11 HAZARD TO LIFE

11.1 Background

11.1.1 In accordance with Section 3.4.12 of the EIA Study Brief (ESB-332/2020), a hazard to life assessment should be conducted to evaluate the risks associated with the existing Liquefied Petroleum Gas store in Tuen Mun Area 44 (the LPG Store) to the Project during the construction and operation phases.

11.1.2 The LPG Store is designated as a Potentially Hazardous Installation (PHI) owing to its bulk storage of over 25 tonnes of LPG, with a Consultation Zone (CZ) of 150m radius from its centre. CZs are established around PHIs to control development in the vicinity and prevent population accumulating to a point where societal risks may become unacceptable. Part of the proposed railway alignment and some works areas/works sites of the Project will be located within the CZ of the LPG Store (Figure No. C1502/C/TME/ACM/M59/101 refers), and therefore a hazard to life assessment is required to confirm that both the individual risk and societal risk during the construction and operation phases of the Project comply with the risk criteria stipulated in Annex 4 of the EIAO-TM.

11.1.3 The hazard to life assessment was conducted according to the requirements in Appendix J of the EIA Study Brief (No. ESB-332/2020). This section of the EIA Report presents the analysis and findings of the hazard to life assessment undertaken for the Project.

11.2 Assessment Criteria

11.2.1 Annex 4 of the EIAO-TM stipulates that the maximum level of offsite individual risk should not exceed 1 in 100,000 per year, i.e. 1×10^-5 per year. For societal risk, the criteria are expressed in the form of a chart presented in Diagram 11‑1, with the cumulative frequency (F) of N or more fatalities from incidents at a hazardous installation plotted against the number of fatalities (N). Three regions are delineated in the chart to represent societal risk at levels that are “acceptable” or “unacceptable”; or to be reduced to levels that are “as low as reasonably practicable” (ALARP) with the adoption of practicable and cost-effective measures.

Diagram 11‑1 Societal Risk Guidelines

11.3 Assessment Approach

11.3.1 The assessment consisted of the following six main tasks:

a) Data / Information Collection and Update: relevant data / information essential for the hazard to life assessment were collected;

b) Hazard Identification: a credible set of hazardous scenarios associated with the operation of the LPG Store was identified;

c) Frequency Estimation: the frequencies of each hazardous event leading to off-site fatalities based on the collected data and operation data for the LPG Store were estimated with the support of justifications from reviewing historical accident data and previous hazard to life assessments;

d) Consequence Analysis: the consequences of the identified hazardous scenarios were analyzed;

e) Risk Assessment and Evaluation: the risks associated with the identified hazardous scenarios were evaluated. The change in the risk level in terms of individual risk and societal risk due to the proposed development with the increase of the populations (with and without the project scenarios) was compared with the risk criteria to determine their acceptability; and

f) Identification of Mitigation Measures: where necessary, risk mitigation measures were identified and assessed to comply with the “as low as reasonably practicable” (ALARP) principle. Risk mitigation measures recommended in previous studies were reviewed, and practicable and cost-effective risk mitigation measures were identified and assessed as necessary. The risk outcomes of the mitigated case were reassessed to determine the level of risk reduction.

11.3.2 The hazard assessment covered the following four scenarios:

· Year 2026 without Project (2026 Base Case) – The risk posed by the LPG Store to the planned population in 2026, in the absence of the Project;

· Year 2026 with Project (Construction Phase) – The risk posed by the LPG Store to the planned population in 2026. This also accounted for the presence of the construction workers in close vicinity of the LPG Store and the potential impacts associated with the construction activities. Year 2026 was selected as the assessment year for construction phase of the Project since it was expected to be the peak construction year with maximum number of construction workers and construction vehicles within the CZ of the LPG Store;

· Year 2031 without Project (2031 Base Case) – The risk posed by the LPG Store to the planned population in 2031, in the absence of the Project; and

· Year 2031 with Project (Operational Phase) – The risk posed by the LPG Store to the planned population in 2031 upon the commissioning of the Project.

11.4 Description of Assessment Area

The LPG Store

11.4.1 ExxonMobil Hong Kong Limited (ExxonMobil) is the owner of the LPG Store, and IP&E GBA Limited becomes the Registered Gas Supply Company for the LPG Store starting from July 2021. The risk levels associated with the operation of the LPG Store were previously assessed in the Environmental Impact Assessment for Proposed Complex and Wholesale Fish Market at Area 44, Tuen Mun (Register No.: AEIAR-070/2003) (WFM EIA study)^[^1]. Based on the information presented in WFM EIA study, there were three underground storage vessels with capacity of 10 tonnes each in the LPG Store. LPG was delivered to the LPG Store by LPG road tankers with capacity of 8 tonnes and there were 260 annual deliveries, which were assumed to be made during daytime only^[^1]. A pipework with total length of 35m was assumed for the consideration of vaporisers’ failure for all the 3 vaporisers in the LPG Store.

11.4.2 An information request on the operation of the LPG Store was sent to ExxonMobil on 27 November 2020 to obtain the latest operation details of LPG Store but the request was not entertained for commercial confidentiality reasons.

11.4.3 Based on publicly available information such as the population census^[^2][3] and the latest information on authorised population from the website of the Hong Kong Housing Authority, the number of residents in Butterfly Estate, Wu King Estate and Siu Shan Court has followed a trend of continuous decrease since 2011. A 5-consecutive-day site survey was conducted to estimate the LPG throughput of the LPG Store and to reflect the latest situation of LPG usage. Based on the findings of the site survey, the number of annual LPG deliveries was estimated to be 365 and the time spent on site by a road tanker to complete LPG unloading for each delivery is about 40 minutes. The maximum capacity of LPG road tanker being used in Hong Kong is about 9 tonnes, and therefore 9-tonne LPG road tankers were adopted for a conservative assessment.

11.4.4 The assumptions adopted in the hazard to life assessment of the WFM EIA study^[^1] for the LPG facilities, including the number of LPG underground storage vessels, storage capacity of the storage vessels, and the 35m length pipework for the vaporisers, were also adopted for this assessment.

Population Data

11.4.5 Societal risk is a measure of the consequence magnitude and the frequency of the hazardous events. To establish the impact of any release (expressed as the number of people likely to be affected), it is necessary to have a good knowledge of the future population levels around the LPG Store. This includes residential population, institutional / commercial population, transport population and construction worker population during the construction of the Project, but excludes the staff present at the LPG Store (as they are considered to be voluntary takers of risk).

11.4.6 The locations of population groups covered in this assessment in relation to operation of the LPG Store during the construction and operational phases of the Project are presented in Diagram 11‑2 and Diagram 11‑3 respectively, and the details on the estimated population of each population groups are provided in Appendix 11.1. The locations of population groups assessed for the base case scenarios are also presented in Diagram 11‑4 for easy reference.

Diagram 11‑2 Location of Population Groups in relation to LPG Store (Construction Phase)

Diagram 11‑3 Location of Population Groups in relation to LPG Store (Operational Phase)

Diagram 11‑4 Location of Population Groups in relation to LPG Store (Base Cases for Year 2026 and Year 2031)

Construction Workers for the Project

11.4.7 Based on the construction programme of the Project, construction activities between A16 Station and Tuen Mun River Bridge that fall within the 150 m radius CZ of the LPG Store would be undertaken between 2024 and 2028. The number of construction workers was estimated according to the Engineer’s experience / analysis based on projects of a similar nature and scale.

11.4.8 The land in the immediate vicinity of the LPG Store, i.e. the existing open car park (ID#9b in Diagram 11‑2 refers), the landscaped area surrounding the LPG Store and also the promenade (ID#9a in Diagram 11‑2 refers) next to the LPG Store, would be used as works areas for the unloading and inspection of precast segments, and the contractor’s site office. It was assumed that 50 construction workers would be present in these works areas during the construction period. These areas would be reinstated to their present uses after the completion of the construction of the Project.

11.4.9 In addition to the works areas mentioned above, four works sites for the viaduct construction and road modification works (ID#10a and #10b in Diagram 11‑2 refers) would fall within the CZ of the LPG Store. It was assumed that not more than 100 construction workers would be present in these works sites at any one time during the construction period of the Project.

Land and Building Population

11.4.10 Population estimations and projections to future assessment years were based on desktop review of the WFM EIA study and verified / updated through site observations. The maximum number of population in each area is listed in Table 11.1, and the details on the estimated population for each population group at different time modes are provided in Appendix 11.1. The population was estimated with the following assumptions:

a) The population at Tuen Mun Road Safety Town is zoned as “Open Space” in the Outline Zoning Plan S/TM/35. was estimated by using a population density of 0.01 person per m²;

b) The population at Tuen Mun Area 44 Joint User Complex and Wholesale Fish Market, Tuen Mun Wu Hong Clinic and the promenade were based on the assumptions adopted in the WFM EIA study;

c) The population at the Wu Shan tennis court was estimated based on the assumption of four persons using each of the three tennis courts;

d) The population at the open car park was estimated based on-site observations; and

e) An average of 5 % of the building population was taken to be outdoors; while 100% of the population at the works areas for the unloading and inspection of precast segments, and the contractor’s site office was taken to be outdoors.

Table 11.1 Population within Consultation Zone of the LPG Store

ID	Description	Maximum Population
ID	Description	Construction Phase (Year 2026)	Operational Phase (Year 2031)	2026 Base Case and 2031 Base Case
1	Wu Shan Recreation Playground	0	108	108
2	Open Car Park	0	20	20
3	Promenade	0	10	10
4	Tuen Mun Area 44 Joint User Complex and Wholesale Fish Market	147	147	147
5	Wu Shan Tennis Court	12	12	12
6	Tuen Mun Wu Hong Clinic	193	193	193
9a	Land WA 5.4	20	0⁽¹⁾	0
9b	Land WA 5.5	30	0⁽¹⁾	0
10a	Lands WS 5.2, 5.3 & 5.6	88	0⁽¹⁾	0
10b	Land WS 5.8	12	0⁽¹⁾	0

Note:

(1) The works areas and works sites (i.e. population ID#9a, #9b, #10a and #10b) will be reinstated to their present uses (i.e. population ID#1, #2 and #3) after the completion of the construction of the Project.

Road Population

11.4.11 The road population considered in this assessment included the population travelling in motor vehicles on Hoi Wong Road, Wu Shan Road and Wu Hong Street. The speed of travelling on these three roads was assumed to be 50km/hr. The traffic population was derived with the following equation:

11.4.12 Based on “The Annual Traffic Census 2019” ^[4], the occupancies for each vehicle type were taken as the average at the Screenline Y-Y (Boundary between Tuen Mun and Yuen Long), which were considered representative of the road traffic in the assessment area.

11.4.13 The traffic population was assumed to be 100% outdoor. The estimated road population in the construction phase scenario (i.e. Year 2026) and the operational phase scenario (i.e. Year 2031) are presented in Table 11.2 and Table 11.3, respectively, and the detailed calculations are provided in Appendix 11.1.

Table 11.2 Road Population in Construction Phase Scenario (Year 2026)

Population ID	Description	Maximum Population ^{[Note 1]}
Population ID	Description	Daytime	Night-time
8a	Hoi Wong Road	25	12
8b	Wu Shan Road	13	7
8c	Wu Hong Street	2	1

Note 1: Road population was estimated based on Traffic Impact Assessment forecasted for Year 2026.

Table 11.3 Road Population in Operational Phase Scenario (Year 2031)

Population ID	Description	Maximum Population ^{[Note 1]}
Population ID	Description	Daytime	Night-time
8a	Hoi Wong Road	25	12
8b	Wu Shan Road	10	5
8c	Wu Hong Street	2	1

Note 1: Road population was estimated based on Traffic Impact Assessment forecasted for Year 2031.

Railway Population

11.4.14 The railway population considered in this assessment included the on-train population of the LRT and the Project. The railway population on the LRT was estimated based on the design capacity of a LRT Vehicle as quoted in LC Paper No. CB(4)854/15-16(07) and the train frequency, the calculated railway population on the LRT was considered to be conservative and was adopted for all the assessment scenarios; while the railway population of the Project was estimate based on the average train loading and the train frequency during the peak hour. The detailed calculations for the railway population are shown in Appendix 11.1.

11.4.15 The population along a railway segment was calculated using the following formula:

11.4.16 The population associated with the railway traffic was modelled as 100% outdoor. The estimated railway population are presented in Table 11.4.

Table 11.4 Train Population in Operational Phase Scenario (Year 2031)

Population ID	Description	Maximum Population
Population ID	Description	0700 – 2300 hours	2300 - 0700 hours
7	LRT^{[Note 1]}	296	148
11	TME Railway Population	932	666

Note 1: The population of LRT was estimated based on the design capacity of a LRT Vehicle and the existing train frequency, and the population for both the construction phase scenario and the operational phase scenario were assumed to be the same.

Pedestrian Population

11.4.17 Pedestrian flow on pavements and cycling tracks near the LPG Store was assessed by site survey. The pedestrian density was estimated by the following equation:

11.4.18 Roads to be covered in the assessment included the cycling tracks, Hoi Wong Road, Wu Shan Road and Wu Hong Street. The population associated with pedestrians and cyclers was modelled as 100% outdoor. The estimated pedestrian population are presented in Table 11.5.

Table 11.5 Pedestrian Population

Population ID	Description	Maximum Population ^{[Note 1][Note 2]}
8a	Hoi Wong Road	3
8b	Wu Shan Road	4
8c	Wu Hong Street	2
12	Cycling Track - Near LPG Store	2
13	Cycling Track - Near Wu Shan Recreation Playground	1

Note 1: The pedestrian population was estimated based on site survey carried out during daytime of 17-18 September 2021. Night-time population was assumed to be same as the daytime population as a conservative approach.

Note 2: The pedestrian population along Hoi Wong Road, Wu Shan Road and Wu Hong Street was combined with the road population in the model.

Time Modes

11.4.19 Four representative time modes were identified to address the variation in the level of activities that could lead to a release and the variation in population in the assessment area with time. Table 11.6 shows the time periods used in this assessment.

Table 11.6 Definitions of Time Modes

Time Period	Definition	Proportion of Time
Weekday Day	Mon – Fri, 7am – 7pm	35.71%
Weekend Day	Sat – Sun, 7pm – 7am	14.29%
Night (1900-2300)	All days, 7pm – 11pm	16.67%
Night (2300-0700)	All days, 11pm – 7am	33.33%

Meteorological Data

11.4.20 Meteorological data is required for consequence modelling and risk calculation. Consequence modelling (dispersion modelling) requires wind speed and stability class to determine the degree of turbulent mixing potential whereas risk calculation requires wind-rose frequencies for each combination of wind speed and stability class.

11.4.21 Meteorological data, including wind speed, stability class, weather class and wind direction, recorded at Tuen Mun Weather Station for 2020 were obtained from Hong Kong Observatory (HKO). This set of data represents the weather conditions for the whole year in 2020 and has already taken into account of seasonal variations, and therefore was considered applicable for the assessment. Table 11.7 shows the wind speed-stability frequencies.

Table 11.7 Stability Category-Wind Speed Frequencies at HKO Tuen Mun Weather Station

Daytime
Wind Speed (m/s)	A	B	C	D	E	F	Total (%)
0.0-1.9	12.39	5.55	0.00	7.65	0.00	9.26	34.85
2.0-3.9	7.35	18.77	8.15	11.12	4.26	0.51	50.16
4.0-5.9	0.00	5.80	3.55	4.38	0.30	0.00	14.03
6.0-7.9	0.00	0.00	0.16	0.78	0.00	0.00	0.94
Over 8.0	0.00	0.00	0.00	0.02	0.00	0.00	0.02
All (%)	19.74	30.12	11.86	23.95	4.56	9.77	100.00
Night-time
Wind Speed (m/s)	A	B	C	D	E	F	Total (%)
0.0-1.9	0.00	0.00	0.00	1.45	0.00	51.11	52.56
2.0-3.9	0.00	0.00	0.00	13.01	22.05	3.16	38.22
4.0-5.9	0.00	0.00	0.00	7.70	0.99	0.00	8.69
6.0-7.9	0.00	0.00	0.00	0.44	0.00	0.00	0.44
Over 8.0	0.00	0.00	0.00	0.09	0.00	0.00	0.09
All (%)	0.00	0.00	0.00	22.69	23.04	54.27	100.00

11.4.22 According to Table 11.7, 6 combinations (3B, 1D, 3D, 6D, 2E and 1F) and 5 combinations (1D, 4D, 6D, 2E and 1F) of wind speed and stability class were chosen for daytime and night-time meteorological conditions, respectively. These combinations were considered adequate to reflect the full range of observed variations in these quantities. It would not be necessary and efficient to consider every combination observed. The principle is to group these combinations into representative weather classes that together cover all conditions observed.

11.4.23 Once the weather classes have been selected, frequencies for each wind direction for each weather class can then be determined. The frequency distributions for the daytime and night-time meteorological conditions are summarised in Table 11.8.

Table 11.8 Weather Class-Wind Direction Frequencies at HKO Tuen Mun Weather Station

Daytime
Direction	3B		1D		3D		6D		2E		1F		Total (%)
0 – 30	6.90		1.12		2.84		0.05		1.24		1.78		13.93
30 – 60	4.59		0.84		7.16		0.36		2.08		0.84		15.87
60 – 90	0.05		0.10		0.13		0.00		0.08		0.25		0.61
90 – 120	0.51		0.05		0.43		0.03		0.15		0.18		1.35
120 – 150	2.82		0.20		3.07		0.53		0.38		0.36		7.36
150 – 180	16.01		0.46		11.24		1.40		1.80		0.66		31.57
180 – 210	6.78		0.30		3.07		0.03		0.71		0.28		11.17
210 – 240	5.94		0.30		1.47		0.00		0.56		0.18		8.45
240 – 270	2.49		0.43		0.41		0.00		0.23		0.20		3.76
270 – 300	0.56		0.18		0.13		0.00		0.18		0.43		1.48
300 – 330	1.42		0.15		0.25		0.00		0.05		0.20		2.07
330 – 360	1.52		0.23		0.25		0.00		0.15		0.23		2.38
All (%)	49.59		4.36		30.45		2.40		7.61		5.59		100.00
Night-time
Direction		1D		4D		6D		2E		1F		Total (%)
0 – 30		0.42		0.95		0.03		3.65		7.88		12.93
30 – 60		0.19		8.33		0.40		10.09		6.29		25.30
60 – 90		0.00		0.03		0.00		0.19		0.61		0.83
90 – 120		0.00		0.26		0.16		1.40		1.56		3.38
120 – 150		0.11		2.51		0.37		4.50		2.27		9.76
150 – 180		0.05		9.87		0.19		14.75		3.17		28.03
180 – 210		0.08		0.71		0.00		4.39		1.96		7.14
210 – 240		0.00		0.29		0.00		2.46		1.30		4.05
240 – 270		0.00		0.08		0.00		0.79		1.24		2.11
270 – 300		0.00		0.03		0.00		0.48		2.17		2.68
300 – 330		0.00		0.11		0.00		0.19		1.51		1.81
330 – 360		0.05		0.29		0.00		0.42		1.22		1.98
All (%)		0.90		23.46		1.15		43.31		31.18		100.00

11.5 Hazard Identification

11.5.1 A hazard is described as the property of a material or activity with the potential to do harm. A release of flammable gas such as LPG has the potential to cause fire or explosion if ignited. Without ignition, the gas vapours will disperse harmlessly. Under normal conditions, the LPG at the existing LPG Store will be stored and handled under contained and controlled manners. For LPG to pose a hazard to the people in the surrounding area, a release must occur as a result of a failure of that containment or as a result of faulty transfer procedures.

11.5.2 The failure rates adopted for this assessment were quoted from the paper on “Quantitative Risk Assessment Methodology for LPG Installations (Reeves, Minah and Chow, 1997)” ^[5]. In addition, reference for certain frequencies were drawn from approved EIA Reports ^[6][7] and QRA studies ^[8][9] where necessary and appropriate. In addition, possible initiating events were identified.

Behaviour of LPG Releases

11.5.3 LPG is a mixture of butane and propane. The gas is twice as heavy as air. For a release of LPG, the nature of the combustion will depend on the timing of ignition and the size of the release.

11.5.4 A release of several tonnes of LPG, if ignited immediately, will produce a fireball. Initially, the gas concentration in the mixture will be above the Upper Flammability Limit (UFL). As burning occurs around the edges of the release, this will entrain more air into the mixture and more combustion will take place. The process accelerates until the mixture rising above the ground as a ball of fire. A fireball may also result from a boiling liquid expanding vapour explosion (BLEVE). This results from the bursting of a vessel (owing to a high internal pressure and a weakening of the vessel material, as a result of a fire for example). The vessel contents rapidly vaporise and are ignited.

11.5.5 If not ignited immediately, the gas will disperse and dilute. If ignition occurs when the gas concentration is between the lower Flammability Limit (LFL) and the Upper Flammability Limit (UFL), a flame front will propagate to produce a flash fire.

11.5.6 For small releases, immediate ignition will produce a long vigorous jet flame from the point of release. As for large releases, delayed ignition will generally produce a flash fire.

11.5.7 For all sizes of release, the LPG will disperse harmlessly if there is no source of ignition.

Hazard Analysis

Spontaneous Failures

Failure of Storage Vessel

11.5.8 Failure of a vessel can result from: (i) a cold catastrophic failure leading to instantaneous release of the full inventory; and (ii) a partial failure leading to continuous release of the full inventory via a 25mm hole. The causes of failure are summarised as follows:

a) Spontaneous failure due to corrosion, fatigue, etc.

b) Overfilling

c) Earthquake

Failure of Road Tanker

11.5.9 The causes of failure of a road tanker are similar to those of a storage vessel. Furthermore, road tankers are vulnerable to collision with other road vehicles during delivery.

Guillotine Failure of Liquid Filling Line to Storage Vessel

11.5.10 Failure of the liquid line is possible as a result of corrosion or fatigue, vehicle impact and external events. Only guillotine failure of the LPG pipework was considered in this assessment as partial failure of the pipeworks is an insignificant contributor to the overall risk levels. The failure would result in LPG leaking from the full bore of the pipe. Moreover, part of the pipeworks is installed aboveground. Failure of the aboveground portion of the liquid filling line can result from vehicle impact while failure of the underground portion of the liquid filling line can result from earthquakes.

Guillotine Failure of Liquid Supply Line to Vaporisers

11.5.11 The liquid supply line connects the underground storage vessel and vaporisers. Failure of the liquid line is possible as a result of corrosion or fatigue, vehicle impact and external events. Only guillotine failure of the LPG pipework was considered in this assessment as partial failure of pipeworks is an insignificant contributor to the overall risk levels. The failure would result in LPG leaking from the full bore of the pipe. Since the pipework is protected by fencing, vehicle impact is not considered credible. However, failure of the liquid line can result from earthquakes.

Failure of Vaporisers

11.5.12 Three units of vaporisers are installed at the LPG Store. Apart from spontaneous failure and loading failure, failure of the vaporisers can result from earthquakes and aircraft crashes.

Guillotine Failure of Liquid Line from Tanker Pipe to Loading Hose

11.5.13 The cause of failure of this line is similar to that of the liquid filling line to the storage vessel, namely mainly corrosion or fatigue. Moreover, the failure can be due to vehicle impact and other external events.

Failure of Flexible Hose

11.5.14 The loading hose could fail from the following causes:

a) Fatigue

b) Hose misconnection

c) Hose disconnection during loading or unloading process

d) Operator / driver error

Loading / Unloading Failures

11.5.15 When LPG releases occur as a direct result of the road tanker unloading operation, the failure events can be regarded as loading failures.

11.5.16 The failure events that could be categorised as loading failures are as follows:

a) Hose misconnection and disconnection error

b) Tanker drive-away error

c) Road tanker collision

d) Vehicle impact with road tanker during unloading

e) Storage vessel overfilling

f) Over-pressurisation of pipework

Hose Misconnection and Disconnection Error

11.5.17 A significant release of LPG during its transfer from road tanker to storage vessel could occur as a result of the failure of the transfer hoses and coupling, human error, or vehicle impact.

Tanker Drive-away Error

11.5.18 This error could result from: (i) repositioning of the road tanker during delivery; and/or (ii) the driver driving the road tanker away before the delivery is completed.

Road Tanker Collision

11.5.19 Road tanker collision refers to an event in which an LPG road tanker strikes the facilities of the LPG Store and causes damages to these facilities. Provision of dedicated road tanker parking area and unloading area, implementation of speed control and well-adopted training system are safety measures commonly adopted to avoid serious collision incidents. Moreover, kerbs with appropriate markings are provided between the unloading area and the LPG facilities to prevent direct impact. Road tanker collision leading to failure of the road tanker itself is considered to be insignificant. Underground facilities such as LPG storage vessel and pipework would not be affected by this event since they are installed underground. Collision of an LPG road tanker with other road tankers is not possible as the LPG Store can only accommodate one LPG road tanker on site at a time.

Vehicle Impact with Road Tanker during Unloading

11.5.20 Dedicated road tanker parking area and unloading area is provided, impact on the road tanker by other vehicles during unloading is thus considered unlikely.

Storage Vessel Overfilling

11.5.21 Failure of the LPG storage vessel could occur as a result of overfilling of LPG from the road tanker to the vessel.

Over-pressurisation of Pipework

11.5.22 Over-pressurisation could be caused by continuing unloading operation when a storage vessel is overfilled or the isolation valves at the receiving storage vessel are closed.

External Events

11.5.23 A LPG release event could occur as a result of external events and the consequences could be catastrophic. The related external events are listed as follows:

a) Earthquake

b) Aircraft crash

c) Landslide

d) Severe environmental event such as typhoon or tsunami

e) Subsidence

f) Lightning

g) High wind loading; and

h) External fire

11.5.24 According to Reeves et al. (1997) ^[5], an earthquake of Modified Mercali Intensity (MMI) VIII could provide enough intensity to result in damage to the storage vessel or pipework, and therefore earthquake was considered in this assessment.

11.5.25 Aircrafts crashing into the LPG Store during take-off and landing as well as arrival / departure flight paths were taken into account in this assessment. The method given in HSE (1997) ^[10] for the calculation of aircraft crash frequency was adopted.

11.5.26 The LPG Store is not situated near any hillside and LPG release events could not result from landslides, and therefore this external event was not further considered in this assessment.

11.5.27 According to BDEIA ^[6], loss of LPG content owing to severe environmental event such as typhoon or tsunami (i.e. a tidal wave following an earthquake) was considered to be insignificant as the installation of LPG vessels is situated underground and the LPG Store is located within the Tuen Mun Typhoon Shelter. Subsidence is usually slow in movement and such movement can be observed and remedial action can be taken in time. Based on the above, the probabilities of severe environmental events and subsidence are very small or negligible so these external events were not further considered in this assessment.

11.5.28 The LPG Store is located within the Tuen Mun Typhoon Shelter and its southern direction facing the seashore is surrounded by tall buildings that shield the LPG Store from damages by high winds. Frequency of high winds damaging the LPG Store and lightning strike on the LPG Store was assumed to be less than the credible frequency of 1×10^-9 per year. An LPG release due to high wind and lightning was therefore not further considered in this assessment.

11.5.29 The proposed works area adjoining the LPG Store during construction phase of the Project could have potential risk of fire affecting operation of the LPG Store. It was confirmed that the works area would be used for the unloading and inspection of precast segments, and the contractor’s site office, and hot works will be banned within this works area. In addition, most of the LPG facilities in the LPG Store are either housed in concrete structure (i.e. the vaporisers) or installed underground (i.e. the LPG storage vessels). It is considered that the above-mentioned measures along with the fire fighting system installed in the LPG Store could isolate fire at the works area until the arrival of fire services. Therefore, this external event is not considered further in this assessment.

Safety Features

11.5.30 Safety features installed in the facilities of the LPG Store can act in different combination to mitigate LPG releases. Such features are highlighted in the following sections.

Non-return Valve

11.5.31 Non-return valve on the liquid filling line can isolate release immediately. If it functions properly, there will be no significant consequence.

Excess Flow Valve

11.5.32 Excess flow valve installed at the LPG road tanker and the storage vessel is expected to mitigate a release from guillotine failure of the pipework or the flexible filling hose.

Emergency Shutdown System

11.5.33 Such system is installed on both the road tankers and the vessels. For a release from a road tanker, the emergency isolation system and the engine emergency stop system can be activated to isolate the release caused by equipment failure and human error. For a release from the vessel, the emergency isolation system can be triggered to prevent a release on the filling line or downstream of the hose connection.

Breakaway Coupling

11.5.34 One problem identified with road tankers is the possibility of them being driven away whilst the hose is still connected, thereby causing damage to the facilities of the LPG Store and resulting in the release of LPG. The breakaway coupling is installed to prevent undue spillage of LPG owing to the movement of road tankers.

Manual Isolation System

11.5.35 A manual valve is installed for the operators / drivers to shut off the delivery connection manually in case of failure.

Double-check Filler Valve

11.5.36 Double-check filler valve is provided at the hose connection point on the liquid filling line to prevent release to be fed back from the vessel. The design of this valve is essentially 2 non-return valves in series.

Relief Valve

11.5.37 Relief valve is employed to ensure that the vessel is not subject to an excessive internal pressure that may cause a failure as a result of overfilling. It also offers protection against excessive pressure build-up within the vessel in case of fire.

Human Error

11.5.38 When a failure of equipment or loading process occurs, it is possible for the staff to rectify the problem before a hazard event occurs. Human error of this nature was regarded as a failure case.

Fire Protection / Fighting System

Chartek Coating

11.5.39 Chartek coating is a safety feature of all road tankers. The coating has been reported to provide protection for at least 30 minutes in the case of a jet fire. The coating could prevent a hot spot from developing in a jet fire attack on the road tanker, which can cause thermal weakening of the road tanker wall leading to BLEVE.

Water Spray System

11.5.40 There is a water spray system installed on site. This fire services installation is provisioned for fire associated with a road tanker so that the fire can be extinguished or surface of the road tanker can be cooled down to avoid BLEVE, and it was accounted in the event tree analysis.

Fire Services

11.5.41 The nearest Castle Peak Bay fire station is located at 8 Tuen Yee Street, Tuen Mun and the travelling distance from the fire station to the LPG Store is around 1km. The fire services will be available within a few minutes in case of a fire. The extinction of fire by fire fighters prevents BLEVE from occurring.

Escalation

11.5.42 BLEVE of an LPG road tanker can happen if the road tanker is impinged by jet fire from the failure of aboveground LPG facilities listed below:

a) Cold partial failure of road tanker

b) Guillotine failure of liquid filling line to vessel

c) Guillotine failure of liquid supply line to vaporiser

d) Failure of flexible hose during loading to storage vessel

e) Failure of liquid line from tanker to loading hose

f) Vaporiser failure

Summary

11.5.43 The possible hazard events for the day-to-day operations of the LPG Store were identified and reviewed in the previous sections. Only those possible failure cases considered to have the potential to cause off-site fatality are summarised in Table 11.9.

Table 11.9 Identified Failure Cases for the LPG Store

Failure Types	Failure Cases
Spontaneous Failure of Pressurised LPG Equipment	· Storage Vessel Failure · Road Tanker Failure · Pipework Failure · Hose Failure · Vaporiser Failure
External Event	· Earthquake MMI VIII · Aircraft Crash
Delivery Failure	· Hose Misconnection Error · Hose Disconnection Error · Tanker Drive-away Error · Road Tanker Collision during Unloading · Storage Vessel Overfilling · Over-pressurisation of pipework
Safety System Failure	· Pressure Relief Valve Failure · Non-return Valve Failure · Excess Flow Valve Failure · Emergency Shutdown System Failure · Double-check Filler Valve Failure · Breakaway Coupling Failure · Human Error · Manual Isolation Valve Failure
Fire Fighting System Failure	· Fire Services Failure · Chartek Coating Failure · Water Spray System Failure

11.6 Hazard Occurrence

Introduction

11.6.1 Subsequent to the hazard identification in the previous section, the next step is to estimate the likelihoods of various release scenarios. There are combinations of hazard initiating events, as identified in previous section, which would lead to release scenarios.

11.6.2 Fault Tree Analysis (FTA) permits frequency of the hazardous incident (“Significant Failure Events”) to be estimated from a logical model of the failure mechanism of a system. The model is based on the combinations of failures of more basic components, safety systems and human errors.

11.6.3 FTA is the use of a combination of simple logic gates, “AND” and “OR” gates, to synthesise a failure model of the hazardous installation. The “Significant Failure Events” frequency is calculated from failure data of more simple events.

11.6.4 A basic assumption in FTA is that all failures in a system are binary in nature, a component or operator either performs successfully or fails completely. In addition, the system is assumed to be functioning if all sub-components are operating properly.

11.6.5 The stepwise procedure for undertaking FTA is presented below:

· Hazard identification and selection of the “Significant Failure Events”;

· Construction of fault trees; and

· Quantitative evaluation of the fault trees.

Frequency Estimation

Spontaneous Failures

Storage Vessel Failure

11.6.6 A release of LPG could occur as a result of catastrophic failure or partial failure of the storage vessel and such a failure would lead to either a loss of entire contents of the vessel or a continuous release of LPG to atmosphere.

11.6.7 A generic failure rate of 1.8×10^-7 per vessel year ^[5] was adopted for cold catastrophic failure and a generic failure rate of 5.0×10^-6 per vessel year ^[5] was applied for cold partial failure.

Road Tanker Failure

11.6.8 The definitions of catastrophic and partial failures of the road tanker are similar to those of the storage vessel. It is generally considered that the catastrophic failure rate for LPG road tankers could be higher than that for a fixed storage vessel because of a) stresses experienced by the road tanker owing to vibration during transportation; and b) cyclic loading associated with filling/unloading the road tanker.

11.6.9 A rate of 2.0×10^-6 per tanker year ^[5] was adopted for catastrophic tanker failure and a failure rate of 5.0×10^-6 per tanker year ^[5] was applied for partial failure of road tanker.

Pipework Failure

11.6.10 Reeves et al. (1997) ^[5] indicated that releases from pipework partial failures were insignificant contributors to the overall risk levels. Based on this, this assessment only considered guillotine failure of LPG pipework as the contribution of a release from the partial failure of pipework to the overall risk levels was insignificant. A generic rate of 1.0×10^-6 per metre per year for guillotine failure of the pipework was adopted.

Vaporiser Failure

11.6.11 The effect of partial failure of the vaporiser was ignored. A generic rate of 1.0×10^-6 per metre per year ^[5] for guillotine failure of the vaporiser coil was adopted.

Hose Failure

11.6.12 The effect of partial failure of the hose was ignored. A generic guillotine failure rate of flexible hose was taken to be 1.8×10^-7 per transfer ^[5] or 9.0×10^-8 per hour ^[5].

Loading / Unloading Failures

Hose Misconnection Error

11.6.13 A significant release of LPG during its transfer from the road tanker to the storage vessel could occur as a result of the failure of the transfer hoses and coupling, human error, or vehicle impact. The likelihood of such an event was taken as 3×10^-5 per operation ^[5].

Hose Disconnection Error

11.6.14 A rate of 2.0×10^-6 per operation ^[5] was adopted for this failure case.

Tanker Drive-away Error

11.6.15 Tanker drive-away error refers to an event in which the tanker moves away with the hose still connected. It could result from the tanker driver inadvertently driving the vehicle away before delivery is completed. It was considered that drive-away was unlikely. Even if such errors do occur, it is highly likely that the failure can be immediately rectified since the delivery process would not go unattended. A failure rate of 4.0×10^-6 per operation ^[5] was adopted.

Tanker Collision during Unloading

11.6.16 A release of LPG cloud occurs as a result of an incident involving an LPG tanker and LPG equipment during delivery. It was assumed that the failure rate of tanker impact during unloading was 1.5×10^-4 per delivery ^[5].

Overfilling of Storage Vessel

11.6.17 The practice on-site in unloading LPG to the storage vessel is that the vessel will be only filled to 85% of the maximum capacity. It was considered that the probability of the driver overfilling a storage vessel is low. A rate of 2.0×10^-2 per operation ^[5] was adopted for this failure case.

Over-pressurization of Pipework

11.6.18 This event has been taken into account by the pipework and hose failure data discussed in Sections 11.6.10 and 11.6.12. It was not considered separately in the assessment.

External Events

Earthquake MMI VIII

11.6.19 The probability of 1.0×10^-5 per year was adopted for the occurrence of an MMI VIII earthquake. The rate of failure of pipework and partial failure of underground vessel owing to earthquakes was assumed to be 0.01 ^[6], whereas the probability failure for road tanker was considered to be zero.

Aircraft Crash

11.6.20 The distance between the nearest arrival flight path for the Hong Kong International Airport (HKIA) and the LPG Store is approximately 4.7km. The distance between the LPG Store and HKIA is about 7.5km, which is within the criteria of 5 miles (8 km) for the consideration of airfield accident. At such distances, the LPG Store comes into the flight paths of the critical takeoff and landing phases, and therefore the background crash rate, airway crash rate and airfield crash rate were taken into account. The frequency of aircraft crash was estimated using the methodology of the HSE (1997) ^[10]. The detailed calculation of aircraft crash frequency is shown in Appendix 11.2.

11.6.21 The total frequency of aircraft crash impacting the LPG Store at year 2031 was estimated to be 1.61×10^-9 per year.

Safety System Failure

11.6.22 If the safety system operates as designed, then releases will not present an off-site hazard. There is, however, a potential for failure of the safety system. A typical safety system involves pressure relief valve, non-return valve, excess flow valve, emergency shutdown system, breakaway coupling and double-check filler valve.

Pressure Relief Valve Failure

11.6.23 Pressure relief valve avoids the LPG pipework or underground storage vessel from getting overpressure. A generic failure rate of 1×10^-4 for the pressure relief valve per demand ^[5] was adopted.

Pump Overpressure Protection System

11.6.24 Such system is installed on LPG road tankers to control the maximum outlet pressure of the pump. In addition to internal pump overpressure by-pass, the pump or adjacent pipework is fitted with a separate by-pass valve set at a lower differential pressure to automatically carry any excess liquid back to the road tanker vessel when the delivery valve is closed.

11.6.25 A generic failure rate of 1×10^-4 for the pump overpressure protection system per demand ^[5] was adopted.

Non-return Valve Failure

11.6.26 Non-return valve is intended to avoid the back flow of LPG. A generic failure rate of 0.013 per demand ^[5] was adopted.

Excess Flow Valve Failure

11.6.27 The excess flow valve installed at road tanker and storage vessel is expected to be functional when guillotine failure of pipework or flexible hose occurs. A generic failure rate of 0.13 per demand ^[5] was adopted for the line to vaporiser.

Emergency Shutdown System Failure

11.6.28 A generic failure rate of 1.0×10^-4 per demand ^[5] was assumed.

Breakaway Coupling Failure

11.6.29 A generic failure rate of 0.013 per demand ^[5] was adopted for road tanker.

Double-check Filler Valve Failure

11.6.30 A double-check filler valve prevents the LPG release to be fed back from the storage vessel. The design has two non-return valves in series. A generic failure rate of 2.6×10^-3 per demand ^[5] for common mode failure was adopted.

Manual Isolation Valve

11.6.31 A manual valve is installed for the operators / drivers to shut off the delivery connection manually in case of failure.

11.6.32 A generic failure rate of 0.5 per demand ^[5] was adopted.

Human error

11.6.33 A probability of 1.5×10^-3 per demand was assumed to account for the human error in which the operators fail to rectify the problem before any hazard event occurs.

Fire Fighting System Failure

Water Spray System Failure

11.6.34 A generic failure rate of 1.5×10^-2 per demand ^[5] was adopted to account for the common problems of the water spray system: block nozzles and malfunction of the fire detectors.

Failure of Fire Services

11.6.35 It was assumed that the Fire Services would always be available, and therefore zero probability was applied for the failure of “fire services arrive late”. A generic failure rate of 0.5 per demand ^[5] was assumed for the fire services to be ineffective against a fire attack.

Chartek Coating Failure

11.6.36 A generic failure rate of 0.1 per demand ^[5] was applied for the Chartek coating failing to prevent a hot spot from developing on the road tanker in a jet fire attack owing to poor maintenance.

11.6.37 The identified failure cases and their corresponding failure rates adopted are presented in Table 11.10.

Table 11.10 Summary of Identified Failure Cases and Their Associated Failure Rates

Failure Cases	Failure Rates	Reference Source
*Spontaneous Failure of Pressurised LPG Equipment*
Catastrophic Failure of Storage Vessel	1.8×10^–7 per vessel year	Reference [5]
Partial Failure of Storage Vessel	5.0×10^-6 per vessel year	Reference [5]
Catastrophic Failure of Road Tanker	2.0×10^-6 per tanker year	Reference [5]
Partial Failure of Road Tanker	5.0×10^-6 per tanker year	Reference [5]
Guillotine Failure of Pipework	1.0×10^-6 per meter per year	Reference [5]
Vaporiser Failure	1.0×10^-6 per meter per year	Reference [5]
Hose Failure	1.8×10^-7 per transfer or 9.0×10^-8 per hour	Reference [5]
*External Event*
Earthquake MMI VIII	1.0×10^-5 per year	Reference [5]
Aircraft Crash	1.61×10^-9 per year	Appendix 11.2
*LPG Loading Failure*
Hose Misconnection Failure	3.0×10^-5 per operation	Reference [5]
Hose Disconnection Failure	2.0×10^-6 per operation	Reference [5]
Tanker Drive-away Error	4.0×10^-6 per operation	Reference [5]
Road Tanker Collision	1.5×10^-4 per operation	Reference [5]
Storage Vessel Overfilling	2.0×10^-2 per operation	Reference [5]
*Safety Features Failure*
Pressure Relief Valve Failure	1.0×10^-4 per demand	Reference [5] based on ESD system
Failure of Pump Over-pressurization Protection	1.0×10^-4 per demand	Based that for failure of pressure relief valve
Non-return Valve Failure	0.013 per demand	Reference [5]
Excess Flow Valve Failure	1.00 per demand for liquid filling line and flexible hose 0.13 per demand for line to vaporiser	Reference [5]
Manual Isolation Valve Failure	0.5 per demand	Reference [5]
Emergency Shutdown System Failure	1.0×10^-4 per demand	Reference [5]
Breakaway Coupling Failure	0.013 per demand for tanker	Reference [5] Conservative estimate, based on breakaway coupling for road tanker
Double-check Filler Valve Failure	2.6×10^-3 per demand	Reference [5]
Operator fails to Rectify Problem	1.5×10^-3 per demand	Reference [11]
*Fire Protection / Fighting System Failure*
Water Spray System Failure	1.5×10^-2 per demand	Reference [5]
Failure of Fire Services	0.5 per demand	Reference [5]
Chartek Coating Failure	0.1	Reference [5]
*Failure Probability*
Catastrophic failure of vessel provided over-pressurization	0.01	Reference [11]
Partial failure of vessel provided over-pressurization	0.1	Reference [11]; 10 times of catastrophic failure
Probability of catastrophic / guillotine failure due to aircraft crash ^{[Note 1]}	1	Assume 100% failure leading to rupture / guillotine failure
Probability of partial failure due to aircraft crash ^{[Note 1]}	0	Assume 100% failure leading to rupture / guillotine failure
Probability of equipment failure due to earthquake	0.01	Reference [6]
Probability of catastrophic / partial failure in earthquake	0.5 / 0.5	It is more likely for an earthquake to lead to the failure of the pipeline connection rather than the failure of the storage vessel, for which damages would be prevented by the buffering effect of the washed sand. Pipeline failure has already been accounted for in other hazardous events, and therefore a 50:50 split was conservatively adopted for vessel failure events.

Note 1: The probability of road tanker rupture and road tanker partial failure due to aircraft crash are considered as 1 and 0 respectively, which assumes only catastrophic failure of road tanker will be resulted in the event of aircraft crash. The probability for storage vessel failure due to aircraft crash will be significantly less as compared to other equipment since storage vessels are located underground. Hence, 0.01 and 0.09 were adopted for catastrophic and partial failure of storage vessels respectively.

Escalation

11.6.38 Escalation refers to the situation in which a relatively insignificant accident causing an event with much more significance to occur. This was addressed in this assessment with the event tree analysis in Appendix 11.4.

Frequency of Occurrence

Fault Tree Analysis

11.6.39 Fault tree analysis was used to provide models for the calculation of failure rates or the probabilities of the hazardous scenarios described in Table 11.11. The fault tree diagrams for this assessment are provided in Appendix 11.3.

Event Tree Analysis

11.6.40 The event trees evaluate the hazard event outcomes for the LPG events assessed are shown in Appendix 11.4.

11.6.41 Potential hazardous event outcomes following an LPG release can be BELVE, fireball, jet fire, vapour cloud explosion (VCE) and flash fire.

11.6.42 In this assessment, it was considered that no significant areas of confinement/congestion would exist for generating the turbulence required for a vapour cloud explosion upon ignition of a flammable gas cloud, and therefore the probability of occurrence of a VCE was assigned a value of 0 for all LPG release events.

11.6.43 The frequencies of the hazardous outcomes assessed are summarised in Table 11.11.

Table 11.11 Summary of Identified Failure Cases and Their Associated Failure Rates

Ref.	Event Description	Outcome Event	Event Frequency per year	Outcome Probability	Probability of Failure to isolate	Total Outcome Frequency /year
F1	Cold Catastrophic Failure of Storage Vessels	FBL	5.90E-07	9.00E-01	1.00E+00	5.31E-07
F1	Cold Catastrophic Failure of Storage Vessels	VCE	5.90E-07	0.00E+00	1.00E+00	0.00E+00
F1	Cold Catastrophic Failure of Storage Vessels	FFR	5.90E-07	1.00E-01	1.00E+00	5.90E-08
F2	Cold Partial Failure of Storage Vessels	JFI	1.505E-05	5.00E-02	1.00E+00	7.53E-07
F2	Cold Partial Failure of Storage Vessels	BLEVE	1.51E-05	0.00E+00	1.00E+00	0.00E+00
F2	Cold Partial Failure of Storage Vessels	VCE	1.505E-05	0.00E+00	1.00E+00	0.00E+00
F2	Cold Partial Failure of Storage Vessels	FFR	1.51E-05	9.50E-01	1.00E+00	1.43E-05
F3	Cold Catastrophic Failure of Road Tanker	FBL	5.56E-08	9.00E-01	1.00E+00	5.00E-08
F3	Cold Catastrophic Failure of Road Tanker	VCE	5.56E-08	0.00E+00	1.00E+00	0.00E+00
F3	Cold Catastrophic Failure of Road Tanker	FFR	5.56E-08	1.00E-01	1.00E+00	5.56E-09
F4	Cold Partial Failure of Road Tanker	JFI	1.39E-07	5.00E-02	1.00E+00	6.94E-09
F4	Cold Partial Failure of Road Tanker	BLEVE	1.39E-07	0.00E+00	1.00E+00	0.00E+00
F4	Cold Partial Failure of Road Tanker	VCE	1.39E-07	0.00E+00	1.00E+00	0.00E+00
F4	Cold Partial Failure of Road Tanker	FFR	1.39E-07	1.90E-01	1.00E+00	2.64E-08
F5.1	Guillotine Failure of Liquid filling Line to Vessel (fed from Tanker)	JFI	1.85E-07	5.00E-02	4.53E-03	4.19E-11
F5.1	Guillotine Failure of Liquid filling Line to Vessel (fed from Tanker)	BLEVE	1.85E-07	6.18E-05	4.53E-03	5.18E-14
F5.1	Guillotine Failure of Liquid filling Line to Vessel (fed from Tanker)	VCE	1.85E-07	0.00E+00	4.53E-03	0.00E+00
F5.1	Guillotine Failure of Liquid filling Line to Vessel (fed from Tanker)	FFR	1.85E-07	1.90E-01	4.53E-03	1.59E-10
F5.2	Guillotine Failure of Liquid filling Line to Vessel (fed from Vessel)	JFI	1.85E-07	5.00E-02	1.27E-02	1.18E-10
F5.2	Guillotine Failure of Liquid filling Line to Vessel (fed from Vessel)	BLEVE	1.85E-07	6.18E-05	1.27E-02	1.45E-13
F5.2	Guillotine Failure of Liquid filling Line to Vessel (fed from Vessel)	VCE	1.85E-07	0.00E+00	1.27E-02	0.00E+00
F5.2	Guillotine Failure of Liquid filling Line to Vessel (fed from Vessel)	FFR	1.85E-07	1.90E-01	1.27E-02	4.47E-10
F6	Guillotine Failure of Liquid Supply Line to Vaporiser	JFI	2.01E-05	5.00E-02	1.30E-01	1.31E-07
F6	Guillotine Failure of Liquid Supply Line to Vaporiser	BLEVE	2.01E-05	6.18E-05	1.30E-01	1.62E-10
F6	Guillotine Failure of Liquid Supply Line to Vaporiser	VCE	2.01E-05	0.00E+00	1.30E-01	0.00E+00
F6	Guillotine Failure of Liquid Supply Line to Vaporiser	FFR	2.01E-05	1.90E-01	1.30E-01	4.97E-07
F7	Guillotine Failure of Liquid Filling Line to Flexible Hose	JFI	2.55E-07	5.00E-02	7.55E-03	9.64E-11
F7	Guillotine Failure of Liquid Filling Line to Flexible Hose	BLEVE	2.55E-07	6.18E-05	7.55E-03	1.19E-13
F7	Guillotine Failure of Liquid Filling Line to Flexible Hose	VCE	2.55E-07	0.00E+00	7.55E-03	0.00E+00
F7	Guillotine Failure of Liquid Filling Line to Flexible Hose	FFR	2.55E-07	1.90E-01	7.55E-03	3.66E-10
F8	Vaporiser Failure	JFI	3.51E-05	5.00E-02	1.30E-01	2.28E-07
F8	Vaporiser Failure	BLEVE	3.51E-05	6.18E-07	1.30E-01	2.82E-12
F8	Vaporiser Failure	VCE	3.51E-05	0.00E+00	1.30E-01	0.00E+00
F8	Vaporiser Failure	FFR	3.51E-05	1.90E-01	1.30E-01	8.67E-07
F9.1	Guillotine Failure of Flexible Hose (fed from Tanker)	JFI	1.02E-04	5.00E-02	7.55E-03	3.86E-08
F9.1	Guillotine Failure of Flexible Hose (fed from Tanker)	BLEVE	1.02E-04	1.240E-04	7.55E-03	9.54E-11
F9.1	Guillotine Failure of Flexible Hose (fed from Tanker)	VCE	1.02E-04	0.00E+00	7.55E-03	0.00E+00
F9.1	Guillotine Failure of Flexible Hose (fed from Tanker)	FFR	1.02E-04	1.90E-01	7.55E-03	1.47E-07
F9.2	Guillotine Failure of Flexible Hose (fed from Vessel)	JFI	1.02E-04	5.00E-02	1.30E-03	6.64E-09
F9.2	Guillotine Failure of Flexible Hose (fed from Vessel)	BLEVE	1.02E-04	1.24E-04	1.30E-03	1.64E-11
F9.2	Guillotine Failure of Flexible Hose (fed from Vessel)	VCE	1.02E-04	0.00E+00	1.30E-03	0.00E+00
F9.2	Guillotine Failure of Flexible Hose (fed from Vessel)	FFR	1.02E-04	1.90E-01	1.30E-03	2.52E-08

Note 1: FBL – Fireball; BLEVE – Boiling Liquid Expanding Vapour Explosion; VCE – Vapour Cloud Explosion; JFI – Jet fire; FFR – Flash Fire

11.7 Consequence and Impact Analysis

Introduction

11.7.1 Consequence and impact analysis was conducted to provide a quantitative estimate of the likelihood and number of deaths associated with the range of possible outcomes (i.e. fireball, jet fire, flash fire) which resulted from failure cases identified in the previous sections. PhastRisk 6.7 was used for consequence modelling.

11.7.2 The underground LPG storage vessels in the LPG Store are situated inside a concrete chamber filled with washed sand. A BLEVE of the LPG tank as a result of a fire underneath the tank or a jet fire was considered impossible, and therefore no BLEVE occurrence was assumed for this scenario.

Modelling Input

11.7.3 Failure events identified in previous sections were considered and evaluated through consequence analysis. Taking into account the safeguard measures, the layout plan of the LPG Store and effect distances of failure events, some failure events would have insignificant off-site impact. Those failure events having potential off-site impact are listed as follows:

a) Rupture of storage vessel

b) Rupture of road tanker

c) Partial failure of storage vessel

d) Partial failure of road tanker

e) Guillotine failure of liquid filling line to flexible hose

f) Guillotine failure of flexible hose

g) Guillotine failure of liquid filling line to storage vessel

h) Guillotine failure of supply line to vaporisers

i) Vaporiser failure

j) BLEVE of road tanker

11.7.4 There are three storage vessels, each with a storage capacity of 10 tonnes, at the LPG Store. The storage vessels were assumed to be filled to a maximum permissible level (85% of the maximum capacity) in this assessment. Replenishment of LPG was assumed to be arranged during daytime only for risk modelling purpose.

Physical Effects Modelling

11.7.5 The following paragraphs give a brief description of the physical effects models that were used to assess the effect zones for the following outcomes:

· Fireball / BLEVE;

· Jet fire; and

· Dense gas dispersion / flash fire and Vapour Cloud Explosion (VCE).

11.7.6 The consequence distances of the failure events presented in Table 11.10 are summarised in Appendix 11.5.

Fireball / BLEVE

11.7.7 A fireball / BLEVE could result from the instantaneous release and ignition of LPG following the catastrophic failure of an LPG pressure vessel. The term ‘fireball’ was applied to a fire event following ignition of a spontaneous catastrophic failure of the LPG vessel while BLEVE refers to the fire event due to fire-induced failure of the LPG vessel.

11.7.8 Cold catastrophic failures are characterised by a rapid propagation of a crack leading to a sudden release of the contents inside the pressure vessel. The contents under pressure will equilibrate with atmospheric conditions causing part of the liquid to flash into vapour. This ‘flash fraction’ can be calculated using a constant enthalpy assumption. For LPG stored under ambient temperature, the flash fraction is usually between 15 and 40%.

11.7.9 For immediate ignition of an instantaneous gas release, a fireball will be formed. Fireball is more likely for immediate ignition of instantaneous release from LPG vessels/ tankers due to cold catastrophic failure although it is possible for late explosion. Instantaneous ignition of a certain mass of fuel (flammable gas/LPG) results in explosion and fire of hemispherical shape. Heat is evolved by radiation. The principal hazard of fireball arises from thermal radiation. Due to its intensity, its effects are not significantly influenced by weather, wind direction or source of ignition. Sizes, height, shape, duration, heat flux and radiation were determined in the consequence analysis.

11.7.10 A fireball may also result from a BLEVE. This results from the bursting of a vessel (due to a high internal pressure and a weakening of the vessel material, due to a fire for example). The vessel contents rapidly vaporise and are ignited.

Jet fire

11.7.11 A jet fire is typically resulted from ignition of gas/ liquid discharging from a pressurised containment. Major concerns regarding jet fire are jet flame and the heat radiation effect generated from the jet flame. Thermal effect of the jet fire on adjacent population was quantified in the consequence model.

Dense gas dispersion / flash fire and Vapour Cloud Explosion (VCE)

11.7.12 A flash fire is the consequence of combustion of gas cloud resulting from delayed ignition. The flammable gas cloud can be ignited at its edge and cause a flash fire of the cloud within the LFL and Upper Flammable Limit (UFL) boundaries. In case of continuous release, fire is flashed back to the release source and leads to jet fire. Major hazards from flash fire are thermal radiation and direct flame contact. Since the flash combustion of a gas cloud normally lasts for a short duration, the thermal radiation effect on people near a flash fire is limited. Humans who are encompassed outdoors by the flash fire will be fatally injured.

11.7.13 A Vapour Cloud Explosion (VCE) can occur when a flammable cloud is ignited in a confined or partially confined situation. The principal factor determining the maximum overpressure produced by an LPG dispersion is the degree of confinement. In this assessment, it was considered that there are no significant areas of confinement / congestion to generate the turbulence required for a VCE upon ignition of a flammable gas cloud. Therefore, VCE was not further considered.

Ignition Source

11.7.14 To calculate the risk from flammable materials, information on ignition sources presented in the assessment area needs to be identified. Such data was included in the risk model for each type of ignition source (i.e. point sources, line sources and area sources). The risk calculation program (MPACT) is a module in PhastRisk. MPACT calculates the impact of the release of a toxic or flammable chemical on the population. It takes the results of the consequence calculations of the toxic and flammable effects, together with additional data on wind direction, ignition sources, event location and frequency and superimposes them on the population to calculate the fatality risk in the surrounding area. It then predicts the probability of a flammable cloud being ignited (delayed ignition) as the cloud moves downwind over ignition sources.

Point Sources

11.7.15 Since there is no significant ignition source at the LPG Store, no point source was applied to the risk model. Instead, the line sources and area sources described in the following sections were applied.

Line Sources

11.7.16 Roads are defined as line sources in PhastRisk. The following assumptions were applied to estimate the presence factor of the line source and the ignition probability:

a) The probability of ignition for a vehicle was taken as 0.4 in 60 seconds; and

b) The traffic density was based on the projected peak traffic flow as shown in Table 11.12 and Table 11.13.

Table 11.12 Summary of Road Ignition Sources for Construction Phase (Year 2026)

Line Source	Peak Hour Traffic Density (veh / hr)	Daytime Traffic Density (veh / hr)	Night-time Traffic Density (veh / hr)	Average Traffic Speed (km / hr)
Hoi Wong Road	1,223	1,223	573	50
Wu Shan Road	618	618	298
Wu Hong Street	309	309	149

Table 11.13 Summary of Road Ignition Sources for Operational Phase (Year 2031)

Line Source	Peak Hour Traffic Density (veh / hr)	Daytime Traffic Density (veh / hr)	Night-time Traffic Density (veh / hr)	Average Traffic Speed (km / hr)
Hoi Wong Road	1,275	1,275	600	50
Wu Shan Road	591	591	286
Wu Hong Street	296	296	143

11.7.17 In addition, the LRT and the Project are also defined as line sources in PhastRisk. The following assumptions were applied to estimate the presence factor of the line source and the ignition probability:

a) The probability of ignition was taken as 0.75 in 60 seconds; and

b) The presence factor of the ignition source was assumed to be 1.

Area Source

11.7.18 PhastRisk considers a residential population to an ignition source (as a result of activities such as cooking, smoking, use of heating appliances etc.). The ignition probability was derived from the population densities in the concerned area by the software.

Ignition Probability

11.7.19 Immediate ignition probabilities of 0.9 and 0.05 ^[1] were adopted for instantaneous release and continuous release of LPG, respectively. These ignition probabilities were applied to event trees as shown in Appendix 11.4.

Protection Factors

11.7.20 With reference to previous practice of assessments with PhastRisk in Hong Kong, protection factors were considered and applied to the concerned population groups if applicable.

Protection afforded to persons indoors in a building

11.7.21 It was generally assumed that the respective outdoor/ indoor population are 5% and 95% at the time of an accident ^[1].

11.7.22 For flash fire consequence, the fatality rate for indoor persons was assumed to be one tenth of the outdoor fatality rate.

11.7.23 For fireball, it was assumed that 50% of indoor persons would be killed.

Protection afforded to persons by being on the upper floors of building

11.7.24 Cloud height decreases further away from the source. Most dispersed clouds for LPG will have a cloud height lower than 10m ^[1]. It is equivalent to have only population on the lowest two floors of a building (including ground level) being affected. TME The actual population affected by release events was dependent on gas dispersion results modelled in PhastRisk. Height protection factors were applied to various population types for flash fire events accordingly. The track of TME will be more than 10m above ground level, protection factor was thus adopted for the passengers on TME trains. The actual population affected by flash fire events are detailed in Appendix 11.1.

11.7.25 Jet fire events had been assumed to only affect population below 10m elevation in previous similar assessment, which was confirmed by the modelling results in this assessment. All jet fires in the model were assumed to be horizontal or near-horizontal therefore reaching their maximum footprint radii. As with flashfires, only the population exposed (i.e. the population below 10m elevation) were considered in the risk summation for jet fire events, and the rest was excluded by the use of protection factor. The actual population affected by jet fire events are detailed in Appendix 11.1.

11.8 Risk Evaluation

11.8.1 The risks arising from the LPG supply facilities were evaluated in terms of both individual and societal risks in this assessment.

11.8.2 Individual risk is a measure of the risk to a chosen individual at a particular location. As such, this is evaluated by summing the contributions to that risk across a spectrum of incidents that could occur at a particular location.

11.8.3 Societal risk is a measure of the overall impact of an activity upon the surrounding community. As such, the likelihood and consequences of the range of incidents postulated for that particular activity are combined to create a cumulative picture of the spectrum of the possible consequences and their frequencies. This is usually presented as an FN curve and the acceptability of the results can be judged against the societal risk criterion under the risk guidelines.

Individual Risk

Risk Level

11.8.4 The predicted individual risk levels for the LPG Store are shown in Diagram 11‑5. The associated risk levels were based on 100% occupancy with no allowance made for shelter or escape, as specified in the user manual of PhastRisk. Since neither the construction activities nor operation of site areas “Land WA 5.4” and “Land WA 5.5” (population ID#9a and #9b) will induce additional hazard to the LPG Store, the individual risk plot is applicable to all the assessed scenarios.

Diagram 11‑5 Individual Risk Contours

Acceptability

11.8.5 As observed in Diagram 11‑5, the 1×10^-6, 1×10^-7, 1×10^-8 and 1×10^-9 per year contours extend approximately 60m, 75m, 130m and 200m from the storage vessels of the LPG Store, respectively. Given that there was no offsite risk with a frequency greater than 1×10^-5 per year, the level of individual risk associated with the operation of the LPG Store and the individual risk is considered acceptable and in compliance with the relevant criterion in Annex 4 of EIAO-TM.

Societal Risk