• Geological prediction inaccuracy and delays

    Understanding geological investigation reports and applying them to the design and supply of the TBM was a central focus of the presentation by Lok Home about the experience gained on three of the most challenging TBM projects in the Robbins portfolio.

    Geologist Trevor Carter picked up the points and warns of the delusion that fancy geological computer models constitute an adequate substitute for real site investigation data and understanding. The delusion abounds, he states, to the extent that all too often wishful thinking geology results in poor TBM designs.

    Consultant David Fawcett, agrees that real site investigation and understanding is the fundamental basis of any tunnelling project with geological modelling used in the first instance to assist in specifying the initial site investigation.

    Consultant Nick Barton adds to the discussion identifying the deceleration of progress rates over tunnel length and the added facilities needed on TBMs to minimise delays to be anticipated due to encounters with adverse geological conditions.

    Read the full contributions from Carter and Barton on the Feedback page and at the bottom of the article page. Contribute to the discussions via the Feedback facility.


Learning from difficult rock TBM drive experiences 28 Jan 2021

Shani Wallis, TunnelTalk

From high energy rock bursting, to high volume, high pressure water inundations, to fault zones of flowing materials and squeezing ground, to rock of massive high strength and extreme abrasivity, all have been experienced on Robbins TBM projects and are at the centre of two clear messages of experience offered by Lok Home, President of The Robbins Company in his BTS British Tunnelling Society meeting lecture in January 2021.

The first message is that the geological investigation report, whether it be 100 pages or thousands of pages in detail, will never be right. A geological report can only be an interpretation he said, and the conditions actually encountered, compared to those predicted by the geological report, will have a degree of accuracy or inaccuracy. Therefore, his message two is to always include insurance in the TBM design and its capabilities. It is much easier, and without doubt cheaper, he explained, to include insurance features, such as extra thrust, extra power, extra ground conditioning and water control systems, and support installation equipment as part of the initial TBM order and its manufacture than to excavate bypasses around a trapped TBM or upgrade a machine once it is launched and well into its drive.

Geological profiles are interpretations and will never be right
Geological profiles are interpretations and will never be right

To illustrate the points, Home reviewed three current or recent long-distance rock TBM drives from the extensive Robbins portfolio of experience, that qualify them as among the most difficult of recent times.

Gone are the days, said Home, that the most difficult geological conditions are assigned to drill+blast. Today TBMs are being applied to projects of extreme conditions, “and that is okay,” he said. “TBM manufacturers have stepped up our game to design and manufacture TBMs that can deal with difficult conditions and to be now the safest and most practical method” providing, he warned, that the experiences of the contractor and the machine manufacturer are included in the project development at an early stage to prepare as best possible.

The first project of experience is the AMR irrigation project in India where two 10m diameter double-shield Robbins TBMs are working from each portal to complete a 43.5km long segmentally lined tunnel. The machines have been working on the drives since 2008 and are about 70% or 32.5km complete. When TunnelTalk visited the project in 2009, the greatest challenge was evident – extremely hard, massive and abrasive granite. The rock of up to 300 MPa in UCS with high quartz content has been punishing the TBMs for more than 10 years, causing the need for three main bearing replacements and wearing through cutters at an expensive and much higher rate than predicted. The time and cost estimates of the project significantly increased as a result.

Low advance with high disc consumption in the massive, abrasive granite of the AMR project
Low advance with high disc consumption in the massive, abrasive granite of the AMR project
Mud and water inflows that defeated the first machines on the Gerede water project in Turkey
Mud and water inflows that defeated the first machines on the Gerede water project in Turkey

The project has been on stop for the last two years with the contractor requiring more money to keep going and the State Government as owner, having no further funds available. Home is confident that the extra funding will be raised and that the tunnel will be finished.

Conditions encountered on the second project, the 31.6km long Gerede water transmission project in Turkey, were about 40% right and 60% wrong compared with the geological investigation report. Where the geological report and profile predicted 30 fault zones, Home said more like 148 were encountered, many with free-flowing mud and material in high water inrushes. Three non-pressurised TBMs from a different manufacturer were procured for the project. Robbins was called in to complete the last 11.7km when the three original machines came to a standstill in much more difficult geology than predicted. Coping with the conditions presented Robbins the opportunity to work with the contractor and the owner to develop its Crossover machine technology. The concept is a cross between an EPB pressurised operation and open TBM boring. In rock the cutterhead operates at 6-8 rev/min and at low torque. In EPB mode, with the same horsepower, torque is higher and the cutterhead rotation is lower at 2-3 rev/min. Shifting between the two modes uses the Robbins torque shifting system which is similar to the gear shifts in a 4-wheel drive vehicle.

Flooding in the Yin Han Ji Wei project, with temperatures in the tunnel reaching 36℃ with 90% humidity
Flooding in the Yin Han Ji Wei project, with temperatures in the tunnel reaching 36℃ with 90% humidity

The third of the three projects is the Yin Han Ji Wei water diversion project in China where an 8m diameter Robbins main beam gripper TBM was selected specifically to face the predictably difficult and predictably unpredictable conditions as the machine excavated the 11km alignment under an overburden of up to 2,000m through the Qinling mountains. The Robbins machine has experienced extreme conditions of high energy rock bursting, much higher than expected cutter consumption, and high-volume water ingress at high temperatures that flooded the down-gradient heading and took 75 days to recover. Home paid tribute to the Chinese engineers and workers on the project for being well experienced and thoroughly dedicated to progress the tunnel against all the odds.

In the question and answer period after the presentation, Home addressed specific questions about geological investigations and predictions, and the specifications for the design and manufacture of the machines for a TBM project.

Rather than more geological investigations in the planning stage, Home recommended to equipping the machine with features to cope with geological surprises. He is also a proponent of investigating the geology using modern probe drilling and system to record the geology as the heading advances. “What I would want the geologists to do is to be in the tunnel 24/7, seeing what comes in on those geology recording devices, interpreting it and understanding it. There are not enough geologists underground and not enough geologists that work with the TBM management crew to warn of conditions that they can see coming in the meters ahead.” A better method than predicting the geology ahead of contract award he said, is to develop in-tunnel geological investigation methods and with more in-tunnel requirements for mapping the geology as the heading advances to get it right, especially in long, deep tunnels with agreed complex geology.

McNally support system of steel elements extruded from pockets in the roof shield of an open-type TBM
McNally support system of steel elements extruded from pockets in the roof shield of an open-type TBM

Home added that, rather than try and predict the geology and put it all into a big contract, better to make the contracts flexible and able to deal with the geological conditions as they are identified at 50m in front. He admitted that there is a tendency to over state the worst expected conditions in contracts, which lead to expensive bids, but added that the owner should pay for his geology. Risk cannot be shifted all onto the contractor, he said, and the contractor cannot put it all on the TBM manufacturer.

He agreed that the concept of including insurance features on the TBM needs to be an all-party discussion and agreement and added that the contractual arrangement has to adapt to that. “The more complex the geology is to interpret, the more you should buy insurance.”

In referencing questions about what should be included in TBM specifications for particular projects, Home admitted that the contractors and manufacturers are protective of their knowhow. This is often given freely at the start of a project he said, and ends up, with competitor recommendations, in the project TBM specification. The request then, to the same manufacturers, is for the lowest and best price and longest payment terms. “There has to be a better way to do it,” he said, “to give your experience to the industry and to get properly paid for it on projects. There are ways of improving practice to get the machine right, get the geology right, and get the pricing right for the project. I have been concentrating on building better machines over the years, and not on creating better contracts, but as an industry we are not striving in a good way to achieve that improved practice goal.”

A Robbins rock TBM prepared for 360 degree pre-excavation grouting
A Robbins rock TBM prepared for 360 degree pre-excavation grouting

In answer to a question about who should specify the machine, Home said it should be the person at the table with the most experience to interpret the geology and understand what that geology means to machine design. “Without a big knowledge of both, you are not equipped to specify a machine,” he said, “and there are very few consultants in the industry that have a good depth of knowledge in both those disciplines.. I will let the consultants answer the question from there.”

In being asked how should the TBM manufacturer be selected, Home suggested that the machine partner should be selected up front and as part of the contractor’s tender and there should be no change after that. “For a contractor to get its bid right, it has to get its machine right, so there should be close partnership with the selected machine supplier prior to submitting the contract bid," he said. "If the contractor has shared all the geological data available and if he wants to be fair with the supplier, and the price for the machine has been agreed, there should be no changing the TBM supplier after that. If the machine is re-specified after the tender has been submitted it means the contractor has not done his homework and will have problems in his contract with the owner if the machine is not right for the job.”

In addition to the hardware features on a TBM to better cope with difficult conditions, Home said that most assuredly foams and polymers and other ground improving chemical products are part of the armoury for being prepared for coping with conditions as encountered.

So much more was presented and discussed in the presentation and a recording of the meeting and its question and answer session is available to view on the BTS YouTube Channel.


Delusional expectations for predictably difficult TBM drives

Feedback from: Dr Nick Barton

Dear TunnelTalk,

The lessons by Lok Home are nice to see. Such things, particularly geologically focused delays, have been the objective of the Qtbm model for the last 20 years, with, unlike many other methods, a specific expectation of a tunnel-length dependent utilization – because that is what the case records have been telling us for a long time. This is the case also with double-shield TBMs, though with greater efficiencies and less delays in general.

It has therefore been encouraging to see the gradually increasing number of pre-injection ports on shield TBMs in addition to improved probe-drilling facilities over recent decades, a lot led by Robbins and on Herrenknecht and other TBM manufacturers machines as well.

We have to be prepared for many eventualities. For example, the mean PR (penetration rate) for 36km of recent TBM excavation projects has been close to 2m/hr due to the hard and quite massive rock, while the mean AR (advance rate) taking all time into account, including pre-injection needs and as an end of project mean for a 36km total, has been 500mm/hr. This is due to deceleration with time over tunnel length (Figs 1 and 2). The declining lines are based on open-gripper case records(1), but the TBM world records (held mostly by Robbins TBMs) follow similar trends. It is recognised that a very efficient double-shield operation may have half the deceleration gradient of that of an open-gripper TBM.

Figs 1. Open-gripper case record trends(1)

Fig 2. TBM world record trends(2)

  1. Barton, N. 2000. TBM tunnelling in jointed and faulted rock. 173p. Balkema, Rotterdam.
  2. Barton, N. 2013. TBM prognoses for open-gripper and double-shield machines: challenges and solutions for weakness zones and water. Fjellsprengningsteknikk, Bergmekanikk, Geoteknikk, Oslo, 21.1-21.17, Nov. 2013.

Geological modelling to aid real site investigation studies

Article references

Learning from difficult rock TBM drive experiencesTunnelTalk January 2021
Editor's Desk comment TunnelTalk Alert, 28 January 2021

Geotechnical reports are an important part of TBM specification and success but are no substitute for real site investigation and understanding. Finding the sweet spot between geological modelling and understanding actual ground conditions and behaviour is the challenge.

Feedback from: David Fawcett

Dear TunnelTalk,

As a given, geotechnical reports should be the result of real site investigation and understanding. This is the fundamental basis of any tunnelling project.

With regard to geological modelling this should be used in the first instance to assist in specifying the initial site investigation. It is in no way any alternative to physical site investigation. Site investigation is an iterative process that should be designed by a suitably qualified and knowledgeable geotechnical engineer.

Determining the necessary detail to enable TBMs to be appropriately specified, and their likely behaviour in the ground to be understood, is an issue. Each type of TBM and each manufacturer will have differing parameters that are used to design machines. Each type of machine will have differing ground behaviour parameters that are important to its successful operation.

Thus, the whole process remains iterative and interactive until the machines have been designed and the operators have fully understood how to use them successfully.

There is no substitute for a knowledgeable geotechnical team leading, and remaining involved, in the site investigation and geotechnical reporting process from project conception through to successful tunnel completion.

Geological modelling can only be the first step in this complex process.

Geological modelling can however be used to assist in predicting likely and /or possible ground conditions in instances where physical site investigation is not practical, for example in high mountain environments. In such situations the modelling must carry a health warning and the chosen TBM must be versatile enough to overcome a much wider variety of ground conditions than when real information is available.

Best regards,
David Fawcett
Tunnelling Consultant
Past Chair of the BTS, British Tunnelling Society, 1995-1997


Feedback from: Dr Trevor Carter

It was excellent to hear Lok Home give such a frank presentation of ground control difficulties experienced in the three challenging Robbins rock TBM projects he discussed in the BTS meeting. As he stated, this sort of detail is seldom publicised even though many are aware of these types of situation not being uncommon.

The comments he made about prediction inaccuracy of technical bid documents (only 40% accurate in one case) resonate with a theme of increasingly delusional expectations that seem to be prevailing across the industry. The delusion is that fancy geological computer models are necessarily reliable, and worse still that they constitute an adequate substitute for real site investigation data and understanding. The delusion abounds to the extent that all too often wishful thinking geology results in poor TBM designs! Unfortunately, in many cases, when this is found out, it is months after the TBM has been built and when it is now deep under the mountains. Lok’s comments that equipping machines with gear for better forward probing and a need for more mapping in every type of tunnel are spot on, but these techniques are directed only at providing data to help negotiate the now identified problems, but after the fact. By this time the TBM is already built, starting maybe 18 months earlier, based on perhaps erroneous geology data and often based on a far too optimistic or unreliable GBR.

That is where I see the real problem lies – getting better geological and geotechnical understanding into early stages of projects, so that, rather than having to try to retrofit an already built machine to cope with ground conditions that might have been an expected possibility, the necessary capabilities are built into the machine design right from square one.

Fig 1. Searching for the sweet spot between how much needs to be known and site investigation investment(1)

Fig 1. Searching for the sweet spot between how much needs to be known and site investigation investment(1)

Yet again we heard this message from Lok, but this is part of the delusion and applies also to other major parts of a project. This is an old message that seems to be increasingly ignored. My perception is that with increasingly analytical sophistication in geotechnical engineering and enhanced 3D geological modelling and visualization as part of design interpretation, matters concerning understanding actual ground conditions and behaviour (which should be what controls the design and specification for the TBMs according to the ground prediction reports) is getting worse, not better. There is a worrying trend prevalent industry-wide of the increasing use of more and more modelling tools based on less and less real data. This trend towards generating pretty models (with miserable data) is leading to an increasing lack of real understanding. However, the parallel trend to production of GBRs (geotechnical baseline reports) on all projects, while applauded, must be tempered by the fact, as Lok also alluded to, that many are unrealistic, either biased and optimistic at one extreme or too risk averse at the other. Preparation of GBRs that lack a firm factual foundation that provides some bracketing to hard data are merely adding to a delusion of improved risk management, when in fact the reverse is the reality. Risk registers are only part of the answer. Education to counter the delusion that current state-of-practise leads to is the other part of the equation.

As an aid to addressing this worrying trend to delusion through visually impressive modelling, I have upgraded my 1992 risk diagram (Fig 1) to help address three key questions:

  1. How much understanding at any geological complexity level is needed;
  2. How much investigation to achieve that understanding is enough; and
  3. If a given amount of money is spent on geological investigation, can this realistically reduce the risks so well illustrated in Lok’s presentation.

The updated graph (which draws on more recent publications relating to better assessing geological reliability for tunnelling(2,3)), puts into perspective real underground risk versus perceived risk, with data benchmarked to real levels of understanding based on:

  • How much is a project willing to spend on site investigation;
  • How much is known about the complexity of the geology; and
  • The reality of diminishing returns of more extensive site investigation in more complex geology.

Obviously, there is a sweet spot on all projects for deciding how much needs to be known. After that, more expenditure is counterproductive, but below which, too little hard data is dangerous.

On the basis of pay now or pay later, up front expenditure is always better than the reality of downstream claims and remedial works costs. However, nowadays there seems even more reluctance to spend money on upfront site investigation. Hopefully, the graph in Fig 1 can prompt deeper thinking about this. Perhaps at least it can provide some guidelines that project planners can use to start to define how much needs to be spent to adequately de-risk a project, based on where they sit in the scale from simplicity to complexity in their project’s geology. Perhaps with this as a guide we can reduce the number of projects where there is less than 50% accuracy in the geological predictions – such as the three highlighted by Lok.

Dr Trevor Carter P.Eng., C.Eng., C.Geol.

  1. Carter TG (1992) Prediction and uncertainties in geological engineering and rock mass characterization assessment. In: Proceedings of the 4th Italian rock mechanics conference, Torino, pp 1.1–1.22
  2. Carter, T. G., & Marinos, V. (2020). Putting geological focus back into rock engineering design. Rock Mechanics and Rock Engineering, 53(10), 4487–4508. doi:10.1007/s00603-020-02177-1
  3. Venturini, G., Bianchi, G.W., and Diederichs, M. (2019). How to quantify the reliability of a geological and geotechnical reference model in underground projects. 2019 RETC Rapid Excavation and Tunneling Conference


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 :

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