Opencast mining has been and continues to be a favoured method for the extraction of the
vast coal reserves in the Highveld of South Africa. Previously backfilled and restored open cast
areas are generally zoned for agricultural uses, but with growing pressures on land use, such
areas are increasingly being considered for the expansion of infrastructure and regional
development. Understanding the backfill permeability and hydraulic behaviour is therefore an
important component in defining the land use restrictions placed on a previously backfilled
area. Centrifuge modelling provides a means of better understanding the hydraulic behaviour
and measuring the permeability of opencast backfill under controlled laboratory conditions.
Based on a preselected backfill prototype, an appropriate centrifuge model was developed.
Using miniature pore pressure transducers, the pore pressures were measured at discrete
locations in the model during falling head tests in a geotechnical centrifuge. Using the
measured volumetric discharge, spacing between the transducers and the measured pore
pressures, the permeability of the backfill was calculated. Due to polarized opinions on the
scaling of permeability in a centrifugal field, a control model was tested at 1g and 23g to
validate this scaling law. It was demonstrated that the respective permeabilities calculated at
1g and 23g were effectively the same and that it is in fact the hydraulic gradient that is scaled
N times in the centrifuge. Knowing this allowed the calculated centrifuge permeabilities to be
directly related to the prototype represented by the model. To determine the accuracy of the
centrifuge model, the results of field percolation tests were compared to the results of an
analogous centrifuge model. There was no correlation between the results and it was not
considered meaningful to compare the results, as the model and percolation test site
(prototype) conditions differed significantly. To simulate the preselected prototype backfill
sequence, a model configuration that represented the geometry and material properties of the
prototype was tested at 35g (half scale) and 70g (full scale) in the centrifuge. The results of
the centrifuge model were used to make reasonable predictions on the long term permeability
and hydraulic behaviour of the backfill prototype. It was found that the permeability of the
backfill is likely to decrease over time due to consolidation settlement. The bottom of the backfill
sequence is expected to have the lowest permeability and the top is likely to maintain a higher
permeability. It was further demonstrated that the horizon interfaces acted as flow restrictors
and resulted in poor vertical permeability between the horizons in the backfill sequence.
Overall the centrifuge methodology provided a unique and efficient means of modelling the
long term permeability and hydraulic behaviour of the backfill sequence.