The stimulating question from Rafik and the answers from Neil lead to the need for possible further discussion, because the strongly anisotropic nature of flysch (interbedded, cyclical sandstones and shales with marked strength and stiffness differences) represents a challenge to 'all' rock mass classification methods. It is not necessarily solved by a 'GSI sketch' of an inter-bedded rock (often folded) if the subsequent H-B equations produce an 'isotropic solution' in terms of strength and deformation parameters. If using Q, a strongly anisotropic RQD is needed(probably '0'/10 in one direction) and Jr/Ja may be strongly adverse, likewise SRF when weathered, as the weak shale is a significantly adverse feature. The key question is the most representative value to take for Jn the number of joint sets. Cross-jointing may give the clue, one way or the other.

Dear Rafik,

GSI is commonly mistaken as a rock-mass classification system. GSI is NOT a classification system as it cannot be applied for on the spot decision making where total strenght conditions must be considerd. GSI is a design tool, rather a σci reduction factor in order to simulate σcm in effective strength fully drained terms, and it should be treated as such. The applicability of either RMR or Q to flysch is difficult. From locally gained experience it seemed to me that RMR was working better with moderately to thinly bedded flysch materials whereas Q seemed to fit flysch with dominant thick bedded coarse grained facies. Having said so, I must warn you that this is just a localised experience. It could be that in your case my conclusions do not fit. Please do take very seriously the previous comments by the above internationally acclaimed experts.

Have a great festive season

George

I agree with Georgios. Q-system has been developed for hard igneous rocks of Scandinavian Mountains. Such a heterogenic formation as flysch needs a big experience to be evaluated by an engineer. See the comments given by Nick. RMR system is quite better or easier to apply.

There is a flysch rock mass in Poland which differs significantly in the different sites and even using GSI system for flysch is sometimes equivocal.

So Rafik - look into the last V. Marinos papers (2014, 2017) where the nice and clear guideline about GSI is shown. It will help, I guess. :)

Piotr!

You are making some serious mistakes concerning the development of the Q-system! The 212 case records dating from 1973 or before were 60% from Scandinavia, and both here and in all the other case records from outside this region there were a lot of different rock types (50 different types - please see some back-ground on Q so as not to spread, no doubt unintended, but nevertheless incorrect opinions). You might be interested to know that in the approx. 100km x 200km of land-area mostly to the south of today's Oslo - where the Q-system was developed, also using local non-hard-rock exposures as examples, there are 10 to 12 named collapsed caldera. Tunnel and cavern case records constructed in your assumed 'hard igneous rocks' would only have been of interest in shear zones and clay-bearing faulted 'hard' rock....we needed B+S(mr) cases, not 'no support needed'. The Q-system does a very good job of characterization, matching tunnelling observation very accurately, at least down to 4 to 9 MPa chalk marl, and of course must be pushed into (< 1 MPa) saprolites on occasion, where Q is usually

Thank You Nick for the answer. Of course,, I should have said "above all for hard igneous rocks". I know, that O-system have been tested also in soft rocks, but here flysch formation is the point. Strongly heterogenic and I don't know Q-system application for such rocks in Greece or Slovakia or Poland. If You have any experience or can recommend any papers - give me some hints, please.

I thank the peers for the rich discussions.

Indeed, the flysch is a delicate rock mass to parameterize, and this is demonstrated with each intervention.

The flysch is characterized by a significant persistence of the stratification joints, and an intense deformation (related to the movements associated with the orogenesis) and this point influences the behavior of the rock mass. The GSI takes this situation into account, especially with V. Marinos papers.

The Q-system takes into account the number of joints, but not directly their persistence. There is only the note related to the RQD that gives some indication.

The question that arises is that generally in deep grounds, the discontinuities are closed, and it is rather plans of weakness (the RQD can be high). In this case, it will be a gap between the Q-system and the GSI?

Most of the examples, even the pictures presented, relate to the surface ground where the joints are always open.

I do not recall a note about persistence and RQD in Q-characterization articles - please put me right as needed. But discontinuous jointing and eventual healed joints have specific suggested ratings of Jr = 4, and Ja = 0.75. For sure the dominant nature of the interbedded sandstone and shale

is likely to be as low as Jr/Ja = 1/4 if in the weathered zone. Significantly different at depth. There is no 'adverse' rating for bedded sedimentary rock in the Q-system - because of bedding persistence. Perhaps there is none because e.g. a 5 or 10 m span tunnel will have stability or instability and support needs that are not unduly affected whether bedding joints can be traced for only 50m instead of 100m - or much more. I am sure there are nice multiple-option sketches of flysch in Marinos's articles re GSI - but is the next step (if needed) going to be representative: seriously anisotropic deformability, shear strength, seismic velocity? See Barton and Quadros, 2015 regarding the anisotropic behaviour - of anisotropic rock masses.

Dear Rafik, Dr Barton is right. More over, GSI was in principle designed for homogenously fractured rock masses. Marinos and Hoek extension to flysch is based on extensive examples along the mountainous sections of Egnatia mototrway, which dissects the Greek part of the Alpine orogen. However I have not seen up to now similar papers (i.e. use of GSI on flysch) from other areas of the world. I'd be glad to know. I am sceptical about the general applicability of GSI for tunnelling design on flysch because of its frequently pronounced anisotropic characteristics. You have to be very careful with strong anisotropy. Dear Rafik let me remind you once again that GSI is not a rock mass classification system but a design tool for the application of H-B failure criterion. Good luck..

Dear Rafik returning to your initial question, I think it is necessary to make clear that there is not a more accurate geomechanical classification than another. Everything depends on the design and the geomechanical context in which the classification is to be used.

Regarding the specific case of galleries inside the flysh, we need to understand how we intend to proceed with the design. The direct use of the charts proposed by N. Barton or Z.Bieniawski for design is not advisable in these rocks.

The use of classifications to derive the geomechanical parameters to be introduced into numerical codes is possible, but the clay and arenaceous horizons must be characterized separately. In this case, in my experience, I suggest to take into account the fact that the deformability parameter of clayey horizons is strongly overestimated by both RMR and Q. It is also inadvisable to use parameterization using GSI unless you do not use with a finite element number code, but this causes the loss of discontinuities information.

I suggest obtaining the geomechanical parameters using laboratory tests and the formulas proposed by the classifications Q, RMR, GSI; then comparing the results and using a distinct element code for the project.

I hope to be proved helpful.

Dear Giovanni, as you know flysch is a mixed turbiditic sequence. During the construction of Egnatia motorway three different geotectonic sequences of flysch (3 different nappes) were encountered as well as post Alpine molassic strata. So the choice of tool for design requires engineering geological and tunnelling experience in order to set up a numerical model. The knowledge of materials, excellent quality of sampling etc are a must. Understanding and sellecting probable modes of failure in very important too. E.g. if the tunnel face is dominated by massive and highly disturbed fine grained shaley layers, it is likely that the tunnel could suffer from face stability and heave. If thick bedded sandstone or conglomeratic beds dominate then structurally controled failures are more likely. Indeed the intermadiate mixed facies are tricky to model aadditionally because of the preferred ground water paths. I would not right off GSI for flysch. There has been plenty of tunnelling experience incorporated in the Marinos Hoek charts. Simply one has to decide whether H-B criterion is applicable or not and what is the probably dominant mode of failure. This judgement is not a novice's job. Happy new year to all.

Has the ground been adequately explored?

What is the rock mass class that has been arrived at independantly by Q and RMR systems ?

Any permeability data is available?

Is it a theoretical exercise or a Construction project if so in what stage.

With my experience in Himalayan rocks I can state that, Heterogeneity in field does not matter as nature is heterogeneous.

I thank you all for these rich interventions.

It help me a lot.

Extensive analyzes must be performed in the flysch grounds.

A use of the three estimates of the rock masses quality (Q-system, RMR, GSI) allows a better understanding of the specificities of the grounds.

Each classification having strengths.

Thank you All.

Rafik.

Once again dear Rafik, GSI is a design tool. It is not a rock mass classification. Please understand the difference between the two, where RMR and Q have been primarily made for construction (on-site) decision making. GSI is a factor to obtain design parameters only and it is not related at all with stand up time and support classes as Q and RMR do. Of course Q and RMR have been used also for design purposes (extraction of rockmass strength parameters) under very particular restrictions. Many people involved in design only, without construction tunnelling experience make this mistake of calling GSI a classification. Please reconsider.

Dear Georgios,

I absolutely agree with you,

I think that to date, there is no study linking the GSI with tunnel construction solutions (someday there may be).

My comments concern rather the quality of the rock masses, and in this case, the GSI is a solution for the estimation of there quality. In this sense it is a kind of classification of the quality of the rock masses (very bad, bad, average, good, very good), without relation with the construction of the tunnels.

The deficiencies in this classification were well presented by Dr. Barton.

The RMR and Q-system classifications give indications on the solutions in the tunnels, but as their bases is the estimation of the quality of the rock mass, they have other applications as for the stability of the slopes, where even the foundations.

The relations that can be put between GSI, RMR and Q-system is to confirm the quality of the rock masses. I remind you that several equations relate these three parameters;

I think if we manage to confirm the quality of the rock masses, all the rest becomes easier.

Thank you very much.