Modelling of Brittle Pillar Behaviour

Abstract

Brittle rock failure can have a dominant influence on the stability of underground excavations. It is therefore important that such material behaviour is realistically represented in any model used to simulate the rock mass response to mining. In this paper, the example of pillar failure is used to quantify the sensitivity of modelling results to the so-called strain-softening rate, as well as to element size. This strain-softening rate, in combination with the element size, determines the effective brittleness of the model material. In principle, it should be possible to calibrate a unique combination of element size and strain-softening rate that correctly reflects actual material behaviour. However, it is important to realise that failure localisation processes control the overall brittle behaviour. If failure localisation is not realistically represented or accounted for, even this calibration is invalid. In a complex situation, such as brittle pillar failure, which includes punching of surrounding strata, different localisation mechanisms can take place. In principle, each such localisation process would need to be calibrated individually. This paper discusses practical aspects of modelling brittle pillar failure and highlights common pitfalls.