D5a.1 Quantitative risk assessment of hydrogen in the built environment - effect of mitigating measures

To assess the risks of using hydrogen in distribution and transport networks compared to natural gas, it is essential to understand the differences in probability and consequence. Within the HyDelta program, the work package 'Hydrogen and Safety' has been established, with the primary objective of:

Mapping out risks related to the behaviour of hydrogen in case of leaks in homes and in the distribution network and defining control measures based on these risks.

For this purpose, a quantitative risk assessment (QRA) was conducted. This analysis compares the risks between the current natural gas distribution system and the potential future hydrogen distribution system and also compares them with other risks in our society. The total risk assessment includes risks arising from leaks in the distribution network and from leaks in homes. Given the approximation of reality, the model provides the individual risk resulting from a fire or explosion. The results of this analysis provide a quantitative basis to assess whether hydrogen distribution poses more risk to society and which measures are most effective in mitigating these risks.

Risks are small, but what does small mean?

The individual risks of hydrogen and natural gas in distribution networks are minimal and remain well below the acceptable 1x10⁻⁶ contour per year. However, public risk perception can vary depending on emotional, normative, and information-related factors. It is important that risk assessments are not only presented with figures but also through understandable methods or comparisons with other risks, to accurately inform and involve the public in decision-making without exaggerated or misleading information.

The calculated risks for using hydrogen in the case-study neighbourhood in the reference scenario without additional risk-reducing measures is ~0.2x10^-6/year. With the most effective risk reduction measure in this study, this risk is reduced to ~0.05x10^-6/year. The magnitude of these risks is on the same order as the number of lightning strike fatalities in recent years. Carbon monoxide poisoning fatalities are twice as high compared to the reference situation (0.4x10^-6/year). Annually, about 10 times as many people die in building fires (2x10^-6/year) or 25 times as many from accidental drowning (5x10^-6/year). Road traffic fatalities are about 250 times higher (50x10^-6/year). Compared to other causes of death, the risk of hydrogen (and natural gas) in the built environment is therefore very small.

Even though the risks are minimal, public perception can quickly change due to initial incidents. Therefore, it is important to proceed cautiously when introducing new technology, such as hydrogen in the built environment through pilot projects. Since new technology does not yet have established statistics on its failure, accidents can occur. However, by implementing a large number of extra safety measures at the introduction, the public might also perceive the new technology as inherently unsafe. It is therefore important to understand the magnitude of the risks and the contribution of different measures in reducing this risk.

A calculation of the risks for a case-study neighbourhood provides insight

To gain a better understanding of the relative effects of leaks behind the meter and from the distribution network, this study analysed a representative case-study neighbourhood. This neighbourhood consists of 57 homes connected to a 100 mbar main pipeline via service lines. The 100 mbar network is fed by a steel 8 bar pipeline running through the neighbourhood. The 100 mbar network is modelled in several segments with different materials and diameters. The homes are modelled based on their area and include detached houses as well as semi-detached houses. Additionally, the risk from leaks behind the meter for each home was determined. Based on failure frequencies in homes, it is confirmed that the individual risk for natural gas from behind-the-meter leaks closely matches the (limited) real-world data.

The analysis shows that the individual risk for hydrogen is greater than for natural gas due to more violent explosions, but smaller when the risk of carbon monoxide poisoning is included. Carbon monoxide poisoning is a common consequence of incomplete combustion of natural gas but does not occur with hydrogen due to the absence of carbon. When this risk is included in the comparison, a shift occurs from the reduced risk of CO poisoning to an increased risk of explosions. The total individual risk, given the chosen set of assumptions and without additional measures, is lower for hydrogen than for natural gas.

Different mitigating measures with different effects

The study calculated the effect of several mitigating measures. These measures can affect the likelihood of a leak, for instance, through regular inspection of pipes and equipment, or by limiting the consequences of a leakage. For regular inspection, a distinction is made between an annual check of the installation, with an assumed 20% reduction in the likelihood of spontaneous leakage, and an automatic daily inspection with an 80% reduction in the likelihood of spontaneous leakage. To limit the consequences of a leak, measures such as excess flow valves, sensors whether or not connected to an automatic shut-off valve, or increasing ventilation can be considered.

When modelling the measures, it is assumed that all measures are 100% effective. This means, for example, that excess flow valves always closes when the maximum flow is exceeded, and that every concentration above the threshold value is detected by the sensor at the location where it is mounted. However, when implementing the mentioned measures in practice, one must account for uncertainty in this effectiveness. Some measures lie within the domain of network operators, such as the excess flow valve, while many of the other measures will be installed in the home and are therefore the responsibility of the homeowner. This means that in addition to the estimated risk reduction, the measures each also have their own advantages and disadvantages.