Risk-based Regulatory Design for the Safe Use of Hydrogen

23. Quantitative risk assessment: Hydrogen versus conventional fuel

Abstract

This report consists of five quantitative risk assessments of hydrogen versus conventional fuel in five accident scenarios. The scenarios presented are: accident during production (electrolyser), leakage from a high-pressure pipeline, hydrogen leakage from a truck driving in a built-up area, traffic accident involving a hydrogen city bus driving in a tunnel and accident at a hydrogen fuel station.

Scenario 1 – Production-electrolyser

A semi-Quantitative Risk Assessment (sQRA) has been carried out for a theoretical electrolyser and hydrogen storage facility with the aim to establish an approximate level of risk and demonstrate the differing risks associated with a comparative system, namely a Liquefied Petroleum Gas (LPG) bulk store facility.

For the sQRA, consequence modelling was carried out on a selection of pre-defined scenarios with individual risk calculated for static receptors at intervals from the equipment. These scenarios are selected to demonstrate the worst-cases from perspectives of both consequence (typically lower frequency events) and frequency (typically lower consequence events).

The results showed that only results of the immediate ignition scenarios, resulting in a jet fire or localised explosion/fireball, were suitable for comparative analysis using the available software for the calculations. This is largely due to the uncertainty of ignition location for delayed ignition (i.e. flash fires and explosions) thus the large number of variables that would need to be considered to produce simple outputs.

The comparative analysis for the immediate ignition scenarios showed that the risk for LPG storage (including delivery and distribution) is clearly greater than the in-situ generation of hydrogen (including compression, bulk storage and distribution). It is proposed that this is not only due to the physical properties of the two materials, but the more rigorous incorporation of safety features assumed to be present in the hydrogen design.

However, a further high-level sensitivity analysis carried out to compare the above scenario to the import and use of natural gas from a pipeline (see Scenario 2) suggests that the assumed hydrogen facility still poses a greater risk to populations in the vicinity of releases from the equipment.

On this basis, the management of hydrogen risk to populations could more easily be controlled through simpler measures such as the separation of plant from buildings and equipment.

Scenario 2 – Pipeline transport (leakage from a high-pressure pipeline)

A comparative semi-Quantitative Risk Assessment (sQRA) has been carried out for high-pressure pipelines of hydrogen and methane with the aim to establish an approximate level of risk and demonstrate the differing risks associated with hydrogen and conventional fuels, methane in particular.

For the sQRA, consequence modelling was carried out on a selection of pre-defined scenarios with individual risk calculated for static receptors at intervals from the equipment. These scenarios are selected to demonstrate the worst-cases from perspectives of both consequence (typically lower frequency events) and frequency (typically lower consequence events).

The results showed that only results of the immediate ignition scenarios, resulting in a jet fire, were suitable for comparative analysis using the available software for the calculations. This is largely due to the uncertainty of ignition location for delayed ignition (i.e. flash fires and explosions) thus the large number of variables that would need to be considered to produce simple outputs. To a lesser extent, the simplistic nature of the modelling for deflagrations and inability for the software to model detonations, limits any meaningful interpretation of explosion results.

The comparative analysis for the immediate ignition scenario showed that the increase in risk for hydrogen is negligible compared to natural gas when the ignition probabilities proposed by Tchouvelev (Tchouvelev, Hay and Benard, 2008[1]) were used. However, when ignition probabilities based on RIVM MRCB methodologies (Rijksinstituut voor Volksgezondheid en Milieu (The Netherlands), 2020[2]) were applied there was a clear order of magnitude risk increase for hydrogen versus methane throughout most of the individual risk calculation intervals. In both instances, however, hydrogen risk tails off at further distance, where the model predicts a longer methane flame.

Scenario 3 – A hydrogen transport truck driving through a built-up area experiences a leak

This report studies the effects of leaks of various sizes experienced by a hydrogen delivery truck, from a small leak to a major outflow. In contrast with Scenario 4, the environment of Scenario 3 is not tightly confined.

This QRA utilises PHAST to model the phenomenon, in line with the expertise and time constraints of the contractors employed for the study. It quantifies risk in terms of location-specific individual risk (LSIR), which denotes the annual probability of injury or death in specific places, and allows expressing results in terms of iso-risk domains. The study models the effects of thermal radiation and the overpressure caused by explosions to quantify this LSIR.

The conditions of the blast are split between high or low ignition energy (e.g. an initial explosion vs a spark, respectively), then then further by obstruction level (high, low or none) and type (whether there are obstructions on at least 2 sides, or not). The model of explosion used is the analysis is the Boiling Liquid Expanding Vapour Explosion (BLEVE).

Individual risk is then estimated as the product of the event frequency, the occupancy (probability to be present at a certain location) and the vulnerability. Vulnerability is quantified for indoor and outdoor cases for various values of overpressure, as well as for a range of thermal radiation intensities. The large number of possible factors in such an accident (geometries of the affected area, traffic, etc.) forced to confine the study to a handful of well-determined cases.

The results show that in gas phase, hydrogen and methane have similar effects, with slightly larger danger distances for methane. In liquid phase, the results are also similar, with hydrogen exhibiting a slightly larger danger area. Catastrophic failure of the tank could yield a fireball of 56 metres in diameter for gaseous hydrogen and 81 metres for methane, while failure of the loading valve could cause an 8-metre hydrogen flame. With liquids, the values reach 116 metres in diameter for the hydrogen fireball. Finally, BLEVE-type explosions, where an external fire triggers evaporation of fuel within the tank and its eventual explosion, would yield a similar 116-metre fireball, compared to 129 metres for liquified natural gas. To illustrate the frequency of such accidents, in the last 30 years, three have occurred in France.

Scenario 4 – A hydrogen city bus driving in a tunnel is involved in a traffic accident

The modelling results of Scenario 4 – a hydrogen city bus driving in a tunnel is involved in a traffic accident (variations e.g.: small leakage/large tank rupture) are reported in this report.

A comparative QRA study is performed between a hydrogen and a methane city bus located in the center of a 1.2 km one-way tunnel. The comparative analysis has conducted both on the jet fire scenarios from TPRD (Thermal pressure relief device) and on the catastrophic tank rupture.

The QRA methodology combines event tree analysis for probabilistic analysis with engineering correlations (i.e., for hydrogen jet flames (Molkov and Saffers, 2013[3]) and for the blast wave decay in a tunnel from for consequence analysis (Molkov and Dery, 2020[4]).)

In the vicinity of the bus (20 m) higher IR values are calculated for jet fire scenarios than for catastrophic tank rupture, while the IR values of the tank rupture scenario are predominant for the whole tunnel.

For jet fire, the hazard distances from the bus are slightly higher for H2 than for methane. Furthermore, the frequency of jet fire events is higher for H2 than for methane.

For catastrophic tank rupture a similar profile is observed for both methane and H2, it can be seen that the overpressure decreases rapidly along the tunnel, especially within the first 50 m. For this scenario the individual risk is higher for hydrogen than for methane and correspondingly larger hazard distances are evaluated for hydrogen.

The results of the consequence modelling found that both the results of the immediate ignition scenario, resulting in a jet fire, and those of the catastrophic tank rupture scenario are suitable for comparative analysis without the use of more complex software (e.g. for modelling H2 release from TPRD and deflagration).

The comparative analysis for the immediate ignition scenario has showed that the frequency (events per year) is slightly higher for hydrogen than for methane, and slightly higher hazard distances are calculated for hydrogen than for methane. The results of the catastrophic tank rupture scenario showed that the individual risk is higher for hydrogen than for methane and correspondingly higher hazard distances are evaluated.

Scenario 5 – Accident at a hydrogen fuel station

A comparative QRA study between a hydrogen and a methane refuelling station is performed. The comparative analysis refers to two configurations of the refuelling station: one with discontinuous supply of H₂ by means of tube trailer mobile storage, and the other with continuous supply via pipeline. In addition, the analysis is performed for a hydrogen filling station with on‑site H₂ production via electrolyser, comparable to that of the 2019 Norway incident.

The software used is HyRAM+ developed at Sandia National Laboratories. The HyRAM+ software toolkit provides a basis for conducting quantitative risk assessment and consequence modeling for hydrogen, methane, and propane systems.

The QRA results are reported in terms of Average Individual Risk (AIR), which expresses the average number of fatalities per exposed individual. It is based on the number of hours the average occupant spends at the facility (i.e., 2 000 exposed hours per occupant per year).

The analysis is performed considering the various sections of the hydrogen refuelling station at different pressures, specifically a high-pressure storage module with compressor at 90 MPa is assumed. For the other components of the hydrogen refuelling station (i.e., electrolyser, pipeline, tube trailer, and dispenser) the specific pressure is considered (i.e., 3 MPa, 10 MPa, 20 MPa, and 70 MPa respectively). For the CNG refuelling station a maximum pressure of 25 MPa is assumed for a conservative estimate.

The results of the comparative analysis show higher AIR values for CNG compared to the H₂ refuelling station in both configurations with pipeline gas supply and via tube trailer. For both gases, the refuelling station configuration with continuous gas supply via pipeline led to lower AIR values than the configuration with discontinuous supply via tube trailer.

In particular, for a hydrogen refuelling station the main components that contribute to the AIR are respectively the high-pressure storage module with compressor and the dispenser. Therefore, the lower AIR is for the configuration with onsite H₂ production via electrolyser where most of the H₂ is stored at 20 MPa. Decreasing the capacity of the hydrogen refuelling station to 500 kg/day results in a slight decrease in AIR.

It should be noted that the methane ignition probability used in HyRAM+ is verified to provide conservative estimate of AIR values.

The consequence analysis of the current HyRAM+ includes the hazard from hydrogen jet fires (for immediate ignition) and unconfined explosion (for delayed ignition). For a jet fire, hazard distances are higher for H₂ than for methane, due to the higher pressures encountered in the high-pressure storage module, compressor, and dispenser. Similarly, for an unconfined explosion the hazard distances are higher for H₂ than for methane, with possible escalation to detonation in case of high congestion.

References

[4] Molkov, V. and W. Dery (2020), “The Blast Wave Decay Correlation for Hydrogen Tank Rupture in a Tunnel Fire”, International Journal of Hydrogen Energy, Vol. 45/55.

[3] Molkov, V. and J. Saffers (2013), “Hydrogen jet flames”, International Journal of Hydrogen, Vol. 38/19, pp. 8141-8158.

[2] Rijksinstituut voor Volksgezondheid en Milieu (The Netherlands) (2020), Manual Risk Calculation Bevi. Version 4.2, Bilthoven: RIVM.

[1] Tchouvelev, A., R. Hay and P. Benard (2008), “Comparative Risk Estimation of Compressed Hydrogen and CNG Refuelling Options”, http://www.tchouvelev.org/Documents/UpdatedDocs/NHA_2007_Conference_RiskEst.pdf (accessed on 5 May 2023).