Before the Baseline

Bill Newns · CPEng, MCSM, MSc, Dip Arb · NovoConsult New Zealand · 2026

The views expressed in this paper are the personal views of the author. They are intended to contribute to industry discussion and should be read as such.

What happens before the baseline should determine what baselines are required

Underground infrastructure is frequently undervalued in business cases. Tunnelling carries visible upfront risk and cost. The long-term benefits — capacity, reduced surface disruption, and structural durability under conditions that damage surface infrastructure — are harder to quantify and easy to discount. The post-earthquake record is instructive: underground structures generally outperform surface infrastructure in seismic events, though the margin depends on depth, ground conditions, and structural form — bored tunnels in rock behave differently from shallow cut-and-cover boxes in soft ground. At Kaikōura, rail tunnels damaged in the 2016 earthquake were used to move personnel and plant safely through the debris zone while the surface corridor remained impassable, and were returned to service well ahead of the surface infrastructure.^[4] That resilience — the ability to recover quickly and keep functioning while surface alternatives failed — was a function of physics, not foresight, and it is the kind of long-term argument that infrastructure business cases rarely quantify adequately.

Getting a major project off the ground requires optimism and determined advocacy. That is not a flaw — it is how important infrastructure gets built. However, objective analysis is needed to confirm that what seemed like a good idea actually is one, and that it represents good use of limited public funds. Infrastructure that serves multiple generations need not be fully funded from a single government balance sheet at day one; staged financing is a legitimate tool. The argument that private finance produces better risk pricing is not well supported by the evidence: the higher borrowing costs are real, the risk transfer is frequently illusory,^[3] and when projects fail the public balance sheet absorbs the loss regardless of the contractual structure. The business case should test the financing assumptions as rigorously as the demand ones.

The institutional response to optimism bias — gateway reviews, investment logic cases, better business case frameworks — has not consistently resolved the problem.^[12] The quality of any review depends on the independence and capability of the reviewer, and on whether there is genuine openness to the findings. A reviewer whose conclusions challenge established momentum risks marginalisation; one who validates it tends to be appointed again. The machinery follows the politics, not the evidence. Where underground infrastructure is built, the long-term value proposition is often vindicated. The problem is that construction cost outcomes in many jurisdictions have consistently exceeded business case projections by margins that call the appraisal assumptions into question^[1][2] — and that pattern erodes public and political confidence in the case for underground solutions.

The client organisation that commissions underground infrastructure is itself a variable. The special purpose vehicle client — formed once, for one project, with limited institutional memory of what went wrong on the last one — now carries a growing share of the world’s underground infrastructure risk. The institutional client — the rail authority, the water utility, the highway agency with decades of project experience — retains organisational knowledge the SPV model cannot replicate. As megaprojects become more common and SPV structures more prevalent, the gap between the knowledge available in the industry and the knowledge present in the client at the moment of procurement may be widening. The accountability is highest where the capability is lowest.

Principle one

Safety duties begin before procurement — and cannot be contracted away

Client decisions in planning and procurement are among the primary safety determinants of an underground project. In many jurisdictions this is not merely good practice — it is a legal obligation. Modern health and safety frameworks impose overlapping duties across the contracting chain: client organisations, managing contractors, designers, and subcontractors each carry concurrent obligations as persons conducting a business or undertaking (PCBUs), and those obligations do not extinguish each other. PCBUs must consult, cooperate with, and coordinate activities with all other PCBUs they share overlapping duties with. They cannot contract out of those duties — but they can enter reasonable agreements about how duties are met, provided each party continues to monitor compliance with what was agreed. The greater the influence and control a PCBU has over a work site or a health and safety matter, the more responsibility they carry. Decisions that preclude the elimination of risks can only be made in the early stages of the project lifecycle — and the client, as the party with the greatest influence over project definition, carries a corresponding weight of that responsibility.

Experience suggests that proper attention to the management of health and safety risks is indicative of a superior risk management culture within an organisation. A client who manages safety well is managing their project well. The converse is also true.

The procurement model is therefore a safety instrument, not only a commercial one. The choice of model determines who controls the design of ventilation systems, plant selection, dust suppression, work scheduling, and the occupational health provisions that govern the working environment. Under design and construct, design control transfers to the contractor. The regulatory duty does not.

The safety implications of procurement model choice are a duty consideration, not a preference. The client asking which model produces the best commercial outcome should be asking the same question about safety outcome.

Principle two

Who controls the detailed design controls most of the risks

Among the most consequential decisions a client makes is not which contract form to use. It is who controls the detailed design and when scope is defined — because that decision determines what risk allocation, GBR preparation, change management, and innovation can actually achieve. Everything downstream follows from it.

Owner-controlled design means the contractor is pricing a defined scope. Competition is on construction method and price rather than on who makes the most optimistic assumptions about what the design will require. Scope risk is substantially resolved before procurement begins. Ground risk is separately allocated — through the GBR and the contract conditions — against a design the client already controls.

Design and construct offers the broadest tender market and the fastest route to procurement. Anything not expressly captured in the Employer’s Requirements becomes a point of commercial contention as design develops. A rigorous reference design and explicit allowances for design development can manage that gap; where they are absent, alternative designs and innovative approaches — often the basis on which a contractor wins — can materially change the risk profile without changing the price. The interface between temporary and permanent works design is a further source of risk that Employer’s Requirements may not address: under D&C the contractor owns the construction method and the consequences of it, but many principals attempt to transfer the whole of the design risk at tender regardless of whether that transfer is achievable in practice. The interface between excavation sequence, temporary support, and permanent lining is where that gap most often surfaces. In thin markets the problem is compounded: experienced underground designers are scarce, and a contractor who cannot find genuinely independent design expertise is carrying a risk the procurement model assumed away.

Collaborative models — Alliance, CM at Risk, and variants — are most balanced under uncertainty, but carry the narrowest tender market and the highest governance requirement. Risk is owned collectively. The GBR becomes a shared reference rather than a contractual weapon. Adversarialism — itself a documented root cause of tunnel failure — is structurally reduced rather than managed by contract terms.

The prior question is not which model is best in principle. It is which model fits this project’s ground conditions, scope definability, client capability, designer pool, and legal jurisdiction. The industry asks this question less often than it should — and the cost of not asking it is paid in the claims, disputes, and losses that follow.

Principle three

Express allocation is superior to retrospective inference

The legal framework is a further determining variable in procurement model selection, not a footnote. What the contract says it transfers and what it actually transfers are not always the same thing — a distinction courts and tribunals have been willing to examine regardless of how the Employer’s Requirements are drafted. Alternative procurement models have not resolved this tension. They have relocated it.

Courts and arbitral tribunals allocating ground risk after the fact are working with incomplete information, adversarial framings, and the systematic distortions of hindsight. They are reconstructing, from silence and inference, what should have been agreed in advance. This process is expensive, slow, and rarely serves either party’s original interests.

Agreed risk allocation at contract formation produces more reliable value outcomes than retrospective allocation. The Privy Council made the point directly in Mitsui Construction Co Ltd v Attorney General of Hong Kong^[14] — a tunnel contract where the bills of quantities carried estimated lining quantities that proved wildly wrong, the project overran by 784 days, and the question before the court was whether a rate adjustment clause was triggered. The government relied on the schedule and a raft of other provisions to argue that the contractor had accepted full ground risk. The Privy Council rejected that reading: if the government were right, contractors tendering for the work would have to either gamble on ground conditions or price the worst case — and a responsible public authority could not have intended either. The risk allocation had not been made clear. The court was left to reconstruct from commercial sense what the contract should have said expressly.

The GBR is one instrument of express allocation. Its function is to define the ground conditions against which the contractor priced the work — it is the client’s statement of conditions, not a cap on the contractor’s foreseeability. The risk allocation principle that follows from this is that the client owns the ground risk up to the baseline: anything that is not baselined, or that falls outside what the GBR describes, remains with the client unless expressly allocated otherwise.^[6] Using the GBR to argue that unbaselined conditions were nevertheless foreseeable by the contractor inverts the instrument’s purpose and recreates precisely the inference problem express allocation is designed to prevent. The principle extends to other sources of subsurface uncertainty — utilities, contamination, existing structures — and to three distinct categories of uncertainty: natural variability in geological materials; knowledge uncertainty arising from the limits of investigation and interpretation; and ontological uncertainty — the semantic and interpretational gap where different parties attribute different meanings to the same terms.^[8] The first two categories are familiar. The third is the one that produces the most expensive disputes. Lawyers and judges are not well placed to resolve ontological uncertainty — they are working after the fact, with adversarial framings, in a domain requiring technical judgment that the contract should have provided. Express allocation is the only mechanism that addresses it in advance, before the lawyers and judges are asked to reconstruct who understood what and when.

Principle four

Ground risk can only be allocated — not eliminated

Contracts allocate risk. They do not change geology. Exculpatory language — the sentences that purport to transfer all risk to the contractor regardless of what the ground is or does — creates only the appearance of finality. Even when a contractor has accepted full ground risk, the commercial incentive is not to collaborate.^[16] In some circumstances full risk transfer does not eliminate ground risk for the client; it eliminates the reporting of it — or the independent assessment of it — until it is too late.^[12]

The Geotechnical Baseline Report is a well-established instrument for express risk allocation.^[7][11] When properly prepared and contractually binding, it replaces retrospective inference with agreed definition. The legal default — silence treated as foreseeability, the contractor deemed to have anticipated what the data did not show — is displaced by a reference condition both parties understood at contract formation.

A GBR does not predict ground conditions. It does not warrant them. It establishes the agreed basis upon which time and cost risk will be allocated when conditions differ from those described. That is a profoundly different function — and one the industry has not yet consistently applied.

It is axiomatic to tunnelling that a client organisation pays for a thorough ground investigation whether or not it does one.

Under-investment in site investigation is a primary contributor to tunnel losses. The investment required to adequately characterise ground conditions — typically one to three percent of project CAPEX^[5] — is recoverable in reduced claims, disputes, and commercial losses.^[6]

Ground risk extends beyond geotechnical conditions. The baselining principle applies wherever material uncertainty is capable of distorting procurement outcomes — wherever silence in the contract will later be treated as foreseeability. Utilities with incomplete or unreliable records present construction risk that is frequently underpriced at tender. Flood levels and climate design assumptions embedded at business case may not reflect the conditions the asset will operate in across its design life. Contamination from previous land use is frequently undisclosed or inadequately characterised. Artificial obstructions in brownfield environments carry similar uncertainty. The pattern recurs: the available information is incomplete, the contract allocates the uncertainty through silence or disclaimer, and the gap emerges as a claim when conditions differ from what was priced. Sometimes the client withholds what it knows — Transport for NSW withheld Ausgrid utility requirements from Acciona until after contract signing on the Sydney CBD and South East Light Rail project, a misrepresentation claim that settled for A$576 million.^[15] More often, nobody knows — and the contract fails to say so.

Principle five

The system should not depend on the individual

The quality of a project outcome — measured against objectives, scope and requirements — is a function of resources, know-how, information and communication, and motivation.^[16][17] Of these, motivation is the most powerful and the most neglected — and unlike the other terms it functions as an exponent: a negative motivation term does not merely subtract from the outcome, it scales the whole product downward. In an adversarial procurement environment, the motivation term is negative: the incentive structure actively works against the sharing of information about active risk. Full risk transfer degrades the information environment on which safe construction depends. You cannot collaborate with someone you do not trust, and adversarial contract forms are instruments of structured distrust.

A compliance culture alone is not sufficient. Project culture is a learned group behaviour — it is shaped by the procurement model, the contract form, and the behaviours that leadership rewards or tolerates.^[9] Regulations and codes are typically recreated after catastrophic failures, producing a system that is reactive, rigid, and forgetful. The same causes — inadequate ground investigation, fragmented responsibility, adversarial procurement, failure to act on monitoring warnings — recur across decades and jurisdictions.^[12] High Reliability Organisation principles from aviation and healthcare offer a more mature model: avoiding the oversimplification of root causes; deference to expertise over organisational rank; just culture over blame culture.^[10]

Most significant tunnel failures involve two things simultaneously: the ground did something unexpected, and someone made a poor decision. Human and organisational factors appear as consistently in the failure record as geological and mechanical ones.

Closing

“Re-establish good faith. Give the estimation of the work and do not refuse a reasonable payment to a contractor who will fulfil his obligations. That will always be the best transaction you will be able to find.”Vauban, 1683 — cited in CIRIA 79, Sir Alan Muir Wood, 1978

That observation is three hundred and forty years old. Vauban’s point was that selecting a contractor is a judgment about character and intent — about who will fulfil their obligations — not a comparison of prices. The industry is still circling around it.

Three centuries later the observation still holds. Successful underground infrastructure depends less on contract form than on institutional capability: an informed client, express allocation of ground risk, adequate site investigation, and procurement structures that encourage the flow of information rather than suppress it. Where those conditions exist, tunnelling projects routinely succeed. Where they do not, the failures that follow are institutional long before they become technical.

References

Flyvbjerg, B., Bruzelius, N. and Rothengatter, W. (2003).Megaprojects and Risk: An Anatomy of Ambition.Cambridge University Press, Cambridge.
Flyvbjerg, B., Holm, M.S. and Buhl, S. (2002). Underestimating costs in public works projects: error or lie?Journal of the American Planning Association,68(3), pp. 279–295.
National Audit Office (2018).PFI and PF2.HC 718, Session 2017–2019. National Audit Office, London.
Newns, W.R., Kotze, R., Sierra Ballen, R. and Walters, H. (2017). Design and construction of remedial works for rail tunnels impacted by the November 2016 Kaikōura earthquake.Proceedings of the 16th Australasian Tunnelling Conference,Sydney.
National Academy of Sciences (1984).Geotechnical Site Investigations for Underground Projects.Vol. 1. National Academy Press, Washington DC.
Newns, W.R. (ed.) (2026).A Guide for Geotechnical Baseline Reports in New Zealand.2nd ed. New Zealand Tunnelling Society, Wellington.
Essex, R.J. (ed.) (2022).Geotechnical Baseline Reports for Construction.2nd ed. American Society of Civil Engineers, Reston VA.
Newns, W.R. and Grasmick, J. (2025). Ground risk management using geostatistics: new ways to integrate risks from concept to completion.Proceedings of the World Tunnel Congress 2025,Stockholm. Paper ID 75943.
Stephenson, G.R. (1966). Cultural acquisition of a specific learned response among rhesus monkeys. In: Starek, D., Schneider, R. and Kuhn, H.J. (eds.)Progress in Primatology.Gustav Fischer Verlag, Stuttgart, pp. 279–288.
Weick, K.E. and Sutcliffe, K.M. (2015).Managing the Unexpected: Sustained Performance in a Complex World.3rd ed. Wiley, Hoboken NJ.
Gomes, A., Ertl, H., Marulanda, A., Neuenschwander, M. and Newns, W.R. (2025). ITA guidelines for the preparation of the GBR for the NEC4 Engineering and Construction Contracts — Options B and D with Bill of Quantities.Tunnelling and Underground Space Technology,Vol. 163.
International Tunnelling and Underground Space Association (ITA-AITES) and IMIA (2023).Code of Practice for Risk Management of Tunnel Works.3rd ed. ITA-AITES, Geneva.
Mitsui Construction Co Ltd v Attorney General of Hong Kong (1986) 33 BLR 1 (PC).
Ehrbar, H. (2025).Key Success Factors for the Design and Construction of Major Underground Constructions.Sir Alan Muir Wood Lecture, World Tunnel Congress 2025, Stockholm, May 2025. ITA-AITES.
Newns, W.R. (2025). ‘Principles, Systems and Methods to Minimise Failures and Losses in Tunnel Works.’ International Tunnelling Technology Special Lecture, Japanese Tunnelling Association 50th Anniversary, Tokyo, December 2025.
Transport for NSW v ALTRAC Light Rail Pty Ltd / Acciona Infrastructure Australia Pty Ltd. Supreme Court of New South Wales. Settled June 2019. Settlement amount A$576 million. See also: NSW Parliamentary Inquiry into the Sydney Light Rail Project (2018–2019).