13 Quantitative Risk Assessment

This section of the EIA presents a summary of the analysis and findings of the Quantitative Risk Assessment (QRA) study undertaken for the proposed LNG Terminal at South Soko and associated facilities including a subsea pipeline from South Soko to Black Point Power Station (BPPS) and a Gas Receiving Station (GRS) at BPPS.

This section is divided into four sub sections: section 1 relates to the general aspects of the QRA study, section 2 relates to the LNG Terminal, section 3 relates to the subsea pipeline while section 4 relates to the GRS.

Further details of the analysis pertaining to each facility are presented in the respective annexes; Annex 13A covers the LNG Terminal QRA study details, Annex 13B covers the subsea pipeline while Annex 13C covers the GRS.

Additional annexes are provided to describe the Safety Management System (Annex 13D) and summarise all the assumptions adopted in the QRA study (Annex 13E).

Sub-Section 1: General

13.1 Legislation Requirement and Evaluation Criteria

The key legislation and guidelines that are considered relevant to the development of the proposed LNG Terminal and associated facilities are as follows:

· Gas Safety Ordinance, Chapter 51

· Hong Kong Planning Standards and Guidelines (HKPSG), Chapter 12

· Dangerous Goods Ordinance, Chapter 295

· Environmental Impact Assessment Ordinance (EIAO), Chapter 499

· The EIA Study Brief, Section 3.7.9.1

There is some overlap in the requirements of the various pieces of legislation and guidelines. The requirement for a Quantitative Risk Assessment study is contained in the EIAO and HKPSG. Such a study, although not required explicitly in the Gas Safety Ordinance, is implied in the regulations and has been an established practice for similar installations in the SAR.

13.1.1 EIAO Technical Memorandum (EIAO-TM)

The requirement for a QRA of projects involving storage, use and transport of dangerous goods where risk to life is a key issue with respect to Hong Kong Government Risk Guidelines (HKRG) is specified in Section 12 of the EIAO-TM.

The relevant authority for an QRA study relating to an LNG Terminal and associated facilities is the Gas Standards Office (GSO) of the Electrical and Mechanical Services Department (EMSD), as specified in Annex 22 of EIAO-TM.

Annex 4 of EIAO-TM specifies the Individual Risk and Societal Risk Guidelines.

13.1.2 Risk Measures and Hong Kong Government Risk Guidelines (HKRG)

Individual risk is the predicted increase in the chance of fatality per year to a hypothetical individual who remains 100% of the time at a given stationary point.

The individual risk guidelines require that the maximum level of off-site individual risk associated with a hazardous installation should not exceed 1 in 100,000 per year i.e. 1 x 10^-5 per year.

Societal risk expresses the risks to the whole population. The HKRG is presented graphically in Figure 13.1. It is expressed in terms of lines plotting the frequency (F) of N or more deaths in the population from incidents at the installation. Two FN risk lines are used in the HKRG to demark “acceptable” or “unacceptable” societal risks. The intermediate region indicates the acceptability of societal risk is borderline and should be reduced to a level which is “as low as reasonably practicable” (ALARP). It seeks to ensure that all practicable and cost-effective measures which can reduce risks will be considered.

Figure 13.1 Hong Kong Government Risk Guidelines

13.2 Study Objectives and Methodology

The objective of the QRA study is to assess the risk to life of the general public including the workers of nearby plants from the proposed facilities during its operational phase. The results of the QRA are compared with the HKRG.

The detailed requirements of the study are (see Section 3.4.9.1 of the EIA study brief):

· To identify all credible hazardous scenarios associated with storage, handling and operation of the LNG facility, which has potential to cause fatalities;

· To carry out the QRA expressing population risks in both individual and societal terms;

· To compare the individual and societal risks at the proposed development sites with the HKRG;

· To identify and assess practical and cost effective risk mitigation measures as appropriate;

· To identify all LNG leakage scenarios and propose a safety management system for the operational phase of the project with an aim to contain any accidental leakage in short notice and to prevent and/or minimise any leakage.

The elements of the QRA are shown schematically in Figure 13.2.

An overview of the methodology employed is provided here to briefly introduce the study approach, while the details are included in the respective sections/ annexes.

Relevant data on the proposed facilities such as their preliminary layout drawings and design basis as well as population data in the vicinity were collected and reviewed.

A Hazard Identification (HAZID) Study was conducted to identify all hazards, both generic and site specific. A review of literature and accident databases were also undertaken. These formed the basis for identifying all hazardous scenarios for the QRA Study.

The frequencies, or the likelihood, of the various outcomes resulting from an LNG/gas release scenario were derived from historical databases and, where necessary, these were modified to take into account local factors.

For all identified hazards assessed as having a frequency of less than 10^-9 per year, the frequency assessment will be documented but no quantification of consequences will be performed.

For hazards with frequencies greater than 10^-9 per year, the consequences of each release were modelled.

Hydrocarbon releases have been modelled using the PHAST consequence modelling package developed by Det Norske Veritas, Inc. (DNV)

The consequence and frequency data were subsequently combined using ERM’s proprietary software Riskplot^TM to produce the required risk calculations.

Finally, the results from the risk assessment were compared with the HKRG and found to be acceptable. No mitigation measures are therefore proposed.

Figure 13.2 Schematic Diagram of QRA Process

Sub-Section 2: Terminal

The QRA study for the terminal includes all planned facilities at the site, including unloading operations at the jetty, LNG storage tanks, sendout pumps, LNG vaporisers and the boil-off gas system.

This section presents a summary of the QRA study for the facilities at the terminal while Annex 13A gives further details.

As per the Study Brief, the QRA study for the terminal is also required to include the marine transit risks for LNG carriers within 500m of the jetty. A Marine Quantitative Risk Assessment (MQRA) study has been separately conducted by DNV. The risk results from the MQRA study for the 500m section at the jetty have been combined with the risk results for the facilities at the terminal to produce an overall risk result for the terminal, which is presented in this section. For further details on marine transit risks, DNV’s report may be referred.

13.3 Facility Details

13.3.1 Site Facilities

The proposed LNG Terminal and associated facilities will be built to provide a peak natural gas sendout capacity of 1000 million standard cubic feet per day (MSCFD).

The Terminal will comprise the following primary components:

· one jetty with a berth for LNG carriers;

· initially up to two 180,000m³ full containment LNG storage tanks (for Phase I,) followed by one additional tank (for Phase II);

· two in-tank LNG single stage centrifugal pumps for each tank, capable of delivering LNG at about 7 barg;

· four High Pressure LNG Booster pumps including one spare for Phase I and three additional pumps for Phase II, to deliver at about 101 barg;

· four including one spare (plus one additional for Phase II) open-rack seawater vaporisers.

The key features of the proposed LNG Terminal are depicted in the preliminary terminal layout diagram (See Figure 3.1 in Section 3). LNG is transferred by pumps under cryogenic conditions from the carrier to the tanks and from the tanks to the vaporisers. LNG is stored at near atmospheric pressure under cryogenic conditions. The vaporisers convert the LNG into gas phase for sendout to the pipeline.

Figure 13.3 shows a schematic of the overall process in an LNG terminal. A more detailed Process Flow Diagram for the Terminal, including details on the design features and the operating philosophy are included in Annex 13A1. Further details of safety features and the Safety Management System are provided in Annex 13D.

Figure 13.3 Process Overview

13.3.2 Land Use in the Vicinity

On South Soko Island, a number of unused Private Lots and derelict buildings are present as well as a recently refurbished Tin Hau Temple, but there is no permanent development or resident population at either South Soko or North Soko. A government owned low level radioactive waste storage facility is located on North Soko Island but this has no permanent staff and waste material will be delivered just twice a year.

13.3.3 Population Data

There is no residential population within 2km of South Soko Island and none proposed. The nearest areas of population are about 6km away on Lantau Island, where a population of about 10,000 is spread along the southern coast. The island of Cheung Chau has a population of 21,000 but is 13km away.

Marine population in the vicinity has been considered based on the marine traffic data provided by BMT [1]; approximately 0.4 person/km² is estimated in the vicinity of South Soko indicating very little marine activity. The Adamasta Channel is about 4km from the South Soko site and comprises of mainly ferries to Macau and some ocean going vessels.

Further details on land use adjoining the proposed site, as well as the land and marine population surrounding South Soko, are presented in Annex 13A2.

13.3.4 Weather Data

Weather data for the South Soko site is based on data from Cheung Chau weather station which is the closest and most relevant. Details are presented in Annex 13A3.

13.4 Hazard Identification

Hazards associated with the LNG terminal have been identified based on a detailed review of known incident records worldwide and experience gained from operations at similar facilities. In addition, a systematic Hazard Identification (HAZID) process was undertaken to identify any local or site specific factors.

13.4.1 Hazards from LNG

LNG is an extremely cold, non-toxic, non-corrosive and flammable substance. If LNG is accidentally released from a temperature-controlled container, it will likely contact warm surfaces and air that transfer heat into the liquid. The heat input begins to vaporise some of the liquid, returning the liquid to the gaseous phase. The relative proportions of liquid and gaseous phases immediately following a release depend on the release conditions. The liquid phase will form an LNG pool on the ground which will begin to “boil”, due to heat input from the surrounding environment.

Immediately following vaporisation, the gas is colder and heavier than the surrounding air and forms a vapour cloud. As the gas disperses, it mixes with the surrounding air and warms up. The vapour cloud will only ignite if it encounters an ignition source while concentrated within its flammability range.

Downstream of the vaporisers the natural gas will be in the gas phase. A release from this piping and equipment will result in a gaseous phase release directly.

13.4.2 Hazard Effects

In the event of an accidental release of LNG from piping or equipment, the characteristics of the possible hazardous effects are described below.

Pool Fire

A pool fire occurs when a flammable liquid is spilt onto the ground and ignited. A pool formed from the release of liquid LNG will initially spread due to the gravitational and surface tension forces acting on it. As the pool spreads, it will absorb heat from its surroundings causing evaporation from the pool surface. Ignition of this vapour leads to a pool fire.

Jet Fire

Jet fires result from ignited releases of pressurised flammable gas or superheated/pressurised liquid. The momentum of the release carries the materials forwards in a long plume entraining air to give a flammable mixture. Jet fires only occur where the LNG is being handled under pressure or when handled in gas phase and the release is unobstructed.

Flash Fire

Following an LNG release, a large proportion of the liquid will evaporate immediately to form a cloud of methane, initially located around the release point. If this cloud is not ignited immediately, it will move with the wind and be diluted as a result of air entrainment. Similarly, a gas release may not be ignited immediately and will disperse in the air.

The dispersing vapour cloud may subsequently come in contact with an ignition source and burn rapidly with a sudden flash. If the source of material which created the cloud is still present, then the fire will flash back to the source giving a pool fire or, if under pressure, a jet fire. Direct contact with the burning vapours may cause fatalities but the short duration of the flash fire means that thermal radiation effects are not significant outside the cloud and thus no fatalities are expected outside of the flash fire envelope.

Vapour Cloud Explosion

A flash fire is the most likely outcome upon ignition of a dispersing vapour cloud from an LNG release. If ignited in open (unconfined) areas, pure methane is not known to generate damaging overpressures (explode). However, if the gas is ignited in areas where there is significant degree of confinement and congestion an explosion may result.

Fireball

Immediate ignition of releases caused by a rupture in a gas piping may give rise to a fireball upon ignition. Fireballs have very high thermal radiation, similar to jet fires although the duration of the event is short.

To summarise, a liquid phase release may result in a flash fire, vapour cloud explosion, pool fire or jet fire. A gas phase release can result in a flash fire, fireball or jet fire.

13.4.3 Review of Industry Incidents

A review of industry incidents at LNG terminal facilities was carried out. Incident records over the last few decades show small LNG vapour releases and minor fires with impact limited to within the plant boundary. These were associated with leaks from valves and process equipment. There have been no instances of leaks to the environment from full containment tanks. There have been no injuries or fatalities recorded outside a plant boundary since 1944. Other incidents have occurred during the construction and repair of LNG facilities but no LNG was directly involved.

In general LNG facilities have shown an exceptionally high safety record due to the high level of safety features incorporated in an LNG terminal design including the use of full containment tanks and emergency shutdown systems.

13.4.4 HAZID Study

A Hazard Identification (HAZID) Study was conducted in October 2005 involving representatives from the Project Proponent: CLP and ExxonMobil and their expert consultants: ARUP, Foster Wheeler and ERM. The potential hazards posed by the facility were identified based on the HAZID team’s expert opinion, past accidents, lessons learnt and checklists. The details of the HAZID study can be found in Annex 13A4.

A systematic approach was adopted, whereby the facility was divided into a number of “subsystems” based on the layout and the process; guidewords from the checklist (see Annex 13A4) were then applied to each subsystem as relevant.

The Study considered each area of the LNG Terminal and identified any potential hazards that apply to it. The study output served as a basis for identification of scenarios for the QRA study.

13.4.5 Scenarios for QRA Study

Scenarios for the QRA study were identified based on the HAZID Study as well as a review of incident records. Loss of containment events have been identified for each section of the terminal, corresponding to the relevant process conditions, as listed in Table 13.1.

A detailed discussion on the hazards, particularly in relation to the LNG storage tanks, is given in Annex 13A5.

Table 13.1 Scenarios for QRA Study

Plant Section	Initiating Event	Potential Outcome Scenario
Jetty Area Unloading arm Piping & equipment at the jetty	Leak, rupture	Pool fire/Jet fire, Vapour dispersion/ Flash fire
Transfer Piping on Trestle Piping	Leak, rupture	Pool fire/Jet fire, Vapour dispersion/ Flash fire
Tank Area Piping on tank roof Storage Tank	Leak, rupture Rupture	Pool fire/Jet fire, Vapour dispersion/ Flash fire Pool fire, Vapour dispersion/ Flash fire
Process Area (HP Pumps, Recondenser, Vaporisers) Piping/equipment	Leak, rupture	Pool fire/Jet fire, Vapour dispersion/ Flash fire /Vapour cloud explosion
Process Area (Compressors) Piping/equipment	Leak, rupture	Jet fire, Gas dispersion/Flash fire, Fireball
Sendout Piping	Leak, rupture	Jet fire, Gas dispersion/ Flash fire, Fireball

13.5 Frequency Analysis

This includes an assessment of the likelihood or the frequency of events resulting in a hydrocarbon release from piping and equipment and the subsequent potential outcomes such as fires. Details of the frequency analysis are provided in Annex 13A6.

Release frequencies have been derived from generic data on loss of containment events. Reference has been made to a number of sources. A summary is presented in Table 13.2. Release scenarios include a range of hole sizes from small leaks to catastrophic rupture.

The frequency of various outcomes following a loss of containment event is estimated using an event tree model. The various outcomes considered include pool fire, jet fire, flash fire and vapour cloud explosions for liquid releases, jet fire and flash fire for continuous gas releases and fireball and flash fire for instantaneous gas releases.

Table 13.2 LNG Release Event Frequencies

Equipment	Release Scenario	Release Phase	Release Frequency	Unit	Reference
Process Vessels	i) 10 & 25mm hole	Liquid	1.00E-05	per year	Crossthwaite et al [2]
	ii) 50 & 100mm hole	Liquid	5.00E-06	per year	Crossthwaite et al
	iii) Full bore rupture	Liquid	1.00E-06	per year	Crossthwaite et al
Pumps	i) Leak	Liquid	1.00E-04	per year	COVO Study [3]
	ii) Full bore rupture	Liquid	1.00E-05	per year	COVO Study
Unloading Arm	i) Leak	Liquid/ Gas	4.05E-03	per year	COVO Study
	ii) Full bore rupture	Liquid/ Gas	4.05E-05	per year	COVO Study
Pipe size 600mm to 750mm	i) 10 & 25mm hole	Liquid/ Gas	1.00E-07	per meter per year	Hawksley [4]
	ii) 50 & 100mm hole	Liquid/ Gas	7.00E-08	per meter per year	Hawksley
	iii) Full bore rupture	Liquid/ Gas	3.00E-08	per meter per year	Hawksley
Pipe size 150mm to 500mm	i) 10 & 25mm hole	Liquid/ Gas	3.00E-07	per meter per year	Hawksley
	ii) 50 & 100mm hole	Liquid/ Gas	1.00E-07	per meter per year	Hawksley
	iii) Full bore rupture	Liquid/ Gas	5.00E-08	per meter per year	Hawksley
LNG Storage Tank	i) Rupture	Liquid	1.00E-08	per tank per year	“Purple Book” [5]

13.6 Consequence Analysis

This section gives a brief summary of the approach adopted to model the consequences of an LNG/natural gas release. Details are given in Annex 13A7.

A range of hole sizes from small leaks to full bore ruptures is considered in the analysis. Discharge rates, dispersion modelling, pool fire modelling, jet fire modelling, fire ball modelling and vapour cloud explosion modelling are considered and are all performed using the PHAST suite of models.

The plant was divided into twenty three isolatable process sections based on the provision of emergency shutdown valves. Physical properties of the fluid (pressure, temperature, density, phase) and equipment dimensions (pipe diameter and length) for each section were applied from the heat and mass balances to estimate the maximum release rate and the inventory in each section.

Fire radiation contours are calculated to 7.3, 14.4, 20.9 and 35.5 kW/m², and the fatality to people within each contour calculated. Overpressure effects from vapour cloud explosions are calculated to 5psi and 2psi contours. Dispersion of vapour clouds is determined to 0.85 of the lower flammability limit. A range of weather conditions is also considered, to represent a full year of conditions that occur within Hong Kong. Details of the consequence modelling and the results obtained are given in Annex 13A7.

13.7 Risk Results

13.7.1 Individual Risk Results

The individual risk (IR) contours associated with the LNG terminal are shown in Figure 13.4. The maximum off-site risk is less than 1 x 10^-5 per year at the site boundary, and hence meets the HKRG requirements.

Figure 13.4 Individual Risk Contours

13.7.2 Societal Risk Results

The societal risk for the South Soko site has been estimated based on marine and land population in the area. Three cases are considered: year 2011, year 2021 “no Tonggu” and year 2021 “with Tonggu”. The potential development of the Tonggu Waterway will increase the marine traffic through the Adamasta Channel and also south of South Soko so this possibility was considered in the analysis.

The societal risk results for the onshore terminal facilities have been combined with the risk results for the LNG carrier during berthing manoeuvres within 500m of the jetty to produce the overall societal risk results (Figures 13.5-13.7). The results for the berthing manoeuvres are taken from the Marine Quantitative Risk Assessment (MQRA).

Most of the points on the curve arise from scenarios involving the catastrophic failure of the LNG storage tanks. This is a very low frequency event but shows up in the FN curves because South Soko is remote and releases from other sections of the terminal are unable to reach populated areas.

Slight changes in the marine traffic for the three cases have resulted in small differences in the risk. The 2021 “No Tonggu” case has an increase in marine traffic predicted compared to 2011 and so the risks increase slightly. For 2021 “with Tonggu”, the marine population in the vicinity of South Soko is the same as the 2021 “No Tonggu” case, but the increased traffic south of South Soko increases the frequency of carrier collisions. Hence, the risks increase slightly (Figure 13.7). Also shown in the figures are results for large (215,000 m³) and small (145,000 m³) carriers. If LNG is delivered from smaller carriers, the number of transfers required per year will be higher. The frequency of possible releases therefore increases, but the consequences would be less severe.

The risks for all cases are well within the Acceptable Region as per HK EIAO.

The Potential Loss of Life (PLL), or equivalent fatalities per year, are given in Table 13.3. The total PLL for the whole terminal is very low at 3.4 x 10^-7 per year, or equivalently, one fatality every 3 million years.

13.3 Potential Loss of Life

Section		2011		2021
		PLL	%	PLL	%
T2 T1 T3 L05	LNG tank 2 LNG tank 1 LNG tank 3 Liquid unloading line from shore to tank	1.18 x 10^-7 1.14 x 10^-7 1.08 x 10^-7 1.13 x 10^-10	34.7 33.5 31.8 0.07	1.18 x 10^-7 1.14 x 10^-7 1.08 x 10^-7 1.13 x 10^-10	34.7 33.5 31.8 0.07
	Total	3.39 x 10^-7	100	3.39 x 10^-7	100

Figure 13.5 FN Curve for 2011

Figure 13.6 FN Curve for 2021 "No Tonggu"

Figure 13.7 FN Curve for 2021 "With Tonggu"

13.8 Conclusions of QRA Study for Terminal

The results indicate that the societal risks from the proposed facility are within the Acceptable Region of the HK EIAO. The individual risks also meet the requirements of the HKRG.

Sub-Section 3: Pipeline

The proposed subsea pipeline will transport compressed natural gas from the LNG terminal in South Soko Island to CAPCO’s Black Point Power Station. This section presents a summary of the QRA study for the subsea pipeline while Annex 13B gives further details.

Three marine traffic scenarios are considered in the analysis. The base case uses marine traffic data for 2011. To take into consideration the impact of future developments, predictions for 2021 traffic volume are also used, with and without the development of the Tonggu Waterway.

13.9 Pipeline and Marine Data

13.9.1 Pipeline Route

The proposed pipeline takes a subsea route from the LNG terminal at South Soko to Black Point Power Station (Figure 13.8), crossing the waterways of the Adamasta Channel and Urmston Road (not a designated channel).

The pipeline with a total length of about 38km will be buried to 3m below the seabed with varying levels of rock armour protection (Figures 13.8 and 13.9). Type 1B protection provides 1m of rock armour backfill and 2m of natural backfill above the pipeline. This provides protection for anchors up to 2 tonnes, essentially protecting against anchors from all ships below about 10,000 dwt. Trench types 2A/B are used on the shore approaches and are designed for protection from 2 tonne anchors and any future construction vessels. Trench types 2A/B are also designed to protect against scouring effects from wave action so that the pipeline is sufficiently protected when it makes the transition from subsea to land. The waterways of Urmston Road and the Adamasta Channel will have type 3A or 3B trenches consisting of 3m of rock armour backfill. Types 3A and 3B are essentially similar and are designed to protect against 20 tonne anchors. This covers the full range of ships currently operating in Hong Kong and also those expected in future.

13.9.2 Marine Traffic

The marine traffic report [1] divides the pipeline route into sections using ‘gate posts’ that roughly correspond to key locations along the route. Radar tracks of marine vessel movements are then used to determine the number of vessels crossing between pairs of gate posts each day. Based on the vessel speed and apparent size from the radar returns, vessels are also divided into six categories (Table 13.4). The same marine vessel classes as that used in the marine traffic report are used in this QRA study, although some interpretation of the data was required; to distinguish between fast ferries and fast launches in vessel class A2, for example.

Figure 13.8 Pipeline Route

Figure 13.9 Pipeline Trench Types

Table 13.4 Vessel Classes Adopted for Assessment

It was also necessary to make some assumptions regarding the population of each class of vessel. These are given in Table 13.5.

Table 13.5 Vessel Population

Class	Population
Fishing vessels Rivertrade coastal vessels Ocean-going vessels Fast launches Fast ferries Other	5 5 21 5 450/350/280/175/105/35* 5
* A distribution was assumed for the fast ferry population to reflect the occupancy at different time periods so that on average, the population is similar to the average load factor published by the Marine Department.

13.9.3 Segmentation of the Route

Based on considerations of the marine traffic data and the level of rock armour protection proposed for the pipeline, the pipeline route was divided into 12 sections for analysis (Table 13.6, Figure 13.10).

Table 13.6 Pipeline Segmentation

Section		Gate [1]		Kilometre Post		Length (km)	Typical water depth (m)	Trench type
		From	To	From	To
1 2 3 4 5 6 7 8 9 10 11 12	South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	LP2 IP1 IP2 IP4 IP5a IP7 IP7a IP10 IP10a	IP1 IP2 IP4 IP5a IP7 IP7a IP10 IP10a LP1	0 1.6 4.5 9.8 14.2 19.5 22.2 31.6 33.5 34.7 37.0 37.8	1.6 4.5 9.8 14.2 19.5 22.2 31.6 33.5 34.7 37.0 37.8 38.3	1.6 2.9 5.3 4.4 5.3 2.7 9.4 1.9 1.2 2.3 0.8 0.5	5 8 25 20 17 7 6 4 20 20 5 4	2A 1 3B/3A 3A 3B 1 1 1 3B/3A 3A/3B 1 2B

Figure 13.10 Segmentation of the Route

In some sections, the gate post locations were modified in this study to reflect changes in conditions. For example, the section from gates IP10a to LP1 spans different trench types and a sharp change in traffic intensity. In this case, the pipeline was divided into smaller sections and assumptions made regarding the marine traffic distribution based on the radar tracks (overlaid in Figure 13.10). Following this interpretation of the marine date, the traffic used for this study is as summarised in Table 13.7. Similar traffic tables were constructed for the future 2021 scenarios (Tables 13.8 and 13.9). With the Tonggu Waterway development, an increase in ocean-going vessels is expected to pass through the Adamasta Channel with a corresponding reduction in ocean-going vessel traffic in Urmston Road.

Table 13.7 Traffic Volume Assumed for Base Case 2011

		Traffic volume (ships per day)
Section		Fishing	River-trade	Ocean-going	Fast Launch	Fast ferry	Other	Total
1 2 3 4 5 6 7 8 9 10 11 12	South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	0 21 126 11 42 37 79 21 21 250 11 2	0 0 16 2 1 12 22 3 2 265 13 3	0 0 7 3 4 0 0 0 6 144 0 0	1 2 83 4 7 5 28 24 23 117 5 2	0 6 260 9 12 11 44 31 30 150 7 0	0 4 4 4 4 6 27 8 2 5 2 0	1 33 496 33 70 71 200 87 84 931 38 7
	Total	621	339	164	301	560	66	2051

Table 13.8 Traffic Volume Assumed for 2021 “No Tonggu” Case

		Traffic volume (ships per day)
Section		Fishing	River-trade	Ocean-going	Fast Launch	Fast ferry	Other	Total
1 2 3 4 5 6 7 8 9 10 11 12	Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	0 22 132 11 44 39 83 22 22 262 11 2	0 0 17 2 1 13 24 3 2 290 14 3	0 0 8 3 4 0 0 0 7 239 0 0	1 2 91 5 8 6 31 26 25 128 6 2	0 7 307 10 14 13 52 36 35 177 8 0	0 5 5 5 4 7 30 9 2 6 2 0	1 36 560 36 76 78 220 96 93 1102 41 7
	Total	650	369	261	331	659	76	2346

Table 13.9 Traffic Volume Assumed for 2021 “With Tonggu” Case

		Traffic volume (ships per day)
Section		Fishing	River-trade	Ocean-going	Fast Launch	Fast ferry	Other	Total
1 2 3 4 5 6 7 8 9 10 11 12	Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	0 22 132 11 44 39 83 22 22 262 11 2	0 0 17 2 1 13 24 3 2 290 14 3	0 0 162 3 4 0 0 0 7 85 0 0	1 2 91 5 8 6 31 26 25 128 6 2	0 7 307 10 14 13 52 36 35 177 8 0	0 5 5 5 5 7 30 9 2 6 2 0	1 36 714 36 76 78 220 96 93 948 41 7
	Total	650	369	261	331	659	76	2346

13.9.4 Pipeline Protection

Varying levels of rock armour protection are proposed for each section of the pipeline based on a qualitative assessment of the hazards identified by the pipeline engineering consultant (Aker Kvaerner) and marine traffic consultant (BMT). These levels of rock armour protection are assumed in the base case analysis as well as future traffic scenarios presented in this report.

13.10 Methodology

Key elements of the risk assessment methodology are described in the following sections.

13.10.1 Hazard Identification

Hazards were identified by reviewing worldwide databases and reports on incidents related to subsea pipelines. A HAZID (Hazard Identification) workshop was also conducted for the proposed pipeline to identify any route/site specific issues. The details of the hazard identification process are presented in Annex 13B.

The main hazard associated with a subsea pipeline is loss of containment resulting in gas release which could be ignited by a passing marine vessel in the vicinity. A loss of containment could occur from:

· Failures due to external impact (such as anchor drag)

· Spontaneous failures from corrosion and material/weld defects

· Natural hazards

13.10.2 Frequency Estimation

Frequency assessment is the estimation of the likelihood of occurrence of each scenario based on the hazard identification exercise. The approach adopted here for estimating frequency of pipeline failure is to apply worldwide historical data, with appropriate modifications for the specific pipeline environment.

The database that is most comprehensive and relevant is PARLOC 2001 [6]. This covers 300,000 km-years of subsea pipeline experience dating from the 1960s to 2000. This database provides failure frequencies for different causes such as corrosion, material defects, external impact etc. It also provides a breakdown for different diameter pipelines, location and contents of pipeline.

To validate this approach, particularly for anchor/impact damage where the specific marine traffic environment is more relevant, alternative calculations were performed for comparison. These were based on marine incident rates in Hong Kong waters, from which likelihood of emergency anchoring events were estimated. Frequency of anchoring was also estimated from the number of anchor marks found on the seabed based on recent geophysical surveys [7]. These alternate approaches were found to give similar failure frequencies to that derived from the PARLOC data. The frequencies used in the analysis are summarised in Table 13.10 while details are presented in Annex 13B.

The CAPCO pipeline will have rock armour protection along its whole length. To allow for this, protection factors are incorporated into the analysis. Trench types 1 and 2A/B are designed to protect against 2 tonne anchors. They are assumed to be 99% effective. They are also assumed to provide some protection (50%) against larger anchors. Trench type 3A/B is designed to protect against 20 tonne anchors, covering all ships currently operating in Hong Kong and those expected in the future. This trench type was assumed to be 99% effective for large anchors and provide even greater protection (99.9%) against smaller anchors, i.e. below 2 tonnes (see Table 13.10). Note that the rock armour design will be finalised during the engineering stage based on these performance considerations. The above assumptions are therefore conservative.

The frequencies used in the analysis are summarised in Table 13.10 while details are presented in Annex 13B. The probability of damage to the pipeline leading to a gas release is estimated as 0.37% for the 38km section during the lifetime of the facility, assumed as 30 years.

Table 13.10 Summary of Failure Frequencies Used

Pipeline section	Trench type	Corrosion /defects (/km/year)	Anchor/Impact			Others /km/year	Total /km/year
			Frequency (/km/year)	Protection factor (%)
				anchor<2	Anchor>2
South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	2A 1 3B/3A 3A 3B 1 1 1 3B/3A 3A/3B 1 2B	1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6 1.18 x 10^-6	1.37 x 10^-5 1.37 x 10^-5 8.6 x 10^-4 1 x 10^-4 8.6 x 10^-4 1 x 10^-4 1 x 10^-4 1 x 10^-4 8.6 x 10^-4 8.6 x 10^-4 1 x 10^-4 1 x 10^-4	99 99 99.9 99.9 99.9 99 99 99 99.9 99.9 99 99	50 50 99 99 99 50 50 50 99 99 50 50	1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6 1.34 x 10^-6	2.66 x 10^-6 2.66 x 10^-6 3.49 x 10^-6 2.70 x 10^-6 3.82 x 10^-6 3.52 x 10^-6 3.52 x 10^-6 3.52 x 10^-6 3.93 x 10^-6 4.58 x 10^-6 3.52 x 10^-6 3.52 x 10^-6

13.10.3 Scenario Development

The outcome of a hazard is predicted using Event Tree Analysis (ETA) to investigate the way initiating events could develop. This considers the cause of failure, the hole size distribution, the likelihood that a marine vessel will be in the area and the probability that the gas will be ignited. Historical data is used where appropriate; for the hole size distribution and ignition probability. The probability that a ship will pass through the flammable plume is calculated based on the size of the plume (obtained from dispersion modelling) and the marine traffic density.

13.10.4 Consequence Analysis

In the event of loss of containment in a subsea pipeline, the gas will release as a jet but is expected to lose momentum and bubble to the sea surface and disperse into the atmosphere as a buoyant gas. The dispersing plume may encounter an ignition source, say from a passing vessel, while within its flammable limits, leading to a flash fire, which will propagate through the gas cloud.

The flash fire could cause injury to personnel on marine vessels. It may also cause secondary fires on the vessel.

If a vessel passes close to the ‘release area’ (where bubbles of gas break through the sea surface), the consequences will be more severe. 100% fatality is assumed for this scenario. Once a fire has ignited, it is presumed that no further ships will be involved because the fire will be visible and other ships can take action to avoid the area. In other words, it is assumed that at most, only one ship will be affected.

Additional consequences may also arise from the proposed Hong Kong to Zhuhai Macau bridge (HKZM). Although the alignment of the bridge is not yet finalised, it is expected to straddle the CAPCO pipeline within the Tai O section (Figure 13.8). Impact on vehicle population due to flash fire events was considered in the analysis, as detailed in Annex 13B.

The pipeline alignment brings the North Lantau section of the pipeline within 3.7km of the thresholds for runways 07L and 07R of Chep Lap Kok Airport. Due to the shallow approach angle of aircraft, the possibility of a vapour cloud due to a large leak/rupture from the gas pipeline affecting a passing aircraft on the final approach to the airport was included in the analysis. See Annex 13B for details.

Helicopters plying to and from Macau fly at about 500 feet altitude. They cross the pipeline at the Adamasta Channel section on the Macau bound journey. The possibility of large leaks/ruptures affected a passing helicopter was included in the analysis.

13.11 Risk Results

13.11.1 Base Case 2011

The individual risk (IR) and potential loss of life (PLL) are given in Table 13.11. These risks are expressed in terms of per km to give a uniform basis for comparison between the various sections. The individual risk is less than 1 x 10^-5 per year for all sections of the pipeline. The total PLL, or equivalent annual fatality, for the whole length of pipeline is 1.4 x 10^-4 per year.

Table 13.11 Risk Results Based on Estimated 2011 Marine Traffic

Section		IR (per km-year)	PLL (per km-year)
1 2 3 4 5 6 7 8 9 10 11 12	South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	8.6 x 10^-9 1.1 x 10^-8 1.1 x 10^-7 1.3 x 10^-8 2.3 x 10^-7 7.9 x 10^-8 7.0 x 10^-8 8.0 x 10^-8 1.2 x 10^-7 3.9 x 10^-7 8.0 x 10^-8 6.7 x 10^-8	4.3 x 10^-8 3.8 x 10^-7 7.6 x 10^-6 5.1 x 10^-7 3.6 x 10^-6 3.5 x 10^-6 3.1 x 10^-6 5.2 x 10^-6 5.6 x 10^-6 6.6 x 10^-6 2.9 x 10^-6 3.3 x 10^-7

On a per km basis, the highest risks come from the Urmston Road and Adamasta Channel sections of the pipeline. This is because of the high traffic volume along these sections.

The FN curves for each section are presented in Figure 13.11. These are also expressed on a per km basis for comparison with the HKRG. The FN curves for all sections of the pipeline lie within the Acceptable Region.

The FN curves also show that the highest risks are associated with Urmston Road and the Adamasta Channel. Despite the high level of pipeline protection, the marine traffic volume is very high along these sections. The curves for these two sections actually cross over. The Adamasta Channel has a higher frequency of larger number of fatalities due to the relatively high number of fast ferries along this section. The Urmston Road has a higher frequency of lower number of fatalities due to the higher number of ocean-going vessels.

The Adamasta Channel was assigned a high failure frequency based on the high traffic volume but the data suggests that the marine vessel incident rate in this area, and hence the likelihood of emergency anchoring, is actually low. The approach has therefore been conservative.

The Sha Chau and North Lantau sections display the next highest risks. These sections were assigned an intermediate failure frequency based on the marine traffic, however, the shallow water along these sections means that only small draft vessels are present. The results are hence conservative.

For the Tai O section, results are presented inclusive of the HKZM bridge. The presence of the bridge does increase the risks slightly but the risks are still comfortably within the Acceptable Region. Similarly, aircraft on the approach to the airport and helicopters travelling to Macau make small contributions to the North Lantau and Adamasta Channel sections respectively, but the risks are in the Acceptable Region per HK EIAO.

The lowest risk occurs in the South Soko Approach and West Soko sections. This arises from the low traffic volume in this area.

Figure 13.11 FN Curve for each Section for Base Case 2011

13.11.2 2021 “No Tonggu” Scenario

The estimated future marine traffic scenario for year 2021 without the Tonggu Waterway development was also assessed, based on the future marine traffic predictions provided by BMT [1]. The frequency of damage due to anchor drop/drag was assumed to increase 15% in line with the average increase in density of marine vessels from 2011 to 2021. The frequency of corrosion failures and other types of failures was left unchanged.

The IR and PLL values are shown in Table 13.12. The greatest increase in risk occurs in Urmston Road and the Adamasta Channel. These are the main routes for fast ferries which are predicted to have the greatest increase in traffic. The IR for all sections remain below 1 x 10^-5 per year. The total PLL for the 38km of pipeline increases to 1.7 x 10^-4 per year.

Table 13.12 Risk Results Based on 2021 “No Tonggu” Case

Section		IR (per km-year)	PLL (per km-year)
1 2 3 4 5 6 7 8 9 10 11 12	South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	9.8 x 10^-9 1.2 x 10^-8 1.3 x 10^-7 1.4 x 10^-8 2.4 x 10^-7 9.1 x 10^-8 8.1 x 10^-8 9.3 x 10^-8 1.5 x 10^-7 5.9 x 10^-7 9.3 x 10^-8 7.7 x 10^-8	4.9 x 10^-8 4.7 x 10^-7 9.2 x 10^-6 5.8 x 10^-7 4.2 x 10^-6 4.2 x 10^-6 3.8 x 10^-6 6.3 x 10^-6 6.8 x 10^-6 9.1 x 10^-6 3.5 x 10^-6 3.8 x 10^-7

The FN curves (Figure 13.12) also show a slight increase but still lie in the Acceptable Region.

Figure 13.12 FN Curve for each Section for 2021 “No Tonggu” Case

13.11.3 2021 “With Tonggu” Scenario

Results from the 2021 future scenario with Tonggu Waterway development are shown in Table 13.13 and Figure 13.13. In this simulation, the frequency of anchor damage was again increased by 15% compared to 2011 in line with the average increase in traffic volume, while the frequency of failure due to corrosion and other causes was left unchanged.

The development of the Tonggu Waterway diverts many of the ocean-going vessels away from Urmston Road and into the Adamasta Channel. The IR, PLL and FN curves therefore show an increase in risk in the Adamasta Channel and a corresponding decrease for Urmston Road. The total PLL for the 38km pipeline remains the same at 1.7 x 10^-4 per year. All IR values are below 1 x 10^-5 per km-year and all FN curves remain in the Acceptable Region.

Table 13.13 Risk Results Based on 2021 “With Tonggu” Case

Section		IR (per km-year)	PLL (per km-year)
1 2 3 4 5 6 7 8 9 10 11 12	South Soko Approach West Soko Adamasta Channel West Lantau Tai O North Lantau Sha Chau North Lung Kwu Chau Urmston Road West Urmston Road Central Urmston Road East Black Point Approach	9.8 x 10^-9 1.2 x 10^-8 3.5 x 10^-7 1.4 x 10^-8 2.4 x 10^-7 9.1 x 10^-8 8.1 x 10^-8 9.3 x 10^-8 1.5 x 10^-7 3.5 x 10^-7 9.3 x 10^-8 7.7 x 10^-8	4.9 x 10^-8 4.7 x 10^-7 1.1 x 10^-5 5.8 x 10^-7 4.2 x 10^-6 4.2 x 10^-6 3.8 x 10^-6 6.3 x 10^-6 6.8 x 10^-6 7.0 x 10^-6 3.5 x 10^-6 3.8 x 10^-7

Figure 13.13 FN Curve for each Section for 2021 “With Tonggu” Case

13.12 Conclusions of Pipeline QRA Study

A QRA study for the proposed CAPCO pipeline was conducted. The study considered the loss of containment that may occur due to corrosion, material defects and third party damage from ship anchor drops/drags. Based on a review of the hazards, the marine traffic density and pipeline rock armour protection, the 38km proposed route was divided into twelve sections for assessment. Risks have been presented for each section on a per-km basis to provide a uniform basis for comparison.

The base case calculation used marine traffic data for 2011 and levels of rock armour protection for each section as proposed in the pipeline design.

The calculated levels of risk were compared with the HK EIAO and the following conclusions were drawn:

· The FN curves for all sections of the pipeline lie within the Acceptable Region.

· The highest risks are generally associated with the Adamasta Channel and Urmston Road where the marine traffic has the highest density.

· IR for all sections are predicted to be less than the 1 x 10^-5 per year as per HK EIAO criterion.

Future Marine Traffic Scenarios

For the future 2021 “No Tonggu” case, a 15% increase in marine traffic is expected compared to 2011. This increased the risks marginally. The FN curves still lie within the Acceptable Region. The IR is also below the 1 x 10^-5 per year as per HK EIAO criterion.

The future 2021 “with Tonggu” scenario redirects a significant number of ocean-going vessels from Urmston Road to the Adamasta Channel. This increases the risk in Adamasta Channel but values are still in the Acceptable Region. The IR remains below the 1 x 10^-5 per year as per HK EIAO criterion.

It is concluded that for all sections, the risks are acceptable per HK EIAO and no further mitigation measures are warranted.

Sub-Section 4: Gas Receiving Station (GRS)

The proposed pipeline from South Soko to Black Point Power Station (BPPS) will terminate at the gas receiving station (GRS) which will be located at the BPPS site. The gas is filtered, heated and letdown to 39 bar for delivery to the generation units.

This section presents the QRA results for the GRS.

13.13 Methodology

As explained in Section 3.3.4 of the EIA report, the GRS will contain a pig receiver, inlet filter-separators, metering, pre-heaters and a pressure letdown station. An emergency shutdown valve will be provided at the inlet to the station and also for individual section isolation in the event of any emergency. Preliminary site layout for the GRS along with Process Flow Diagrams and stream details are included in Annex 13C.

The methodology for the QRA of the GRS is similar to that adopted for the LNG terminal. The LNG terminal also contains sections of high pressure gas piping (downstream of the vaporizers) which are similar in design to the gas piping in the GRS. The LNG terminal contains submerged combustion vaporizers which are broadly similar to the pre-heaters proposed in the GRS.

The hazards associated with the GRS are mainly accidental releases from the high pressure gas piping. Upon ignition of this flammable gas, this may lead to a jet fire and/or flash fire.

The main population in the vicinity of GRS is the marine traffic along Urmston Road. Details are included in Annex 13C.

Meteorological data for the GRS is obtained from the Sha Chau Weather Station. The details on meteorological data, frequency and consequence parameters are included in Annex 13C.

13.14 Risk Results and Conclusion

13.14.1 Individual Risk Results

The individual risk is less than 1 x 10^-5 per year everywhere on site and at the site boundary, and hence meets the HKRG requirements.

13.14.2 Societal Risk Results

The potential loss of life for the gas receiving station is given in Table 13.14. There is essentially no change between 2011 and 2021 and values are very low given the low population in the vicinity. The total PLL is 2.8 x 10^-8 per year, or equivalently, one fatality every 35 million years.

Table 13.14 GRS Potential Loss of Life

Section		2011 (per year)		2021 (per year)
G2 G1 G3	Gas heater piping Gas piping from shutdown valve through gas filter to control valve Pressure control assembly	1.5 x 10^-8 7.5 x 10^-9 5.0 x 10^-9	54% 27% 18%	1.5 x 10^-8 7.5 x 10^-9 5.0 x 10^-9	54% 27% 18%
	Total	2.8 x 10^-8		2.8 x 10^-8

Figure 13.14 shows the FN Curve for the GRS at the BPPS. The curves are similar for Year 2011 and Year 2021.

It can be seen that the societal risk for the GRS is within the Acceptable Region as per HK EIA Ordinance.

Figure 13.14 FN Curve for GRS, Year 2011 and 2021

References

[1] BMT Asia Pacific Ltd, Marine Impact Assessment for Black Point & Sokos islands LNG Receiving Terminal & Associated Facilities, Pipeline Issues, Working Paper #3, Issue 6, May 2006.

[2] Crossthwaite, P.J., Fitzpatrick, R.D. & Hurst, N.W., Risk Assessment for the Siting of Developments Near Liquefied Petroleum Gas Installations, IChemE Symposium Series No 110, 1988.

[3] Rijnmond Public Authority, A Risk Analysis of Six Potentially Hazardous Industrial Objects in the Rijnmond Area – A Pilot Study, COVO, D. Reidel Publishing Co., Dordrecht, 1982.

[4] Hawksley, J.L., Some Social, Technical and Economic Aspects of the Risks of Large Plants, CHEMRAWN III, 1984.

[5] Guidelines for Quantitative Risk Assessment – Purple Book, 1^st Edition, Committee for the Prevention of Disasters, 1999.

[6] PARLOC 2001: The Update of Loss of Containment Data for Offshore Pipelines, 5^th Edition, Health & Safety Executive, 2003

[7] Hydrographic and geophysical Survey for Proposed LNG Terminal, Final Survey Report, EGS Earth Sciences & Surveying, 2005.

Sub-Section 5: Black Point and South Soko LNG Terminal Marine Quantitative Risk Assessment