The design of support pillar layouts in shallow South African mines has historically been based on comparisons of pillar strength to pillar loading, providing some acceptable factor of safety. For this evaluation of pillar stability, empirical strength formulae were compared to either tributary area or modelled loading.

Empirically founded pillar strength formulae are based on recorded cases of stable and failed pillars in specific mining areas. These methods inherently include area-specific behaviour of both the overburden system that loads pillars and the responding pillars. In applying empirically derived formulae to a project area, it is assumed that the relationship between the loading system and pillar behaviours is similar to that in areas for which the formulae were derived. However, the same pillar strength formulae appear to have been applied on most of the South African gold, platinum, chrome, and manganese hard-rock mines, all of which have substantially different loading environments.

Despite this possible incorrect application of empirical formulae, vast mined-out areas supported by pillars so designed have historically remained stable. However, it is not known whether these pillar systems are in a state of unstable equilibrium that will lead to system failure, or so stable that substantial amounts of ore are unnecessarily locked up in the pillars.

In this paper, a method for support pillar layout design that promotes the determination of area-specific loading system and pillar behaviour is described, making use of computer-based numerical modelling techniques. The method is then used to evaluate the potential for optimizing the pillar layout at a bord-and-pillar coal operation. The results indicate huge potential for optimization, but also the critical need for substantial rock property data and instrumentation for calibration.


Historically, support pillar layouts in shallow South African hard-rock and coal mines have generally resulted in stable workings. In these layouts, pillar dimensions were mostly derived using empirical pillar strength relationships, derived from back-analyses of stable and/or failed pillars. Despite the apparent success, pillar behaviour modelling using empirical methods is explained solely by a factor of safety (FoS). The description of pillar behaviour is more complex than just quoting a single number, and should also include reference to issues such as pillar stiffness, foundation failure/punching, yield point, and peak strength, as well as failure mode.

This content is only available via PDF.
You can access this article if you purchase or spend a download.