D1A.1 Insights from the Hy4Heat and H21 projects, translated to the Dutch situation

This report is part of the national hydrogen research programme HyDelta Work Package 1A Safety and hydrogen. It focuses primarily on creating the framework conditions for safely applying hydrogen in urban areas. Extensive research has been carried out in recent years in the United Kingdom concerning the safe application of hydrogen in urban areas. This report summarises the most important results of the H21 and Hy4heat studies that have been published in the United Kingdom. The key question in this report is as follows:

What are the safety risks associated with the distribution of hydrogen in the distribution network and in indoor installations with regard to the nature and extent of leakages, dispersion and inflow into other areas, the risk and consequences of ignition and what measures are suitable in this regard.

This question has been answered by translating the results of the aforementioned British studies to the Dutch situation. In addition to the aforementioned reports, several other Dutch reports have also been included in order to reflect the latest available knowledge.

As part of Work Package 1A Safety, a report will also be published on the first adjustments to the risk models that were established in the United Kingdom for hydrogen as well as any possible mitigating measures for pilot projects in the Netherlands. For this, please see HyDelta D1A2 [1]. This report focuses on research conducted in the United Kingdom (UK) and its application for the Netherlands, including possible mitigating measures in the UK. The two reports therefore supplement each other.

Scope

H21 and Hy4Heat were very extensive programmes that spanned a period of three years at £25 million per project. Much of what has been published has led to follow-up research programmes in the UK.  The publications describe references to literature, historical data and experiments carried out up to and including the most recent experiments in 2020. This report is limited to urban areas – distribution pipes in the street – connecting pipes to low-rise housing and the gas installations in low-rise housing with a heat demand of < 70 kW.

Risk of leakage: nature and extent

Hydrogen will not affect the existing natural gas grid; using hydrogen instead of natural gas will not cause more leaks. The volume of hydrogen that is released in the event of a leak is, however, larger than with natural gas. For small leaks (max. 1 litre per hour) that amounts to about 30% more volume, and for larger leaks it is up to 190% more (also referred to as a factor of 1.3 to a factor of 2.9). In addition, permeation through the pipe wall can be up to 5 times higher. However, in absolute volumes this is very low.

• The low-pressure distribution network in the United Kingdom is constructed in part using the same materials as in the Netherlands, although in different proportions in terms of length. The exception to this is that in the Netherlands, (impact-resistant) PVC is mostly used in the 100/200 mbar distribution network. The Netherlands is unique in this respect. Leakage tests conducted in the UK on the same materials are therefore applicable to the Netherlands. The H21 study shows that parts of the distribution network that are leak tight for natural gas are also leak tight for hydrogen. When natural gas leaks were repaired, this also proved effective for hydrogen. The main source of distribution leaks was found in the UK’s cast iron distribution network. H21 therefore recommends replacing cast iron in the distribution network with the more modern polyethylene (PE) pipe material. In addition, H21 also recommends using a gas stopper in the branching of the connecting pipe as an extra precaution.
• Hy4heat has conducted research on leakages in indoor installations (the part downstream of the gas meter set-up). In the United Kingdom, gas meters are usually installed outside or in a kitchen cupboard. Installation in meter boxes as in the Netherlands (in combination with the electricity meter) is not done in the UK. This makes the research results less directly applicable to the Dutch situation in some cases. It also holds true for indoor installations that hydrogen does not lead to more leaks. Installation errors in the pipes in particular are the cause of the largest leaks. This applies in the Netherlands just as it does in the United Kingdom. No fitting types were found that leaked hydrogen but not natural gas. No fittings were found to be unsuitable for hydrogen in Hy4Heat. This is not directly applicable to the Netherlands because other fittings are used here. Research has been conducted in the Netherlands that supports the leakage rate found in Hy4heat: An existing leak with natural gas results in an approx. 30% higher flow rate with hydrogen for fittings.

Dispersion and inflow into other areas

As far as the dispersion of hydrogen in the open air is concerned, i.e. as a result of leakages in the distribution section, it holds true that, due to the lower density, the gas will rise more quickly than natural gas and will therefore not lead to higher risks in the open air. In general, it can be said that hydrogen does not disperse further in the soil than natural gas does. Inflow from leaks into an enclosed space can lead to unsafe situations in particular. However, this also applies to natural gas.

Dispersion in spaces was examined in Hy4heat, Hyhouse and the Gas Dispersion Analysis report (sub-report of Hy4heat). Here, the primary argumentation was based on a leak in the pipework in the indoor installation. In the case of a hydrogen leak, homogeneous gas concentrations first form at the top of the room (gas stratification). The volume of hydrogen from leakage in a pipe is 1.2 to 1.8 times that of natural gas. The influence of ventilation openings is considerable, see also Hy4heat reference [29] page 39 ff, and significantly greater than, for example, mechanical ventilation. At the highest tested leakage rate of 78.6 m³/hour, the addition of natural ventilation ensured that the concentration at the top of the room was reduced from ~60% full gas to ~40% full gas. This kept the mix ratio within the explosion limits and closer to the stoichiometric ratio. Please note that this tested leakage rate does not occur in reality in a domestic environment in the Netherlands. The maximum leakage rate of natural gas when a gas cooker pipe is penetrated is approximately 10 m³ per hour; for hydrogen this would be 20-30 m³ per hour.If hydrogen spreads to other rooms, the hydrogen concentration will quickly decrease due to natural ventilation. Formation of an explosive mixture with the most common small leaks (<10 m³/h) and normal ventilation does not appear to be realistic. In the event of a hydrogen leak in a room without ventilation, in accordance with building regulations and closed/door windows, a very high hydrogen concentration may be formed at the site of the leak, depending on the pressure, leakage and volume of the room where the leak is located. The most effective measure for preventing an explosive mixture from forming is to combine box ventilation (e.g. air vents in meter boxes of at least 0.01 m²) with room ventilation from the adjacent room. As a comparison, the NEN2768 now prescribes an upper and lower grid for a meter box of net 0.02 m² each, which is already more than the aforementioned requirement of 0.01 m².The flammability limit in the source space (where the leakage occurs) is not achieved and by ventilating the space this is likely to be further reduced. In addition, with natural gas the vast majority of leaks are noticed due to the odour present and are subsequently repaired before a dangerous quantity of gas is able to escape and ignite. UK research therefore suggests that odourisation is an effective tool for hydrogen as well, in addition to flow protection such as an EFV or gas stopper adjusted to the maximum consumption of an appliance.

Risk of ignition

The risk of ignition for hydrogen is different from that of natural gas. At the same gas pressure, hydrogen can ignite at distances that are up to 25% further [2] from the gas outlet. In practice, this means that the presence of ignition sources at a greater distance may cause ignitions. Various studies have been conducted on the influence of ignition sources on ignition in urban areas. Mechanical extractors and light fixtures do not cause ignition in hydrogen under normal operating conditions. This is important, because these potential ignition sources are often located at the top of the physical space, where the hydrogen concentration first accumulates. When testing white goods (various household appliances present in the kitchen, but also outside this area: freezers, hair dryers, hoovers, etc.) as ignition sources, no difference was found between the risk of ignition for natural gas and for hydrogen. In addition, the risk of igniting hydrogen concentrations is further reduced by the fact that the ignition source is often located at a low level in the room, while the hydrogen gas rises quickly and is first concentrated at the top of a space. The risk of ignition is then further reduced by ventilation in the box/cupboard or room. The studies in the UK have focused on kitchen areas as gas meters may be present in a kitchen cupboard, several gas appliances may be installed in the kitchen, and several ignition sources may be present. For the Netherlands, the most relevant room is the meter box because of the presence of electrical distribution boards, followed by the kitchen where a central heating appliance may be installed.

Consequences of ignition

Combustion of gas, and therefore of hydrogen, releases flue gases and heat that may cause a build-up of pressure in an enclosed space. If this combustion gas is unable to escape, the pressure will continue to build up. In the UK, studies have been conducted into the consequences of ignition of hydrogen and natural gas.

The results are shown in “ISO damage charts”. The data have been further developed into several different concentration bands, expressed in the percentage of gas in air (GIA).

A summary of the results of the data obtained on ignition of the gas is broken down into concentration bands, with the comparison between natural gas and hydrogen and the consequences in general terms:

1. 0 – 10 vol% - Hydrogen may be less severe than natural gas
2. 10 – 15 vol% - Comparable damage between hydrogen and natural gas
3. 15 – 20 vol% - Hydrogen may be more severe than natural gas
4. 20 – 25 vol% - Hydrogen is likely much more severe than natural gas
5. >25 vol% - Hydrogen is likely much more severe than natural gas

Consequences of the aforementioned bands are subject to experimental conditions / environments

In the open air and at low concentrations, a fire will first occur at a gas concentration of >LFL value. Then a hydrogen fire is possible without overpressure. In closed areas (indoor installation situations) or at higher concentrations, an explosive ignition can occur with potentially more far-reaching consequences, as indicated in the 5 points mentioned.

• A hydrogen fire with the same energy outflow as natural gas has equal or lower heat radiation. The heat radiation of hydrogen becomes equal to that of natural gas if dust/earth are also present in the flame.
• A hydrogen explosion may, under the same conditions, cause greater consequential damage than a natural gas explosion due to the higher burning rate. At low gas concentrations (<10 vol%) the consequential damage with hydrogen is lower; starting from 15 vol% it is more severe. In the case of an explosion with a stoichiometric (30 vol%) hydrogen mixture, pressure build-up can cause an overpressure of over 100 mbar and, in non-ventilated spaces, up to 7 bar. This could lead to walls collapsing or houses being destroyed

Assessment of the overall risks and appropriate control measures

From the UK studies, the following overall risk assessments from fires and explosions are apparent: the overall risks from hydrogen may be higher than from natural gas, with higher risks of explosion, though this is partly offset by lower risks of fires. However, in the UK, the risks from the incomplete combustion of natural gas, which produces carbon monoxide, have not been taken into account. Unfortunately, in the UK the use of natural gas causes a significant proportion of casualties due to carbon monoxide (approximately 20 incident reports with casualties per year). This is also the case in the Netherlands (39 natural gas incidents with carbon monoxide poisoning out of a total of 69 downstream of the gas meter between 2010-2020). When using 100% hydrogen, these casualties will no longer occur because 100% hydrogen does not release carbon monoxide.

In the UK, the total risk is calculated in the quantitative risk models and compared with the field data for natural gas. The risk from natural gas distribution is shown as the potential number of casualties per year: Potential loss of life (PLL). As an illustration: in air traffic, the number of casualties per X million flights is used. On average, it is claimed that per 30 million flights worldwide each year, there are approximately 600-1000 unfortunate casualties. In the Netherlands, we do not work with this kind of an approach, because there are hardly any casualties, which is certainly the case when carbon monoxide is not taken into account. In the UK, the results of the calculated PLL values for natural gas are also higher than actually measured, which is why the results of these models are considered conservative.

The main differences between the UK and the Netherlands are as follows:

• In the UK there are proportionally more steel and cast iron pipes in the low-pressure distribution network than in the Netherlands. In the Netherlands, approximately 80% of the distribution network is already made of plastic. Both countries have replacement programmes for the ageing network sections made of materials such as cast iron. Using plastic materials instead of cast iron reduces the PLL. The ongoing replacement programmes in both countries are therefore contributing to a lower PLL.
• In the UK, 50% of gas meters are placed in kitchen cupboards and 50% are placed in the exterior facade. In the Netherlands, indoor gas meters are currently placed in the meter box (during the large scale introduction of natural gas in the 60’s/70’s there was more variation here), which also houses the electricity distribution board, meaning that the risk could be different.
• In the UK, there are proportionally more older houses as compared to the Netherlands. As a result there are more cracks, less mechanical ventilation as well as ventilation that does not comply with the existing regulations in the UK. Ventilation has a major impact on the PLL.

The PLL for hydrogen in 2032 in the UK is 1.88 times higher than the PLL for natural gas in 2020, with 83% of the risk attributed to the metal networks that remain in the grid even with the current replacement plans in the UK. If all remaining iron pipes in the UK low-pressure and medium-pressure networks are replaced, the PLL for hydrogen could fall to 0.18.

Due to the above mentioned differences between the UK and the Netherlands, the total risk as described for the UK cannot be translated into the same figure for the Netherlands. This will be further explored in HyDelta 2. [1].

There are control measures available to bring the risks posed by hydrogen in urban areas to the same level as for natural gas. These have been considered for the Dutch situation from the UK’s proposed measures:

• Nature and extent: lowering the gas pressure of the network is possible and would reduce the PLL by 0.02 per year. The use of a different medium does not cause more leaks to occur.
• Nature and extent: in the case of hydrogen, the risk of leakage appears to be greater. Therefore, more frequent searches for leaks could be undertaken to reduce the number. The QRA models indicate which parts of the network (material as well as pressure) contribute most to leakages and could therefore be checked more often.
• Nature and extent: the installation of excess flow valves (EFV) would barely reduce the PLL in the main distribution network (i.e. the section before the connecting pipe and meter set-up), because most of the hydrogen leaks to be expected in a main network with older materials will have already been managed after the replacement programmes. Using a gas stopper in the branching between the main pipe and the connecting pipe will ensure a significant reduction in the risk associated with damage caused by excavations. There would also a risk reduction in the event of failure of the gas installation in the house, including the gas meter set-up, as a result of fires in the house involving a leak in the 100 mbar connecting pipe. Depending on the set values of the gas stopper, the UK has recommended using a gas stopper when the building is entered. Discussions are ongoing with smart gas meter manufacturers for integrating gas stoppers in gas meters.
• Nature and extent: replacement of cast iron and ageing iron assets (pipes and components) with plastic pipe material reduces the likelihood of leakage in the UK. These replacement programmes have been active in the Netherlands for years (replacement of cast iron and steel connecting pipes, for example).
• Dispersion in spaces: most leaks in connecting pipes, meter connections and indoor pipes occur as a result of work activities. Odorization allows for early detection of leaks (this would be very effective, as in the case of natural gas) and could reduce the extent of the leakage and therefore prevent the formation of an explosive mixture.
• Dispersion in spaces: installing gas meters outdoors reduces the PLL by 0.01 per year in the UK (this is already the case for half of the situations). However, as this would constitute a major change in the Netherlands, it is uncertain whether this control measure is realistic for the Dutch situation.
• Dispersion in spaces: pipes that are leak tight when used with natural gas appear to be so for hydrogen as well. When a new natural gas installation is installed, a leak tightness test is carried out. It is only logical to do this for hydrogen as well, together with a visual inspection for existing pipe systems.
• Dispersion in spaces: ventilation is a very effective measure. Applying box ventilation (grating in the meter box) and room ventilation (air vents, etc.) in accordance with the applicable building regulations ensures sufficient ventilation, so that the formation of a mixture that could catch fire or explode is prevented or significantly delayed. Just as with natural gas: ensure that there is sufficient ventilation in basements, crawl spaces or other enclosed spaces where gas can accumulate. Or apply other control measures here.
• Ignition prevention: there is little difference in the risk of ignition sources (fittings, fans, white goods) between hydrogen and natural gas. The most effective control measure is again ventilation in the box/cupboard or room. In addition, hydrogen can be ignited up to 25% further away from the ignition source than natural gas. Where possible and feasible, potential ignition sources in the vicinity of a gas installation should be avoided or combined with adequate room ventilation.
• Consequences of ignition: the risk of ignition (consequential damage) is clearly reduced by the above measures. If an ignition does occur, the pressure has to be released. In practice this happens when the explosion forces a door or window open (sometimes even walls and/or a ceiling). The intensity of the explosion can be reduced by lowering the pressure wave of the explosion. Again, sufficient ventilation is the most effective solution.

In addition to the control measures mentioned (see also HyDelta WP1A D1A.2 Part 3), general control measures (which also apply to the use of natural gas) are also applicable. This includes adequately competent staff, procedures and measuring equipment, training and independent monitoring during large-scale pilot projects in order to establish additional control measures