54 2 Pells
54 2 Pells
54 2 Pells
ABSTRACT
The classification system developed in 1978 by a sub-committee of the Australian Geomechnics Society in association
with guideline allowable bearing pressures on the Triassic sandstones and shales of the Sydney Basin has proven to be
useful as a communication tool on many and varied projects in Sydney and its surrounds. However, the classification
system has also been misused.
In this 2nd update of the original paper, the authors:
• indicate uses of the associated classification system, or components thereof, for design of prestressed ground
anchors, and for preliminary groundwater analyses,
• show that a modified system is appropriate for linkage to tunnel support design,
• document why the system should not be used in assessments of slope stability, excavatability and rock cutting,
and
• expand the guidelines for foundation bearing pressures and socket side shear values based on extensive research
testing, and field experience, gained in the 40 years since the original publication.
1 INTRODUCTION
From 1976 to 1978 a sub-committee of the Australian Geomechanics Society developed a classification system for the
Triassic-aged sandstones and shales of the Sydney Basin (primarily Hawkesbury Sandstone and Ashfield Shale) with the
purpose of providing design guidelines for foundations of major structures (Pells et al, 1978).
Over the following two decades it transpired that the classification system proved useful in communications between
various professionals, well beyond design and construction of foundations, sometimes appropriately but sometimes not.
By the mid-1990's substantial additional field and laboratory research work was done at Monash University and Sydney
University on design of foundations on rock and rock socketed piles. Therefore in 1998 an update of the classification
system was produced, with additional information on design parameters and foundation design methods (Pells et al, 998).
The author changes for the 1998 publication simply reflected the passing away of Don Douglas and retirement of others.
It is now 40 years since the original publication.
The original guidelines can still be applied for foundation design with cognisance of the settlement criterion which
controlled those guideline values (see Figure 1). However, four decades of experience and further research germane to
foundations on rock, allow a more detailed computation of allowable bearing pressures. Such computations are
appropriate for major structures.
This 40-year update addresses:
• appropriate and inappropriate use of the associated classification system for works other than foundation
design,
• computation methodology for determining foundation bearing pressures beyond the tabulated guideline values,
and
• design of prestressed ground anchors.
The author changes from the 1998 update reflect Bruce Walker’s retirement and incorporation of more widespread
experience.
Figure 1: Plot from Pells et al (1978) which showed that for available case studies at that time the recommended
allowable bearing pressures would be greater than measured values at settlement of 1% of minimum footing
dimension
2 HISTORY
2.1 THE 1978 PUBLICATION
The impetus for the 1978 publication was the status that allowable footing pressures were dictated by a Local Government
Ordinance 70, 1974 and earlier, which applied at least until 2002 under the guise of the NSW Building Code (see Table
1). The Ordinance 70 allowable values were considered by geotechnical specialists to be very conservative. Application
of the Ordinance was entirely controlled by Council inspectors. This process required use of the "infamous" spoon (a
length of reinforcing bar with a bent spoon-shaped end), which was used, very effectively, to scrape the sides of a
percussion hole drilled in the base of every footing so as to measure the locations and thicknesses of seams within a depth
of 1.5 times the footing width. It is, and should be, still used for such measurement.
The recommendations in the 1978 paper, and the associated classification system, were directed to vertically loaded
footings, hence the emphasis on proportion of 'clay seams' within the load influence zone; such seams are a particular
feature of Hawkesbury Sandstone, and formed a key part of Ordinance 70 (see Table 1).
The development of the 1978 paper was undertaken in conjunction with the Sydney City Council, and was formally
presented to Council by Dr Tony Phillips, on behalf of the Australian Geomechanics Society, on 24 December 1987.
That was critical to the change from footings being the province of Council Inspectors to being designed and inspected
by professional engineers.
Table 1: Portion of Table of Allowable Bearing Pressures as per Ordinance 70 of 1974 and NSW Building Code.
Allowable bearing
MATERIAL
pressure (kPa)
Soft shale 400
Weathered rock or medium shale 600
Shale rock at 600mm from the boundary 1000
Shale rock at the boundary 650
Soft sandstone, free from defects to a depth of 450mm and with a total seam thickness not exceeding 1300
20mm for the next 450mm of depth, where the footing is 900mm or more from the boundary
Soft sandstone, free from defects to a depth of 450mm and with a total seam thickness not exceeding 850
20mm for the next 450mm of depth, where the footing is at the boundary
Medium sandstone, free from defects to a depth of 600mm and with a total seam thickness not 2100
exceeding 20mm for the next 600mm of depth, where the footing is 1200mm or more from the
boundary
Medium sandstone, free from defects to a depth of 600mm and with a total seam thickness not 1400
exceeding 20mm for the next 600mm of depth, where the footing is at the boundary
Hard sandstone, free from defects to a depth of 900mm and with a total seam thickness not exceeding 3200
20mm for the next 900mm of depth, where the footing is 1800mm or more from the boundary
Hard sandstone, free from defects to a depth of 900mm and with a total seam thickness not exceeding 2100
20mm for the next 900mm of depth, where the footing is at the boundary
It is considered that the classification system should not be used for detailed evaluation of the excavatability of sandstones
and shales, whether by blasting, ripping, cutting or drilling. Such evaluations encompass rock substance and rock mass
characteristics, and machine characteristics, which are not captured in an appropriately quantitative way in the1978/1998
classification system.
In particular, no judgement should be made on drilling production and cutter wear for rock socket holes on the basis of
the classification used for design of foundations.
However, broadly, it is reasonable to conclude that:
• Ripping of Class I and Class II sandstone requires a very heavy bulldozer (Cat D10 or heavier) probably
equipped with an impact ripper, and is likely to produce a large volume of large blocks of sandstone, and
• Class V sandstone or Class IV and Class V shale can be excavated by a heavy excavator with a narrow bucket
equipped with rock teeth.
The classification system should not be used in evaluation or design of excavated slopes in sandstones and shales. The
level of safety of a cut slope in any rock mass is dictated by particular consideration of site specific geological structures
and groundwater pressures. These are not captured by the classification system.
Figure 2: Correct and incorrect classification (from Bertuzzi and Pells, 2002)
Now, 20 years later, the classification system has become adopted for all kinds of projects, including rock excavation,
anchored retention systems, pile boring, slope stability, and design of major tunnels and underground caverns. This has
seen application of the system well outside the original intent, and has seen engineering properties ‘attached’ to the
different classes of sandstone and shale without presentation of laboratory or field test data to justify such properties.
As set out in this paper, there exist a few situations where the 'foundation-based' classification system can be usefully
used for other kinds of structures. But there are also situations where this is inappropriate.
Uncritical adoption of the1978/1998 recommendations has led to situations, for example, where a site with small footings
800mm wide and large footings 4500mm wide, could with the recommended bearing pressures have settlements of about
8mm and 45mm; yet the structural serviceability requirements could be identical. Overlooking this aspect could result in
unacceptable differential settlement between two adjacent footings.
Figure 3: Bearing pressure versus displacement on Class I and II Sandstone (Pells and Turner, 1982)
Proper design requires explicit consideration of the allowable deformations of the super-incumbent structure, expressed
as serviceability settlements of all the footings. Bearing pressures (pserv) are then calculated from the assessed
serviceability settlements (δallow) using the mass modulus values given in Tables 3 and 4. These mass modulus values are
similar to those in the 1978 and 1998 papers, modified herein on the basis of additional field test data as summarised in
Table 5.
Note 1: Lower values for lowest strength, closest defect spacing or maximum seam % in each Class.
Note 2: Clean sockets of roughness category R2 or better (Walker and Pells, 1998).
The guideline bearing pressures in Tables 3 and 4 have been revised from those in Pells et al (1998) to make clear that
bearing pressure is lower in the lower range of each Class, independent of the UCS of the intact rock.
Serviceability bearing pressures (pserv) may be calculated from designated settlements (δallow) using the following
equations (from Poulos and Davis, 1974 and Whitman and Richart, 1967); similar equations for other geometries are
provided in Poulos and Davis. More sophisticated analyses using numerical methods may be appropriate in many design
cases.
Rigid circular footing (diameter D) pserv = 4 δallow/ (πD (1- 2))
Rigid Rectangular Footing (L by B) pserv = δallow/ (B (1- 2))
L/B 1 2 4 10
1.07 0.78 0.60 0.44
Note the expressions above are for footings on semi-infinite masses and thus may be conservative.
Settlements may be reduced where footings are recessed below the surface of the design-quality rock. However, caution
must be exercised when using settlement reduction factors from continuum elasticity theory (Pells, 1982) because tensile
fracturing of the sandstone, or the effects of bedding planes, are likely to make inappropriate the use of the theoretical
elasticity reduction factors.
For flexible footings appropriate equations are given in Poulos and Davis (1974).
Calculations of design bearing pressures using the modulus values in Tables 3 and 4 may give pressures greater than the
nominal recommended values for 1% settlement. This is not surprising given the cautious approach taken in 1978, as
discussed in the Introduction.
Calculation of probable settlements of socketed footings, invoking both side shear and end bearing, requires special
techniques such as given by Rowe and Armitage (1984), summarised in Pells (1999) and Randolph (2003).
The validity of computed settlements is entirely dependent on the adopted mass modulus. If a designer proposes bearing
pressures substantially greater than the revised guidelines in Tables 3 and 4 then it behoves that designer to confirm the
adopted design moduli by site specific investigation and verification during construction. This would include spoon
testing of all footings.
The second columns in Tables 3 and 4 give an indication of the stress level to which bearing pressure versus settlement
behaviour is approximately linear. This information is only to indicate the stress levels to which it is reasonable to
calculate settlements using equations from the theory of elasticity and does not imply that it is appropriate to adopt bearing
pressures many times greater that the substance unconfined compressive strength of the rock
In simple cases where all footing are approximately the same size, and the 1% criterion is acceptable, the revised values
in Tables 3 and 4 are appropriate.
4.2 FOOTINGS ADJACENT TO CUT FACES
The ultimate end bearing pressure of a footing on a laterally extensive rock mass relies upon mobilising a large volume
of rock mass in forming a failure mechanism and is many times the unconfined compressive strength of the rock forming
the mass.
There is an unconfined stress state along the face of an excavation, and thus a footing at the crest of an excavation is
loading unconfined rock. A narrow footing at the crest of a high excavation has an ultimate bearing capacity, ignoring
adversely oriented defects, approximately equal to the unconfined compressive strength of the rock mass. As the
dimension of the footing normal to the excavation increases the remote edge approaches the geometry of a footing on a
semi-infinite mass.
The unconfined compressive strength of the rock mass must allow for the possible existence of kinematically admissible
wedges formed by adversely oriented defects daylighting, or nearly daylighting in the cut face. Usually these can only
be identified with any confidence by inspection and/or probing of the excavation face during construction. If such defects
exist the ultimate bearing pressure may be effectively zero and the design would need revision, or measures must be taken
to reinforce the cut face as was done in the Eastern Distributor (Pells, 2001).
In addition, a footing near a cut face will settle more than indicated in Section 4.1 above and, further, will translate and
rotate under vertical centric loading compared with a centrically loaded footing on a laterally extensive surface.
For relatively simple geometries, the following guidelines, derived from generic 2D FEM analysis undertaken by the
authors, may be useful:
• The vertical settlement at serviceability loads is approximately twice that for a footing on a laterally extensive
rock mass and thus pserv is approximately half that given in Section 4.1.
• The lateral movement towards the excavation due to the vertical load is approximately half the vertical
settlement (thus coincidently approximately the vertical settlement given in Section 4.1). This is independent
of the lateral movement associated with in situ stress relief on excavation (Pells, 1990) which should also be
considered by the designer.
• The rotation of the footing into the excavation is approximately 0.5 to 1.0 times p/E.
Alternatively the designer should analyse the specific case.
Footings for existing buildings at the crest of a new excavation will already have settled under their actual working loads
and may be expected to undergo some additional settlement due to excavation. The major risk in this situation is either
bearing failure if the loads approach the unconfined compressive strength of the rock or kinematic failure if adversely
oriented defects daylight in the excavation face.
It is recommended that a designer specify adequate inspections to ensure adversely oriented defects are discovered if they
exist and/or revise the design where appropriate.
5 PRESTRESSED ANCHORS
5.1 INTRODUCTION
We deal here only with that facet of ground anchor design germane to assessing the anchor bond length from consideration
of shear strength between grout and rock.
Ultimate average side shear resistance is primarily a function of the rock substance unconfined compressive strength and,
secondly, lateral confinement (see Williams and Pells, 1981). Therefore side shear resistance for piles or anchors is
dominated by only one of the three factors in the classification system detailed above. Therefore bond shear resistance
cannot be directly related to the Class of sandstone or shale.
Given that the key parameter for anchor design, substance UCS, is captured as part of the classification system it is
obvious that guidelines can be formulated for strand or bar anchors with conventional cement-grout bond zones, of the
kind illustrated in Figure 4.
.
It is assumed that anchor holes are drilled with some form of percussion equipment giving a sidewall roughness of R2 or
better (see Pells et al 1978 for definitions of roughness). It is also assumed that anchor drill holes are cleaned of crushed
and smeared rock on the sidewalls of the bond zone by flushing with water or compressed air and water. In this regard it
is noted that the New South Wales Roads and Maritime Services Specification B114 for ground anchors includes the
following statement:
"Cleaning of boreholes is a Critical Anchor Activity that must be carried out under the supervision of the Anchor
Supervisor.
On completion of drilling or reaming, clean and then seal the boreholes to prevent contamination. Carry out
cleaning by flushing the borehole at least three times with water and air until the emerging water is clear, to
remove all smearing and drill cuttings from the borehole walls and bottom."
Anchors which have plates, mechanical devices, large nuts or under-reams at the distal end (furthest from the rock face)
behave differently compared with the simple conventional grouted strand anchor shown in Figure 4, and are not dealt
with here. Barley (1988) showed that under-reams have no significant benefit in rock with UCS greater than about 5MPa.
5.2 CODE REQUIREMENTS
Standards such as AS5100.3:2017 (Bridge design) and AS4678-2002 (Earth-retaining structures) require prestressed
anchors to be designed using Ultimate Limit State procedures such that:
Table 6: Contradictory importance category reduction factors; ( n from AS5100.3 and k from AS4678)
Minimum proof load/working load
Category n ( k)
(AS 4678)
Temporary with up to 6 months
1.0 (0.9) 110%
service life
Temporary with service life to 6
0.93 (0.85) 125%
years
Permanent and temporary where
0.7 (0.8) 150%
consequences of failure are serious1.
(1) The word 'serious' is not explicitly defined, but may be intended as "significant damage or risk to life"
(Table 1.1 AS4678), where "significant" is not defined!
The two codes are also contradictory in respect to g (geotechnical strength reduction factor). AS4678 provides a blanket
value of 0.7, whereas in AS 5100.3, depending on the method used to assess the ultimate bond, the value ranges between:
Figure 5a: Depiction of progressive failure (after Post Tensioning Institute, USA, PTI DC35.1-14)
Figure 5b: Finite element analysis of progressive failure of 2m anchor in Class I/II Sandstone; peak bond
strength 800kPa, residual bond strength 100kPa
The above considerations, coupled with the paucity of documented ultimate values as set out in Section 5.4, below, means
it is logically inappropriate to adopt the philosophy embedded in Codes such as AS5100 or Eurocode 7 (EN 1997-1)
which depend fundamentally on an a priori knowledge of ultimate bond strength. As a matter of logic, the Code
requirements are absurd.
This view is supported in a paper by Brady and McMahon (1997) which discusses Limit State Design of anchors. Two
pertinent comments from that paper are:
“the strength of the grout/ground interface....is the principal source of uncertainty in the design of ground
anchorages”
and:
"The role of proof testing in a partial factor of safety approach to design should be reviewed; having set a
proof load it may be unnecessary, in many cases, for the designer of the anchored structure, or the client, to
insist on minimum factors of safety against pullout, particularly where the margin of safety over and above
that provided by tendon rupture cannot be exploited."
Dr Devon Mothersille, technical author for BS8081:2015, and Chairman of TC 182/W3 (Testing of grouted anchors),
informed the authors on 15 September 2018 that there are considerable differences of opinion in Europe and USA
regarding the quantifying the ultimate bond strength. Denmark, Germany, US Dept. of Transport and Bureau Securitas
adopt a critical creep load and assess the magnitude of ultimate bond stress from data derived prior to complete pull-out
of the anchor.
Dr Mothersille also indicated that the term 'anchor failure load' is frequently used in the UK, representing all ground
failure mechanisms controlling ultimate load. Prior to the recent publication of EN ISO 22477-5:2018, UK practice
followed BS8081-2015 so that the magnitude of the anchor failure load in an investigation test was generally established
using an observational method:
"In an investigation test, the ultimate ground/grout interface resistance should be estimated by back analysis
from the failure load. The failure load is identified as the maximum load attainable by the test anchor in spite of
ongoing displacement.
NOTE: A prudent approach when confirming the failure load would be to closely observe the behaviour of the
after 15 min since this could be an indication that
the failure load for the test anchor is approaching. Additional cycles of unloading and reloading might also
serve to confirm the failure load.”
is the elastic extension of the tendon at the proof load.
Dr Mothersille notes that the approach described above provided a pragmatic method of identifying and quantifying the
failure load from which ultimate bond stress was derived, but that acceptance criteria based on the above may be seen to
have a bias towards the acceptance of anchors with longer free lengths. Thus it was possible that adjacent anchor
installations with different free tendon lengths would have different acceptance criterion. For this reason, and in
embracing the concept of European compatibility, the UK has now amended its approach and resorted to adopting
measurement of creep rate as an acceptance criterion in the new testing standard. By doing this each anchor installation
will be subjected to a single acceptance criterion for creep behaviour irrespective of the tendon free length.
It is the authors’ opinion that regardless of the criterion used to define failure load, the fundamental mathematical problem
remains that, unless one adopts a specially designed short bond length anchor, it is impossible to measure the parameter
termed 'ultimate bond strength' in AS5100.3.
5.4 AVAILABLE DATA ON ULTIMATE BOND STRENGTH
An examination of published test data and that otherwise available to the authors, shows that:
• of 84 entries in the famous series of papers by Littlejohn and Bruce (Ground Engineering 1975 to 1976, collated
in 1977) there is only one measure of ultimate bond stress in sandstone, with no technical information about
that rock,
• of 87 entries for anchors in sandstone by Barley (1988) there are 12 measures of ultimate bond stress, with
limited technical information about the sandstones, and
• the present authors have less than ten ultimate results from about 200 tested rock bolts and anchors.
The available data are given in Table A1 in Appendix A. The data in Table A1 are insufficient and inadequate for
determining the parameter of Equation 1, let alone . Therefore, we are left with a situation that bond shear strength
values cannot be determined with reasonable confidence from existing anchor test data for Hawkesbury Sandstone or
rock similar to Hawkesbury Sandstone,.
5.5 PROOF TESTS AND DESIGN VALUES
Post-installation testing of anchors, and even prestressing them to the desirable lock-off load, places them in a different
category to almost every other kind of structure, such as socketed bored piles, pad foundations, on-ground slabs, bridges
and buildings. It also means that there is a substantial database of proof test bond shear values which can be used to
develop design guidelines. While it may be desirable to have a safety factor of 2.0 on ultimate bond strength, we have
demonstrated above that this is not a parameter which is knowable with confidence.
The authors’ recommended approach is to base designs on proof test data, verifying such designs by proof loading of
every anchor.
As stated by Byrne, Everett and Schwartz (1995):
"The stressing and testing of anchors is one of the most important phases of any anchor installation. During
this phase each anchor is tested to a specified percentage above its design load and the performance of the
anchor recorded on a stress/strain graph which is compared to predicted performance. In addition, creep
relaxation is recorded which allows a prediction of the long term behaviour of each individual anchor."
Littlejohn and Bruce (1976) record 41proof test results on various rocks, with 7 proof tests in sandstone and 5 in shale.
Barley (1988) records proof test bond values for 66 different projects in sandstone, and 50 different projects in shale and
mudstone.
Unfortunately, for most of the Littlejohn and Bruce cases the authors have been unable to find data on the rock strength
properties. The authors therefore sourced the cited references and were disturbed to find that many of the entries in
Littlejohn and Bruce (1977), of relevance to this paper, contain significant errors. For the authors’ database the authors
have reverted to the source documents.
Barley (1988) gives strength assessments in terms of the 1970 Geological Society UK system. In addition, several of his
results are for Bunter Sandstone. Roxborough (1982) proved that in respect to strength, cuttability and abrasivity there is
"statistically no significant difference between" Hawkesbury Sandstone and Bunter Sandstone.
Table A2 in Appendix A summarises available proof test data from anchors in sandstone where there is information on
unconfined compressive strength (UCS).
The data from the second last column of Table A2 represent field tests loaded to about 120% of the average bond stress
adopted for those particular projects. Therefore the data can be thought of as proven design values, but without knowledge
of what were the levels of safety against ultimate bond strength. The authors believe that this level of safety is substantially
unknowable, and in any event unnecessary. The authors are of the view that the values in Table A2 can be used as a
guide to design values for Hawkesbury Sandstone. Such design values must, in accordance AS 4678, be proof tested.
Figure 6 gives the tested design values from Table A2 plotted against UCS, together with the authors’ recommended line
for design average anchor stresses in Hawkesbury Sandstone with UCS>1MPa.
Figure 7 (overleaf) gives the authors’ recommendations for design values in shale, on the same basis as for sandstone.
For a bond length of greater than about 4 m in sandstone or shale, the distal part of the bond zone contributes nothing of
substance to ultimate or serviceability performance. A dangerous design combination is therefore an aggressive design
bond strength with a long bond length. If bond zone lengths of greater than about 4m are necessary it may be appropriate
to adopt a single-bore multiple-anchor system.
The authors were kindly provided with pull-out test results by Mr Ron McKenzie of Megabolt, for tests in unweathered
Triassic and Permian shales and sandstones in the Sydney Basin. The data could not be included in Figures 6 and 7 as
neither UCS values nor specific lithological units were recorded. Also it is not recorded whether it was the cable, the
grout/sheath or grout/rock contact that failed. However, for the record the data are presented in Figure 8, as box and
whisker plots. The results give comfort that the design recommendations given in this paper are conservative.
5.6 UNTENSIONED FULLY BONDED DOWELS
It is not possible to provide generalised guidelines for fully grouted dowels in jointed Hawkesbury Sandstone and Ashfield
Shales. The reasons are as follows.
• Passive-resistance bond-lengths in a particular dowel, relative to bond-lengths along the same dowel within
zones of active movement (restrained by the passive lengths) are not definable in a general manner as is the
case with prestressed anchors.
• In many situations the primary action of fully bonded dowels is to provide shear resistance across bedding
planes and joints; requiring specialist testing and analyses (Pells, Pells and Pan, 2018).
6 APPLICATION TO HYDROGEOLOGY
Groundwater modelling requires parameterisation of:
• Hydraulic conductivity of a hydrogeological unit. A hydrogeological unit is a zone or layer judged to have
reasonable uniform macro hydrogeological properties; choice and volumes of units depends on the particular
problem under consideration.
• Drainable storage (an index of the volume of water that can be drained from the pores – termed Specific Yield).
• Compressible storage (the quantity of water released under changes in stress).
If an analysis assumes saturated conditions the hydraulic conductivity is taken as a constant for each unit. In reality,
hydraulic conductivity changes as a function of saturation (and also of stress, although this effect is typically small enough
to be considered negligible). The change in hydraulic conductivity considered in unsaturated analysis requires the
additional functions:
• stored water content vs suction pressure, and
• hydraulic conductivity versus suction pressure function.
Herein the authors deal only with saturated conditions.
Hydraulic conductivity of a rock mass is a function of flow through the rock substance (primary) and flow through the
defects (secondary). The combined conductivity estimated by packer tests and, less often, pumping tests. There is an
extensive database of such tests in Hawkesbury Sandstone and the Wianamatta Group.
Drainable storage, or specific yield, is taken as the difference between total porosity and residual water content. These
values are seldom measured directly in the field this parameter can be estimated from pumping tests that include
monitoring of piezometers; but in reality they can be reasonably estimated.
Specific storage (which is Hydraulic Conductivity divided by Hydraulic Diffusivity) cannot be determined arbitrarily
because it is, from consideration of applied mechanics, a function of the compressibility of the mass, compressibility of
water, and porosity: as per Equation 2.
Ss = specific storage =
(
(
)(
)
)
(2)
≥ 25 1A 1B 1C 1D
10 – 25 2A 2B 2C 2D
3 – 10 3A 3B 3C 3D
Notes to Table 9
(a) Intact rock strength and typical defect spacing (true, not apparent) are assessed across the excavated face and bolted zone.
Bedding partings and joints are assessed independently. The lesser typical spacing governs.
(b) The ground type is downgraded from 'A' to 'B', 'B' to 'C' or 'C' to 'D' if a sub-horizontal feature occurs within 1/10 x span
above the crown: sub-horizontal features include; bedding parting, bedding shear, clay seam and fragmented / highly
weathered rock
(c) Minor discrete faults which are clean and tight and which apart from small bedding offset are indistinguishable from joints
are to be treated the same as joints.
The defect spacing of 0.5 m is essentially the same as the 0.6 m value in the foundation classification system. However,
it was found necessary to introduce 1m and 1.5 m categories to cover the sensitivity of support design to defect
spacing. Consideration was also given to using strength ranges of the foundations system. However, the two different
strength ranges for sandstone and shale classes in the foundation system and the experience in estimating intact strength
in practice as discussed in Nash et al (2017), persuaded the authors to use the CANMET (1977) strength categories for
both sandstone and shale.
Hence, according to Table 9, a sandstone mass with intact UCS≥25MPa classifies as Type 1; with intact UCS ≥10MPa
and < 25 MPa is Type 2, and so on. The appropriate suffix A, B, C or D is controlled by the true defect spacing.
As opposed to the foundation classification system, the 12 categories of sandstone or shale determined according to Table
6 are not directly related to support design. Such design must take into account:
• tunnel span,
• in situ stress conditions, being a function of depth of cover, and
• potential kinematic failure blocks which are a function of tunnel orientation relative to defect orientations.
8 CONCLUSIONS
The classification developed in 1978 by a sub-committee of the Australian Geomechnics Society for foundation designs
on the Triassic sandstones and shales of the Sydney Basin has proven to be useful as a communication tool on many and
varied projects in Sydney and its surrounds. However, it has also been misused.
In this 2nd update of the system, forty years after the first publication, the authors have:
• sought to clarify the guidelines in respect of foundation design,
• indicated uses of the classification system, or components thereof, for design of prestressed ground anchors,
and for preliminary groundwater analyses, and
• shown that a modified classification system is appropriate for linkage to tunnel support design.
It is noted that the 1978/1998 classification system should not be used in assessments of slope stability, rock
excavatability and rock cutting.
9 ACKNOWLEDGEMENTS
Derek Anderson provided substantial data on testing of cement grouted dowels and cable anchors in Wianamatta Shales.
Dr Steven Pells provided much of the material in respect of hydrogeology. A particular thanks to Ngaire Petit-Young of
Information First who found for us old and obscure references cited in the database of Littlejohn and Bruce (1977) and in
the text book by Hanna (1982). The authors thank Ron McKenzie for providing access to Megabolt’s pull-testing database
The authors thank Dr Chris Haberfield of Golder Associates, Dr Devon Mothersille of Geoserve Global, and Mr Phil
Clarke of Roads and Maritime Services, NSW for their helpful comments on early drafts. The authors also thank the
unknown reviewer from Australian Geomechanics who was, correctly, pretty ruthless in regard to the first submission
and thereby guided the authors to produce a better document
10 REFERENCES
Anderson, D.A. (2018) Personal Communication, Rockbolt and Anchor Test Results in Sydney Tunnels, 1999 to 2018.
Barley, T. (1988). Ten Thousand Anchorages in Rock. Ground Engineering.
Bertuzzi, R. and de Ambrosis, A. (2017). Alternative rock mass classification for tunnel design in Sydney, 16th
Australasian Tunnelling Conference, Sydney 30 Oct – 1 Nov 2017.
Bertuzzi, R. and Pells, P. J. N. (2002). Geotechnical Parameters of Sydney Sandstone and Shale. Australian
Geomechanics Journal Vol 37, No 5 p 41-54.
Brady, K. C. and McMahon, W. (1997). Limit State Design in Ground Anchorage Practice. Ground Anchorages and
Anchored Structures, Thomas Telford, London.
Bruce, D. and Wolfhope, J. (2007). Rock Anchors for North American Dams: The Development of the National
Recommendations (1974-2004). Ground Anchors and Anchored Structures in Service 2007. Proceedings
International Conference, Institution of Civil Engineers, London
Byrne, G., Everett, J. P. and Schwartz, K. (1975). A Guide to Practical Geotechnical Engineering in South Africa.
Frankipile South Africa Pty Ltd.
CANMET (1977). The Pit Slope Manual. Mining Research Laboratories, Canada.
De Nicola, A. and Randolph, M.F. (1993). Tensile and Compressive shaft Capacity of Piles in Sand. ASCE Journal of
Geotechnical Engineering, Vol. 119, No. 12, pp 1952-1973
Haberfield, C. M. and Baycan, S. (1997). Field Performance of the Grout/Rock Interface in Anchors. Ground Anchorages
and Anchored Structures, Thomas Telford, London.
Hanna, T. H. (1982). Foundations in Tension. Ground Anchors. Trans Tech Publications Series on Rock and Soil
Mechanics Vol. 6.
Hegazy, Y. A. (Undated). Reliability of Estimated Anchor Pullout Resistance.
http://www.dappolonia.com/pdfs/techpaper6.pdf
Kenny, T. C. (1977). Factors to be Considered in the Design of Piers Socketed into Rock. Conference Design and
Construction of Deep Foundations, Canadian Society for Civil Engineering, Sudbury, Ontario, February.
Littlejohn, G. S. and Bruce, D. A. (1977). Rock anchors – state of the art. Foundation Publications Ltd.
Longworth, C. R. (1971). The Use of Stressed Anchors in Open Excavations and Surface Structures. The Australasian
Institute of Mining and Metallurgy Illawarra Branch, Symposium on “Rock Bolting”.
Maddox, J. M., Kinstler, B. E. and Mather, R. P. (1967). Meadowbank Dam - Foundations. Civil Engineering
Transactions, Institution of Engineers, Australia, October.
Morris, S. S. and Garrett, W. S. (1956). The raising and strengthening of the Steenbras Dam. Proceedings, Institution of
Civil Engineers, Part I, Vol. 5, No. 1.
Nash, T. R., Shen, B. and Bertuzzi, R. (2017). Geotechnical Validation for Sydney Tunnels. 16th Australasian Tunnelling
Conference 2017, 30 October – 1 November 2017.
Obergfell, M. N. (1987). Construction and Performance of a Permanent Anchor (Tieback) System for the Stanford Linear
Collider. Stanford Linear Accelerator Centre, Publication 4211, February.
Pells, P. J. N. (1983). Plate load tests on soil and rock. Chapter 6 In-Situ Testing for Geotechnical Investigations. Ed M.
C. Ervin, Balkema, 1983.
Pells, P. J. N. (1990). Stresses and displacements around deep basements in the Sydney area. 7th Australian Tunnelling
Conference, Inst. Engineers Australia, Sydney, p241.
Pells, P. J. N. (1999). State of Practice for Design of Socketed Piles in Rock. 8th Australian - NZ Geomechanics
Conference, Hobart.
Pells, P. J. N. (2001). A Note on Rockbolt Design for Load Supporting Ledges in Hawkesbury Sandstone. Australian
Geomechnics, December.
Pells, P. J. N., Douglas, D. J., Rodway, B., Thorne, C. P. and McMahon, B. K. (1978). Design Loadings for Foundations
on Shale and Sandstone in the Sydney Region. Australian Geomechanics Journal, G8, pp 31-39.
Pells, P.J.N. and Turner, R.M. (1982) End bearing on rock with particular reference to Sandstone. Structural Foundations
on Rock, Balkema, Rotterdam.
Pells, P. J. N., Mostyn, G. and Walker, B. F. (1998). Foundations on Shale and Sandstone in the Sydney Region.
Australian Geomechanics Journal, December pp 17-29.
Pells, P. J. N., Pells, S. E. and Pan, L. (2018) On the resistance provided by grouted rock reinforcement to shear along
bedding plane and joints. Australian Geomechanics Journal, V53, No 2, June 2018, pp 55-74.
Post Tensioning Institute, USA (2014) Recommendations for Prestressed Rock and Soil Anchors. 5th Ed, Farmington
Hills, Michigan.
Poulos, H. G. and Davis, E. H. (1974). Elastic solutions for soil and rock mechanics. John Wiley, New York.
Poulos, H. G., Best, R. J. and Pells, P. J. N (1993). Load Test on Instrumented Socketed Pile for new Glebe Island Bridge.
Australian Geomechanics Journal, No 24.
Radhakrishna, H. S. and Klym, T. W. (1980) Behaviour of rock anchored foundations subject to shear and moment loads.
IEEE Trans. On Power Apparatus and Systems, Vol. PAS-99, No 2, pp 760-4.
Randolph M. (2003) RATZ Manual (v4.2) Load transfer of axially loaded piles.
Rowe, K. R. K. and Armitage, H. H. (1984). The design of piles socketed into weak rock. GEDT-11-84, University of
Western Ontario.
Saliman, R. S. and Schaefer, R. H. (1968). Anchored Footings for Transmission Towers. ASCE Annual Meeting, Preprint
753, Pittsburgh, October.
Standards Australia (2011) AS4678-2002 Earth-retaining structures, Standards Australia
Standards Australia (2017) AS5100.3:2017 Bridge Design Part 3 – Foundation and soil supporting structures, Standards
Australia.
Thornton, L. E. (2006). Rock Strength: A control of Shore Platform Elevation. Journal of Coastal Research, V22, No 1.
Walker, B. and Pells, P. J. N. (1998) The Construction of Bored Piles Socketed into Sandstone and Shale. Australian
Geomechanics Journal, No33, Part 3.
Williams, A. F. and Muir, A. G. (1972). The Stabilization of a Large Moving Rock Slide with Cable -anchors. Country
Roads Board, Victoria.
Williams, A. F. and Pells, P. J. N. (1981). Side resistance rock sockets in Sandstone, Mudstone and Shale. Canadian
Geotechnical Journal 18, No 4.
Whitman, R. V. and Richart F. E. (1967). Design procedures for dynamically loaded foundations. University Michigan
Wong, P. K. (2013). Case Studies of Cost-effective Foundation Design in Rock. Proc 18th Int. Conf. Soil Mechanics and
Geotechnical Engineering, Paris pp 2901-2904
Wong, P. K. and Oliveira, D. (2012). Class A Prediction versus Performance of O-Cell Load Tests in Sydney Sandstone.
Australian Geomechanics Journal., Vol 47, No.3, pp 89-96.
Wong, P. K. (2018) Unpublished back-analyses of recent rock anchor proof test results in Sydney Sandstone and Shale
(Coffey Services Australia projects).
Wyllie, D. C. (1999). Foundations on Rock, E & FN Spon (Pub), Chapter 9 Tension Foundations, pp300-359.
Zicko, K. E., Bruce, D. A. and Kline, R. A. (2007) The Stabilization of Gilboa Dam, New York, Using High Capacity
Rock Anchors: Addressing Service Performance Issues. Ground Anchors and Anchored Structures in Service. Inst
of Civil Engineer, London.
APPENDIX A
DATABASES
Through the process of back-analysing some of the tests conducted in Sydney, the authors have become aware of error
arising from in testing rock anchors that have been full column grouted, including the free length. In such cases the test
zone must include part of the free length above the bond zone even though the free length tendon is greased and
sleeved. As such, the performance testing of the bond zone may be meaningless, with only the elastic elongation of the
tendon tested and confirmed at best.
Table A1: Reported Ultimate Average Bond Strength for Sandstone (OPC grout)
Ultimate
Unconfined Average Approximate
Formation and Compressive Length & Bond Shear
Description Location Factor Reference
Age strength diameter [m] Strength ( ave/UCS)
[MPa]
[kPa]
Medium grained Steenbras dam
Table Mountain Morris and
quartzitic
Cape Town 50 1.1 / 0.0613 3620 0.07 Garrett
interbedded with Ordovician
South Africa (1955)
weaker sandstone
Table A2: Data from Anchor Proof Tests in Sandstone (OPC grout)
Intact Strength
Average
Formation
Description Location Description Assessed Test Bond Reference
and Age
(Barley, UCS [kPa]
1988) [MPa]
Sandstone 50 840
Sandstone with interbedded Meadowbank Dam Longworth (1971).
Triassic 30 1100
siltstone/ mudstone Tasmania, Australia Maddox et al (1967)
Stanford Linear
Moderately to well
Miocene Accelerator 1 228 Obergfell (1987)
cemented sandstone
California, USA
80 973
80 1576
Mod Strong
Sandstone Glasgow 80 1215
to very strong
80 1250
80 1250
Sandstone with siltstone River Forth Strong 70 483
Fine grained Glasgow Strong 70 1036
Sandstone Jarrow Strong 70 802
70 940
70 1370
Sandstone with mudstone Heysham power
Strong 70 940
flakes Station
70 928
70 940
Table A2 (continued): Data from Anchor Proof Tests in Sandstone (OPC grout)
Intact Strength
Average
Formation
Description Location Description Assessed Test Bond Reference
and Age
(Barley, UCS [kPa]
1988) [MPa]
Barley (1988)
1 692
Very weak to
Silty sandstone Birmingham 1 360
weak
1 450
Williams & Muir
Great Ocean Road
Hard sandstone Jurassic 70 1240 (1972)
Australia
Thornton (2006)
Gilboa Dam 110 1903
Sandstone Zicko et al (2007)
New York, USA 110 1455
Highly weathered Torlesse
Wellington NZ 18 437 Wightman (2017)
(greywacke/ sandstone)
Glebe Island Bridge Poulos et al (1993)
15 1420
Sydney, Australia
1060
1780
M2 Clarke (2012)
25 1520
Sydney, Australia
1060
1060
Circular Quay
455
Sydney, Australia
272 Wong (2018)
Hawkesbury 25
Sandstone Barangaroo
Medium grained, fresh 247
Triassic Sydney, Australia
284
30 2000
30 2000
30 2000 Salcher & Bertuzzi
Metro NW
(2018)
Sydney, Australia 15 500
15 500
15 500
Table A3: Reported Ultimate Bond Strength Shale, Siltstone and Mudstone (OPC grout)
Ultimate
Average Bond Approximate
Formation and UCS Length &
Description Location Shear Strength Factor Reference
Age [MPa] diameter [m]
( ave/UCS)
[kPa]
4100 0.41
Siltstone with 0.25 / 0.098 3100 0.31
some sandstone, Scoresby
Melbourne 4 to 15 4000 0.4 Haberfield
moderately Melbourne
Mudstone (1997 and
weathered, Mean 10 1700 0.17
Silurian Australia 2018)
horizontal
bedding 0.5 / 0.3 1800 0.18
4500 0.45
Lincolnshire
Mudstone 1 10 / 0.127 190 0.19
England
Bridgewater Barley
Silty mudstone 1 10 / 0.105 240 0.24
England (1988)
Leeds
Shaly mudstone 2 6 / 0.112 316 0.158
England
Table A4: Data from Anchor Proof Tests in Shale, Siltstone and Mudstone (OPC grout)
Intact Strength
Length & Average
Formation and Description
Description Location Assessed diameter Test Bond Reference
Age Geological
UCS [m] [kPa]
Society UK
[MPa]
(1970)
Fractured shale 25 708
Bragdon Saliman & Schaeffer
(75%) and California, USA
Formation USBR (1968)
siltstone 8 339
Siltstone / Charlestown Moderately
25 8 / 0.105 580
mudstone UK strong
Pennine Coal
Mudstone Measure Group Newcastle, UK Strong 40 6 / 0.105 505
Carboniferous
Halifax
Mudstone Carboniferous Strong 40 8 / 0.105 503
Yorkshire, UK
Siltstone / Carboniferous / Bridgenorth Strong to very
60 8 / 0.105 360
mudstone Permian Shropshire, UK strong
Mudstone and
Carboniferous Kielder, UK Strong 50 6 / 0.112 505
sandstone Barley (1988)
Pennine Coal
Mudstone coal
Measure Group Newcastle, UK Very weak 1 8 / 0.105 308
bands
Carboniferous
8 / 0.105 550
Middlesborough Very weak to
Mudstone Jurassic 3 12 / 0.105 550
Yorkshire, UK mod. Weak
8 / 0.105 270
Siltstone and Wigan, Weak to mod.
Permo-Triassic 5 12 / 0.105 455
mudstone Lancashire, UK weak
Table A4 (continued): Data from Anchor Proof Tests in Shale, Siltstone and Mudstone (OPC grout)
Intact Strength
Length & Average
Formation and Description
Description Location Assessed diameter Test Bond Reference
Age Geological
UCS [m] [kPa]
Society UK
[MPa]
(1970)
Middle Coal Barnsley
Mudstone with
Measures South Yorkshire SPT = 90 1 6 / 0.105 250
sandstone
Carboniferous UK
Lower Coal Sheffield
Mod weak to
Siltstone Measures South Yorkshire 12 5 / 0.105 240
mod.strong Barley (1988)
Carboniferous UK
Hanley (Stoke-
Upper Coal
Mudstone and on-Trent)
Measures Weak 2 10 / 0.105 200
siltstone Staffordshire
Carboniferous
UK
Gilboa Dam 65 1.49/0.380 1944
Shale New York, Zicko et al (2007)
USA 65 3.02/0.380 1441
Macdonaldtown
Extremely
Sydney, 1 8.5 / 0.15 105 Wong (2018)
weak to weak
Australia
2 3 / 0.075 277.3
10 1 / 0.075 769.5
Olympic Park
10 1 / 0.075 727.9
Ashfield Shale Rail Loop This paper
Siltstone Sydney, (Anderson)
Triassic 10 1 / 0.075 769.5
Australia
10 1 / 0.075 665.5
10 1 / 0.075 727.9
4 1 / 0.075 811.1
Olympic Park
4 1/0.075 831.8
Rail Loop This paper
Sydney, (Anderson)
1.5 1/0.075 831.8
Australia
1.5 1/0.075 831.8
2 1.05/0.09 640
2 1.23/0.09 561
4 1.2/0.045 1690
4 1.2/0.045 1267
Siltstone
14 0.6 955
4 1.2 477
14 0.6 955
Metro NW
Ashfield Shale Bertuzzi and Salcher
Sydney, 14 0.6 955
Triassic (2018)
Australia
4 1.2 477
14 1.65 857
14 1.65 857
Table A4 (continued): Data from Anchor Proof Tests in Shale, Siltstone and Mudstone (OPC grout)
Intact Strength
Length & Average
Formation and Description
Description Location Assessed diameter Test Bond Reference
Age Geological
UCS [m] [kPa]
Society UK
[MPa]
(1970)
14 1.65 857
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 3.55 359
14 8 159
14 8 159
14 8 159
Metro NW
Ashfield Shale Bertuzzi and Salcher
Siltstone Sydney, 4 8 159
Triassic (2018)
Australia
4 8 159
4 8 159
14 1 1273
14 1 1273
14 1 1273
14 0.75 858
14 0.75 858
14 0.75 858
14 1 1123
14 1 1123
14 1 1123
4 1 1123
4 1 1123
4 1 1123
The authors were kindly provided with pull-out test results by Mr Ron McKenzie of Megabolt, for tests in unweathered
Triassic and Permian shales and sandstones in the Sydney Basin. The data could not be included in Figures 6 and 7 as
neither UCS values nor specific lithological units were recorded. Also it is not recorded whether it was the cable, the
grout/sheath or grout/rock contact that failed. However, for the record the data are presented below, as box and whisker
plots.
Data from pull-testing in sandstone and shale by Megabolt Australia Pty Ltd.