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

net/publication/269196437

Analytical Study of Dolomite Sinkholes in Centurion, South Africa

Conference Paper  in  Geotechnical Special Publication · May 2014

DOI: 10.1061/9780784413388.066

CITATION READS

1 2,206

2 authors:

Daniel Avutia Denis Kalumba

University of Cape Town University of Cape Town
2 PUBLICATIONS   2 CITATIONS    78 PUBLICATIONS   185 CITATIONS   

SEE PROFILE SEE PROFILE

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

A Comparative Study of 2D and 3D Slope Stability Analysis using Shear Strength Reduction Finite Element Technique: Case Study of Sishen Open Pit Mine Slope, South
Africa View project

Multi-layer Interface Shear Testing View project

All content following this page was uploaded by Denis Kalumba on 30 April 2018.

The user has requested enhancement of the downloaded file.

 ANALYTICAL STUDY OF DOLOMITE SINKHOLES IN CENTURION,
SOUTH AFRICA

D. Avutia1 and D. Kalumba2

1
SRK Consulting, Johannesburg, South Africa, [email protected]
2
University of Cape Town, Cape Town, South Africa, [email protected]

ABSTRACT: Karst landscapes cover 15% of the earths’ topography with an

estimated 1.5 billion inhabitants residing on the land. Therefore restricting land uses
and preventing infrastructure developments on dolomite land may be deemed highly
impractical. As a result, infrastructure restrictions in karst areas have encouraged
geotechnical professionals into developing geological models to simulate the
magnitude of catastrophic subsidence experienced due to dolomite sinkholes.
Sinkholes in dolomite are triggered by the alteration of the existing groundwater
level within the karstic strata. In this study, the appraisal of sinkhole propagation was
facilitated with the geological data acquired along the Gautrain route through
Centurion, South Africa. This paper presents the conceptualized ‘angle of draw’ of
dolomite overburden layers into cavities, with Terzaghi’s arching in soil equation.
The analytical results illustrated constant vertical drawdown in the WAD and
incremental cavity propagation in the frictional chert residuum. The sinkhole
propagation results replicated preceding physical modelling and empirical findings.
This idealized analogy of sinkhole propagation compliments the quantitative
assessment of empirical data, limit analysis and expert judgement applied to estimate
dolomite sinkhole hazards.

INTRODUCTION

Dolomitic land represents a quarter of South Africa’s Gauteng province with

4 million people residing in this region (Watermeyer et al. 2011). The Gauteng
transportation system was enhanced by the construction of the Gautrain Rapid Rail
Link, connecting Johannesburg with Pretoria through Centurion. The Centurion
‘deep dolomite’ (bedrock depth > 20 m) section of the Gautrain was constructed on a
3km long viaduct structure, spanning 65 pier foundations. The stratum characteristics
of each pier foundation presented a unique geotechnical challenge, such as
undulating dolomite bedrock with cavities and the highly erodible weathered altered

Page 1
dolomite (WAD). The evaluation of the probable sinkhole development size was
required for the design and construction of each pier foundation. This study
attempted to calculate the propagation of dolomite sinkholes with Terzaghi’s arching
in soil equation, which was prompted by the requirements by the South African
National Standards SANS (1936) for the quantification of possible dolomite
sinkholes in the strata. Therefore the objectives of this paper are to illustrate the
‘angle of draw’ of dolomite residuum above a cavity to facilitate the quantification of
sinkholes at the surface.

DOLOMITE SUBSIDENCE STUDY

Dolomites from the Chuniespoort Group of the Transvaal Supergroup are

2500 million years old with a mean annual rainfall of 700 mm of the last century,
suggesting chemical weathering of the carbonate rock. The karstic strata of
Centurion is characterised by bedrock pinnacles and troughs overlain by dissolution
products such as chert gravels and WAD. The chert gravels near the surface may
possess adequate shear strength and stiffness characteristics, contrary to the
compressible and erodible WAD sediment. The particle structure of the WAD
sediment contributes significantly to the mechanism of sinkhole failure, as WAD
may exist in two distinct forms - laminated WAD (compressible) or massive WAD
(erodible) (Buttrick. 1986). The majority of dolomite sinkholes have been attributed
to the infiltration of mismanaged surface water Watermeyer et al. (2011), suggesting
the erosion of the massive WAD sediment into cavities in the stratum. Consequently,
defining the WADs’ particle hydrodynamic stability towards erosion is a
fundamental step during subsidence investigations. The displacement of dolomite
residuum into conduits is defined as the ‘angle of draw’ Fig. 1 to which a particular
mobilizing agency will become operative (Buttrick. 1992).

Fig. 1. Sinkhole ‘angle of draw’ ( Buttrick. 1992).

The geological mechanisms attributed to dolomite sinkholes have been discussed

by Venter (1981), Buttrick (1992), Schijf and Bryne (2007), who specified erodible
residuum and cavities, as critical stratum characteristic for sinkhole formation. These
stratum characteristics that facilitate the development of dolomite sinkholes have

Page 2
been explored through physical modelling (centrifuge) and analytical techniques to
determine the collapse load of residuum above a cavity. Physical modelling studies
performed in centrifuges have illustrated the weakly cemented residuum material to
fail as semi-rigid blocks forming steeply inclined inverted cone sections above a
cavity (Abdulla and Goodings. 1996). However, Drumm et al. (1990) utilised the
theory of plasticity with an associated flow rule to obtain the directional failure path
of soil at failure. The orientation of the slip surface was controlled by the material
strength, overburden thickness and cavity diameter.

The analytical equation used in this paper incorporates the Terzaghi (1936) arching
in soil calculation of the vertical stress on a yielding soil strip. The infancy of this
approach in the study of dolomite subsidence, required limit equilibrium assumptions
to be applied to the yielding strip above the cavity. The material weight was assumed
to be sufficient to produce failure, subsequently; no external loading was
incorporated in the analysis. The analytical study attempts to simulate the descriptive
observations and centrifuge modelling results achieved in preceding studies.

EXISTING DOLOMITE CLASSIFICATION METHODS

Several classification systems proposed by De Beer (1981), Venter (1981),

Wagener (1982) and Buttrick (1992) have contributed immensely to the evaluation of
dolomitic land in South Africa. However, only two of the most widely used and
appropriate dolomite classification concepts proposed by Venter (1981) and, Buttrick
and van Schalkwyk (1995) were adopted for the propagation analysis.

The Venter (1981) classification categorized dolomitic stratum by its geo-

mechanical behaviour and geological characteristics (engineering geological
homogeneity) to provide quantitative data to facilitate the utilization of karstic land.
Venter (1981) classification system groups the material strength, erodibilty and
geometry of the bedrock into inhibiting and inducing features relating to the
development of dropout sinkholes.

The method of scenario supposition by Buttrick and van Schalkwyk (1995) was
established in an attempt to evaluate the maximum potential size of a dolomite
sinkhole. The cavities identified in the strata were assumed to have sufficient
capacity to accommodate all the overburden sediment. The dolomite overburden
layers were grouped by their susceptibility to erosion and geological origin. For
scenario supposition a generalized ‘angle of draw’ is assigned to each material group,
which constitutes the metastable slope to which a particular mobilizing agency will
displace the material.

Page 3
 The following definitions form an integral part of the method of scenario
supposition by Buttrick and van Schalkwyk (1995) Fig. 2:

 Receptacles: Cavities present in the bedrock or an opening/ void in the

overburden layers.
 Blanketing layer: The overburden sediment layers directly above the dolomite
receptacles. The sediment is assumed to be responsive to erosion.
 Mobilizing agents: The ingress of surface water and the extraction of
groundwater are primary mobilising agents. However dynamic loading and
ground vibrations may serve as mobilising agents.
 Mobilization: The displacement of overburden sediment via a mobilising
agent.

Fig. 2. Idealized schematic for the method of scenario supposition (after

Buttrick and van Schalkwyk. 1995).

It must be noted that the maximum potential size of a sinkhole is a function of the
‘angle of draw’ and depth of the respective sediment below the surface. This concept
has been included in the latest legislature; Annexure B of the SANS (1936) for the
qualitative categorization of dolomite land amongst inherent sinkhole hazards.

The Venter (1981) approach does not elaborate on the effects of the groundwater
table in relation to the WAD residuum and cavities in the stratum. Yet the location of
the water-table in relation to the erodible residuum and cavities in the strata may
control dolomite instability. A shallow water-table may enhance the stability of the
dolomite strata assuming no erosion is experienced below the phreatic surface
(Warrick. 1987).

The method of scenario supposition by Buttrick and van Schalkwyk (1995) was
developed to estimate the maximum development space of a sinkhole, which is
suitable for qualitative zoning applications rather than quantitative geotechnical
design requirements of foundations. Instead of refining the ‘angle of draw’

Page 4
conventions in different dolomite layers, new approaches have developed towards
the quantitative risk assessment and the implementation of mitigation strategies on
dolomite sites.

The review of prevalent dolomite classification methods facilitated the assumptions

of the analytical approach for the propagation of dolomite sinkholes. Thus, the
displacement of dolomite residuum into cavities was calculated with Terzaghi (1936)
arching in soils equation and limit equilibrium analysis. Terzaghi’s calculation of the
vertical stress applied to yielding sediment and the approximated cavity diameters
enabled the determination of the cavity propagation in dolomite residuum. The cavity
propagation results from sediment displacement, attempt to emulate the Buttrick
(1992) ‘angle of draw’ by incorporating soil mechanics principles to illustrate the
mobilization of sediment into a cavity.

TYPICAL DOLOMITE PROFILE IN CASE STUDY AREA

The ‘deep dolomite’ bedrock section, beneath the viaduct structure of the Gautrain
route through Centurion is characterized by undulating bedrock and cavities in the
stratum. However the material above the bedrock maybe generalised into four typical
lithological materials, namely colluvium, chert, WAD and the occasional highly
weathered dolomite floater. The vertical extent and lateral range of these materials is
highly variable, however for the implementation of the soil arching analysis in this
investigation only the material above the cavity was considered. Externally applied
loading (surcharge) was not included in the scope of the study.

Fig. 3 presents a typical Chuniespoort chert-rich dolomite profile encountered

along the Centurion section of the Gautrain route. The materials at depths below 9m
were assumed to be moist.

Fig. 3. Typical Centurion dolomite profile above cavity.

The colluvium, chert and WAD material were assumed to possess engineering
geological homogeneity with autonomous unit weight, shear strength and lateral
earth stress coefficients values. Table. 1 summarizes typical dolomite strata values
retrieved from laboratory tests performed by Wagener (1982) and Buttrick (1986).

Page 5
 Table 1. Typical Dolomite Material Properties.

Although the WAD values may vary considerably within a cubic meter of karstic
strata, it was not the intention of this work to explore the particle orientation and
structural fabric variation of the WAD material.

APPLICATION OF TERZAGHI’S SOIL ARCHING

The method of scenario supposition in Fig. 2 forms the basis for the incorporation
of Terzaghi (1936) equilibrium of a yielding strip of sediment calculation. The
arching in soils Eq. 1 calculates the vertical stress development above a known
width, which was used in the primary calculations to ultimately determine the
drawdown angle of the respective material above a cavity.

Chevalier et al. (2007) performed trapdoor laboratory experiments that indicated

the theoretical Terzaghi equation results over-estimate the vertical stress on a
yielding strip at depth/width ratio’s greater than 1. Iglesia et al. (2013) confirmed that
the theoretical and experimental (Centrifuge test) vertical stress correlation results of
soil arching were most reliable for (z/B) = 1 with the incorporation of Rankine’s
lateral earth stress coefficient into Eq. 1. The following notations were applied in Eq.
1, cavity width B (m), thickness of slice z (m), unit weight γ (kN/m3), angle of
friction (Ø ), cohesion c (kN/m2), surcharge pressure q (kN/m2) and the lateral earth
stress coefficient K (unit less) parameters. However, the Terzaghi (1936)
assumptions of a constant lateral earth stress coefficient of (K = 1) and the assumed
mobilised shear zone of (Z/B = 2) were not incorporated in the analysis, due to the
conservatively high theoretical vertical stress values achieved from these
assumptions.

= . [1 - ]
(1)

The incremental resultant vertical stress was calculated at 1 m depth intervals from
the existing cavity to the surface. The values in Table. 1 were incorporated into
Terzaghi's arching in soil equation to determine the resultant vertical stress on a

Page 6
yielding strip. These resultant vertical stress vs. depth values were then plotted in
Fig. 4 to establish the linear or polynomial relationship dependent on the respective
material. The resultant vertical stress x (kN/m2) maybe attained by substituting the
known depth y (m) into the equations in Fig. 4.

Fig. 4. Vertical stress from Terzaghi (1936) soil arching equation.

Assuming the material was displaced as semi-rigid block at a constant force,

facilitated the equating of the equations in Fig. 5. This limit-equilibrium principle
was then applied to determine the trajectory of the width at a fixed z/B ratio of 1. Fig.
5 illustrates how the Terzaghi (1936) vertical stress values over a predetermined
width maybe idealized as a semi-rigid block being displaced into an opening.
Assuming the semi-rigid block is displaced at a constant resultant force F (kN)
enables the top and bottom resultant forces to be equated. The following notations
are used in Fig. 5, vertical stress at the bottom QZ1 (kN/m2), vertical stress at the top
QZ2 (kN/m2), width at bottom B1 (m) and width at the top B2 (m).

Fig. 5. Semi-rigid yielding strip in equilibrium.

SUBSIDENCE RESULTS AND DISCUSSION

The analytical results of dolomite subsidence were plotted to illustrate the effects
of material strength and weight on the propagation of dolomite sinkholes. In Fig. 6,
the WAD material produced near vertical drawdown angles, which may be attributed
to the low frictional strength caused by prolonged erosion of the dolomite rock. The
moist chert and WAD interface showed a reduction in the incremental resultant
vertical stress due to the increased cohesion of the WAD in comparison to the low
unit weight of the material. The stresses at each interface reduce the original width of
the sinkhole.

Page 7
 The incremental cavity propagation was prevalent in the chert and colluvium
frictional material. The colluvium material located at shallow depths of (0 m to 5 m)
possesses minimal cohesion which promotes the rapid increase of the cavity width.
The increase in initial cavity width (B) reduces the drawdown angle of the frictional
material, which suggests that small cavities propagate at a slower rate.

Fig. 6. Sinkhole size vs. depth.

The drawdown results are plotted in Fig. 7 which shows that the increase in
cohesion values of the respective materials raise the drawdown angle of the strata
approaching near vertical 90 . At cohesion values greater than 20 kN/m2, the angle of
friction has minimal influence on the drawdown angle. The measurement of the
angle of draw was calculated with the horizontal axis at 0 , counter-clockwise to a
vertical axis of 90 .

Fig. 7. Drawdown angle vs. material Cohesion.

The general results and shape of the sinkhole propagation resemble the proposed
‘actual sinkhole development’ in Fig. 1 and the steeply inclined inverted cone section
proposed by Abdulla and Goodings (1996), but the colluvium exaggerates the
drawdown angle near the surface. No externally applied loading was included in the
study so the lack of cohesion was not exploited to illustrate puncturing shear failure
in the colluvium. This alteration would result in minimal increase in the cavity width
of the sinkhole in the colluvium geological unit, thus reducing the sinkhole size at the
surface.

Page 8
 The material interface boundaries lead to a reduction in the incremental resultant
vertical stress on the yielding strip, which correlates to a reduced cavity width. This
isolated result may be ignored due to the rational analogy of a yield strip of soil
remaining constant or increasing in width towards the surface.

CONCLUSION

The prevalence of karstic land in the Centurion district and the rapid development
of infrastructure have amplified the need for understanding the behaviour of WAD
residuum and the propagation of dolomite sinkholes. However, recent studies of the
sinkhole hazard have been heavily reliant on the statistical analysis of empirical data,
which may not be available at alternate karst sites. So the quantitative analytical
method described in this paper represents one of the preliminary studies
incorporating soil mechanics principles into the propagation of dolomite sinkholes.

The rigorous analytical results achieved with this approach may be utilized to
calibrate laboratory analysis of subsidence in centrifuges. This idealized analogy of
sinkhole propagation compliments the quantitative assessment of empirical data,
limit analysis and expert judgement applied to estimate dolomite sinkhole hazards.

ACKNOWLEDGEMENT

The authors wish to acknowledge the kind permission given by the Gauteng
Provincial Government, Bombela Concession Company and Aurecon (SA) to
publish the data in this paper.

REFERENCES
Abdulla, W.A & Goodings, D.J. (1996) “Modeling of sinkholes in weakly
cemented.” ASCE Geotechnical Engineering, Volume 122: 988-1005.

Buttrick, D.B. (1986) “Wad and Ferroan Soil Developed in the Dolomitic area south
of Pretoria.” Pretoria: M.Sc Thesis. University of Pretoria.

Buttrick, D.B. (1992) “Characterisation and approriate development of sites on

dolomite. “Pretoria: Ph.D Thesis, University of Pretoria.

Buttrick D.B, Van Schalkwyk A. (1995) “The method of scenario supposition for
stability evaluation of sites on dolomitic land in South Africa.” South Africa
Institution of Civil Engineering Journal, 4h Quarter: 9-14.

Page 9
 Chevalier B., Combe G., Villard P. (2007) “Experimental and numerical studies of
load transfers and arching effect in trapdoor problem.” Laboratoire Sols,
Solides, Structures. Grenoble, France:.

de Beer, J.H D. (1981) “Evaluation of dolomite sites.” Seminar on the engineering

geology of dolomite areas. Pretoria: University of Pretoria.

Drumm E.C, Kane W.F, Yoon C.J. (1990) “Application of limit plasticity to the
stability of sinkholes.” Engineering Geology, Elsevier Science Publishers,
Amsterdam. vol 29: 213-225.

Iglesia G.R, Einstein H.H,Whitman R.V. (2013) “Investigation of Soil Arching with
Centrifuge Tests.” Journal of Geotechnical and Geoenvironmental
engineering, no. July: 1-42.

SANS, 1936-2. (2013) “Development of dolomite land - Part 2.” Geotechnical

investigations and determinations.

Schijf J., Bryne R. (2007) “Progressive Dolomization of Florida limestone recorded

by alkaline earth elements concentrations in Saline, Geothermal, Submarine
springs.” Journal of Geophysical Research (Marine science faculty), Vol.
112: 1-17.

Terzaghi, K. (1936) “Stress distribution in Dry and Saturated Sand above a Yielding
Trapdoor.” First International Conference on Soil Mechanics and
Foundation Engineering. Cambridge: Havard University:307-311.

Venter, I.S. (1981) “Evaluation of dolomite areas.” Pretoria: Seminar on the

engineering geology of dolomite areas. Pretoria: University of Pretoria.

Wagener, F. (1982) “Engineering construction on dolomite.” Durban: PhD Thesis,

Department of Civil Engineering, University of Natal.

Warrick, D. (1987) “Shallow Ground-water tables and their relation to Dolomitic

Stability.” Pretoria: S.A Geological Survey Unit, University of Pretoria.

Watermeyer R., Buttrick D.B ,Trollip N.Y.G, Gerber A.A ,Pieterse N. (2011) “A
Performance based approach to the development of dolomite land.”
Evironmental Earth Sciences, Volume 64, Issue 4:1127-1138.

Page 10

View publication stats