You are on page 1of 7

Lining of tunnels under groundwater pressure

T. Franzn SvBeFo Swedish Rock Engineering Research T.B. Celestino University of So Paulo

ABSTRACT: Since the 1980s there has been a promising development worldwide towards more efficient lining techniques , where heavy cast concrete linings have been substituted by single shell, mesh or fibre reinforced shotcrete linings. However, under certain circumstances cast concrete linings including sealing membranes are still needed, for example under high water pressure, strict demands on sealing or a long service life. Some design aspects that are not very well resolved for such cases are related to durability, water infiltration, design calculations with respect to membranes, stress relaxation in concrete etc. A short survey of typical solutions is given for waterproofing for functional reasons and for water sealing to prevent water inflow, usually in order to maintain an existing groundwater level. Design considerations are briefly illustrated by references to the Channel tunnel crossover, the Hallandss railroad tunnel in Sweden and tunnels for the So Paulo ring road and others in soft rock. An important by-product of the above-mentioned criteria is related to lining durability. Chemically aggressive water or inflows of large amounts have resulted in, or raised concerns about concrete deterioration. An example is the tunnels for the Kanmon Strait crossing, where the concrete lining was damaged by sea water (Miyaguchi, 1986). In such cases, waterproofing has to be considered. For tunnels in stiff clay in So Paulo, excavated in the 80s with single shell lining consisting of mesh reinforced shotcrete, water ingress has varied between 0.5 and 5.5 l/min per100 m of single tunnel length. Those extreme figures occur for two parallel tunnels under water head in excess of 20 m (Celestino et al. 2001, Celestino 1997). The low permeability of the ground mass plays an important role in such cases. The statements above demonstrate that it is not always obvious what waterproofing means. All the above mentioned situations might be referred to as waterproof tunnels. The objective of this paper is to give some references, which illustrate this situation, and possibly clarify some aspects of waterproofing and sealing. We would prefer that waterproofing be used for the functional aspect, to prevent water from affecting the function of a tunnel, whereas water sealing be reserved for preventing water ingress, which may be set to specified limits.

1 INTRODUCTION Water sealing of tunnels is done for two main reasons, either functional or for preventing a lowering of the ground water level above the tunnel. In cases where leakage cannot be tolerated from the roof or the walls for functional reasons, a drained tunnel can be accepted as long as dripping or disturbing moisture can be avoided, i e an umbrella solution. What must be evaluated then, is the cost for drainage, which must be acceptable during construction as well as the permanent stage, where pumping costs as well as maintenance of the drainage system must be thoroughly considered. If maintaining the ground water level is the critical issue, during or after the construction stage, a strict limit on tolerable water ingress to the tunnel is often set in the specifications. An alternative would be to specify an acceptable lowering of the ground water table, but very few contracts are made that way. For shallow tunnels, cement grouting of the rock mass, in combination with a high standard sprayed concrete lining, may be sufficient to meet such requirements. In urban areas in Sweden infiltration rates are often set in the order of 2-5 l/min per 100 m of tunnel. For deeper tunnels with a ground water pressure of more than a few bars, or a water head of 10-30 m, a full sealing structure is often necessary. This implies that the sealing and supporting structure must be designed to stand full water pressure.

2 SHOTCRETE AS A SEALING MEASURE For waterproofing and water sealing, different solutions can be used. Conventional techniques include: shotcrete lining as primary support, a geotextile with a plastic membrane sheet, or a sprayed on material fo llowed by a cast, heavy concrete lining, (alternatively, precast concrete segments). These solutions are very expensive, and therefore cheaper alternatives have been looked for and tried. Two examples will be briefly referred to, which represent two different traditions as the basis for their presentations. The first one is discussing different ways to modify the conventional technique by using shotcrete to replace the cast concrete and including a sealing agent in a single shell solution. The second case is arguing that shotcrete in itself can be made water tight, provided that modern technology is used in mix design as well as during application. The two papers were presented at the ITA Congress Tunnels and Water in Madrid 1988, and illustrate the different views, based on local traditions in Germany and Norway. From STUVA, experiments were presented, where single shotcrete shells had been tested in different combinations with sealing membranes or agents (Shreyer 1988). The idea was to find a cheaper solution than the established practice with heavy concrete linings including welded plastic membranes. The proposed shotcrete shell included various reinforcement solutions, similar to what would have been used in a concrete lining. Hoses were installed for injection of epoxy or polyurethane in fractures, from the early stages of hardening. After a series of tests of different concepts it was stated that a single-shell lining made of shotcrete should initially only be used in zones with minimal groundwater pressure, up to 1 bar, and the shotcrete layer must not be regarded as watertight in any case. As a consequence, special seals must be used. The author concludes that the next decades investments in tunnels in Germany estimated at 15 billion DM would involve a potential of saving 1-2 billion DM if one half of the tunnels were constructed with single shell solutions. Thus, it was stated that the still unresolved issues for creating single-shell tunnels should be clarified as soon as possible. One decade later, Pttler & Klapperich (2001) still call attention to the importance of and the need for implementation of single shell tunnel projects. However, that type of lining was apparently adopted only for trial sections or adit tunnels of the high speed German Railway. They also showed examples

of economy achieved with the single-shell approach. Nine transportation tunnels constructed during the 80s according to that principle resulted in savings in the range of 10-15% with respect to the conventional double-shell approach. The authors go even further, saying that savings could be more significant with more flexible design philosophy, still not implemented in Germany at the time. This subject will be further discussed in item 6 of this paper. In great contrast, the other paper, presented by NOTEBY, Norway, aimed at demonstrating that modern shotcrete technology could meet the stricter demands on water tightness, which are now called for (Astad et al. 1988). Based on a general tradition to use shotcrete and bolts in fairly good rock, but even in weak or fractured zones, they meant that shotcrete could be used for waterproofing even for an undrained solution. With reference to scepticism in the construction community, they argued that seepage through shotcrete linings was due to the design, where watertightness had not been a distinct requirement. The paper presents basic guidelines for the manufacturing of watertight shotcrete, its poss ible applications as well as its limitations. To this end, a concrete mix is described which within good margins satisfies the specifications for a watertight concrete given in the Norwegian standards. A mix designed for a compressive strength of 35-40 MPa, including microsilica for low porosity, is recommended, a w/c ratio below 0.5 and steel fibres for stress distribution and crack-arresting effects. Curing is a crucial factor, to minimise shrinkage and improve the bond strength. Welldesigned and applied linings could result in a permeability of 10-12 m/s, which is considered as watertight for conventional concrete. The authors also comment on how to consider this lining as being one part of the waterproo fing. They again point at the importance of the bond, as it will prevent any pressure build up behind the shotcrete layer. Thus the water pressure is taken by the rock mass in interaction with the shotcrete, which supports and seals the rock su rface. For the sake of comparison, back analysis of the permeability of different lining systems based on real water ingress into tunnels led to shotcrete permeability values lower than 10-10 m/s in three cases, and in the range of 10-9 m/s in another case (Celestino et al. 2001). In all the cases, the tunnels were constructed in the 80s and many of the mix design characteristics advocated by Astad et al. (1988) had not been implemented yet. The value for in-situ permeability mentioned by them seems therefore feasible when so many technological advances have been incorporated.

Fourteen years after Tunnels and Water, one can state that specific demands on water sealing are more frequent and stricter in many cases. Shotcrete has been developed as a material, which is effectively produced with a well-controlled mix design, including steel fibres and additives to improve its properties for certain objectives. It is widely accepted and adopted as a useful structural element, not only for primary support, but as the final supporting element (Franzn 1992, Franzn et al. 2001). At the same time we find that we have not really seen the anticipated development discussed in the German and Norwegian papers. We have not seen a working sandwich structure or another successfully implemented combination of shotcrete and sealing agents with the capacity to replace a traditional water sealing technique for water pressures of any significant magnitude. Nor can we argue that shotcrete linings are water tight in the sense that they could be used for 100 percent water sealing. Shotcrete, even if well mixed and applied, will - like cast concrete - never be water tight in a strict sense but can form a useful barrier to water leakage if a minor amount of leakage or moisture can be tolerated and specified. However - a s mentioned before - under reasonable water pressures, grouting in combination with shotcrete can still meet a fairly tough requirement on sealing. 3 TYPICAL CASES A simplified structure of cases, which call for different solutions is illustrated in Figure 1. The referred cases above could be classified as types (b) and (c) respectively. Below, cases falling under (c) (d) are discussed. 4 TWO CASES WHICH NEED WATERPROOFING AND SEALING Two current examples are of interest for comparison. The Channel tunnel, 50-100 m below the sea, including a crossover chamber of substantial dimensions, and the 2x8 km long parallel rail tunnels, through the Hallandss ridge, with a ground water head of up to 150 m above the tunnel. 4.1 UK Crossover The crossover section on the UK side of the Channel tunnel is a shotcrete supported, membrane and concrete lined tunnel chamber. It is 165 m long, 22 m wide and 15 m high, with 35 m of rock cover and above that 35 m of seawater, an unlimited source of potential leakage. Most of the tunnel, including the crossover, is situated in chalk marl, a rock mass which is favourable for machine excavation as well

as for avoiding water leakage, Figure 2. The cavern was mechanically excavated from the service tunnel, driven by a 5.4 m TBM. The primary support was shotcrete, according to NATM principles, i. e. a comprehensive set of deformation measurements was used as a basis for the dimensioning of the

(a)

(b)

(c)

(d)

(a) Waterproofing for functional reasons Low groundwater pressure Umbrella by shotcrete or concrete elements Simple drainage (b) Watersealing for limited inflow of groundwater Low groundwater pressure Grouting and shotcrete lining Simple drainage (c) Watersealing for limited inflow of groundwater Moderate groundwater pressure Shotcrete, membrane, shotcrete or concrete lining Tentative dual drainage system (d) Water sealing for zero inflow to tunnel and no drawdown of groundwater table Membrane and concrete (or shotcrete) lining Full water head as design basis No drainage Figure 1. Waterproofing and sealing for typical situations

shotcrete lining. No certain measures had to be taken for sealing, as the rock structure was merely watertight, fractures typically 1-2 m apart and not water bearing. In this respect, the construction of the chamber was quite straightforward, whereas the challenge of the logistics and other aspects of this impressive engineering task should of course not be underestimated (Birch et al. 1991).

component of the design and to be relied upon as a water pressure relief system over a long time. However, we know that dam engineers have already learned what to do to rely on permanent drains for safety purposes. It could also be pointed out, when comparing with the next case, that for the crossover cavern, in principle no upper limit had to be set for an allowable ingress of water. Still, the design task was obvious. Even if there was minor leakage during construction, the final design must accept a tentative full water pressure as a most probable load case.

Figure 3. Load case 2, full water head, for the UK crossover cavern, from Hawley & Pttler 1991.

4.2 Hallandss railway tunnel


Figure 2. Geology and cross -section of the crossover cavern, from Fug eman et al. 1991.

The design calculations for the final support had to take into account different loading cases. Even if the rock structure was practically watertight in itself, the final lining of course included a sealing membrane, and full water pressure had to be one (worst) load case. This had to be checked even if it was not likely to develop, because of the non-waterproofed running tunnels nearby, cf Figure 3. Figure 2 shows the drainage pipes along the abutments of the arch. We have not investigated whether this drainage is meant to be a system, which can be checked and maintained. In any case, it is obvious that systems like this can be very difficult to design and construct if they are to work as a vital

The Hallandss tunnel is different in many respects. According to the current Water Act, the groundwater level is to be largely maintained, definitely in the permanent stage, but even to a certain level during construction. This requirement was transformed into a specification of maximum 33 l/s of water ingress for the whole tunnel, or a mean of 12 l/min per 100 m of each single tunnel. That is 2-3 times more than what is generally set for urban tunnels in Sweden tod a y , d e p e n d i n g o n l o c a l h y d rogeological conditions. Normally, this sealing requirement would have been met in most Scandinavian hard rock conditions, even when passing some heavily fractured or weathered zones. One would use cement grouting and possibly some chemical compounds like polyurethane in difficult parts. As is now well known, this technique was not successful in Hallandss, even after the

ground water had been considerably lowered (well below what was anticipated in the Water Act). The fracture pattern and the hydrogeological conditions proved to be a challenge for the client and contractor. After testing several techniques and grouting materials, they tested acrylic amide, which gave a very good sealing effect but was neither satisfactory for workers health and safety, nor with respect to pollution of the ground water. To make it short in this presentation: The northern parts of the currently excavated 1/3 of the tunnels have now been lined and sealed in the conventional way, i. e. shotcrete, geotextile, plastic membrane and an unreinforced, cast concrete lining, which is designed to withstand the full ground water head of 150 m, Figure 4. 4.3 Comments Based on different requirements, these two cases have been designed to withstand full water load on a watertight lining. For the final stage however, the difference is not only the greater load to be taken by the Hallandss lining, but also another specific requirement. This tunnel, floating in the groundwater, is not only to be designed to take full water pressure, in principle a trivial design task for a circular tube. The design must also guarantee that only a limited drainage will occur along the tunnel, towards the entrance parts of the tunnels or the outer slopes of the ridge. Any drainage pipes from the construction stage must be eventually closed. Only minor leakage can be accepted along the outside (bottom half) of the concrete tunnel, where a narrow slit of up to 10-20 mm will occur when the concrete shrinks during hardening, deforming and creeping as full pressure is being built up on its contour. Even this tentative drainage must fall within the stipulated allowed leakage of totally 33 l/s for the two tunnels. 5 HALLANDSS LINING DESIGN Late 2001 the decision has been taken by the Swedish government to finance the remaining excavation and completion of the Hallandss tunnel. Still all formalities as regards environmental issues are not solved and no contracts have yet been signed. A tentative restart of construction is estimated for next summer and the finalisation would take several years. The final design and construction methods for the remaining parts to be excavated have not yet been decided in detail, but most of the tunnels will have to be fully water sealed. Figure 4 shows a cross section of the currently lined parts, in total about 1200 m. The design is based on full water pressure. This means that the concrete ring will act under

compression and is therefore unreinforced, except the invert. This fact facilitated the installation and welding of the 4-mm thick polythene membrane, which is sensitive to mechanical damage. To avoid any longitudinal leakage along the tunnel, as a consequence of concrete shrinking, transverse stops have been installed every 100m. Thus, a swelling material has been applied between the shotcrete and the blinding, and on the inner side of the blinding a conventional waterstop is welded to the membrane. I n t h e p r e p a rations for continued works at Hallandss, a report has been prepared for the National Rail Administration, in order to compare different solutions used in Scandinavia and middle Europe (Steiner et al. 2000). The report states, as we do in this paper, that underlying philosophies can be quite different depending on geological conditions, environmental considerations and specifications, as well as local traditions in the tunnelling industry in different countries. Therefore different solutions are chosen. The report recommends that more detailed studies should be initiated, due to the complexity of the answer.

Figure 4. Cross-section of the lined parts of the Hallandss tunnel (Swedish National Rail Administration 1998). Drainage pipes are temporary, and have now been closed in order to let the groundwater head raise to its original level.

6 DESIGN CONSIDERATIONS It is important to note that different design philosophies have been adopted in different parts of the world, and some times by different agencies in the same country, leading to different concepts of safety and prices. Pttler & Klapperich (2001), for instance, present data about lining thickness, type of

ground mass, maximum overburden and water pressure above the roof for 11 tunnels constructed in the 80s in Germany, following the single shell principles. As mentioned before, savings with respect to double-shell concept was in the range of 10-15%, and the authors made the comment that savings could be even more significant with different design philosophy. Considerations about the load (other than permanent water pressure) on the second layer of the lining have varied, ranging from full load (complete deterioration of the first layer, therefore double-shell concept) to variable contribution of the first layer. In the cases presented by Pttler & Klapperich (2001), the total shell thickness for single track tunnels with maximum overburden in the range of 619 m varied from 25 to 55 cm. Water pressure above the roof was 1.2 bar for the case of maximum thickness (55 cm). The ground mass was described as solid marl. For comparison, the previously mentioned single track tunnels constructed in the same period for the So Paulo Subway used for the back analysis by Celestino et al. (2001) have 40 m overburden of stiff clay and compact sand, water pressure of 2 bars, and 25 cm total shotcrete thickness. The performance of these tunnels in terms of both hydraulic and structural aspects has been adequate, despite the high permeability of a sand layer intercepting the tunnel sections for more than one half their lengths. Certainly the difference in shotcrete thickness comes from differences in design concepts. Pttler & Klapperich (2001) suggest that the evaluation of the effectiveness of the second layer shou ld be made by bringing the safety philosophy of the Eurocode, introducing a reduction of strength parameters of the ground mass after installation of the second layer. That was exactly the procedure adopted for the design of the Brazilian tunnel, as is well documented by Domingues & Palermo (1985). Figure 5 shows an example of this design procedure presented by those authors for a double track tunnel with total shotcrete thickness of 40 cm, 25 cm in the first layer and 15 cm in the second. (Figure 5a). Ground mass strength envelopes are schematically indicated in Figure 5b. The short-term envelope, used for the construction period, is labelled for time = 0, whereas the long -term envelope, adopted after the second layer has been installed, is labelled for time . Bending moments and axial forces are shown in Figure 5c. Lines of maximum design internal forces (Md x Nd) are shown for total thickness h = 0.25 m (first layer only), h = 0.40 and 0.45 m (first layer and two options for the second layer of the shell). Internal forces for all the points of

the cross section at two locations of the tunnel for different situations are shown: at the portal (lower overburden) and central length (high overburden), for (1) the end of construction, (2) long term conditions, and (3) long term plus water pressure. It is clear from that diagram that the design internal forces at the end of construction are compatible with the 0.25 m thick first layer, whereas an addition of 0.15 m is enough for the long term load plus water pressure. The tunnel was constructed in 1984 with a 0.40 m thick shell, according to this design procedure.
Interface Graund Mass 1st Layer (0.25m) 2nd Layer(0.15m)

t= 0 t

(a) h = 0.45 200 Md (kN.m/m)

(b)

h = 0.40

100
E3

C3

E : Portal C : Central 1 : end of construction 2 : long term 3 : 2+ water pressure

h = 0.25
E2 C1

C2

E1 500 (c) 1000

Nd (kN/m) 1500

Figure 5. Definition of shotcrete lining thickness for a double track tunnel. From Domingues & Palermo 1985.

Among other cases of moderate water ingress under pressure in the range of 2 bars, the Ibirapuera Tunnels (urban highway) are worth mentioning. Since they are underneath the Ibirapuera Park in So Paulo, concerns were raised during design stages about the possibility of the water table being lowered to the extent of affecting trees and lawns. The tunnels were not provided with sealing membranes and have been operating for 5 years with no sign of ground water table lowering. Again, the low permeability of the stiff clay has played an important role. Previously mentioned solutions adopted in Sweden for moderate water pressure have also been adopted in Brazil. Castro et al. (2001) report the use of cement grouting the rock mass for the 200 m 2, 4-

lane tunnels for the So Paulo Ring Road currently under construction. The lining consists of mesh and steel fibre reinforced shotcrete and the water pressure is in the range of 1.5-2 bars. No sealing membrane has been used. 7 CONCLUSIONS Reading the papers referred to above, as well as personal communications with people involved in this kind of design issues, one can find several matters which are not fully understood. Therefore, solutions on the safe side have to be chosen. We would very much like to see more international exchange of ideas which govern specifications for waterproofing, as well as design concepts, which could lead to cheaper tunnel construction under complex hydrogeological conditions and in environmentally sensitive areas. ACKNOWLEDGEMENTS The assistance from Sweco, C-O Sder and Mats Burtu, and the permission from the Swedish National Rail Administration to use design documents for the Hallandss tunnels are gratefully acknowledged. REFERENCES

Escavaes Subterrneas, ABGE, Vol. 1, pp 180-194, Rio de Janeiro Franzn, T. 1992, Shotcrete for underground support A state of the art report with focus on steel fibre reinforcement, Proc. Rock Support in Mining and Underground Construction, Kaiser & McCreath (eds), Balkema Franzn, T., Garshol, K.G., Tomisawa, N., 2001 in press, Sprayed concrete for final linings: ITA working group report, Tunnelling and Underground Space Technology, Elsevier Fugeman, I. C. D. et al. 1991, The Channel tunnel: development of design and construction methods for the UK undersea cross over, Proc. Tunnelling 91, Institution of Mining and Metallurgy, London, Elsevier Hawley, J. Pttler, R. 1991, The Channel tunnel: numerical models used for design of the UK undersea crossover, Proc. Tunnelling 91, Institution of Mining and Metallurgy, London, Elsevier Pttler, R. & Klapperich, H., 2001, Singleshelled shotcrete linig aspects and application in Central Europe, Proc. Shotcrete for Underground Support VIII, ASCE, pp. 174-192 Schreyer, J. 1988, Sealing tunnels in single shell shotcrete, Proc. Tunnels and Water, Serrano (ed), Balkema Steiner, W., Malmtorp, J., Rosengren, L. 2000, Scandinavian and continental tunnelling practice. A comparison of solutions for waterproofing, Techn. Rep. BB 00:01, Swedish National Rail Administration Swedish National Rail Administration, 1998, Design criteria for the structural design of concrete lining in hard rock formations, Report by VBB anlggning for the National Rail Administration

Astad, U., Heimli, P., 1988, Waterproofing of tunnels by the use of shotcrete, Proc. Tunnels and Water, Serrano (ed), Balkema Birch et al. 1991, The Channel Tunnel: geotechnical aspects of the design and construction of the U K crossover. Proc. Tunnelling 91, Institution of Mining and Metallurgy, London, Elsevier Castro, G.R., Takahashi, J., DeMarco, L.A., Samara, V. & Kochen, R., 2001, So Paulo ring road tunnels (In Portuguese), Rev. Engenharia, 58, N. 543, pp. 70-73. Celestino, T.B. 1997, Long term monitoring and maintenance of tunnels, (In Portuguese), Proc. 2nd. Simp. Tneis Urbanos, ABGE, pp. 101-120, So Paulo Celestino, T.B., Giambastiani, M., & Bortolucci, A.A., 2001, Water inflows in tunnels: back-analysis and role of different linig systems, Proc. 2001 World Tunnel Congress, V.II, pp.547-554, Milano Domingues, L.C.D. & Palermo, G., 1985, Introducing safety for soft ground NATM tunnel design, Proc. 2nd. Simp. Bras. -

You might also like