Hydrogen Safety. Progress Made Knowledge Gaps Role of CFD. Dr. Prankul Middha

Transcription

1 Hydrogen Safety Progress Made Knowledge Gaps Role of CFD Dr. Prankul Middha 28 th FLACS User Group (FLUG) Meeting Technip, Paris, France 8-9 November, 2016

2 Hydrogen - how to ensure and prove acceptable safety? Consequences of an internal explosion in hydrogen cars? Risk in connection to new H 2 infrastructure, transport, production, storage, etc.? How to dimension a container or pipes with flammable H 2 -air? Consequence of H 2 leaks inside a congested process facility, e.g. chemical industry, nuclear facility or semiconductor plant? Consequences of a significant leak in a vehicle? What if inside tunnel or workshop?

3 Background Hydrogen quite different from conventional fuels e.g. Natural Gas o Buoyancy will quickly remove hydrogen when unconfined o Wide flammability limits (however not so different on the lean side) o Hydrogen much more reactive (ca. 10x Natural Gas), high sensitivity to DDT o Lower energies needed to ignite H 2 (ca. 1/10 gasoline vapours) o Negative J-T coefficient, leaking gas warms, spontaneous ignition frequent More background information: o o E-book by Molkov: PhD thesis Middha: H 2 disperses upwards after release, limits cloud size with low confinement Spontaneous ignition much more frequent for H 2 than NG (Groethe, 2005).

4 Background Hydrogen safety issues are a crucial aspect for wide spread deployment and use of hydrogen and fuel cell technologies Hydrogen technologies must have the same or lower level of hazard and associated risk compared to conventional fossil fuel technologies Reliable analysis and R&D leading to the development standards and norms is important appropriate Several EU/global joint R&D projects HySafe, IEA Task 19/31/37, HyIndoor, NaturalHy, HySEA, SUSANA, 4

5 CFD (Computational Fluid Dynamics) Validated CFD modelling based on experimental work allows for a detailed representation of physical phenomena that govern the realisation for a hydrogen hazard: Facilitate precise representation geometry and effect of active / passive barriers Includes required physics for hydrogen (equation of state, density, combustion models, ) We cannot rely on work done for natural gas alone due to key differences in properties Balloon explosion (hydrogen) Balloon explosion (propane)

6 CFD Context Computational Fluid Dynamics (CFD) is increasingly used to perform safety analysis of potential accident scenarios (production, storage, distribution of hydrogen and its use in fuel cells) CFD is a powerful numerical tool that can provide useful data and insights but it also requires a high level of competence and knowledge in order to be used in a meaningful way To apply CFD with a high level of confidence on the accuracy of the simulation results, two main issues have to be addressed: o the capability of the CFD models to accurately describe the relevant physical phenomena o the capability of the CFD users to follow the correct modelling strategy. The reliability/accuracy of the CFD results remains a significant concern.

7 CFD tools available hydrogen deflagrations 1. FLACS Gexcon. 2. ANSYS Fluent (+UDFs) Ulster university 3. COM3D FZK/KIT 4. REACFLOW JRC 5. Recent development: ANSYS Fluent (+UDFs) NRG/Delft Univ

8 x, m FLACS Validation hydrogen deflagration Overpressure (kpa) Overpressure (kpa) Vented tube (HYCOM) Congested pipe grid (Shell/HSL) Blind simulations in a highly congested, repeated pipe grid 18 m 3 cloud, 26 mm pipes (125 mm spacing) Traffic tunnel (SRI) Experiments 12 1/5 scale (2.5 m diameter, 80 m long) 37 m 3 stoichiometric H 2 -air in middle Refuelling Station (Shell/HSL) 10 13% H2/air x ign = 0 m Xign BR=0.3, D=174 mm Simulations Experiments Simulations polyethylene sheet 6 4 ignition venting ratio, % vehicle model Experiments Distance (m) Simulations t, s Time (s)

9 Benchmarking hydrogen dispersion Low momentum release (INERIS) Room 7.4 x 3.8 x 2.7 m, sealed except 2 small openings 4 minute leak 1 g/s, 1.5 hour waiting time after leak Blind simulations before the experiments Details in Venetsanos et al. (2009)

10 Refueling Station (Shell / HSL) Representation of a HRS, 2 dispensers & 1 vehicle Homogeneous, stoichiometric H 2 -air cloud (90 m 3 ) Details in Makarov et al. (2009)

11 Focus: Turbulent burning velocity models FLACS (commercial) Bray correlation not optimised for hydrogen however, other corrections specific to hydrogen and validation work show reasonable agreement S 1.81 u ' S l T L I New work with Markstein number dependency (ongoing development work, Hisken) provides much better results for e.g. vented (congested) deflagration tests (FM Global); A promising model by Zimont et al much better physical dependencies (pressure/temperature effects). New validation carried out by NRG for THAI facility, etc (Nuclear safety) however wider validation effort (for hydrogen) needed

12 Improvements in Turbulence Model (FLACS) Very high over-pressures predicted in hydrogen explosion simulations (FLUG 2010/11) o IRSN two cylindrical tank case o IRSN three hydrogen truck case Hydrogen Refueling Station (Shell/HSL) o (a) on-grid case o (b) porous case Modified sub-grid turbulence model to switch off unrealistic production of turbulence in partially porous regions (FLUG 2012) Porous case On-grid case

13 Results Porous case On-grid case Hydrogen refueling station (Shell/HSL) IRSN Two cylindrical tank case

14 Significant experimental work in recent years High pressure releases (different pressures and orifice sizes) Development of correlations similarity law

15 Example of ongoing research HySEA HySEA: Improving Hydrogen Safety for Energy Applications through pre-normative research on vented deflagrations ( ) Consortium: Gexcon (Coordinator), University of Warwick, University of Pisa, Fike Europe, Impetus Afea and University of Science and Technology of China (self-funded) Total budget: About 1.5 MEUR Started 1 September 2015 Website: Details presented earlier by T Skjold

16 Concentration (vol. %) Gaps in CFD modelling Identification of gaps in CFD modelling of hydrogen release and combustion e.g. absence of a single model to represent all modes of flame propagation One of the main gaps: the lack of a Model Evaluation Protocol (MEP) for safety of hydrogen technologies Key concern lack of convergence between different CFD codes No systematic study of mesh/time step sensitivity Experiment (0.75 m) Experiment (1.5 m) Experiment (2.25 m) Simulations (0.75 m) Simulations (1.5 m) Simulations (2.25 m) Distance (m) FLACS over-predicts concentration for high pressure releases from small oriffices

17 The SUSANA project SUpport to SAfety ANalysis of Hydrogen and Fuel Cell Technologies ( The SUSANA project (co-funded by the Fuel Cell and Hydrogen Joint Undertaking): producing a Model Evaluation Protocol for hydrogen technologies safety (HyMEP) 18

18 The SUSANA project HyMEP: o The first MEP for hydrogen technologies safety o To evaluate the accuracy of the CFD models o To assess user capability of correctly using CFD codes Current version of database ( including ~30 experiments Current Status: o Limited benchmarking and progress made as a part of the SUSANA project to resolve differences seen (special edition of FABIG meeting) o Not a very large participation o Significant work remains to be done 19

19 How about DDT (Deflagration/Detonation transition)? FLACS DDT criteria initially developed for hydrogen Middha & Hansen (2006/07) criteria based on DPDX (spatial pressure gradient) presented Evaluated against several experiments with good precision DPDX-criterion: => May be DDT >5 => Likely DDT Hotspot region should be large enough to propagate detonation (wrt cell size) However: Description of DDT by FLACS an engineering approach Other techniques available but difficult to take complex geometries into account Haven t seen much progress in the past few years DDT not a part of SUSANA project!

20 What about modelling DDT and detonations? Recent efforts to model detonation (following confirmed DDT) as two-step simulations by tweaking FLACS parameters (Hansen, UKELG 2013) Approach worked well (accurate predictions) against several experiments Simulation shown from Buncefield DDT test 2 (Spadeadam)

21 Hydrogen Refueling Station (HRS) process map

22 Chronological development of regulations and norms

23 Applicability to hydrogen refueling stations

24 Current status Safety Distances Several experimental efforts carried out e.g. HSL s work on LH2 releases (2014): Realistic ignited spill of LH2 14 tests (4 not ignited) Fire (and weak secondary explosion in some cases) No harmonized approach for safety distances yet further work is needed. CFD can have an important role here to fill gaps

25 Current status Regulation and Approvals Italy has a national regulation; other countries rely on existing international standards Still a need for one overlying standard/regulation document Nonetheless, HRS development is progressing well: o o o 2015: 184 hydrogen refuelling stations in operation all over the world (82 in Europe) 129 planned hydrogen stations (2015 figures) (53 in Europe) No major incidents so far! However, hydrogen lags far behind electric as the green fuel of choice

26 Nuclear Safety CFD plays an important role Ongoing development and revamp requires extensive analysis of hydrogen hazards and how they can be mitigated Only CFD has the capability to carry out effective risk and safety studies in complex geometries Benefits/added value for design presented in presentation by Gexcon during presentation at nuclear safety seminar last week Important to understand hazards to prevent incidents such as Fukushima

27 Final remarks Tools for paving the way for a safe transition to a hydrogen economy are needed due to its many hazards Nonetheless, even if hydrogen incidents can lead to quite large consequences, risk contribution in general is low o Inventories not large o Jet fires more common than explosions CFD modelling of an accidental sequence of events is very useful and can support: o Understanding the hazards o Optimization/increasing performance of the mitigation measure (gas detection, ventilation rate,...) o Reducing the potential consequences and design accordingly o Verifying the design and demonstrate compliance o Reducing risk and increase asset integrity However, many gaps remain in the reliability of CFD models (and also regulatory framework which needs to be harmonised) Possible to derive useful trends and results if one appreciates the limitations and use CFD tools carefully