Degree Name

Doctor of Philosophy


Department of Civil and Mining Engineering


This thesis describes the development of physical models specifically to predict the behaviour of strata and artificial supports in the vicinity of underground mining openings. The application of the technique was initially designed for coal mining but was extended to include metalliferous mining structures.

The factors which caused the observable effects underground were either unknown or improperly understood. Physical models were considered, and later proved, to be capable of displaying the breaking behaviour of the strata, leading to methods to control it or avoid problems so that mining could take place efficiently, safely and economically.

At the time of commencement of this research there was little prior information to establish and extend modelling techniques. Therefore it became necessary for the author to envisage all possible applications and to design test rigs, hydraulics for load application and modelling methods thus providing the greatest flexibility in operation.

The thesis includes sections dealing with modelling theory, design philosophy, modelling techniques and the acquisition of data such as rock properties and strata stresses from bore cores or in the mine, together with an extensive testing programme for the development of model materials to simulate mine rocks. Examples by way of case studies and the conclusions derived are given. There are various types of models of which true models were considered the only satisfactory category for the purposes envisaged.

Rig design was influenced by the strengths of the mine rocks, selection of model materials, geometric and strength scale factors. Maximum rig loads were affected by space limitations and design criteria, dictated by the appropriate structural design codes. Following the construction of test rigs and some models it was found that there were factors which were not common to model and prototype. Limited physical dimensions of models, geometric scale, opening dimensions compared with platen sizes, friction between models and their test rigs, platens and other surfaces became vital and significant problems which had to be overcome before accurate predictions became possible. These deficiencies or apparent departure from true modelling conditions first required recognition of their existence followed by methods to ameliorate or avoid certain problems. Despite considerable devotion to the problem of friction there always remained problems relating to overstressing or understressing a model. Therefore, the problem of friction has been given detailed attention in the thesis but in each case, new developments have been discussed together with their various ramifications.

Model materials were found to be highly abrasive and even took their toll on tungsten carbide tipped tools. Dissection of complex models after completion of testing was achieved using large crosscut saws with water flushing. Mining of models required innovative methods. Electric detonators were used to progressively mine coal or metalliferous pillars while the model was under load. Contrary to popular opinion, testing of specially constructed sections determined that numerous detonators were required to be fired simultaneously for the desired results and without damage to other parts of the model. Flexible, fibre-optic devices were used through tubing built into the model during construction to observe progressive failure. Strain gauges, adhered to surfaces of openings, recorded changes as various portions fractured or failed. Simple masonry tools and steel tubing with flared, segmented ends were ideal for extracting model coal seams. Pneumatic tools were also used.

In order to evaluate the accuracy of prediction, models of known situations were constructed and tested. Later, models which had been constructed, tested and results reported were found to have predicted the behaviour of the prototype with acceptably good accuracy. For example, in the prediction of longwall caving, model tests were better than 90% of actual prototype results in features such as caving span and support pressures. Other conclusions derived from model tests showed that pillars greater than a certain size had no deleterious effects upon roadways. Results of this nature were subsequently confirmed by mathematical modelling, specifically finite element analysis and it is an important fact that the results from physical models which predicted the behaviour of underground structures were able to correct and refine computer programmes. Physical models, acting in an intermediate role enabled precise refinements since all the necessary properties (input data) such as external stresses, material strengths, geometric details and resultant closures or deformations were available. Many of these properties were, and still are unable to be derived from the mine.

Physical models not only display a realistic appearance of the structure being simulated but have been shown to accurately predict the behaviour of the prototype.