Table of Contents

9. hazard to life.. 9-1

9.1 Introduction. 9-1

9.2 Yau Kom Tau Water Treatment Works Location and Operations. 9-3

9.3 Tsuen Wan Road Upgrading Project 9-6

9.4 Meteorological Conditions and Population Data. 9-6

9.5 Hazard Identification. 9-23

9.6 Consequence Analysis. 9-35

9.7 Rationalisation of Chlorine Release scenarios And estimation of Scenario Frequencies. 9-45

9.8 Quantitative Risk Assessment 9-54

9.9 Conclusions and Recommendations. 9-61

9.10 References. 9-62

List of Tables

Table 9-1 Plant Operating Data for Yau Kom Tau WTW

Table 9-2 Meteorological Data Used in the Study

Table 9-3 Population Data Assumed for Different Scenarios

Table 9-4 Physiological Effects of Chlorine (IChemE, 1989)

Table 9-5 Summary of Chlorine Release Quantities

Table 9-6 Summary of Injuries

Table 9-7 Summary of Fatalities

Table 9-8 Injuries to Fatality Ratio

Table 9-9 Location of Incidents

Table 9-10 Primary Causes of Incidents

Table 9-11 Hazards Identified during HAZOP Study for Yau Kom Tau WTW

Table 9-12 Failure Modes of Contain and Absorb System Identified at HAZOP Study

Table 9-13 Characterisation of Chlorine Release Scenarios

Table 9-14 Chlorine Toxicity Relationship

Table 9-15 Effective Outdoors Probability of Fatality

Table 9-16 Summary of Source Term Modelling Results for Yau Kom Tau WTW

Table 9-17 Summary of Wind Tunnel Tests Results for Yau Kom Tau WTW

Table 9-18 Summary of CFD Modelling Results

Table 9-19 Summary of DRIFT Modelling Results

Table 9-20 Chlorine Cloud Heights

Table 9-21 Rationalisation of Chlorine Release Scenarios

Table 9-22 Release Scenarios Included in YKT WTW QRA

Table 9-23 Release Scenarios Categorised by Leak Quantity

Table 9-24 Base Failure Rate Data

Table 9-25 Event Frequencies - Base Case of WTW QRA

Table 9-26 Frequencies (times per year) of Chlorine Release Events used in the Present Study

Table 9-27 Breakdown of the Case 4 FN Data by Release Scenario

Table 9-28 Total PLL Values

Table 9-29 Breakdown of Case 4 and Case 5 PLL by Population Area

Table 9-1 Plant Operating Data for Yau Kom Tau WTW

Table 9-2 Meteorological Data Used in the Study

Table 9-3 Population Data Assumed for Different Scenarios

Table 9-4 Physiological Effects of Chlorine (IChemE, 1989)

Table 9-5 Summary of Chlorine Release Quantities

Table 9-6 Summary of Injuries

Table 9-7 Summary of Fatalities

Table 9-8 Injuries to Fatality Ratio

Table 9-10 Primary Causes of Incidents

Table 9-11 Hazards Identified during HAZOP Study for Yau Kom Tau WTW

Table 9-12 Failure Modes of Contain and Absorb System Identified at HAZOP Study

Table 9-13 Characterisation of Chlorine Release Scenarios

Table 9-14 Chlorine Toxicity Relationship

Table 9-15 Effective Outdoors Probability of Fatality

Table 9-16 Summary of Source Term Modelling Results for Yau Kom Tau WTW

Table 9-17 Summary of Wind Tunnel Tests Results for Yau Kom Tau WTW

Table 9-18 Summary of CFD Modelling Results

Table 9-19 Summary of DRIFT Modelling Results

Table 9-20 Chlorine Cloud Heights

Table 9-21 Rationalisation of Chlorine Release Scenarios

Table 9-22 Release Scenarios Included in YKT WTW QRA

Table 9-23 Release Scenarios Categorised by Leak Quantity

Table 9-24 Base Failure Rate Data

Table 9-25 Event Frequencies – Base Case of WTW QRA

Table 9-26 Frequencies (time per year) of Chlorine Release Events Used in the Present Study

Table 9-27 Breakdown of the Case 4 FN Data by Release Scenario

Table 9-28 Total PPL Values

Table 9-29 Breakdown of Case 4 and Case 5 PLL by Population Area

List of Figures

Figure 9-1 Site Layout of Yau Kom Tau Water Treatment Works

Figure 9-2 Site Location of Yau Kom Tau WTW

Figure 9-3 Tsuen Wan Road Upgrading Project and theeh Developments Planned in its Vicinity

Figure 9-4 Population Map

Figure 9-5 Scheme of Traffic Lanes included in the Traffic Impact Assessment for the Project

Figure 9-6 FN Curves for Background General Population

Figure 9-7 FN Curves for Updated General Population

Figure 9-8 Individual Risk Levels

List of Appendices

Appendix 9-A DRIFT Flat Terrain Dispersion Modelling

Appendix 9-B Wind Tunnel Test Results

Appendix 9-C CFD Modelling Results for Sha Tin WTW and Tai Po Tau WTW

Appendix 9-D Application of Dispersion Modelling Results in QRA

Appendix 9-E Chlorine Cloud Height Predictions

Appendix 9-F Modelling of Escape from Chlorine Cloud

Appendix 9-G Seismic Hazard Assessment

Appendix 9-H Aircraft Crash Frequency Assessment

Appendix 9-I Frequency Estimation

Appendix 9-J Traffic Projections from the Project Traffic Impact Assessment

Appendix 9-K Probability of Fatality for the Indoor Population

9. hazard to life

9.1 Introduction

Background

9.1.1 Hazard to life assessment has been conducted, in association with Environmental Resources Management (ERM), during the construction and operational phases of the Project following the criteria for evaluating hazard to life as stated in Annexes 4 and 22 of the TM (Hong Kong Risk Guidelines). 

9.1.2 As stipulated in Section 3.4.7.1 of the EIA Study Brief No. ESB-141/2006, the hazard assessment will include the following:

(i) Identification of all credible and applicable hazardous scenarios requiring Quantitative Risk Assessment (QRA) associated with the chlorine storage and on-site transport in the Yau Kom Tau Water Treatment Works (YKTWTW);

(ii) Execution of a QRA expressing population risks in both individual and societal terms;

(iii) Comparison of individual and societal risks with Hong Kong Risk Guidelines (Annex 4 of TM); and

(iv) Identification and assessment of practicable and cost-effective risk mitigation measures.

9.1.3 Hazard assessment that is consistent with the previous QRA for YKTWTW (ERM, 2001) has been adopted. As suggested in the EIA Study Brief, the similar methodology of the earlier QRA for Route 8 (formerly Route 16) which also crosses Consultation Zones of Water Treatment Works is also taken into account. This Report outlines our methodology and presents the results of the Hazard to Life Assessment.

Previous Studies

QRA for YKT WTW

9.1.4 The hazard assessment for Yau Kom Tau Water Treatment Works (ERM, 2001) was conducted on behalf of the Water Supplies Department as part of a larger project: “Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works” (ERM, 1997a).

9.1.5 The study included detailed Hazard Identification, Consequence Analysis (which involved physical modelling of the dispersion of chlorine clouds in a wind tunnel as well as numerical modelling using CFD and dispersion models), Event Frequency Estimation and Quantitative Risk Assessment exercises. The results of the QRA were expressed in terms of individual and societal risk and compared with the Hong Kong Risk Guidelines as expressed in TM (EPD, 1997). A GIS-based software application GISRisk, custom-built for the study is described in Section 8.1.

EIA for Route 8 (formerly Route 16)

9.1.6 EIA for that route was conducted by Scott Wilson (Hong Kong) Ltd and Parsons Brinckerhoff (Asia) Ltd in joint venture, in association with ERM Hong Kong (responsible for the Hazard to Life Assessment) and MVA Asia.

9.1.7 Two separate road alignments were considered (Scott Wilson, 1998a,b), both passing in close vicinity of a number of Water Treatment Works (WTWs) which were considered the principal source of hazard for both the construction and operational phase of Route 16. Approach to the risk assessment related to potential chlorine spills at WTWs was very similar to the detailed studies of eight WTWs for the Water Supplies Department (ERM, 1997a).

Technical and Geographical Scope of the Study

9.1.8 This study is based on the results of QRA for the Yau Kom Tau WTW study (ERM, 2001) and is similar in scope. However the previous input data to the risk modelling have been updated and the modelling repeated in order to take into account the proposed road works under the project as well as the latest traffic and population projections for the area. The study is concerned only with the hazards posed by the storage, handling and on-site transport of chlorine at the YKTWTW.

9.1.9 The geographical scope includes all the areas that, based on the results of the previous study could be adversely affected by a potential chlorine spill at YKTWTW (refer to Figure 9-2). It should be noted that while such areas may extend beyond the 1 km WTW Consultation Zone, they do not cover the whole site of the Tsuen Wan Road Upgrading Project, of which only the western part is included in this assessment.

Approach

9.1.10 The approach to the study follows that of the WTW QRA (ERM, 1997ab, 2001) but takes into account the details of the Tsuen Wan Road upgrading and the latest traffic and population projections for the area. The resulting societal and individual risk levels are evaluated for both the construction and operational phases of the Tsuen Wan Rd upgrading project. The background scenario, assuming that no Tsuen Wan Rd upgrading is performed is also evaluated. The societal risk is expressed in terms of PLL (potential loss of life) values and FN curves, showing a cumulative frequency (F) of various levels of fatalities (N). They can be directly compared with Hong Kong Risk Guidelines (Annex 4 of TM) as well as the Government regulations concerning Potentially Hazardous Installation (PHIs). The individual risk levels express a risk for a hypothetical individual spending 100% of their time outdoors at a specific location.

9.1.11 Four separate scenarios are assessed in this study:

· Case 1: 2006 background conditions;

· Case 2: 2030 background conditions (no Tsuen Wan Road upgrading project);

· Case 3: Construction Phase of the project; and

· Case 4: Operational Phase, 2030 projections of traffic and population.

· Societal risk results for the Tsuen Wan Rd population only, for the construction and operational phases of the project are also presented.

Structure of this Chapter

9.1.12 The remaining part of the report is divided into the following sections:

· Section 9.2 provides a description of Yau Kom Tau WTW operations;

· Section 9.3 describes the proposed Tsuen Wan Rd Upgrading Project;

· Section 9.4 describes the meteorological and population data used in the assessment;

· Section 9.5 deals with hazard identification;

· Section 9.6 presents the results of the consequence analysis;

· Section 9.7 concerns the chlorine release scenarios and their frequencies

· Section 9.8 describes the results of the QRA in terms of the societal and individual risk levels;

· Section 9.9 presents the conclusions and recommendations of the study;

· Section 9.10 includes a list of references; and

· Appendices 9-A to 9-K include supporting information related to different parts of this report.

9.2 Yau Kom Tau Water Treatment Works Location and Operations

Location

9.2.1 Yau Kom Tau WTW is situated about 1 km west of Tsuen Wan. The WTW is located on a level platform cut into the hillside immediately above the Tuen Mun Road. To the east and south-west of the WTW lie the high rise residential developments such as Greenview Terrace, Greenview Court, Bayview Garden , Belvedere Garden, Hanley Villa and Sunny Villa. Access to the site is via Castle Peak Road leading up to Yau Kom Tau village. The Tsuen Wan Bypass, Widening of Tsuen Wan Road between Tsuen Tsing Interchange and Kwai Tsing Interchange and Associated Junction Improvement Works project site is located to the east and south-east of YKT WTW site, with its western boundary about 1 km to the east from the WTW.

9.2.2 The site location and the site layout plan are shown in Figures 9-1 and 9-2.

Operations

Delivery, Storage and Handling of Chlorine

9.2.3 Chlorine is delivered to Yau Kom Tau WTW in batches of up to 6 x 1 tonne drums. Unloading takes place inside the chlorine store with the doors closed in a designated truck unloading bay. The movement of drums within the storage area and unloading bay is carried out using one of two hoist/monorail systems with purpose-built lifting beams. Each system covers two of the four rows of drums within the chlorine store. Prior to usage, the drums are stored on cradles within the chlorine storage area.

9.2.4 Table 9-1 provides the basic plant operating data for Yau Kom Tau WTW according to the latest information received from WSD. It can be seen that the average chlorine stock level, chlorine dosage and estimated chlorine usage are now significantly lower than those reported in ERM (2001). Nevertheless, since due to increased demand the chlorine usage at WTW can increase in the future, the value of 1187 tonnes per year, based on the maximum plant capacity is used in the modelling.

Table 9-1 Plant Operating Data for Yau Kom Tau WTW

Items	Units	1996	2007	Assumed in modelling
Type of container in use	-	1 tonne	1 tonne	1 tonne
Design capacity of plant	Mld	250	250	-
Plant throughput (average)	Mld	120	120	-
Chlorine stock level (average)	tonnes	45	32	64
Chlorine dosage (average)	mg/l	13	2
Estimated chlorine usage	tonnes per year	350	100	1187
Chlorine storage capacity (including duty and standby containers)	tonnes	64	64	64
Number of draw-off units	-	2	2	2
Chlorine container lifting device	-	Hoist/Monorail	Hoist/Monorail	Hoist/Monorail
Scrubber capacity	tonnes chlorine	1	1	1
Distance travelled by chlorine truck along site access road	km	0.2	0.2	0.2

Chlorination System

9.2.5 The draw-off units comprise pairs of drums, one drum on duty, the other serving as standby. The number of drums on line is subject to the raw water quality. Changeover panels automatically change the draw-off from duty to standby when the draw-off pressure falls below a preset level. The changeover is achieved by electrically actuated isolating valves provided for each drum.

9.2.6 Liquid chlorine is drawn from the 1 tonne drums and is passed to the evaporators for conversion into the gaseous state. The gaseous chlorine passes through the chlorinators and dissolves in water at ejectors to form a chlorinated water solution for feeding into the bulk water stream during the treatment process.

9.2.7 The chlorinators are of vacuum venturi type and thus the section of line between the regulator and the chlorinator is at negative pressure. Double non return valves are provided within the chlorinator units.

Ventilation System

9.2.8 The chlorine drum storage/evaporator area and chlorinator rooms are normally ventilated via a supply of fresh air at high level and extracted at low level. On detection of chlorine levels above 3 ppm there are visual and audible alarms and the ventilation extract fans stop.

Chlorine Scrubbing System

9.2.9 An emergency chlorine scrubbing system is installed to remove any leaked chlorine in the chlorine handling and storage areas. The system is a packed tower utilising sodium hydroxide as the neutralising agent. The plant and equipment are installed in a separate scrubber room.

9.2.10 On detection of a chlorine concentration of 3 ppm or above in the chlorine handling or storage areas, the scrubbing system will activate automatically. The air/chlorine mixture in the affected areas is drawn into the scrubber by the scrubber fan via ducting connected to the normal ventilation system.

9.2.11 The scrubber system is normally set at auto control mode to recycle air back to the drum storage area. However, the treated air may be discharged to atmosphere at roof level when the chlorine concentration is below 3 ppm. The control for recycling or discharging air to atmosphere is effected by means of a pair of electrically operated change-over dampers which can also be manually controlled from the local control panel. A continuous chlorine monitor is installed at a point downstream of the packed tower and upstream of the vent/recycle changeover dampers. It has a high level alarm which sounds at both the local control panel and in the main control room when the chlorine concentration exceeds a pre-set level.

9.2.12 The sodium hydroxide solution is of 10-12% concentration and is held in a solution tank beneath the packed tower. When the system is in operation, the sodium hydroxide is recirculated by a pump to the distributor at the top of the packed tower to provide adequate irrigation to the packing. Sufficient solution is provided to absorb 1 tonne of chlorine. A mist eliminator is provided at the top of the packed tower to prevent entrainment of liquid into the treated air.

9.2.13 The scrubber is provided with the following additional features: a sampling point, a top entry mixer (for in-situ preparation of the sodium hydroxide solution), a direct reading transparent level gauge, an inspection window and level indication with high and low level alarms, and a temperature measurement device for monitoring the temperature of caustic solution during the preparation process.

Emergency Repair/Stoppage Kit for Chlorine Spillage/Leakage

9.2.14 According to Fire Services Department's fire safety requirements, a set of emergency repair/stoppage kit for chlorine spillage/leakage is provided and maintained in good working condition at all times for use by the trained persons and stowed adjacent to but outside the store/plant room. Regular drills are conducted to train personnel on the proper use of the breathing apparatus and protective clothing.

Surrounding Topography

9.2.15 Yau Kom Tau WTW is situated on a hillside at about 80 m above Principal Datum (PD). To the north of the site is a steep hill rising to above 400m PD. To the south, the slope falls towards Rambler Channel. High rise developments lie to both the east and southwest of the site. Both the topography and nearby high rise buildings have been taken into account in the chlorine dispersion modelling.

9.3 Tsuen Wan Road Upgrading Project

Location and Scope of the Project

9.3.1 The existing TWR is generally a dual-three lane carriageway, dropping to 2 lanes through both Tai Chung Road and Tsuen Tsing Interchanges with single lane up and down ramps at Kwai Tsing Road, Texaco Road and Tai Chung Road. It is an elevated structure between Tuen Mun Road and Tsuen Tsing Interchange.

9.3.2 Under the project, two single lane viaducts with connection to Tuen Mun Road will be provided on the two sides of the existing elevated section of TWR. The proposed viaducts will be separate from the existing structure. Near Hoi Kok Street, these two single lane viaducts will link with the up and down ramps from/to Hoi On Road and become two 2-lane viaducts. The proposed 2-lane viaducts will also be separate from the existing structure and located on the two sides of the existing elevated section of TWR. A down ramp branching off from the Kowloon bound 2-lane viaduct would be provided near Tsuen Tsing Interchange for the traffic to Texaco Road. The proposed 2-lane viaducts will end after joining the at-grade TWR near Tsuen Tsing Interchange, where the existing at-grade TWR would be widened from dual-3 to dual-5 until the Project limit near Kwai Tsing Interchange.

9.3.3 The EIA study area covers the existing urban area of Tsuen Wan West and Kwai Chung. This dense urban area includes a variety of landuse such as industrial/commercial buildings, high rise residential towers, schools, open spaces and recreation facilities. In addition, several West Rail property developments are planned in the immediate neighbourhood of the Project. The location of the project, new roadways to be constructed and the planned MTR developments are shown in Figure 9-3.

Construction Phase

9.3.4 Construction works are tentatively planned to commence in 2011 for completion in Year 2015. A maximum of 840 construction workers are expected to work on-site with additional 110 in the office to be located in the western part of the project site.

Operational Phase

9.3.5 Completion of the project will result in improving driving conditions on the Tsuen Wan Road and an overall increase of the traffic volumes. Increased traffic volumes are also expected on Tuen Mun Road. The details of the traffic volumes predicted at various locations are given in Appendix 9-J.

9.4 Meteorological Conditions and Population Data

Meteorological Conditions

9.4.1 For the sake of consistency with the previous assessment, the meteorological data to be used in this study is the data recorded at the Shell Tsing Yi weather station in the year 1996 by the Hong Kong Observatory which was also used in the hazard assessment for Yau Kom Tau Water Treatment Works (ERM, 2001). The weather data have been rationalised into different combinations of wind direction, speed and atmospheric stability class. The probabilities of occurrence of each combination during day and night are presented in Table 9-2. The Pasquill-Gifford stability classes range from A through F. Class A represents extremely unstable conditions which typically occur under conditions of strong daytime insolation. Class F on the other hand represents moderately stable conditions which typically arise on clear nights with little wind. Turbulent mixing, which will affect the dispersion of a chlorine cloud, increases through the stability class range from F to A.

Table 9-2 Meteorological Data Used in the Study

	Probability
Wind Speed (m/s):	Day				Night			Total
	2.5	2.5	5	2	2.5	5	2
Atmospheric Stability:	B	D	D	F	D	D	F
Direction
N	0.0018	0.0078	0.0019	0.0070	0.0068	0.0043	0.0315	0.0611
NE	0.0011	0.0043	0.0001	0.0051	0.0000	0.0001	0.0348	0.0455
E	0.0134	0.0335	0.0497	0.0166	0.0387	0.0442	0.1198	0.3159
SE	0.0398	0.0452	0.0528	0.0088	0.0106	0.0173	0.0777	0.2522
S	0.0487	0.0252	0.0147	0.0046	0.0036	0.0038	0.0161	0.1167
SW	0.0143	0.0039	0.0009	0.0009	0.0010	0.0001	0.0035	0.0246
W	0.0181	0.0080	0.0036	0.0009	0.0030	0.0010	0.0057	0.0403
NW	0.0137	0.0329	0.0216	0.0085	0.0206	0.0058	0.0403	0.1434
Total	0.1509	0.1608	0.1453	0.0524	0.0843	0.0766	0.3294	1

Population in the Areas Surrounding YKT WTW

Population Assumptions of the YKT WTW QRA Study

9.4.2 Since the present assessment is based on the previous QRA Study for YKT WTW (ERM, 2001) we start with a description of the population assumptions of that study and then will describe their modifications introduced for the present assessment to make the population information up to date.

9.4.3 The population data for the area surrounding Yau Kom Tau WTW have been gathered from a variety of sources:

· Planning Department;

· Social Welfare Department;

· Education Department;

· Transport Department; and

· surveys undertaken by the Consultants.

9.4.4 Detailed information has been gathered for the area up to 2 km from the chlorine store at Yau Kom Tau WTW.

9.4.5 The population data which was included in the Hazard Assessment is all population within the Consultation Zone (CZ), in addition to population outside the CZ which can be affected by the LD03 contour (refer to Section 9.6), up to a maximum of 2 km from the WTW.

9.4.6 The population data is presented in five time periods: night, working day, weekend day, peak hour and 'jammed peak', the last time period representing conditions under which traffic on some roads is at a standstill, 'bumper-to-bumper'. The definition of the time periods is given in Appendix 9-I.

9.4.7 For the purposes of risk modelling (refer to Section 9.8) the population was grouped in a number of separate point, line and polygon GIS themes shown in Figure 9-4. The corresponding population numbers assumed for these areas are presented in Table 9-3. Note that except for major roads (33 to 39), the population units with a letter reference lie within the 1 km Consultation Zone, while the numbered units are located further away from YKTWTW. Population units labelled V, W and 40 have been added for the purposes of this assessment.

9.4.8 While two separate population datasets reflecting the 1996 situation and 2006 population projection were used in ERM (2001) only the latter case is relevant in this study. The detailed population data of the 2006 case of the previous assessment are listed in column “Case 1” of Table 9-3.

9.4.9 Sensitive populations such as homes for the elderly, kindergartens and hospitals (vulnerable population factor 3.3) are separately identified from other populations (vulnerable population factor 1). The definition of the vulnerable population factor and its use in the QRA are explained in Section 9.6.

9.4.10 The modelling of the impact of chlorine releases on road populations follows the approach in HSE (1998) and is explained further in Section 9.6

General Population Assumptions for the Present Assessment

Population Scenarios

9.4.11 Five separate population scenarios are considered in this study:

· Case 1 is a scenario using the 2006 population estimates of the previous assessment (ERM, 2001). It is included here as a background case, for comparison with other scenarios;

· Case 2 uses the same general population data as Case 1, but the road populations reflect the traffic projections for 2030, after the completion of the project;

· Case 3 is similar to Case 2 but the road populations reflect the 2015 construction phase of the project;

· Case 4 uses the updated general population data and the 2030 projection for the road populations; and

· Case 5 involves the updated general population data, but concerns the 2015 construction phase of the project.

9.4.12 As can be seen our five scenarios include different combinations of two general population data sets: the 2006 projection of ERM (2001) and the updated population data as well as three road populations: the 2006 projection of ERM (2001), the 2015 construction phase, and 2030 operational phase of the project.

Updating the General Population Data

9.4.13 The following paragraphs discuss updating of population assumptions for various population units listed in Table 9-3. Note that only the population totals, where necessary, are subject to change. The assumptions regarding the population presence during different time periods, vulnerability factors etc remain unchanged. The road populations are separately discussed in Sections 9.4.28 to 9.4.53.

9.4.14 Note that the same “updated” general population estimates are applicable to Case 4 reflecting the 2015 situation and to Case 5 which concerns year 2030. The reason is that reliable population forecasts for the area are not available beyond 2011 (PlanD, 2006). Thus, where applicable (ie principally for the village populations only), the 2011 data have been assumed for both 2015 and 2030. This is considered a conservative assumption, since in general the population in this area is declining over the years. Note that the PlanD 2003-based TPEDM population projections up to 2021 by “PVS-338 Zones” have also been consulted, but those projections are older, applicable to strategic studies only and their PVS-338 Zones appear less adequate to the specific areas considered in this study than the Tertiary Planning Units (TPUs) used in PlanD (2006). Thus, the PlanD 2011 of PlanD (2006) projections will form the basis for our village population estimates. It should however be noted that also according to the TPEDM projections for areas relevant to this project their 2021 population estimates do not exceed the 2006 levels.

YKT WTW Staff Quarters (Ref A)

9.4.15 36 residents were assumed for 2006 to live in the YKT WTW Staff Quarters (Table 9-3 Ref A). However, according to the recent information provided by WSD, there are currently only 10 people living there and that number is not expected to increase in the future. Thus, the population of 10 is assumed for cases 4 and 5 of the present study.

Golden Villa and the new Residential Developments nearby (Ref V, W)

9.4.16 A residential development Golden Villa is located on the north side of Castle Peak Road, just to the south of Blocks 5, 6 and 7 of Hanley Villa. Golden Villa was not specifically mentioned in the previous assessment. The age of the buildings is not known, so it is not certain if its population was included in the total for the Sunny and Hanley Villas and Keymount Lodge (Table 9-3 Ref F) or was simply not taken into account, since it was non-existent at the time. It is a two-block, 30 storey development with two flats on each floor. Additionally, a new residential development is being constructed just to the east of Golden Villa. We were advised by Paul Y Builders, Ltd who is the construction contractor that this will be a one-block, 33 storey development also with 2 flats at each floor.

9.4.17 To accommodate the population of these two developments in the risk assessment, a new separate population unit V has been created. Its population is estimated assuming 2.5 people per flat ie a number slightly greater than that used for the Sunny and Hanley villas estimates in the previous assessment as 2.5 x 2 x 2 x 30 + 2.5 x 2 x 1 x 33 = 465 and this estimate is used for Cases 4 and 5 of the present study.

9.4.18 Another new residential development, close to YKT WTW has been recently constructed on Po Fung Road, south of Tuen Mun Rd. It is about 25 storeys high with three flats per floor, while at present the building is not yet in use, its future population could be estimated at 25 x 3 x 2 5 = 188. This development will be represented by a new population unit W.

Tsuen Wan Bay Further Reclamation (Ref M)

9.4.19 Under the previous assessment, over 25,000 people were assumed to live in 2006 in the Tsuen Wan Bay Further Reclamation area (TWBFR). However, no reclamation works have been undertaken to date and according to the latest Tsuen Wan Zoning Plan No S/TW/24, the project has been cancelled. Thus, the possible future population of TWBFR is not taken into account in Cases 4 and 5 of the present assessment.

Village Population

9.4.20 Since the specific village population projections for the area are not available, it has been decided to apply a general Planning Department population forecast for separate Tertiary Planning Units (TPUs).

9.4.21 Thus, the future village populations (refer to Table 9-3 Ref 11, 12, 22, 24-29, B) all located within the PlanD TPU 3.2.3 are estimated based by applying a factor of 0.975 to their 2006 population levels assumed in the previous assessment. The factor is based on the general population for 2006-2011 projection for TPU 3.2.3 (see Table 15, PlanD, 2006). Similarly, for the village population unit 31, located within TPU 3.5.1 a factor of 0.974 will be applied

9.4.22 Note that the assumed decline of the village populations may seem contradictory to the general HK population trends[1]. However, in our opinion, even if the general HK population will follow the high growth scenario, the population growth will be concentrated in the newly developed areas, while the population of the already developed areas such as Tsuen Wan, and in particular the village population in such areas would remain steady or decline, as shown in PlanD (2006).

9.4.23 Note that PlanD projections are not available beyond 2011 so we conservatively assume that after 2011 no further population decline will occur in the area.

Yau Kom Tau Squatter Population (Ref E)

9.4.24 Due to its proximity to YKT WTW and low protection provided by squatter dwellings, Yau Kom Tau Squatter population estimated at 676 during the previous assessment, is of particular interest. On a site visit on 12 September 2007 the number of squatter dwelling in the area, some of which seemed abandoned, was estimated as not more than 20. While, according to the information obtained from Lands Department there are no specific clearance plans for the area, it is reasonable to expect that in the future that population will further decline. Nevertheless, for the Cases 4 and 5 of this assessment, the squatter population E is conservatively assumed at 100.

Schools, Hospitals, Kindergartens etc.

9.4.25 Population of the schools, hospitals, kindergartens and similar institutions (refer to Table 9-3 Ref 2, 3, 4, 5, 6, 7, 14, 15, 16, 17, N, O, P, Q, R, S, T, U) are assumed at their 2006, Case 1 levels. Note that in the previous assessments it was also assumed that their population does not change, and the 2006 projections were equal to actual 1996 numbers.

Specific Areas of Residential and Workforce Population

9.4.26 Specific residential and hotel developments that were included in the previous assessments (refer to Table 9-3 Ref 1, 8, 9, 10, 19, 23, C, D, F, G, H, I, K, L,) are also assumed to retain populations at their 2006 levels. Note that this is a conservative assumption, since in the previous assessment (ERM, 2001) for this category of buildings the population was assumed to decline, and the projected 2006 numbers were markedly lower than those for 1996.

9.4.27 Similarly, for the following other units the populations are assumed at their 2006, Case 1 levels: 13 (Discovery Shopping Mall), 18 (Factories and Godown), 20 (Fire Station), 21 (Hing Shing Temporary Housing Area), 30 (Cement Plant), 32 (Tsing Yi Jetties and Shipyards), and J (Bus Terminal). It should be noted that for the unit 18 with a particularly high population number at 12,000, this is a conservative assumption, since the HK manufacturing activities tend to decline over the years.

Road Populations

9.4.28 Since the current assessment concerns the Tsuen Wan Road Upgrading project and the Tuen Mun Road population contributed over 30% of the total PLL under the previous assessment (see Table 7.3b of ERM, 2001) projections of the road population require a special attention.

9.4.29 The methodology of the road population estimation does in general follow that of the previous assessment, however the estimates are now based on the traffic information from the Traffic Impact Assessment (TIA) for the Tsuen Wan Rd Upgrading Project (refer to Appendix 9-J) that has been endorsed by Transport Department and other relevant parties, or, where unavailable, from the latest traffic census. Note that the former case applies to all the major roads in the vicinity of YKT WTW, ie not only Tsuen Wan Rd, but also Tuen Mun Rd, eastern sections of Castle Peak Rd and Hoi On Rd, so the 2006 traffic census information is only used for more remote roads such as Ting Kau Bridge or the western part of Castle Peak Rd. The following paragraphs describe population assumptions for all roads that are included in the assessment for Case 4 - the 2030 operational phase of the project. The methodology for the Case 5, is similar, only the data from different TIA tables are used, so the computations are not repeated here and only their results are listed in Table 9-3. The Case 5 population of Tsuen Wan Rd is additionally discussed in Section 9.4).

9.4.30 Generally, the peak hour populations for free flowing roads are estimated from the relevant traffic data assuming vehicle speed of 35 km/h and 3.3 persons per vehicle on the average. It should be noted that the latter assumption, used both in ERM (2001) and in the present assessment is significantly higher than 2.3 persons per vehicle assumed in the previous Route 16 assessments (Scott Wilson, 1998a,b).

9.4.31 The pertinent data from the Traffic Impact Assessment for the Tsuen Wan Rd Upgrading Project are for the reader’s reference included in Appendix 9-J to this report.

Tuen Mun Road (Ref 33)

9.4.32 The 1996 estimate of 2593 people present on the 3.158 km stretch of Tuen Mun Rd (refer to Figure 9-4) during the peak hour used in the previous assessment was obtained from the AADT value of 99180 from the 1996 Traffic Census multiplied by a factor of 0.00943 and the length of the road (3.158).

9.4.33 The factor 0.00943 is equal to 0.10 x 3.3 / 35 where 0.1 is the assumed fraction of AADT during the peak hour, 3.3 the average number of people per vehicle and 35 the assumed peak hour traffic speed in km/h.

9.4.34 Note that even during the “jammed peak” period the traffic is assumed free-flowing, with above estimate valid for both the “peak” and “jammed peak” periods.

9.4.35 Similar approach is adopted for the present assessment, however the peak hour traffic volumes are now taken from the Traffic Impact Assessment undertaken for the Tsuen Wan Rd Upgrading Project.

9.4.36 For the case “with TWR upgrading”, the TIA 2030 estimate of the Tuen Mun Rd two-way traffic (Sections E+D, see Figure 4.2 and Annex J) the morning peak hour traffic volume is 7650.

9.4.37 Thus, applying the same assumption as in ERM (2001) we obtain 7650 x 3.158 x 3.3 / 35 = 2278. Note that while this value is lower than 2354 used for 2006 in the previous assessment, it should not be surprising, since according to the 2005 Traffic Census the traffic volumes on Tuen Mun Rd (Core Station 5035) significantly declined between 2002-2005 and in 2006 the AADT value at 68530 was again lower than in 2005, so it can be assumed that this trend would continue. If the factor 0.00943 x 3.146 was applied to the actual 2006 AADT, the population estimate would stand at 2040, significantly lower than the 2006 projection made for the previous assessment.

9.4.38 Following ERM (2001) the time distribution of the Tuen Mun Rd, as well as all the other free-flowing roads is assumed as the percentage of the peak as follows:

· night: 5%

· jammed peak: 100%

· peak hour: 100%

· weekend day: 50%

· working day: 55%

Ting Kau Bridge (Ref 34)

9.4.39 Since the Ting Kau Bridge is not covered by the TIA, the traffic estimates are based on the latest Traffic Census. Since the bridge lies on the periphery of the study area, this approach is considered adequate for the purposes of the study. According to the 2006 Traffic Census the AADT on Ting Kau Bridge was 76300 in 2005 and 75380 in 2006. The length of the road section included in the modelling is 1.147 km. Using the 2006 AADT and applying the same 0.00943 factor (see above) we obtain the number of people on the bridge at the peak hour as 75380 x 0.00934 x 1.147 = 808. The above estimate, significantly higher than the 2006 projection used in the previous assessment is assumed for the Cases 2 - 5 of the present study.

9.4.40 As in the previous assessment, similar to Tuen Mun Rd, the Ting Kau Bridge traffic even during the “jammed peak” period the traffic is assumed free-flowing, with above estimate valid for both the peak and jammed peak periods.

Tsuen Wan Road (Ref 35)

9.4.41 Tsuen Wan Rd was in the previous assessment modelled as a single line object. Despite the addition of new lanes planned under the present project, since they run parallel to the existing lanes, that approach is considered adequate for the present assessment as well, especially that the road lies quite far from YKT WTW.

9.4.42 Population estimate for TWR on the 2030 traffic projections provided in the Traffic Impact Assessment for the Project was referred in this assessment. The traffic estimate is based on the western (closer to WTW) segment of the road with the heaviest traffic, ie the sum of TIA sections BO, AK, AL, AG, J, G, F and B (refer Figure 9-4). The morning peak hour traffic for the case “With TWR Upgrading” is projected for these sections as 10400 vehicles (refer to Appendix 9-J). Applying the same assumptions as in the case of Tuen Mun Rd, the number of people on the 1.570 km segment of Tsuen Wan Road included in the modelling can be estimated as 10400 x 1.571 x 3.3 /35 = 1540.

9.4.43 As in the previous assessment, similar to Tuen Mun Rd, the Tsuen Wan Rd traffic even during the “jammed peak” period the traffic is assumed free-flowing, with above estimate valid for both the peak and jammed peak periods.

Castle Peak Section 1, from Sham Tseng to Bayview Garden, Ref 36

9.4.44 The 2006 Traffic Census gives the AADT at station 6209, located on Castle Peak Road to the east of Ting Kau Bridge as 6690, 1.5% lower than in 2005 and over 33% lower than in 1996. Since suitable traffic projections for the future years on this section of the Castle Peak Road are not available, this number will form a basis for our population estimate. It is considered conservative, since similar to other neighbouring roads, the traffic is has been declining over the recent years. The length of this section is 2.616 km so the population estimate can be obtained as 6690 x 0.00934 x 2.616 = 163, a number significantly lower than used in the previous assessment.

9.4.45 As in the previous assessment, similar to Tuen Mun Roadd, the Tsuen Wan Road traffic even during the “jammed peak” period the traffic is assumed free-flowing, with above estimate valid for both the peak and jammed peak periods.

Castle Peak Road Section 2 (from Bayview Garden to Tuen Mun Road) Ref 37

9.4.46 This Castle Peak Road section was the only road for which in the previous assessment the traffic in one of the directions during the jammed peak period was considered jammed, ie ‘bumper to bumper’. The same assumption will be used in the present study, so the population estimation methodology is in this case slightly different than for the other roads.

9.4.47 Similar to the previous assessment (ERM, 2001), the population of the jammed lane is estimated assuming that each vehicle occupies on average 6 m of the road. With the road length of 0.951 km and assuming as in the previous cases 3.3 persons per car, the jammed lane population is 3.3 * 951/6 = 523.

9.4.48 The population of the other lane is computed as for other free-flowing roads, based on the peak hour data from the project Traffic Impact Assessments. The 2030 traffic volume on this section of Castle Peak Road is given in Appendix 9-J as 800 (Section BR – see Figure 9-5) or 1050 (Section BD). For the one way traffic estimate we will use one half of the average of these values, ie (800 + 1050)/4 = 463. Thus, assuming as for other roads the average speed of 35 km/h and 3.3 persons per vehicle, the population of this lane is 463 x 0.951 * 3.3 /35 = 42. The total jammed peak population of this road section is therefore 523 + 42 = 565.

9.4.49 The road population for the peak hour, where both lanes are assumed free-flowing is simply 84, the double of the free-flowing lane population obtained for ‘jammed peak’. Note that this number constitutes 15% of the jammed peak population.

9.4.50 For the other time periods we use the same fractions of the peak periods that are used for other free flowing roads (see sub-section on Tuen Mun Rd above). Thus, the corresponding percentages of the Cases 2 and 4 jammed peak population can be obtained as:

· night: 5% * 15% = 0.8%

· jammed peak: 100%

· peak hour: 15%

· weekend day: 50% * 15% = 7.5 %

· working day: 55% * 15% = 8%

9.4.51 Note that the corresponding jammed peak population of this section of Castle Peak Rd is for Cases 3 and 5 is 538 and the peak hour population is 29, so the night, peak, weekend day and working day fractions can be similarly established at 0.3%, 5.4%, 2.7% and 3.0%.

Castle Peak Road Section 3 (from Tuen Mun Road to Tai Chung Road) Ref 38

9.4.52 For this part of Castle Peak Rd, which is also assumed as free-flowing, the 2030 peak hour traffic volume is taken as the sum of TIA sections A + BH + H (refer to Figure 9-5), and according to Table J2 in Appendix 9-J amounts to 3600 vehicles. The length of this road section assumed in the modelling is 1.040 km. Thus the peak hour population is estimated as 3600 x 1.040 x 3.3 / 35 = 353.

Hoi On Road Ref 39

9.4.53 Hoi On Rd traffic is assumed as an average of TIA sections BS and BE (see Figure 4.2) which according to Table J2 in Appendix 9-J is equal to 3700 vehicles. The length of the road assumed in modelling is 0.935 km. Thus Case 2 and 4 of 2030 peak hour population can be estimated at 3700 x 0.935 x 3.3 /35 = 326.

Construction Phase Population

9.4.54 For most population units the Case 3 and 5, ie Construction Phase population is simply based on the appropriate population projections and the 2015 TIA traffic forecast ‘without the project’. However the construction workforce for the project needs also to be accounted for. According to the information provided by CEDD, the maximum number of construction workers onsite will be 840 working on the Tsuen Wan Rd plus 130 in temporary office located in the western part of the project site.

9.4.55 For the office workers a separate population unit 40 is created. The exact location of the office is not yet known, but it is understood that two possible locations are considered. In order to account for the worst case, we assume that it will be located at a site closer to the WTW, in the western part of the Project site area (refer to Figure 9-4).

9.4.56 The remaining workforce is assumed to be evenly distributed along the Tsuen Wan Rd, so this population component can simply be added to the corresponding working day traffic population (Ref 35).

9.4.57 Similar to Cases 2 and 4 above, the jammed peak and peak hour Case 3 and 5 traffic population of Tsuen Wan Rd, can be obtained as the sum of TIA sections BO, AK, AL, AG, J, G, F and B multiplied by 1.571 x 3.3 /35, ie (see Table J1 of Appendix 9-J) 5850 x 1.571 x 3.3 /35 = 867. The working day traffic population is assumed as 55% of the peak level, i.e. 477. Note that of the about 3.25 km of Tsuen Wan Road within the Project site area, only its western 1.571 km is modelled in this assessment, since the remaining part lies beyond the area of potential risk associated with YKT WTW. So, only 840 x 1.571/3.25 = 406 construction workers need to be added to the traffic population of that section of the road. Thus, the working day population of Tsuen Wan Rd during the construction phase can be estimated as 477 + 406 = 883. The peak and jammed peak population at 867 constitutes now 98% of this maximum, the night population is 5% x 98% = 5% and the weekend day population 50% x 95% = 49%.

Table 9-3 Population Data Assumed for Different Scenarios

Ref	Description	Case 1 Background 2006 (Previous Assessment)	Case 2 Project Operation + Case 1	Case3 Project Construction +Case 1	Case 4 Project Operation + Revised Population	Case 5 Project Construction + Revised Population	Night	Jammed Peak	Peak Hour	Weekend Day	Working Day	Portion Indoors	Vulnerable Pop. Factor	Storeys
01	Allway Gardens	9753	9753	9753	9753	9753	100%	50%	50%	70%	50%	99%	1	27
02	Kei Wai Primary School CCC HK Council	840	840	840	840	840	0%	50%	50%	0%	100%	95%	1	5
03	YCH Lim Por Yen Secondary School	1080	1080	1080	1080	1080	0%	50%	50%	0%	100%	95%	1	5
04	PLK Lee Shing Pik College	1160	1160	1160	1160	1160	0%	50%	50%	0%	100%	95%	1	5
05	Chai Wan Kok Catholic Primary School	840	840	840	840	840	0%	50%	50%	0%	100%	95%	1	5
06	Textile Int Acc Woo Hon Fai Prevocational School	1000	1000	1000	1000	1000	0%	50%	50%	0%	100%	95%	1	5
07	Tsuen Wan Adventist Hospital	400	400	400	400	400	100%	100%	100%	100%	100%	99%	3.3	5
08	Sheeny Terrace, Joyful Building, Kam Fung Garden	2508	2508	2508	2508	2508	100%	50%	50%	70%	50%	99%	1	27
09	Proposed Residential Development	3456	3456	3456	3456	3456	100%	50%	50%	70%	50%	99%	1	35
10	Tsuen King Garden, Tsuen Wan Centre	21483	21483	21483	21483	21483	100%	50%	50%	70%	50%	99%	1	32
11	Squatters and Villages	31	31	31	30	30	100%	50%	50%	70%	50%	0%	1	1
12	Squatters and Village Housing	1838	1838	1838	1792	1792	100%	50%	50%	70%	50%	0%	1	1
13	Discovery Park (Shopping Mall)	1000	1000	1000	1000	1000	0%	50%	50%	100%	50%	99%	1	4
14	Allway Kindergarten	210	210	210	210	210	0%	50%	50%	0%	100%	95%	3.3	1
15	Annuciation Catholic Kindergarten	180	180	180	180	180	0%	50%	50%	0%	100%	95%	3.3	1
16	Sun Island English Kindergarten	294	294	294	294	294	0%	50%	50%	0%	100%	95%	3.3	1
17	Morning Sun Kindergarten	455	455	455	455	455	0%	50%	50%	0%	100%	95%	3.3	1
18	Factories and Godown	12000	12000	12000	12000	12000	10%	50%	50%	20%	100%	99%	1	20
19	Proposed Hotel Development	780	780	780	780	780	100%	50%	50%	60%	60%	99%	1	15
20	Fire Station	30	30	30	30	30	100%	100%	100%	100%	100%	95%	1	5
21	Hing Shing Temporary Housing Area	0	0	0	0	0	100%	50%	50%	70%	50%	90%	1	2
22	Village Housing	46	46	46	45	45	100%	50%	50%	70%	50%	99%	1	1
23	Proposed Hotel Development	780	780	780	780	780	100%	50%	50%	60%	60%	99%	1	14
24	Residential Village Housing	85	85	85	83	83	100%	50%	50%	70%	50%	99%	1	3
25	Ting Kau Village	174	174	174	170	170	100%	50%	50%	70%	50%	99%	1	3
26	Village Housing	11	11	11	11	11	100%	50%	50%	70%	50%	99%	1	1
27	Village Housing	319	319	319	311	311	100%	50%	50%	70%	50%	99%	1	3
28	Village Housing	6	6	6	6	6	100%	50%	50%	70%	50%	99%	1	1
29	Beach Building, Village Housing	5	5	5	5	5	100%	50%	50%	70%	50%	99%	1	2
30	Cement Plant	10	10	10	10	10	10%	50%	50%	20%	100%	50%	1	1
31	Village	120	120	120	117	117	100%	50%	50%	70%	50%	99%	1	3
32	Jetties and Shipyards	100	100	100	100	100	10%	50%	50%	20%	100%	50%	1	1
33	Tuen Mun Road	2354	2278	1772	2278	1772	5%	100%	100%	50%	55%	50%	1	0
34	Ting Kau Bridge	533	808	808	808	808	5%	100%	100%	50%	55%	75%	1	0
35	Tsuen Wan Road	1304	1540	883	1540	883	5%^(1)	100%^(1)	100%^(1)	50%^(1)	55%^(1)	75%	1	0
36	Castle Peak Road Section 1 (from Sham Tseng to Bayview Garden)	275	163	163	163	163	5%	100%	100%	50%	55%	75%	1	0
37	Castle Peak Road Section 2 (from Bayview Garden to Tuen Mun Road)	752	565	538	565	538	1%^(2)	100%	14%^(2)	7%^(2)	7%^(2)	75%	1	0
38	Castle Peak Road Section 3 (from Tuen Mun Road to Tai Chung Road)	542	353	338	353	338	5%	100%	100%	50%	55%	75%	1	0
39	Hoi On Road	170	326	238	326	238	5%	100%	100%	50%	55%	75%	1	0
40	Temporary Office for Tsuen Wan Rd Upgrading Project	0	0	130	0	130	0%	0%	0%	0%	100%	100%	1	3
A	YKT WTW Staff Quarters	36	36	36	10	10	100%	50%	50%	70%	50%	99%	1	2
B	Yau Kom Tau Village Extension	476	476	476	464	464	100%	50%	50%	70%	50%	99%	1	3
C	Greenview Court and Belvedere Garden	10394	10394	10394	10394	10394	100%	50%	50%	70%	50%	99%	1	33
D	Gardenview Terrace and Blossom Terrace	587	587	587	587	587	100%	50%	50%	70%	50%	99%	1	3
E	Yau Kom Tau Squatters	676	676	676	100	100	100%	50%	50%	70%	50%	0%	1	1
F	Sunny and Hanley Villa, Keymount Lodge	1794	1794	1794	1794	1794	100%	50%	50%	70%	50%	99%	1	25
G	Long Beach Garden	142	142	142	142	142	100%	50%	50%	70%	50%	99%	1	3
H	The Bay Bridge	438	438	438	438	438	100%	50%	50%	50%	50%	99%	1	13
I	Bayview Garden	3419	3419	3419	3419	3419	100%	50%	50%	70%	50%	99%	1	30
J	Bus Terminal	150	150	150	150	150	10%	100%	100%	30%	30%	0%	1	1
K	Belvedere Garden, The Panorama	9915	9915	9915	9915	9915	100%	50%	50%	70%	50%	99%	1	30
L	Serenade Cove	2257	2257	2257	2257	2257	100%	50%	50%	70%	50%	99%	1	30
M	Tsuen Wan Bay Further Reclamation (Public and Private Housing)	25757	25757	25757	0	0	100%	50%	50%	70%	50%	99%	1	39
N	HK Baptish Convention Primary School	1050	1050	1050	1050	1050	0%	50%	50%	0%	100%	95%	1	5
O	Future Secondary School	1080	1080	1080	1080	1080	0%	50%	50%	0%	100%	95%	1	5
P	Future Primary School	840	840	840	840	840	0%	50%	50%	0%	100%	95%	1	5
Q	Creative KG	185	185	185	185	185	0%	50%	50%	0%	100%	95%	3.3	1
R	St. Monica's Play School	200	200	200	200	200	0%	50%	50%	0%	100%	95%	3.3	1
S	Belvedere Anglo-Chinese KG	130	130	130	130	130	0%	50%	50%	0%	100%	95%	3.3	2
T	Po Leung Kuk Fong Tam Yuen Leung Nursery	114	114	114	114	114	0%	50%	50%	0%	100%	95%	3.3	1
U	Tsuen Wan St. Andrew Anglo-Chinese Kindergarten	115	115	115	115	115	0%	50%	50%	0%	100%	95%	3.3	1
V	Golden Villa and a new residential development nearby	0	0	0	465	465	100%	50%	50%	70%	50%	99%	1	32
W	New residential development on Po Fung Rd	0	0	0	188	188	100%	50%	50%	70%	50%	99%	1	25

Notes:

(1) Valid for Cases 1, 2 and 4 only. For Cases 3 and 5 fractions see Section 9.4.

(2) Valid for Case 1 only. For Cases 2-5 fractions see Section 9.4.

9.5 Hazard Identification

Hazardous Characteristics of Chlorine

9.5.1 The following paragraphs summarise some of the key hazardous characteristics of chlorine (ICI Chlorine Handbook, 1995):

· Chlorine gas is heavier than air and as a result will tend to accumulate in low places when released to the atmosphere and flow downhill in still air. However, slight breezes or thermal turbulence will cause it to move upward, so people are not necessarily safe simply because they are above the point of release.

· Chlorine gas has a greenish-yellow colour which is only visible at concentrations (above approximately 500ppm) many times higher than the danger level (see Table 5.1 below).

· Chlorine gas is a respiratory irritant. Symptoms caused by inhalation of chlorine include: headaches, pain, difficult breathing, burning sensation of the chest, nausea and watering of the eyes.

The physiological effects of chlorine are summarised in Table 9-4.

Table 9-4 Physiological Effects of Chlorine (IChemE, 1989)

Concentration (ppm)	Effects
0.2-3.5	Threshold of odour perception in most individuals
3-5	Tolerated without undue ill effect for half to one hour.
5-8	Slight irritation of the mucous membranes of the upper respiratory tract and of the eyes.
15	Effects are immediate. Irritation of nose, throat and eyes with cough and lachrymation.
30	Immediate cough with a choking sensation, retrosternal chest pain and a sense of constriction in the chest.
40-60	Development of a chemical tracheo-bronchitis and pulmonary oedema.
1000	Concentration likely to be fatal after a few deep breaths.

Hazard Identification Study

Review of Past Incidents

9.5.2 A survey of worldwide incidents involving chlorine drums and cylinders has been conducted and presented in Annex A of Methodology Report (ERM, 1997a). There have been a total of 86 incidents in the previous 77 years.

9.5.3 The following reference data have been searched and consulted:

· MHIDAS Database (MHIDAS is a Major Hazard Incident Data Service developed by the Safety and Reliability Directorate of the UK Atomic Energy Authority. MHIDAS contains incidents from over 95 countries particularly the UK, USA, Canada, Germany, France and India. The database allows access to many other important sources of accident data, such as the Loss Prevention Bulletin, and is continuously updated);

· HSELine (The Library and Information Services of the UK Health and Safety Executive has accumulated in a computer database documents relevant to health and safety at work. HSELine contains citations to HSE and Health and Safety Commission publications, together with documents, journal articles, conference proceedings, etc.)

· Lees (1996);

· AQUALINE;

· Chlorine Institute;

· Chlorine Transport Risk Study, DNV Technica (1997)

9.5.4 The search undertaken was for all incidents involving chlorine 'drums', 'cylinders' or 'containers' OR incidents occurring at treatment works.

9.5.5 The results are summarised in the following tables.

Table 9-5 Summary of Chlorine Release Quantities

Release Quantity (in kg)	Number of incidents	Percentage
0 - 9	4	4.7%
10 - 99	8	9.3%
100 - 499	4	4.7%
500 -1000	9	10.5%
1000+	2	2.3%
Not specified	59	68.6%
TOTAL	86	100.0%

9.5.6 The above data shows a significant proportion of the incidents having a release quantity of 500 kg or more. A number of these incidents concerned faulty or damaged valves or connecting pipework, illustrating the difficulty in being able to isolate even relatively small leaks.

Table 9-6 Summary of Injuries

Number of Injuries	Number of incidents	Percentage
0 - 9	23	26.7%
10 - 49	10	11.6%
50 - 100	6	7.0%
100 - 499	8	9.3%
500+	2	2.3%
Not specified	37	43.0%
TOTAL	86	100.0%

Table 9-7 Summary of Fatalities

Number of Fatalities	Number of incidents	Percentage
0	29	33.7%
1	7	8.1%
2	5	5.8%
3	2	2.3%
4 - 9	3	3.5%
10+	1	1.2%
Not specified	39	45.3%
TOTAL	86	100.0%

9.5.7 Of the 86 recorded incidents, 8 specified both injuries and fatalities. The overall injury to fatality ratio is 48 to 1 as shown below in Table 9-8.

Table 9-8 Injuries to Fatality Ratio

Date of Incident	Injuries	Fatalities
10/1/47	104	1
8/5/69	33	2
27/12/74	85	2
14/8/76	1	2
12/6/85	100	5
24/5/91	600	5
4/7/95	19	3
6/7/95	73	1
TOTAL	1015	21

Table 9-9 Location of Incidents

Location	Number of incidents	Percentage
Water Treatment Plant	22	25.6%
Sewage Treatment Plant	7	8.1%
Swimming Pool/Water Park	2	2.3%
Pumping Station	2	2.3%
Prechlorination House	2	2.3%
Road Transportation	14	16.3%
Rail Transportation	2	2.3%
Marine Transportation	4	4.7%
Loading/Unloading	5	5.8%
Elsewhere	16	18.6%
Not specified	10	11.6%
TOTAL	86	100.0%

9.5.8 The above table shows most incidents occurring either at water treatment plant (25.6%) or during transportation (23.3%).

Table 9-10 Primary Causes of Incidents

Primary Cause	Number of incidents	Percentage
Equipment Failure	13	15.1%
Human Error	7	8.1%
Corrosion	4	4.7%
Fire/Overheating	1	1.2%
Contamination	5	5.8%
Road accident	5	5.8%
Marine Accident	5	5.8%
Rail accident	1	1.2%
Loadshedding	5	5.8%
Unknown/not specified	40	46.5%
TOTAL	86	100.0%

9.5.9 The table shows that the primary cause of incidents was equipment failure, eg faulty valves or pipework (15.1%). However, accidents during transportation of chlorine containers are also significant (12.8%).

HAZOP Study

9.5.10 A Hazard and Operability (HAZOP) study was conducted for Yau Kom Tau WTW to provide a full and systematic identification of the hazards associated with delivery, storage and handling of chlorine (see ERM, 1997b, 2001). The HAZOP technique provides a means of examining deviations from the design intent, their causes, consequences and safeguards, in a structured manner.

9.5.11 The primary focus of the HAZOP was on the hazards posed to people off-site. 'Internal' as well as' external' hazards were considered, ie those within the control of the operating staff, such as the hazards arising during drum connection/disconnection, as well as those outside their control such as an external fire. The information provided for the HAZOP included the site layout plan, Process and Instrumentation Diagram (P&ID), chlorine store layout plan, as well as the Operations and Maintenance Manual.

9.5.12 The HAZOP sessions considered each of the following aspects of the design and operation of the WTW:

· transport of chlorine containers along the site access road (including manoeuvring of the truck outside the entrance to the truck unloading bay);

· handling of containers within the store;

· containers in storage;

· connection and disconnection of containers;

· chlorination system (including the liquid chlorine pipework, evaporators, chlorinators and ejectors);

· Contain and Absorb system.

9.5.13 The HAZOP considered the various operating modes of the plant (auto/manual) as well as planned maintenance operations. Prior to the HAZOP study, previous HAZOP studies of WTWs and chlorine leak incidents were reviewed to provide additional input to the identification of the chlorine release scenarios.

Review of the Hazard Identification Results

9.5.14 In light of the recent judgement of the Court of Final Appeal (FACV 28/2005) the previous Hazard Identification Results have been carefully reviewed for the purpose of the current assessment. In particular the latest version of the world wide accident database MHIDAS (Major Hazard Incident Data Service) has been independently reviewed in order to update the Hazard Identification conclusions. However, only a few relevant chlorine incidents occurred worldwide since the previous review, and after examination of their nature it has been concluded that noonly minor revisions of the previously identified hazard scenarios are necessary for this study. They are incorporated in the following sections where applicable.

Results of the Hazard Identification Study

9.5.15 Table 9-11 shows the hazards which were identified for Yam Kom Tau WTW as a result of the review of past incident and the HAZOP studies.

9.5.16 In summary, the primary hazard arises from a loss of containment of chlorine with subsequent acute exposure of people off-site leading to injuries or fatalities. Releases may range in size from a small leak (eg via a valve gland), through to dislodgement of a fusible plug (¾" diameter) or catastrophic failure of the container itself. The releases may be isolatable (ie via closure of the changeover valves or drum valve) or non-isolatable (ie a leak from the shell of the container).

9.5.17 The physical state of the release may be gas, liquid or two-phase depending on the precise location, eg a small leak downstream of the evaporator is likely to be gas, a leak from the 'pigtail' connection is likely to be two-phase (due to flashing in the line), whereas a leak from the container shell itself is likely to be liquid. The release may arise from failure of the chlorine equipment itself or failure induced by an external event such as an earthquake or landslide. The quantity of chlorine released may vary from a few kilograms to several tonnes released instantaneously, eg in the case of a severe external event such as an earthquake.

9.5.18 For releases occurring within the chlorine store a Contain and Absorb system is provided to minimise the likelihood of the release escaping to atmosphere. The principal failure modes of the Contain and Absorb system, as identified at the HAZOP, are summarised in Table 9-12.

Table 9-11 Hazards Identified during HAZOP Study for Yau Kom Tau WTW

Hazard

1. ACCESS ROAD

1.1 Fire on the truck leading to melting of the fusible plugs on one or more containers

1.2 Fire on the roadside leading to melting of the fusible plugs on one or more containers

1.3 Impact with object during truck manoeuvring

1.4 Lorry over-turns

1.5 Collision with another vehicle

1.6 Loadshedding

1.7 Spontaneous container failure

2. CONTAINER HANDLING

2.1 Dropped container

2.2 Collision of container with another object

2.3 Accidental impact of drum on pigtail during setdown at standby position

2.4 Overextension during use of truck crane (not normally used)

2.5 Section of monorail track incorrectly aligned leading to a dropped drum

3. CONTAINERS IN STORAGE

3.1 Leaking chlorine containers

3.2 Overfilled containers leading to overpressurisation on thermal expansion

3.3 Impurities in chlorine containers, in particular nitrogen trichloride (leading to explosion) or moisture (causing accelerated corrosion)

3.4 Object falls onto chlorine containers

3.5 Fire (external or internal)

3.6 External explosion

3.7 Lightning strike

3.8 Extreme wind

3.9 Flooding

3.10 Construction activities

3.11 Subsidence

3.12 Landslide

3.13 Earthquake

3.14 Aircraft crash

3.15 Sabotage

3.16 Vehicle crash

3.17 Electromagnetic interference

4. CONNECTION AND DISCONNECTION OF CHLORINE CONTAINERS

4.1 Human error or equipment failure during connection/disconnection of containers

5. CHLORINATION SYSTEM

5.1 Trapping of liquid chlorine between closed valves and subsequent thermal expansion leading to overpressurisation of pipework

5.2 Pigtail failure

5.3 Failure of fixed chlorine pipework

5.4 Open end on pipework due to operator or maintenance error

5.5 Corroded pipework

5.6 Presence of nitrogen trichloride in evaporator leading to explosion

5.7 Presence of moisture in evaporator leading to accelerated corrosion

Table 9-12 Failure Modes of Contain and Absorb System Identified at HAZOP Study

Hazard

1. LEAK DETECTION SYSTEM

Chlorine leak detector 'cell' malfunction

2. CONTAIN SYSTEM

Door left open

Air inlet louvres fail to close on demand

Normal ventilation system left in manual mode, therefore does not shutdown on detection of chlorine leak

Failure of chlorine absorption system (see below) with dampers set to 'exhaust to atmosphere' position and failure to detect chlorine breakthrough

Failure of door seals

Outage due to maintenance

System disabled by operator

3. CHLORINE ABSORPTION SYSTEM

Caustic pump failure

Blockage in caustic pipework

Overloading of scrubber

Caustic valve inadvertently closed

No caustic in tank

Caustic pipe leakage

Scrubber nozzles blocked

Initial make-up of caustic incorrect

Leakage of water into tank causing dilution of caustic and reduced efficiency

Degradation of caustic over a period of time

Wrong chemical in tank

Failure due to power failure

Outage due to maintenance

System disabled by operator

Characterisation of Chlorine Release Scenarios

9.5.19 Having identified various possible mechanisms for a chlorine release (Table 9-11), the next step in the Hazard Assessment is to characterise these release scenarios in terms of the releasing inventory, hole size and phase of release (Table 9-13). This follows the approach outlined in ERM (1997a), however the potential scenarios were re-examined and the table revised in light of the recent CFA (FACV 28/2005) requirements with regards to Hazard Assessment.

9.5.20 Note that, following the approach of (ERM, 2001) we conservatively assume that there still can be “old” and “new” type chlorine canisters in use, the former having six fusible plugs, instead of one and thus capable of releasing larger amount of chlorine. According to the information received from WSD only the “new” type is currently in use. However, as the number of fusible plugs is not specified in the contract, the chlorine supplier is free to make future deliveries in canisters with a larger number of plugs. We therefore decided to keep the original approach.

Table 9-13 Characterisation of Chlorine Release Scenarios

Chlorine release scenario	Outcome	Releasing inventory (tonnes)	Hole size (diameter)	Phase
1. ACCESS ROAD
1.1 Truck fire	Considered to result in melting of the fusible plugs on up to three drums. Three cases are considered according to the number of 'old' and 'new' drums^(1): Case (i): 3 new drums, no old drums Case (ii): 2 new drums, 1 old drum Case (iii): No new drums, 3 old drums	3 3 3	3x6mm 2x6mm (new drums) 6x6mm (old drums) 18x6mm	liquid liquid liquid
1.2 Fire on the roadside	Considered to present a negligible off-site risk as truck does not park on site outside the chlorine building	-	-	-
1.3 Manoeuvring accident	Considered to result in a single drum –small leak (eg valve gland failure)	1	3mm	liquid
1.4 Rollover	Single drum - small leak (eg valve gland failure) Single drum - medium leak (eg guillotine failure of drum valve)	1 1	3mm 8mm	liquid liquid
	Three drums - medium leak (eg guillotine failure of drum valves on three drums)	3	3x8mm	liquid
	Fire (outcomes as item 1.1 above)
1.5 Collision	Single drum - rupture Fire (outcomes as item 1.1 above)	1	-	liquid
1.6 Load-shedding	Single drum - small leak	1	3mm	liquid
	Single drum - medium leak	1	8mm	liquid
1.7 Spontaneous container failure	Single drum - medium leak	1	8mm	liquid
	Single drum - large leak (eg dislodgement of a fusible plug)	1	20mm	liquid
	Single drum - rupture	1(inst)	-	liquid
2. DRUM HANDLING
2.1 Dropped drum	Single drum - medium leak	1	8mm	liquid
	Single drum - large leak	1	20mm	liquid
	Single drum - rupture	1 (inst)	-	liquid
2.2 Collision of drum with another object	Single drum – medium leak	1	8 mm	liquid
2.3 Accidental impact of drum on pigtail during setdown at standby position	Pigtail - guillotine failure	1	4.5mm	two-phase
2.4 Dropped drum due to incorrect alignment of monorail truck or overextension of truck crane	Single drum –medium leak	1	8mm	liquid
3. CONTAINERS IN STORAGE
3.1 Leaking chlorine drums	Single drum - medium leak	1	8mm	liquid
	Single drum - large leak	1	20mm	liquid
	Single drum - rupture	1 (inst)	-	liquid
3.2 Overfilled drums leading to overpressurisa-tion on thermal expansion	As item 3.1 above
3.3 Impurities in chlorine drum leading to explosion or leak	As item 3.1 above
3.4 Object falls onto chlorine containers	Single drum – medium leak	1	8 mm	liquid
3.5 Fire (external or internal)	Since chlorine stores are 2 hour fire-rated structures, the most significant internal source of fire is considered to be the chlorine truck. However, pessimistically, all truck fires are modelled as occurring outdoors, thus this scenario is already included in Scenario 1.1.	-	-	-
3.6 External explosion	Single drum – medium leak	1	8 mm	liquid
3.7 Lightning strike	Single drum – medium leak	1	8 mm	liquid
3.8 Extreme wind	Considered to present no off-site risk as chlorine store is designed for typhoon loading	-	-	-
3.9 Flooding	Considered to pose no risk as flooding could only affect empty drums.	-	-	-
3.10 Construction activities	Construction activities could pose a significant risk to chlorine drums. However it is considered that this 'once-off' risk should be addressed through a HAZOP study prior to commencement of construction.	-	-	-
3.11 Subsidence	Single drum – medium leak	1	8 mm	liquid
3.12 Landslide	Considered to present no off-site risk - failure of cut slope (toe 10m from back of chlorine store) would not cause sufficient damage to store to affect chlorine containers (Ove Arup, 2000).	-	-	-
3.13 Earthquake^(2)	Roof collapse: Multiple drum-rupture	11 (inst)	-	liquid
3.14 Aircraft crash	Roof collapse: Multiple drum-rupture similar to earthquake)	11 (inst)	-	liquid
3.15 Sabotage	Single drum – medium leak	1	8 mm	liquid
3.16 Vehicle crash	Single drum – medium leak	1	8 mm	liquid
3.17 Electromagnetic interference	Considered to present no off-site risk as suitable precautions are adopted in the design of the electrical systems.	-	-	-
4. CONNECTION AND DISCONNECTION OF CHLORINE CONTAINERS
4.1 Human error or equipment failure during connection or disconnection of drums	Pigtail - guillotine failure	1	4.5mm	two-phase
5. CHLORINATION SYSTEM
5.1 - 5.5 Failures associated with the chlorination system pipework	Liquid chlorine pipework - guillotine failure	1.05^(3)	4.5mm^(4)	two-phase
5.6 - 5.7 Failure of Evaporator	Evaporator - leak or rupture	1.05	4.5mm	two-phase

Notes:

(1) 'New" drums have a single fusible plug on each drum and make up about 85% of the current drum stock. "Old" drums have six fusible plugs per drum and make up the remaining 15% of the drum stock. The fusible plugs have a 6mm fusible core (old and new drums)

(2) For assessment of effects of earthquakes on chlorine store refer to Appendix 9-G and Ove Arup (2000)

(3) Inventory of drum (1 tonne) and evaporator (50kg)

(4) Diameter of liquid chlorine pipework is 20mm but limiting orifice size is that of pigtail, ie 4.5mm.

9.6 Consequence Analysis

Methodology

9.6.1 The assessment of the consequences of a chlorine release essentially involves three steps:

· modelling the initial release of chlorine (whether inside or outside the chlorine building);

· modelling the dispersion of chlorine in the atmosphere;

· assessing the toxic impact to people off-site (whether indoors or outdoors).

9.6.2 Methodology of this study follows that of the YKT WTW QRA as described in ERM (1997a, 2001). The most important points are summarised and discussed below.

Initial Release of Chlorine

9.6.3 The initial release of chlorine or 'source term' is modelled using standard discharge rate formulae as detailed in ERM (1997a). Releases direct from the chlorine container are the most significant and, in the case of chlorine drums, these are modelled as liquid releases.

9.6.4 The rapid flashing of chlorine which occurs following a liquid leak from a drum is conservatively assumed to result in 100% entrainment of the liquid as aerosol with no rain-out. For catastrophic (instantaneous) liquid releases the rapid boiling of the chlorine on contact with the ground is assumed to result in entrainment of twice the initial flash fraction as aerosol, following Lees (1996). The remainder of the liquid chlorine is modelled as a spreading, evaporating pool.

9.6.5 For releases of chlorine within the chlorine building, a simple 'perfect mixing' model is used to account for the initial dilution of chlorine, based on Brighton (1989a, b) and Porter (1991). Instantaneous releases of 1 tonne of chlorine are assumed to result in pressurisation of the building to the extent that there could be a release of chlorine via weak points in the building structure, eg door seals. Continuous releases are assumed to be entirely contained, except in the event of failure of the Contain and Absorb system for which two modes of failure are considered: normal ventilation remains on or a door is left open.

Dispersion of Chlorine in the Atmosphere

9.6.6 Advanced techniques had been used for prediction of the dispersion of chlorine in the atmosphere. The effects of buildings and variable ground terrain on the dispersion of chlorine were modelled. The modelling involved three elements:

· wind tunnel simulations;

· Computational Fluid Dynamics (CFD); and

· flat terrain dispersion modelling.

9.6.7 The wind tunnel and CFD studies represent the 'state of the art' in dense gas dispersion modelling and provide the only rigorous means of accounting for the effects of buildings and complex terrain. Wind tunnel testing has been used in this study to investigate a range of release scenarios, wind directions and wind speeds in near-neutral atmospheric conditions. CFD has been used to determine the influence of atmospheric stability on the dispersion of chlorine and provide a broad comparison against the wind tunnel results for neutral stability. Note, that while the wind tunnel tests were conducted specifically for YKT WTW, the results of CFD dispersion modelling from for two other sites representing the extremes of topography (Sha Tin WTW and Tai Po Tau WTW) were used to supplement the wind tunnel test results. Both the wind tunnel testing and CFD modelling besides the major topographical features have included off-site high rise buildings as well as on-site buildings, as these have a significant influence on the dispersion of the chlorine.

9.6.8 The role of the flat terrain dispersion modelling had been to provide the 'source term' for both the wind tunnel and CFD studies. The model used in the study was DRIFT (Webber et al, 1992), an integral dispersion model developed by AEA Technology under the sponsorship of the UK Health and Safety Executive. DRIFT contains the necessary thermodynamics and heat transfer sub-models to be able to simulate the dispersion of a cold, aerosol-laden cloud typical of the early stage of a chlorine release. As DRIFT runs were used to simulate the full range of chlorine release rates and weather conditions. In conjunction with the wind tunnel and CFD, this provided all the data needed for input to the QRA.

Toxic Impact Assessment

Chlorine Probit Equation

9.6.9 The following probit equation has been used to estimate the likelihood of fatality due to exposure to chlorine:

Pr = -14.3 + lnC2.3t

where

Pr = probit value

C = chlorine concentration (mg/m³)

t = exposure time (minutes)

9.6.10 This probit equation is recommended for use in QRA studies by the Dutch Government (TNO, 1992) and incorporates the findings of most recent investigations into chlorine toxicity.

9.6.11 Table 9-14 shows the relationship between the chlorine concentration and the probability of fatality for the TNO probit, assuming 10 minute exposure duration.

Table 9-14 Chlorine Toxicity Relationship

Chlorine concentration (ppm)	Probit value for 10 min exposure	Probability of Fatality (LD = Lethal Dose)
251	3.17	0.03 (LD03)
557	5.00	0.50 (LD50)
971	6.28	0.90 (LD90)

Modelling of Escape from the Chlorine Cloud

9.6.12 In risk assessments for toxic gas releases it is common practice to take into account the possibility of escape of exposed persons. This is because at lower concentrations of the gas people may be able to obtain protection by moving indoors or directly out of the cloud.

9.6.13 Appendix 9-F provides details of the modelling of escape from a chlorine cloud. The methodology followed is similar to that developed by the UK Health and Safety Executive (Lees and Ang, 1989). It assumes that a person out of doors will have a probability of escape dependent on the chlorine cloud concentration, with escape occurring either directly out of the cloud or to a nearby building. The methodology takes into account the dose received during escape as well as the subsequent dose in the place of refuge. Suitable conservative assumptions are made for the time of escape bearing in mind the debilitating effect of the chlorine gas.

9.6.14 Incorporating all the above considerations it is possible to calculate an 'effective' outdoors fatality probability, ie the fatality probability that can be applied to the total outdoor population at any given location taking into account the probability of escape.

9.6.15 The consequence analysis gives three fatality probability contours for each release scenario, corresponding to 3%, 50% and 90% nominal outdoor fatality probability. The effective outdoors fatality probabilities corresponding to these levels of fatality are shown in Table 9-15.

Table 9-15 Effective Outdoors Probability of Fatality

Nominal outdoor fatality probability (for a person remaining outdoors)	% of population attempting escape	Effective outdoor fatality probability (taking into account the probability of escape)
90%	0%	90%
50%	80%	31%
3%	80%	0.7%

Protection for Persons Indoors

9.6.16 Based on a detailed analysis presented in Appendix 9-K, the probability of fatality for a person indoors is assumed as 5% of that for a person remaining out of doors, i.e. nominal outdoor fatality probability.

9.6.17 Protection is also considered for people on the upper floors of high-rise buildings. This is based on data on the typical height of a chlorine cloud provided by the dispersion modelling.

Sensitive Populations

9.6.18 Certain groups of people, ie the very young, the elderly and the infirm will be more sensitive to the effects of chlorine than others. This is taken into account in the QRA by increasing the fatality rate applied to certain sensitive receivers such as nurseries, old people homes and hospitals. The fatality rate for these groups of people is set a factor of 3.3 higher than for the average population (refer to Table 9-3).

Road Populations

9.6.19 The approach adopted for assessing the risk to road populations essentially follows that developed under research work undertaken on behalf of the UK Health and Safety Executive (HSE, 1998).

9.6.20 All roads within the study area that carry significant quantities of traffic are specifically modelled in the study. Data on traffic flows, obtained from the Project Traffic Impact Assessment and the latest Annual Traffic Census are described in Section 9.4.

9.6.21 The analysis takes into account the variation in traffic flows by modelling five separate time periods: night, working day, weekend day, peak hour and ‘jammed peak’, the last period representing conditions under which traffic is at a standstill, eg due to an accident, lane restriction etc.

9.6.22 One of the key factors in modelling the effects of chlorine on road vehicles is assessing the extent to which chlorine builds up inside the vehicle and whether this could impair driving ability. HSE (1998) shows that vehicles generally afford only limited protection from toxic gas ingress and that for a major highway located 100m from a chlorine storage facility, releases of the order of 10 kg/s could give rise to concentrations sufficient to cause vehicles to come to a standstill.

9.6.23 Thus, in this study the population associated with road vehicles is modelled as partially outdoors. To account for the worst case, the traffic is considered stationary, due to the presence of traffic lights, road junctions etc or the possibility that the chlorine itself may be of sufficient concentration to bring traffic to a halt. As for other outdoors populations, allowance is made for the possibility of escape of exposed persons either directly out of the cloud or to a nearby building, although with a minimum exposure duration of 5 minutes. Calculation of the indoor population factors for different roads is explained below.

9.6.24 Tuen Mun Road is located underneath the WTW and constitutes a significant portion of risk in comparison to other roads. For the hazard scenarios with some longest hazard range, such as those resulted from 8tonne instantaneous release, LD90 contour is approximately 250m from the release point. LD50 is 250m further away from LD90, and LD03 is 500m further away from LD50 (see Annex B). Therefore the total distance of Tuen Mun Road affected by LD90, LD50 and LD03 are 500m, 500m and 1000m respectively.

9.6.25 It is assumed that under LD90 chlorine concentration, vehicle passengers are exposed in equivalence to 0% indoor. Vehicle itself provides no protection under such a high chlorine concentration. For LD50 the assumption is slightly relaxed and passengers are assumed to be 50% indoor. For LD03, 75% indoor is assumed. At low concentration of LD03, vehicles can provide a reasonably good protection for passengers if they stay within. However, 25% outdoor is again assumed for some passengers attempting to escape out of cars.

9.6.26 Based on the discussion, the equivalent indoor proportion for Tuen Mun Road vehicle users is:

(500x0% + 500x50% + 1000x75%)/(500+500+1000) = 50%

9.6.27 All other roads which are further away from the WTW are assumed to be affected by LD03 at most under the worst case scenarios and the assumption of 75% indoor is applied generically.

Consequence Analysis Results

Initial Release of Chlorine

9.6.28 The results of the 'source term' modelling of chlorine releases is summarised in Table 9-16 below according to ERM (2001).

Table 9-16 Summary of Source Term Modelling Results for Yau Kom Tau WTW

Release case	Hole size (mm)	Phase	Mode of release to atmosphere (for internal release cases only)	Release rate to atmosphere or instantaneous release quantity	Release duration (Note 1)
External releases (1 tonne drum)
Small leak	3	Liquid	-	0.2 kg/s	83 min
Medium leak	8	Liquid	-	1.4 kg/s	12 min
Multiple medium leaks	6 (x6)	Liquid	-	4.8 kg/s	3.5 min
Large leak	20	Liquid	-	8.8 kg/s	114 s
Rupture	-	Liquid	-	1000 kg	-
Internal releases (1 tonne drum or chlorine pipework)
Pigtail - guillotine failure	4.5	Two-phase	Normal ventilation remains on	0.027 kg/s	10 min (Note 2)
			Door left open	0.011 kg/s	10 min
Medium leak from drum	8	Liquid	Normal ventilation remains on	0.30 kg/s	10 min
			Door left open	0.13 kg/s	10 min
Large leak from drum	20	Liquid	Normal ventilation remains on	0.56 kg/s	10 min
			Door left open	0.29 kg/s	10 min
Rupture	-	Liquid	Pressurisation of chlorine store - release via weak points (Note 4)	5.6 kg/s	10 s (Note 3)

Note 1: assumes no intervention by operating staff

Note 2: upper limit of 10 min set for duration of releases from chlorine building (by which time action would be taken to shut-off ventilation, close doors etc.)

Note 3: assumed release duration for catastrophic failure of a drum, eg a split along a weld (QRA not sensitive to this assumption)

Note 4: 'Normal ventilation remains on' and 'Door left open' are not included for this mode of release since 'Pressurisation of chlorine store - release via weak points' will be more dominant in the QRA.

9.6.29 It is apparent from Table 9-16 that releases from a drum due to melting of the fusible plugs (4.8 kg/s) or dislodgement of a plug (8.8 kg/s) occur sufficiently rapidly to cause emptying of the drum in a short period of time (within a few minutes). Therefore these release cases are treated as effectively instantaneous releases.

9.6.30 It is also apparent that the chlorine building has a significant effect in modifying the release of chlorine to the atmosphere, given failure of the Contain and Absorb system. The rate of chlorine release is reduced dramatically (eg for a medium leak the rate of chlorine to atmosphere is reduced from 1.4kg/s to 0.30kg/s or 0.13kg/s) and the chlorine becomes diluted in the building air. The failure mode of the Contain and Absorb system 'Normal ventilation remains on' is a more severe case than 'Door left open' in terms of the chlorine release rate to atmosphere. This is because the normal ventilation (typically 2.6 air changes per hour) provides a more rapid release of chlorine to the environment than if a door is left open (normal ventilation shutdown, chlorine scrubber system in operation).

Dispersion of Chlorine in the Atmosphere

Wind Tunnel Modelling Results

9.6.31 The results of the wind tunnel testing for Yau Kom Tau WTW are summarised below in Table 9-17 and Figures B.1 to B.13 in Appendix 9-B.

Table 9-17 Summary of Wind Tunnel Tests Results for Yau Kom Tau WTW

Release case	Release location	Description	Weather Class	Wind directions (Note 1)	Maximum extent of LD03 contour (m) (Note 2)
50 kg (vapour) instantaneous	Chlorine store	Chlorine vapour release due to rupture of chlorine drum leading to pressurisation of store and release via weak points.	D2	S E N W	195 No LD03 contour off-site 169 273
1 tonne (aerosol) instantaneous	Access Road	Catastrophic failure of a chlorine conrtainer	D2	S SE E NE N NW W SW	435 468 520 663 390 468 585 696
1 tonne (aerosol) instantaneous	Access Road	Catastrophic failure of a chlorine container	D5	W NE	332 299

Note 1: Direction which wind blows from

Note 2: Downwind distance to 3% nominal outdoor fatality probability, ie not taking into account escape and assuming 10 min exposure duration or cloud passage time (whichever is the shorter)

9.6.32 From the results in Table 9-3 above, Appendix 9-B and RWDI (1998), the key findings of the wind tunnel testing may be summarised. The downwind hazard range is significant for both internal 1 tonne instantaneous releases (modelled as 50kg (vapour) instantaneous releases) (169-273m) and external 1 tonne instantaneous releases (390-696m). Topography significantly influences the dispersion characteristics of chlorine releases at Yau Kom Tau WTW, as manifested by:

· flow reversal due to creation of a wake zone at the site of the WTW for winds from the N (ie over the ridge) (Figures B2 and B8 in Annex B);

· curtailment of dispersion up the slope on which the WTW is situated leading to shorter, broader clouds (Figures B4 and B5 in Annex B);

· channelling of chlorine releases along Tuen Mun Road and around nearby high rise estates (Figures B6, B9 and B10 in Annex B).

9.6.33 The wind tunnel testing for Yau Kom Tau WTW did not include the 1.4 kg/s continuous release case, as was done for Sha Tin, Pak Kong and Tai Po Tau WTWs. The reason for this is that an independent technical review of the wind tunnel testing (Webber, 1998) highlighted the limitation of modelling this type of release in the wind tunnel (1:500 scale) due to the difficulty of accurately simulating turbulence close to the ground near the source of the release. The QRA for Yau Kom Tau WTW therefore makes use of the results of the CFD modelling for this type of release, described below.

CFD Modelling

9.6.34 CFD modelling was performed for Sha Tin WTW and Tai Po Tau WTW. Details can be found in ERM (2001), HSL (1998a) and HSL (1998b). The summary of these results is provided in Table 9-18 and the LD contours are shown in Appendix 9-C.

Table 9-18 Summary of CFD Modelling Results

Release case	Weather class	Maximum Extent of LD Cobtour (m)
		LD90	LD50	LD03
Sha Tin WTW
1.4 kg/s continuous	D2	110	140	205
	F2	145	155	225
1 tonne instantaneous	D2	170	200	255
	F2	220	255	275
Tai Po Tau WTW
1.4 kg/s continuous	D2	135	165	265
	F2	130	180	330
1 tonne instantaneous	D2	200	215	255
	F2	180	255	355
	B2	75	95	105

9.6.35 The key findings of the CFD modelling may be summarised as follows:

· atmospheric stability does not significantly influence the hazard range of either a 1.4 kg/s continuous release of chlorine or a 1 tonne instantaneous release of chlorine for the two weather conditions of most interest in this study (ie D - neutral stability and F - stable conditions). This is because, in the presence of buildings and complex, heavily-vegetated terrain, atmospheric stability has less of an influence on chlorine dispersion;

· for B (unstable conditions) the CFD results for Tai Po Tau WTW indicate that the chlorine hazard range is significantly reduced compared to neutral conditions (ie a factor of 2.5 shorter for a 1 tonne instantaneous release). It is indicated that this is due to the unstable wind field which significantly enhances vertical dispersion of the chlorine. However, as B conditions account for no more than 20% of the weather in Hong Kong, this is not considered a significant factor for the QRA (ie risks are not considered to be significantly overestimated by ignoring B conditions); and

· for F (stable conditions) the CFD results for Tai Po Tau WTW indicate that, whilst the chlorine hazard range is not significantly affected by atmospheric stability, the direction of travel of the chlorine cloud may be affected. At Tai Po Tau WTW, the chlorine releases in F conditions more closely followed the topographic contours than the equivalent releases in D conditions, which followed the direction of the wind.

Flat Terrain Dispersion Modelling Results

9.6.36 The results of the flat terrain dispersion modelling using DRIFT are summarised in Table 9-19 below. Annex A contains the results of all the DRIFT runs undertaken.

Table 9-19 Summary of DRIFT Modelling Results

Release case	Weather class	Downwind distance (m)
		LD90	LD50	LD03
0.2 kg/s continuous	D2	86	119	182
1.4 kg/s continuous	D2	268	362	550
1 tonne instantaneous	D2	325	425	600
3 tonne instantaneous	D2	586	735	1044
10 tonne instantaneous	D2	1004	1286	1790

9.6.37 From the results in Table 9-18 it is possible to derive a relationship between the chlorine release rate (or release quantity) and the downwind hazard range. This is shown graphically in Figures A1 and A2 in Appendix 9-A. This relationship is used in the QRA, as described below.

Comparison of Results of Wind Tunnel Testing and DRIFT Modelling

9.6.38 Table 6.1shows the relationship between the chlorine concentration and the probability of fatality for the TNO probit, assuming 10 minute exposure duration.

· the chlorine hazard range predicted by the wind tunnel testing is generally shorter than that predicted by the DRIFT flat terrain dispersion modelling. This highlights the importance of modelling the effects of buildings and complex terrain, which act to increase turbulence and cause greater mixing of the chlorine. (It should also be noted that there is an inherent limitation in models such as DRIFT, whereby the surface roughness chosen must be small in relation to the cloud height. For dense gas release this limits the scope of DRIFT-type simulations to relatively smooth terrain, which is not applicable to Hong Kong conditions);

· the hazard range predicted by the wind tunnel for 1 tonne instantaneous releases are greater than those predicted by the CFD modelling for Sha Tin WTW and Tai Po Tau WTW. The reason for this is not clear, however, as the wind tunnel results err on the conservative side (whilst eliminating the pessimism in the DRIFT-type predictions for these releases) they will be used in preference in the QRA.

Rationalisation of Chlorine Dispersion Modelling Results

9.6.39 The preceding sections have discussed the results arising from the various strands of work on chlorine dispersion modelling. The following paragraphs summarise how these results will be applied in the QRA. More details are provided in Appendix 9-D.

9.6.40 Wind tunnel testing: the wind direction-specific cloud shapes generated in the wind tunnel will used directly in the QRA. This will achieved through use of Graphical Information Systems (GIS) software which is described in further detail below. Another output of the wind tunnel testing was the influence of wind speed on the chlorine hazard range. From the wind tunnel test results for all eight WTWs a simple scale factor was derived to modify the cloud contours for the 2m/s wind speed case to determine those for the 5m/s case.

9.6.41 CFD modelling: the CFD modelling results show no significant influence of atmospheric stability on the chlorine hazard range (for D and F conditions), therefore this parameter is not considered further in the QRA. However, the CFD results for the 1.4 kg/s continuous release case (D2 weather conditions), which are consistent for Sha Tin WTW and Tai Po Tau WTW, are used in the QRA for YKT WTW.

9.6.42 DRIFT modelling: the DRIFT flat terrain dispersion modelling results are not used directly in the QRA. However the relationships derived from the DRIFT modelling for the chlorine release rate/quantity versus hazard range are used to scale the wind tunnel results for the complete range of release scenarios which need to be considered in the QRA. The method of simple, uniform scaling provides a sufficiently accurate approach where the terrain surrounding the WTW is relatively flat. However, where there are significant topographic features, then the uniform scaling approach provides only a crude approximation to the cloud shape. In these situations, where the uniform scaling approach does not provide sufficient accuracy, the topographic features (e.g. hills) have been taken into account by allowing them to modify the shape of the chlorine cloud, whilst preserving the cloud area.

9.6.43 One further aspect which needs to be considered in applying the results of the wind tunnel testing in the QRA is the number of individual wind tunnel directions which are considered. In the wind tunnel testing up to eight wind directions were typically modelled for the most important release scenarios (e.g. a 1 tonne instantaneous release). However, in a QRA, it is usually necessary to consider a greater number of possible directions, in order to eliminate any spurious, numerical error in the risk results. The process of interpolating between the modelled wind directions is called 'wind smoothing' (achieved mathematically in software such as RISKPLOT the Consultants' proprietary risk integration tool).

9.6.44 The application of wind smoothing in this study was considered in detail in Technical Note 1 (ERM, 1998). It was concluded that for sites with relatively flat surrounding terrain, wind smoothing could be achieved by the simple method of cloud 'rotation' (ie rotation of clouds to fill the directional 'gaps' left by the wind tunnel). However for sites with complex terrain and/or high rise buildings, such as Yau Kom Tau WTW, it would be necessary to demonstrate that the important effects of the topography and buildings had been adequately captured in the raw wind tunnel data. However, in the case of Yau Kom Tau WTW sufficient number of wind directions have been modelled in the wind tunnel to capture the important topographic influences.

Chlorine Cloud Height

9.6.45 Information on the height of a chlorine cloud has been obtained from the wind tunnel simulations (Sha Tin WTW), CFD modelling (Sha Tin WTW and Tai Po Tau WTW) and DRIFT flat terrain dispersion modelling. This is useful for determining the degree of protection of people inside high rise buildings. The data, presented in detail in Appendix 9-E, have been rationalised for use in the QRA as shown in Table 9-19 below.

Table 9-1920 Chlorine Cloud Heights

Release case	Chlorine cloud height (m) (Note 1)	Equivalent number of storeys (Note 2)
1.4 kg/s continuous	30	10
1 tonne instantaneous	6	2
10 tonne instantaneous	9	3

Note 1: Note that this is not the full height of the chlorine cloud. It is the height up to which the ground level chlorine concentration is assumed to apply for the purpose of calculating numbers of fatalities in tall buildings.

Note 2: Assumes 3m per storey

9.7 Rationalisation of Chlorine Release scenarios And estimation of Scenario Frequencies

Rationalisation of Chlorine Release Scenarios

9.7.1 The consequence analysis from wind tunnel testing and CFD modelling shows that only certain, severe types of chlorine release which could produce fatal off-site concentrations of chlorine (refer to Tables 9-17 and 9-18). The release cases which fall into this category are external continuous releases of 1.4 kg/s or more (equivalent to guillotine failure of a drum valve) and instantaneous releases of 1 tonne or more whether external or internal.

9.7.2 Appendix 9-F summarizes the results of modelling of escape from chlorine cloud.

9.7.3 These results mean that some of the chlorine release scenarios identified in Section 5 (Table 9-8) can be eliminated from further consideration in the QRA. Table 9-21 considers each release scenario in turn and, based on the results of the consequence analysis, determines whether the scenario poses an off-site hazard. Table 9-22 then summarises the results of the analysis in Table 9-21 by grouping the release scenarios into 'events' having identical release characteristics (ie the same release rate, duration and phase of release), eg

· 'truck fire - 3 new drums' (2.4 kg/s) is grouped together with 'rollover - 3 drums medium leak' (4.2 kg/s) into RU1TMML (4.2 kg/s). This is a simplification erring on the conservative side for the truck fire release;

· the group RU1TSML (1.4 kg/s) has the following release scenarios grouped together: 'rollover - single drum medium leak', 'loadshedding - single drum medium leak' and 'spontaneous container failure - single drum medium leak'..

· the group RU1TSRU (1 tonne) has the following release scenarios grouped together: 'truck fire - 2 new drums/1 old drum' (1.6 kg/s plus 4.8 kg/s), 'collision - single drum rupture' (1 tonne), 'spontaneous container failure - single drum large leak' (8.8 kg/s), and 'spontaneous container failure - single drum rupture' (1 tonne). The truck fire scenario was put in this group since the leak from the 'old' drum (4.8 kg/s) empties the drum so quickly such that it can be considered to be an instantaneous release from the drum. Similarly so for 'spontaneous container failure - single drum large leak' (8.8 kg/s).

9.7.4 Table 9-21 shows the events grouped according to the leak quantity. The classification as a “significant off-site hazard” is based on the results of consequence analysis (refer to Section 9.6).

Table 9-21 Rationalisation of Chlorine Release Scenarios

Chlorine release scenario	Outcome	Hole size	Phase	Chlorine release rate from primary source (kg/s)	Chlorine release quantity (tonnes)	Chlorine release rate (or quantity) to atmosphere	Significant off-site hazard ? (Y/N)	Event Ref^(2)
1. ACCESS ROAD
1.1 Truck fire	Considered to result in melting of the fusible plugs on up to three drums. Three cases are considered according to the number of 'old' and 'new' drums: Case (i): 3 new drums, no old drums Case (ii): 2 new drums, 1 old drum Case (iii): No new drums, 3 old drums	3x6mm 2x6mm (new drums) 6x6mm (old drums) 18x6mm	liquid liquid liquid	2.4 1.6 4.8 14.4	3 2 1 3	2.4 kg/s^(3) 1.6 kg/s 4.8 kg/s^(4) 14.4 kg/s^(4)	Y Y Y	RU1TMML RU1TSRU RU1TMRU
1.3 Manoeuvring accident	Single drum - small leak	3mm	liquid	0.2	1	0.2 kg/s	N	-
1.4 Rollover	Single drum - small leak Single drum - medium leak	3mm 8mm	liquid liquid	0.2 1.4	1 1	0.2 kg/s 1.4 kg/s	N Y	- RU1TSML
	Three drums - medium leak	3x8mm	liquid	4.2	3	4.2 kg/s	Y	RU1TMML
	Fire (outcomes as item 1.1 above)
1.5 Collision	Single drum - rupture Fire (outcomes as item 1.1 above)	-	liquid	-	1 (inst)	1 tonne	Y	RU1TSRU
1.6 Loadshedding	Single drum - small leak	3mm	liquid	0.2	1	0.2 kg/s	N	-
	Single drum - medium leak	8mm	liquid	1.4	1	1.4 kg/s	Y	RU1TSML
1.7 Spontaneous container failure	Single drum - medium leak	8mm	liquid	1.4	1	1.4 kg/s	Y	RU1TSML
	Single drum - large leak	20mm	liquid	8.8	1	8.8 kg/s^(4)	Y	RU1TSRU
	Single drum - rupture	-	liquid	-	1(inst)	1 tonne	Y	RU1TSRU
2. DRUM HANDLING
2.1 Dropped drum	Single drum - medium leak	8mm	liquid	1.4	1	0.30 kg/s	N	-
	Single drum - large leak	20mm	liquid	8.8	1	0.56 kg/s	N	-
	Single drum - rupture	-	liquid	-	1 (inst)	57 kg	Y	IU1TSRU
2.2 Collision of drum with another object	Single drum - medium leak	8mm	liquid	1.4	1	0.30 kg/s	N	-
2.3 Accidental impact of drum on pigtail during setdown at standby position	Pigtail - guillotine failure	4.5mm	two-phase	0.12	1	0.027 kg/s	N	-
2.4 Dropped drum due to incorrect alignment of monorail track or overextension of truck crane	As item 2.1 above
3. CONTAINERS IN STORAGE
3.1 Leaking chlorine drums or Scenarios , 3.11, 3.15 of Table 5.5 which are considered to result in a medium leak only)	Single drum - medium leak	8mm	liquid	1.4	1	0.30 kg/s	N	-
	Single drum - large leak	20mm	liquid	8.8	1	0.56 kg/s	N	-
	Single drum - rupture	-	liquid	-	1 (inst)	56 kg	Y	IU1TSRU
3.2 Overfilled drums leading to overpressurisation on thermal expansion	As item 3.1 above
3.3 Impurities in chlorine drum leading to explosion or leak	As item 3.1 above
3.4 Object falls onto chlorine containers	Single drum - medium leak	8mm	liquid	1.4	1	0.30 kg/s	N	-
3.6 External Explosion	As item 3.4 above
3.7 Lightning Strike	As item 3.4 above
3.11 Subsidence	As item 3.4 above
3.12 Earthquake	Roof collapse: multiple drum-rupture	-	liquid	-	11 (inst)	7.7 tonnes^(1)	Y	EU1TMRU EU1TMRUH
3.12 Aircraft crash	Roof collapse: multiple drum-rupture (similar to earthquake)	-	liquid	-	11 (inst)	7.7 tonnes^(1)	Y	AU1TMRU
4. CONNECTION AND DISCONNECTION OF CHLORINE CONTAINERS
4.1 Human error or equipment failure during connection/disconnection of drums	Pigtail - guillotine failure	4.5mm	two-phase	0.12	1	0.027 kg/s	N	-
5. CHLORINATION SYSTEM
5.1 - 5.5 Failures associated with the chlorination system pipework	Liquid chlorine pipework - guillotine failure	4.5mm	two-phase	0.12	1.05	0.027 kg/s	N	-
5.6 - 5.7 Failure of Evaporator	Evaporator - leak or rupture	4.5mm	two-phase	0.12	1.05	0.027 kg/s	N	-

Note (1): For large instantaneous releases, 70% of the chlorine is estimated to be released instantaneously to atmosphere as vapour and entrained aerosol. This comprises the initial vapour flash fraction (19%) plus the entrained aerosol (2 x 19%) plus the contribution from the evaporating chlorine pool over the first minute (around 10% depending on the pool size).

Note (2): Key to event ref

E R (Road) or E (Earthquake) or A (Aircraft crash) or I (internal)

U U (Unisolated) or I (Isolated)

1T 1T (1 tonne drums) or 50 (50 kg cylinders)

M S (Single) or M (Multiple)

RU RU (Rupture), LL (Large Leak), ML (Medium Leak) or SL (Small Leak)

H Earthquake of higher ground acceleration

Note (3): This release (2.4 kg/s) treated as a multiple medium leak for simplification (conservative assumption)

Note (4): These releases treated as effectively instantaneous releases due to short release duration

Table 9-22 Release Scenarios Included in YKT WTW QRA

Event Ref	Component scenarios	Release rate (or quantity) to atmosphere	Type of release	Release location
IU1TSRU	Spontaneous failure Dropped drum	56 kg	Instantaneous	Chlorine store
RU1TSML	Rollover Loadshedding Spontaneous failure	1.4 kg/s	Continuous	Access road
RU1TMML	Rollover Truck fire	4.2 kg/s	Continuous	Access road
RU1TSRU	Truck impact Truck fire Spontaneous failure	1 tonne	Instantaneous	Access road
RU1TMRU	Truck fire	3 tonnes	Instantaneous	Access road
EU1TMRU	Earthquake: roof collapse, ground acceleration 0.5g	7.7 tonnes	Instantaneous	Chlorine store
EU1TMRH	Earthquake: roof collapse, ground acceleration 0.8g	7.7 tonnes	Instantaneous	Chlorine store
AU1TMRU1G	Aircraft crash: roof collapse,	7.7 tonnes	Instantaneous	Chlorine store

Table 9-23 Release Scenarios Categorised by Leak Quantity

Leak Quantity (kg)	Event Ref
0-10	None
10-100	IU1TSRU
100-1000	RU1TSML, RU1TMML, RU1TSRU
>1000	RU1TMRU, EU1TMRU, EU1TMRUH AU1TMRU

Frequency Estimation

Base Case of WTW QRA

9.7.5 Having identified the chlorine release scenarios of interest (Section 9.5) the next step in the Hazard Assessment is to determine their frequency of occurrence. This is based on the approach outlined in the Consultants' Methodology Report (ERM, 1997a).

9.7.6 In light of the recent judgement of the Court of Final Appeal (FACV 28/2005) that methodology has been carefully reviewed for the purpose of the current assessment. In particular the latest version of the world wide accident database MHIDAS (Major Hazard Incident Data Service) has been independently reviewed. However, only a few relevant chlorine incidents occurred worldwide since the previous review, and after examination of their nature it has been concluded that no significant revisions of the previous event frequencies are necessary for this study.

9.7.7 It should be noted that for the sake of consistency and to allow for a potential increase in the future, the estimated frequencies of truck-related incidents are still based on the 1187 tonnes per year chlorine use assumed in the previous assessment, while according to the latest information the current chlorine use at YKT WTW is only 100 tonnes per year (refer to Table 9-1). Similarly, as discussed in Section 9.5, the “old type” chlorine containers, that can release more chlorine in accidents are still accounted for in the present assessment, even that they are currently not in use.

9.7.8 Table 9-24 summarises the base data which has been used in the frequency calculations. Appendix 9-I contains the details of how this data has been used to determine the frequency of each of the events listed in Table 9-22. Table 9-25 summarises the results of the frequency calculations.

Table 9-24 Base Failure Rate Data

Data item	Value	Units	Source
1. Chlorine container
1.1 (i) Spontaneous container failure frequency (ii) Conditional probability of catastrophic failure (iii) Conditional probability of medium leak (iv) Conditional probability of large leak	1.5E-4 0.027 0.22 0.081	per year - - -	Methodology Report (based on review of worldwide failure data for chlorine containers and generic pressure vessel failure data)
1.2 (i) Probability of dropped container (ii) Conditional probability of catastrophic failure	7.7E-6 1.0E-4	per lift -	Methodology Report (based on Hong Kong data for number of lifts which have occurred without incident)
2. Chlorine delivery vehicle
2.1 (i) Frequency of load shedding (ii) Conditional probability of a medium leak	1.1E-7 6.3E-2	per truck-km -	Chlorine Transport Risk Study, DNV (1997)
2.2 (i) Frequency of rollover (ii) Conditional probability of a medium leak of a single drum (iii) Conditional probability of medium leak of multiple drums	1.9E-7 1.5E-1 1.1E-2	per truck-km - -	Chlorine Transport Risk Study, DNV (1997)
2.3 (i) Frequency of vehicle impact (ii) Conditional probability of drum rupture	4.0E-7 1.7E-2	per truck-km -	Chlorine Transport Risk Study, DNV (1997)
2.4 (i) Frequency of spontaneous truck fire (ii) Probability of having 2 or 3 'old' drums among the 3 ruptured drums (iii) Probability of having 1 'old' drum among the 3 ruptured drums (iv) Probability of having nil 'old' drums among the 3 ruptured drums	4.0E-9 0.061 0.33 0.61	per truck-km - - -	Chlorine Transport Risk Study, DNV (1997)
3. External events
3.1 (i) Frequency of earthquake of 0.5g ground acceleration^(1) (ii) Probability of roof collapse in an earthquake of 0.5g ground acceleration	3.3E-6 0.1	per year -	Cook et al (1993) Water Treatment Works Seismic Hazard Assessment, Ove Arup (2000)
3.2 (i) Frequency of earthquake of 0.8g ground acceleration^(2) (ii) Probability of roof collapse in an earthquake of 0.8g ground acceleration	1.4E-7 0.5	per year -	Cook et al (1993) Water Treatment Works Seismic Hazard Assessment, Ove Arup (2000)
3.3 Frequency of aircraft crash	1.2E-8	per landing	Based on US National Transportation Safety Board aircraft crash data (Appendix 9-H)

Note 1: Approximately equivalent to MMX

Note 2: Approximately equivalent to MMXI-XII

Note 3: Seismic Hazard Assessment is attached in Appendix 9-G

Table 9-25 Event Frequencies - Base Case of WTW QRA

Event Ref	Component scenarios	Frequencies (per year)	Time periods during which event could occur
IU1TSRU	Dropped drum Spontaneous drum failure	1.83E-6 2.60E-4	All except Night All
	Total	2.62E-4
RU1TSML	Rollover Loadshedding Spontaneous failure	1.13E-6 2.75E-7 1.79E-7	All except Night All except Night All except Night
	Total	1.58E-6
RU1TMML	Rollover Truck fire	8.29E-8 9.73E-8	All except Night All except Night
	Total	1.80E-7
RU1TSRU	Truck impact Truck fire Spontaneous drum failure Total	2.69E-7 5.15E-8 8.78E-8 4.08E-7	All except Night All except Night All except Night
RU1TMRU	Truck fire	9.63E-9	All except Night
EU1TMRU	Earthquake	3.30E-7	All
EU1TMRUH	Earthquake	7.00E-8	All
AU1TMRU	Aircraft crash	1.36E-8	All

Frequencies used in the Present Study

9.7.9 Following the QRA for the base case WTW operation, a number of possible mitigation measures were discussed and three of them recommended for implementation in order to reduce the societal risk As Low As Reasonably Practicable (see ERM, 2001).

9.7.10 These mitigation measures included provision of a containment barrier at the S boundary of the WTW site, improvements to the access road and improved protection to the WTW staff quarters. Since such measures had been since implemented, the present study is based on the mitigated scenario of ERM (2001). Thus, in accordance with the mitigated scenario of the previous study, the event frequencies for RU1TSRU and RU1TMML have been reduced by 33% and those for RU1TSML by 45% as compared to those listed in Table 9-25 and Annex 9-I. The frequencies for all other events remain unchanged.

9.7.11 The event frequencies that are used in this study, ie basically those used for the mitigated case in of the previous assessment are listed in Table 9-26 for each of the time periods considered in the study.

Table 9-26 Frequencies (times per year) of Chlorine Release Events used in the Present Study

Event Ref	Peak Hour	Jammed Peak	Night	Working Day	Weekend Day
IU1TSRU	3.52E-05	2.34E-06	1.30E-04	6.88E-05	2.50E-05
RU1TSML	2.33E-07	1.55E-08	0.00E+00	4.56E-07	1.66E-07
RU1TMML	3.23E-08	2.14E-09	0.00E+00	6.32E-08	2.30E-08
RU1TSRU	7.34E-08	4.88E-09	0.00E+00	1.44E-07	5.22E-08
RU1TMRU	2.58E-09	1.71E-10	0.00E+00	5.04E-09	1.83E-09
EU1TMRU	4.42E-08	2.94E-09	1.65E-07	8.64E-08	3.14E-08
EU1TMRUH	9.37E-09	6.23E-10	3.50E-08	1.83E-08	6.66E-09
AU1TMRU	1.82E-09	1.21E-10	6.80E-09	3.56E-09	1.29E-09

9.8 Quantitative Risk Assessment

Risk Assessment Methodology

9.8.1 The QRA combines information on the consequences of chlorine releases with information on the likelihood of releases to generate two measures of risk -individual risk and societal risk. Individual risk is the chance of death per year to a specified individual at a specific location. Societal risk is the risk to the population as a whole.

9.8.2 The QRA will be undertaken using a GIS-based software application, GISRisk developed specifically for the WTW project. The GIS component of the software enables the complex cloud shapes generated by the wind tunnel to be input directly into the QRA. GISRisk is an application of standard, well-validated, commercial software, ie ESRI's ARCVIEW GIS software, Microsoft Access and Microsoft Excel.

9.8.3 Associated with the GIS software is a database containing all the relevant information relating to the WTW, the defined events, the meteorological data, the population data and the chlorine cloud coordinates. The database contains the routines for the calculation of individual and societal risk.

9.8.4 For the purposes of the present study the database pertinent to the ERM-HK (2001) case for the 2006 population and assuming implementation of the recommended mitigation measures has been modified to include the latest population estimates (refer to Section 9.4). The main outputs from the software are as follows:

· Individual risk in the form iso-risk contours overlaid on a base map of the area;

· Societal risk in the form of an FN curve, which is a graphical representation of the cumulative frequency (F) of N or more fatalities plotted against N on a log-log scale; and

· Societal risk in the form of a Potential Loss of Life (PLL) value, which expresses the risk to the population as a whole and for each scenario and its location.

The PLL is an integrated measure of societal risk obtained by summing the product of each f-N pair, as below:

PLL = f₁N₁+f₂N₂+f_nN_n....

Scenarios Studied

9.8.5 As outlined in Section 4.2.2, the QRA is conducted in this Study for a number of scenarios differing by their population assumptions. Five detailed population data sets reflecting the 2006 projections of ERM (2001) and various estimates for the present study have been discussed in Section 4.2 and listed in Table 4.2. Note that while the road populations could be reliably estimated based on the detailed Traffic Impact Assessment forecasts for the project, adequate general population projections are not available so our estimates could only be based on publicly available data such as Planning Department forecasts (PlanD, 2006) and the information gathered during the site visits. Thus, while our population projections form a basis for the principal scenarios, as a control exercise we also repeat the model runs using the same general population data as in the previous study (ERM, 2001).

9.8.6 Note that all these cases differ only by their population assumptions. All other assumptions of the study remain the same for all scenarios. The following scenarios are being considered:

· Case 1 is a scenario using the 2006 population estimates of the previous assessment (ERM, 2001). It is included here as a background case, for comparison with the other scenarios;

· Case 2 uses the same general population data as Case 1, but the road populations reflect the traffic projections for 2030, after the completion of the project;

· Case 3 is similar to Case 2 but the road populations reflect the 2015 construction phase of the project;

· Case 4 uses the updated general population data and the 2030 projection for the road populations; and

· Case 5 involves the updated general population data, but concerns the 2015 construction phase of the project.

· Societal risk results for the Tseun Wan Rd population only, during the construction and operational phase of the Project are also presented

Risk Criteria

9.8.7 The Hong Kong Planning Standards and Guidelines (HKPSG), Chapter 12 require that Potentially Hazardous Installations (PHIs) comply with Government Risk Guidelines. The same risk guidelines are stipulated under EIAO in Annex 4 of Technical Memorandum on Environmental Impact Assessment Process (EIAO-TM). Acceptable risk levels are defined as follows:

· Individual Risk: The maximum involuntary individual risk of death associated with accidents arising at PHIs should not exceed 1 chance in 100,000 per year (10^-5/yr).

· Societal Risk: The societal risk associated with a PHI should comply with the FN diagram (Fig 3 in HKPSG, Ch 11, also shown here in Figures 8.1 and 8.2). Three areas of risk are shown:

(i) Acceptable where risks are so low that no action is necessary;

(ii) Unacceptable where risks are so high that they should usually be reduced regardless of the cost or else the hazardous activity should not proceed; and

(iii) ALARP (As Low As Reasonably Practicable) where the risks associated with each probable hazardous event at the PHI should be reduced to a level ‘as low as reasonably practicable’, usually measured as a trade off between the risk reduction afforded and the cost of that reduction. Risk mitigation measures may take the form of either engineered measures at the PHI or development (ie population) controls in the vicinity of the PHI.

Societal Risk Results

FN Curves

9.8.8 The societal risk results for the areas surrounding YKT WTW (refer to Figure 9-5) are presented in Figure 9-6 and 9-7 in the form of FN curves. Both the “background” general population (Case 2 and Case 3, Figure 9-6) and the “updated” general population scenarios (Case 4 and Case 5, Figure 9-7) are compared with the HKPSG Risk Guidelines and the Case 1 results, for the “background” scenario of ERM (2001). The FN curves for the construction and operation phases excluding the surrounding populations are also included in Figure 9-6 and repeated in Figure 9-7. These include the Tsuen Wan Rd population only, and for the construction phase also the temporary project office (items 35 and 40 of Table 9-3)

9.8.9 As can be seen, the shape of the FN curves is mostly influenced by the general population projections – the cases differing only by the assumed road populations (Cases 1-3 in Figure 9-6 and Cases 4 and 5 in Figure 9-7) have very similar FN curves, with the only exception of F for N=1, which in both figures is significantly lower for the construction phase scenarios. An analysis of the detailed results suggests that the difference for N=1 is a result of the short-range IU1TSRU event and the different road populations of Tuen Mun Road assumed for the construction and operational phases. For other N values neither the constructional nor operational phase of the project has a great influence on societal risk results.

9.8.10 On the other hand, the introduction of the “updated” general populations that besides adding a few new residential development involved reduction of the population of WTW Staff Quarters, Yau Kom Tau squatters and removing the population of Tsuen Wan Bay Further Reclamation resulted in a significant lowering of the F levels in the 100 to 800 N interval.

9.8.11 Nevertheless, similar to the ERM (2001) results, for N greater than 20 and less than about 800 all of these FN curves exceed are in the ALARP criterionregion, and, but remain well below the “Unacceptable” levels.

9.8.12 The FN results for the Tsuen Wan Rd population only (that for the construction phase includes both the project workforce, and the vehicular population) are low, and well within the “Acceptable” zone.

9.8.13 The details of FN curve for the most relevant Case 4 (operational phase of the Project) are given in Table 9-27, which also shows the breakdown of the F values between the different chlorine release scenarios (refer to Table 9-22).

9.8.14

Table 9-27 Breakdown of the Case 4 FN Data by Release Scenario

N	F :Total	EU1TMRU	EU1TMRUH	AU1 TMRU	RU1 TMML	RU1TSML	RU1 TMRU	RU1 TSRU	IU1 TSRU
1	3.14E-06	3.30E-07	6.98E-08	1.36E-08	5.96E-08	2.31E-07	9.53E-09	1.79E-07	2.25E-06
2	8.47E-07	3.30E-07	6.95E-08	1.36E-08	5.05E-08	1.98E-07	9.51E-09	1.76E-07
3	8.07E-07	3.28E-07	6.64E-08	1.36E-08	4.73E-08	1.81E-07	9.45E-09	1.61E-07
4	7.61E-07	3.24E-07	6.54E-08	1.34E-08	4.44E-08	1.56E-07	9.37E-09	1.49E-07
5	7.36E-07	3.21E-07	6.50E-08	1.33E-08	4.40E-08	1.42E-07	9.31E-09	1.41E-07
6	7.04E-07	3.07E-07	6.32E-08	1.31E-08	4.27E-08	1.35E-07	9.14E-09	1.34E-07
7	6.87E-07	3.07E-07	6.18E-08	1.31E-08	4.21E-08	1.31E-07	8.91E-09	1.24E-07
8	6.59E-07	3.02E-07	6.18E-08	1.27E-08	4.16E-08	1.13E-07	8.79E-09	1.19E-07
9	6.39E-07	2.92E-07	6.18E-08	1.26E-08	4.14E-08	1.08E-07	8.57E-09	1.15E-07
10	6.33E-07	2.92E-07	6.18E-08	1.24E-08	4.10E-08	1.05E-07	8.29E-09	1.13E-07
20	5.07E-07	2.67E-07	5.26E-08	1.16E-08	3.70E-08	3.13E-08	7.13E-09	1.01E-07
30	4.27E-07	2.45E-07	5.13E-08	1.05E-08	2.86E-08	3.95E-09	6.60E-09	8.16E-08
40	3.81E-07	2.32E-07	4.89E-08	9.97E-09	1.73E-08		5.77E-09	6.68E-08
50	3.60E-07	2.26E-07	4.65E-08	9.57E-09	1.01E-08		5.46E-09	6.23E-08
60	3.37E-07	2.20E-07	4.56E-08	9.27E-09	5.74E-09		4.90E-09	5.18E-08
70	3.21E-07	2.15E-07	4.44E-08	9.07E-09	1.67E-09		4.50E-09	4.62E-08
80	3.03E-07	2.09E-07	4.43E-08	8.88E-09	6.79E-10		4.24E-09	3.50E-08
90	2.88E-07	2.07E-07	4.27E-08	8.62E-09	4.57E-11		3.86E-09	2.50E-08
100	2.65E-07	2.01E-07	3.55E-08	8.34E-09			3.43E-09	1.61E-08
200	1.04E-07	8.57E-08	1.27E-08	4.30E-09			1.20E-09
300	4.90E-08	4.06E-08	6.00E-09	2.15E-09			2.72E-10
400	2.40E-08	1.99E-08	3.04E-09	1.10E-09			2.69E-12
500	1.27E-08	1.08E-08	1.29E-09	5.86E-10
600	4.65E-09	3.63E-09	6.37E-10	3.84E-10
700	2.16E-09	1.85E-09	1.68E-10	1.44E-10
800	1.00E-09	7.91E-10	1.17E-10	9.55E-11
900	8.90E-11	4.93E-11	7.07E-12	3.26E-11
1000	2.03E-12			2.03E-12

9.8.15 As can be seen, while the truck incident-related scenarios have some importance for the lower N values, practically only the two earthquake scenarios EU1TMRU and EU1TMRUH and aircraft crash AU1TMRU have a potential of causing 100 or more fatalities. The spontaneous failure or dropped drum event IU1TSRU is important for N=1 only.

PLL Values

9.8.16 The total Potential Loss of Life values obtained for different scenarios considered in this assessment are shown in Table 9-28.

Table 9-28 Total PLL Values

Population Scenario	PLL
Case 1	1.24E-04
Case 2	1.23E-04
Case 3	1.14E-04
Case 4	8.43E-05
Case 5	7.56E-05
Operation only	4.45E-07
Construction only	3.37E-07

9.8.17 As can be seen, the Case 1 and Case 2 PLL values are practically identical – with the same general population, the increase of the traffic population on some roads was apparently compensated by the decrease on others. On the other hand, the revision of the general population data resulted in a marked decrease for the predicted risk levels, with Case 4 and Case 5 PLL significantly lower than for Cases 2 and 3. Also, due to the lower road population levels, PLL for the construction phase is slightly lower than for the corresponding operational scenarios.

Table 9-29 Breakdown of Case 4 and Case 5 PLL by Population Area

Ref	Description	Case 4 PLL		Case 5 PLL
		per year	% of total	per year	% of total
33	Tuen Mun Rd	3.678E-05	43.6%	2.86E-05	37.8%
C	Greenview Court & Belvedere Garden	1.008E-05	12.0%	1.01E-05	13.3%
B	Yau Kom Tau Village Extension	8.31E-06	9.9%	8.31E-06	11.0%
E	Yau Kom Tau Squatters	7.065E-06	8.4%	7.07E-06	9.4%
J	Bus Terminal	2.832E-06	3.4%	2.83E-06	3.8%
W	New development at Po Fung Rd	2.74E-06	3.3%	2.74E-06	3.6%
K	Belvedere Garden, The Panorama	2.04E-06	2.4%	2.04E-06	2.7%
D	Gardenview & Blossom Terrace	1.99E-06	2.4%	1.99E-06	2.6%
I	Bayview Garden	1.58E-06	1.9%	1.58E-06	2.1%
V	Golden Villa + new development	1.35E-06	1.6%	1.35E-06	1.8%
O	Secondary Scool	1.17E-06	1.4%	1.17E-06	1.6%
N	HK Baptist Convention School	8.34E-07	1.0%	8.34E-07	1.1%
	- - - - -	- - -	- - -	- - -	- - -
34	Ting Kau Bridge	5.141E-09	0.0%	5.14E-09	0.0%
35	Tsuen Wan Rd	4.45E-07	0.5%	3.37E-07	0.5%
36	Castle Peak Rd S 1	2.29E-07	0.3%	2.29E-07	0.3%
37	Castle Peak Rd S 2	3.39E-07	0.4%	1.48E-07	0.2%
38	Castle Peak Rd S 3	8.10E-08	0.1%	7.75E-08	0.1%
39	Hoi On Rd	6.63E-07	0.8%	4.84E-07	0.6%
40	Construction Site Office	0	0	9.71E-10	0.0%
	- - - -	- - -	- - -	- - -	- - -
Other		5.77E-06	7.3%	5.71E-06	8.2%
	Total	8.43E-05		7.56E-05

9.8.18 Table 9-29 shows that of the roads included in the assessment only Tuen Mun Rd located close to YKT WTW significantly contributes to the total PLL values. PLL values for other roads are significantly lower, in particular the Tsuen Wan Rd which is the subject of the present project contributes only about 0.5% to the total PLL, and that value includes not only the traffic but (for Case 5) construction worker populations as well. PLL for the Project Construction Site Office is even smaller, well below 0.001% of the total.

9.8.19 EU1TMRU event contributes about 80% of the total Case 4 PLL at Tsuen Wan Rd while over 19% is contributed to the other earthquake scenario EU1TMRUH and the aircraft crash event AU1TMRU. The role of other scenarios considered in the assessment is negligible. Similar proportions can be obtained for Cases 1, 3 and 5, and for the project site office for cases related to the construction phase.

Individual Risk Results

9.8.20 The individual risk levels for the areas surrounding Yau Kom Tau WTW are presented in Figure 9-8 for all cases. The individual risk levels are calculated for a hypothetical person spending 100% of their time outdoors, with no provision for escape from the chlorine cloud. This is an upper-bound estimate of the risk to actual individuals.

9.8.21 It should be noted that since the individual risk levels depend on the assumed event frequencies and location only, but do not depend on the population data, the risk level shown apply to all five cases considered in this assessment.

9.8.22 As can be seen from the figure, the individual risk levels are well below the HKPSG criterion of 10^-5. In fact most of the Tsuen Wan Rd Upgrading Project site lies outside the area of any significant individual risk, with only its western part located within the 10^-8 to 10^-7 risk area.

Mitigation Measures

9.8.23 Following the QRA assessment for YKTWTW (ERM, 2001) several mitigation measures have been implemented at the WTW to reduce the risk As Low As Reasonably Practicable. This study confirmed that those measures are adequate even for the general and road populations projected up to 2030, ie the risk levels remain well within the ALARP zone (refer to Figures 9-6 and 9-7). On the other hand the Tsuen Wan Rd Upgrading Project itself has a very little effect on the societal risk levels in the YKT WTW vicinity. It is therefore concluded that no additional mitigation measures at YKT WTW are reasonably practicable.

9.8.24 Similarly, the risk levels to the road and workforce population of Tsuen Wan Rd are relatively low due to its significant distance to YKT WTW, and since the risk to this population is dominated by an earthquake event affecting both the YKT WTW and the Tsuen Wan Rd and subsequent chlorine release at WTW, no mitigation measure that could significantly reduce such risk is considered cost-effective or practicable.

9.8.25 This can be illustrated by the following simple cost-benefit analysis, similar to that applied in Maunsell (2007).

9.8.26 The cost effectiveness mitigation measures may be estimated by calculating the Implied Cost of Averting Fatality (ICAF) which may then be compared with the statistical value of life to determine whether a mitigation measure is reasonably practicable to implement. The value of life is taken as HK$660M (including an ‘aversion factor’ of 20), same as used in the approved Tuen Mun Rd EIA (Maunsell, 2007).

9.8.27 Assuming optimistically that a potential mitigation measure could remove all risk to life to the Tsuen Wan Rd operational phase population (PLL = 4.45 x 10^-7per year, see Table 9-29) and assuming that it will remain effective for a period of 50 years, in order to obtain ICAF lower than $660M, the maximum cost of a mitigation measure would have to be $660,000,000 x 4.45 x 10^-7 x 50 = $14,685

9.8.28 It is clear that no conceivable mitigation measure that could significantly reduce the risk can be designed within the above cost. For example, Maunsell (2007) estimates the cost of a road enclosure at HK150,000 per metre. No mitigation measures implemented at Tsuen Wan Rd could therefore be considered cost-effective. Similar conclusion can also be drawn for the operational phase population (PLL = 3.37 x 10^-7 per year).

9.8.29 Nevertheless two measures previously recommended in the EIA for Route 8 (formerly Route 16) warrant consideration:

· Introduce the ‘no stopping zones’ on the western part of Tsuen Wan Rd and on all major roads within the area where the individual risk levels exceed 1 x 10-9 (see Figure 8.39.8). This measure (if not already implemented) would also be particularly effective on Tuen Mun Rd.

· Provide traffic signals to stop in case of emergency the traffic on western part of Tsuen Wan Rd and other major roads within the area that could be affected by a chlorine spill at YKT WTW.

9.8.30 Similarly, implementation of the following measures for protection of the Project construction workers should be considered:

· The number of workers on site during construction stage should not exceed the levels assessed in this report.

· Emergency evacuation procedures should be formulated and all workers on site should be familiar with these procedures as well as the route to escape in case of gas release incident . Relevant Departments, such as WSD and FSD, should be consulted during the development of Emergency procedures. Diagram showing the escape routes to a safe place should be posted in the site notice boards and at the entrance/exit of site.

· The emergency procedures should specify means of providing a rapid and direct warning (e.g. Siren and Flashing Light) to construction workers in the event of chlorine gas release in the YKTWTW.

· The construction site officer should establish a communication channel with the YKTWTW operation personnel during construction stage. In case of any hazardous incidents in the treatment works, operation personnel of YKTWTW should advise the site officer to evacuate the construction workers.

9.9 Conclusions and Recommendations

9.9.1 This QRA study re-assessed the hazard to life associated with a potential chlorine spill at the Yau Kom Tau Water Treatment Works (YKT WTW) in light of the increased road and workforce population associated with the construction and operational phases of the Tsuen Wan Rd Upgrading Project. The study methodology is based on the previous YKT WTW assessment (ERM, 2001) however the updated newest projections of the general population in the area have been taken into account. Note that very similar methodologies have been adopted for the previous chlorine spill hazard assessment (Scott Wilson, 1998a,b) for Route 8 (formerly Route 16) and the recently approved EIA of Tuen Mun Rd widening (Maunsell, 2007). While the ERM (2001) methodology has been carefully re-examined, based on the review of the recent chlorine incidents worldwide, it proved robust and adequate for the present assessment. Thus, most revisions introduced to the present methodology relate to the population assumptions.

9.9.2 The QRA shows that the road population increases during the operational phase of the Tsuen Wan Rd Upgrading Project as well as the workforce population during the construction phase have little effect on the total societal risks in the vicinity of YKTWTW, as expressed by the total PLL value and the FN curves. While these risks, similar to the ERM (2001) assessment, remain within the ALARP region of the HKPSG Risk Guidelines, it should be noted that due to the updated population projections introduced in this study, the general risk levels even for the 2030 project operational phase are slightly lower than those predicted for 2006 in the ERM (2001) assessment.

9.9.3 Since all the mitigation measures at YKT WTW suggested in ERM (2001) to reduce the risk to As Low As Reasonably Practicable levels have already been implemented, no new mitigation measures are recommended in this study. Similarly, since the risk levels to the road and workforce population of Tsuen Wan Rd are relatively low due to its significant distance to YKT WTW, and since the risk to this population is dominated by an earthquake event affecting the YKT WTW and the Tsuen Wan Rd and subsequent chlorine release at WTW, as shows a simple quantitative cost-benefit analysis, no mitigation measure that could significantly reduce such risk is considered cost-effective or practicable. Nevertheless a number of measures suggested in Section 9.8 such as introduction of no stopping zones and emergency traffic signals on the relevant roads as well as establishing adequate emergency procedures for the Project workforce are recommended for the Project Proponent consideration.

9.10 References

(1) Brighton, P W M. (1989a), "Pressures produced by instantaneous chlorine releases inside buildings" UKAEA Report SRD/HSE/R467.

(2) Brighton, P W M. (1989b), "Continuous chlorine releases inside buildings: concentrations on emission to atmosphere" UKAEA Report SRD/HSE/R468.

(3) Cook, I., N. J. Holloway, W. Nixon and D.W. Phillips (1993), Consultancy on the Environmental Aspects of the Daya Bay Nuclear Power Station for the Government of Hong Kong - Risk Assessment Report.

(4) DNV(1997), Quantitative Risk Assessment of the Transport of Chlorine in Hong Kong for Environmental Protection Department.

(5) EIAO-TM, Technical Memorandum on Environmental Impact Assessment Process, available at: http://www.epd.gov.hk/eia/english/legis/index3.html .

(6) ERM (1997a), Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works: Methodology Report, for Water Supplies Department.

(7) ERM (1997b), Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works: Hazard Identification Report, for Water Supplies Department.

(8) ERM (1998), Technical Note 1: Application of Wind Tunnel Test Results in Hazard Assessments, Report for Water Supplies Department CE14/96 Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works, for Water Supplies Department.

(9) ERM (2001) Reassessment of Chlorine Hazard for Eight Existing Water Treatment Works: Hazard Assessment for Yau Kom Tau Water Treatment Works, for Water Supplies Department.

(10) HKPSG, Hong Kong Planning Standards and Guidelines, available at: http://www.pland.gov.hk/tech_doc/hkpsg/english/index.htm

(11) HSL (1998a), CFD Modelling of Chlorine Dispersion at Sha Tin WTW, The Health and Safety Laboratory, March 1998.

(12) HSL (1998b), CFD Modelling of Chlorine Dispersion at Tai Po Tau WTW, The Health and Safety Laboratory, December 1998.

(13) HSE (1998), The Implications of Major Hazard Sites in Close Proximity to Major Transport Routes, Contract Research Report 163/1998 prepared by WS Atkins Safety & Reliability.

(14) IChemE (1989), "Chlorine Toxicity Monograph - Updated Version", The Institution of Chemical Engineers.

(15) ICI (1995), "Chlorine Handbook", ICI Australia.

(16) Lees F P (1996), Loss prevention in the Process Industries, Butterworth Heinemann.

(17) Maunsell (2007) Widening of Tuen Mun Road at Tsing Tin Interchange - Environmental Impact Assessment EIA 142/2007, approved under EIAO on 23 January 2008

(18) Ove Arup (2000), Water Treatment Works Seismic Hazard Assessment - Seismic Assessment of Chlorine Storage Buildings and Chlorine Containers.

(19) PlanD(2006) http://www.pland.gov.hk/info_serv/statistics/tables/WGPD_Report_e.pdf

(20) Porter (1991), Risk Mitigation in Land Use Planning: Indoor Releases of Toxic Gases, International Conference on Modelling and Mitigating the Consequences of Accidental Releases of Hazardous Materials, 1991.

(21) RWDI (1998), Wind Tunnel Study - Reassessment of Chorine Hazard for Yau Kom Tau WTW, Rowan Williams Davies and Irwin Inc., January 1998.

(22) Scott Wilson (1998a) Agreement No CE 42/96, Route 16 from West Kowloon to Sha Tin: EIA Study, Final Assessment Report, for Highways Department.

(23) Scott Wilson (1998b) Agreement No CE 42/96, Route 16 Alternative Alignment: EIA Study, Final Environmental Assessment Report, for Highways Department.

(24) TNO (1992), "Methods for the Determination of Possible Damage to People and Objects Resulting from Releases of Hazardous Materials", CPR 16E, Green Book, 1992.

(25) Webber, D M, S J Jones, G A Tickle, and T Wren (1992), UKAEA Report SRD/HSE R587 “ A model of dispersing gas cloud, and the computer implementation DRIFT – II Steady Continuous Releases”

(26) Webber (1998), Brief Review of RWDI and DRIFT Results, 16 February 1998.

---

[1] ^{PlanD: Hong Kong 2030: Planning Vision and Strategy study, Working Paper No 42 (http://www.hk2030.gov.hk/eng/wpapers/pdf/wpaper42.pdf)}