by Ross Seedsman, Seedsman Geotechnics

Recent work on longwall face weightings and longwalls in general is showing that the behaviour of overburden strata is much more complicated that previously thought. In the past few years, at least five Australian longwalls have experienced heavy weighting events or windblasts that have severely impacted on mine profitability. This has prompted a number of research projects, some completed and others recently commenced. A review of results to date suggests that many of the researchers are failing to adequately considered the nature of the rock mass in which mining takes place - a jointed and bedded sequence.

The nomogram on longwall conditions under conglomerates in the Newcastle Coalfield produced in 1997 deserves much more attention than it appears to be receiving. This nomogram relates face weightings to panel width and massive unit thickness. It highlights the need to consider the longwall problem in three dimensions:

Panel width - a horizontal dimension across the face line.

Massive unit thickness - the vertical dimension.

Periodic weighting - a horizontal dimension at right angle to the face line.

The need to consider all three dimensions contributes to the complexity of the geotechnical design of longwall faces, and this is made even more complex by adding the requirement to consider rock structure, rock strength and ground stresses. Seedsman Geotechnics approaches this design problem by making the key observation that any successful design method must consider panel width as a key variable. It has been conclusively shown at mines such as Newstan and Clarence that reducing panel widths is one of the few control options available to the longwall miner.

SGPL is developing a design approach based on panel width and this differentiates SGPL from other contemporary researchers such as CSIRO and also the mining literature that analyses vertical sections normal to the face line.

Fortunately, by explicitly considering joints we are able to simplify the design process significantly. Consider the case of longwall face aligned parallel to the dominant vertical joint set in the overburden (Figure A1). Looking down on the longwall panel, the roof is divided into a number of joint blocks of finite width. Consider the case of a rock unit that cannot span the panel. When the face line passes under one of the vertical joints, horizontal stress relief into the goaf means that this new joint block separates from the joint block further outbye. This new joint block has yet to fail because the majority of the block is supported by the coalface. The end result is that the longwall face is exposed to very little loading.

Now consider what happens as the face line approaches the next joint (Figure A2). The coalface gives little if any support to the joint block. The block must therefore span across the panel between the pillars. The stage is reached when the reduced restraint from the face is not sufficient and the joint beam fails. The vertical forces associated with the isolated joint block may be very high and be sufficient to cause new rock fracturing in the immediate roof rocks. An observer at the face sees new rock fractures even if the driving forces are related to joints. Note that the magnitude of the loading will depend on the thickness of the massive unit and the spacing of the joints.

From a design perspective, the spanning ability of rock beams can be readily analysed using jointed rock beam theories and the calibrations to mining experiences (such as the diagonal line on the Newcastle nomogram) have not required strength reductions.

To date, the design method has been used as a consulting tool at Newstan, West Wallsend, and Crinum and the predictions have been validated by subsequent longwalling.

This general mechanism can be extended to the case of mining oblique to the joints (Figure A3). The same stress relief occurs into the goaf that allows the problem to be assessed in two dimensions. The difference is that at one end of the face there may be light loadings because the effective (that is, diagonal) span is low, and at the other end of the face the loadings may also be low because the rock beam has already failed. The heavy weightings are concentrated in the middle of the face where the effective span approaches the failure span of the beam. Note that the weightings are of equal magnitude as for Figure A2, but are over a much-restricted length of the face. The model can be applied for the case of two conjugate joint sets.

The model locates new rock fracturing ahead of the longwall face as well as above the chain pillars. These locations are compatible with the microseismic studies being conducted by CSIRO.

It is important to realise that the location of some of the predicted events can be invisible to the microseismic arrays if the travel path for the event is through the goaf. The failure mechanism of the joint beams is compressive failure, with the induced failure surface heading over the goaf. This is the orientation reported by CSIRO from Gordonstone and Appin.

The model can also assist in the understanding and prediction of wind blasts. Wind blasts are associated with the delayed caving of thick units in the immediate roof. From a design viewpoint, delayed caving can be related to beams with a factor of safety against overall failure just less than 1.0.

Alternatively, for very thick units that do not fail, the underside of the beam towards the centre of the span is a zone where there is no lateral confinement. Should there be discontinuous bedding surfaces in this area, over time they may fail under the tensile stresses developed by their self- weight. The model predicts at least two different mechanisms for the source of precursor microseismic noise for windblasts. We understand that different microseismic patterns have been recorded from Moonee and West Wallsend, and we await eagerly the publication of the ACARP research.

In applying these concepts, the major issue becomes assessing the presence of massive units. The sedimentology of the overburden gives some indication, with heavy weightings and wind blasts having been found in association with braided river channels, barrier beach sands, and reworked marine units; significantly not point bar deposits. SGPL has had a great deal of success in the use of geophysical logs, and particularly the sonic log, to identify massive units in bore holes. Once this issue is addressed, the next challenge is to assess the spacing of the master joint sets so that the size of blocks in the overburden can be assessed. There is some initial evidence that the vertical lines in the Newcastle nomogram can be related back to some of the very early ideas about detached blocks and longwall support capacity.