535

A publication of

CHEMICAL ENGINEERING TRANSACTIONS

VOL. 77, 2019 The Italian Association
of Chemical Engineering
Online at www.cetjournal.it

Guest Editors: Genserik Reniers, Bruno Fabiano

Copyright © 2019, AIDIC Servizi S.r.l.
ISBN 978-88-95608-74-7; ISSN 2283-9216 DOI: 10.3303/CET1977090

Safety of Cryogenic Storage Facilities of Flammable Gases

Tommaso Iannacconea, Federica Ovidib, Giordano Emrys Scarponia, Gabriele
Landuccib,*, Valerio Cozzania
a
Department of Civil, Chemical, Environmental and Materials Engineering – University of Bologna, Via Terracini 28, 40131
Bologna (Italy)
b
Department of Civil and Industrial Engineering - University of Pisa, Largo Lucio Lazzarino 2, 56126 Pisa (Italy)
[email protected]

In recent years, cryogenic storage of flammable gases spread beyond conventional applications. Actually,
besides large-scale cryogenic transport by ship, loading/unloading and bulk storage, a number of smaller
scale applications are proposed, in particular for liquefied natural gas (LNG). However, this growth comes with
safety concerns arising from the flammable nature of natural gas and the amounts of this hazardous material
stored and processed in LNG facilities. A long-lasting experience is present in the chemical industry
concerning the medium and small-scale cryogenic storage of flammable gases as e.g. ethylene and
propylene. In the present study, the operating experience deriving from such technologies is used as a basis
to assess the safety of new supply chains based on small-scale storage of LNG and other cryogenic fuels.
Preliminary hazard identification studies were used to identify credible accident scenarios. Based on current
industrial practice and on proposed new installations, reference technologies were defined and considered in
the study. A comparative analysis of the relevance and severity of reference accident scenarios was carried
out. Safety key performance indicators based on the concept of inherent safety were defined, providing a
simplified tool to support safety assessment and emergency response in cryogenic storage facilities.

1. Introduction
The limitation of pollutant emission related to navigation activities has favoured the use of alternative fuels; in
particular LNG is a promising candidate for replacing fuel oil in specific sensitive areas subject to emission
control (Baresic et al., 2018). The development of a small-scale distribution grid for LNG is a required step in
order to foster the use of LNG as a fuel. This kind of installations are likely to be constructed in proximity of
populated areas, such as coastal or harbour zones mainly due to logistic needs (e.g. ease of supply, access to
main routes, etc.). As a result of heightened public concern, LNG onshore siting is strictly regulated, see e.g.
NFPA 59A (NFPA, 2016). As outlined in a previous work (Tugnoli et al., 2010), the societal acceptability of
newly-built storage and distribution facilities requires a proof that risk for population and environment is
negligible.
However, a long-lasting experience is present in the chemical industry concerning the medium and small-
scale cryogenic storage of flammable gases. For instance, storage, handling and transportation of refrigerated
ethylene is a well-established industry practice that can be taken as a reference when considering risk
mitigation solutions for cryogenic processes. A number of guidelines and recommendations concerning
cryogenic ethylene have been issued, see (EIGA, 2017).The assessment of safety performance for process
installations can be performed using a risk-based approach, also useful to support design solutions and to
take into account accident scenarios that may arise when processing hazardous cryogenic substances, as
illustrated in (Paltrinieri et al., 2011).
The aim of this work is to evaluate the inherent safety level of small-scale onshore LNG installation and
compare the results with existing industrial ethylene storage plants. Possible accident scenarios were
identified following a preliminary hazard analysis. A consequence-based methodology was used to produce an
inherent safety score of the considered installation, then the possibility of accident escalation (Landucci et al.,
2013) was questioned considering equipment damage thresholds for consequence assessment and

Paper Received: 6 December 2018; Revised: 29 May 2019; Accepted: 19 June 2019

Please cite this article as: Iannaccone T., Ovidi F., Scarponi G., Landucci G., Cozzani V., 2019, Safety of Cryogenic Storage Facilities of
Flammable Gases, Chemical Engineering Transactions, 77, 535-540 DOI:10.3303/CET1977090
536

calculating dedicated performance indicators that were used to quantify domino events impact for the
evaluated cryogenic facilities.

2. Methodology
2.1 Overview
The methodology described allows the application of a consequence-based approach for the comparison of
the inherent safety performance of different cryogenic onshore installations. Main concepts behind this
approach are adapted from a previous work (Tugnoli et al., 2007). A set of key performance indicators (KPIs)
was defined, based on results obtained from consequence assessment, providing an effective and reliable
metric for the comparison. Flowchart reported in Figure 1 summarizes the main steps of the methodology.

Step 1 Step 2 Step 3 Step 4 Step 5

Reference Process LOC definition Consequence Calculation of
scheme characterization and credit modelling KPIs
definition factors
assignment

Figure 1: Proposed methodology flowchart

The first step is the definition of a reference process scheme representative for the small-scale onshore
installation, outlining the required process units and the facility layout. Then, process data are gathered for
each assessed unit, defining hazardous substance physical state and quantity present at the facility. Once
having defined all process equipment needed, possible accident scenarios are identified following an event
tree analysis: the initiating release events and the relevant frequencies (expressed by a credit factor) are
defined on the basis of suggestions from recognized technical literature (Uijt de Haag and Ale, 2005). A
damage distance is then calculated using well-known consequence analysis models, as those proposed by
(Van Den Bosh and Weterings, 2005): the maximum distance at which dangerous effects threshold values
(such as those proposed by national laws) are reached is the input for the following KPI calculation.
The set of indicators adopted in this work includes two main categories of indices, depending on the damage
target considered:
1. Human hazard indicators,
2. Escalation hazard indicators.
The first category reflects hazards affecting plant operators and/or residential population, since damage
distances are calculate using threshold values for humans, specifically intended for this purpose. The second
group considers domino effect thresholds for the accidental scenario outcomes considered (Landucci et al.,
2008). As reported in details in (Landucci et al., 2008), a potential hazard index and an inherent hazard index
are defined. The first one expresses the maximum area that can be affected by the worst-case scenario
dangerous effects originating from the i-th process unit, while the latter introduces the impact of the loss of
containment event frequencies. Table 1 reports a summary of considered indicators, together with their
definition.

Table1: Summary of considered key performance indicators

Indicator Units Equation Target Eq.
Unit Potential hazard m2 = ℎ, Human (1)
2
Unit Inherent hazard m /y =∑ , ∙ℎ, Human (2)
Overall Potential hazard m2 =∑ Human (3)
Overall Inherent hazard m2/y =∑ Human (4)
Unit Potential Escalation hazard m2 = , Equipment (5)
2
Unit Inherent escalation hazard m /y =∑ , ∙ , Equipment (6)
Overall Potential escalation hazard m2 =∑ Equipment (7)
Overall Inherent escalation hazard m2/y =∑ Equipment (8)

The damage distance calculated for the j-th LOC of the i-th process unit considered is represented by (hi,j),
while , is the credit factor of the j-th LOC event, and ni is the number of LOCs considered for the i-th unit.
 537

The overall values of the above-mentioned indicators are obtained summing over all process units identified
for the considered reference scheme, as expressed by Equations 3 and 4 in Table 1. Similarly, the escalation
KPIs are defined as indicated by Equations 5 and 6 in Table 1. The escalation vector (ei,j) contains the domino
damage distances calculated. Subscripts have the same meaning as for the human target KPIs.

3. Definition of the case study

3.1 Process description
The case study proposed in this paper covers a small-scale onshore terminal located in a coastal area. This
facility can receive the liquefied gas directly from carriers thanks to a 2 km dedicated cryogenic pipeline
installed on a jetty. The terminal features four tanker truck loading bays that can provide an overall loading
3
capacity of 240 m /h. Liquefied gas is transferred using 3” diameter loading arms. Seven total containment
3
cryogenic pressure vessels allow an overall storage capacity of 10,000 m of cryogenic liquid. These tanks are
connected with a 16” manifold and to a distribution pipeline which serves the loading bays. Cryogenic pumps
and a boil-off gas (BOG) re-liquefaction system are also present. A simplified process flow diagram of the
considered onshore facility is shown in Figure 2.

Figure 2: Simplified scheme considered for onshore storage & distribution facility. Continuous lines represent
liquid pipelines, dashed ones are vapour lines associated with the BOG.

As mentioned in Section 1, two different substances, i.e. LNG and ethylene, have been considered as stored
fluid for the small-scale terminal shown in Figure 2. Table 2 and Table 3 show the operating conditions
implemented for the LNG and the ethylene case studies, respectively.

Table 2: Process conditions considered for LNG onshore terminal. For unit tags refer to Figure 2.
Unit
Parameter Jetty Storage tank Tank Loading bay Loading bay BOG
pipeline D01 – D07 manifold pipeline transfer arm pipeline
Inventory (kg) - 648,648 - - - -
Flow-rate (kg/s) 1,543 - 30.57 7.63 7.63 0.15
Pressure (MPa) 0.50 0.45 0.45 0.70 0.70 0.45
Temperature (K) 114 130 128 129 129 135
Nominal diameter (mm) 914.4 - 406.4 101.6 76.2 254.0
State Liquid Liquid Liquid Liquid Liquid Vapour
538

Table 3: Process conditions considered for ethylene onshore terminal. For unit tags refer to Figure 2
Unit
Parameter Jetty Storage tank Tank Loading bay Loading bay BOG
pipeline D01 – D07 manifold pipeline transfer arm pipeline
Inventory (kg) - 746,031 - - - -
Flow-rate (kg/s) 1,842 - 34.78 8.38 8.38 0.16
Pressure (MPa) 0.50 0.45 0.45 0.70 0.70 0.45
Temperature (K) 180 199 197 198 198 205
Nominal diameter (mm) 914.4 - 406.4 101.6 76.2 254.0
State Liquid Liquid Liquid Liquid Liquid Vapour

3.2 Definition of hazardous scenarios

A selection of loss of containment (LOC) events was considered following guidance provided by technical
literature (Uijt de Haag and Ale, 2005). Each LOC event is associated with a credit factor that represents the
frequency of occurrence of the considered release for a specific process unit. Frequency values were
retrieved from technical literature (Uijt de Haag and Ale, 2005) and release frequency database (IOGP, 2010).
Release frequency for the tanker truck loading arm (expressed in events per hour) was multiplied by the yearly
operational hours of all the four loading bays (estimated in a total of 14000 hours per year), in order to
consider the simultaneous use of all the bunker lots.
Depending on the category of evaluated process units, the release events reported in Table 4 were deemed
possible.

Table 4: Definition of LOC events

LOC Description
R1 Small leak, continuous release from a 10 mm equivalent diameter hole
R2 Catastrophic rupture, release of the entire inventory in 600 s
Catastrophic rupture, instantaneous release of the entire inventory and release from
R3
the full-bore feed pipe
R4 Pipe leak, continuous release from a hole having 10% of pipe diameter
R5 Pipe rupture, continuous release from the full-bore pipe

The final dangerous outcomes originating from the considered release events were obtained performing an
event tree analysis and following suggestion by (Vílchez et al., 2011). Dangerous scenarios evaluated in the
present study were pool, jet, and flash fires, in addition to vapour cloud explosions (VCE). The latter scenario
has been considered to account for the presence of potentially congested areas on small scale terminals,
where confinement can result from the limited space available and may induce flame acceleration (Van Den
Bosh and Weterings, 2005).
3.3 Consequence assessment
Accident scenario consequences were evaluated implementing well known integral models for the estimation
of physical effects (Van Den Bosh and Weterings, 2005). The consequence modelling of hazardous LNG
scenarios was carried out disregarding mixture composition, i.e. considering natural gas as pure methane, to
avoid results variability. The cryogenic pipeline used for cargo ships unloading operations has a total length of
approximately 5 km. Despite this, consequence evaluation for this piece of equipment only focused on the
aboveground pipeline section, which is the one on the jetty, placed at a distance of about 3 km from the tank
farm.
Damage distances refer to the worst case among that from the Pasquill weather categories 2F and 5D, that
are typically considered in risk assessments. Release sources were set at 1 m height and modelled as non-
impinging horizontal. Cryogenic liquid pool evaporation rates were calculated using available pool evaporation
sub models; the resulting vapour cloud was used to estimate flash fires and VCE consequences. Explosions
outcomes were estimated using the simplified approach provided by the TNT equivalency model (Van Den
Bosh and Weterings, 2005), accounting for the different substance reactivity by changing the explosion
efficiency parameter. According to what suggested by (Health and Safety Executive, 1986), explosion
efficiency for natural gas was considered equal to 3%, whereas VCE resulting from ethylene were modelled
using 6% efficiency.
 539

4. Results and discussion

4.1 Inherent safety assessment
An overview of inherent safety KPIs estimated considering the physical effects on humans is given in Tables 5
and 6, respectively for LNG and ethylene. Among the LNG terminal units, jetty pipeline shows the largest
values for both UPI and UHI (see Table 5). Instead, for the ethylene case, UPI value for storage tanks results
the highest among all process equipment considered, while the jetty pipeline is still deemed as critical due to
the highest value of UHI (Table 6). The relevant value of inherent hazard index for the jetty pipeline is mainly
related to the combination of high credit factor and wide damage distances. The relative weight of the different
accident scenarios considered over inherent human and escalation hazard is shown in Figure 3a. Flash fires
and VCEs are the sole scenarios contributing to the overall value of HI. In particular, flash fires give the higher
contribution in the LNG terminals, while VCEs consequences play a limited role. This trend is inverted when
considering ethylene facility as a result of the higher severity of VCE, due to the higher ethylene reactivity.

Table 5: Results of consequence assessment and calculated KPIs for LNG onshore facility. Values marked
with a * were not considered for calculation of escalation KPIs.
Unit LOC Cfi,j (1/y) hi,j (m) ei,j (m) UPI (m2) UHI (m2/y) UPD (m2) UHD (m2/y)
Storage tanks D01 - 07 R1 1.00 × 10-5 48.08 44.21 5.43 × 106 5.07 4.55 × 106 2.84
R2 5.00 × 10-7 2,161 2,134
R3 5.00 × 10-7 2,330 1,040
-3 6
Jetty pipeline R4 1.00 × 10 308.40 291.43* 6.96 × 10 1,487 - -
R5 2.00 × 10-4 2,638 2,625*
Tank manifold pipeline R4 2.50 × 10-5 248.74 239.13 1.58 × 105 2.34 1.54 × 105 2.20
R5 5.00 × 10-6 397.62 391.86
Pipeline to loading bay R4 2.00 × 10-4 57.40 55.15 5.49 × 104 2.31 5.10 × 104 2.14
R5 3.00 × 10-5 234.37 225.85
Loading bay arm R4 4.15 × 10-3 44.51 43.39 5.49 × 104 31.00 5.10 × 104 28.96
R5 4.15 × 10-4 234.37 225.85
BOG pipeline R4 5.00 × 10-6 19.79 11.78 2.88 × 105 0.29 2.76 × 105 0.28
R5 1.00 × 10-6 536.45 524.93

Table 6: Results of consequence assessment and calculated KPIs for ethylene onshore facility. Values
marked with a * were not considered for calculation of escalation KPIs.
Unit LOC Cfi,j (1/y) hi,j (m) ei,j (m) UPI (m2) UHI (m2/y) UPD (m2) UHD (m2/y)
Storage tanks D01 - 07 R1 1.00 × 10-5 76.03 72.05 6.83 × 106 5.66 6.72 × 106 5.37
R2 5.00 × 10-7 2,613 2,593
R3 5.00 × 10-7 2,094 1,979
Jetty pipeline R4 1.00 × 10-3 359.55 334.85* 4.56 × 106 1,040 - -
R5 2.00 × 10-4 2,134 2,108*
Tank manifold pipeline R4 2.50 × 10-5 288.25 239.13 8.94 × 105 6.55 4.67 × 105 3.77
R5 5.00 × 10-6 945.54 683.72
Pipeline to loading bay R4 2.00 × 10-4 70.24 67.7 8.27 × 104 3.47 7.33 × 104 3.12
R5 3.00 × 10-5 287.55 270.79
Loading bay arm R4 4.15 × 10-3 88.15 83.64 8.27 × 104 66.52 7.33 × 104 59.42
R5 4.15 × 10-4 287.55 270.79
BOG pipeline R4 5.00 × 10-6 21.20 12.42 3.58 × 105 0.36 3.38 × 105 0.34
R5 1.00 × 10-6 598.42 581.44

4.2 Accident escalation assessment

The evaluation of domino-potential consequences was performed disregarding all dangerous scenarios not
capable to affect the tanks with hazardous radiation levels. In particular, all the accidental events investigated
for the jetty pipeline were not able to affect the storage tanks; hence, they were not taken into account for
calculation of the escalation indices. Tables 5 and 6 report the results of the analysis for the LNG and
ethylene, respectively. On the basis of results, loading bay arms are deemed as critical units, in terms of
inherent escalation hazard, due the high frequencies of loss of containment, although the escalation distances
are not particularly extended. Considering only the maximum affected area, storage tanks are characterized
540

by the highest values of unit potential escalation index for both the LNG and ethylene case. Considering the
contribution of the single scenarios on the overall inherent escalation index, VCE scenarios account for the
totality of HD index scores, due to large escalation distances. A minor contribution is related to jet fires for the
ethylene facility, as can be seen in Figure 3b.
100% FF VCE 100%
90% 90% JF VCE
80% 80%
70% 70%
60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
a) LNG Ethylene b) LNG Ethylene

Figure 3: Accident scenario contribution for HI (a) and HD (b) indices. FF= Flash fire; JF= Jet fire; VCE=
Vapour cloud explosion.

5. Conclusion
An inherent safety comparison of cryogenic substance storage and distribution facilities was made. A
consequence-based method was implemented in order to calculate performance indicators. The aim was to
identify the criticalities that need to be further investigated during the design stages. Comparison of an
ethylene terminal with an equivalent LNG facility shows a lower safety performance of the former. The
analysis of critical equipment identifies the loading bay transfer arms as most critical units. The presented
method may serve as first guidance for addressing further accident escalation studies, highlighting the
dangerous scenarios that may impact over storage tanks or tanker trucks. A further step, representing an
important contribution for safety studies, may be the integration of the proposed methodology with
computational models for prediction of thermal and mechanical response of fire-exposed vessels (Landucci et
al., 2013).

References
Baresic D., Smith T., Raucci C., Rehmatulla N., Narula K., Rojon I., 2018, LNG as a Marine Fuel in the EU:
Market, Bunkering Infrastructure Investments and Risks in the Context of GHG Reductions, UMAS,
London, UK.
EIGA, 2017, Safety in Storage, Handling and Distribution of Cryogenic Ethylene, EIGA, Brussels, Belgium.
Health and Safety Executive, 1986, The effect of explosions in the process industries, Loss Prevention
Bulletin, 68, 37–47.
IOGP, 2010, Process Release Frequencies, Risk Assessment Data Directory, London, United Kingdom.
Landucci G., Cozzani V., Birk M., 2013, Heat Radiation Effects, Chapter In: Reniers G. and Cozzani V. (Eds.),
Domino Effects in the Process Industries: Modelling, Prevention and Managing, Elsevier, Amsterdam, The
Netherlands, 70-115.
Landucci G., Tugnoli A., Cozzani V., 2008, Inherent safety key performance indicators for hydrogen storage
systems, Journal of Hazardous Materials, 159, 554–566.
NFPA, 2016, Standard for the Production, Storage, and Handling of Liquefied Natural Gas (LNG) NFPA 59A,
National Fire Protection Agency, Quincy (MA), USA.
Paltrinieri N., Tugnoli A., Bonvicini S., Cozzani V., 2011, Atypical scenarios identification by the DyPASI
procedure: Application to LNG, Chemical Engineering Transactions, 24, 1171–1176.
Tugnoli A., Cozzani V., Landucci G., 2007, A consequence based approach to the quantitative assessment of
inherent safety, AIChE Journal 53, 3171–3182.
Tugnoli A., Paltrinieri N., Landucci G., Cozzani V., 2010, LNG regasification terminals: comparing the inherent
safety performance of innovative technologies, Chemical Engineering Transactions, 19, 391–396.
Uijt de Haag P.A.M., Ale B.J.M., 2005, Guidelines for Quantitative Risk Assessment (Purple Book), Committee
for the Prevention of Disasters, The Hague, the Netherlands.
Van Den Bosh C.J.H., Weterings R.A.P., 2005, Methods for the Calculation of Physical Effects (Yellow Book),
Committee for the Prevention of Disasters, The Hague, the Netherlands.
Vílchez J.A., Espejo V., Casal J., 2011, Generic event trees and probabilities for the release of different types
of hazardous materials, Journal of Loss Prevention in the Process Industries, 24, 281–287.