By popular perception, tunnel projects are accident-prone. This belief is not without supporting evidence from a few recent notorious instances of major ‘failed’ tunnelling projects. Apart from the rarer circumstance of physical collapse, a ‘failed’ project is also one that has suffered gross over-runs in time and money, however laudable the technology. A cynic has described a tunnel as ‘a long cylindrical hole through ground in a state of plane strain, with a geologist at one end and a group of lawyers at the other’; yet more dire is the present-day phenomenon of lawyers at each end! Sir Alan Muir Wood’s object here is to set such experiences into context, to explain the reasons for failure and to insist that success in the future needs to recover principles learned from success in the past – and the present – set into the complexities of modern tunnelling technology.
Special features of tunnelling
Are there special features of tunnelling which set it apart as a distinctive sub-species of civil engineering? Tunnelling shares characteristics with other forms of construction which are dependent upon the ground. The ground is more difficult to explore and characterise in fine detail at the depth of a tunnel. A particular feature of tunnelling lies in the interdependence between design and construction. The product, the project and its component parts, is inseparable from the process of construction. The intermediate phases of construction of a tunnel represent the period of greatest exposure to hazard. As a linear operation, tunnelling has special problems in access and logistics in overcoming unforeseen problems. The strategy for a project needs therefore to be fashioned in considerable detail before major resources are committed. While the account that follows is written specifically with tunnelling in mind, there are many parallels with those other forms of construction that depend significantly on the forces and processes of nature.
Above all, successful tunnelling depends on management of uncertainty of the ground and how it may affect a specific scheme of construction within a specific project. Sir Harold Harding, a founder Fellow of The Royal Academy of Engineering, set the issue pithily in remarking that in tunnelling, ‘the engineer should expect to be surprised but never astonished’. Success is marked by the avoidance of the embarrassment of astonishment and the ability to anticipate, in the full sense of the word, ‘surprise’.
Technological advances in all aspects of tunnel design, geotechnics, special techniques of construction, in purpose-made plant, in the range of available materials and tools, also the diversity of special expedients, all make their contribution to capability. The variety of choice also contributes to the complexity of engineering management. Success depends essentially upon the system whereby all these aspects may be combined as processes. Ignorance and neglect of the system are the main culprits for the egregious failures of tunnel projects.
The ‘system’: process management
All engineering enterprise – or simply all enterprise – depends on the management of a combination of processes as a system. Within living memory, tunnel construction was a traditional craft depending largely on the practical skill of artisans, who adapted the operation of tunnelling, itself comprising familiar elements, to variations of the ground. Exploration of the ground was fairly rudimentary, interpretation qualitative. In these circumstances, there might be a linear sequence of:
investigation of the ground
a simple basic design
followed by construction.
As each process became more complex and refined, and as tunnels were required to penetrate ground presenting greater technical difficulty, so have the interactions between the several processes become increasingly interactive and iterative.
Each project must have its impresario, the owner or client. The owner has initial objectives for the project, to be achieved through the processes of planning, design and construction. Operational objectives may be in part imperatives, in part desirable aims if found affordable. Operational objectives may be simple or complex. The performance of a tunnel as a conduit for water transfer, for example, could be expressed as a set of hydrodynamic time-dependent characteristics which, when combined with associated provisions for pumping, could enable the preferred scheme to be selected for minimum full-life costs. A transport tunnel, particularly where this forms part of a road or rail network, offers greater opportunity for different options, possibly making provisions for future measures to prolong the life of the project, increase its capacity or to achieve greater safety as demand increases.
Enlarging an existing tunnel is a costly and disruptive expedient; this is especially so in open water-bearing ground. Hence, Link (A) of Figure 1 provides for essential interaction, probably iterative, between ‘operation’ and ‘planning’. But planning cannot advance far without site investigation to reveal the characteristics of the ground, to provide the basis for how and where the tunnel is to be built. In consequence, additional iterative links operate as (B1), (B2) and (B3), attaching planning, studies (which include site investigation) and design of the works to be constructed. Design has then to be developed around preferred scheme(s) of construction, which may well entail the use of special plant and expedients whose specifics are dictated by the characteristics of the ground and the project. Link (C) in Figure 1 will, among other objectives, help to ensure that site investigation is directed to the special needs for designing the processes of construction. Link (D) completes the iterative circuit by ensuring linkage between construction and operation, whose requirements may well be developing while construction is taking place. Modifications may then be fed back into design and into the phasing and progress of construction. The system naturally incorporates many linked sub-systems to achieve optimisation within current levels of uncertainty. There will also be interactions between different processes of the system and external agencies concerning, for example, planning, the environment and safety.
Evolution of modern tunnelling
As the techniques of tunnelling have become more complex and the conditions more predictable, so have the benefits of purposeful continuity between aspects of operation, design and construction become more evident. Where, as in France for example, the tradition was for tunnelling to be largely undertaken for the State, bodies such as the Corps de Ponts et Chaussées provided an organisation familiar with operation, design and the principles of construction. Much of the ‘system’ could therefore be operated within a single body. In Britain, with a less centralised structure and with a late interest in engineering education, a different tradition evolved, with the engineer – for a significant tunnelling project, a consulting engineer – providing the dual function of translating the owner’s requirements into design and in directing the work of construction. This required the engineer to have an understanding and capability to bear much of the responsibilities represented by links (A), (B), (C) and (D) of Figure 1, with the most vital iterations of (B) proceeding within the engineer’s organisation. Thus, for example, as the scheme evolved, so could the site investigation be developed in detail and possibly extended to address the areas of uncertainty important to the choice of form of construction and the possible use of special expedients. ‘Special expedients’ are here to be understood to comprise particular measures for improving the characteristics of the ground. Temporary improvement might be by freezing or through the use of compressed air. Permanent improvement might use chemical injection or ground reinforcement. The success of the scheme depends therefore greatly upon the all-round competence of the engineer, including an ability to understand the owner’s interests as well as the existing and potential capabilities of construction techniques.
Figure 3 (right) The 10.3 metres internal diameter Cargo Tunnel was driven in London clay by shield beneath Runways 5 and 6 at Heathrow Airport in 1968, at a depth of 7–8 metres below the ground surface. The boltless 0.3 metres thick lining comprises 27 precision cast concrete segments, whose butting ends were shaped to allow rotation as each ring was expanded and stressed directly against the ground, by means of jacks inserted at axis level.
Figure 2 (above) The early railway tunnels in Britain were generally built in brickwork or masonry within the protection of timbering. Large sewer tunnels also used the method into the 1930s. Longitudinal timber bars, propped at the leading end and supported at the rear on the last length of brickwork, were boarded against the ground. The brickwork was advanced thus in lengths of 3 to 6 metres depending on the nature of the ground. The method may be seen as a forerunner of the tunnel shield, subsequently evolved in a diversity of tunnel boring machines, in sizes and types adapted to the particular demands of the project, the ground and external water pressure.
Figure 4 (right) The North Downs Tunnel of the Channel Tunnel Rail Link has an internal height of 10.7 metres and a width of 13 metres. It was advanced through chalk by top heading, two bench headings and excavation for the invert. The primary support was by sprayed concrete, rock dowels and spiles (bars driven into the ground), followed by a secondary lining of cast in-situ concrete. Detailed knowledge of the ground, which varied along the tunnel, and of the time-dependent properties of the concrete in primary and secondary lining allowed the geometry and details of the lining to be optimised. © Ros Orpin/Rail Link Engineering
Innovation in tunnelling has been a key to economy and safety. Every project is different, so in preparing the contract for construction of the project, the engineer would understand the nature of the risks most important to the project and how best to provide for their elimination or mitigation. Economy and efficiency would be served by freeing the contractor from the need to carry responsibility for incalculable risks beyond his control, by the use of such features as ‘reference conditions’ which set clear limits on physical features, usually based on characteristics of the ground, that fall within the contractor’s liabilities.
Trends in tunnelling have included exposure to more difficult ground, such as high pressures of ground water in sub-sea tunnels or the ‘squeezing rocks’ at depth beneath the Italian Apennines. More refined techniques of exploration, testing and geophysics play an increasingly important role in the design of a tunnel, choice of route and design of its construction. A common feature of today’s design techniques is that of exploiting the ability of the ground to support itself to the greatest possible degree, commensurate with safety and control of settlement. This capability can be demonstrated to allow major economies while adding to the structural security of the tunnel. Meanwhile, operational requirements have become more refined and subject to more rapid change. Modern tunnel construction techniques, fast and economic where the ground is favourable, are only tolerant within ranges of variations of the ground narrower than traditional methods. It is therefore apparent that understanding and operating the system become increasingly vital for successful tunnelling.
To illustrate the change from craft to technology, Figure 2 depicts construction of a late nineteenthcentury railway tunnel in clay; Figure 3 shows the Heathrow Cargo Tunnel of the 1960s using a boltless expanded concrete lining; and Figure 4 the Channel Tunnel Rail Link currently under construction, using sprayed concrete for initial support.
Learning from failure: principles for success
The traditional procedures were bound to change for the complex project as operational requirements began to become more intimately involved in the engineering of construction. Instead, however, of exploring the deficiencies of the practice, potentially good but fraying at the edges, the construction process, from around 1980, became generally fragmented. The civil engineering profession made inadequate representation as to the consequences of this disintegration, particularly evident in relation to tunnelling where the ‘system’ was so dominant for success. The changes were undoubtedly influenced by lawyers, mistakenly supposing that the transfer of risk to others would benefit their clients, several of which were newly privatised organisations with limited technical strengths. Control of project cost could not be achieved by controlling separately the cost of each isolated element. The engineer was appointed in competition in such a manner that the appointed engineer could no longer undertake essential synthesis, in the extreme being reduced to the title of ‘design contractor’. By the inappropriate attachment of risk to the contractor, the objectives of the ‘system’ were thwarted, since the terms of the contract predicated against variation in design or in co-operation between construction and design.
The secret of success in tunnelling is to recognise the ubiquity of uncertainty which attends work beneath, and sometimes a long way beneath, the surface of the ground. The nature of the uncertainty needs to be identified and a management strategy for its containment developed, specific to the project. The object is to avoid the progression:
uncertainty -> hazard -> risk
It may be possible, by investigation and analysis, to contain the uncertainty before construction begins. Often, however, this is impracticable. It is then that an observational approach is justified.
Suppose that a particular feature of suspected variability is a feature of design. Economy and safety may be achieved by an initial scheme of construction only adequate for a certain degree of variability, combined with a contingent plan of modification. The modification must be adequate to meet requirements revealed by observation of performance and capable of achievement in the time available. This is observational design, or by more specific definition the observational method. Observational design is a powerful tool in tunnelling. Successful application requires management and engineering fully coupled, continuity in project concept and an appropriate form of contract. The principle is illustrated by Figure 5.
The ‘ground model’ represents the conceptual characterisation of the ground relevant to features of tunnelling.
The optimal form of tunnel project management is specific to the particular project, the nature of uncertainty, the owner’s experience, the operational complexities, location and logistics. Continuity in conceptual development is essential, so that the system of Figure 1 can be respected. The present trend of successful tunnelling projects, such as the Øresund road and rail link between Denmark and Sweden and the Channel Tunnel rail link, have each addressed such issues as these deliberately and constructively:
The engineering of construction is closely linked to the engineering of operation.
The acceptability to external agencies is established ahead of construction.
Meticulous attention is given to problems at contractual interfaces.
There are rewards for innovative improvements to the scheme of work.
The development of relationships is comparable to partnering, with the team assembled on merit as the project demands.
Information relevant to construction is available transparently to all parties.
The basic elements of tunnel construction are simple: excavation and disposal of spoil, control of groundwater, support of the resulting cavity. For variety, there may be questions of prevention or control of consequential damage by ground settlement, entry of methane or problems of high temperature. A stalled tunnel is costly; an unpremeditated change of plan a major problem. The collective memory of tunnelling provides many examples of such problems prevented by foresight. What is essential remains an understanding of the ‘system’ and the principles vital to allow the linkages between the several processes to be operated effectively.
Uncertainty is a feature that is unavoidable in tunnelling. But it can be understood and controlled so that it does not cause damaging risk.
Sir Alan Muir Wood spent much of his professional career with Halcrow, retiring as Senior Partner in 1984 and remaining a Consultant to the Group. He has been associated with many major tunnelling projects, and is the author of books on tunnelling and coastal engineering. He was elected first President of the International Tunnelling Association in 1974 and subsequently Honorary Life President.