See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/293064090

A critique of Q TBM

Article · June 2005

CITATIONS READS

4 770

2 authors, including:

Nick Ryland Barton

Nick Barton & Associates
309 PUBLICATIONS   12,397 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Tunnelling and Q-system application View project

Illustrated guide to the Q-system View project

All content following this page was uploaded by Nick Ryland Barton on 16 June 2017.

The user has requested enhancement of the downloaded file.

Comments on ’A critique of QTBM’ by Blindheim
Dr Nick Barton replies to the recent critique of the as an example, represent 3m above to 3m below
Qtbm model given by Blindheim in T&TI, June the (future) tunnel level. The ‘blocky rock’ problems
2005. He disputes many of the comments of occurred when relative block size (RQD/Jn) and
Blindheim, acknowledges some limitations of the inter-block shear strength (Jr /Ja) were both below
that predicted. This latent, and unexpected condition
QTBM prognosis model, and explains some
was a part of the basis for successful claims by the
improvements that are planned. He believes that contractor.
Blindheim has misunderstood several aspects of
TBM, and of the QTBM method.

As a developer of something new, it is inevitable

that one virtually ‘invites’ critique by those who
assume they are ‘the establishment’. Ideally both
parties, and the advancing subject matter, benefit
from the process. It is to be hoped that this will be
the case here.
In this reply to Blindheim will be found elements
of agreement, much dissent, and places where one
must assume that our opinions differ due to
experience of contrary behaviour.
In his introduction to the need for practical and
reliable TBM performance models, Blindheim claims
that I attempt to cover the ‘whole range’
(geological/geotechnical factors, machine factors,
organisational factors). While both the ‘geo-factors’
are squarely addressed in the QTBM method and in
Barton, 2000, there has (of course) been no attempt
to address organisational factors, unless he means I
should not address hours/week etc. Apart from
(average) cutter thrust, TBM diameter, utilization Fig 1 – A demonstration of reduced RQD/Jn ratios in a
(%) and support needs, other machine factors are TBM tunnel compared to those predicted. Barton,2004.
presently neglected. The important open or single-
shield, double-shield aspect will be discussed later.
Blindheim proceeds through the parameters
used in the QTBM model in a thorough manner,
mostly as translated from his 2002 article in
Norwegian. Unfortunately he starts by giving the first
version of the ‘QTBM’ formula in his Figure 1, from
an early chapter of my book, not the final version
nor that coded in the numerical QTBM -model.

RDQo/Jn (‘block size’)

Based on Palmström et al. 2002 and earlier
Palmström opinions, Blindheim claims that RQD ‘is
insensitive for high and low frequencies of joints’,
and ‘RQD/Jn poorly characterises the block size’. In
fact, RQD/Jn has remarkable sensitivity to situations
causing blocked buckets, blocked schutes, and Fig 2 – An exposure of the Brisbane NFB meta-sediments
clogged or damaged conveyors. In particular it is its that illustrates the importance of a combined RQD/Jn and
combination with the inter-block shear strength (i.e. Jr /Ja on block release. PR and AR may be affected in
Jr /Ja) that is important. opposite directions: easier boring, subsequent delays.
Blindheim appears not to fully accept that when
the Q-parameters are used both for PR prognosis, When promoting the idea of volumetric joint count
through QTBM, and for utilization and AR prognosis, (Jv = sum of number of joints/m for each set) in
through the gradient of deceleration (-)m, Q may place of RQD and Jn, both Blindheim and
act in (legitimately) ‘opposite’ directions. Fast PR for Palmström should realize that one cannot assume
short periods can be associated with a low that RQD ceases to have value when average
utilization and low actual advance rate AR, in the blocks are below 10-3 m3 (the 10 cm ‘limit’) or above
same tunnel length, due to the stop for support. about 0.4m3. The reality of the rock mass is a
Figure 1 shows the core-logged values of distribution of block sizes, which ‘stretches’ the use
RQD/Jn from two campaigns of site investigation in of RQD a long way into both low and high Jv. RQD
Brisbane in 1993-95, and 1998.The ‘+3/-3’ columns, does not cease to have value when the average
block size is about 10cm. At the other end of the
scale, is anyone but a dimension-stone-producer PR. Indeed, this has been especially noted following
interested if blocks are 100 or 1000 m3 in volume? the re-grip delay, declining towards the end of each
RQD may be no larger than 90% even when stroke.
average block size is several m3. It is both incorrect Blindheim is rightly concerned that an increased
and misleading to show a ‘short’ (and artificial) RQD SRF due to higher stresses may indicate increased
scale next to a ‘long’ Jv scale, and then say that stability problems. The QTBM method allows PR to
RQD is insensitive to block size, due to this increase due to stress-induced cracking, but it will
(artificial) comparison. first reduce due to the tangential stress on the face
(the σФ /5 ‘depth’ factor shown in Figure 4).
Jr /Ja (‘inter-block shear strength’)
For estimation of QTBM, it has been recommended
(Barton, 2000, p. 155) that the joint set (or
discontinuity) ‘most assisting (or hindering) cutter
penetration’ should be evaluated when assessing
an oriented Jr/Ja. Blindheim misquotes, and perhaps
has therefore misunderstood this aspect.
However, Blindheim makes a useful comment
concerning the comparison of a joint set with large
joint spacing, and a set with small spacing. In the
former, an orientation at an acute angle to the
tunnel axis may be favourable (as in Figure 3), while
in the latter a ‘blunt’ angle will be more favourable.
The difference has also to do with how long the
differently oriented joint sets intersect the tunnel.

Fig 4 – The QTBM formulation in complete form. The

given PR and AR curves refer to a worked example in
Barton, 2000. The maximum assumed hours for a day,
week and month are of course modified if available time is
reduced by different shift agreements.

If the cracking caused by the stress actually

causes the operator to reduce thrust, as suggested
by Blindheim, then there is a powerful increase in
Fig 3 – Classic demonstration from NTH 1994 of the the QTBM value (as seen by inspecting Figure 4),
potential effect of ‘single’ intersecting joints on the and the predicted PR reduces.
penetration rate.
SIGMA/(F10/209), (strength and thrust)
Jw /SRF (‘active stresses’) Thrust per cutter (F) as used in the QTBM
Application of the Q-system, and especially SRF estimate, is the obvious ratio of thrust available for
has never been something for the faint-hearted, and boring divided by the total number of cutters, when
Jw and SRF have been set to 1.0 by many users, only this information was reported. Blindheim is
occasionally by the undersigned too, but not once quite right that available torque is not at present
the real situation is understood, i.e the presence of included in the QTBM model, mainly due to lack of
faulting, or inadequate strength/stress in a hard rock readily available information on the separate effect
situation, or in relation to potential squeezing in a of cutter load and torque on PR, especially in
soft rock situation. Lugeon testing helps to judge Jw. weaker materials where friction across the cutter-
Blindheim makes the predictable comment head is of particular importance, and where release
that ‘Jw as an indicator of PR is highly suspect’. from fault zones is hoped for, without a by-pass.
When Jw is an extreme value (how often is this Blindheim seems confused by the fact that PR
correctly predicted ahead of time?) it is no doubt sometimes reduces as increased thrust is applied
correct to expect an unwanted effect on PR, and of (‘Barton…refers….in an unclear manner’). One of
course on AR, which is ‘correctly’ linked to the the several figures used in Chapter 6 of Barton,
reduced Q-value through a potentially steeper 2000 is reproduced here in Figure 5. It is a classic
deceleration gradient (-)m, and reduced utilization. demonstration from Nelson et al. 1983 of the
But for Jw values of 1, 0.66, 0.5, even 0.33 there consequences of insufficient thrust. Other examples
is little doubt in my mind that PR can be assisted by are also given in Barton, 2000.
the water, the water pressure, and by cutter cooling. A further example of this ‘phenomenon’ (from
The fact that AR may ‘soon’ be adversely affected is Chapter 12 of Barton, 2000) is shown in Figure 6,
a separate issue, and of course it may become the from the two 5.5km long Clermont water tunnels,
most important issue. Blindheim appears to not be where Blindheim was a co-author. He presumably
fully comprehending that ‘Q’ is legitimately used in introduced the NTH model available at the time, but
two different ways in ‘the QTBM method’. unfortunately it came up with a remarkably poorly-
Concerning SRF, Blindheim acknowledges that fitting ‘theoretical’ curve, as shown at the base of
high stress (in a massive, hard rock situation) can the figure. The reason for missing the ‘reduced PR
result in stress-induced cracking that allows higher with increasing cutter force’ is understandable, but
 Fig 7 – The inverse (F/mm) versus just the Q-value, at a
TBM tunnel in granites in Malaysia, with UCS (mean) of
182 MPa. Sundaram and Rafek, 1998.
Fig 5 – PR reduces with increased thrust due to the
increased % of 130 MPa limestone compared to shale of
‘operator usually reduces thrust’ (for QTBM < 1) could
roughly half the strength. Nelson et al. 1983.
be applied. Naturally, this is not meant as guidance
not the general misfit. Perhaps the NTH model was to an operator, who will never (poor fellow) have any
incorrectly applied, or that this NTH fracture class idea what the QTBM value is (unless equation 1 in
(F1) was not well adjusted for the positive effect of Figure 4 was evaluated in real-time, on his screen).
jointing on penetration. My insert on the curve in Figure 4, about the
The lack of a UCS/F ratio (or equivalent) in the operator, is rather an emphasis of one of the
NTH model, will tend to have been penalized by the uncertainties when making prognoses. The operator
very hard to extremely hard quartzitic sandstones, has experience-based ‘gut-feelings’ and may reduce
sandstones and siltstones, which reportedly had an thrust, which we have to respect and live with.
overall mean UCS of 170 MPa. There appears The machine parameter (F) does not have a
room, at least, for an alternative model, where rock place in the site classification system, as feared by
mechanics factors are given more weight. Blindheim, but afterwards, in a legitimate attempt to
make a TBM prognosis, when the factor F can be
varied, to help make a suitable choice of the
necessary range of F that should be available, in
relation to the rock mass strength estimate(s), given
by SIGMA. The QTBM numerical model helps here.
This process of thrust/strength comparison was
clearly inadequate (or absent) in the case of the
NTH-based Figure 6 performance prediction, and it
seems to be lacking in many TBM projects.
Consideration of F (a machine parameter), in the
QTBM calculation has nothing whatsoever to do with
‘obscuring comparisons of expected and
encountered conditions’, as Blindheim fears.
Fig 6 – Measured performance versus NTH model
prediction of penetration rate versus thrust. McKelvey et al.
Advance rate, utilization
1996.
Since 110 hours per week is not by any means a
universal rule for TBM tunnelling contractors, it was
Although Blindheim and his co-authors in
logical to use the maxima of 24, 168 and 720 hours
Palmström et al. 2002 have roundly criticised the Q-
as reference hours for a day, week and month. Of
system, and the extension of the Q-system into
course, other labour hours and numbers of shifts
other fields than D+B tunnel support and general
were sometimes the basis for reported utilizations,
rock mass classification, it has to be said that even
and these were not changed when assembling sets
a Q-value on its own, used by a dedicated Q-logger,
of reported data. There is no ‘new definition’ of
apparently does a good job in the case illustrated in
utilization, as Blindheim fears. Does he really think
Figure 7: a tunnel in variably weathered granites.
an intelligent person would treat national vacations
If logging had been by RMR, and attempts made
‘in the same manner as delays due to geological
to ‘retro-fit’ a Q-value and then a QTBM value, using
conditions and machine problems’?
guesses for several of the other parameters shown
in Figure 4, it is clear that ‘a beautiful shotgun plot in
‘m’ (‘gradient of deceleration’)
a logarithmic diagram’ could well be obtained when
Blindheim seems to struggle to comprehend the
plotting AR (rather than PR) against QTBM, Such
reasons for the different gradients of deceleration,
was recently reported from retro-analysis of three
which I would suggest, have not been clearly
tunnels in Italy, and used by Blindheim, in his clear
acknowledged in the tunnel-reporting literature. As
attempts to discredit Q and QTBM.
from a current difficult TBM project in Spain
Later in the same section concerning thrust and
(Abdalajis), we hear only of the best day (34m) in
(rock mass) strength, Blindheim wonders how
the ‘easiest’ section of the tunnel so far driven. The
 enormous torque, while surely necessary, is not more of such information, and attempt to provide
linked to the actual AR through geotechnically simple guidelines separating open TBM from
described rock. So one is not able to estimate (-)m. double-shield-push-off-the-liner TBM.
In massive rock, where Blindheim strangely A second edition of Barton, 2000 has been
expects highest utilization, he forgets to mention requested by the publisher, where several updates
cutter change delays, which is an important part of are planned, including a CD with the QTBM program.
the reason behind the input data used to estimate I will also attempt to reflect some important recent
(-)m, including the NTH/NTNU cutter life index CLI, advances in machine designs, such as the
quartz content and porosity. He suggests, shortened, and reducing diameter, double-shield
remarkably, that self-sharpening ‘thereby increases’ DSU-TBM described by Weber et al. in T&TI, June,
cutter life per m3 rock excavated. In fact too much 2005.
abrasive material flowing past the sides of the
sharpened cutter are paid for by reduced cutter life. Concluding remarks
This information, as far as I remember, is from his Blindheim clearly does not recommend use of the
old institute, NTH/NTNU. QTBM method, and the writer does not recommend
that Blindheim use the method either, with so much
Open and double shield machines negative pre-occupying him. Others are invited to
Although there is mention of the advantages of keep an open mind, realizing the complexity of what
double-shield machines, especially in Chapter 17 TBM prognosis-makers are dealing with. There is
concerning ‘Unexpected events and their Q-values’, room for several comprehensive models, even a
I was unable, in Barton 2000, to give general second one from Norway. The obvious miss in
guidelines concerning the reduced gradients of prediction in the case shown in Figure 6 is no
deceleration (-)m that would generally be expected reason that the model should not work well in other
when able to push off PC-element lining while re- circumstances. Probably the NTH/NTNU model has
setting the grippers. For obvious reasons, there was been altered since this time. I have visited, and
limited data on rock conditions encountered in such lectured, in this NTNU institute, and respect the
case records, with the notable exception of Figure 7 contribution they have made over many years. Their
from Grandori et al. 1995. However, the two bottom ‘CLI’ has a prominent place in QTBM, and is used
lines had to be added by the writer, in an attempt to twice.
interpret ‘RMC III’, ‘RMC IV’, etc. They may be
inaccurate, unless RMR was used as a basis for REFERENCES
‘RMC’. 1. N Barton, 2000. ‘TBM Tunnelling in Jointed and
Faulted Rock. Balkema, 173p.
2. A Palmström, OT Blindheim & E Broch, 2002.
‘The Q-system – possibilities and limitations’.
Bergmekanikkdag, Oslo (in Norwegian).
3. R Grandori, M Jaeger, F Antonini & L Vigl, 1995.
’Evinos-Mornos Tunnel - Greece. Construction
of a 30 km long hydraulic tunnel in less than
three years under the most adverse geological
conditions’. Proc. RETC. San Francisco.
4. J G McKelvey, E A Schultz, T A B Helin, & O T
Blindheim, 1996. ‘Geotechnical analysis in S.
Africa’. World Tunnelling, November 1996, 377-
390.
5. P Nelson, T D O'Rourke, & F H Kulhawy, 1983.
‘Factors affecting TBM penetration rates in
sedimentary rocks’. 24th US symposium on rock
mechanics, Texas A&M Univ.
6. NTH. 1994. ‘Full face tunnel boring’. [In
Norwegian: ’Fullprofilboring av tunneler’.]
Prosjektrapport, anleggsdrift 1-94. Trondheim:
NTH.
7. N M Sundaram, & A G Rafek, 1998. ‘The
Fig 8 – Average performance of four TBM from the Evinos- influence of rock mass properties in the
Mornos project, showing some general advantages of assessment of TBM performance’. Proc. 8th
double-shield machines in much of the poor conditions. IAEG congress, Vancouver.
However, note the ‘stand-stills’ caveat. Grandori et al. 8. W Weber, H Daoud & E Fernandez, 2005.
1995. ‘Challenging TBM tunnelling at Abdalajis’, T&TI,
May, 2005, 36-38.
Since the recent execution of some more
milestone projects such as the 28.7 km long
Guadarrama Tunnels in Spain, using four
‘competing’ double-shield TBM, it has been possible
to gather some updated data.
Early analyses of Guadarrama performances
suggest excellent (-)m values as low as (-)0.12,
despite the need for frequent cutter change and only
fair-to-poor PR in the hard, massive granites. Now
completed, and with data presented in a July 2005
seminar in Madrid, it will be possible to analyse

View publication stats