Your browser is no longer supported

For the best possible experience using our website we recommend you upgrade to a newer version or another browser.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

Deep trouble

Report on the joint BGS/ICE Ground Board informal discussion 'Geotechnical aspects of shaft sinking' held at the Institution of Civil Engineers on 8 April, by John Cutler, Golder Associates.


Dr Bill Craig, chairman of the BGS, introduced the evening's speakers, Dr Chris Snee of Donaldson Associates and David Hartwell, groundwater management consultant. Dr Snee apologised on behalf of Dr Jin Chin Chern, of Sinotech Engineering Consultants, Taiwan, who was unable to attend due to unavoidable commitments in Taiwan. Dr Snee would present material on behalf of Dr Chern.

Snee outlined the presentation, which would include a review of some particular shaft sinking problems with the description of one current and two completed research projects, and would conclude with some considerations of groundwater in relation to shaft sinking.

Deep shafts and the geotechnical challenge

Chris Snee chose to concentrate on deep shafts in extreme ground conditions where the challenges tend to be unusual. Shallow and smaller diameter shafts, and those in soft ground would not be considered specifically.

For a shaft in a deep environment, the geological structure may be heterogeneous, and high anisotropic insitu stresses may exist. It is important to appreciate how significantly these ground conditions could vary from a less demanding environment. In particular, the implications for the design of temporary support and permanent linings can be significant.

Snee described how the Kirsch equations, while applicable to determining the stress distribution in two-dimensions around a circular opening in an elastic material, clearly are not applicable to situations where the media is heterogeneous, the insitu stresses are anisotropic, and the excavation shape is dictated mainly by functional requirements. Investigations for these environments often include in situ stress measurements and extensive geophysical testing.

Not all deep shafts suffer problems related to variable geology and stresses, and many are sunk to depths in excess of 2,500m in competent dry rock, without difficulty. However, there are shafts that suffer problems, and this is where the interest lies. Snee summarised items to be considered during the design of shafts in such environments (Table 1).

USBM - 1970s

During the 1970s, the United States Bureau of Mines undertook a study into ways to reduce the damage to shaft linings caused by excessive ground movements. Such damage resulted in significant costs in shaft repair and maintenance, and a consequent loss in production.

The investigation by the USBM included lining instrumentation in shafts from 3.5m to 7m diameter, and between 1,000m and 2,000m deep. In one instrumented shaft, the magnitude of tangential stress measured in the linings was seen to increase and rotate around the shaft over a period of months.

The research concluded that:

Simple stress analysis does not account for significant variations in tangential stress.

Yield zones are limited to a thickness of less than one shaft radius.

Linings should be maintained at least one shaft diameter from the shaft floor, at which point most of the elastic deformation has occurred.

Radial stresses on the linings can be highly anisotropic due to the geological and in situ stress conditions.

Shaft enlargements can change pressures on the lining.

Orientation of the principal stresses changes as the shaft floor advances.

Selby Coalfield - 1980s

The development of the superpit at Selby included one of the most detailed investigations into underground space ever undertaken in the UK, comprising five sites with twin shafts to depths of between 370m to 1,008m.

This investigation was aimed at improving the understanding of shaft construction, and in particular the design of linings. Instrumentation was designed to measure lining strains, rock pressures, water pressures and temperature changes. The geological sequence was the Permo-Triassic overlying the coal measures, and included weak rocks and potentially squeezing material.

The study concluded that, in competent rock, the loading on the concrete lining of a conventional drill and blast shaft was solely due to water pressure. Even at several hundred metres depth, rock loads were shown to be insignificant.

Two important findings were made during development around the shaft, where small scale geological features were present. During the construction of the 42m high Maltby bunker, at a depth of 720m, the lining was instrumented. The strain rate in the rings was observed to be related to the advance rate of the floor. Strains were eccentric, reaching 450 microstrain in some places, with the greatest strains occurring in the more deformable and discontinuous rocks.

From the investigations at this bunker, and those at the Daw Mill bunker at a depth of 600m, it was concluded that:

Significant benefits can be gained by delaying installation of the lining, to reduce the total strain and minimise cracking.

Eccentric loading of a lining can be due to geological anisotropy, and specifically highly discontinuous and deformable rocks.

Critical geological features included listric surfaces, fissility and micro faulting.

When the deformation of these materials has passed a critical point, the final loads on the permanent lining should only be due to groundwater pressure.

Although these conclusions are now well known, it should be noted that this investigation was concurrent with the development of NATM in this country. As a result of the fourth conclusion, sprayed concrete became accepted as an alternative to heavily reinforced concrete linings, and significant cost savings were made.

However, problems were still being encountered where changes in excavation geometry or direction occurred. Snee described how, at an inset in the North Selby shaft at a 980m depth, potential problems were avoided. Pre- reinforcement, comprising concentric cones of Macalloy bars, was installed in the roof of the inset, thereby reducing excessive strains and cracking of the lining, and the stiffness of the support for the adjacent excavations was progressively decreased away from the inset, which resulted in lower loads being imposed on the lining.

Taiwan - 1990s

The final part of Snee's presentation was a description of a current research project in Taiwan into the construction of deep shafts. Snee emphasised that the level of research at Selby up to the mid-1980s had been unprecedented, and has not been superseded in the UK since, although the tools that were used were very different from those that are in use on prestigious shallow shaft and tunnelling projects today. Technology has advanced significantly, but modern techniques are not adapted easily to the stronger, stiffer, discontinuous and anisotropic materials found at greater depths.

In 1994 Hencher presented a summary of advances in rock engineering made during the previous decade, summarised here in Table 2. The combination of research developments and advances in rock engineering contribute to the current understanding of deep shaft construction.

The current infrastructure construction programme in Taiwan is extensive, involving many kilometres of highways and numerous tunnels (Figure 1). Taiwan is young in geological terms, located on the tectonically active Pacific rim, and comprising meta-sedimentary rocks that have suffered extreme tectonic deformation. Construction of some of the shallower tunnels has encountered considerable difficulties. One example occurred during construction of the Mu-cha tunnels, where inward movements of the roof exceeded 1,200mm (Figure 2) where the cover was 160m. Re-mining of the tunnel was necessary.

Concern arose over the construction of the deeper tunnelled routes, such as the Central Cross Island Highway, and in particular the construction of the associated deep shafts, and this led to the collaboration between Sinotech and Snee.

Experience in Taiwan had shown that creep deformations, ie time dependent deformations under constant load, occurred in the soft and heavily sheared weak rocks, and in areas of high in situ stress. Consideration had to be given to this creep effect on the primary support and permanent lining for the tunnels and shafts.

Analytical tools are available to investigate creep effects, including well established theoretical and empirical solutions, with numerical solutions becoming more widely used. Practical support and lining design methods are essentially empirical. The current research programme encompasses the development of a numerical creep model, with validation of the model using physical modelling

in the laboratory, and calibration by back analysis using field data. Sinotech is carrying out development of the numerical creep model, with the use of FLAC. The model is based on Burger's creep model, and is able to describe elastic strain, and primary and secondary creep.

The physical modelling was undertaken at the University of Bradford. The testing comprised biaxial loading of specimens of a cement bentonite mix, which were 1m square by 200mm thick, and which contained a central circular hole 150mm in diameter. The cement bentonite mix (water:cement 2.5, bentonite: water 7%) had an unconfined compressive strength (UCS) of 1.1MPa and an elastic modulus of 300MPa. Extensive triaxial tests were carried out to determine its compliance with Burger's creep model.

The testing rig comprised a stiff frame, capable of loading four sides of the specimen via 20 hydraulic jacks, with instrumentation to measure displacements of the frame, loading plates and specimen. A plane stress configuration was selected to make observations and comparison with the numerical model easier.

Elastic theory shows that failure will occur in the shaft, that is to say the central hole in the specimen, at boundary stresses of 50% of the UCS. At low boundary stresses up to 25% of the UCS,

the convergence normally stabilised within about 30 minutes, and with no apparent creep deformation. At moderate boundary stresses of between 25% and 40% of the UCS, the characteristics of primary and secondary creep were observed. At approximately 50% of the UCS, cracks propagated from the shaft circumference, leading to failure of the block.

For validation of the numerical model, it is necessary to know the bulk modulus, shear modulus and viscosity of the material. These parameters were obtained by curve fitting of the laboratory data. There was a close correlation between the numerical model and the physical testing, and it was concluded that the numerical model is capable of simulating primary and secondary creep for soft rocks.

The design and construction of the shafts is yet to commence, but the current research has increased the confidence in the techniques that will be used. The conclusions drawn from the research programme to date are:

Back analysis of construction behaviour is required in determining the creep parameters of representative rock masses.

Reliance should not be placed solely on laboratory or field testing.

Groundwater control during,shaft sinking

David Hartwell described some aspects of groundwater that should be considered during shaft design and construction.

Is groundwater control required ?

Shafts below the water table yield water to some degree, ranging from situations where evaporation exceeds inflow, to conditions where inundation can occur. It is necessary to decide at what point control of inflows is required. Often this decision must be made during the feasibility or tendering stage of a project, and this can have a significant impact.

When considering the effect of inflows on shaft construction, it is useful to ask the question: 'At what point are inflows more than a nuisance?' When inflows exceed the nuisance level, control may

be required, and production, quality, morale and safety can be affected.

Are there surface implications?

Shaft construction in urban areas is common, with shafts sometimes being located in water supply aquifers. Consideration needs to be given to the impact of the shaft construction on the aquifer, and the impact at the surface of changes in the groundwater levels associated with the construction.

Where shafts are to be constructed in water supply aquifers, special consideration must be given to the impact on groundwater quality and supply, with recent examples being in the chalk in London, and the Sherwood Sandstone in Liverpool.

Shaft construction can affect groundwater levels to the extent that lowered water levels can cause surface settlements and damage surface structures. The extraction of relatively small quantities of water can have a significant impact at the surface, where drainage of weak deposits occurs, with recent examples occurring in Stockholm, Hong Kong and Jersey. However, in many urban areas ground-water levels have risen in recent years, due to surface development and reduced aquifer abstraction. Consequently, the potentially damaging effects of shaft construction may not arise, as water levels may remain above historically lower levels, such as during the MEPAS scheme in Liverpool and the Jubilee Line Extension in London.

Clearly it is important to have an answer to the question 'What is the allowable drawdown?' Increasingly, the answer given is zero, due to restrictions imposed by environmental and engineering considerations.

To assess the impact of shaft construction on groundwater levels, it is necessary to have an idea of the likely magnitude of groundwater inflows to a shaft, and for this, an understanding of the permeability of the ground is required.

Variations of permeability

Permeability is a measure of the hydraulic conductivity of uniform homogeneous materials, such as sands or gravels, which obey D'Arcy's Law. Variations in the homogeneity of a material can have a significant effect on its permeability, with one example being the presence of joints in a rock mass, and this can lead to misuse and misunderstanding of the term permeability.

In rock masses, often the hydraulic conductivity is measured in Lugeons, determined by pumping water into a section of borehole isolated by packers. Permeability values may be reported for a length of borehole, over which it has been assumed that the permeability is uniform. This is unlikely to be a valid assumption. In reality, flow is likely to have been restricted to discrete fissures.

Transmissivity, the product of permeability and aquifer thickness, may be more useful in representing the hydraulic conductivity of the ground. Pumping tests can be used to determine transmissivity, and at this stage flow zones can be identified using a geophysical flow logging tool.

Flow logs measure the velocity of the water travelling up the borehole. Changes in velocity can be used to identify inflow zones (Figure 3), and hence assist in understanding the variations in permeability.

Predicting shaft inflows

Despite the problems with determining the requirement for groundwater control, the impact of shaft construction at the surface, and a meaningful permeability value, Hartwell acknowledged that the most pressing question is often 'What will the shaft inflows be?'.

Hartwell's initial advice was 'Don't predict shaft inflows', but accepted that this was probably a little unhelpful to the practising engineer. However, it is a good starting point in helping to appreciate the risks involved.

A reasonable approach may be to calculate inflows by a number of methods, examine the limitations of those methods, and then make a judgement, based on experience, as to what inflows might be. Consideration should also be given to the use of flow net solutions, as another indicator of likely inflows.

Most flow prediction formulae derive from D'Arcy's law, where flow is the product of the permeability, the hydraulic gradient, and the cross- sectional area through which flow is occurring. Among the formulae to consider is Hvorlsev's formula, which is semi-empirical and used primarily to determine permeabilities based on flows to vertical boreholes, Dupuit's formula, which was developed for similar applications, and Sichardt's formula, which is empirical and based on maximum well inflow data.

Strictly, the first two formulae are applicable in laminar flow situations only, and therefore do not allow for the turbulent flows that are likely to occur adjacent to shafts. The latter formula, based on maximum well flows takes this factor into account but, actually, is dimensionally incorrect. Despite this, it is in common use.


Brian Skipp opened the discussion by describing the Monkton Hall bunker, constructed in the 1960s, which was one of the largest chambers at depth at that time. Skipp raised various points of interest relating to Snee's presentation. How much was the client prepared to pay to determine the in situ state of stress at depths of 1km to 2km? To what extent are theoretical models useful to engineers, and is there a cost benefit in determining the insitu state of stress, or is guessing better ?

Snee acknowledged the significance of cost in determining the in situ stresses, suggesting costs might be of the order of pounds100,000. Numerical models were proving capable of predicting deformations, such as those at the Mu-cha tunnels.

Alan Auld, of Alan Auld Associates, considered creep a critical issue, and in particular the amount of deformation that may occur by this mechanism. Auld considered that after characterising the material, reasonable deformation predictions could be made using hand calculations and/or standard software

such as FLAC.

Snee acknowledged that this was a reasonable approach, but given the size of the schemes in Taiwan, with ventilation shafts extending to 2,000m depth, considered that research was justified.

John Hislam, of Applied Geotechnical Engineering, agreed that when considering groundwater there was a large span between nuisance water and inundation.

Bob Irvin, of Gibb, described how techniques used to measure ground-water flows on the Nirex project had differed from those described by Hartwell. Low flows were being investigated, where temperature and conductivity profiles in a borehole identified inflow zones.

Bob Pine, of Golder Associates, considered that flow logs recorded permeabilities in the range relevant to shaft construction, and that temperature and conductivity logs could be beneficial where resolution was required, but considered that fracture network characterisation, such as that carried

out by Nirex, was not particularly relevant to more typical shaft projects.

Hartwell considered the combined flow log and temperature/conductivity log good value for money.

Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Please note comments made online may also be published in the print edition of New Civil Engineer. Links may be included in your comments but HTML is not permitted.