Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected consequences of slow drawdown, soft bank materials and anthropogenic change

Abstract

Slope stability modeling has been used to investigate river bank failures that occurred on the lower Murray River between 2008 and 2010. The larger of the known failures commonly present as deep-seated, circular failures and are developed in clays which form the floodplains and channel margins. Most of the modeled failures occurred during the peak of the so-called Millennium Drought when the water surface of the river fell to a level one metre below sea-level as the barrages located at river mouth prevented seawater incursion. Bank failure during a drought is unusual as most large-scale river bank failures occur due to toe-scour or post-flood draw-down effects associated with flood recession. Slump rupture surfaces commonly crest within constructed levees and channel margin fills, while the toes of the rupture surface are located near the break in slope where the inclined channel margin joins the channel floor. Modeling indicates that the failures are related to anthropogenic channel modification, ie., i) pool level manipulation; and ii) levee construction, but that these anthropogenic modifications to the channel have probably accelerated and amplified natural processes of channel-change.

Full text

7th Australian Stream Management Conference - Full Paper
Geomechanical modeling of the Murray’s Millennium Drought river bank
failures: a case of the unexpected consequences of slow drawdown, soft
bank materials and anthropogenic change
Hubble T1, De Carli E1, and Airey D2
1 School of Geosciences, The University of Sydney, 2006. tom.hubble@sydney.edu.au
2. School of Civil Engineering, The University of Sydney, 2006.
Key Points
• The lower Murray River’s channel margins and adjacent floodplains are predominantly comprised of
unconsolidated, low shear-strength, low-permeability muds
• Slope stability models indicate that the banks and channel margin slopes are commonly inclined at angles
close to, or at the limit of stability making the banks prone to destabilization by other factors
• Lowering of the water surface during the Millennium Drought to unusually low, below sea-level elevations
probably triggered widespread deep-seated circular failures
• Constructed levees, embankments or man-made fills placed at or near the water’s edge increases the
likelihood of bank failure
Abstract
Slope stability modeling has been used to investigate river bank failures that occurred on the lower Murray River
between 2008 and 2010. The larger of the known failures commonly present as deep-seated, circular failures and are
developed in clays which form the floodplains and channel margins. Most of the modeled failures occurred during the
peak of the so-called Millennium Drought when the water surface of the river fell to a level one metre below sea-level as
the barrages located at river mouth prevented seawater incursion. Bank failure during a drought is unusual as most
large-scale river bank failures occur due to toe-scour or post-flood draw-down effects associated with flood recession.
Slump rupture surfaces commonly crest within constructed levees and channel margin fills, while the toes of the rupture
surface are located near the break in slope where the inclined channel margin joins the channel floor. Modeling indicates
that the failures are related to anthropogenic channel modification, ie., i) pool level manipulation; and ii) levee
construction, but that these anthropogenic modifications to the channel have probably accelerated and amplified
natural processes of channel-change.
Keywords
Slope stability, river banks, slumps, draw-down, undrained shear strength
Introduction
River banks commonly collapse during floods or during the recession of the flood waters (e.g. Galay 1983; Hubble and
Rutherfurd 2010; Thompson and Crooke 2013) due to toe-scour and the rapid drawdown effect (Morgenstern 1963). A
series of highly unusual riverbank collapses occurred on the lowermost reaches of Australia’s Murray River between
2008 and 2010 during the peak of the extended ‘Millennium’ drought (Jaksa et al 2013) when the river level, which can
be manipulated by a barrage located at the river mouth, gradually fell almost two metres below normal pool level over a
period of twelve months (Coffey 2012; Jaksa et al 2013). Most of these failures occurred without warning and one event,
at Long Island Marina in February 2009, very nearly caused a fatality and took three cars into the Murray. Despite an
intensive search which employed high-resolution multibeam bathymetry only one of these vehicles has been recovered.
Initial geotechnical investigations into the causes of these failures that were commissioned to understand the problem
and evaluate the risk to the public were somewhat equivocal but in general agreed that the primary contributing factors
responsible for the collapses were the presence of a soft clay in the channel margins and historically low river levels
brought about by severe drought conditions and blocking of upstream seawater icursions by the system of barrages
located at Goolwa (e.g. Arup 2008; SKM 2010; Coffey 2012). This paper extends this work and reports some of
preliminary findings of an integrated geomorphologic, geological and geotechnical, multi-agency research project
investigating the Millennium Drought failures which is sponsored by South Australia’s Department of Environment,
Water and Natural Resources (DEWNR-SA).
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 278

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Field Area, Specific Sites and Methods
The focus of this study are the lowermost reaches of the Murray River which developed within the bedrock gorge
located downstream of Lock One (Blanchetown) and upstream of Lake Alexandrina. Specific sites of interest that have
been used to develop representative bank profiles for investigating river bank response to the events the banks
experienced during the Millennium Drought are located at a variety of places near the town of Murray Bridge;
specifically, White Sands (WS) and Thiele Reserve (TR) (Figure 1).
Figure 1. (A) Map of field area and LIDAR image of the study area showing the location of sites discussed in the text.
LA-Lake Alexandrina, W-Wellington, WS-White Sands, TR-Thiels Reserve, M-Mannum, P-Purnong.
Channel margin materials were physically sampled with two to four meter long push cores (thick walled 90 mm PVC
pipe) deployed at appropriate intervals from the side of a houseboat to establish bank stratigraphy – the horizontal
layering evident in the each core enables the overlapping of successive cores taken from progressively deeper water.
Cores were visually logged and sub-sampled to establish sediment grain size distributions using a Mastersizer 2000. Cone
penetrometer (CPT) profiles for each site were acquired to confirm vertical continuity of the stratigraphy with an
instrument provided by the University of Adelaide; these results were also used to determine the material’s undrained
shear strength. Channel margin and bank geometry was established from multibeam echo-sounding survey records
provided by South Australia Water. Slope stability modeling was undertaken using the computer program XSLOPE
(Version 4.6, Balaam 2010) using Bishop's Slip Circle method to generate factors of safety (Bishop 1955).
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 279

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Figure 2. (A) Core photographs and CPT record of horizontally laminated muds sampled from the channel margin at
Riverglen Marina, White Sands (at left). (B) Photographs showing the recovery of a push core at Wellington (upper
right) and deployment of the CPT at Monteith, Bells Reserve (lower right). The CPT record indicates a monotonic
increase of end-bearing and sleeve resistance with depth indicating continuity of the low strength muds to the base of
the record (13 m below river level).
Sediment Type and Characteristics
Photographs of a typical channel margin core and a CPT profile recorded at the same site are presented in Figure 2. The
photographs show exquisitely laminated, light-grey to medium-grey, well-sorted, unconsolidated muds and clays of the
Coonambidgal Formation which is mid-Holocene in age at its base (Twidale et al. 1978) to near present-day in age at the
top of the unit (Hubble and De Carli unpublished data). Laminations vary in thickness between ~0.2 mm to 2 cm and
close inspection of the laminations reveals a variety of soft sediment deformation features, e.g. distorted layering and
small faults (evident at ~0.7m ~1.4m and ~1.6m depth). In situ the material is completely saturated and the thumb
penetration test (USBR 2001) demonstrates that the muds are very soft. CPT test profiles acquired at all of the study
sites confirms that the Coonambidgal muds are low strength materials with cone tip resistance typically increasing
monotonically from a starting resistance of 0.2 MPa, just below the channel floor at the river’s edge to 0.8 MPa, 20
metres below the starting point of the test. Sleeve resistance increases from 0 MPa to 0.01 MPa over the same range
with friction ratios generally varying between 0.8% and 1.2%. These values are typical of the soft muds and clays
retrieved in push cores taken at this site. Typical undrained shear strengths for materials with this low cone tip pressures
are 10 kPa + ~2kPa in the near surface materials rising evenly to between 30 kPa + ~4 kPa at a depth of 20 m below the
surface (cf USBR 2001).
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 280

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Slope Stability Modeling
XSlope model results for the Thiele Reserve field site are presented in Figure 3 which is located adjacent to a cliff which
forms a part of the inside a meander bend formed in the bedrock valley in which the Murray’s floodplain materials have
been deposited. An inferred bedrock bench is included in the model as these have been observed elsewhere when the
margin of the channel abuts the valley walls. The sands of the Monoman Formation which underlie the Coonambidgal
Muds at a depth of 15 metres below river level at this site are also included in the analyses. The models presented in this
study are preliminary, simplified representations of the site’s stratigraphy and materials. One set of analyses models the
actual conditions of this site; a second set of analyses is included to demonstrate how the presence of a constructed
embankment would affect the stability of the channel margin at this site. Levees and roadways have been built at many
sites adjacent to the channel margin in the surrounding area, indeed some of the larger and more problematic failures
have occurred where a levee or deep fill has been constructed. A five-layer geomechanical model has been used to
represent banks and adjacent floodplain. It consists of a surface sandy fill 0.5 m thick (plus the embankment in models C
and D); a very soft near-surface mud layer (undrained strength ~10 kPa) approximately four metres thick; a deeper and
stronger mud layer (undrained strength ~20 kPa); which overlie the Monoman Sand and then Bedrock.
Table 1. Factors of Safety calculated for Thiels Reserve for lowered pool levels and a range of undrained shear
strength values for the near-surface mud layer
XSlope Model Runs
Upper Clay Layer
Undrained Shear
Strength
Factor of Safety at
Normal Pool Level
Factor of Safety
One Metre Below
Normal Pool Level
Factor of Safety
Two Metres Below
Normal Pool Level
Thiele Reserve Profile
8.5 kPa
1.20
1.13
1.01
10 kPa
1.26
1.19
1.11
11.5 kPa
1.27
1.20
1.10
Thiele Reserve with
embankment
8.5 kPa
1.10
0.99
0.85
10 KPa
1.17
1.10
0.99
11.5 kPa
1.17
1.12
1.03
Two key parameters have been varied in order to understand their effect on the stability of the banks. These are: firstly,
the previously identified trigger for failure, pool level, which slowly lowered over a period of six-months to levels
between one and two metres below normal; and b) the undrained strength of the near surface muds. The results of
eighteen model runs are summarized in Table One. Note that the generated Factor of Safety (FoS) is a ratio that divides
the restoring force acting on a slide mass by the disturbing force acting on it. A FoS > 1 indicates stability, FoS <1
indicates instability, and a FoS=1 indicates that the disturbing and restoring forces are critically balanced. However
caution is normally exercised in interpreting FoS values because natural materials vary laterally such that the model
parameters may not be completely representative of the actual situation. Consequently, consideration of the change in
FoS is often just as useful as the FoS value with increases indicating greater stability and decreases indicating lower
stability (cf. Hubble and Rutherfurd 2010). Decreases of FoS values to near unity are commonly accepted to indicate a
high likelihood of failure when considering river bank stability (cf Hubble 2010).
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 281

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Figure 3. Representative XSlope model results for the failures that occurred at Thiele Reserve during the unusual
lowered river elevations experienced during the Millennium Drought. (A and B) Actual conditions at the site; (C-D)
Hypothetical Embankment included illustrating the consequences of loading the channel margin with a levee or road
embankment. Note that in models A to C are for a very soft, near-surface mud layer (8.5 kPa) which tends to generate
shallow failure surfaces while Model D is for a slightly stronger near-surface mud layer (11.5 kPa) which generates
more deeply located rotational circles.
The model results demonstrate several characteristics that have been noted and documented in the geotechnical
reports and studies produced during or just after the Millennium Drought (e.g.SKM 2010; Coffey 2013; Jaksa et al 2013
and De Carli et al 2013). In general critical circles are located in positions where actual failures occurred for situations
similar to the models (cf. SKM 2011; Coffey 2013 Jaksa et al 2013). In particular the positioning of the shallow failure
circles in the unmodified bank (Figure 3A) replicates very closely the shallow failure style and geometry described for
White Sands (Jaksa et al 2013) and the failures that occurred at Thiele Reserve (De Carli et al 2013; see also De Carli et al
this volume). The more important findings of this modeling are the following:
A) the failure style is very sensitive to the undrained shear strength of the near surface muds which form the
banks. The lower end of the range of undrained shear-strengths (8.5 kPa) indicated by CPT tests favors the formation of
shallow, wide-diameter, slip circles and produce a failure-surface geometry that approaches a planar slide (Figures 3A,
3B, 3C). Stronger near-surface muds (10 and 11.5 kPa) favor the formation of more deeply located, smaller-diameter, slip
circles (See Figure 3D).
B) Lowered pool levels destabilizes both a bank with normal channel margin geometry and banks where an
embankment has been formed. In the normal margin geometry case failures would be expected at the maximum
lowered pool level of two metres below normal operating. A bank with a constructed channel margin embankment or
levee becomes unstable once the pool level is lowered by one metre.
C) The presence of a near channel margin embankment significantly reduces the factor of safety and tends to
induce deep-seated rotational failure (see Figure 3D) except when a very soft near surface mud layer is present (see
Figure 3C). The presence of an embankment might be expected to compact and strengthen the underlying sediments
(Foott and Ladd 1981) however, the fact that the larger and more problematic failures have occurred at sites where
embankments are present indicates that this effect must be overwhelmed by the head-loading of the potential slide by
the embankments.
Lowered Pool
Lowered Pool
Normal Pool
Bedrock
Monoman Sand
Soft Mud
Normal Pool
A
B
C
D
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 282

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Discussion
The geomechanical models presented above demonstrate that the majority of the Lower Murray’s banks and channel
margin would have probably remained stable and resisted collapse if the river pool level had not withdrawn to below
sea-level elevations during the Millennium Drought. The pool levels fell to these levels: firstly, due to the lack of water
delivered to the Murray from the upstream catchment; and secondly because the barrage located at Goolwa was
deployed to prevent seawater incursion in to the Coorong and then further upstream into Lake Alexandrina and Lake
Albert. The lowering of the Lower Murray’s river pool level was slow and it took approximately six months for the river
level to reach this nadir. This represents a slow drawdown due to an anthropogenic intervention. The other major factor
that has contributed to the collapses is the predominance of the very soft clays that form the banks and channel
margins. These are particularly weak materials, and it obvious from the core photographs presented in Figure Two that
the laminated clays are not typical of floodplain materials. In particular the absence of intercalated silts or sandy lenses
in cores leads us to suggest that these clays are probably lacustrine rather than riverine in origin. This probability will be
investigated in forthcoming works such as Hubble et al (2014). Note that it is also suspected that the impermeability of
the soft muds will influence the pore-pressure distribution within the muds which has implications for channel margin
stability but investigation of this phenomenon is beyond the scope of this work (and the available space). Other work
(Jaksa et al 2013; De Carli et al 2013; and De Carli et al this volume) has demonstrated that the Murray has been
entrenched in its current planform by incision of the channel thalweg into the soft muds. The normally expected
response to deep incision of a thalweg, is widening of the channel which in this case has occurred by scour and mass
failure of the banks. Processes which have dominated channel widening in other rivers both in Australia (e.g. Rutherfurd
2000) and overseas (cf Schumm 1977). In this sense the anthropogenic factors contributing to the bank failures
described here have exacerbated and amplified the natural riverine processes.
Conclusions
1) Slope stability modeling has been used to demonstrate that the lowering of the water surface during the
Millennium Drought to unusually low, below sea-level elevations probably caused widespread mass failure of
the Lower Murray’s banks and channel margins
2) Constructed levees, embankments or man-made fills placed at or near the water’s edge increased the likelihood
of bank failure.
3) The very soft muds into which the Murray has incised to form the channel margins presents a situation that
favors channel widening by mass failure processes.
Acknowledgments
This study is supported by a Goyder Institute Research Program, Environmental Water Grant (Project E.1.8) that is an
initiative of the South Australian Government. The many University of Sydney students and community volunteers,
particularly David and Marie Mitchell, who ensured the success of the field program on the MV Breakfree are thanked
for their enthusiasm and assistance, as are the many riverfront residents of the Lower Murray towns who have allowed
easy access to their properties and shared their wealth of local knowledge with us. Richard Brown, Gareth Carpenter and
Jai O’Toole of South Australia Water are also thanked for their encouragement and support during the project.
References
ARUP, 2008. Assessment of soil cracking and river bank slumping in the lower Murray. Part1 and Part 2 Reports. ARUP
for South Australia, Department of Water, Land and Biodiveristy Conservation.
Balaam, N.P., 2010. XSlope Version 4.6, Users Manual. USyd. Center for Geotechnical Research.
Bishop, A.W., 1955. The use of the slip circle in the stability analysis of earth slopes. Geotechnique.
De Carli, E., Hubble T., Jaksa, M., Clarke S., Airey, D., O’Toole J., Carpenter G., Scientific Party M.V. Breakfree, 2013.
Riverbank collapse on the lower Murray River, recent phenomenon or long-term geomorphic process? Abstract
NH13A-1595, Fall Meeting, American Geophysical Union.
Coffey, 2012. Review of management options for four river bank collapse high risk sites. Final Report. Coffey
Geotechnics Pty Ltd (Adelaide Office).
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 283

7ASM Full Paper
Hubble et.al. - Geomechanical modeling of Murray River Millennium Drought bank failure
Foott, R., Ladd, C.C., 1981. Undrained Settlement of Plastic and Organic Clays. Journal of the Geotechnical Engineering
Division, American Society of Civil Engineers.
Galay V.J., 1983. Causes of river bed degradation. Water Resources Research
Jaksa, M., Kuo, Y., Hubble T., De Carli E., 2013. Riverbank collapse in the lower Murray River, Knowledge Gap Report.
Goyder Institute Research Project E.1.8.
Hubble, T.C.T., Rutherfurd, I.D., 2010. Evaluating the relative contributions of vegetation and flooding in controlling
channel widening: the case of the Nepean River, southeastern Australia. Australian Journal of Earth Sciences.
Hubble, T.C.T., 2010. Improving the stream of consciousness: a nomenclature for describing the factor of safety in river
bank analysis. Ecological Engineering.
Hubble, T., Hamilton, R., De Carli, E., Penny, D., Petley, D., 2014. The Lake Mannum Muds, a lacustrine origin for the
lower Murray Gorge’s uppermost valley fill. Australian Earth Sciences Convention.
Morgenstern, N.R., 1963. Stability charts for earth slopes during rapid drawdown. Geotechnique.
Rutherfurd, I.D., 2000. Some human impacts on Australian channel morphology. In: Brizga,S., Finlayson, B., (Eds.),
River Management – The Australasian Experience. Wiley & Sons, Brisbane.
Schumm, S.A., 1977. The fluvial system. Wiley, New York.
SKM, 2010. Study into river bank collapsing, Lower Murray River: Geotechnical Investigation Report. Sinclair Knight
Merz for South Australia, Department of Water, Land and Biodiveristy Conservation.
Thompson, C., Croke, J., 2013. Geomorphic effects, flood power, and channel competence of a catastrophic flood in
confined and unconfined reaches of the upper Lockyer valley, southeast Queensland, Australia. Geomorphology.
Twidale, C.R., Lindsay, J.M., & Bourne, J.A.. (1978). The geomorphology of the Murray River and Gorge in South
Australia. Proceedings of the Royal Society of Victoria.
USBR. (2001). Engineering Geology Field Manual, 2nd Edition, Volume I. United States Department of the Interior,
Bureau of Reclamation.
Hubble, T., De Carli, E. & Airey, D. (2014). Geomechanical modeling of the Murray’s Millennium Drought river bank failures: a case of the unexpected
consequences of slow drawdown, soft bank materials and anthropogenic change, in Vietz, G; Rutherfurd, I.D, and Hughes, R. (editors), Proceedings of
the 7th Australian Stream Management Conference. Townsville, Queensland, Pages 278-284. 284