10.7 New tunnelling techniques in relation to rock mechanics
10.7.1 The new Austrian tunnelling method (NATM); examples
According to Rabcewicz, a layer of shotcrete only 15 cm thick, applied to a tunnel of 10 m diameter, can safely carry a load of 45 tonnes/m2 corres- ponding to a burden of 23 m of rock, more than has ever been observed with roof falls. If a steel support structure incorporating 'number 20' type wide flanged arches at 1-m centres were used under these conditions, it would fail at about 65 % of the load carried by the shotcrete lining.
Shotcrete becomes strong very quickly (fig. 10.35), which is very important when a rapid high-bearing capacity is needed. Its immediate flexural-tensile
250.
200-
J50"
100-
0.15 0.5 1.0 7 10 days
Fig. 10.35 Shotcrete strength: (1) crushing strength; (2) shear strength; (3) tensile strength.
strength amounts to 50% and 30% of the compressive strength after 2\ days.
A recently introduced hardening-accelerating admixture based on silicifica- tion gives even better results. In one of the tunnels at the Kaunertal hydro- electric scheme a 2-cm jet of water was satisfactorily plugged with shotcrete
alone, without even installing a relief pipe. Bernold recommends the adjunc- tion to the shotcrete of steel needles, 2-5 cm long, 0-4 mm in diameter, in a proportion of 2 to 3 % of the solid concrete weight.
Rabcewicz writes that a shotcrete layer applied immediately after opening up a new rock face turns a rock of doubtful stability into a stable one. The close interaction between shotcrete and rock leaves the neighbouring rocks almost unaltered, enabling them to participate effectively in the arch action.
It also absorbs the tangential stresses which build up to a peak close to the surface of a cavity. The zone of arch action can be increased at will by rock bolting.
Russian experts tested shotcrete for the large Inguri tunnel (Masur, 1970).
Drilling through the shotcrete and the rock, they checked that the shotcrete really forms a solid bond with the rock. They tried grouting behind the shot- crete layer with pressures up to 15 or 20 kg/cm2, which was entirely successful.
Fissures and joints were well cemented by the injections.
The first successful application of a shotcrete lining was carried out at Lodano-Mosagno tunnel (Maggia hydroelectric power system) in Switzer- land in 1955. Rabcewicz and others (see Suttcliffe, 1969; Abel, 1970; Cecil, 1970) acquired considerable experience in the new methods. Some Austrian engineers were slow in adopting them, but the technique rapidly spread from Austria to Switzerland, Scandinavia, France, Australia, South Africa, Canada and the United States.
Rabcewicz himself commented on the probable sequence of events which take place during the loosening of rock masses around a tunnel (see section 10.5.2). A better description of the behaviour of rock masses around a tunnel or gallery was given many years before Rabcewicz's proposals in a brilliant paper by Robert Maillart (1922), commenting on some remarks by the great geologist Albert Heim (1878) and some subsequent comments by Charles Andrea (1926) (one of the builders of the Lotschberg tunnel). Further theoretical research by Talobre (1957), Kastner (1962), Lombardi (1970, 1972), Egger (1973) and Duffaut & Piraud (1975) has contributed much to the understanding of the efficiency of combining rock bolting and shotcreting.
In their papers, Maillart (1922, 1923) and Andrea (1926, 1961) carefully examine the experience gained during excavation of the Simplon tunnels, under well over 2000 m overburden. They explain how the rock mass, and sometimes the horse-shoe shaped masonry supporting it, deforms, the rock being progressively decompressed as decompressed rings are being formed above the tunnel. Sometimes rock spalls at the rock surface, mainly when the tunnel radius is too large, but finally settles, a relatively thin lining (0-40 m for the Simplon tunnel) being sufficient as rock support, in spite of the fact that the crushing strength of the concrete is less than the crushing strength of the rock. This analysis which is more realistic than Rabcewicz's approach could have led to an early discovery of the NATM. It did not.
A very lucid analysis of the problems of excavations and rock supports
264 Underground excavations
and of the historical development of the theories and methods has been given by Lombardi in his address to the Lucerne Symposium (1972).
Shotcrete, especially when combined with rock bolting, has proved excellent as a temporary support for all qualities of rock. In very bad cases steel arches are used for reinforcement of the weaker tunnel sections. Rabcewicz mentions a tunnel of 8 m2 cross-sectional area for a hydroelectric scheme in the Austrian Alps. Originally it had been driven without shotcrete, using only steel arches and steel lagging. When the tunnel reached a zone of kaolinized gneiss under an overburden of 250 m with heavy water inflow of 35 1/s, the pressure became so heavy that the arches were deformed and their footing forced into the ground. The heavy inflow of water could only be relieved slightly, as the discharge pipes became clogged shortly after placement.
Excavation had to be stopped. After re-driving the roof in the deformed portion, new steel arches had to be placed at 60-cm centres on heavy wooden sills and another arch interposed between each set. As soon as a set was placed the surface was immediately shotcreted to a complete ring. This difficult situation, aggravated by unsuccessful attempts at driving, was brought under control using these methods.
The Kaunertal tunnels and the pressure shafts are another example worth studying. The 5-15-cm thick shotcrete is now part of the complete thickened final lining. At some places the shotcrete lining failed because the 'Perfo- anchors' and bottom bracing were not applied in good time (delays up to one full year). The Serra Ripoli super highway twin tunnel (Italy) is another example where conventional tunnelling methods failed, but the shotcrete method reinforced by sectional steel was successful.
Experience by others than Rabcewicz can be mentioned. The Mont Blanc tunnel (France) (Panet, 1969) is a typical example of rock bolts being used mainly to avoid rock bursts. Under a rock overburden of 2000 m and with a fc-value of only 0-40, the Mont Blanc granite was overstressed. Bolts 1-50 to 3-50 m long, designed for a maximum load of 14 tonnes and a minimum spacing of 0*50 m were able to stabilize roof and walls. Rock bolting can be used either for pinning locally unstable rock masses or as a systematic rock support. Talobre (1957) and Rabcewicz & Rescher (1968) have developed methods of designing rock bolts for the purpose of stabilizing rock vaults.
The natural tendency for a 'sustaining arch' to be formed in overstressed rock masses, as shown by Hayashi's finite element method analysis (1970) explains the success of this practical method. Rock bolts, when stressed, apply an average pressure p* on the surface of the rock and consolidate the natural arch formed in the mass. Rock bolts designed to cause pressures p* = 0-70 kg/cm2 to 1-5 kg/cm2 are usually satisfactory. Higher values are often required for p* for large excavations like underground power houses, where anchored cables are being used. Veytaux underground power station (Rescher, 1968) and Lago Delio (Mantovani & Bertacchi, 1970) are good early examples of the NATM; Waldeck II (chapter 16) is more recent.
The NATM has also been widely used in the construction of underground railways, where difficult local tunnelling problems had to be convincingly resolved. The construction of instrumented test sections provided the most reliable design data for major tunnel schemes. This enabled the consulting engineers to control the tunnel work rigorously. Almost negligible settle- ments and consequently no damage to buildings is an essential condition of such tunnelling. Furthermore this method makes it possible to compare the calculation routine of the project stage with the reality of the final phase.
In Frankfurt, a 150-m-long test section was installed. In Munich, the test section was 60 m long. It showed that temporary ground stability could be achieved in difficult formations by groundwater lowering, combined with the instant installation of ground anchors and the application of a protection skin of shotcrete of 50 mm thickness, which was soon afterwards strengthened by a further layer of pneumatically applied concrete of 200 to 250 mm thickness.
Information on the use of the NATM in highly stressed rock formations (Tauern mountain chain in Austria) was given at the 24th International Symposium of the Austrian Society for Geomechanics, Salzburg, 1975 (summarized in International Construction, 1975, pp. 29-41). Experience has shown that under extreme stress conditions wire mesh, steel arches and shot- crete layers provide only protection against rock falls, water seepage and atmospheric influences, etc., while rock anchors attain the most dominant structural function by including the rock itself into the support system up to anchor length at least. The implication is that for every type of rock there is an optimum number of anchors of a certain length to suit the particular tunnel dimensions. This anchor system can safely compensate for the dis- turbed equilibrium of forces which was caused by the tunnel excavation.
Progressive rock movements up to 400 or 500 mm have been tolerated. In order to allow for such deformations, gaps in the shotcrete skin are left between the side walls and the tunnel roof. There the steel mesh can deform and the steel arches can be adjusted. Before the rock comes to rest, it is convenient for a second shotcrete lining to be installed which can accept and resist stresses in a similar manner to the first.
Similar news concerning the use of the NATM comes from the large Gotthard road tunnel (Lombardi, 1974) now under construction in Switzer- land, from the Snowy Mountains (Australia), from the 82-km long Orange- Fish tunnel (South Africa) and from Ruacana power station (South West Africa).
Very recently the construction of a section of the Munich Underground, in the middle of the town, in 'soft ground' (Golser et al., 1977), using the NATM techniques, has allowed engineers to develop some of their advanced theories (Rabcewicz et al., 1972). The very large Munich Underground tunnel was excavated with top heading and a bench near the head. In order to get to work on a nearly circular shotcrete-lined section as quickly as
266 Underground excavations
possible a temporary shotcreted invert was built on top of this bench, full excavation proceeding later.
The radial deformations of the excavations were measured systematically during the three successive phases of the lining:
(a) a first layer of shotcrete, including the supporting steel arches, (b) a system of anchoring,
(c) the second layer of shotcrete.
The importance given to systematic measurement of radial deformations is also characteristic of Lombardi's methods which will be discussed in the next chapters.
The new NATM techniques touch the whole field of modern tunnelling.
It has been used in the most diverse situations, but there is no agreed theory explaining how the thin shotcrete layer works so efficiently. Some authors give credit to its bending strength, others believe its efficiency is due to its shearing strength combined with the consolidation of the rock vault or rock walls, an opinion which seems to be corroborated by tests in the Inguri tunnel and others by Bernold. The effect of short rock bolts keeping rock blocks solidly together forming an arch, is easier to understand: the Romans knew how to build masonry arches without even using mortar. (See section 7.4 on the clastic theory of rock masses and section 10.7.2 on the Bernold sheet system and its failure by shearing.)
10.7.2 The Bernold sheet system (Wohlbier, 1969; Bernold 1970; Jaeger, 1975/76)
Bernold has developed a technique of thin concrete lining behind perforated steel sheets, which is applied immediately to the rock surface after blasting, before rock deformations due to internal rock ruptures and plastic deforma- tions develop. With this method rock strains are maintained within limits and a radial rock pressure is transmitted at an early stage to the concrete lining.
This technique is important when watertightiiess is imperative for hydraulic tunnels.
The technique developed by Bernold which has been used for all types of tunnels or galleries, some of them very large, as in the Gotthard road tunnel now under construction, is based on similar but not identical prin- ciples to the NATM. The basic idea is to build a thin concrete lining imme- diately behind the heading, the vertical face of the heading stiffens the rock vault nearby during early stages of concreting.
Figures 10.36 and 10.37 illustrate the technique favoured by Bernold. A ribbed reinforcement sheet (fig. 10.36) provided with slots or weep holes, rests provisionally on steel arches, supporting the steel sheet during con- creting of the lining (fig. 10.37). The concrete lining is 15 to 25 cm thick. A rule of thumb given by Bernold suggests a ratio djR = 1/15 or even 1/20
(d = thickness of the concrete lining, R = radius of the concrete vault).
During concreting, about 3 per cent of the concrete is lost through the weep holes. After hardening of the lining, grouting the contact zone with the rock is essential to fill any void. Then the arches are removed, but the steel sheet remains a permanent part of the lining.
y / A_/ \ / V
grouting
temporary installation arch
Fig. 10.36 Longitudinal section of the Bernold concrete lining system (dimensions in mm) (after Bernold).
overlapping of steel sheets
connection rod
continous stiffening rib section AA
Fig. 10.37 Plan and sections of the Bernold sheet system used to help prevent deformations (after Bernold).
268 Underground excavations
Bernold categorized rocks in the following way:
(a) slightly ficable (shearing strength larger than tangential tensions);
(b) ficable;
(c) very ficable (shearing strength smaller than tangential tensions);
(d) rock under stress,
The temporary stability of the rock roof is estimated at about 24-48 h for (a), 8-18 h for (b), 4-12 h for (c) and 0-2 h for (d).
Table 10.4 summarizes some technical data concerning the reinforcement sheet, of standard size 1200 X 1080 mm ( = 1-296 m2) and with 12 cm overlap.
Table. 10.4 Reinforcement sheet data
Sheet thickness 1 mm 2 mm 3 mm 5 mm Total weight (including
connection rod) per m2 110 kg 210 kg 31-0 kg 520 kg Steel cross-section of
one reinforcement rib 0-57 cm2 109 cm2 1-62 cm2 2-7 cm2
The thin concrete lining has been tested in Japan at the Public Works Research Institute, Ministry of Construction. The test arch had a radius of 4-128 m and a thickness of 25 cm with 2 mm steel sheet (1st test series).
Other tests were with thickness 15 cm, with 2 mm steel sheet (2nd test) and 25 cm without sheet. Rupture occurred by shear near the soffit of the arch.
Similar ruptures have been observed in situ on tunnel linings.
Theoretical research confirmed that a thin lining can be ruptured by high tangential compression stresses, ou in the arch, by buckling, bending or by shear. The most likely cause of failure is shear, and Kurt W. Weirich from the Bernold Office has published diagrams which explain how a lining has to be developed against possible rupture by shear (fig. 10.38).
Rupture by excessive tangential stress at is not likely to occur. It will be remembered that the finite element method analysis of rock stressing is supposed to yield the at values around the excavation, which, according to Bernold is not the most important. Buckling and rupture by bending would occur by lack of bondage of the lining with the rock, which justifies the grout- ing of the concrete-rock contact zone.
Gunite is used to cover the rough steel sheet and smooth the tunnel surface. In some cases the Bernold linings have been protected with a second concrete ring, concreted at a later stage. An impervious sheet of bitumen was laid between the two linings.
The Bernold system has been used mainly for large or very large tunnels but also for hydro-power galleries. It is very important to emphasize that a Bernold concrete lining put into place a few hours after rock excavation is really put under compression by rock deforming radially. When hydraulic
/>(t/m2)
Fig. 10.38 Possible method of failure of a lining by the rupture of the lining due to shearing.
pressure is applied to a pressure tunnel, these compression stresses can balance some of the tensile stresses induced in the lining.