Design charts for hard rock pillars that are based on the actual
stability data can then be considered as alternative design tools and
starting points before a more sophisticated engineering analysis is
conducted. These charts have been proposed over the years [2–9]
where the latest one included all stability compiled previously and
suggested “The Confinement Formula” to generate pillar stability
charts. Research reported in this paper utilised the database used in
Ref. [9] and further analysed the data by implementing the multinomial
logistic regression that is basically a similar approach to the
binary logistic regression for coal pillars that was given in Ref. [10].
The main difference of these two approaches lies in the stability
data. In Ref. [10] the stability data were grouped into two, which
were stable and failed, whereas in this research the stability data
were extended into stable, unstable, and failed as shown later in
this paper.