Australia operates some of the safest mines in the world - and the picture is improving.

Over the past 15 years, the annual number of 'lost time injuries' has been reduced by a factor of about four across both the mineral industry and the underground coalmining sector, according to figures from the Minerals Council of Australia. But that doesn't mean that the battle for safety is won.

Between 1999 and 2002, the Minerals Council reported that 10 people died in Australia's underground coalmines, and in 2001-02 there were more than 500 lost time injuries. One of the most common causes of these events was rock falls - face, roof and rib falls that can occur at various stages of longwall mining operations.

These falls also cause significant production losses and delays. A day of unexpected longwall stoppage at a large mine could cost a mining company up to $1 million in lost coal production.

So it's no wonder that the industry has given strong support to a CSIRO-Japan Coal Energy Center project to develop the capacity to predict roof falls with a real-time monitoring system of the rock strata above the mining area.

"We want to determine the proper warning signs of rock instability," says Hua Guo, the CSIRO principal geotechnical engineer who is leading the project.

"But more than that, we want to be able to collect data and carry out real-time interpretation, so that we can provide information in a timely fashion to help mine operators make decisions. We hope the system will have a predictive capacity - so that you would know before something happens and be able to take remedial measures in time."

Preliminary research found three common warning signs that could be used to predict a roof fall - fracturing in the overlying rock, which can be picked up as microseismic vibrations (see earthmatters, December 2003, Page 16); deformation of the roof itself, usually measured by displacement sensors known as extensometers; and stress concentration in the roof, detected by stressmeters.

The concept, which dovetails neatly into plans for CSIRO's new mines communications system, NEXSYS, is to develop a warning system that integrates data from all three sources.

Clearly this warning system has to be reliable, but it also has to be automatic, to reduce reaction time in an emergency. So development of sensors that can report remotely - without staff having to read them manually - is a significant part of the project.

In particular, Dr Guo and colleagues have been developing an extensometer for measuring roof displacement automatically and remotely. One type of extensometer commonly used sits in a borehole. It consists of a series of rods, each of which is anchored into the rock at one end. The rods are of different lengths so the anchor points can measure the movement at different heights, up the borehole.

At the other end the rods feed into an instrument known as a potentiometer that changes electrical resistance in proportion to displacement. In other words, the movement of the rod with respect to its anchor is reflected in variations in electric current moving through the potentiometer circuits.

Traditionally such extensometers were read manually. "People needed to go to the borehole," said Dr Guo. "Not only does this take time, but there are less and less staff available in mines to do so. And when roof conditions are bad, it's better to take the reading remotely."

So the team has designed an electronic system in which the potentiometer is in wireless 'contact' with the mine's communication network.

Readings are normally transmitted to the computer on the surface at preset times - once a minute, for instance. There, they are stored and analysed. If rock movement is detected, the system can demand an increase in the frequency of readings, to monitor the trends. It warns the mine operator at pre-determined levels of displacement.

The design of the extensometer has been completed and a working laboratory version built. Dr Guo says the team plans to have a production model developed by June. It will be battery powered, enclosed in a steel box, and needs to be intrinsically safe - that is, designed to operate in underground coalmine environments.

The design approach to the wireless communication system is applicable to other instruments, such as sensors that monitor the closure and tilt of the longwall support shield (chocks).

The team is also developing software which will analyse data from the extensometers, and integrate readings from the microseismic array and stressmeters to give a more complete picture of what is happening in the rock strata above the work areas of the mine.

The underground models of the wireless extensometers and longwall shield sensors are scheduled to be finalised for intrinsic safety approval processes in Queensland and New South Wales for the second half of 2004."There are 26 longwall mines in Australia where this system would be appropriate," says Dr Guo.

He says the real time sensors and data interpretation system could also be used for other purposes such as monitoring civil engineering structures like bridges and tunnels.