The detection of spontaneous combustion has traditionally relied on indicators sourced from laboratory testing or events in underground coalmines that occurred many years ago. This approach is flawed, according to Cliff, because there is no universal indicator for spontaneous combustion. Furthermore, often the conditions in mines from which historical indicators are derived, bear no similarity with modern underground coalmines.
“Historically, for example, spontaneous combustion events would often occur in the pillars of roadways and were detected by smell or a rise in CO make,” Cliff said. “Now the majority of incidents occur in the goaf some distance behind the longwall where there is no externally defined ventilation circuit. Thus it is unreasonable to expect that textbook definitions of indicators can be routinely applied without significant modification and testing for relevance.”
Cliff said recent events such as at Southland, North Goonyella and Dartbrook indicate this detection process is less than perfect, mostly due to the size of the area to be monitored and an inability to sample within goafs. A lack of defined airflows also hampers early detection.
“Indeed, due to the difficulties of monitoring for the presence of heatings in the goafs of modern longwall mines, there needs to be a shift from detection of a heating to detection to prevent a heating,” he said.
In each of the three cases previously cited, there was no way of knowing what caused that particular area of goaf coal to abnormally oxidise and not the millions of tonnes of other coal in the goaf all around it.
Circumstances at the point of combustion must just have been right for it to spread, Cliff said. “The initiation of the event in each case probably occurred months beforehand and the oxidation stewed away until conditions favoured acceleration. In two of the cases this was caused by sudden influx of additional air due to seal failures. In the third case it was probably simply the case that the longwall had been stationary for a number of weeks and air was able to continually flow to the heating site, under conditions that favoured abnormal oxidation.”
Because of the difficulties of detecting an active heating, Cliff argues the focus should shift to prevention and comprehensive monitoring.
Monitoring strategies defined by an early response should be triggered by such things as:
The detection of oxygen in areas of the goaf where it should not be. This does not immediately cause trouble but it will be the catalyst if this condition remains in place for any length of time. Remedial action to reduce the oxygen supply can avert a heating. Such action could include proactive inertisation, tightening of seals and reducing the pressure difference across the face of the longwall.
The ability of oxygen to pass into areas of particular coal in the goaf for longer than normal; eg if the longwall stops for any length of time or is reduced to slow production rates.
Pressure differences across seals that are not what is expected – this, of course, presumes that you know what to expect. Abnormal pressures differences often indicate leaking seals and air ingress into goaf areas.
Also, because mines often collect inadequate amounts of data from far too few monitoring locations, when something abnormal is detected, the situation is often serious and evacuation of the mine is the only option. Mines need to be aware of typical goaf seal behaviour including:
pressure differentials across seals and around goafs as a function of distance from the face and other factors such as the change in the pressure difference across the face;
gas evolution and derived indicators as a function of distance from the face; this is especially important where factors such as goaf drainage and or back bye ventilation is used to reduce seam gas impacts on the face; and
longwall return concentrations and derived indicators such as CO make as a function of operating parameters including size of goaf, rate of retreat, etc.
Characterising goaf behaviour would require an intensive effort initially until a benchmark could be established for expected behaviour, Cliff said.
Techniques to detect spontaneous combustion include the measuring of: carbon monoxide concentration and hydrogen concentration – both unreliable in isolation; ethane – a common seam gas not to be used as an indicator; of significant ethylene concentration – a reliable indication severe abnormal oxidation has occurred; CO make - only valid in roadways with defined, known ventilation; Graham’s ratio – but can hide a small intense heating in the effects of a large-scale low-level oxidation; and CO/CO2 ratio – beware of other sources of CO2 such as seam gas.
Cliff warned that inertisation techniques can distort parameters and give artificial readings because a number of computer programs calculate ratios using preset factors for such things as the ratio of oxygen to nitrogen in inlet air.
“In summary, prevention is better than cure,” Cliff said, “especially where there is no guarantee that a heating can be detected at a stage early enough to control it quickly and easily.
“Comprehensive monitoring systems need to be established to establish normal mine environment behaviour and understand the factors that can affect gas concentrations in all areas of the mine, including monitoring pressure differences around the mine, air flows and temperatures. Proper maintenance and personnel skilled in understanding mine monitoring systems and interpretation of mine atmospheres must support these systems.”
Based on the paper, “The Ability of Current Gas Monitoring Techniques to Adequately Detect Spontaneous Combustion”, presented at AusIMM conference COAL 2005. MISHC is the Minerals Industry Safety and Health Centre, Sustainable Minerals Institute, University of Queensland.
