Learnings

Outcomes

The main conclusions of the work were:

Currently the UK GS(M)R limits gas Wobbe Index (WI) to between 51.41 and 47.20 MJ/m3 during normal times. These limits may be extended to 52.85 to 46.50 MJ/m3 during a supply emergency. Almost no hydrogen addition is permitted. Appliances of category 2H and which are fitted with an ODS as a primary safety device are only tested with a reference gas (G20 - WI of 50.72 MJ/m3) to confirm effective ODS operation during normal appliance certification, hence the interest in this work. It has always been assumed that this was sufficient to demonstrate effectiveness with all the gases distributed in the UK gas network. In this work the 8 appliances have been tested with gases ranging from WI 45.56 up to 54.7 MJ/m3 and with gases containing various levels of non-methane gases. This is a very wide and complex range of gases.

The widening of WI bands would reduce or eliminate the need for costly and possibly wasteful ballasting of imported Liquefied Natural Gas (LNG). It would also permit use of low WI gas from the North Sea and ease the injection of biogas (of low carbon footprint) without having to add propane (of high carbon footprint). Additionally, it would allow the injection of hydrogen which would lower UK greenhouse gas emissions, the effect of such widening will have a complex effect on the performance of appliances and (where fitted) their ODS. Taking addition of hydrogen as an example, the addition of hydrogen reduces CO levels from gas appliances and potentially reduces the risk of low level CO poisoning from low level spillage of combustion products into a room or living space. However, in the event of a 100% flue blockage, the hydrogen does detract from ODS operation in some appliances.

There is no routine testing of ODS performance of appliances fitted with ODS on gases other than G20 (methane) for the UK market, primarily because it is a reference gas that is similar in properties to the UK Normally Distributed Gas quality (NDG). Such testing does happen for appliances distributed in European countries, but tests are performed with different reference gases which have similar properties to the NDG in their networks. The primary factor affecting the appliance manufacturers’ selection of an ODS was reliability and performance of the ODS; cost was not a factor.

In summary, the majority of ODS operated very well on the widest range of test gases. In the event of a 100% flue blockage, the ODS turned the appliance off before dangerous levels of CO were reached. However, one appliance had its performance somewhat degraded and one severely underperformed. This is not unsurprising; the chemistry of the combustion of these test gases is different and the flame speeds increase significantly with the addition of more hydrogen and/or higher hydrocarbon content gas such as ethane or propane. The ODS were designed and optimised for use on G20 (pure methane). 

Tests of ODS in isolation 

When tested in isolation, individual ODS designs reacted differently to each gas mixture under the same test condition. Some ODS shut-off more rapidly, and at a higher concentration of room oxygen (vitiation), compared to other types. In all tests, ODS 6 always operated first, operating at a higher level of room oxygen, demonstrating that it was most sensitive to low levels of room vitiation compared to the other types. In most instances, ODS 7 performed the worst and operated at lower room oxygen concentrations.

In most cases, ODS performance was better in isolation compared to when fitted to an appliance. In one case, an ODS showed improved performance when switched from the appliance it was supplied with, to another appliance.

Combustion performance at normal conditions cannot be used as a proxy to predict the outcome of the ODS test because other factors, such as flame impingement or combustion performance at blocked flue condition can influence this. The environment in which the ODS is fitted in the appliance can also affect performance.

Wobbe Index alone is not the main driver for ODS performance, as previously thought.

There is a strong linear correlation between Weaver flame speed factor and ODS performance. Increasing Weaver flame speed factor increased the ODS shut-off time, meaning there was reduced oxygen content at shut-off. This resulted in higher levels of room carbon monoxide at the point of ODS operation. Adding propane or ethane (which increase the WI) or adding hydrogen (which reduces the WI) all had an adverse effect on ODS performance, because the Weaver flame speed factor of the gas mixture was increased. It therefore seems that the flame stability of the gas mixture is being changed by the addition of other gases. The addition of nitrogen did not show a corresponding effect.

Tests of ODS in appliance

When the flue was partially blocked, the point at which the ODS operated was delayed compared to a fully blocked flue. It was possible for the appliance to operate and spill combustion products to produce room carbon monoxide levels above the Health and Safety Executive current recommended Long Term Exposure Limit and Short Term Exposure Limit. The ODS was ineffective between the point of spillage and before full flue blockage, i.e. it did not shut off despite the spillage of flue gas into the room.

All the open flue appliances tested with reference gas G20 met the ODS test criterion, and four met the criterion with all gases.

The flueless appliance (containing ODS 4) met the ODS test CO criterion with all gases, despite being one of the worst performers in isolation. However, it only met the CO2 limit (0.80-1.50%) with G23, G20 and 53.25 P, and exceeded it with the others. 

Other factors affecting ODS operation

Increasing the aeration port size of the ODS resulted in a faster shut-off because the volume of air entrained into the aeration port increased which in turn increased the speed of the gas-air mixture compared to the flame speed and made the flame more unstable at both normal and vitiated conditions.

There was some variability between individual measurement points for different magnetic units. Variable production tolerances for magnetic units of the same type can significantly affect ODS performance, so it is essential that ODS and magnetic unit combinations are carefully selected to account for these.

There was minimal or no impact on ODS performance caused by injector diameter, thermocouple lead length, ambient air temperature or temperature effects caused by gas quality alone.

Real-world operation of ODS

Throughout this project, contacts have been maintained with manufacturers of ODS and gas fire manufacturers and the Heating & Hotwater Industry Council (HHIC) Fires Group. They have been informed of progress during the testing programme. Extensive enquiries of the GasSafe registered trade, and appliance manufacturers found no hard evidence of ODS devices operating in the event of a flue gas blockage, although there was anecdotal evidence that if the device started to malfunction (and thus shut down the appliance) it encouraged householders to have their appliance serviced. It should be noted that this would also apply to standard flame failure devices. These can also become linted which will stop the appliance from operating and trigger a service call. There was a reluctance by some to criticise a safety device with a long history (even if unproven in recent years) whereas others did indicate they believed that ODS did very little.

Historical data sources do not demonstrate a clear and direct relationship between the incidence of unintentional carbon monoxide poisoning fatalities associated with natural gas fuelled fires, and the introduction of CE marking and ODS (or alternative devices) with their associated testing requirements from the early 1990s onwards. The annual number of CO fatalities associated with open-flued gas is around 20-25 % of the total CO fatalities from natural gas use. This proportion varies from year-to-year due to the very low number of deaths. CO deaths from natural gas open flued appliances (of a type likely to be fitted with ODS) are very rare with the last reported as occurring 2015. Classically, this situation would have called upon a quantitative risk assessment (QRA) but such analysis depends upon statistically significant events and the incidence of ODS operation is very rare, very uncertain and probably dependent upon the details of the incident. A grossly defective appliance can still injure or kill without being turned-off even by a correctly performing ODS, if the CO level in the flue is too high.

The increasingly widespread use of CO alarms, the introduction of a registered installers scheme and a decrease in the number of installed open-flued gas fires mean it is not possible to independently judge the effectiveness of ODS. This decrease in the population of open flued appliance is likely to continue, particularly when householders move to the use of decarbonised gas (100% hydrogen) or all electric heating (heat pumps or resistive heating). Industry stakeholders did not have readily accessible data on the effectiveness of ODS.

There was no strong consensus between the stakeholders as to whether the introduction of ODS has improved product safety. The majority believed they had, however there was also a view that ODS have an indirect impact on safety. Through more rigorous testing for CE marking, appliance designs had been improved to comply with more stringent combustion requirements. This meant the appliances were inherently safer. The tendency for ODS to become linted or covered in dust also triggered occupiers to get their appliance serviced, when a significant proportion of them would not have done so otherwise. However, this does not solely apply to ODS, other FFDs can also become linted and trigger service visits.

Impacts of Widening the Wobbe Range and introduction of Hydrogen

The output from the project was also used to support other NIA projects such as IGEM Gas Quality Working Group and the HyDeploy project work being undertaken on a QRA for the gas networks. 

Evidence shared from the ODS project was used to form part of the safety case to the HSE to support the move away from GS(M)R to a Gas Quality standard that could support the proposed changes to WI and introduction of hydrogen as a blend into the gas network.

The following reports were submitted to the HSE by IGEM on 18th of June 2021.

• IGEM report IGEM/TSP/258 “Project final report Investigation of the Performance of Oxygen Depletion Sensors (ODS) Kiwa 30961”

    This paper details the final report investigation by Kiwa into the performance of Oxygen Depletion Sensors affected by changes in Gas Quality specification.

• IGEM/TSP/21/239 “A quantification of the risk likely to arise from a widening of the permitted UK Wobbe Indices relating to the performance of ODS sensors”

    This research explains that a modest change in GB gas quality is unlikely to affect current CO incident rates, especially with reference to possible poor performance of some ODS sensors. The latter was investigated using 15 test gases of different blends of methane, nitrogen, propane, butane and up to 30% hydrogen.

Lessons Learnt

CO alarms are a relatively low cost and effective method of warning people about any CO emissions from effective gas appliances and, more advantageously, if the conditions are such that the fault is unlikely to trigger operation of ODS. It is thus suggested that if the UK wishes to proceed with all aspects of widening gas quality, all rooms containing open flued appliances could be fitted with CO alarms. Note this recommendation does not include room sealed (balanced flue) appliances or flueless appliances of low heat input e.g. gas hobs. This should be discussed with the appropriate regulatory authority who would be expected to carry out a cost benefit analysis. It has recently been announced that the Scottish Government is considering mandatory CO alarms (November 2020) and the English authorities may follow. 

If it is considered essential to retain the requirement for ODS to be fitted to gas fires, and the broadening of the gas quality limits goes ahead, then appliance and ODS manufacturers may need to carry out additional development work to ensure that all new designs operate with a wider range of gas compositions. This development work would be guided by the results of the test work reported here, which show the factors which are important in ODS and appliance design. 

This would lead to a requirement for standards to incorporate testing with gases other than G20 to prove ODS operation on a wider range of gas qualities. The criteria for test gas selection should: 

• have a higher WI compared with the new limits to allow “headroom” between it and the NDG. 
• include gases with higher Weaver flame speed factors because this property has been shown to have a significant effect on ODS performance.