12A Quantitative Risk Assessment for Submarine Gas Pipelines

12A.1 Introduction

This Annex covers details of the Quantitative Risk Assessment (QRA) for the two subsea pipelines from the Mainland China to Black Point Power Station. Details of the methodology are presented here whilst the results and conclusions are given in Section 12 of the EIA Report.

Two 20 km pipelines are proposed, although and only 5 km of the pipeline alignment lies within Hong Kong SAR waters. It is this 5 km of pipelines that are the subject of this analysis.

The first pipeline will likely be completed towards the end of 2011 and so the assessment considers the population in this year as the base case. Construction of the second pipeline will likely take place in 2014. A future scenario is also considered for the year 2021 when both pipelines will be operational.

12A.2 Data Collection & Review

The proposed pipelines from the Mainland China to Black Point are in many ways similar to the subsea pipeline that was proposed as part of the LNG terminal project for South Soko. The section of pipeline within Hong Kong SAR waters, for example, follows a similar alignment. Relevant information from these earlier studies has therefore been referenced where appropriate. The following information was reviewed and formed the basis of this study:

· Project Profile, ERM [1];

· Drawing HKLNG-WPL-00-PIIP-PL-009 detailing the pipeline trenching and backfill details, Worley Parsons [2];

· Marine vessel density data, BMT [3];

· Marine traffic data in Hong Kong waters, Marine Department (MD) [4, 5];

· UK Loss of Containment Database for Offshore Pipelines [6];

· Hydrographic & Geophysical Survey of the Seabed, EGS [7]; and

· Environmental and Risk Assessment Study for a LNG Terminal in Hong Kong, ERM [8].

12A.3 Pipeline & Marine Data

This section of the report describes the subsea pipelines, their environment and details of marine traffic along the proposed route.

12A.3.1 Subsea Pipelines

The proposed pipelines take a subsea route from the Mainland China to Black Point Power Station. The pipelines will cross the Urmston Road waterway (not a designated channel) and only 5 km is within Hong Kong SAR waters (Figure 12A.1). The Hong Kong section passes about 100-200 m north of the existing Yacheng pipeline. The seabed for much of the route is classed as very soft clay [7]. The water depth varies between 2 and 20 m, with the deeper sections corresponding to the busy Urmston Road fairway.

Details of the pipeline are preliminary at the time of writing but will likely consist of two pipes of between 32” and 42” diameter. These will be located in separate trenches constructed about 2 years apart. These differences are not expected to make significant differences in the risk results but where there is uncertainty in the design, the analysis has made assumptions that err on the conservative side. For example, the larger diameter of 42” has been assumed in the analysis since this creates a larger gas inventory. Construction of the pipelines at different times has also been considered.

The operational pressure within the pipelines is expected to be 63 barg, however, the maximum operating pressure of 100 barg (design pressure) is used in the analysis, again as a conservative upper limit. The pipelines will have an anti-corrosion coating and sacrificial anodes for external corrosion protection and an outer layer of reinforced concrete for buoyancy control and to provide mechanical protection during pipeline installation and trenching operations. A summary of the pipeline details is given in Table 12A.1.

The composition of the gas is mainly methane (85-99.5 mol%) and is such that no internal corrosion is expected.

The pipelines will be buried below the seabed with varying levels of rock armour protection (Figures 12A.1 and 12A.2). Type 1 trenching will be used for the approach to Black Point. The type 1 trench involves dredging with 1.5m of rock armour backfill (measured from the top of the pipeline). This provides protection for anchors up to 3 tonnes, essentially protecting against anchors from all ships below about 10,000 dwt. Trench type 2 is used in shallow water areas away from the busy marine fairways. Type 2 consists of post-trenching with about 5 m of armour rock and natural backfill. This is designed for protection from 3 - 5 tonne anchors (i.e. from all ships below about 10,000 dwt) and any future dredging work.

Table 12A.1 Summary of Pipeline Details

Parameter	Details
Location Length Outside diameter Nominal wall thickness Line pipe grade External coating Cathodic protection Design flowrate Design pressure Delivery pressure Maximum delivery pressure Pressure assumed for analysis Operating temperature Water depth Seabed soil Pipeline protection Design life	Mainland to Black Point Power Station 20 km 42” 1” API 5L X70 anti-corrosion coating Aluminium based sacrificial anodes 1200 MSCFD 100 barg 63 barg 100 barg 100 barg 12 °C 2 – 20 m Very soft clay becoming firmer with depth Up to 3m cover with rock armour backfill 25 years

The busy waterway of Urmston Road will have type 3 trenches. This consists of pre-trenching with 3m of rock backfill. This is designed to protect against 19 tonne anchors. This covers the full range of ships currently operating in Hong Kong and also those expected in future.

Figure 12A.1 Pipeline Alignment and Trench Type

Figure 12A.2 Pipeline Trench Types

12A.3.2 Marine Traffic

The marine traffic influences the risks from the pipeline in two ways:

· It increases the potential for damage due to interference such as anchor drop/drag incidents; and

· In the event of a pipeline failure, marine traffic could exacerbate the consequential effects causing fatalities.

The marine vessel traffic volume was surveyed by BMT [3] using tracks of vessel movements obtained from radar (Figure 12A.3). Details from the BMT report that are pertinent to the current study are summarised below.

Marine Vessel Activity along Pipeline Route

The marine traffic report [3] divided the previous South Soko to Black Point pipeline route into sections using ‘gate posts’ that roughly corresponded to key locations along the alignment. Three of these gate posts remain applicable to the current study and were used to estimate marine traffic crossing the 5 km of pipeline within Hong Kong SAR waters (Figure 12A.3).

Figure 12A.3 Radar Tracks of Marine Traffic

The section between gates 1 and 2 is used by fishing vessels and some rivertrade vessels en route between Tuen Mun and Zhuhai. The water is shallow in this region, ranging from 2 - 5 m deep. This precludes its use by large draft vessels.

Gate posts 0 and 1 span Urmston Road. Urmston Road is the main route for container ships, rivertrade vessels and fast ferries plying between Hong Kong and the ports of the Eastern Pearl River Delta.

Vessel Types

The marine traffic consultant calculated the marine traffic volume between pairs of gate posts based on radar tracks [3]. The vessel speeds and apparent size from the radar returns are interpreted into 6 marine vessel categories (Table 12A.2). The same categories are used for the current study.

Table 12A.2 Vessel Classes Adopted for Assessment

Based on this vessel classification, the population used in this study are as given in Table 12A.3. The maximum population of fast ferries is assumed to be 450, based on the maximum capacity of the largest ferries operating in the area. However, the average load factor of ferries to Pearl River ports is only 31.8% [9]. Hence, a distribution in ferry population was assumed as indicated in Table 12A.3. This distribution gives an overall load factor of about 58% which is conservative and covers any future increase in vessel population. There is an additional category in the traffic volume data called ‘Others’ (see Section 12A.3.2). These are assumed to be small vessels with a population of 5.

Table 12A.3 Vessel Population

Class	Population
Fishing vessels Rivertrade coastal vessels Ocean-going vessels Fast launches Fast ferries Others	5 5 21 5 450 (largest ferries in peak hours, 4 hours a day) 350 (average ferry in peak hours, 4 hours a day) 280 (80% capacity, peak hours, 4 hours a day) 175 (50% capacity, daytime operation, 9 hours a day) 105 (30% capacity, late evening, 4 hours a day) 35 (10% capacity, night time, 7 hours a day) 5	3.75% of trips 3.75% of trips 22.5% of trips 52.5% of trips 12.5% of trips 5% of trips

Traffic Volume

The traffic volume as provided by BMT [3] is given in Table 12A.4. This is based on radar tracks for the year 2003. The current study takes year 2011 as the base case since this is the expected year of completion of the pipeline. A future case, year 2021, is also considered. BMT provided predictions for the traffic increase to years 2011 and 2021 (Table 12A.5). The traffic growth rates presented in Table 12A.5 do not take into account the development of the Tonggu Waterway which has been implemented recently. This is expected to shift ocean-going vessels away from Urmston Road and into Tonggu instead. This is taken into account in the assessment by conservatively assuming that ocean-going vessel traffic remains at 2003 levels.

The data in Table 12A.4 required further interpretation. Vessel class A2 is described as fast launches and fast ferries. The population of a fast launch is very different from that of a fast ferry and so a more precise breakdown is required. Some of these A2 fast ferries clearly belong in class B2 with the other fast ferries. Taking into consideration the timetable of ferries serving the Pearl River ports and information provided by the marine consultant [10], it was established that 55% of fast vessels along Urmston Road are fast ferries.

Class C2 is described as fast ferries and ocean-going vessels. Since all fast ferries have now been accounted for, class C2 are assumed to comprise of cargo ships only.

The data shows a small number of ocean-going vessels (class C1 and C2) along the route between gates 1 and 2. The shallow water along this section negates the possibility that these are large vessels. They must be vessels at the smallest end of the distribution of ocean-going vessels, no more than 100m long [10]. More likely, they are rivertrade vessels. They were therefore treated as smaller vessels in the analysis by reclassifying them as either rivertrade or ‘other’ vessels.

Table 12A.4 Traffic Volume across Gate Sections (Daily Average, 2003)

Vessel Class									Total
Vessel Speed (m/s)		0-5			5-25			Others
Vessel Length Range (m)		0-30	30-75	75+	0-30	30-75	75+
From Gate	To Gate	“A1” Fishing Vessels and Small Craft	“B1” Rivertrade Coastal Vessels	“C1” Ocean-going Vessels	“A2” Fast Launches and Fast Ferries	“B2” Fast Ferries	“C2” Fast Ferries & Ocean-going Vessels
0	1	250	265	45	150	110	40	5	865
1	2	40	5	1	50	50	5	10	161

Notes: Values >5 are rounded to nearest 5

Daily values based on 9 day record. Some rounding applies

Table 12A.5 Traffic Growth Forecast

Vessel Type	2011 compared to 2003	2021 compared to 2003
Ocean-going Vessel* Rivertrade Coastal Vessel Fast Ferry Fishing Vessel/ Small Craft/ Fast launch Others	-5% +5% +10% +5% +5%	+10% +15% +30% +15% +15%

* The traffic growth forecasts for 2011 and 2021 do not take into account the development of the Tonggu Waterway. This waterway is expected to shift ocean-going vessels away from Urmston Road, resulting in a net decrease in traffic for large vessels. The analysis therefore retains 2003 traffic volumes for ocean-going vessels.

12A.3.3 Sectionalisation of the Pipeline

Based on the above discussions and the level of pipeline protection, the pipeline route was divided into 4 sections for analysis (Table 12A.6). The four sections include:

· Black Point Approach - the 0.1 km shoreline approach to BBPS;

· Black Point West - the shallow water section between Black Point and Urmston Road (between kilometre posts KP2.52 and KP4.78);

· Urmston Road - the Urmston Road fairway between KP0.73 and KP2.52 ; and

· Boundary Section - the shallow water section between Urmston Road and the boundary of Hong Kong SAR waters (KP0 to KP0.73).

Given that the pipeline sections do not correspond exactly with the gate posts used for determining the marine traffic, redistribution of the marine data was required. With insight gained from the radar tracks, knowledge of water depth and ferry activities, the following assumptions were made:

· 95% of the marine traffic between gates 0 and 1 is assumed to pass through Urmston Road;

· The remaining 5% of traffic observed between gates 0 and 1 are assigned to the Black Point West section. These marine vessels are assumed to be rivertrade vessels, fast launches, ferries and fishing boats. No large ocean-going vessels are expected here due to the shallow water;

· Although no radar tracks are observed within the 100m shore approach, a small number of small crafts are assigned to this section as a conservative measure;

· The mix of vessels observed between gates 1 and 2 is assumed to be representative of vessels crossing the Boundary Section of the pipeline. Half of the traffic observed between gates 1 and 2 is assumed to traverse the Boundary Section of pipeline.

Table 12A.6 Pipeline Segmentation

	Section	Kilometre Post		Length (km)	Typ. Water depth (m)	Trench type
		From	To
4	Boundary Section	0	0.73	0.73	2-20	2
3	Urmston Road	0.73	2.52	1.79	20	3
2	Black Point West	2.52	4.78	2.26	5	2
1	Black Point Approach	4.78	4.89	0.11	2	1

Based on the above assumptions, the marine traffic volume used in the present analysis is summarized in Table 12A.7.

Table 12A.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
4	Boundary Section	21	3	0	24	30	8	86
3	Urmston Road	250	265	81	118	150	5	869
2	Black Point West	12	16	0	5	8	2	43
1	Black Point Approach	1	0	0	0	0	0	1
	Total	284	284	81	147	188	15	999

Tables of traffic volume for the 2021 future scenario were created in a similar manner. This is given in Annex 12 of the EIA Report.

Ocean-Going Vessel Distribution

All classes of ship, with the exception of ocean-going vessels, have anchor sizes below 2 tonnes (Table 12A.2), and it is noted that the entire length of the proposed pipeline will have rock armour protection designed to protect against at least 3 tonne anchors. Ocean-going vessels cover a very wide range of size. A breakdown of the size distribution for this class of marine vessels is given in Table 12A.8 [3, 10]. These vessels are predominantly found in Urmston Road which has type 3 rock armour protection to protect against anchors up to 19 tonnes. From the size distribution, it can be seen that the majority of these ships are below about 110,000 tonnes displacement and so the majority of anchors are below 12 tonnes. 60% have anchor size below 5 tonnes.

Table 12A.8 Size Distribution of Ocean-Going Vessels

Size Range (dwt) ^†	Displacement (tonnes)*	Length (m)	Anchor Size (tonne)	Proportion of Ships (%)
1,500 – 25,000 25,000 – 75,000 75,000 – 100,000	1,500 – 35,000 35,000 – 110,000 110,000 – 150,000	75 – 200 200 – 300 300 – 350	2 – 5 5 – 12 12 – 15	60 35 5
† Dead Weight (dwt) = Cargo + Fuel + Water + others * Displacement = Total Weight = Hull + Machinery + Outfit + Dead Weight Displacement has been assumed to be ~ 1.4 ´ dwt

12A.4 Hazard Identification

This section identifies the main hazards from the subsea gas pipelines. Hazard identification is based on a literature review of past incidents as well as HAZID studies (Section 12A.4.2) conducted for the proposed pipeline. Hazards identified from these studies are then carried forward for further consideration in the QRA.

12A.4.1 Literature Review

Incident Databases and Pipeline Reports

The Consultants (ERM) have examined incident databases such as the MHIDAS [11] and the IChemE Accident Database [12]. Only two pipeline incidents in offshore Vietnam have been reported in the Asia-Pacific region. These occurred at White Tiger and Vung Tau, both in 1994 and both were caused by anchor damage. No injuries were reported.

Relevant reports on major subsea pipeline failures (that caused fatality) by the National Transportation Safety Board have also been reviewed [13, 14]. A summary of a few main incidents from these sources are included in the following paragraphs.

Tiger Pass, Louisiana, 1996

On October 23, 1996, in Tiger Pass, Louisiana, the crew of the dredge Dave Blackburn dropped a stern spud (a spud is a large steel shaft that is dropped into the river bottom to serve as an anchor and a pivot during dredging operations) into the bottom of the channel in preparation for continued dredging operations. The spud struck and ruptured a 12" diameter submerged natural gas steel pipeline. The pressurised (about 930 psig) natural gas released from the pipeline enveloped the stern of the dredge and an accompanying tug. Within seconds of reaching the surface, the natural gas ignited and the resulting fire destroyed the dredge and the tug. All 28 crew members from the dredge and tug escaped into water or onto nearby vessels. No fatalities resulted.

The incident occurred due to incorrect information on the location of the gas pipeline that was passed on by the gas company to the dredging operator. The investigation report on the incident (by the National Transportation Safety Board) recommended that all pipelines crossing navigable waterways are accurately located and marked permanently.

Mississippi River Delta, 1979

In an incident in the Mississippi River Delta in 1979, four workers drowned attempting to escape a fire that resulted when a crane barge dropped a mooring spud into an unmarked high pressure natural gas pipeline.

Louisiana, 1987

In July 1987, while working in shallow waters off Louisiana, a fishing vessel, the menhaden purse seiner Sea Chief struck and ruptured an 8" natural gas liquids pipeline operating at 480 psi. The resulting explosion killed two crew members. Divers investigating found that the pipe, installed in 1968, was covered with only 6" of soft mud, having lost its original 3-foot cover of sediments.

Sabine Pass, Texas, 1989

A similar accident occurred in October 1989. The menhaden vessel Northumberland struck a 16" gas pipeline in shallow water near Sabine Pass, Texas. The vessel was engulfed in flames; 11 of the 14 crew members died. The pipeline, installed in 1974 with 8 to 10 feet of cover, was found to be lying on the bottom, with no cover at all.

Pipeline Failure Databases

There are a few international failure databases for gas and liquid transmission pipelines which are useful in identifying potential hazards and estimating the frequency of loss of containment incidents.

The most comprehensive database on offshore gas pipeline failures is available in a report published by the UK Health and Safety Executive entitled 'PARLOC 2001' [6]. The most recent version of this database covers incidents from the 1960s up to 2000. The information in this database is based on data obtained from regulatory authorities in the UK, Norway, the Netherlands, Denmark and Germany, Operators in the UK, Dutch and Danish sectors and published sources. The main causes of pipeline failure, as identified from a review of the PARLOC 2001 data, are listed in Table 12A.9. Based on this, it can be seen that anchor/impact followed by internal corrosion are the main contributors to subsea pipeline failures.

A similar database on incidents involving offshore pipelines in the US has also been referred to [15]. This is based on incidents that are required to be reported to the US Department of Transportation (DOT) under the Federal Regulations. Out of 109 incidents reported during the period 1985 to 1994, only one incident involved a fatality, and only one incident involved leak ignition. The main causes of pipeline failure, as identified from a review of the US DOT database, are listed in Table 12A.10. Based on this, it can be seen that third party damage and internal corrosion (characteristic of well fluid pipelines) are the main contributors to subsea pipeline failures.

Table 12A.9 Causes of Subsea Pipeline Incidents from PARLOC 2001 [6]

Main cause	Detail	No. of Incidents of Loss of Containment
		Platform Safety Zone^(1)	Subsea Well Safety Zone^(2)	Mid-line
ANCHOR	Supply Boat	6	-	-
	Rig or Construction	-	-	-
	Other/ Unknown	0	-	2
	Total	6	-	2
IMPACT	Trawl	-	-	6
	Dropped Object	-	-	-
	Wreck	-	-	1
	Construction	1	-	-
	Other/ Unknown	-	-	1
	Total	1	-	8
CORROSION	Internal	3	4	7
	External	1	-	2
	Unknown	1	-	2
	Total	5	4	11
STRUCTURAL	Expansion	-	-	-
	Buckling	-	-	-
	Total	-	-	-
MATERIAL	Weld Defect	2	-	1
	Steel Defect	2	1	1
	Total	4	1	2
NATURAL HAZARD	Vibration	-	-	-
	Storm	-	-	-
	Scour	-	-	-
	Subsidence	-	-	-
	Total	-	-	-
FIRE/ EXPLOSION	Total	-	-	-
CONSTRUCTION	Total	-	-	-
MAINTENANCE	Total	-	-	-
OTHERS	Total	2	1	4
TOTAL		18	6	27
(1) Platform safety zone and subsea safety zone refer to pipelines located within 500m of an offshore platform and subsea well respectively (2) Mid-line refers to pipelines located more than 500m from a platform or subsea well.

Table 12A.10 Causes of Subsea Pipeline Incidents from US DOT Database [15]

Cause of Failure	Description of Cause	No. of Incidents	% of Total Incidents	Incidents Considered^(1)
1. EXTERNAL FORCE		25	29.8%	24
Earth Movement	Subsidence, landslides	2	2.4%	2
Heavy Rains/Floods	Washouts, floatation, scouring	1	1.2%
Third Party		21	25.0%	21
Previously Damaged Pipe	Where encroachment occurred in the past	1	1.2%	1
2. CORROSION		45	53.6%	3
External Corrosion	Failure of coating/CP	3	3.6%	3
Internal Corrosion		42	50.0%
3. WELDS & MATERIALS		4	4.8%	4
Defective Fabrication Weld	Welds in branch connections, hot taps, weld-o-lets, sleeve repairs	2	2.4%	2
Defective Girth Weld		2	2.4%	2
4. EQUIPMENT & OPERATIONS		3	3.6%
Equipment Failure	Malfunction of control or relief equipment, failure of threaded components, gaskets & seals	3	3.6%
5. OTHERS		7	8.3%	7
Unknown		7	8.3%	7
TOTAL		84	100%	38
1. Only these incidents are considered relevant to the proposed pipeline.

Incident Records and Protection Measures for Pipelines in Hong Kong SAR Waters

A review of existing and proposed subsea pipelines in Hong Kong waters including the level of protection provided are reviewed in the following paragraphs.

Subsea Pipelines

Existing subsea pipelines in Hong Kong waters are as follows:

· The 28" natural gas pipeline from Yacheng Field, South China Sea (90km south of Hainan Island) to CAPCO’s Black Point power station was constructed in 1994/95. The total pipeline length is 778 km. Within Hong Kong waters, the length of pipeline is about 5 km and the water depth varies from 4 m to 25 m. The pipeline is trenched with a minimum of 1m rock armour protection at sections where it crosses the shipping route Urmston Road and at the anchorage areas near the shore. Similar protection (i.e. 1m rock armour and 1 m backfill) is also provided outside Hong Kong waters at the Lingding channel crossing and Jiuzhou channel crossing. The pipeline is laid on the seabed for the remaining length. There has been no incident of damage reported in Hong Kong waters although an incident occurred during construction when the unprotected section of the pipeline was buckled by the anchor lines of the barge laying the rock armour.

· the 20" dual aviation fuel pipelines between Sha Chau jetty and the airport (about 5km length), installed in 1997, are laid in a 2.2 m trench and provided with sand cover plus rock armour protection. The water depth along the route varies from 4-7 m. There has been no incident of damage reported;

· the Airport Authority propose to construct another 5 km submarine aviation fuel pipeline from Sha Chau jetty to the new tank farm in Tuen Mun. The pipeline will be crossing the Urmston Road shipping route and similar protection as for the existing pipelines (i.e. rock armour protection) is proposed;

· the town gas subsea pipelines are also reported to have no damage record. These pipelines are laid at a depth of 2 to 3 m below seabed and protected by engineering backfill materials;

· the Hongkong Electric Company Limited recently laid a pipeline from its Lamma Power Station Extension to Shenzhen LNG Terminal. The pipeline is jetted to 3 m below seabed and protected with rock armour in high risk areas near the anchorages and shore approaches; and

· the recently installed town gas subsea pipeline from Shenzhen to Tai Po is jetted to 3 m below seabed with additional rock armour protection in high risk areas.

By comparison, the proposed CAPCO pipeline will be laid in waters between 2 and 20 m deep. The pipeline will be provided with 3 m of rock cover except in areas of shallow water where it will have 1.5 – 5 m of rock/ natural fill cover. These rock cover requirements are based on water depth (which determines the size of vessels) and marine traffic volume. The measures proposed are in line with, or exceed, comparable pipeline installations.

12A.4.2 HAZID Report

A Hazard Identification (HAZID) workshop was held in September 2009 as part of the risk assessment to identify issues specific to locality of the pipeline. The worksheets from this workshop are presented in Table 12A.11.

Table 12A.11 HAZID Worksheets

System: 1. Pipeline
Subsystem: 1. Operational
Hazards/ Keywords	Description/ Causes	Consequences	Safeguards	Recommendations
1. Internal corrosion	1. No issue for non corrosive, clean and dry gas
2. External corrosion	1. Sea-water; corrosive environment	1. Loss of wall thickness leading to potential leak	1. Coating system
			2. Sacrificial anode system
			3. Designed for intelligent pigging
3. Pressure cycling	1. Pipeline pressure will vary with time of day, loads etc	1. Metal fatigue leading to crack	1. Design will consider pressure cycles
4. Material defect/ construction defect		1. Possible leaks	1. Quality control during manufacture and construction
5. Impact from one pipeline to the other	1. No issues identified during operation
6. Maintenance	1. Possible damage to one pipeline during maintenance/intervention on the second.	1. Possible damage to pipeline leading to potential leaks	1. Maintenance procedures: - proper equipment - surveying (GPS positioning) - marker buoys

System: 1. Pipeline
Subsystem: 2. External hazards
Hazards/ Keywords	Description/ Causes	Consequences	Safeguards	Recommendations
1. Anchor Drag	1. Emergency anchoring for vessel underway due to loss of steerage, power or control, either due to mechanical problems or due to collision events.	1. Possibility of damage to external coating, damage to pipe requiring remedial action.	1. Engineered rock protection with respect to vessel sizes/types.	1. Periodic survey along the route to be carried out to ensure integrity of the protection.
	2. Drag from anchorage areas under storm conditions.	2. Potential loss of containment leading to gas release. Impact on passing vessels and shore population. Vessel involved in the incidents may sink due to loss of buoyancy cause by the gas bubbling.	2. Depth of cover.
	3. Anchoring by vessels outside anchorages.	3. Disturbance to the rock cover protection. Possible exposure of the pipe.	3. Route avoiding anchorage areas.
			4. Concrete external coating.
			5. Heavy wall pipe in shore approaches.
			6. Marking marine charts of the pipeline route.
			7. Shore population is at least 3km away along the route except near the shore approach.
2. Anchor Drop	1. Same as cause 1 & 3 of anchor drag hazard	1. Same as consequence 1, 2 & 3 of anchor drag hazard but less severe.	1. Same as for anchor drag hazard.
3. Dropped Object	1. Loss of cargo	1. Same as consequence 1, 2 & 3 of anchor drag hazard but less severe.	1. Same as safeguards 1, 2, 4, 5, & 7 of anchor drag hazard.
	2. Construction activities
4. Dumping	1. Dumping of construction waste and other bulk materials outside of designated dumping grounds.	1. Minor surface damage.	1. Same as safeguards 1, 2, 4, 5, & 7 of anchor drag hazard.
5. Grounding	1. Navigation error, loss of control due to mechanical or adverse weather.	1. Same as consequence 1, 2 & 3 of anchor drag hazard.	1. Burial depth appropriate to the type of shipping activities
		2. Displacement of the pipeline leading to exposure	2. Comprehensive risk based design has been conducted and the pipeline alignment minimizes exposure to major shipping lanes. Pipeline is routed through shallow water as far as possible.
6. Vessel Sinking	1. Collision, foundering.	1. Same as consequence 1, 2 & 3 of anchor drag hazard.	1. Comprehensive risk based design has been conducted and the pipeline alignment minimizes exposure to major shipping lanes. Pipeline is routed through shallow water as far as possible.
7. Fishing & Trawling	1. Operation of trawl board and other fishing/trawl gear.	1. No damage to the pipeline.	1. Pipeline is buried below the seabed with rock cover flush with seabed.
8. Dredging	1. Impact from dredge bucket or drag head.	1. Same as consequence 1, 2 & 3 of anchor drag hazard but less severe.	1. Burial depth appropriate to the type of shipping activities based on Marine Department and CEDD guidelines.
			2. Engineered rock protection with respect to vessel sizes/types.
			3. Depth of cover.
			4. Marking marine charts of the pipeline route.
			5. Concrete external coating.
9. Service crossing or other services in the vicinity	1. No crossings envisaged		1. Surveys have demonstrated no other services along the pipeline route

System: 1. Pipeline
Subsystem: 3. Natural hazards
Hazards/ Keywords	Description/ Causes	Consequences	Safeguards	Recommendations
1. Scouring	1. Current and wave actions	1. Possible reduction of cover	1. Alignment is away from areas of high currents	1. Periodic survey along the route to be carried out to ensure integrity of the protection.
			2. Engineered rock cover
2. Seismic event	1. Low seismic activity area	1. No damage	1. None required
3. Subsidence	1. No issue

System: 1. Pipeline
Subsystem: 4. Construction / future developments
Hazards/ Keywords	Description/ Causes	Consequences	Safeguards	Recommendations
1. Damage to pipeline during construction of second pipeline	1. Damage from construction activities	1. Damage to pipe and possible loss of containment	1. Design for appropriate separation distance	2. Design for second pipeline should be taken into account during construction of the first. Critical areas such as the shore approach should be pre-constructed in parallel for the two pipelines.
			2. Construction procedures: - proper equipment - surveying (GPS positioning) - marker buoys.
			3. Pipeline protection design covers foreseeable marine activities including dredging and anchoring
2. Reclaimed land over first pipeline	1. Weight of overburden may lead to subsidence and damage to first pipeline.	1. Overstressing of the first pipeline leading to catastrophic failure	1. Conservative design taking into account the overburden.

12A.4.3 Hazardous Properties of Natural Gas

The natural gas to be transmitted by the pipeline predominantly contains methane (85 - 99.5 mol%). It is a flammable gas that is lighter than air (buoyant). The properties of natural gas are summarised in Table 12A.12.

Table 12A.12 Properties of Natural Gas

Property	Natural Gas
Synonyms State Molecular Weight Density (kg/m^3) Flammable Limits (%) Auto-ignition Temperature (°C)	Methane Gas 16.0 - 18.7 0.55 (at atmospheric conditions) 5 - 15 540

12A.4.4 Discussion on Subsea Pipeline Hazards

The incident records highlight the potential for damage to subsea pipelines from marine activity such as fishing, dredging and anchoring as well as the potential for the vessel (that caused damage) to become involved in the fire that follows.

A review of subsea pipeline incidents in Europe and the US suggests that third party damage (including anchor and impact incidents) and internal corrosion are the main contributors to subsea pipeline failures.

It is noted that the above databases cover a large proportion of well fluid pipelines where internal corrosion is relevant as compared to clean natural gas as considered in this study.

Most existing pipelines in Hong Kong waters have some rock cover protection in addition to being buried, although it is noted that these pipelines are either crossing shipping channels or laid in waters with high levels of marine activity.

A brief description of the main causes of failure of a subsea pipeline is included in the following paragraphs.

External Impacts

Anchor drop/drag is the dominant cause of potential failure or damage to a subsea pipeline. This occurs when a ship anchor is dropped inadvertently across the pipeline. The type of damage that could be caused will vary depending on the size of anchor and other factors such as pipeline protection.

Anchor Drop

The decision for a mariner when to drop an anchor depends on the particular circumstances and the proximity of the pipeline route to the flow of marine traffic, port/harbour areas and designated anchorage locations. In fairways, traffic will normally be underway where the necessity to drop anchor is expected to be low. Consistent with normal practice, the pipeline route will be identified on nautical charts. The mariner is then provided with the necessary information to avoid anchoring where the pipeline could be damaged.

Emergency situations may arise such as machinery failure or collision thereby limiting the choice where to drop anchor. Such a decision will, as part of a mariner’s responsibility, be influenced by the particular circumstances and the pipeline route delineated on the navigation chart.

Although it is expected that vessels should be aware of all subsea installations (including gas pipelines) since these are marked on the admiralty nautical charts, erroneous dropping of anchor (i.e. error in position at the time of deployment) are known to occur.

Anchor Drag

Anchor drag occurs due to poor holding ground or adverse environmental conditions affecting the holding power of the anchor. The drag distance depends on properties of the seabed soil, the mass of ship and anchor and the speed of the vessel. If there is a subsea pipeline along the anchor drag path, anchor dragging onto the pipeline may result in localised buckling or denting of the pipeline, or over-stressing from bending if the tension on the anchor is sufficient to laterally displace the pipeline. A dragged anchor may also hook onto a pipeline during retrieval causing damage as a result of lifting the pipeline.

Vessel Sinking

Vessel sinking in the vicinity of the pipeline may cause damage to the pipeline resulting in loss of containment. Vessel sinking will depend on the intensity of marine activity in a given area. For the years 1990 to 2007, there were 492 incidents of vessel sinking in Hong Kong waters [16]. This averages 27 cases per year. Most of the recorded incidents occurred in Victoria Harbour and the Ma Wan Channel and involved mainly smaller vessels of less than 1,000 dwt, which will have less impact on a pipeline buried below the seabed. The probability that a vessel sinking incident will impact the proposed pipeline is therefore considered to be low, in comparison to anchor impact damage. Additionally, pipeline damage due to vessel sinking is included in the historical pipeline failure data for external impact used in this study (see Table 12A.9).

Dropped Objects

Objects other than anchors may be dropped from vessels passing over the pipeline or vessels operating in the vicinity. The dropped objects may include shipping containers, construction/maintenance equipment, etc. The pipelines will be lowered to at least 1 m below seabed and protected by rock armour. Given the likely sizes of dropped objects and the level of pipeline protection provided, loss of containment due to dropped objects is not considered to be a significant contributor to the risk. Such events will in any case be included in the historical pipeline failure data for external impact used in this study.

If any future construction work is conducted in the vicinity of the pipeline, procedures will be developed to safeguard the pipeline during the construction activities.

Aircraft Crash

The proposed pipeline route does not lie close to Chek Lap Kok or Macau airports or flight paths. As such, the possibility of an aircraft crashing onto the pipeline has a very small probability. Also, aircraft are constructed from light weight materials such that even a fully loaded Boeing 747 weighs only 400 tonnes. Aircraft also readily breakup on impact with water, scattering the debris over a larger area. Given that the pipeline is buried and protected and aircraft have limited weight, it is considered not possible for an aircraft to damage the pipeline.

Fishing Activity

Based on the BMT report [3], there is fishing activity along the proposed pipeline route. Many of the techniques involve towing of a variety of equipment along the seabed. Pipeline damage from fishing gear can occur due to impact, snagging of nets or trawl door on the pipeline or a "pull over" sequence. Impact loads mainly cause damage to the coating whilst pull over situations can cause much higher loads, which could lead to damage of the steel pipeline itself.

The vessels of concern are stern trawlers with lengths up to 30 m. Considering the size and weight of trawl gear and since the pipeline will be lowered to at least 1 m below seabed and protected by rock armour for the entire route, pipeline damage due to trawling activities are not possible and are not considered further.

Dredging and Construction Activities

Dredging vessels could cause damage due to dredging operations involving cutting heads. They could also cause damage to the pipeline by anchoring.

It is assumed that dredging operations will be closely monitored and controlled and therefore there is negligible potential for pipeline damage due to dredging.

Spontaneous Failures

Corrosion

Corrosion is one of the main contributors to pipeline failures. Corrosion is attributed mainly to the environment in which they are installed (external) and the substances they carry (internal).

The proposed pipeline will be protected against external corrosion by sacrificial anodes in addition to an anti-corrosion coating. However, ineffective corrosion protection due to a failure or breakdown of the protection system could cause external corrosion resulting in general or local loss of wall thickness leading to pipeline failure.

Historically, internal corrosion is a greater cause of pipeline failure compared to external corrosion. However, the proposed pipeline will transport gas that does not contain components that induce corrosion such as water/moisture, carbon dioxide, hydrogen sulphide, etc. This will largely reduce the chance of internal corrosion.

Despite these considerations, loss of containment due to corrosion (both internal and external) remains a possibility and is included in the analysis.

Mechanical Failure

Mechanical failure of the pipeline could occur for various reasons, including material defect, weld failure, etc. Stringent procedures for pipeline material procurement, welding and hydrotesting should largely mitigate against these hazards. In any case, it remains a credible scenario and is included in the frequency data.

Natural Hazards

Natural hazards such as subsidence, earthquake and typhoon may cause varying degrees of damage to pipelines.

Soft soil can sometimes suffer from localised liquefaction which can result in pipelines floating out of their trenches. The pipeline will be designed to withstand such loads, based on detailed seabed investigations.

Environmental loads (currents and waves) on the pipeline during the construction phase can compromise the lateral and vertical on-bottom stability of the pipeline on the seabed. This problem becomes more acute in shallower waters (near the shore) where the pipeline attracts a higher level of environmental loads. The pipeline will be designed to withstand these environmental loads. Once it is lowered below the seabed, it would not be exposed directly to 100 year return wave loads.

Based on the above considerations, it is considered that there is no disproportionate risk to the pipeline from natural hazards. These causes of failure are in any case included in the generic failure rates derived from historical incidents, as used in this study.

12A.5 Frequency Analysis

12A.5.1 Overview

This section presents the base failure frequency data for the hazards identified in Section 12A.4. The approach to frequency analysis is based on the application of worldwide historical data for similar systems, modified suitably to reflect local factors such as proximity of the pipeline to busy shipping channels and anchorages.

Event tree analysis was used to determine the probabilities of various hazard outcomes (such as flash fire) occurring, following a release.

12A.5.2 Historical Data

The international database that is most comprehensive in its coverage of subsea pipelines is PARLOC 2001 [6]. The most recent version of this database which was used in this study covers incidents from the 1960s until 2000. Incidents recorded in the database have been classified according to several categories, including:

· Failure location, i.e. risers, pipelines within 500 m of an offshore platform, pipelines within 500 m of a subsea well and mid-line (pipelines located more than 500 m from a platform or a subsea well). Failure data pertaining to risers is not relevant to this study and has therefore been excluded;

· Pipeline contents. The database includes both oil and gas pipelines. Where the contents in the pipeline have an impact on failure rate, such as corrosion, only incidents pertaining to gas pipelines are considered; and

· Pipeline type, i.e. steel pipelines (both pipe body and fittings) and flexible lines. Only failures involving the pipe body of steel pipelines are considered here.

A breakdown of the incidents recorded in PARLOC 2001 by failure location is shown in Table 12A.13. The number of incidents of loss of containment that have occurred within 500 m of a platform or a subsea well is almost equal to the number of incidents that have occurred away from it (i.e. mid-line). The higher failure rate in the vicinity of an offshore installation (one to two orders of magnitude higher than mid-line) is due to the effect of increased ship/barge movements in the vicinity and the potential for anchor damage as a result.

The proximity of some sections of the proposed pipeline route to high marine traffic environments could be regarded as similar to the environment in the vicinity of the platform safety zone although it is not strictly comparable.

Table 12A.13 Failure Rate Based on PARLOC 2001 [6]

Region of Pipeline	Operating Experience	No. of Incidents	Failure Rate
Mid-line	297,565 km-years	27	9.1´10^-5 /km/year
Platform safety zone	16,776 years (8,388 km-years)^*	18	1.1´10^-3 /year (2.1´10^-3 /km/year)
Subsea well safety zone	2,586 years (1,293 km-years)^*	6	2.3´10^-3 /year (4.6´10^-3 /km/year)
Total	307,246 km-years*	51	1.66´10^-4 /km/year

* The number of years in the case of platform and subsea well safety zone is multiplied by 0.5km of safety zone to obtain corresponding km-years

The main causes of pipeline failure are summarised in Table 12A.14, based on the causes identified in PARLOC 2001. As discussed earlier, anchor/impact and internal corrosion are the main contributors to pipeline failure.

Table 12A.14 Main Contributors to Subsea Pipeline Failure (PARLOC 2001)

Cause	Platform Safety Zone	Subsea Well Safety Zone	Mid-line	Total
Anchor/Impact	7 (39%)	-	10 (37%)	17 (33%)
Internal corrosion	3 (17%)	4 (67%)	7 (26%)	14 (27%)
Corrosion -others	2 (11%)	-	4 (15%)	6 (12%)
Material defect	4 (22%)	1 (17%)	2 (7%)	7 (14%)
Others	2 (11%)	1 (17%)	4 (15%)	7 (14%)
Total	18	6	27	51

Analysis of Failure Causes

The failure frequency derived from the PARLOC 2001 data is further filtered to take into account the local conditions in Hong Kong.

Corrosion and Material Defect

Based on experience in Europe (Table 12A.14), internal corrosion tends to be a greater problem than external corrosion. For the proposed pipeline, failures due to internal corrosion are expected to be less likely as the gas handled is clean, unlike sour gas transported from wells/platforms which may contain moisture and hydrogen sulphide. Also, it is assumed that the condition of the pipeline will be monitored periodically and maintenance work carried out as necessary.

Failures due to defects in materials and welds are also expected to be lower than implied by the historical record due to technological improvements. The database for PARLOC 2001 dates back to the 1960s; there have been significant improvements in pipe material and welding over the last 10 to 20 years. An 80% reduction is therefore assumed for all forms of corrosion and material defects.

Taking the mid-line data as the most representative for the proposed pipeline, the failure rate is therefore derived as 13 incidents in 297,565 km-years with 80% reduction, giving 8.7´10^-6 /km/year.

The PARLOC 96 report [17] provides a breakdown of loss of containment incidents due to corrosion and material defect for gas pipelines greater than 5km in length. The failure rate for such pipelines is lower at 5.9´10^-6 /km/year (0.7 failures in 119,182 km-years; the km-years are lower because only gas pipelines are considered). This value is considered more appropriate for the proposed pipeline. Unfortunately, a more current value could not be extracted from PARLOC 2001 due to a difference in presentation format of the data. However, a downward trend in failure frequencies is to be expected as technology improves and so 5.9´10^-6 /km/year is considered to be reasonable. Incorporating an 80% reduction again gives a corrosion/defect frequency of 1.18´10^-6 /km/year.

Anchoring/Impact Incidents

There is a significant difference in the failure rate due to anchor/impact incidents for pipelines within 500m of an offshore platform (8.3 ´ 10^-4/ km/ year) as compared to mid-line (3.4´10^-5 /km/year). Further breakdown of incidents based on pipeline diameter is given in Table 12A.15.

Table 12A.15 Frequency of Loss of Containment Incidents due to Anchor/Impact- Breakdown by Pipe Diameter & Location

	Frequency (per km per year)
Location	<10" diameter	10 to 16" diameter	18 to 24" diameter	24 to 40" diameter
Mid-line	1.53´10^-4	2.26´10^-5	1.76´10^-5	1.37´10^-5
Safety zone	6.68´10^-4	1.94´10^-3	4.24´10^-4	8.6´10^-4

It is seen from the above that the failure rate (for mid-line) for larger diameter pipelines is lower by an order of magnitude in comparison to smaller diameter pipelines.

As discussed previously, it is considered that the likelihood of pipeline damage due to anchor/impact incidents may be related to the level of marine activity (this is taken to be a combination of marine traffic and anchoring activity). The frequency of pipeline failure due to these causes has therefore been derived as a function of three levels of marine activity: high, medium and low. Frequency values are based on the large diameters pipes of 24-40” as given in Table 12A.15 since these are the most relevant to the proposed CAPCO pipeline.

For locations with high marine activity, a frequency of 8.6´10^-4 /km/year is adopted. For low marine activity, 1.37´10^-5 /km/year is used. An intermediate value of 10^-4 /km/year is also applied to locations with medium levels of marine activity. This is discussed further in Section 12A.5.3 where alternative calculations based on emergency anchor deployment frequency are also presented for comparison.

These failure frequencies from PARLOC assume minimal protection for the pipeline. The proposed CAPCO pipeline will be provided with rock armour protection over its entire length. To allow for this, the failure frequencies are reduced by appropriate factors as discussed in Section 12A.5.4.

Other Causes

“Other” causes include blockages, procedural errors, pressure surges etc. As with corrosion, improvements in technology and operating practices are expected to reduce this significantly and so a general 90% reduction is assumed for failures due to other causes. This gives a frequency of
1.34´10^-6/km/year (4 cases in 297,565 km-years with 90% reduction).

12A.5.3 Alternate Approach to Anchor Damage Frequency

While international data is commonly applied to infer failure rates for Hong Kong subsea pipelines, in this section an alternative approach is adopted for comparison. This is based on marine traffic incident rates, since such incidents are more likely to result in emergency anchoring. In the first instance, the effects of rock armour protection are neglected to allow these calculations to be compared with historical data from PARLOC. The effects of rock armour protection are then incorporated as described is Section 12A.5.4.

Frequency of Anchor Drop

Emergency Conditions

Vessels may drop anchor due to emergency conditions such as fog, storm, or due to collisions or machinery failure. The likelihood of anchoring due to adverse weather conditions is expected to be low especially for the larger vessels who will determine whether dropping an anchor is the safest option. Furthermore, knowledge of vessel position from onboard navigation systems should prevent inadvertent dropping of an anchor onto a pipeline delineated on the navigation chart.

To estimate the frequency of emergency anchoring, data from the Marine Department of Hong Kong [5] is used. The distribution of incidents of all types (Figure 12A.4) shows that most incidents are concentrated in the harbour regions near Yau Ma Tei, Tsing Yi and Tuen Mun. The region near the proposed pipeline indicates low incident rates, although some areas of Urmston Road have slightly higher incident rates. This is due to the higher traffic density in this area.

Average values of 0.3 appearing in Figure 12A.4 clearly refers to a single incident that occurred during the 3-year period from 2001 to 2003. The size of each cell in Figure 12A.4 is one nautical mile, or approximately 1.852 ´ 1.852 = 3.4 km². A value of 0.3 refers then to an incident frequency rate of 0.3/3.4 @ 0.1 /km²/year. This incident rate is taken to be appropriate for sections of the pipeline away from the busy Urmston Road.

The incident rate for Urmston Road is a little higher and is taken to be 1/nm²/year (Figure 12A.4) or 0.3 /km²/year.

For comparison, the total number of incidents from 1990-2008 in the 1830 km² area of Hong Kong SAR waters was 6491 [16]. This gives a territory average of 0.19 /km²/year.

Figure 12A.4 Average Annual Incident Distribution (2001-2003)

The distribution by types of incidents (Figure 12A.5) shows that most incidents are collisions or contact. Not all incidents will result in an anchor drop. Most collisions, for example, are not serious. It is assumed therefore that only 10% of incidents will result in an emergency anchor drop.

Once the anchor is dropped, it may fall directly on the pipeline causing damage. A greater concern is the possibility of an anchor being dragged across the seabed and into the pipeline. In an emergency situation such as mechanical failure, it is possible that the vessel is still moving when the anchor is deployed. Since anchors can be dragged significant distances, the resulting pipeline contact frequencies tend to be higher compared to a simple anchor drop. In most instances, however, the ship master’s first action will be to reduce speed to near stationary and then drop anchor if necessary. For the purpose of this analysis, it was assumed that 90% of ships drop anchor at near rest (1 knot), while the other 10% drop anchor at 4 knots due to mechanical failure and the uncontrolled advance of the vessel.

Figure 12A.5 Distribution of Incident Types (1990-2008)

The efficiency of an anchor is defined according to its holding capacity:

Holding capacity = anchor weight ´ efficiency

The efficiencies for different classes of anchor [19] are given in Table 12A.16. It is believed that types E and F are common on large commercial vessels.

Table 12A.16 Anchor Efficiency

Class	Efficiency
A	33-55
B	17-25
C	14-26
D	8-15
E	8-11
F	4-6
G	<6

This definition can be used to calculate the drag distance. The work done in dragging an anchor through some distance must be equal to the change in kinetic energy in bringing the ship to rest.

Anchors are designed to penetrate into the seabed for maximum holding capacity. As an anchor is dragged across the seabed, it will begin to penetrate into the mud; the softer the soil, the greater the penetration. Maximum holding capacity is only reached once the maximum penetration depth has been reached i.e. the efficiency is a function of penetration depth. As a conservative approach, the lowest efficiency anchor, type E, is assumed for the calculations. The efficiency is halved again to allow for the varying restraining force with depth. The efficiency is therefore assumed to be 2.

Table 12A.17 gives some drag distances resulting from these calculations. It can be seen that most vessels will drag an anchor for less than about 20m. Ocean-going vessels can drag an anchor over significantly greater distances due to the larger mass and hence kinetic energy of the ship. This class of ship is subdivided into different sizes to reflect the distribution of ships expected along the proposed pipeline route (see Table 12A.8). A 150,000 tonne ship is the largest of ships visiting Hong Kong and this provides the upper limit to the drag distance of about 170m.

Table 12A.17 Drag Distances

Class	Size Range (dwt)	Displacement (tonnes)		Anchor (tonnes)	Drag Distance (m)
Fishing vessel Rivertrade coastal vessels Ocean-going vessels Fast Launches Fast ferries Other	1,500 – 25,000 25,000 – 75,000 75,000 – 100,000	400 1,500 1,500 – 35,000 35,000 – 110,000 110,000 – 150,000 150 150 200	(60%) (35%) (5%)	1 2 2 – 5 5 - 12 12 - 15 0.1 0.5 0.2	7 13 13 – 118 118 – 154 154 – 168 25 5 17

The frequency of anchor drag impact can then be calculated as:

Impact freq =
incident freq (/year/km²) ´ probability of anchor drop ´ drag distance/1000 (1)

where the drag distance is in metres. This gives an impact frequency per km of pipeline per year. If an impact occurs, the damage may not be severe enough to cause containment failure. Based on PARLOC 2001, approximately 22% of anchor /impact incidents result in containment failure when considering all pipe diameters. Larger pipes, however, fail three times less often. This suggests that 7% of incidents would result in a loss of containment.

This approach was applied to each section of the pipeline and to each class of vessel. The marine traffic incident rate was assumed to apply equally to all classes of vessel.

The hydrographic survey [7] identifies seabed conditions as very soft clay. Under these conditions, significant anchor penetration can occur [19]. For example, a 15 tonne anchor can penetrate to 17m, and a 2 tonne anchor can penetrate to 9m. These data apply to high efficiency anchors and less penetration is to be expected for the commonly used types E and F, but nevertheless, it is likely that a wide range of anchors sizes will be able to achieve 3m penetration during emergency anchoring scenarios and hence may interact with the proposed pipeline.

MARAD Study

An alternative to using the incident frequency from Figure 12A.4 is to use data from the MARAD study [18] which reported that the frequency of collisions in Hong Kong waters of ocean-going vessels as 56 per million vessel-km. Since only 70% of incidents are collisions, this value of 56 per million vessel-km was scaled upwards to estimate the number of incidents of all types. 90% of these incidents resulted in only minor damage and so again it is assumed that only 10% will result in an emergency anchor drop. The approach is then similar to that described above for anchor dragging.

The results from this analysis are compared in Figure 12A.6. Also shown are the loss of containment frequencies obtained from PARLOC 2001 for the platform safety zone and mid-line. These are assumed to be representative of areas of high and low marine activity respectively. It can be seen that there is some spread in the predictions. The platform safety zone and mid-line frequencies differ by almost two orders of magnitude but effectively bound most of the other predictions.

Figure 12A.6 Anchor Damage Frequency Based on Marine Incidents

The calculations are broadly consistent with failure frequencies from PARLOC 2001. The frequency obtained from PARLOC 2001 for the mid-line is appropriate for regions of low marine vessel volume. The platform safety zone frequency is regarded as appropriate for the failure frequency in locations of high marine traffic. Some sections have intermediate levels of marine activity and so a frequency of 10^-4 per km-year is adopted for these sections.

Based on the above considerations, the failure frequencies due to anchor impact used in this study are as summarized in Table 12A.18. A low frequency was assigned to the Black Point approach since no vessel movements were observed in this area from the marine radar tracks. Urmston Road was assigned a high frequency owing to the high marine traffic in this section. The remaining sections of pipeline were assigned a medium frequency.

Table 12A.18 Anchor Damage Frequencies used in this Study

Pipeline section	Frequency (/km/year)	Comment
Boundary Section	1´10^-4	Medium marine traffic
Urmston Road	8.6´10^-4	High marine traffic
Black Point West	1´10^-4	Medium marine traffic
Black Point Approach	1.37´10^-5	Low marine traffic

12A.5.4 Pipeline Protection Factors

Many pipelines are trenched to protect them from trawling damage. In the pipeline database in PARLOC 2001, 57% by length of all lines have some degree of protection, either trenching (lowering) or burial (covering) over part or all of their length. Considering large and small diameter lines, the proportion of lines with some degree of protection are 59% by length for lines <16" diameter and 68% for larger diameter lines. It is, however, concluded in the PARLOC report that there have been insufficient incidents to determine a clear relationship between failure rate and the degree of protection.

The loss of containment frequencies given in Table 12A.18 assume minimal protection since they are based on the PARLOC data. The proposed CAPCO pipeline has rock armour protection specified for its whole length. To allow for this, protection factors were applied. Based on the classes of marine vessel found along the proposed route (Table 12A.2), most classes of ship have anchors below 2 tonnes in weight. Only ocean-going vessels have anchors up to 15 tonnes. The rock armour protection along the route is designed to protect against either 3 – 5 tonne anchors (trench types 1 and 2) or 19 tonne anchors (trench type 3). The analysis therefore assigns protection factors for the rock armour and makes a distinction between ocean-going vessels that have large anchors and other types of vessel which have smaller anchors.

Trench types 1 and 2 were assumed to provide 99%protection for anchors smaller than 2 tonnes. These trench types should also offer some protection against larger anchors. For ocean-going vessels, 60% of them have anchors below about 5 tonnes (Table 12A.8) and so trench type 1 should offer reasonable protection against these vessels. 50% protection was assumed for ocean-going vessels. For simplicity, trench type 2 was treated the same way as type 1 and 50% protection was assumed for large anchors. This is a little conservative since trench type 2 is designed to protect anchors up to 5 tonnes.

Trench type 3 (deigned to protect against 19 tonne anchors) was assumed to provide 99% protection for anchors greater than 2 tonnes, and greater protection of 99.9% for small anchors below 2 tonnes.

12A.5.5 Summary of Failure Frequencies for the Proposed CAPCO Pipeline

Based on the above discussions, the failure frequencies used in this study are as summarized in Table 12A.19.

The failure frequencies specified in Table 12A.19 will apply to each of the two pipelines.

Table 12A.19 Summary of Failure Frequencies used in this Study

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
Boundary Section	2	1.18´10^-6	1´10^-4	99	50	1.34´10^-6	3.5´10^-6
Urmston Road	3	1.18´10^-6	8.6´10^-4	99.9	99	1.34´10^-6	4.1´10^-6
Black Point West	2	1.18´10^-6	1´10^-4	99	50	1.34´10^-6	3.5´10^-6
Black Point Approach	1	1.18´10^-6	1.37´10^-5	99	50	1.34´10^-6	2.7´10^-6

* The calculation of total failure frequency takes into account the size distribution of ships (based on 2011 marine traffic) and the protection factors for anchors

12A.5.6 Scenario Development

The outcome of a hazard can be predicted using event tree analysis to investigate the way initiating events could develop. This stage of the analysis involves development of the release cases into discrete hazardous outcomes. The following factors are considered:

· Failure cause;

· Hole size;

· Vessel position and type; and

· Ignition probability.

The probabilities used in the event trees are discussed below.

Failure Cause

Failures due to corrosion and other events are considered separately from failures caused by anchor impact. This is because the hole size distribution is different in each case, as described below. Also, in the event of failure due to anchor impact, the probability of vessel presence is assumed to be higher, as discussed later.

Hole Size Distribution

The data on hole size distribution in PARLOC 2001 is summarised in Table 12A.20.

This data on hole size distribution is clearly limited, particularly for large diameter pipelines. One approach is to compare this distribution with that for onshore pipelines, which include a much larger database of operating data and failure data. For example, the US Gas database [15] is based on 5 million pipeline km-years of operating data as compared to 300,000 km-years in the PARLOC study.

Table 12A.20 Hole Size Distribution from PARLOC 2001

Pipeline size		Hole size (mm)
	Location	0 to 20mm	20 to 80mm	>80mm
2 to 9"	Safety zone Mid line	6 14	3 (1 rupture) 4 (2 ruptures)	2 1 (1 rupture)
10 to 16"	Safety zone Mid line	1 1	1	4 (3 ruptures) 3
>16"	Safety zone Mid line	1 2		2 (2 ruptures)
Total		25 (55%)	8 (18%)	12 (27%)

An analysis of hole size distribution for onshore pipelines as given in the US Gas [15] and European Gas Pipelines databases [20] provides a hole size distribution as given in Table 12A.21.

Table 12A.21 Hole Size Distribution Adopted for Corrosion and Other Failures

Category	Hole Size	Proportion
Rupture (Half Bore)	22" or 558mm	5%
Puncture	4" or 100mm	15%
Hole	2" or 50mm	30%
Leak	<25mm	50%

The above distribution is largely similar to the distribution derived in the PARLOC report [6]. The only difference is the consideration of a small percentage of ruptures. It is a matter of debate whether ruptures could indeed occur although ruptures extending over several metres are reported in the various failure databases.

In this study, it is proposed that the hole size distribution given in Table 12A.21 be adopted for failures caused by corrosion and ‘other’ failures (including material/weld defect). In the case of failures caused by anchor damage, the hole sizes are expected to be larger. The distribution given in Table 12A.22 is adopted.

Table 12A.22 Hole Size Distribution for Anchor Impact

Category	Hole Size	Proportion
Rupture (Full Bore)	Full bore	10%
Major	22" or 558mm (half bore)	20%
Minor	4" or 100mm	70%

Vessel Position

In the case of failures due to corrosion/other events, the probability of a vessel being affected by the leak is calculated based on the traffic volume and the size of the flammable cloud. Dispersion modelling using PHAST [21] is used to obtain the size of the flammable cloud for each hole size scenario and four weather scenarios covering atmospheric stability classes B, D and F. Once the cloud size is known, the probability that a passing marine vessel will travel through this area within a given time can be calculated. A time period of 30 minutes is used since it is assumed that if a leak occurs, warnings will be issued to all shipping within 30 minutes. Further details on the dispersion modelling are given in Section 12A.6.

In the case of failures due to anchor impact, the following two scenarios are considered:

· “Vessels in vicinity” - the vessel that caused damage to the pipeline (due to anchoring) is still in the vicinity of the incident zone. The probability of this is assumed to be 0.3; and

· “Passing vessels” - ships approach or pass the scene of the incident following a failure. In this case, the probability of a vessel passing through the plume is calculated using the same method as for a corrosion failure; i.e. based on cloud size and traffic volume.

Event trees showing these scenarios are given in Figures 12A.7 and 12A.8. If a vessel passes through the flammable gas cloud, a distinction is further made between vessels passing directly over the release area and vessels passing through other parts of the cloud. This is discussed further in the following section.

Figure 12A.7 Event Tree for External Damage from Anchors

Figure 12A.8 Event Tree for Spontaneous Failures

Corrosion or Other failure	Passing vessel		Release area		Ignition		Outcome
		Yes		Yes		Yes	Flash fire
		No		No		No	No effect
						Yes	Flash fire
						No	No effect
							No effect

It is assumed that at most, only one vessel will be affected by a pipeline failure. Once the flammable plume is ignited, the resulting fire will be visible and other ships will naturally avoid the area.

Vessel Type

The categorisation of vessel types follows those identified from the radar tracks (Table 12A.2), namely:

· Fishing vessels and small crafts;

· Rivertrade coastal vessels;

· Ocean-going vessels;

· Fast Launches;

· Fast ferries;

· ‘Others’ (assumed to be small vessels)

The relative proportion of the different vessel types will vary along the pipeline route, as indicated in Table 12A.4.

Ignition Probability

Ignition of the release is expected only from passing ships or ships in the vicinity. Ignition probabilities derived from offshore pipeline releases in the vicinity of an offshore platform are given in Table 12A.23 [22]. Similar values are adopted in this study, as given in Table 12A.24.

Table 12A.23 Pipeline Hydrocarbon Release Ignition Probability in Platform Vicinity [23]

Typical Ignition Probability (integrated platform)
Location of release	Massive gas release (>20 kg/s)	Major gas release (2-20 kg/s)	Minor gas release(<2 kg/s)
Riser above sea*	0.168	0.026	0.005
Subsea	0.443	0.13	0.043
Typical Ignition Probability (bridge linked platform)
Location of release	Massive gas release (>20 kg/s)	Major gas release (2-20 kg/s)	Minor gas release (<2 kg/s)
Riser above sea*	0.078	0.013	0.002
Subsea	0.14	0.051	0.002

* 'Riser above sea' refers to pipeline riser portion that is above sea level

Table 12A.24 Ignition Probabilities used in Current Study

Release Case	Ignition Probability
	Passing Vessels^(1)	Vessels in Vicinity^(2)
<25mm	0.01	n/a
50mm	0.05	n/a
100mm	0.1	0.15
Half bore	0.2	0.3
Full bore	0.3	0.4
1. Values applied to passing vessels for all types of incidents, i.e. corrosion, others and anchor impact. 2. Values applied only to scenarios where the vessel causing pipeline damage due to anchor impact is still in the vicinity.

12A.5.7 Second Phase Construction Activities

The second pipeline may be constructed concurrently with the first, or two years later in 2014. From a risk perspective, construction of the pipelines at different times may present an increase in risk due to construction activities from the second pipeline impacting on the first operational pipeline.

The project has taken this into consideration with the following safeguards:

· The two pipelines will be located 100 m apart;

· The pipelines are planned to run parallel without any crossing points and without crossing any other existing pipelines;

· Strict procedures for construction activities involving surveys, confirmation of location using Global Positioning Systems and the demarcation of alignment using marker buoys;

· The pipelines are protected against damage from dredging by rock protection along their full length; and

· Design for the second pipeline will be taken into account during construction of the first pipeline. Critical areas, such as the shore approaches will be pre-constructed in parallel for the two pipelines as far as practicable.

The Gas Production & Supply Code of Practice [24] provides a practical guidance in respect of the requirements of the Gas Safety Ordinance Cap 51 and the Gas Safety (Gas Supply) Regulations. Article 23A of these regulations requires that:

· No person shall carry out, or permit to be carried out, any works in the vicinity of a gas pipe unless he or the person carrying out the works has before commencing the works, taken all reasonable steps to ascertain the location and position of the gas pipe; and

· A person who carries out, or who permits to be carried out, any works in the vicinity of a gas pipe shall ensure that all reasonable measures are taken to protect the gas pipe from damage arising out of the works that would be likely to prejudice safety.

Work, ‘in the vicinity’ is defined according to Table 12A.25 and these guidelines apply to both onshore and subsea pipelines. Although many of the activities listed are not directly relevant to the proposed CAPCO pipeline, Table 12A.25 serves to indicate typical effects distances for different types of work and when special precautions are warranted. A separation distance of 100m is very significant compared to distances listed in Table 12A.25. This, combined with the strict procedures that will be followed and the pipeline protection provided, suggests that the likelihood of damage to the first operational pipeline from construction activities during phase 2 will be very low. This is therefore not considered further in this study.

Table 12A.25 Works in the Vicinity of Gas Pipes

Type of Work	Distance
Trench or other excavation up to 1.5m in depth in stable ground	10m
Trench or other excavation over 1.5m and up to 5m in depth	15m
Trench or other excavation in stable ground over 5m in depth	20m
Welding or hot works near exposed gas pipes or above ground installations	10m
Piling, percussion moling or pipe bursting	15m
Works near high pressure pipelines	20m
Ground investigation and any kind of drilling or core sampling	30m
Use of explosives	60m

The construction activities may also increase risk by increasing the population within the vicinity of the operational pipeline. Any incident affecting the operational pipeline may impact on the construction workers and lead to a higher number of fatalities.

The hazard effects exceed 100m only for the half bore rupture case in weather condition 7D (refer to Consequence Analysis). This scenario has a hazard range of 115 m. Geometric considerations (Figure 12A.9) imply that a leak from a section of pipeline just 114m long has the potential to reach the workers 100 m away.

Figure 12A.9 Construction Workers’ Proximity to Pipeline

An incident at the operating pipeline may be caused by internal failure or external impact.

Internal Failure

The failure frequency (Table 12A.19) for internal failure is 2.52´10^-6 /km/year ([1]) . The frequency of events from the operational pipeline impacting on construction workers at the second pipeline may be estimated from:

/year

Where the factor of 114/1000 arises from the geometric considerations and the fact that an incident must occur within a 114m length section of the pipeline to affect the workers. 0.695 refers to the probability of weather category 7D and a factor of 1/6 is applied to approximate the probability of the wind blowing towards the construction workers. The factor of 0.05 corresponds to the probability of the leak size being half bore rupture for internal failures and 0.2 corresponds to the ignition probability for this sized leak.

External Impact

The highest frequency (ie. Urmston Road) for external damage is 1.58´10^-6 /km/year (calculated from Table 12A.19 ([2]) ). The frequency of incidents affecting workers on the second pipeline may be estimated in a similar manner as above:

/year

Where the probability of half bore rupture is taken to be 0.2 for external damage and the factor of (1-0.3´0.3) represents the probability that the vessel causing the damage did not itself ignite the release (0.3 for the vessel that caused the damage is still present and 0.3 for the ignition probability). Other terms are the same as in the internal failure case.

Combining the internal and external failure scenarios gives a total frequency of 1.09´10^-9 per year that the construction workers will be affected by an incident at the operational pipeline. Construction, however, is expected to take 11 months and will take place for 12 hours per day, except for the Urmston Road section where construction will be round-the-clock. Taking into account these presence factors reduces the frequency below the 10^-9 /year and therefore construction related risks are not considered further in the assessment.

12A.6 Consequence Analysis

12A.6.1 Overview

In the event of loss of containment, the gas will bubble to the surface of the sea and then disperse. If it comes in contact with an ignition source, most likely from a passing marine vessel, it could lead to a flash fire which will propagate through the cloud to the point of release and result in a gas fire above the water surface.

If a marine vessel passes into a plume of gas and ignites it, then there is the possibility of fatalities on that ship due to the flash fire. If a vessel passes through the ‘release area’ of the release, the vessel will likely be affected also by the ensuing fire and the consequences will be more severe. If the release gets ignited, it is presumed that no further ships will be involved because the fire will be visible and other ships will naturally avoid the area. In other words, it is assumed that at most, only one ship will be affected.

Further details are described in the following paragraphs.

12A.6.2 Source Term Modelling

The release rate is estimated based on standard equations for discharge through an orifice. The empirical correlation developed by Bell and modified by Wilson [23] is adopted. A maximum operating pressure of 100 barg is assumed.

The results are presented in Figure 12A.10. For holes with equivalent diameter smaller than about 100 mm, the discharge rate diminishes rather slowly because of the large inventory in each pipeline (about 1,380 tonnes). For half and full bore failures, the discharge rate diminishes more quickly over a period of about 30-60 minutes.

Figure 12A.10 Variation of Release Rate with Time

12A.6.3 Dispersion Modelling for Subsea Releases

In the event of a release from the subsea pipeline, the gas jet is expected to lose momentum and bubble to the surface. The simplest form of modelling applied to subsea releases is to assume that the dispersing bubble plume (driven by gas buoyancy) can be represented by a cone of fixed angle (Figure 12A.11) [23]. The typical cone angle is between 10 to 12°. However, Billeter and Fannelop [23] suggested that the 'release area' (where bubbles break through the surface) is about twice the diameter of the bubble plume. Hence, an angle of 23° was recommended and is used in this study.

Based on Figure 12A.11, the water depth is between 2-5m for much of the proposed pipeline route, increasing to 20m in Urmston Road. For this range of water depths, the cone model predicts the ‘release area’ to be in the range of 0.8 to 8 m diameter.

Figure 12A.11 Simple Cone Model for Subsea Dispersion

12A.6.4 Dispersion above Sea Level

The gas will begin to disperse into the atmosphere upon reaching the sea surface. The distance to which the flammable envelope of gas extends will depend on ambient conditions such as wind speed and atmospheric stability as well as source conditions. The extent of the flammable region is taken as the distance to 0.85 LFL (Lower Flammable Limit).

Conditions at the source such as momentum and buoyancy are important. At shallower depths and high release rates, the gas will have a large momentum at the sea surface resulting in a plume extending rapidly upwards into the atmosphere. For smaller releases or release from deeper water, the gas will lose all momentum by the time it reaches the sea surface resulting in a plume of greater horizontal extent. Dimensional analysis using the Froude number [23] suggests that momentum and buoyancy are both important over most release scenarios considered in the current study. Only full bore ruptures in shallow water result in a momentum dominated jet release.

The above sea dispersion was modelled using PHAST [21]. Based on the above discussion, to achieve realistic simulations it is important to give due consideration to the momentum and buoyancy of the source. The gas was assumed to gain heat from the sea water, during transport and following a release. The gas was therefore assumed to be released at 20°C and 100barg. Being lighter than air, natural gas lifts away from the sea surface under all atmospheric conditions.

The cone model is believed to be a reasonable approach for estimating the ‘release area’ for small to moderate releases. The worst scenario is deep water, which produces a large ‘release area’ and hence low efflux momentum for a given mass release rate. The deepest water case of 20m was therefore chosen for analysis. A low momentum gives a lower plume rise and hence a larger hazardous area near the sea surface. The cone model, however, has not been validated for massive releases such as would occur in a half bore or full bore rupture. To err on the cautious side, a larger ‘release area’ was assumed for massive releases. The diameter of the release area was increased by 50% for half bore rupture and by 100% for full bore rupture scenarios. This lowers the source momentum and gives conservative results.

PHAST was used to model the plume dispersion as an area source on the surface of the ocean. The mass release rate, the release velocity and temperature were specified and the release was assumed to be vertical. The surface roughness parameter was assumed to be 0.043, a value appropriate for dispersion over water. Even though the release is a transient, particularly for the large release scenarios, the time constant for the release is still longer than the dispersion time scale. The modelling therefore assumed a steady release of gas at the maximum (initial) release rate. Again, this is conservative. Simulations were performed for atmospheric stability classes of B, D and F to cover the range of meteorological conditions expected. Given that the plume in all cases lifted away from the surface due to buoyancy, the length of the plume was taken to be the maximum extent of the plume in the windward direction up to the ship height which is assumed to be a maximum of 50m.

The relative occurrence of weather conditions 2F, 3D, 7D and 2.5B were taken to be 0.083, 0.070, 0.695 and 0.152 respectively to match conditions measured at the Sha Chau meteorology station (Table 12B.6). This is based on the average of the most recent 5 years of meteorological data from 2004 to 2008.

12A.6.5 Impact Assessment

Impact on Population on Marine Vessels

The hazardous distance was taken to be the distance to 0.85 LFL as discussed above. It was assumed that ships would be at risk for 30 minutes before warnings could be issued to advice vessels to avoid the area. Knowing the marine vessel traffic (in ships per day per km of pipeline), the probability that a passing ship will cross through the flammable plume during this 30 minutes is calculated as:

Prob. = (3)

If a marine vessel comes into contact with the flammable plume and causes ignition, the resulting flash fire may lead to fatalities depending on the type of ship. Small open vessels such as fishing boats are expected to provide less protection to their occupants. Large ocean-going vessels will provide better protection. Fatality factors are therefore applied to each class of vessel to take into account the protection offered by the vessel. These take into consideration:

· The proportion of the passengers likely to be on deck or in interior compartments.

· The materials of construction of the vessel and the likelihood of secondary fires.

· The size of the vessel and hence the likelihood that it can be completely engulfed in a flammable gas cloud.

· The speed of the vessel and hence its exposure time to the gas cloud.

· The ability of gas to penetrate into the vessel and achieve a flammable mixture.

Considering fast ferries; they are air conditioned and travel at high speeds in excess of 30 knots (15m/s). If the occupants are to be affected by a flash fire, gas must penetrate into the interior of the vessel, achieve a flammable mixture and ignite. The time to transit the largest gas cloud of 95m is of the order of 7 seconds. Assuming typical air ventilation rates of 6 to 10 volume changes per hour, a time constant for changes in gas concentration within a ferry can be derived as 6 to 10 minutes. This implies that it would take several minutes for the gas concentration within a ferry to respond to changes in concentration in the ambient air. Given that the exposure time is mere seconds, it becomes apparent that it is very difficult to achieve a flammable mixture of gas within a ferry. Based on these considerations, the fatalities assumed in the current study for fast ferries and other vessels are as given in Table 12A.26.

If a ship enters the ‘release area’ and ignites the gas cloud, the vessel is more likely to be caught in the ensuing fire. This is assumed to result in more severe consequences with potential for 100% fatality of occupants. The probability of this is calculated using a similar equation as above (Equation 3) but replacing the cloud size with the release area diameter.

Table 12A.26 Fatality Probabilities

Class	Fatality
	‘Release area’	‘Cloud area’
Fishing vessels Rivertrade coastal vessels Ocean-going vessels Fast launches Fast ferries Others	1 1 1 1 1 1	0.9 0.3 0.1 0.9 0.3 0.9

If the failure is caused by corrosion, a passing ship may pass through the flammable plume or release area with a probability given by Equation 3. If the failure is caused by third party damage, then two scenarios are considered as mentioned in Section 12A.5. The vessel that caused the incident may still be in the area and may ignite the plume, or if this vessel is no longer present, a passing ship may pass through the plume. The probability that the vessel causing the incident is still present is assumed to be 0.3 and this is assumed to result in 100% fatality.

The analysis limits the number of ships involved to one. It is assumed that once the plume is ignited, other ships will avoid the area.

12A.6.6 Consequence Results

Hazard distances are determined from the dispersion modelling. Given that natural gas is buoyant and tends to lift away from the sea surface, the hazard distance is defined as the gas cloud width near sea level where ignition is possible by passing ships. Specifically, the hazard distance is taken to be the maximum width within 50m of the sea surface (Figure 12A.12). Based on this, the hazard distances obtained from dispersion modelling are summarised in Table 12A.27.

Figure 12A.12 Hazard Distance

Table 12A.27 Hazard Distances for Gas Cloud Dispersion

Hole Size (mm)	End Point Criteria	Marine Vessel Hazard Distance (m)*
		Weather conditions
		2F	3D	7D	2.5B
Full bore	0.85LFL	56	57	82	63
Half bore	0.85LFL	53	53	115	56
100	0.85LFL	59	56	80	43
50	0.85LFL	35	37	52	32
25	0.85LFL	22	27	33	24

* Distances quoted are those for releases from 20m water depth. Deep water releases give higher hazard distances and were used in the assessment as conservative upper limits.

12A.7 Risk Summation

The frequencies and consequences of the various outcomes of the numerous accident scenarios are integrated at this stage, to give measures of the societal risk (FN curves and Potential Loss of Life) and individual risk.

Risk results are compared with the criteria for acceptability as laid down in the Hong Kong Planning Standards and Guidelines, chapter 12 [25] and also in Annex 4 of the Technical Memorandum of EIAO. However, these risk guidelines cannot be applied directly for transport operations (such as pipelines). Since transport operations extend over several kilometres and communities, they cannot be equated with risks from fixed installations (such as an LPG plant, refinery or a petrochemical plant) which have a defined impact zone. As a result, a pipeline of 1 km length is considered as equivalent to a fixed installation for the application of risk criteria. This approach is adopted internationally [26] and was adopted by the consultant in similar studies for onshore and offshore high pressure gas pipelines. Based on this approach, the results are presented on a per-kilometre basis for each section of the pipeline.

The individual risk (IR) criterion for a potentially hazardous installation specifies that the risk of fatality to an offshore individual should not exceed
1´10^-5 per year. It is generally accepted that the same IR criteria should also apply for transport operations.

Risk results are presented in the Section 12 of the EIA Report.

12A.8 References

[1] ERM-Hong Kong, Black Point Gas Supply Project, Project Profile, June 2009.

[2] WorleyParsons Resources and Energy, Drawing HKLNG-WPL-00-PIP-PL-009, revision 0, Gas Pipeline Trenching and Protection, 2008.

[3] BMT Asia Pacific Ltd, Marine Impact Assessment for Black Point & Sokos islands LNG Receiving Terminal & Associated Facilities, Pipeline Issues, Working Paper #3, Issue 5, Apr 2006.

[4] Marine Department Port Statistics, 2004 http://www.info.gov.hk/mardep.index.htm

[5] Marine Department, Marine Traffic Risk Assessment for Hong Kong Waters (MARA Study), March 2004.

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

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

[8] ERM, Environmental and Risk Assessment Study for a Liquefied Natural Gas (LNG) Terminal in the Hong Kong SAR, April 2005.

[9] Marine Department, Hong Kong Government SAR, Port and Maritime Statistics, Passenger Arrivals and Departures by Route, January – December 2008. www.mardep.gov.hk

[10] Personal Communication with BMT.

[11] UKAEA, Major Hazard Incident Database (MHIDAS) Silver Platter.

[12] Institution of Chemical Engineers UK, The Accident Database, Version 2.01.

[13] National Transportation Safety Board, Natural gas Pipeline Rupture and Fire During Dredging of Tiger Pass, Lousiana, October 23, 1998.

[14] National Research Council, Improving Safety of Marine Pipelines, 1994.

[15] PRC International American Gas Association, Analysis of DOT Reportable Incidents for Gas Transmission and Gathering Pipelines – January 1, 1985 Through December 31, 1994 Keifner & Associate Inc., 1996.

[16] Marine Department, Hong Kong Government, Statistics on Marine Accidents, 1990-2008, www.mardep.gov.hk.

[17] Health and Safety Executive UK, PARLOC 96: The Update of Loss of Containment Data for Offshore Pipelines, 1998

[18] Marine Department, The MARAD Strategy Report Comprehensive Study on Marine Activities Associated Risk Assessment and Development of a Future Strategy for the Optimum Usage of Hong Kong Waters, 1997.

[19] Vryhof, Vryhof Anchor Manual, www.vryhof.com, 2005.

[20] European Gas Pipeline Incident Data Group 3^rd EGIG-Report 1970-1997.

[21] DnV Technica, PHAST Release Notes, DnV Technica Inc., Temecula, CA., 1993.

[22] Centre of Chemical Process Safety, Guidelines for Use of Vapour Cloud Dispersion Models, 1996.

[23] P J Rew, P Gallagher, D M Deaves, Dispersion of Subsea Releases: Review of Prediction Methodologies, Health and Safety Executive, 1995.

[24] The Gas Authority, Gas Production & Supply Code of Practice, GPS 01, 1^st Edition, The Government of the Hong Kong Special Administrative Region, July 1997.

[25] Planning Department, Hong Kong Planning Standards & Guidelines Chapter 12, Hong Kong Risk Guidelines for Potential Hazardous Installation, 1992.

[26] M J Pikaar, M A Seaman, A Review of Risk Control, Ministerie VROM (1995/27A), 1995.

---

([1]) Corrosion Frequency (/km/year) + Frequency of Other Causes (/km/year) = 1.18x10-6 + 1.34x10-6 = 2.52 x10-6 /km/year

([2]) Total Failure Frequency (/km/year) - Internal failure Frequency = 4.1x10-6 - 2.52x10-6 = 1.58x10-6 /km/year