be established at the earliest. The behaviour of the rock during the formation of secondary stresses depends on the deformations of the lining. A tunnel section that is not completely closed will not assume the role of a carrier/bearing ring,and hence will lead to the loosening of the rock.

Fig.10. CM 45 support class details, stretch 2-Bolu tunnel by-pass class CM45 initial support (unit: m) (Geoconsult, 2002).

Rounded spacer profiles should be used.The cross-section of the underground cavity should be circular or elliptical, with no protrusions or corners. The first lining should be thin to prevent undesirable bending moments.The inner lining should also be thin and be tightly bonded to the first lining frictionlessly. However, to prevent inner lining from interfering with the loads that may be transmitted from the first lining, it must be placed firmly by using its entire surface, with no friction applied to the inner lining.

Fig. 11. Option 3 intermediate (Bernold) lining longitudinal section, stretch 2 Bolu tunnel Option 3 Bernold initial support (unit: m) (Geoconsult,1999a).

Fig. 12. Bench pilot tunnel support details, stretch 2 Bolu tunnel Option 4 initial support section with deformation elements (unit: m) (Geoconsult,1999b).

Fig. 13. Predicted and practical geological cross-sections at the Elmal?k entrance(Chainage in km and elevation in m) (Geoconsult,1998d).

Fig.14. Predicted and encountered geological cross-sections at the Asarsuyu entrance(Chainage in km and elevation in m) (Geoconsult,1998d).

Fig.15. Vertical deformation in chainages 64 t 117 km to 64 t 246 km.

Fig. 16. Displacement measurements at chainage 64 t 182 km in the left tunnel of Elmal?k (Astaldi,1998).

The stability of the structure must be ensured with prereinforcement procedures. Deformations in the underground rock structure consisting of the lining and surrounding rock should be halted before the inner lining is put in place.The secondary stresses should achieve their balance. The task of the inner lining is to enhance safety and improve architectural appearance.

The water pressure transmitted to the lining via the rock mass should be discharged through drainage. To achieve this, drainage pipes should be installed around the space.

3. Bolu tunnel

Fig. 17. Reinforcement work prior to the earthquake in the section that collapsed during Düzce earthquake (Aygar, 2007).

Fig.18. Extensometer installation on shotcrete surface.

Fig.19. Extensometer installation locations.

Fig. 20. Extensometer measurements performed on ceiling section of Elmal?k left tunnel at chainage 64 t 034 km (Astaldi,1997).

Bolu tunnel located on the_IstanbuleAnkara highway consists of two tunnels, each having three lanes. Excavation initiated at the Asarsuyu (Istanbul) side on 16 June 1993 and at the Elmal?k(Ankara) side on 24 June 1994. The excavation of the tunnel was completed in 2006, and it was open to traffic in 2007(Fig.1).

Between the two tunnels, there is a buttress with 50e60 m width. The tunnels are constructed mostly 100e150 m below the ground surface,with the deepest point at 250 m.With the changes of the ground conditions, lining thickness, and deformations, the cross-sectional area of the tunnel face varies between 133 m2and 260 m2.The equivalent excavation diameters are 13e18.2 m.In the typical excavation cross-sections,the inner surface of the concrete lining has a horizontal opening of 14 m and height of 8.6 m(Fig.2).

Fig. 21. Extensometer measurements performed at the left lower half section of Elmal?k left tunnel at chainage 64 t 034 km (Astaldi,1997).

Fig.22. Extensometer measurements at the right lower half section at the Elmal?k left tunnel at chainage 64 t 034 km (Astaldi,1997).

The tunnel route is located near the active North Anatolian transform fault between the Eurasia plate in the north and Anatolian plates in the south. It is shifting westward from the Anatolian plate.

The general geological structure adopts the form of the North Anatolian metamorphic crystalline base. The Silurian, Devonian,and Carboniferous layers are composed of conglomerates, arkose,sandstone, mudstone, marl, shale, limestone, and dolomitic limestone (Astaldi, 1993), as shown in Fig. 3. The crystalline bedrock consists mostly of granite, grandiorite, quartzdiorite, and diorite,whereas metamorphic rocks in the amphibolite faults contain gneiss and amphibolites as migmatite.The ridges of the crystalline basins within Paleozoic formations have been eroded.

In the young Paleozoic layer,the northern continental basin was separated from the marine basin to the south. This formation resulted from low-grade metamorphism transforming old sediments into marble,phyllite,and schist.Owing to many faults on the route of the Elmal?k tunnel, a large number of rock mass blocks(crystalline bedrock, metasediment rock series, and the relevant sections of the flysch formation) are buried within a large-scale matrix of clayey faults, stretching to several hundred meters.

Fig. 23. Geology of Elmal?k entrance, where a collapse occurred during the Düzce earthquake (Geoconsult,1999c).

Fig. 24. Support system applied in the right tunnel of Elmal?k between chainages 54 t 140 km and 54 t 080 km (Geoconsult,1996).

The geology of the Elmal?k side in the fault zone contains flysch series with largely tectonic and rigidly weighted fault clay infilling with smooth discontinuity surfaces and exhibiting plastic properties. These flysch series comprise of claystone, siltstone, and limestone units. The Asarsuyu side is generally traversed by a metasediment series and cataclastic zones formed by these series.This formation continued until the transition to the Bakacak fault and intersects with Bolu tunnel within a 200 m wide fault zone.The fault is oriented in EW direction and crosses the tunnel route perpendicularly between chainages 62 t 800 km and 63 t 000 km(in the left tunnel).These faults have north-oriented plunge angles on cross-sections and cut the tunnel by means of fault clay materials at 75 m in the left tunnel and 91.5 m in the right tunnel,respectively.This fault clay material is comprised of the mixture of metasediments (metasilt stone, quartz limestone, crystallised limestone,low to medium plastic,dense and well-cemented sandy silty fault clay matrix, smooth and polished surface, and water flow)and quartz rocks t amphibolite t metacrystalline basement.This fault zone is located at the junction of the Asarsuyu and Elmal?k geological formations.

Fig. 25. Excavation works performed at Elmal?k right tunnel between chainages 54 t 140 km and 54 t 080 km.

The units in the Bolu tunnel can be classified into four main groups: metacrystalline base (Yedig?ller formation), metasedimentary base (_Ikizoluk formation), flysch sequence (F?nd?cak formation),and clayey fault zones(Dalg??,2002).The oldest unit in the tunnel route is Yedig?ller formation, located at the Asarsuyu entrance and comprised of metamorphic rocks. Yedig?ller formation is overlapped by _Ikizoluk formation comprised of Devonian metamorphic units with a tectonic contact. These two formations are primarily overlapped by intrusive granite and also sedimentary rock diversified from Upper Cretaceous to Upper Eocene. A lithological section comprised of amphibole and weathered granodiorite units are observed at the beginning of the Asarsuyu entrance.The other sections are presented to be sandstone, quartzite, and weathered granodiorite and amphibole units combined with marble. Contact between the crystalline basement and sedimentary overburden was formed by the activities of low-angle and conjugate faults.The entire sequence has a fractured structure as a result of clayey fault zones diagonally covering the units. The limestone basement at the Elmal?k entrance appears to be scattered sections in between sandstone and clayey strata containing limestone blocks. The entire sequence contains clayey fault zones and fault infillings.

The fault zone material at the tunnel elevation is in a series of two units. The first unit is a dark brown, highly plastic, and smoothly polished surface.Meanwhile,the second unit consists of a reddish brown, medium plastic, highly smooth-hard polished surface,and these materials constitute the Bakacak fault and geology.

Fig. 26. Measurements of vertical displacement between chainages 54 t 141.50 km and 54 t 240 km of the left tunnel of Elmal?k (Astaldi,1997).

4. Support systems applied to the Bolu tunneldC modified support system

Owing to the complex geological and geotechnical conditions in Bolu tunnel, there are large differences between the expected and practical rock classes.In the flysch series and the clayey fault zones,significant displacements have occurred in the support systems due to the swelling and compression characteristics of the ground.Large deformations (approximately 1.5 m) have occurred in the tunnel while crossing the clayey fault zone in the Elmal?k right tunnel.This section was filled with a backfill material to prevent the total collapse of the tunnel (Geoconsult, 1998e). In addition, deformations larger than 1 m occurred in the flysch series and minor fault zones were observed in the left tube of the Elmal?k tunnel,where regular repairs had to be conducted. Apart from these problems at the Elmal?k entrances, a water flow/entry of up to 400 L/s was observed at the metasediment series of the Asarsuyu entrances. This was caused by a large number of instability problems in the tunnel(Aygar,2000).To resolve all these problems,the C modified(CM)support system was developed and applied for the first time in Bolu tunnel (Geoconsult,1997a).

4.1. Primary design phase of the tunnel

In Bolu tunnel, the project was performed according to the NATM principles. Austrian standards (ONORM B2203) were also implemented.Accordingly,rock behaviours were divided into three categories as stable, brittle, and squeezing (Table 1), and the support systems were divided into eight main groups(Table 2).Figs.4 and 5 illustrate the determined support classes.

Fig.27. Measurements of horizontal displacements between chainages 54 t 141.50 km and 54 t 240 km of left tunnel of Elmal?k (Astaldi,1997).

Fig. 28. Displacement measurements in the right tunnel of Elmal?k at chainage 54 t 147 km (Astaldi,1997).

Figs.4 and 5 and Table 3 detail the proposed support system for each rock class.Evidently,all the support systems were applied at the initial design phase depending on the NATM principles and flexible outer lining principle. The thickness of the shotcrete pavement increased from 5 cm to 25 cm. However, due to unpreventable deformations originated from a previous design of shotcrete thickness of 25 cm, the coating thickness had to be increased to 140 cm for the upper half and 300 cm for the bench section.

Fig. 29. Concrete filling implemented at right tunnel of Elmal?k between chainages 54 t 140 km and 54 t 080 km following the collapse (Aygar, 2000).

Fig. 30. General view of concrete filling implemented in Elmal?k’s right tunnel between chainages 54 t 140 km and 54 t 080 km following the collapse (Aygar, 2000).

According to ONORM B2203, excavation would be conducted theoretically as a full face in class A1 rocks. In practise, it is excavated in two parts:top and bottom.Rocks of class A2 are similar to those of class A1. Excavation in class B1 rocks is divided into two parts:top heading and bench.In class B2,two separate excavations were performed for top heading-bench and basement,respectively.In general, excavations were conducted by DB method. Excavation of rock masses sensitive to vibration was performed by excavators.The top and bottom excavations were performed separately in class C1 rocks. Excavation was performed by DB method or excavators.Separate excavations were performed at the top heading and bench section of class C2 rocks. In many cases, reinforcements were required at the face of the top heading.Excavation was performed by DB method or tunnel excavators.

Excavation was conducted for the top heading, middle, and basement sections in class L1 soils. In many cases, reinforcements are necessary for the face of the top heading. Excavation was conducted by vibration-free blasting or tunnel excavators. Blocks and hard rock sections need to be detonated. Staged drilling and implementation of side galleries were required to overcome instability problems at the face in class L2 ground units.Excavation was conducted with tunnel excavators.

In the design phase of the Bolu tunnel,1% A2 rock, 9% B1 rock,40%B2 rock,19%C1 rock,11%C2 rock,10%L1 rock,and 10%L2 rock were expected.The rock classes estimated during the design stage before excavation are presented in the Table 4. The excavation in Bolu tunnel was planned to be carried out as follows:50%on A or B class rock, 30% on C1 or C2 class rock, and 20% on L1 and L2 class rock. The rock classes encountered after completion of tunnel excavation are presented in Table 5 and illustrated in Fig. 6,showing that the rocks upon tunnel excavation is composed of 75%compacted (C) rock class. In addition, 68% of C2M, CM, OP3, and OP4 rock classes were identified during excavation,which were not anticipated during design phase. These segments were entirely designed regardless of the NATM principles.

Fig.31. The approximately 3.5 m buckling that occurred in the tunnel top heading due to excessive displacements, and the mechanism of failure in the top heading invert(Aygar, 2000).

Fig. 32. Damage to the tunnel section due to excessive displacements (Aygar, 2000).

4.2. CM support system

The CM support system is essentially a method developed specifically for Bolu tunnel (Aygar, 2000). The CM support system was based on the solutions proposed for the section of Bolu tunnel where support systems were applied successfully between 1993 and 1999. That is, all these support systems, which differ significantly from each other in terms of their application ranges, are classified as CM support systems.The Bernold lining was applied in the flysch series and minor fault zone sections,whereas the bench pilot tunnel method was implemented for the large fault zones.The support systems were applied in the metasediment series and metacrystalline grounds. All these application methods belong to the CM support system.Different support systems were categorised under a single class according to the requirement of significant changes in the support systems proposed in the project design phase. The CM support system was applied on four main levels(Geoconsult,1998a; b):

(1) Metasediments (observed in Asarsuyu);

(2) Metacrystalline (observed in both Asarsuyu and Elmal?k entrances);

(3) Flysch series(observed only at the entrance of Elmal?k);and

(4) Clayey fault zones (observed in the transitional sections of the ground types mentioned above).

These support systems were grouped under four main headings(Tokgozoglu and Isik, 2002):

(1) Option 1: CM class temporary shotcrete basement with or without invert (CM 35);

(2) Option 2:CM class(flysch series,low swelling potential)with reduced ring closure distance and temporary shotcrete with invert (CM 45);

(3) Option 3:Intermediate(Bernold)lining,temporary shotcrete with invert;and

(4) Option 4: Bench sectiondpilot tunnel excavation.

Option 1 is implemented through metacrystalline layers,whereas Option 2 is generally excavated through fracture-crushed metasediment rocks and flysch series with low swelling potential.In addition, Option 3 is implemented for small fault zones with blocky flysch series exhibiting high swelling potential, whereas Option 4 is designed for the worst-case fault conditions such as thick fault clay layers. In accordance with the above explanations,the details of the support and excavation class of these sections are summarised below(Tokgozoglu and Isik, 2002):

Fig. 33. Cross-section of bench pilot tunnels (unit: m) (Geoconsult,1999b).

(a) Option 1:CM class(Metacrystalline layer):Standard CM class is a tunnel support system having shotcrete lining of a maximum of 35 cm, monolithic basement concrete, and 360 m rock bolt installed in each advance. This class has a maximum ring closure distance of 30 m. The excavation speed remains constant in this ring closure distance (Figs. 7 and 8).

(b) Option 2: CM class (Flysch series) reduced ring closure distance, implementation for more unfavourable rock conditions. Intermediate (temporary) basement concrete is required for the top heading section. Shortening the ring closure distance does not affect the excavation speed. The shortest ring closure distance is 22e23 m. The shotcrete lining is 45 cm (Figs. 9 and 10).

Fig.34. A view during excavation of the bench pilot tunnel method in Elmal?k tunnel.

(c) Option 3: Intermediate (Bernold) lining: Option 3 is applied for weak flysch series and clayey fault zones(length less than 20 m).The purpose of this system is to ensure the stability of the excavation with rigid supports without applying a large amount of shotcrete in the areas where deformation increases. The intermediate linings are applied while the tunnel displacements continue at high speed, aiming to maintain the stability of excavation with an intermediate lining. The Bernold lining is implemented after installing monoblock basement concrete. The ring closure distance ranges from 20 m to 24 m, and a temporary basement is applied to the top heading section. The Option 3 project is prepared for blocky flyer floors.This project can also be used for thin clay layers of fault zones. The basement concrete is installed as a monoblock with iron reinforcements. If necessary,large deformation can be suppressed by applying a temporary basement at the top heading section extending to the bench section. The primary lining in the support systems includes highly tight rock bolts for each interval(generally 1.1 m) with a shotcrete shell supported by TH-29 steel ribs (Geoconsult, 1998c). Fig. 11 illustrates the crosssection of the Option 3 support system (Aygar, 2007).

(d) Option 4 (Lower half pilot tunnel method): Under highly unfavourable ground conditions, pilot tunnel excavation of bench section is the only choice. Pilot tunnel excavation of bench section was performed in almost all the conditions in Bolu tunnel,particularly in clayey fault zones spreading over more than 20 m.In this case,the primary support consisted of 40 cm-thick steel fibre shotcrete applied to the tunnel face.The intermediate lining was located 8e16 m behind the face and consisted of an 80 cm-thick steel fibre lining.A bench and a deep(depth of 5.35 m)monoblock basement concrete arch were formed 22e35 m after the tunnel face so as to complete the tunnel circle (ring closure). The final lining (inner lining concrete)was comprised of 60 cm-thick concrete with class B40 iron reinforcements (Geoconsult, 1999d). Fig. 12 illustrates the support system for Option 4(Aygar,2007).

4.3. Practical and predicted geological conditions in Bolu tunnel and stability problems

Fig.13 illustrates the predicted geological sections and practical sections in the entrance to the Elmal?k tunnel. The predicted and practical geological sections for the entrance of Asarsuyu are presented in Fig.14.

Fig. 35. Bernold lining (unit: m) (Geoconsult,1998c).

The CM support system was implemented following the fundamental principles of the NATM, from 1995 to 1997. No significant differences were imparted to the support systems during this period. In accordance with the progress rate of displacement,there was no change in the support system except for altering the shotcrete thickness, steel ribs, and arrangement of the rock bolts.Meanwhile, in the left tube of the Elmal?k tunnel, the tunnel excavation was interrupted due to the unavoidable displacements.These were overcome by applying 60 cm-thick shotcrete lining to the flysch units. Subsequently, it was understood that Bolu tunnel potentially violated the fundamental principle of the NATM, and a‘flexible outer lining’failed to combat the displacements around the tunnel to form a protective zone. Therefore, it was decided that a more rigid lining method was applied in the flysch series and fault zones of Bolu tunnel. This resulted in implementation of the Bernold lining method(Option 3)(Geoconsult,1998c)and bench pilot tunnel method (Option 4) (Geoconsult, 1999d). Whereas the thickness of the first lining (shotcrete) in the metasediment and metacrystalline ground was 45 cm,this thickness exceeded 1 m in the clayey fault zones and flysch series.

Fig. 36. Application of Bernold lining in the tunnel.

4.4. Asarsuyu entrance

The tunnel behaviours and recorded displacements were consistent with the predicted values in chainages 61 t 900 km(left tunnel) and 51 t 800 km (right tunnel) from the entrance of Asarsuyu. The tunnel support systems were selected based on the rock classes, ranging from A2 to C2. Moreover, the number of rock bolts was altered when the deformation exceeded 1 m at points approximately 150 m from their entrances, and a significant number of remediation processes had to be carried out.As a result,the C2 support class was adapted to C3, and the number of rock bolts installed per meter of progression of the tunnel increased from 123 to 350. Furthermore, additional support elements were used (such as double steel ribs and longer bolts). In spite of these improvements, stability could not be maintained in this area, and the first lining was severely damaged. Given these undesirable scenarios, the faces were temporarily stopped at chainages 62 t 104 km (left tunnel) and 51 t 974 km (right tunnel). The CM class support was implemented for these locations after geological and geotechnical investigations, beginning in September 1996(Aygar, 2000).

4.5. Elmal?k entrance left tunnel

In the Elmal?k tunnels, the class C2 support was applied at chainages 64 t 140 km in the left tunnel and 54 t 074 km in the right tunnel.Subsequently,from this location,the C2 rock class was revised according to the increase in displacement and the deterioration of the ground, and the implementation of the CM support system was initiated. In the left tunnel, an attempt was made to advance from chainages 64 t 140 km to 63 t 880 km without excessive alterations to the support. In this section, the only modifications were to increase the number of rock bolts,alter the steel ribs, and vary the thickness of the shotcrete.

This section of the tunnel is located in the fault zone between chainages 64 t 140 km and 64 t 210 km.This fault zone is a lowangle fault zone with high plastic properties(Aygar,2000).The tunnel was excavated along the fault infilling which was dark brown-reddish brown in colour (Isik and Ozben, 2007). According to the test results, 80% of the material consists of clay fractions. Vertical deformations emerged due to the overlapping low angle fault zone,as shown in Fig. 15. The deformation value reached up to 63 cm. In addition,lateral deformations in this section of the tunnel reached up to 50 cm(see Fig.15).These measurements were taken in the period before 1996.Deformation rise was noticed in progress between 1996 and January 1998. Condition of the deformations is illustrated in Fig.16 with an uppermost value of 100 cm. When the deformation value reached 100 cm,additional shotcrete was applied as an emergency measure,as shown in Fig.17.In other words,the tunnel was at the threshold offailure andthosesupport systems were implemented to prevent tunnel collapse. Unfortunately, when the deformation valuereached100cm,Düzceearthquakeoccurred,and this partof the tunnel completely collapsed under the force of earthquake.

The deformation in the left tunnel led to the observed fractures and cracks in the concrete. Over elapsed time, these movements continued over a distance of 200 m from the tunnel face.Comprehensive repair and strengthening work started in the tunnel (Aygar, 2000). In addition, extensometers with lengths of 6 m,9 m, and 12 m were placed at the tunnel ceiling and lateral walls.The placement of the extensometers is illustrated in Figs.18 and 19.The extensometer measurements were performed at chainage 64 t 034 km of Elmal?k’s left tunnel to gain an understanding of the plastic area around the tunnel.

As presented in Fig.20,an examination of the measurements for the 6 m, 9 m and 12 m long single-point extensometers placed in the tunnel ceiling revealed that the displacements continued at a 12 m section in the tunnel ceiling. The 6 m, 9 m, and 12 m long extensometer underwent a total displacement of 7 cm,5.8 cm,and 2.5 cm,respectively.The extensometer measurements at 6 m,9 m,and 12 m in the left bench of the tunnel are illustrated in Figs. 21 and 22. In the lower left half (Fig. 21), the 12 m, 9 m, and 6 m long extensometers recorded displacements of 13 cm, 8 cm, and 4.8 cm, respectively. The displacements occurring at the right bench of the tunnel (Fig. 22) were observed to be 4.8 cm, 3.8 cm,and 0.8 cm by the 6 m, 9 m and 12 m long extensometers,respectively. The main reason for the differences between the measurements of the left and right benches is that the left tunnel was affected by the right tunnel excavation. Considering that the largest displacement occurred in the 12m long extensometer, it is apparent that the right tunnel severely affected the left tunnel in this section. According to the extensometer measurements, the plastic zone around the tunnel appeared to exceed 12 m. Considering that the length of the longest rock bolt placed in the tunnel system was 12 m,it was apparent that all rock bolts were placed in the plastic zone and cannot reach the elastic zone. This demonstrates that the outer lining of the tunnel did not provide a complete arch around the tunnel.

4.6. Evaluation of the right tunnel support systems

To cross the clayey fault zone in the right tunnel of Elmal?k between chainages 54 t 080 km and 54 t 140 km (Fig. 23), a pilot tunnel was opened in the upper half section (Geoconsult,1996). It was feasible to carry out sufficient geological and geotechnical investigations of the fault zone through this pilot tunnel. A fresh support system was developed. Then, the excavation restarted(Fig. 24). The excavation in the application phase is illustrated in Fig. 25. Despite of the application of the support system following the principle of a ‘flexible outer lining’ within the NATM, a large displacement of 160 cm occurred in this area(Aygar,2000).Severe buckling was observed in the intermediate invert in the top heading(Schubert et al.,1996).The tunnel stability was not assumed.

The displacement measurements between chainages 54 t 141 km and 54 t 240 km in the right tunnel of Elmal?k are illustrated in Fig.26.This section is located in the main fault zone and was involved with the section of the tunnel that collapsed during the earthquake.The measurements were obtained along a 100 m tunnel line,and the vertical displacements exceeded 100 cm over a period of five months.Moreover,a deformation of 65 cm was observed in the horizontal direction (Fig. 27). That is, the tunnel was severely narrowed both vertically and horizontally.The measurements made at chainage 54 t147 km in this section areillustratedin Fig.28.Inthe top heading,a vertical displacement of 1.64 m occurred in section 1(tunnel roof),whereas a vertical displacement of 1.23 m occurred in the left and right sides of the top heading(sections 4 and 5).In this sense, stability could not be achieved in the tunnel, and displacements continued during thisperiod with increasing intensity.Owing to unexpected large displacements, this section was completely abandoned and filled with backfill material to prevent tunnel collapse(see Figs.29 and 30)(Aygar,2000).

Figs.31 and 32 illustrate the damage at the tunnel caused by the displacements in this section. These images were captured during the reprofiling works performed after filling the tunnel with backfill material.

Various solutions were considered for the section that was filled following the collapse. First, it was concluded that the support system should be a method not based on NATM principles. The principle of maintaining an arch through a flexible outer lining lost its validity in the fault zones. Therefore, it was decided to utilise a more rigid lining. Eventually, the most appropriate project design for crossing this fault zone would be applying the bench pilot tunnel method (Figs. 33 and 34). As is evident in Fig. 33 two 5 mdiameter pilot tunnel excavations were made in the bench of these sections, and they were filled with iron reinforcement and concrete.The top heading section comprised 70 cm shotcrete.The fault zone was passed smoothly using the bench pilot tunnel method.Apparently,this method is completely out of the scope of the NATM principles, involving a very rigid lining thickness (Aygar, 2000).

It was not feasible to decide what support class should be implemented in the flysch series of the right tunnel (from chainages 54 t 140 km to 54 t 080 km). Based on the experience gained during applications in the left tunnel,it was clear that a rigid shotcrete layer of 60 cm would be insufficient and a middle lining would be required for these sections. The middle lining (Bernold lining) is a lining located between the shotcrete and the inner lining,for which a lining method was applied after completion of top heading, bench, and invert concrete linings (Fig. 35). The application of the Bernold lining in tunnels is illustrated in Fig. 36.

5. Conclusions

NATM is an approach that covers the principles of general underground support rather than a tunnelling method. The results obtained from Bolu tunnel reveal that it is not always feasible to implement the principle of“ensuring an arch with a flexible outer lining”.Thus,the implementation of this method could not always be an appropriate or economic approach. Implementation of this method resulted in significant increase in the cost of Bolu tunnel.The collapse of a 50 m section in the right tunnel can also be induced by implementation of this method.Furthermore,in the left tunnel,excavation works were halted for approximately 10 months,and continuous repair and reinforcement works were required.

A wide range of opinions is that each tunnel brings with a sitespecific method.The results obtained in a tunnel tend to serve as a guide for other tunnels.By the time,the excavation was completed for Bolu tunnel,which was originally designed according to NATM principles, but completed without it. Implementation of NATM principles and proposing solutions for the encountered problems would be appropriate only if they are revised according to the solutions developed for a case wherein the NATM provides unsatisfactory outcomes.The NATM principles consist of 22 items set out by Müller (1978). The revised ones are as follows:

(1) The reinforcement lining should be in the form of ‘thin shells’. However, when the ground conditions are highly inadequate(such as presence of widespread fault zones and flysch series containing low angle faults having the potential of swelling and squeezing), the outer lining should be significantly more rigid than the thin shell. If necessary, the bench pilot tunnel method should be applied in large fault zones. In sections where the ground is in good condition,both the temporary and permanent reinforcement linings should be thin, shell-shaped, and sufficiently flexible to permit bending. This aids in preventing the formation of bending moments in the shell.It also prevents the formation of associated pull and shear fractures.

(2) Reinforcements should be made of steel meshes, bolts and steel ribs,rather than by thickening the shell,to provide the necessary flexibility to the shell. With the deterioration of ground conditions, a more rigid system such as the Bernold lining will be required as a supplement to the reinforcement elements in the fault zones and flysch series.

(3) The inner lining should be thin and frictionless, yet tightly bonded to the outer lining.The outer lining should be thin to prevent undesired bending moments.To prevent them from interfering with the loads originating from the outer lining,the inner and outer linings should be placed firmly by using their entire surface without friction. Bolu tunnel has demonstrated that thin outer lining is not suitable for all ground conditions. This is why a rigid system should be preferred for the outer lining under weak ground conditions.

(4) The stability of the structure must be ensured during the prestabilisation phase. Deformation in the underground rock units including the crust and surrounding rock should be halted before the inner lining is installed. That is, the secondary stress condition should be finalised in a balanced manner.The purpose of the inner lining is to enhance safety and provide an appropriate architectural appearance. However,when groundwater is present,the inner shell should be sized to receive all loads. In these cases, the layer between the outer and inner shells should be waterproof using bitumen, nylon, membranes, etc. As a highly important element of temporary stabilisation, bolts must be protected against corrosion, so that they can be relied upon for permanent strengthening.When the stability of the structure is not secured by the outer lining,the deformation could not be completely halted, and they continue to a certain extent.Thus, the inner lining should be installed as close to the tunnel face as possible. Moreover, the tunnel should be reinforced with steel bars before deformation rates decrease below 2 mm per month based on the Austrian Standards.

Declaration of Competing Interest

The author declares that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The author thanks to General Directorate of Highways(KGM)for their supports.