Engineering and monitoring behind development on dolomite

October 3rd, 2019, Published in Articles: PositionIT, Featured: PositionIT

Lakeside Office Development is a nine-storey commercial office building in Centurion, and successfully stands on Inherent Hazard Class 8 dolomite, with four known cavities, and a highly-fluctuating rock profile. This requires robust design and geotechnical planning for blasting, excavating, and dynamically compacting the underlying material. The project encompassed a measurement and testing regime to allow for the early prediction of propagating sinkholes as well as a strict water-management system, not to mention an innovative monitoring system ensures an early warning system.

Lakeside Office Development is a nine-storey commercial office building with a GLA of 18 677 m2. Located opposite the Centurion Gautrain station, it successfully stands on Inherent Hazard Class 8 dolomite, with four known cavities, and a highly-fluctuating rock profile. The design required sufficient robustness to adapt to the known and unknown variabilities of the ground, and the potential 15 m wide sinkholes that could develop. A 2,25 m thick concrete raft solution was opted for, in conjunction with a geotechnical system of blasting, excavating, and dynamically compacting the underlying material. The design solution also encompassed a measurement and testing regime to allow for the early prediction of propagating sinkholes. A strict water-management system was implemented to prevent water ingress, the main instigator of dolomitic sinkhole propagation. This project used the mega-structural element of the raft to ensure robustness against severe accidental loading, while the innovative monitoring system ensures an early warning system.

Dolomitic rock

Dolomite is a sedimentary carbonaceous rock similar to limestone. As with limestone, caves or receptacles can form within the dolomitic rock due to interaction with weak carbonic acid.

The unweathered bedrock is typically overlain by partially weathered, segmented bedrock, then by totally weathered and low-strength residual material consisting mainly of weathered altered dolomite (WAD), chert and iron oxides [1].

Dolomite is notorious for ground instabilities which would naturally occur over time due to the nature of dolomite dissolution. Significant effects at natural ground level are noticeable as sinkholes propagate from bedrock due to triggered stability events in the soil, such as fluctuations of water tables (draw downs and leaks, etc.)

Site identification

The site is situated on prime land within a modal hub opposite the Centurion Gautrain station. The Gautrain is a high-speed rail system linking four major cities in Gauteng. The site is bounded on four sides by a road, two commercial office buildings, and Centurion Lake.

The development prior to the New Lakeside Offices had to be demolished, as it showed indications of structural damage. The demolished building occupied approximately one third of the erf size. Originally, three office blocks had been planned for the site in 2005, but the geotechnical report, conducted in 2005, revealed highly-variable dolomite, with several potential cavities across the site. Several deep pockets of WAD, reflecting total air-loss during drilling, were also identified. At the time, these were classified as Risk Class 6 zones, in terms of the dolomite assessment criteria then applicable. As such, the initial development on the site was limited to just one of the planned buildings.

In 2013, a further geotechnical investigation was conducted. This increased the extent of drilling on-site to a total of 47 percussion-drilled, evenly-spread boreholes. Based on the depth-to-bedrock, the air-loss instances, rate of drilling advance, and interpretations of the mobilisation potential and overlying horizons, the site was categorised into three inherent hazard classifications (see Fig. 1): IHC <= 5 (low to high risk of small sinkholes < 5 m); IHC 6 and 7 (high risk of medium and large sinkholes 2 to 5 m and 5 to 15 m respectively) with instances approaching IHC 8 (high risk of a very large sinkhole > 15 m) [2].

Fig. 1: IHC Classifications (Red Class 8; Orange 6-7 and Green =<5).

In 2016, a proof-drilling investigation was conducted. The purpose of this investigation was to increase the number of borehole logs to 106, in an attempt to obtain a “best possible” representation of the ground profile (Fig. 2).

The coordinates and depth-to-rock obtained from each borehole log was used to create a Civil 3D surface that used a linear interpolation model to predict the rock between the logs. Several different interpolation profiles were attempted, but the natural profile of dolomite pinnacles does not adhere to a smoothed profile.

Rock profile modelling

The accuracy of the rock-level model was updated during the blasting process, with the blasting drill logs assisting in modelling the high-lying rock. The rock-profile model developed, illustrated clearly that the site ground conditions were highly variable, with steep pinnacles. In one example on-site, two adjacent boreholes (situated 3 m apart) picked up bedrock at -1 m below natural ground level (NGL) and -45 m below NGL respectively.

This model provided an estimate of the quantity of rock to be blasted. These predictions allowed preliminary costing exercises to be conducted to minimise the risk to the client, without spending exorbitant amounts of money upfront to drill the whole site. Additionally, the model was used to assess that the blasted rock and soil quantities were captured accurately during construction, limiting the potential for disputes at final account stages.

Another advantage of having the Civil 3D model was so that the designers could plot and map out canyons within the site. This allowed the design team to rationalise that only a 15 m wide sinkhole could potentially occur within the site boundaries. If a bigger sinkhole was likely to occur, it would be more linear in nature. The structure would still be able to accommodate this type of sinkhole.

Fig. 2: Interpolated representation of dolomitic bedrock profile on-site based on borehole logs. The proposed founding level is indicated as an olive-coloured plane.

Gravity survey and continuous surface waves

The designers were already in possession of a gravity survey conducted prior to the previous development, but this only provided a potential indication of the depth-to-rock over a third of the site. Another gravity survey was conducted in 2016 to map out the rest of the erf. Unfortunately, along with the resolution of the surveys received, and the nature of the dolomitic ground conditions (very stiff chert bands, dolomite floaters, etc.), these surveys could only indicate areas of stiffer material, but could not distinguish between bedrock, floaters, and the very stiff chert layers, and were therefore unreliable to pick up the localised intruding pinnacles.

Continuous surface waves (CSW) were conducted in three strategic areas of the site to ensure that there were minimal variations between the assumed stiffness profile compared to the actual stiffness profile over the development. Due to the limitations of CSW tests, where testing can only be conducted to a depth of 20 m, and the influence of high-lying rock on the stiffness, the following factors were taken into account in determining the strategic locations of where the three tests were to be performed, as well as obtaining a variable stiffness indication over various profiles:

  • Depth-to-rock in borehole logs in that specific area had to be 30 m and deeper.
  • Each area was attempted to best represent different thicknesses and depth of different horizons, i.e. a high-lying chert horizon with thicker WAD horizons, compared to lower-lying chert horizons with thinner horizons of WAD, etc.

Site development

Two main considerations resulted in terms of site development:

  • How to span the potential 15 m wide sinkhole.
  • How to found the structure in such a way so as to limit the differential movement and moments due to the highly-varied foundation depth.

Fifteen alternative concept designs were modelled by the structural and geotechnical teams. Ground engineering solutions included, jet grouting, compaction grouting, dynamic compaction, soil rafts, and stone columns.

Structural foundation alternatives encompassed concrete rafts, piling solutions (various types of piles), beams spanning between pinnacles, and bridge systems formed from the superstructure.

These solutions were considered both independently, and in combination with each other. A risk-mitigation system, balanced against cost-effectiveness and returns, was used to evaluate the concepts, and isolate the preferred way forward. This method provided the most predictable cost expenditure, while still ensuring the safety and reliability of the structure as a whole.  Each option had advantages and disadvantages, but the final design combined a number of geotechnical and structural solutions.

A 5 to 8 m deep zone of ground improvement was chosen, consisting of the removal of the rock pinnacles, replacing unsuitable pockets of in-situ material, and dynamically compacting the footprint of the building. A structural concrete raft was then cast on top of the geotechnical platform.

Geotechnical platform development

In order to obtain a flat working platform to cast the raft on, the site was excavated to the anticipated founding level (ranging from 7 to 1 m to below the sloping NGL).

All bedrock, in the form of pinnacles, ridges and floaters, was removed for the first 5 m (Fig. 3). In the locations of known high-lying rock, obtained from previous drilling investigations, this rock was drilled and blasted in a highly-controlled environment due to blasting taking place in a densely populated and developed urban area. To limit the potential fly rock, a 1,5 m minimum overburden was maintained during blasting operations, with very stringent overburden particle size limitations in place.

Blast and seismic monitoring

Due to the blasting process being conducted as close as 12 m away from adjacent buildings, as well as the site being in close proximity to the Centurion Gautrain station, continuous vibration monitoring was carried out via three seismographs.

These seismographs were placed at each building adjacent to the Lakeside development, as well as on the boundary of the erf closest to the Centurion Gautrain station. The seismographs not only monitored vibrations and peak particle velocities during each blast, but also monitored vibrations during ambient conditions. This not only allowed the design team to ensure that blasting specifications and regulations were adhered to, but also allowed the team to optimise the blasting process by monitoring each blast continuously.

Numerous test drills to a depth of 6 m below platform level were conducted to ensure that no high-lying pinnacles were present in areas where the rock was anticipated to be deeper than 5 m.

It was aimed to blast the rock in as small fragments as possible so that, when excavated, minimal large fragments were removed to spoil, while some large fragments were retained to be placed on areas surrounding the building as mementos of what went into the design of the building, as well as creating an aesthetically-pleasing atmosphere in the recreation areas.

The small fragments were then blended with an on-site chert residuum mix, and recompacted as fill material. Simultaneously, any WAD in the upper 5 m was removed to spoil. In conjunction with using the smaller blasted rock fragments to be recompacted, building concrete rubble, along with removed reinforcement, and so on from the previously demolished building were also used mixed in with the chert residuum and recompacted as fill material.

Fig. 3: Excavations to remove high-lying rock.

Once the blasting process was complete, the entire building footprint was dynamically compacted (Fig. 4). A 30 m x 30 m test area, with primary impacts at a 10 m spacing, was established to determine how the site material reacted to the designed imparted energy. Based on the plate load test results of the trial region, the primary compaction locations were adjusted to 8 m x 8 m grid spacings, with the secondary 8 m x 8m intervals offset directly from the primary. The tertiary pass then utilised a 4 m x 4 m spacing with fewer impacts.

Fig. 4: Dynamic compaction rig.

A 14 t pounder was dropped approximately 4600 times on-site, followed by an 11 t ironing pounder for the smoothing phase. As intense vibrations can be imparted to the ground when the pounder impacts, the three vibration monitoring stations were maintained for the duration of the dynamic compaction stage. This allowed the project team to ensure that, even on the dynamic compaction work close to the existing buildings, the vibrations remained within the acceptable range.

Each impact hole was inspected for evidence of WAD, and the volume of the crater was measured to evaluate the performance of the underlying material. Where there was a large variation in compaction volumes, further excavations were undertaken to determine if pockets of highly compressible materials or high-lying rock were present.

The rock blasting, followed by dynamic compaction, resulted in the creation of a uniform soil mattress for 5 m above the highly-variable rock profile. It also assisted in softening hard spots below the raft, improving soft spots, and pre-collapsing any shallow cavities.

A total of 23 000 m3 of material was excavated, blended, and recompacted back into place. A total of 14 200 m3 of dolomitic rock was blasted during the earthworks contract in 33 independent blast events, with about 17 700 m3 of material finally removed from site.

Structural solution

Designed foundation

The final structural foundation system was a 2,25 m thick reinforced concrete raft, designed to span a 15 m sinkhole.

The raft was optimised to determine the most appropriate thickness, elevation, and number of suspended floors. A sensitivity analysis was carried out, varying the above parameters, and balancing deflections and moments under serviceability and ultimate limit states. This allowed total cost versus lettable area calculations to determine the viability of the development.

The raft was designed on a mattress of variable spring stiffnesses, adjusted for the depth-to-bedrock, the anticipated depth and stiffness of the WAD material, and the enhanced soil mattress zone to a depth of 8 m. This was updated based on blasthole logs and compaction logs, as well as in-situ plate load tests during the earthworks contract phase, as compared to the Civil 3D rock profile model. The raft was then verified against the potential sinkhole locations.

Raft construction

The raft footprint was approximately 82 x 68 m, necessitating 13 200 m3 of concrete, and 1500 t of reinforcement steel. The raft construction was divided into nine main continuous concrete pours of 1400 m3 each (Figs. 5 to 7), with a smaller tenth pour thereafter.

Fig. 5: Drone footage of Pour 1 construction, Pour 2 being prepared.


Fig. 6: Inside of the raft during Pour 1. First layer being vibrated.


Fig. 7: First pour with protruding reinforcement for Pour 3.

Construction considerations

A pour of this nature is not only technically difficult, it also has extensive construction and logistical considerations. For instance, for each 1400 m3 pour:

  • 18 trucks were used, in continuous rotation, supplied from three batch plants.
  • Accessibility and movement around the site was limited.
  • Two 40 and 60 m boom pumps were used, in conjunction with a static line pump (Fig. 7).
  • Prior to each truck depositing its contents within the receiving bucket of the pump, the slump and temperature of the concrete was assessed and rejected if not within specification.

In the South African summer, temperatures regularly reach 34°C, and mid-afternoon thunderstorms and showers are common. An acceptable concrete temperature for pouring purposes is considered to be less than 28°C.

Wind speed, temperature, and humidity were monitored with the aid of an on-site weather radar, where the rate of evaporation was calculated every 45 minutes during pouring, with misting nozzles laid out to reduce the evaporation rate when this became problematic. Lightning detectors, rain covers, and portable submersible pumps were on-site, and used daily to ensure the pours could continue fluidly when safe to do so.

Fig: 8. Pour 2, temperature probe no. 5 readings.

The thermal peaks and gradients in the concrete while setting were monitored continuously. The theoretical design predicted a peak of 62°C in the centre of the concrete, and no more than a 25°C differential temperature from the thermal centre of the concrete to the surface. This was monitored during the curing process by three sets of five thermal probes, placed at different levels and locations within each pour of the raft. The surface temperature was maintained by blanketing with thick polystyrene sheets and wooden shutter boards to prevent excessive surface cooling during the night.

Fig. 8 indicates the typical pattern of measurements obtained from the thermal probes. In the majority of cases, the maximum temperature reached was 58°C. On two pours, the middle sensor on the middle probe reached 65°C. In all cases, except where damage caused monitoring errors, the maximum thermal differential within the concrete was maintained below 25°C.

Long-term monitoring regime

In order to minimise any residual risk from the ground conditions through construction, and thereafter the lifespan of the structure, a monitoring and tracking regime system was implemented. This tracks a baseline, and then any subsequent ground, raft, and structure movement.

This was achieved by surveying location tags attached to the rafts at each column, surveyed bi-weekly. In addition, three-rod extensometers were installed at three locations on the raft. Each extensometer location was close to a previously-identified high-risk area. The three rods were anchored in interpreted key sectors of the profiles, based on the anticipated locations of the potential movement and the ground mobilisation.

In the short term, the extensometers provide high-precision spot settlement values and, in conjunction with the surveyed locations above, create an accurate depiction of the movement of the structure and underlying ground during construction.

Fig. 9: Vertical settlement along GL 7.

These measurements were plotted against the structural model anticipated settlements. This permitted interpretation of how the actual movements on-site were tracking against the theoretical design settlements. Fig. 9 illustrates readings along one of the column lines, obtained after all floors had been constructed, and while finishes were being installed. The settlement was tracking at approximately a 1/6th of the potential anticipated short-term settlement (3 mm maximum displacement compared to a 15,7 mm theoretical). This is likely due to the conservative approach in settlement prediction utilised during the structural design, as typically the potential worst case is scenario is of critical interest. The pattern locations of maximum settlement are following the overall trends in movement anticipated in the design, but also indicate that the structural and geotechnical solutions are currently minimising the risk of differential settlement.

In the long term, the extensometers are linked to the building management system (BMS), and serve as an early-warning signal for potential ground movement or instability at depth. If one rod were to disappear, the BMS would alert the relevant personnel that ground mobilisation (a potential sinkhole) has occurred. As the rods are placed at various depths, the progressive development of the sinkhole can be assessed in order to predict a timeline for the sinkhole to propagate to the surface. This will ensure that a solution can be implemented timeously to try mitigating any potential further progression of any sinkhole.

Technological systems

All consultants worked on Autodesk Revit to create fully-integrated 3D models of the building. Extensive clash detection was carried out in Navisworks to ensure that the site installation worked smoothly. Virtual reality rooms were used to check and assess the models – for example, to ensure that the reinforcement, detailed in 3D in Revit Rebar, was located appropriately, and covered the necessary locations.

Lakeside’s BMS is fully integrated, where every element of the building is connected back to the control room. Standard elements such as fire, gas, PV, electrical, security, and metering are all connected. In addition, the dolomite requirements led to additional services monitoring, over and above the usual.

Dolomitic sinkholes result primarily from surface water infiltration and reduction in water levels at depth. A key design principle of the project as a whole was the isolation of the ground surface from water ingress. To this end, all building services are fully contained within the structure, with concrete-lined ducts at entry and exit points. External services such as sewers, which are required to run at depth, are run in HDPE pipes within secondary sleeves. If a leak occurs, the secondary sleeve will redirect the flow to the closest manhole. These manholes are all installed with float switches to trigger an alert on the BMS for early leak detection.

The subsoil drains behind the retaining walls are expected to run empty except in the case of a burst municipality service. In this instance, the moisture sensors in the drains will identify the change in moisture conditions, and trigger an alert on the BMS. This is then used by the maintenance team to inform the local municipality of the burst service.

Figure 10. The final building after occupation.


The high-risk class dolomitic conditions under the New Lakeside Office made the development of a new office park very challenging. Detailed investigations, surveys, and models allowed for a cost-effective solution. Only through team collaboration from all aspects of the design and construction could a state-of-the-art, intelligent building be constructed that met the client’s requirements, while still being cost-effective.


[1] AC Oosthuizen, S Richardson. 2011. Sinkholes and subsidence in South Africa. Council for Geoscience Report number: 2011-0010. Council for Geoscience.
[2] SANS 1936:2012 – Development of Dolomite Land.

Note from the editor

This project showcases leading engineering talent and creative thinking, and it is worth noting that this project won the Consulting Engineers South Africa (CESA) Engineering Excellence Award 2019 for the project category: value greater than R250-million. The author, Kim Timm, was also awarded the Mentor of the Year 2019.

Contact Kim Timm, Aecom SA, Tel 012 421-3500,

Related Articles

  • Now Media acquires EngineerIT and Energize from EE Publishers
  • Latest GIS further supports electric and gas networks
  • EE Publishers 2020 Year Planners
  • To readers, customers, suppliers, staff and other stakeholders of PositionIT
  • Mozambique’s land administration modernisation