i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Analysis of hydrogen incidents to support risk assessment

Nouman Rafique Mirza*, Sven Degenkolbe, Werner Witt

Plant Design and Safety Technology, Brandenburg Technical University, Burger Chaussee 2, 03044 Cottbus, Germany

article info abstract

Article history: Hydrogen is an emerging alternative fuel, yet its properties like wide flammability range,
Received 5 April 2011 extremely fast burning rate (order of magnitude larger compared to natural gas) and the
Received in revised form considerably high amount of energy released when it burns or explodes render it as
12 June 2011 dangerous, if not handled with care. Hydrogen Incident Reporting Database (HIRD) is one of
Accepted 14 June 2011 the various databases which have been generated to collect incident information in
Available online 23 July 2011 hydrogen industry. In this study, 32 chosen (from HIRD) hydrogen processing incidents
have been analyzed to learn about their root causes. As a result of the study, statistical
Keywords: values about the effects, causes and consequences as well as a check-list for avoiding these
Hydrogen incidents, have been developed. The support to risk assessment is mainly directed to the
Safety analysis of weak points and system optimization. For support of various aspects of risk
Risk assessment analysis an extension of incident analysis and its documentation is recommended.
Loss prevention Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Incident analysis reserved.

1. Introduction a comprehensive hydrogen incident database, techniques

like Bayesian approach have been used to fit hydrocarbon
A superior fuel must be a convenient transportation fuel, it failure frequency data to hydrogen for its risk and safety
should be versatile (easily convertible to other forms of distances calculations [5e7]. But the Bayesian approach is
energy), it must have high utilization efficiency, must be still in the developing phase [5]. The data on hydrogen inci-
environment friendly, cheap and safe [1]. Keeping these dents is not so scarce, but it is scattered (as shown in Table 4).
factors in mind, hydrogen outshines all of its present In the past, researchers [8,9] have tried to combine the orig-
competitor fuels and comes out to be the most suitable for inal hydrogen incident data and have performed conven-
present and future use [1,2]. Hydrogen’s high energy content, tional statistical analysis to extract useful information on
low ignition energy, fast burning speed, extensive flamma- hydrogen safety. The approach is still useful and provides
bility and detonability ranges make it a highly unsafe fuel, if information on which direction to exert the future research
not handled with care [3]. A pre-requisite for commercial efforts. Such information from the analysis of incidents
application of hydrogen is that the safety of the required can also support some steps in the risk assessment shown
infrastructure is investigated and that its design is made such in Table 1.
that the associated risk is at least not significantly higher than In risk assessment possible incident scenarios have to be
that of existing fuel supplies [3]. defined (steps 3, 6). Potential system and external effects have
Knowledge gained from incident analysis and investiga- to be evaluated (steps 3, 6e8). The risk analysis can be
tions help industries to form a better safety management improved if the scenarios included are not only based on
system which ensures a safer and healthier working envi- hypothetical weak point analysis, but also on real incidents.
ronment in their facilities [4]. Recently, due to lack of Comparing the consequences/effects of real incidents with

* Corresponding author. Tel.: þ49 176 6236 4585; fax: þ49 355 869 2107.
E-mail addresses: mirzanr@hotmail.co.uk (N.R. Mirza), degenks@tu-cottbus.de (S. Degenkolbe), witt@tu-cottbus.de (W. Witt).
0360-3199/$ e see front matter Copyright ª 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2011.06.080
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7 12069

Table 1 e Risk assessment steps. Table 3 e Terms used in the study of H2-incidents.
Step # Description Step # Description Term Definition

1 Task description 7 Weak point analysis/ Incident The loss of containment of material
Risk analysis part 2 or energy.
2 General data 8 Risk evaluation Near miss An event, which under slightly different
3 Risk potential study 9 Optimization I conditions might have become an incident.
4 Working concept 10 Multi-objective optimization Analyzed Possible root causes leading to undesired
(Optimization II) causes event (incident or near miss). They have
5 Specific data 11 Minimization of residual risk been characterized as Primary and
6 Weak point analysis/ Secondary Causes. Primary causes are the
Risk analysis part 1 event-initiating happenings while the
Secondary are those which followed the
Primary causes.
Consequence A measure of expected effects of the
the results of the consequence calculation, it is possible to results of an incident (in terms of effect
to personnel and plant damage).
validate the used models. Furthermore, the incident analysis
gives information about frequency of the initiating event
(leakage, pipe rupture .), and probabilities (ignition/
explosion). these databases, Hydrogen Incident Reporting Database
The present study investigates 32 selected hydrogen inci- (HIRD) [10] was selected for incident analysis due to a number
dents, which have taken place in the hydrogen processing of reasons.
industry. Potential causes of the incidents have been analyzed
and based upon these causes, general recommendations  HIRD is based purely upon hydrogen based incidents.
(lessons learned) have been made to avoid them in future.  Technical information regarding incidents is posted well in
Table 3 represents the terms used in the analysis of these 32 detail on the server which gives a better understanding of
H2-based incidents. incidents to the readers.
 Probable causes are reported along with the scenario
descriptions.
2. Data collection  Consequences are stated as ‘property damage’ and ‘deaths/
injuries’.
Learning from previous incidents is an old and effective tech-  Suggestions regarding ‘lessons learned’ are made.
nique in the process industry. For this reason, various data-  HIRD categorizes incidents into various sections like valves,
bases were generated which started acting as platforms for the pipes, storage vessels etc. This not only makes navigation
collection of incidents related to hydrogen (see Table 4). These into various kinds of incidents easier but it also helps in
databases pose extensive information on H2-incidents. Out of estimating the most vulnerable portions of the process
plant.
 Dates of the incidents are mentioned which makes easier to
predict the incident frequency in various time intervals.
 Database is regularly and frequently updated.
Table 2 e Incident information to support risk
assessment steps, examples. HIRD currently poses a total of 194 incidents (last updated:
Risk analysis/weak point analysis (support of weak 16/05/2011) and is maintained by the Pacific Northwest
point analysis step 6) National Laboratory (PNNL), USA. The main criteria for
 Initiating event (external fire, explosion, valve opening, consideration of any safety event record in HIRD are avail-
power failure, freezing of valves/safety valves, human
ability of sufficient information to establish “lessons learned”
errors at maintenance, control, material,
management, operation
of relevance to hydrogen production, storage, transmission
 Safe guards (common cause failure) and use [4]. These “lessons learned” are part of the database
 System consequence (damage of internals) software used to collect the incident information. To secure
 Incident scenarios (domino effects) the privacy of the industries, all the names of the data
Consequence calculation (validation of consequence providing entities are omitted carefully, and rest of the data,
calculation steps 3, 7)
along with all the technical details provided, is updated on the
 Release (leak size, one or two phase flow)
server.
 Dispersion (heavy/neutrally buoyant gas)
 Consequence/effect (flammability distance, heat radiation, The 32 incidents collected from HIRD for analysis are most
deflagration or detonation) relevant to hydrogen processing industry. These don’t include
Frequency and probability calculation any domestic, refueling station, NASA or transportation inci-
(Risk assessment steps 3, 6, 7) dents. These include incidents related to valves, piping,
 Frequency of initiating events flanges, storage vessels, process vessels etc. While selecting, it
 Failure probability of safeguards
was kept in mind that necessary information about the inci-
 Ignition probability
dent is available, based upon which it was possible to prepare
Recommended improvements (Risk assessment steps 6, 9, 10, 11)
the check-list.
12070 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7

Table 4 e Important databases about industrial incidents.

a b
Database Name Number of Web address Administered by
incidents

HIRD (Hydrogen Incident 194 http://www.h2incidents.org/ Pacific Northwest National

Reporting Database) Laboratory, USA.
HIAD (Hydrogen Incident and 253 https://odin.jrc.ec.europa.eu/hiad/ European Commission’s Joint Research
Accident Database) global_view.hiad Center (JRC), Petten, Netherlands.
ACUsafe (US Chemical Safety No information http://www.acusafe.com/Incidents/ US Chemical Safety Board, USA.
Board monitored database) available. frame-incident.htm
eMARS (Major Accident 743 http://mahb-srv.jrc.it/typo3/?id¼4 Major Accident Hazards Bureau,
Reporting System) JRC (EC), Italy.
FACTS (Failure Accidents 24,100 http://www.factsonline.nl/ TNO Industrial and External Safety
Technical information System) Department, Netherlands.
ERNS (Emergency Response No information http://www.rtknet.org/db/erns/substance OMB Watch (A non-profit organization), USA.
Notification System) available.
ARIA (Analysis, Research and 37,000 http://www.aria.developpement- French Ministry of Ecology and
Information on Accidents) durable.gov.fr/barpi_stats.gnc Sustainable Development, France.
ARIP (Accident Release 4946 http://www.epa.gov/oem/tools.htm#arip Environmental Protection Agency, USA.
Information Program)

a last updated: 16th May, 2011.

b last accessed: 18th October, 2010.

misses. No distinction could be made between spontaneous

3. Data analysis and delayed ignitions from the analysis.
In the study, the analyzed causes have been categorized
For the purpose of analysis of incidents, some abbreviations under five major classes. These are ‘management’, ‘design
have been used. A table has been attached at the end of error’, ‘technical’, ‘maintenance’ and ‘operator error’.
document as Appendix A which shows all abbreviations used ‘Management’ means a wrong decision, action or negligence
in the analysis of incidents. The analysis of the 32 H2-based was undertaken by the section in-charge of the plant which
incidents is presented in Appendix B. resulted in an incident. ‘Design error’ includes all the
In the analysis, the Primary Causes are represented by an possible design flaws which were present before plant
asterisk symbol while all of the other causes (without asterisk) commissioning and later resulted in an incident. ‘Technical’
are Contributing or Secondary Causes. The table also enlists contains all the causes which were resulted due to the wrong
the type of equipment involved in the incident (e.g. valves, decision or an installation of wrong equipment by the on-site
piping, storage vessel etc.) and it also shows the year in which technical staff during plant operations. ‘Maintenance’
incident took place. represents all the causes which appeared during the repair
The analysis of these carefully chosen 32 H2-based inci- work of the plant while the ‘operator error’ category contains
dents showed that 43.7% of them resulted in fires, 31.3% in any wrong actions from the operators which resulted in an
explosions and 15.6% in both fire and explosion (Fig. 1). The incident. (Fig. 2)
lion’s share goes to fire which is due to the low ignition energy The analysis showed that most of the root causes for the
required for hydrogen and its wide flammability range. incidents were ‘Technical’. This meant that an on-site, on-
Explosions are usually caused by the accumulation and igni- spot wrong decision resulted in an uncontrolled situation.
tion of hydrogen in confined or semi-confined areas [3]. The These represented a figure of 33.3% while 23.5% of causes were
study showed that only 9.4% incidents resulted in leak or near

Fig. 1 e Effects of H2-incidents. Fig. 2 e Analyzed causes of H2-incidents.

 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7 12071

ensure a safer operation at hydrogen facilities (Appendix C). A

Table 5 e Primary and secondary causes of H2-incidents.
number of the recommendations in the check-list are already
Analyzed Contribution Contribution Overall mentioned in the best practice guidance [11,12].
cause as primary as secondary contribution
The information presented in the HIRD and other
cause (%) cause (%) (%)
present day databases only support the weak point anal-
Technical 23.5 9.8 33.3 ysis for risk assessment process (see Table 2). Often no
Maintenance 9.8 13.7 23.5 information is given about size of the leak and how much
Design Error 13.7 2.0 15.7
hydrogen was released. Therefore, consequence calcula-
Operator Error 13.7 2.0 15.7
tion models can not be validated. Because of this, an
Management 2.0 9.8 11.8
extension of the incident databanks and reports according
to the needs of risk analysis (see Table 2) is proposed.
Besides, the development of a comprehensive hydrogen
‘Maintenance’. ‘Design Error’ and ‘Operator Error’, each specific databank for determining the frequency of initi-
constituted 15.7% of causes while ‘Management’ was the least ating events and probability of safe guard malfunctions
to be blamed for any incident as they contributed only 11.8% can be accomplished.
in the analysis. The results showed need for more technically
sound personnel in the field of hydrogen handling.
Table 5 classifies the causes as Primary and Secondary
Causes. It is observed that ‘Design Error’, ‘Technical’ and
5. Conclusion
‘Operator Error’ are the most severe ones, as mostly they
appear as Primary Causes. ‘Maintenance’ appears equally as
Present study of 32 H2-based incidents which took place in
‘Primary Cause’ and ‘Secondary Cause’ while the ‘Manage-
industrial hydrogen processing plants showed that the effect
ment’ shows a reverse trend. It generally appears as
of 43.7% incidents was fire, 31.3% explosion, 15.6% both fire
a ‘Secondary Cause’ and does not result into a ‘Primary/Initi-
and explosion while only 9.4% incidents were near misses.
ating Cause’.
The study also concluded that generally ‘Technical’, ‘Operator
Fig. 3 shows the distribution of incidents based upon the
Error’ and ‘Design Error’ causes are the ones which fall in the
consequences. It shows that 59.4% of incidents resulted in
‘Primary Causes’ category (initiating the incidents) while
‘Property Damage’ while 25% of the incidents caused both
‘Maintenance’ appeared equally as ‘Primary Cause’ and
injury/death to personnel and property damage. In 12.5%
‘Secondary Cause’. ‘Management’ showed the reverse trend
of the incidents no significant damage was observed. This
and mostly it acts as a ‘Secondary Cause’ to other ‘Primary
is due to the fact that some of the incidents resulted in
Causes’. 59.4% of incidents resulted in property damage or
only ‘Leaks’ or ‘Near misses’ which constituted no signifi-
plant damage, while 25% resulted into an adverse effect on
cant damage to plant personnel or the plant itself. Only
both plant personnel and plant itself. Based upon the analysis,
3.1% of the incidents resulted in ‘Injury/Death’ of the plant
general recommendations are made which can significantly
personnel.
reduce the root causes of these incidents. There is a need to
The study shows that in 87.5% incidents some adverse
perform more extensive study of the mistakes made in the
effect was observed either on plant personnel or to plant itself.
past and implement the lessons learnt (recommendations)
These statistics clearly highlight the high demand of safety
from them so that in future, more aspects of risk assessment
precautions in H2 facilities.
will be supported.

4. Learning from past Appendix A.

Publications like [8,9] have focused on the causes of the inci-

dents in hydrogen installations. Based upon the analysis of
these 32 hydrogen processing incidents, a check-list along
with general recommendations has been prepared which can Table A.1 e Abbreviations used for analysis of H2-based
incidents.

Effect, analyzed causes and Abbreviation

consequences

Management Mgmt.
Design Error DE
Technical Tech.
Maintenance Maint.
Operator Error OE
Fire Fire
Explosion Expl.
Injuries/Deaths Inj./Deaths
Property damage PD
Fig. 3 e Consequences of H2-incidents.
12072 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7

Appendix B.

Table B.1 e Analysis of H2-based incidents in process industry.

þ
Sr. # Incident Effect Analyzed causes Consequences

Fire Expl. Mgmt. DE Tech. Maint. OE Inj./Deaths PD

1 PIPING Fire and explosion in an effluent pipe of    *  

a refinery connected to a high pressure
reactor (1997) Ref: 1-12-2010
2 Fire due to rupture of stainless steel pipe  * 
in a Steam Methane Reformer (SMR) (1996)
Ref: 1-11-2010
3 Explosion in piping connected to  * 
electrolytic cells in a paper and pulp
chemical plant (2007) Ref: 5-26-2009
4 Fire in a pipe connected to desulfurization  *  
reactor in a refinery (1989) Ref: 8-8-2007
5 Fire and explosion at a hydrogen storage    *  
facility (reported in 2007) Ref: 1-19-2007
6 Explosion in a piping at a hydrogen  *  
producing electrolysis plant (2001) Ref:
1-12-2007
7 Fire in a carbon steel elbow at Resid  *   
Hydrotreater Unit (RHU) in a refinery
(2005) Ref: 1-11-2007
8 Hydrogen leak from a globe valve at  *
a power plant (1987) Ref: 1-10-2007
9 Explosion at piping in a Boiling Water  *  
Reactor (BWR) off-gas system (reported in
2006) Ref: 12-27-2006
10 Fire in a generator at a power plant facility  * 
(reported in 2006) Ref: 12-21-2006
11 VALVES Hydrogen leak from a Gate valve in *
a Hydrotreater Unit of a refinery (2008) Ref:
9-9-2009
12 Fire in a safety valve at a chemical plant  *
(1998) Ref: 6-8-2007
13 Fire in a valve at hydrogen storage facility  * 
(1999) Ref: 12-20-2006
14 GASKETS Fire in a safety valve at a Succinic Acid  *
manufacturing chemical plant (1998) Ref:
6-8-2007
15 Fire in a valve due to a relatively smaller  * 
gasket installation at a chemical facility
(1996) Ref: 5-24-2007
16 FLANGES Hydrogen fire in a heat exchanger due to  *  
a flange leakage (1998)
Ref: 5-31-2007
17 Fire at an Alkylbenzene manufacturing  *  
chemical plant (1994) Ref: 8-1-2007
18 SAFETY DEVICES Explosion in a Hydrogen storage facility   * 
due to premature rupture of burst disc
(2008) Ref: 1-7-2010
19 Hydrogen explosion due to premature   *  
bursting of Pressure Relief Device (PRD) at
a coal-fired power plant (2007) Ref:
2-5-2008
20 PROCESS VESSELS Hydrogen fire and explosion in a refinery   * 
hydrogen production unit during
a turnaround/shutdown (1996) Ref:
7-19-2007
21 Hydrogen explosion due to microbial  * 
hydrogen formation in a pulp storage
tower (1979) Ref: 1-10-2007
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7 12073

Table B.1 (continued)

þ
Sr. # Incident Effect Analyzed causes Consequences

Fire Expl. Mgmt. DE Tech. Maint. OE Inj./Deaths PD

22 STORAGE VESSELS Hydrogen BLEVE (Boiling Liquid  *   

Expanding Vapor Explosion) in a chemical
plant (1974) Ref: 6-2-2010
23 Fire in a hydrogen storage vessel due to  * 
tank rupture (reported in 2007) Ref:
9-10-2010
24 Hydrogen fire and explosion at a hydrogen   *  
producing electrolyzer plant (2001) Ref:
1-12-2007
25 Hydrogen fire at a Hydrogen storage  *  
facility (1999) Ref: 12-20-2006
26 Explosion at a hydrogen storage facility  *   
(reported in 2006) Ref: 12-6-2006
27 MISCELLANEOUS Hydrogen fire in a battery manufacturing  *  
EQUIPMENTS plant (1991) Ref: 6-27-2007
28 Explosion due to hydrogen compressor    * 
failure in an electrolysis plant (reported in
2007) Ref: 1-10-2007
29 Hydrogen explosion at a Ni-Cd battery  *  
charging facility (reported in 2007) Ref:
2-7-2007
30 Hydrogen leak due to H2 vent blow down  * 
at a storage facility (2006) Ref: 1-5-2010
31 Hydrogen explosion in a vent pipe due to   * 
inadequate maintenance at a chemical
plant (2002) Ref: 6-18-2007
32 Hydrogen explosion at a water treatment  *  
facility (reported in 2006) Ref: 12-27-2006
Total (32 incidents) 14 10 6 8 17 12 8 9 27

þ www.h2incidents.org.
* indicates Primary/Initiating Causes (rest are Secondary/Contributing Causes).
12074 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7

Appendix C.

Table C.1 e Recommendations to avoid incidents in hydrogen industry.

a
Analyzed causes Summary Reference General recommendations (with
references)

Management Incomplete Operations & Maintenance 1-12-2010 Written SOPs and O&M procedures
(O&M) procedures i. Written, updated Operations & Mainte-
Poor radio communication. 1-12-2010 nance procedures should always be
Inadequate supervision 1-19-2007 available to operators. [1-12-2010,
Absence of hydrogen detecting equipment 1-10-2007, 1-10-2007]
(lack of response toward technical 6-27-2007 Written emergency procedures
recommendations) ii. Emergency procedures should be
Unavailable written SOPs (Standard 1-10-2007, written down to help the operators in case
Operational Procedures)/Improper shut 1-5-2010 of abnormal operations. [1-12-2010,
down procedures. 1-10-2007, 1-5-2010]
Operator supervision
iii. Operators should be supervised prop-
erly during maintenance work. [1-12-2010,
1-19-2007]
H2 leakage detection
iv. H2 detecting instruments should be
installed on various points in the H2
facility. This helps detection of any
unsighted leak. [1-10-2007, 6-27-2007]
Proper chain of commands
v. Most of the above problems can easily
be avoided if proper chain of commands is
followed. [1-10-2007]
Design Error Stress corrosion cracking by KOH because 1-11-2010 Material selection
of material incompatibility i. Usage of proper alloyed materials should
Corrosion due to NH4OH. 8-8-2007 be preferred to avoid corrosion in
Replacement of alloy steel elbow with 1-11-2007 hydrogen plants. [1-11-2010, 8-8-2007]
a carbon steel elbow which is less ii. Alloyed steel elbows should be used in
resistant to High Temperature Hydrogen hydrogen plants to avoid High Tempera-
Attack (HTHA) ture Hydrogen Attack (HTHA). [1-11-2007]
Bolts and flanges having different 5-31-2007 iii. In case of high temperature/pressure
Thermal Coefficient Of Expansion (TCOE) conditions, both flanges and bolts used
caused loosening of flange at high should be of the same material. This helps
temperatures, thus resulting in H2 escape in avoiding difference of thermal expan-
A single too large (20 in.) Gate Valve was 8-1-2007 sions which can cause loosening of joints.
used to shut off hydrogen supply which is [5-31-2007]
a pretty light (small molecular size) gas iv. Usage of polymeric braces should be
Installation of larger mesh size permeable 12-6-2006 avoided in high temperature applications.
membrane to restrain light H2 gas High melting point metals should be used.
Insufficient ventilation in the 6-27-2007 [1-5-2010]
containment Monitoring material strength
Usage of polymeric braces (having low 1-5-2010 v. Strength of various piping components
melting point) at high temperatures. Wind should be monitored through Positive
drag was also not considered in the Material Identification testing or X-ray
design. Fluorescence testing before and after
carrying out maintenance. [1-5-2010,
1-11-2007]
Selection of flow devices
vi. For light gases like H2, small diameter
valves (e.g. double block and bleed) should
be used. [1-10-2007, 8-1-2007]
Selection of permeable membranes
vii. For sealing purposes in H2 industry,
inert gas should be used instead of air
because air might form flammable
mixtures with H2.
(continued on next page)
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7 12075

Table C.1 (continued )

a
Analyzed causes Summary Reference General recommendations (with
references)

viii. When using permeable membranes to

restrain H2, the mesh size of membrane
should always be kept in mind. Smaller
mesh size membranes (e.g. Neoprene)
should be avoided in H2 systems because
of their poor efficiency to restrain H2.
[12-6-2006]
Plant design and weather conditions
ix. Hardware design should be compatible
with the weather conditions. For example,
in case of open systems, extra stresses on
the system due to wind drag must be
considered in the design. [1-5-2010,
6-27-2007]
Technical Failure of High Purity Hydrogen purity 1-12-2007 System redundancy
analyzers and rupture of welded pipe i. Performance and deterioration of
(weakened due to maintenance). various plant equipments like H2 purity
Failure of Globe Valve packing causing 1-10-2007 analyzers should be regularly monitored.
hydrogen escape. Absence of gas detecting These analyzers should be backed up with
instrument in the facility. Fail-safe mode, so that if these fail, then
Usage of smaller gasket in a safety valve 6-8-2007, 5-24-2007 the vent valve opens up automatically [1-
than the standard compatible size. 12-2007]
Failure of three valves due to inadequate 12-20-2006 ii. Vent valves should work in Fail-safe
monitoring and maintenance. mode so that when any back pressure
Improper nitrogen purging during change 8-1-2007 develops due to blockade, they open
of catalyst in the plant. automatically [1-7-2010]
Rupture disk failure. Vent stack also failed 6-2-2010 iii. There should always be a separate
because of blockade due to frozen water. electricity backup system for the plant.
Poor ventilation considerations for the 1-12-2007, 2-7-2007 Also, it should be assured that devices like
containment. pumps and compressors are completely
Usage of air instead of a noble gas as 12-6-2006 shut off before putting out the main power
a sealing media. supply in the plant [6-8-2007]
Poor electricity backup. 1-10-2007 iv. Valves should be thoroughly moni-
Premature burst disk failure due to stress 6-18-2007, 2-5-2008 tored. They should be checked periodi-
overloading. cally for any imbalance in load
Back pressure development on burst disc 1-7-2010 distribution which might result into a H2
(causing rupture) due to heat transfer leakage [6-8-2007]
from atmosphere. Plant design and weather conditions
Absence of ventilators on a pulp 1-10-2007 v. Civil architect of H2 containments is
processing tower. The biochemical important. Proper ventilation should be
reaction resulted in hydrogen production ensured for such facilities. Also, control
which accumulated and later resulted in rooms or other offices should not be
an explosion during abnormal/shutdown located near the process plant [1-10-2007,
period. 6-22-2007, 1-12-2007]

Selection of rupture disc and gaskets

vi. Conventional rupture discs should be
replaced with spring-style relief valves
which have a better efficiency. If rupture/
burst discs are used, then they should be
regularly monitored for any back-pressure
development which might cause
a premature disc failure [2-5-2008]
vii. Standard sized gaskets should be used
in the safety devices. A smaller gasket
may cause failure of the safety valve [6-8-
2007, 5-24-2007]

Selection of sealing media

viii. For sealing purposes in H2 industry,
inert gas should be used instead of air
because air might form flammable
mixtures with H2 [12-6-2006]

(continued on next page)

12076 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7

Table C.1 (continued )

a
Analyzed causes Summary Reference General recommendations (with
references)

Design of safety devices

ix. Safety devices (e.g. burst discs) should
be designed in such a way that moisture
can’t enter them. Moisture can freeze,
expand and cause stress overloading
which results in premature failure of burst
discs [6-18-2007]
Use of ventilators to flush out H2 in towers
x. In case of biochemical H2 production
during shutdown period, fans along with
some inert media should be used to flush
out biochemical H2 byproduct [1-10-2007]
Emergency alarm
xi. All H2 facilities should be equipped
with an emergency alarm which should be
distinguishable from other system alarms.
Maintenance Inadequate maintenance. 1-12-2007, 1-11-2007, Regular inspection and maintenance
1-7-2010, 2-5-2008, i. Plant should regularly be inspected by
12-20-2006, the responsible technical personnel. Most
2-7-2007, 6-18-2007 of the incidents happen due to insufficient
Missing Bull plug in the Gate valve. 9-9-2009 inspection and maintenance. [1-12-2007,
Welding of support at the pipe dead-end 8-8-2007 2-7-2007, 1-11-2007, 6-18-2007, 1-7-2010, 2-
which increased the rate of corrosion. 5-2008]
Welding at H2 line without following 12-27-2006 Strength of dead-end piping in flow
proper SOPs (Standard Operational systems
Procedures) during abnormal operations. ii. Strength of dead-end piping should be
Improper hot-bolting at high temperature 5-31-2007 continuously monitored in flow systems.
and pressure. [8-8-2007]
Valves
iii. After maintenance, all the valves
should be checked and closed properly.
Gate valves should be checked properly so
that no Bull plugs are missing from them.
[12-20-2006, 9-9-2009]
iv. After maintenance, thorough checkup
of valves and piping should be carried out.
Torque measurements should be per-
formed for valves to check for any imbal-
ance in load distribution. [5-31-2007]
Bolts and flanges
v. Proper hot-bolting should be performed
for high temperature processes. [5-31-
2007]
vi. Welding should be avoided on piping
system involving H2 flow. This increases
corrosion rate. If welding is unavoidable,
then proper SOPs should be followed
under the supervision of concerned tech-
nical authorities. [8-8-2007, 12-27-2006]
Safety devices
vii. Vent pipe caps (in burst discs) should
be regularly checked for maintenance.
Even small amounts of moisture can
freeze, expand and cause stress over-
loading which will result in failure of burst
disc. [6-18-2007]
Operator Error Lack of response from operators. Failure to 1-12-2010 Operator training
activate emergency reactor i. Special training should be provided to
depressurization at abnormally high operators to work on hazardous sites.
temperature. ii. Operators should be thoroughly briefed
Failure to follow SOPs during abnormal 5-26-2009 before carrying out maintenance. [7-19-
operations. 2007, 1-12-2010, 1-19-2007, 5-26-2009]
1-19-2007 Operator response time
(continued on next page)
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 6 ( 2 0 1 1 ) 1 2 0 6 8 e1 2 0 7 7 12077

Table C.1 (continued )

a
Analyzed causes Summary Reference General recommendations (with
references)

Unauthorized and improper purging iii. Proper SOPs and training should be
which resulted in the mixing of H2 & O2 provided to operators to increase their
which formed a flammable mixture. response time. Increased response time,
Welding on an H2 pipeline in which some 12-21-2006 ensures less pressure on operators which
H2 was left even after CO2 purging. results in reduced incidents at the plant
Overfilling of unit without checking the 7-19-2007 facility. [1-12-2010]
level indicator. Operator discipline
Water sprayed (to fight fire) on H2 storage 6-2-2010 iv. Operators should be continuously
tank, which froze in the vent stack, monitored and they shouldn’t be allowed
blocking it and causing an explosion. to carry out any maintenance work
Liquid N2 (used to extinguish fire in 9-10-2010 without prior authorization. [1-19-2007]
another unit) fell on a nearby H2 vessel,
cracking its vacuum jacket which resulted H2O and N2 (as fire extinguishers) on H2
in pressure loss and H2 fire. sites
Use of electrical equipment (pump) in H2 12-27-2006 v. Operators should be properly trained on
environment which caused an explosion. the usage of fire extinguishing materials
like water and nitrogen. Their improper
use can create big mishaps at H2 facilities.
[6-2-2010, 9-10-2010]
Welding and electrical equipments at H2
facilities
vi. Operators should be properly trained
and debriefed on the hazards of improp-
erly using high-temperature or electricity
involving processes in H2 facilities. Care-
less use of such processes has often
resulted in a severe incident. [12-27-2006]

a www.h2incidents.org

references [7] LaChance J, Houf W, Middleton B, Fluer L. Analyses to

support development of risk-informed separation distances
for hydrogen codes and standards. USA: Sandia Natl
Laboratories; 2009. DOE.
[1] Veziroglu TN, Barbir F. Hydrogen: the wonder fuel. Int J
[8] Zalosh RG, Short TP. Comparative analysis of hydrogen fire
Hydrogen Energy 1992;17(6):391e404.
and explosion incidents. Factory Mutual System.
[2] Winsche WE, Hoffman KC, Salzano FJ. Hydrogen: its future
Massachusettes, USA: Factory Mutual Research Corporation;
role in the nation’s energy economy. Science 1973;
1978.
180(Number 4093).
[9] Ordin PM. Review of hydrogen Accidents and incidents in
[3] Middha P. Safety and risk analysis studies for hydrogen
NASA Operations. 9th Intersociety Energy Conversion
Applications. Symposium on Loss Prevention and safety
Engineering Conference 1974; San Francisco, California,
Promotion in the process industries; 2010. Brugge, Belgium.
USA.
[4] Weiner SC, Dean J, Davis PB, Ruiz A. Incident reporting:
[10] www.h2incidents.org, [accessed 18.10.10], administered by
learning from experience. Presented Int Conf Hydrogen Saf
Pacific Northwest National Laboratory, Department of
2007; San Sabastian, Spain.
energy, USA.
[5] Pasman HJ. Challenges to improve confidence level of risk
[11] NASA Safety Standard for hydrogen and hydrogen systems.
assessment of hydrogen technologies. Int J Hydrogen Energy;
Washington: Office of safety and Mission Assurance; 1997.
2010. doi:10.1016/j.ijhydene.2010.05.019.
Report NSS 1740.16.
[6] Haugom GP, Friis-Hansen P. Risk modeling of a hydrogen
[12] ISO/PDTR-15916 Basic considerations for the safety of
refueling station using Bayesian network. Int J Hydrogen
hydrogen systems.
Energy; 2010. doi:10.1016/j.ijhydene.2010.04.131.