Differentiating Gas Leaks from Normal Appliance Use

DNV has carried out an investigation into potential uses for smart gas meter data as part of Phase 1 of the Modernising Energy Data Applications competition, as funded by UK Research & Innovation. In particular, a series of calculations have been carried out to investigate the possibility of differentiating accidental gas leaks from normal appliance use in domestic properties. This is primarily with the aim of preventing explosions, but the detection of leaks also has environmental and financial benefits.
Three gases have been considered in this study:

• A representative UK natural gas composition.
• A blend of 80% natural gas and 20% hydrogen.
• Pure hydrogen.

An established method was used to calculate the gas outflow rate from leaks into buildings, downstream of the meter regulator. These gas leak rates were compared with the gas use by typical domestic appliances, and these leaks were considered in the context of the smart gas meter data that is currently available in 30-minute periods. A representative set of gas accumulation calculations was performed. It was found that the following three general cases apply to all three gases:

• Small holes of up to 1 mm rarely reach flammable gas/air concentrations for any gas, except under the most unfavourable conditions such as small volumes combined with low ventilation rates. These releases would likely be detected within 6 to 12 hours.
• Medium holes between 1 mm and 6 mm give outflow rates equivalent to a moderate to high level of gas use by appliances. The ability to detect these leaks is highly dependent on the hole size, the time at which the leak begins and the normal gas use profile in the building. The larger leaks in this category would be detected within 30 to 60 minutes while the smaller leaks could take several hours to be clearly differentiated from appliance use. This is quick enough to prevent some explosions.
• Large holes of over 6 mm give leak rates greater than any gas use by appliances. These releases rapidly reach a flammable gas/air mixture in most cases, but would typically be detected within the first 30-minute meter output period. Again, some explosions could be prevented in this timescale.

An event tree has been used to quantify the approximate success rate of detecting and preventing explosion events. Using the currently available 30-minute smart meter data and functionality it is estimated that around 7% of explosions could be prevented, with a slight variation depending on the gas type in use. Additional functionality could be included in the meter to improve this performance. Typically, one of the risk mitigation measures in isolation increases the explosion prevention rate to around 15%. If multiple improvements are made, then the detection rate is predicted to increase to around 30%.
An examination of detailed historical incident information suggests that the explosions that lead to fatalities or significant damage to houses are typically of the type that would be more likely to be detected and prevented. It is estimated that between 25% and 75% of the more severe explosions could be prevented, depending on which potential improvements are implemented.
Based on the conservative estimates of explosion prevention, a cost benefit analysis suggests that it is justifiable to spend between around £1 and £10 per meter installed to implement the proposed technology. This is based purely on lives saved, and does not take account of other benefits.