Dr Trevor G. Carter, Golder Associates, Toronto, Canada

In recent years increasing use has been made of TBMs for the excavation of deep rock tunnels. Dealing with adverse geology however can be problematic and can lead to significant delays and cost increases for either drill+blast or TBM approaches, when encountered at significant depth. Dr Trevor Carter argues that even after the tunnelling industry has executed hundreds of kilometres of tunnels and endured numerous collapses, inadequate forecasting of future conditions still emerges as a clear common thread for almost all projects that have suffered geotechnical difficulties and that mitigating delay problems requires considerable foresight and advanced planning.

From the tunnelling perspective, the Himalayas arguably pose the most challenging ground conditions almost anywhere in the world. One of the prime reasons for this is that they are the youngest of the mountain chains. They are demonstrably rising faster than anywhere else. Their composition is also younger generally, and in consequence less well consolidated than older fold belts.

Himalayas present formidable tunnelling challenges

This is consistent with the fact that they constitute one of the most active of the plate margin zones, rising at a rate that is almost double that of the Andes, which, in turn is almost three times that of the Alps. Almost nowhere else, on a world scale, except around the Pacific Ring of Fire, is even on the same active 'stress' scale. As in-situ stress levels are to a large extent geologic age-dependent, the younger the mountain belts the more imbalanced is the stress state.

As a result, stress conditions (magnitude and variation) can be potentially more extreme and adverse on a Himalayan tunnelling project than even has been encountered in some of the worst sub-mountain tunnel drives, including the Olmos and Yacambu Tunnels of Peru and Venezuela respectively, which are landmark projects from the bursting and squeezing perspective. On a ranking scale, these Andean tunnels traversed much worse ground conditions and arguably met greater geotechnical challenges than were encountered anywhere along the 50km+ length of the Lötschberg and St Gotthard deep Alpine railway tunnel drives in Switzerland. One can thus postulate a tunnelling difficulty ranking scale for the mountain chains of the world as:

1. The Himalayas, arguably the most difficult and challenging;
2. The Andes;
3. The Alps, through to the least difficult of the main chains;
4. The Rockies and the Western Cordillera, with 5, 6, 7 corresponding to older mountain cores with the Canadian and Scandinavian Archean, Algonquin and Adirondack age mountain belts being almost totally benign stress-wise.

TBM launches on deep rock tunnel at Olmos, Peru

This is not to say that there are not adverse faults and challenging zones of poor ground in these old mountain range areas. The dominant difference relates to stress state. Based on 'active' stress state alone, therefore, similar length deep tunnel excavations under the Himalayas likely will pose significantly more challenges than an equal length, equal cover drive almost anywhere else in the world. These difficulties of tunnelling at depth through high mountainous terrain pose major challenges not just for TBM drives but also for application of traditional drill+blast and NATM methods. Dealing with adverse geology at any depth can be problematic and can lead to significant tunnelling delays if not adequately foreseen; but geological problem conditions, which might be tractable at shallow depth, with either TBMs or drill+blast approaches, when encountered at significant depth (>1000m) can prove disastrous depending on stress state, rock competence and prevailing groundwater inflows.

Mitigating delay problems associated with exceptionally bad ground at depth requires considerable foresight and advanced planning. The more challenging the ground, the greater the pre-planning required prior.

The fact that within the Himalayas, conditions can be expected to be as bad as has ever been encountered elsewhere, means that there has to be the ability, while tunnelling, to allow changes to be made of driving method and support approaches. This need to adopt flexible solutions is often seen as being at variance with the constraints imposed by the rigidity of design elements incorporated into the fabrication of a typical TBM. As a result, traditionally there has been a reluctance to use machines in these conditions, mainly due to the perceived extremely adverse consequences of entrapping or damaging the TBM. In some part, this is due to the perception that there is more difficulty dealing with adverse ground conditions in the confined working area of a TBM heading, in comparison to dealing with the same problem in the larger working space of a drill+blast or NATM heading. Hard rock machine designs are, however, moving forward to encompass full umbrella forepoling and soft rock machine pre-grouting and ground treatment philosophies in an attempt to combat some of these problems by making the machines sufficiently robust and at the same time flexible enough to be capable of safely and successfully excavating through extremely bad ground.

Problematic geotechnical conditions
Two issues, essentially, control the ability to successfully improve tunnelling effectiveness for traversing through the characteristically complicated ground conditions found beneath the mountainous regions of the world. These are:

• The influence of adverse geotechnics, centred on the ability to deal with difficult ground conditions created by stress state, groundwater conditions and/or prevailing rock quality, and
• The limitations of current tunnelling technology ranging from drill + blast and NATM methods through to various TBM types.

Fig 1. A range of adverse conditions

Tunnelling in the Himalayas, the Andes, and until recently, the Alps, has shied away from TBM use due to perceived inflexibility, and the high likelihood of them getting trapped by adverse ground conditions, either as a result of squeezing or spalling/bursting conditions or because of ground collapses associated with rockfalls or with running or flowing ground within faults, always in these cases complicated by heavy water inflows (Fig 1). Any of these situations can lead to problematic tunnelling at best, and at worst collapses and abandonment. Dealing with such problems is always challenging, but is ten times worse when the tunnel heading is 10km from the nearest portal, as is the case in many TBM drives. The fact that such conditions pose almost as many challenges for conventional drill+blast and NATM methods as for a modern machine drive is often ignored.

Geotechnical considerations
Three main geotechnical elements control the ability to execute trouble free tunnels at significant depth. These are stress state, groundwater conditions and the rock itself.

Adverse characteristics of any one of these three elements can, on its own, compromise drill+blast or TBM tunnelling, but it usually takes a combination of all three being adverse to trap a machine or halt a drill+blast drive, to the extent that a bypass becomes necessary (Fig 2). We have methods to characterize each, at least at an overview level of assessment.

Methods for characterizing rockmass conditions for deep tunnels are briefly examined in a recent paper by Verman et al. (Verman, M., Carter, T.G., and Babendererde, L. TBM vs D&B - A Difficult Choice in Mountain Terrain - Some Geotechnical Guidelines; Proc 34th ISRM Congress, Beijing, 2011). While vast strides have been made with numerical analyses to better understand the behaviour of these difficult rock masses at the two ends of the rock competency scale, application of these methods as a predictive tool, rather than for back-analysis of existing or ongoing tunnelling conditions, is generally not justified at such an early stage, unless previous work has been done on the site or in similar materials, as typically current analytical and numerical assessment capability far outweighs early project ability to properly define appropriate input parameters.

Fig 2. Primary risk elements

Unfortunately it is always at an early stage in a project that decisions about excavation method are needed. Almost always, also, there is inadequate definition of stress state, rock competence and groundwater for most of the tunnel. Estimating conditions in the zones considered as most geologically problematic, therefore, becomes the focus for minimizing risk and maximizing objectivity for decision-making.

Decisions on whether or not to use a TBM for example, remains therefore a matter of judgment with the two most difficult questions being how much of the tunnel length is problematic? and how much of this problematic length is of critical concern?

Alone, no amount of analysis can yield the necessary answers. It requires a combination of information - yielded by the best possible geological assessment of likely conditions along a planned alignment coupled with application of numerical and analytical techniques to back-analyze similar conditions and assess applicability. Such analyses need to be credible, and done in sufficient detail that reasonable estimates can be made of critical yield extent and probable closure magnitudes. Only by such definition can difficult decisions be made on TBM applicability and the suitability of different TBM design types.

Risk reduction 1 - Geological appreciation
In mountainous terrain when considering a decision on whether or not to apply a TBM, and which type of TBM to use for a planned deep tunnel, it must be appreciated that historically, three types of ground conditions have proven the most problematic from the viewpoint of halting tunnel advance.

In order of severity, these three ground conditions are - (i) bad faults, (ii) heavy water ingress and (iii) major stress. Case records suggest that these conditions, either individually and/or in combination, constitute the most problematic ground conditions, almost irrespective of tunnelling method. For deep mountain tunnels, with very few exceptions, major disturbance zones associated with faulting have posed the most problems to tunnelling advance, often historically requiring bypass drifts and significant ground treatment before being able to be traversed.

The need is to look carefully not just at the basic geotechnics of deep tunnel alignments in terms of Q/RMR/GSI and other key rock mechanics parameters, but also at regional structural geology domains. In particular, three key geological factors need consideration over and above straightforward definition. These are: (i) rock mass quality, (ii) cover depth, as an estimate of stress magnitude, and (iii) groundwater conditions.

Although these three geotechnical control indicators give an initial clue to degree of adversity, alone they do not provide the extra insight needed to assess the possible degree of adversity posed by different types of faults likely to be encountered at depth along deep mountain tunnels.

Three additional factors need consideration. (i) the structural geological regime, (ii) the current regional tectonic state, and (iii) the likely palaeostress history. In mountain zones, understanding these factors can help route planning and alignment definition, as they provide clues to probable stress regime variability associated with specific styles of geological faulting.

Defining stress state conditions associated with major faults that cross a planned alignment does not need to wait until the tunnel reaches the fault. Use of palaeostress analysis techniques can allow some estimates to be made. If, based on such analyses, controlling stress magnitudes are anticipated to be very high, then consideration for de-stress blasting can be built into the contract provisions. In the Himalayas extreme ranges of conditions are now known to be common. This should also not be unexpected as this is a typical characteristic of young mountain areas and young faults, where mud rushes or running or flowing ground conditions can be a typical problem.

Dealing with such adverse geology at any depth can however be problematic and can lead to significant tunnelling delays, particularly if not adequately foreseen. The target for investigation is, therefore, attempting to unravel the structural geology and define the faults well ahead of meeting them in the heading and thereby ascertain their influence on stress state and rock mass conditions and competence, as a means for developing a strategy for tackling their crossing with the minimum of fuss.

Risk Reduction 2 - Improved decision-making
The lack of foresight of where adverse conditions can occur is central to many of the problems encountered in deep tunnel execution. It frequently clouds understanding to the extent that oftentimes errors and unnecessary uncertainties are introduced into the decision-making process related to drill+blast versus TBM selection, and even more so related to selection of machine type, if a machine option is favoured (Fig 3).

Fig 3. Decision tree for evaluating TBM operating mode, based on face stability issues.
Source: Source: Fruguglietti, A, Guglielmetti, V, Grasso, P, Carrieri, G, and Xu, S. Selection of the right TBM to excavate weathered rocks and soils. Proceedings of the World Tunnel Congress, Oslo, 1999

A further complication to the decision-making process relates to the timing when making this key decision, as it needs to be made some 12-18 months in advance of the actual start of tunnelling, to provide sufficient lead time for building the machine. However, all too often detailed project site investigations are either incomplete, still ongoing, or in some cases not even started when this key decision is required to be made. Furthermore, once the contract is finally awarded to the contractor, generally after a long and arduous tender process, always insufficient time and/or funds have been allocated to allow the contractor any opportunity for additional customized exploration to support his own excavation technology selection procedures before initiating equipment procurement.

In nearly all projects, the tender exploration data, which can be exceedingly variable in quality, is the only basis on which to make equipment selection. These added constraints create yet more levels of totally unnecessary risk to an already difficult decision.

Risk reduction 3 - Project risk
Unlike other civil geotechnical tunnel works, investigating the tunnel alignment of deep mountain tunnels is a challenge all of its own. In urban areas, tunnel investigations frequently end up with boreholes on 50m centres or closer along the alignment. This is impractical, if not cost prohibitive, for deep mountain tunnels. As a result, heavy reliance is placed on gaining, as best as possible, an appreciation of geological conditions at depth and along the alignment.

Fig 4. Influence of investigation on risk reduction and uncertainty

On the scale of typical project risk reduction (Fig 4 from T.G. Carter, Prediction and Uncertainties in Geological Engineering and Rock Mass Characterization Assessment. Proc. 4th Italian Rock Mechanics Conference, 1992) even for the most heavily investigated of deep tunnels (the Lötschberg and Gotthard Baseline tunnel alignments in Switzerland) only 20-30% understanding of what was finally known was available at the time when decisions were already having to be made about drill+blast versus TBM and with respect to TBM type selection. This possible 70% lack of understanding, arguably led to some of the problems that were ultimately encountered. But these, given the length of the tunnels, were quite minor as a percentage, affecting less than 5% of the length and only a small fraction of the total cost (A. Moergeli, Risk management in action - controlling difficult ground by innovation, RETC 2005).

Plotting on a conceptual graph (Fig 4) suggests that probably significantly greater than 50% risk reduction had been achieved by the site investigations.

The importance of focused site investigation cannot be over-emphasized as it is upon the data acquired from the investigation that the decision must be made between drill+blast or TBM excavation, and about what type of TBM to use, if a machine drive is selected. It is clear that excavation of deep rock tunnels poses several unique challenges which can be daunting, but all of which must be addressed as best possible when considering potential TBM applications.

Risk reduction 4 - Back analysis of case records
Many case records of difficult tunnelling through complex ground conditions in mountainous areas, particularly associated with faults and often with mud and debris inflows, have been reported down through the history of mountain tunnelling. Case records dating back to the turn of the previous century indicate the extensive use of bypass tunnels to navigate around the most difficult fault zones. In fact, review of many documented cases suggests that for a large percentage no forecasting was available before the collapse because the area was not investigated, often because cover was too great or access difficulties prevented drilling to the tunnel horizon or because of contractual arrangements. Turnkey and EPC (engineer-procure-construct) contracts, for example, have gained a notoriety in recent years for their lack of comprehensive investigation compared with most typical owner-administered contracts of earlier years, and before the advent of GBRs (Geotechnical Baseline Reports) and Project Specific Risk Registers as methods of managing geotechnical risk on tunnelling projects.

Sometimes direct causative information can be identified. One common factor is delayed progress through difficult ground. The Amsteg fault zone collapse and subsequent entrapment of the TBM on the west drive of the Gotthard Baseline Tunnel drives in Switzerland, while the TBM on the parallel east drive passed through the zone with only minimal difficulty, can be partially attributed to rate of advance issues.

Several recent TBM problems that have developed in bad ground in Himalayan tunnels have occurred also when production advance was delayed due to holidays and/or maintenance shutdowns.

It is an amazing fact though, that even after the tunnelling industry has executed hundreds of kilometres of tunnels and endured numerous collapses, the majority of which have led to signifcant delays and often serious cost implications for the projects, inadequate forecasting of future conditions still emerges as a clear common thread for almost all cases.

All too often, as indicated by the case records, an incomplete understanding of possible ground conditions and likely ground behaviour during tunnelling has proved detrimental and led to slow progress, costly and time consuming machine modifications underground, and in extreme cases, loss of a TBM or open-face heading. Embarking on better planned, more extensive investigations holds the key to removing much current uncertainty regarding TBM effectiveness and success of drill+blast headings in certain rock conditions along any particular alignment. Understanding rockmass behaviour and how different machines cope in different ground conditions is also key in stimulating innovative thought towards next generation improvements in machine design, excavation processes and ground support systems.

Download a pdf of the full technical paper

Crossing the Himalayas by rail - TunnelTalk, May 2012
Epoch making Gotthard Baseline railway - TunnelTalk, October 2010
Brenner Base Tunnel - let the works begin! - TunnelTalk, April 2011
Andes highway link a priority for Chile-Argentina-Brazil - TunnelTalk, December 2011
TBM excavation conquers Peruvian Andes - TunnelTalk, January 2012
Modern large diameter rock tunnels - TunnelTalk, April 2010
Collapse of Ethiopian headrace tunnel after grand opening - TunnelTalk, February 2010

Add your comment

Thank you for taking the time to share your thoughts and comments. You share in the wider tunnelling community, so please keep your comments smart and civil. Don't attack other readers personally, and keep your language professional.

In case of an error submitting Feedback, copy and send the text to Feedback@TunnelTalk.com

Name :

Date :

Email :

Phone No :

	Refresh
Enter the security code : No spaces, case-sensitive