10                     HAZARD

 

10.1               Objectives

10.1.1         The overall objectives of the hazard assessment are to identify potential hazard concerns, the likely mitigation measures required and to report the findings.

10.2               Scope of Work

10.2.1         The risk to the construction workers and road users due to chlorine usage at WSD Cheung Sha Water Treatment Works shall be assessed. The Applicant shall follow the criteria for evaluating hazard to life as stated in Annexes 4 and 22 of the TM EIAO in conducting hazard assessment and include the following in the assessment:

(i)        Identification of all hazardous scenarios associated with the transport, storage and use of chlorine at the water treatment works, which may cause fatalities on Tung Chung Road during road construction and operation phases.

(ii)      Execution of a Quantitative Risk Assessment expressing population risks in both individual and societal terms during road construction and operational phases.

(iii)     Comparison of individual and societal risks with the Criteria for Evaluating Hazard to Life stipulated in Annex 4 of the TM; and

(iv)    Identification and assessment of practicable and cost effective risk mitigation measures during road construction and operational phases.

10.2.2         The following study cases are evaluated and assessed:

¨              Construction Phase:

-      Transient risk to construction workers only

¨              Operation Phase:

-      Risk to the future new road population only

10.2.3         The risk levels for the population are assessed based on the year 2006 when the new road is undergoing construction completion and about to be operational.  The future population is based on projected population estimates for the year 2021 which represents the ultimate case after 15 years of road operation.

10.2.4         It should be noted that the Cheung Sha WTW is not a PHI site and therefore no hazard assessment of this site has been conducted to date.

10.3               Study Area

10.3.1         Background

10.3.1.1   A visit to the Cheung Sha WTW was conducted to ascertain details on the operation, process configuration, surrounding environment and possible population exposure.  Details are provided below and photographs provided in Appendix J.

10.3.1.2   Cheung Sha WTW is located off the Tung Chung Road on South Lantau approximately 200m north of Cheung Sha Village.  It is located about 100m above sea level on a gentle hill slope overlooking Cheung Sha Beach.  The layout of the Cheung Sha WTW is shown in Figure 10.1, with its general location in South Lantau indicated in Figure 10.2.  The access road to the site is approximately 150m long and runs along the side of a large catchwater.  Between the catchwater and the road there is either fence or a kerb roughly 9 inches high for most of the length, but the kerb is missing for a short stretch of about 10m.

10.3.1.3   The WTW provides treated water for supply locally to Lantau Island and for fire hydrants along Tung Chung road.  The sterilisation is done in 2 separate chlorination rooms and each has a chlorine store.  The WTW uses and stores chlorine in 50kg cylinders.  Chlorine consumption was advised to be 1 cylinder per 10-15 days, with maximum consumption at 1 cylinder per 3-5 days.  An average consumption of 1 cylinder in 8 days was assumed.

10.3.1.4   For the local water supply, 1 duty and 1 standby chlorine cylinders were present in the chlorination room and 29 full and 5 empty cylinders were present in the store.  2 and 34 full cylinders respectively in the chlorination room and storage room will be modelled in the analysis.  For the chlorination room of the fire hydrant supply, while the equipment could handle a maximum of 2 cylinders, only 1 cylinder was observed.  5 cylinders were present in storage, while a maximum of 8 may be present.  All cylinders were modelled as full.

10.3.1.5   The chlorine stores are located at the chlorination rooms, at the centre of the WTW site and at the south-east corner, next to the service reservoir.  The chlorine drum storage, draw off, evaporation and chlorination equipment is all located within the chlorine store.

10.3.1.6   All the cylinders were removed from the delivery racks and chained up in wall racks or the connected position.  Each connected cylinder was on a weigh scale to indicate how full it was.  On each connected cylinder, an outlet valve was screwed onto the cylinder, connected by a coil pipe (8mm) to a short length of fixed 20mm pipework then via a second coil pipe (8mm) to a multiport valve to select the duty cylinder.  Fixed, 20mm pipework fed the chlorine to the vacuum regulator in the chlorinator.  Since the draw off from a cylinder is gas, releases from this pipework would only be gaseous releases.  The total length of pipework in each chlorination room was about 10m.

10.3.1.7   After the vacuum regulator the chlorine gas is at less than atmospheric pressure and does not pose any significant hazard.  The chlorine gas passes the vacuum ejector and dissolves in water in high concentration but this solution is not considered to pose any significant hazard.  The solution is then diluted into the final water of the WTW.

10.3.1.8   Chlorine cylinders are delivered to the works by truck.  A truck can carry a maximum of 9 racks of 5 cylinders each.  Consignments unloaded vary from 10 to 40 cylinders, so an average consignment will be assumed to be 25 cylinders.  As a cylinder is assumed to last 8 days, a delivery would be made every 200 days.  The truck does not spend any additional time on site (e.g. for paperwork) other than for unloading cylinders.

10.3.1.9   The racks are unloaded from the truck using the truck crane.  Cylinders are then removed from the cage and manhandled for approximately 20m into the chlorine stores by rolling on one edge of the base.  It is understood that the truck also makes deliveries to the Tai O WTW.  The typical delivery to Tai O is 1 rack and so the truck leaves the site still carrying a few full cylinders.

10.3.2         Safety Facilities

10.3.2.1   There are numerous safety features of the chlorine facilities.  A loud door open alarm was provided on the external doors to each room.  It is necessary for the door to be opened for 15 to 30 seconds for the loading/unloading of each cylinder.  Each room was provided with a low level extractor fan (12 inch diameter) and a high level inlet vent.  Both were fitted with automatic louvres.  A chlorine gas detector was provided in each room.  On detection of 3ppm chlorine the louvres would be shut and the fan turned off automatically.  It is understood that 1 or 2 false alarms occur per year, generally due to lightning storms.

10.3.2.2   It was observed the extractor fan in the fire hydrant chlorination room was running, but the fan in the local water supply room did not appear to be running, despite being set to automatic.  It is possible that the door alarm triggered during the visit to the site had shut off the fan. 

10.3.2.3   Both chlorination rooms were fitted with a 6 inch high kerb to prevent release of chlorine to outside the building if a chlorine spill occurred.

10.3.2.4   The quantity of chlorine stored at Cheung Sha WTW does not exceed the criterion of 10 tonnes of chlorine in 50kg cylinders or a single 1 tonne drum, and so the WTW is not classified as a PHI.  Therefore no Consultation Zone is defined for Cheung Sha WTW. The areas which could be exposed to fatal effects from a chlorine release at the WTW include the construction site and proposed road, as shown in Figure 10.2.

10.4               Input Data

10.4.1         Meteorological Data

10.4.1.1   The weather data for this study has been referred from the Hazard Assessment for Upgrading of Silvermine Bay WTW (ERM 1996).  Table 10.1 and Table 10.2 show the probabilities for combination of wind speed, direction and Pasquil stability class used in the assessment.  In line with the general geographical area covered by the reference report, DNV regards this weather data as relevant to South Lantau.

Table 10.1       Day Time Weather Probabilities

Wind

Wind Speed and Stability Class

Total

Direction

B

D

F

B

E

D

D

D

 

 

0.5 m/s

0.5 m/s

1.0 m/s

3.0 m/s

3.0 m/s

4.0 m/s

7.0 m/s

15.0 m/s

 

N

0.72

0.20

0.00

0.35

0.00

2.89

0.92

0.28

5.36

NNE

0.50

0.25

0.00

0.27

0.00

4.88

1.18

0.52

7.60

NE

1.09

0.38

0.00

0.27

0.00

2.77

0.37

0.16

5.04

ENE

1.24

0.53

0.00

0.37

0.00

3.34

1.22

0.85

7.55

E

0.68

0.43

0.00

0.61

0.00

7.84

5.78

6.95

22.29

ESE

0.33

0.21

0.00

0.36

0.00

6.17

4.46

4.19

15.72

SE

0.17

0.14

0.00

0.14

0.00

1.47

0.76

0.75

3.43

SSE

0.18

0.11

0.00

0.13

0.00

0.65

0.30

0.30

1.67

S

0.13

0.16

0.00

0.31

0.00

2.42

0.56

0.22

3.80

SSW

0.14

0.17

0.00

0.26

0.00

2.80

0.41

0.09

3.87

SW

0.14

0.12

0.00

0.18

0.00

1.24

0.31

0.09

2.08

WSW

0.23

0.23

0.00

0.25

0.00

1.15

0.22

0.11

2.19

W

0.61

0.31

0.00

1.56

0.00

1.67

0.08

0.05

4.28

WNW

1.34

0.34

0.00

2.17

0.00

1.96

0.16

0.09

6.06

NW

1.35

0.18

0.00

1.04

0.00

2.11

0.57

0.32

5.57

NNW

0.78

0.18

0.00

0.25

0.00

1.50

0.54

0.28

3.53

Total

9.63

3.94

0.00

8.52

0.00

44.86

17.84

15.25

100

Note: The numerical figures are the percentage of the time that the wind speed, direction and stability class occurs. Letters A to F are the Pasquil stability class of the weather. A indicates unstable conditions, more common during the daytime, F indicates stable conditions, more common at night.

 

Table 10.2       Night Time Weather Probabilities

 

Wind

Wind Speed and Stability Class

Total

Direction

B

D

F

B

E

D

D

D

 

 

0.5 m/s

0.5 m/s

1.0 m/s

3.0 m/s

3.0 m/s

4.0 m/s

7.0 m/s

15.0 m/s

 

N

0.00

0.13

0.85

0.00

0.76

1.94

0.62

0.19

4.49

NNE

0.00

0.17

1.51

0.00

2.07

3.27

0.79

0.35

8.16

NE

0.00

0.26

2.02

0.00

1.63

1.86

0.25

0.11

6.13

ENE

0.00

0.35

2.85

0.00

1.96

2.24

0.82

0.57

8.79

E

0.00

0.29

3.70

0.00

3.22

5.26

3.88

4.66

21.01

ESE

0.00

0.14

2.85

0.00

3.14

4.14

2.99

2.81

16.07

SE

0.00

0.09

1.62

0.00

0.79

0.98

0.51

0.50

4.49

SSE

0.00

0.07

1.31

0.00

0.37

0.44

0.20

0.20

2.59

S

0.00

0.11

1.85

0.00

1.33

1.62

0.38

0.15

5.44

SSW

0.00

0.12

1.70

0.00

1.63

1.88

0.28

0.06

5.67

SW

0.00

0.08

1.03

0.00

0.81

0.83

0.21

0.06

3.02

WSW

0.00

0.15

1.22

0.00

0.59

0.77

0.15

0.07

2.95

W

0.00

0.21

1.04

0.00

0.27

1.12

0.06

0.03

2.73

WNW

0.00

0.23

0.89

0.00

0.25

1.31

0.10

0.06

2.84

NW

0.00

0.12

0.54

0.00

0.34

1.42

0.38

0.21

3.01

NNW

0.00

0.12

0.54

0.00

0.41

1.00

0.36

0.19

2.62

Total

0.00

2.64

25.52

0.00

19.57

30.08

11.98

10.22

100

10.4.2         Population

 

10.4.2.1   The population used for this study is considered both for the construction phase (year 2006) and the ultimate case (year 2021). Most of data was derived based on the Working Paper 3, "Traffic Forecasts", "Table 3.4: Population Figures Assumed for Traffic Modelling" covering the population assumption for Lantau obtained by Mouchel (Mouchel, 2001).

Construction Phase

 

10.4.2.2   This new road runs along the hillside on the east of Cheung Sha Valley, continues eastwards to cross the catchwater and connects with South Lantau Road near the YWCA youth camp.  Approximately 800m of the road falls within a 500m radius of the water treatment works.  This section of road stretches from NNW to ENE of the water treatment works and the entire route section is located on higher ground than the water treatment works i.e. between 80m to 150m higher.  The road has a steady gradient of 10% to 12%.

10.4.2.3   Construction work for the road is expected to end in year 2006.  Data for the working population during construction was provided in correspondence with the Mouchel Project Manager (Mouchel, 2001).  The main construction site will be in Tung Chung but an additional proposed construction site may be provided 800m to the east of the WTW, as shown in Figure 10.2. Neither construction site could be affected by releases of chlorine from the WTW.  It is expected that approximately 10 gangs of 10 to 20 workers per gang will be spread along, on or very close to the route.  Hence it is assumed that total of about 5 workers per 100m of new road would be present during the construction phase of the project (Table 10.3). Concerns about more workers temporarily working in a smaller area will be addressed as a sensitivity test of the results. This figure represents the conservative upper bound figure which includes both the construction and supervisory staff.  The workers are present during the 12 hour day time and do not work at night time.

10.4.2.4   The alignment of the road and its work boundary are presented in Figures 2.1 to 2.13 and the scope of the landscape planting proposals in  Appendix I.  These figures show that the alignment in the vicinity of the WTW conforms to the contoured landscape of the hillside and involves minimal slope cutting or landscaping on the southern side of the alignment which is nearer to the WTW.  The worker population is assumed to work within the proposed alignment, and they would not be required to work outside this boundary and hence will not be exposed to the risk of being closer to the WTW.  The risk model has modelled the worker population being restricted along the proposed alignment.

Table 10.3    Construction Worker Data in 2006 (on new Tung Chung Road)

Location

Time Period

Number of Workers

New Tung Chung Road

Day

5 per 100m stretch

 

Night

(no activity)

 

10.4.2.5   The traffic data used throughout this study is provided by MVA Hong Kong Ltd. (MVA), reference (MVA, 2001).  The data selection is based on the more conservative scenario where no TCRPZ permit system is enforced subsequent to the improvement to Tung Chung Road. Peak traffic data for the existing Tung Chung Road and the South Lantau Road during the construction phase was forecasted by MVA and is shown in Table 10.4.  Assumptions were made to account for the average traffic for both the 12 hours day time and 12 hours night time.

Table 10.4    Vehicle Forecast for Construction Phase in 2006

Location

Time Period

Peak Number of Vehicles / hr

Average Number of Vehicles / hr

Existing Tung Chung Road

Day

180

120

 

Night

N/A

12

South Lantau Road

Day

180

120

 

Night

N/A

12

Assumptions

1)         Peak hours assumed to be 4 hours out of the 12 hours of daytime, i.e. 2 hours in the morning and 2 hours in the evening.

2)         Average no. of vehicle is calculated based on 50% peak traffic for the non-peak hours

3)         Night time average is assumed to be 10% of daytime average

 

0.4.2.1       The vehicle breakdown obtained from the same source, reference (MVA, 2001), indicates number of vehicle types per hour.  Analysis of the data gave approximately 61% passenger car, including taxis, 32% goods vehicle and 7% bus. Table 10.5 below tabulates the calculation method to obtain an average of 4.22 people per vehicle. This number corresponds closely to the 4.6 people per vehicle based on the South Lantau Road taken from the Annual Traffic Census Statistics (DNV, 1999) albeit less than 10% variance.

Table 10.5    Calculation for Average Number of People Per Vehicle

Vehicle Type

Fraction of Vehicle Type

*Number of People / Vehicle

Average Number of People / Vehicle

Passenger Car / Taxi

0.61

3

1.83

Goods Vehicle (LGV/HGV)

0.32

2

0.64

Bus / Coach

0.07

25

1.75

Total

1.00

n/a

4.22

Assumptions

·                The passenger car / taxi each has 3 people, the goods vehicle each has 2 people while the bus / coach each carries 25 people.

 

0.4.2.2       The time exposure of the population travelling in moving vehicles is dependant on the speed of the vehicle.  It is assumed that the average speed of the vehicles is 50km/hr, as shown in Table 10.6.

Table 10.6    Road Population for Construction Phase in 2006

Location

Time Period

Average Number of Vehicles / hr

Vehicle Speed in km/hr

Vehicle / km

People / km

Existing Tung

Day

120

50

2.40

10.13

Chung Road

Night

12.0

50

0.24

1.01

 

Operational Phase

0.4.2.3       The transient population for the ultimate case where the road traffic at the new Tung Chung Road reaches saturation was also estimated by MVA for year 2021 and is shown in Table 10.7.  Population densities were calculated by assuming approximately 4.22 passengers per vehicle based on the above.

Table 10.7      Population Data for Operation Phase in 2021 (15 years after opening of new Tung Chung Road)

 

Location

Time Period

Average Number of Vehicles / hr

Vehicle Speed in km/hr

Vehicle / km

People / km

New Tung Chung

Day

667

50

13.33

56.27

Road

Night

66.7

50

1.33

5.63

 

10.5               Hazard Identification

10.5.1         Introduction

 

10.5.1.1   Although chlorine is used in low concentrations for disinfection of water, chlorine particularly if inhaled, is toxic.  The hazard from the use of chlorine at WTWs arises from possible releases of chlorine gas into the atmosphere, which would then be blown with the wind and gradually disperse.  This is referred to as the toxic cloud.

10.5.1.2   Chlorine (as pressurised liquid in 50kg cylinders) is delivered to the WTW by truck, unloaded from the truck within the building by crane and the chlorine is loaded into storage.  As needed cylinders in the duty and standby positions are replaced by cylinders from storage, again using the crane.  Groups of cylinders (draw off units) in the duty and standby positions are connected to the chlorination system.  The chlorine first passes through a regulator which reduces the pressure of the chlorine to below atmospheric.  The chlorinators are ejector type and operate at less than atmospheric pressure.

10.5.1.3   The hazard is therefore significant from when the truck arrives on the approach road (the battery limit of the WTW) to the regulator.

10.5.2         Recorded Chlorine Releases in Hong Kong

 

10.5.2.1   The historical data available in Hong Kong, listed below, gives insight into the magnitude and frequency of the more common occurrences and is considered to realistically represent the smaller, internal events caused by human error and spontaneous failure.  The Hazard Identification exercise has been based on the 6 previous Hong Kong incidents for chlorine releases, and supplemented with larger, internal and external releases cases from previous chlorine studies.

Drum Leak, Tai Po Tau, Hong Kong

 

10.5.2.2   A chlorine leak from a 1 tonne drum occurred on 8 June 1983 at Tai Po Tau Water Treatment Works, Hong Kong.  Following changeover between drums, the copper coil connection failed.  The operators were not carrying breathing apparatus (BA).  When BA sets were obtained from the office, it was impossible to approach the leak because of the pressure of the liquid release.  The emergency services were called 15 min after the start of the incident.  The 9 personnel then evacuated the works.  The emergency services arrived within 6 min, and entered the drum store wearing BA sets and turned off the drum isolating valve 50 minutes after the start of the incident.  A total of 725kg of chlorine had been released.  6 of the operators were treated in hospital for less than 24 hours.  The works are isolated and there were no off-site casualties.

Drum Leak, Tai Lam Chung, Hong Kong

 

10.5.2.3   11.2kg of chlorine leaked from a 1 tonne drum in the chlorine store at Tai Lam Chung Pre-chlorination House in the New Territories on the 28 September 1990.  The leak was repaired on site by the chlorine supplier with the presence of Fire Services personnel by tightening the gland of the drum isolating valve.  There were no casualties in this incident.

Chlorine Charge Line Leak, Sha Tin, Hong Kong

 

10.5.2.4   The incident occurred inside the chlorine transfer room of the Sha Tin Water Treatment Works, the biggest treatment works in Hong Kong, on the 8 September 1992.  Workers were charging liquefied chlorine from 1 tonne drums to bulk chlorine vessels each with a maximum holding capacity of 125 tonnes.  150kg of chlorine leaked from the liquid chlorine charging line.   One staff at the treatment works together with six residents living in the nearby staff quarters 200m away suffered respiratory problems.  The section of the defective pipe was replaced after the incident and the frequency of maintenance inspection was increased.  The chlorine plant and the chlorine bulk storage tank were decommissioned at the end of 1992 and were replaced by a new chlorination house with an absorption system.   Seven persons were injured, other than these injuries there was no other perceivable impact.

Feed Pipe/Drum Connection Leak, Sheung Shui, Hong Kong

 

10.5.2.5   2kg of chlorine leaked from the connection between the feeding pipe and the chlorine drum at the Sheung Shui Water Treatment Works on 25 January 1993.  Fire Services personnel were informed and the leak was sealed up.  There were no casualties in this incident.  A centralised ordering and inspection system was implemented for the lead washers to ensure that they are supplied to good standards.  No person was injured.

Valve Spindle Leak, Pak Yue Kong Swimming Pool, Kowloon

 

10.5.2.6   Liquefied chlorine found leaking from the valve spindle of a 1 tonne chlorine drum inside the chlorine store of the Pau Yue Kong Public Swimming Pool on 18 May 1992.  Fire Services personnel eventually tightened the valve and stopped the chlorine gas from further release.  The released chlorine escaped through the air vent of the chlorine store to the outside of the swimming pool complex.  A passer-by suffered respiratory problems after inhaling toxic gas.  One person was injured.

Leakage at Isolating Valve of Full Drum. Tai Lam Chung Prechlorination House

 

10.5.2.7   In March 1994 leakage occurred at isolating valve of full chlorine drum at Tai Lam Chung Pre-chlorination House.  The chlorine absorber was activated.

10.5.3         Types of Chlorine Releases

 

10.5.3.1   This study has divided potential releases into the following categories:

¨              internal releases;

¨              external releases;

¨              on the road;

¨              while unloading; and

¨              at the store, not contained due to collapse of building due to major external event such as earthquake.

 

10.5.4         Internal Release Cases

 

Accidents within the WTW Building

10.5.4.1   Accidents within the WTW building would lead to internal releases.  Accident types would include:

¨              spontaneous failure (leak or rupture) of a cylinder, the cylinder valve or pipework, perhaps due to corrosion or lack of maintenance;

¨              dropping of containers during truck unloading or loading into the duty position;

¨              fire; and

¨              human error, leading to a leaking connection.

 

Release Categories

10.5.4.2   Three release categories have been used to represent these previous accidents:

¨              50kg leaking through a small hole, equivalent to the diameter of the connecting pipe or a leaking valve (4.53mm);

¨              50kg released effectively instantly, equivalent to a rupture of a cylinder or breaking off of the cylinder valve; and

¨              100kg leaking through two small holes (6.4mm).

 

10.5.4.3   These cases have been used as the basis for the accident rates in the analysis.

Containment System Failures

10.5.4.4   If the containment system operates as designed then internal releases will present a less severe offsite hazard.  There is however potential for failure of the containment system.  The failure modes of this system which are significant are:

¨              containment system not being activated resulting in a release of chlorine through forced ventilation; and

¨              operation of containment system resulting in a release of chlorine driven by the expansion of the release.

 

10.5.4.5   These are discussed in more detail in Section 10.6 below.

10.5.5         External Release Cases

 

Accidents on the Road

10.5.5.1   There is a potential for a release of chlorine during the delivery of chlorine cylinders to the store.  The scope of the study only considers such accidents within the site boundary.  The accident cases which could occur are:

¨              roll-over;

¨              crushed at rear;

¨              crushed at side;

¨              vehicle fire (crash where other vehicle bursts into flames);

¨              tanker fire (crash where other vehicle is a tanker carrying large quantity of flammable DG and bursts into flames);

¨              spontaneous;

¨              load-shedding; and

¨              truck fire (fire on chlorine truck).

 

10.5.5.2   The failure cases considered are based on the Chlorine Transport Study (DNV, 1997):

¨              brief, small gas leak, emergency response succeeds, 5kg, 0.01kg/s (SVCS);

¨              longer duration, small gas leak, emergency response fails, 50kg, 0.01kg/s (SVCF);

¨              small liquid leak, 50kg, 0.1kg/s (SLC);

¨              medium vapour leak, 50kg, 0.1kg/s (MVC);

¨              medium liquid leak, 50kg, 1kg/s (MLC);

¨              cylinder rupture, 50kg, instantaneous (RC); and

¨              multiple cylinder rupture, 500kg, instantaneous (RCM2).

 

Accidents while Unloading

10.5.5.3   The range of incidents which can occur are:

¨              dropped cylinder rack, large vapour release from one or five cylinders. It is considered that toppling of a cylinder during manhandling between the truck and the store would unlikely result in a leak;

¨              accident while manoeuvring, large vapour release from one or five cylinders;

¨              drive-away while unloading, large vapour release from one or five cylinders; and

¨              fire on truck while unloading, multiple cylinder rupture, 500kg.

 

Major External Impact Events

10.5.5.4   Previous studies have considered a range of external events which include:

¨              earthquake;

¨              subsidence;

¨              flooding or dam burst;

¨              aircraft crash;

¨              landslide; and

¨              lightning strike.

 

10.5.5.5   The WTW is constructed on a platform formed partly by cutting into the hillside to the north and partly by an embankment slope to the east.

10.5.5.6   The cut slope to the north is approximately 16m high.  The chlorine stores are more than 40m from the crest of this cut slope, the debris from a slippage of this slope would not reach the chlorine store.

10.5.5.7   The embankment slope to the west is constructed to contain the service reservoir.  It appears that the chlorination room and store for the pumping station could be within the area which could be undermined by slippage of this fill slope, which could in principle be caused by leakage from the storage reservoir.  However, storage reservoirs are regularly inspected and subjected to leakage checks, and the angle of the underlying original ground level will provide for good drainage, even if the original ground is less permeable than the fill.

10.5.5.8   The first store and treatment room (for the “Treatment Works”) and the storage reservoir were built in the mid to late 80s.  The chlorination room and store for the pumping station were a later addition.  Since the formation of GEO / GCE in the mid to late 1970s, it has been required to design and construct fill slopes to Government standards.  This includes consideration of possible failure modes or slip planes within the slope and adequate compaction of the fill material.  The advent of computers has allowed virtually all slip planes to be considered in the design of a slope.  Also, the slope and its as built drawings and other details will be inspected periodically by a competent engineer.

10.5.5.9   Two drainage lines run near to the WTW site.  The one due north of the site drains into the catchwater to the north of the access road and could drain in either direction in case of blockage of the catchwater in one direction.  A second drainage line runs to the east of the WTW through a culvert under the catchwater into a man made drainage channel running north to south bedside the WTW.  The site plinth itself is constructed with a slight angle to allow water to naturally drain away and the slopes to the north will have appropriate drainage provisions, hence subsidence and flooding are thought to be unlikely.  The following factors are also relevant:

10.5.5.10   Dam burst is not credible since there is no reservoir above the WTW.

10.5.5.11   The site is several km from Chek Lap Kok and not directly below the flight path and so aircraft crash is considered incredible.

10.5.5.12   Lightning strike could occur but is likely to result in electrical damage and not likely to directly result in a release.

10.5.5.13   The most significant of these is earthquake.  Failure cases, shown in Table 10.8, have been developed similar to those for the QRA of Tsuen Wan WTW, a similar WTW using 50kg cylinders, (DNV, 1998).

Table 10.8       Summary of Failure Cases due to Earthquake

Reference

Failure Case

Released Mass (kg)

Hole diameter (mm)

Phase

Eq1SVP

1 cylinder valve in either chlorination room

50

8

gas

Eq1SPVM

Piping from valves to vacuum regulator (4.53mm dia)

50

4.53

gas

Eq2SVP

2 cylinder valves in either chlorination room

100

11.28

gas

Eq2SPVM

2 pipe coils from valves to manifold (4.53mm dia)

100

6.4

gas

Eq1Rup

Instantaneous Failure of 1 cylinder in either store

50

rupture

liquid

Eq2Rup

Instantaneous Failure of 2 cylinders in either store

100

rupture

liquid

Eq34Rup

Instantaneous Failure of 34 cylinders in Village Supply Store

1700

rupture

liquid

Eq8Rup

Instantaneous Failure of 8 cylinders in Hydrant Store

400

rupture

liquid

 

10.6               Frequency Estimation

10.6.1         Indoor Failure Cases

 

Frequency of Release within the Building

10.6.1.1   There have been 6 releases of chlorine in Hong Kong during the 14 years period 1983 to 1996.  In 1995 4,270 tonnes of chlorine were used at WTWs in Hong Kong.  Tuen Mun and Tsuen Wan WTW used 308 tonnes in 50kg cylinders.  Thus 10,121 containers (3961 drums and 6160 cylinders) were used in 1995.  Thus releases are expected at the frequency of:

6 / 14 / 10,121 = 4.23 x 10-5 per container per year

10.6.1.2   Of the 6 releases, 4 were in the 0 to 50kg category and 2 in the 50 to 250kg category.  95% of the 50kg releases are assumed to be small leaks and 5% to be ruptures.  The base frequencies calculated are shown in column 5 of Table 10.9.   The indoor release frequency is calculated by multiplying the base frequencies by the number of cylinders used per year, approximately 46 for the uprated Cheung Sha WTW.  The indoor release frequencies are shown in the last column.

Table 10.9       Frequency Calculation

Maximum Released Mass (kg)

Hole Diameter (mm)

Proportion of Releases

Number from the 6 Historical Releases

Base Frequency

 (per Container year)

Number of Cylinders Used (per year)

*Indoor Release Frequency

(per year)

50

4.53

0.95

4

2.68 x 10-5

45.625

1.22 x 10-3

50

Instant

0.05

 

1.41 x 10-6

45.625

6.44 x 10-5

100

6.39

1

2

1.41 x 10-5

45.625

6.44 x 10-4

*          Column 7 is given by the product of Columns 5 and 6.

The Effect of the Containment System

10.6.1.3   The containment system is designed to lessen the risk of releases of chlorine escaping from the building to the atmosphere.  Forced ventilation releases to the atmosphere only occur if this system fails such that the ventilation fan continues to run, otherwise a release driven by the expansion of the liquid chlorine on release occurs.  A standardised approach to the probability of failure of the containment system was discussed during a meeting between consultants and EPD on 18 July 1995.  This suggested use of 1% probability of contain / absorb system failure, which is considered to be applicable to failure of the containment system (such that the ventilation fan remains on) in this study.

Frequency of Releases to Atmosphere

10.6.1.4   Thus depending on whether the fan remains on or release is driven by the natural expansion of the chlorine liquid, the three failure cases identified in Table 10.9 produce 2 groups (forced ventilation and natural ventilation) of the same three failure cases, i.e. the 6 failure cases shown in Table 10.10.

Table 10.10     Frequencies of Release to Atmosphere for Indoor Releases

Reference

Contain and Absorb Failure Case

Released Mass (kg)

Hole Diameter (mm)

Phase

Base Frequency (per year)

Probability of Containment Failure Case

Frequency of Release to Atmosphere (per year)

SVSF

Release, forced ventilation

50

4.53

Gas

1.22 x 10-3

0.01

1.22 x 10-5

RupSF

Release, forced ventilation

50

rupture

Liquid

6.44 x 10-5

0.01

6.44 x 10-7

LVSF

Release, forced ventilation

100

6.39

Gas

6.44 x 10-4

0.01

6.44 x 10-6

SVNV

Release, natural ventilation

50

4.53

Gas

1.22 x 10-3

0.99

1.21 x 10-3

RupNV

Release, natural ventilation

50

rupture

Liquid

6.44 x 10-5

0.99

6.38 x 10-5

LVVF

Release, natural ventilation

100

6.39

Gas

6.44 x 10-4

0.99

6.38 x 10-4

Column 6 is the frequency of the same event taken from column 7 of Table 10.9. Column 8 is the product of column 7 and column 6.

 

10.6.1.5   Indoor release modelling has been carried out based on first principles, assuming that an indoor release mixes completely with the air in the chlorine store and it is the resultant mixture which is released to atmosphere. The concentrations and flow rates are calculated, forming the typical “saw-tooth” shape versus time, and averaged over a period of twice the duration of the release, see Table 10.11. The model assumes 95.2m3 store volume, the vent size is 300mm square and the velocity of forced ventilation is 2.5m/s. The equation for release concentration versus release flow rate is as follows:

V.dc/dt = vc

Where: V is the store volume (m3), c is the concentration in the air within the room and in the expelled air, v is the volumetric flow of air/chlorine mixture from the vent (m3/s) and t is the time (s).

 

Table 10.11     Release Rates and Concentrations to Atmosphere

Reference

Contain and Absorb Failure Case

Store Volume

Indoor release rate (kg/s)

Released mass (kg)

Release rate to atmosphere (kg/s)

Concentration of release to atmosphere

SVSF

Release, forced ventilation

95.2

0.035

50

0.177

3.2%

RupSF

Release, forced ventilation

95.2

instant

50

Instant

15.1%

LVSF

Release, forced ventilation

95.2

0.069

100

0.177

15.6%

SVNV

Release, natural ventilation

95.2

0.0349

50

0.012

10.2%

RupNV

Release, natural ventilation

95.2

instant

50

Instant

15.1%

LVVF

Release, natural ventilation

95.2

0.07

100

0.024

13.2%

 

10.6.2         External Failure Cases

 

Failure Cases on the Approach Road

10.6.2.1   The approach road between the WTW boundary gate and the chlorine store is 155m long. The chlorine truck delivers 5 racks of 5 cylinders (25 cylinders per delivery.  The truck may leave the site carrying a part load.  Thus the truck travels 0.57 km per year on site when loaded.

10.6.2.2   The base event frequencies from the recent QRA of the Transport of Chlorine in Hong Kong (DNV, 1997) have been used to calculate the release frequencies here, this is shown in Table 10.12.

Table 10.12     External Releases on the Approach Road

Ref.

Failure Case

Released Mass (kg)

Hole Diameter (mm)

Phase

Base frequency (per km year)

Distance Travelled per year loaded (km)

Frequency of release to Atmosphere (per year)

SVCS

Small gas leak, emergency response succeeds

5

2.5

Gas

8.90 x 10-10

0.57

5.04E-10

SVCF

Small gas leak, emergency response fails

50

2.5

Gas

8.90 x 10-10

0.57

5.04E-10

SLC

Small liquid leak

50

2.5

Liquid

2.90 x 10-7

0.57

1.64E-07

MVC

Medium gas leak

50

7.5

Gas

8.20 x 10-9

0.57

4.64E-09

MLC

Medium liquid leak

50

7.5

Liquid

7.30 x 10-8

0.57

4.13E-08

RC

Cylinder rupture

50

Rupture

Liquid

1.60 x 10-8

0.57

9.05E-09

RCM2

Multiple rupture

500

Rupture

Liquid

5.10 x 10-9

0.57

2.89E-09

 

Failures During Unloading

 

10.6.2.3   The base frequencies used in the calculation of failure frequency during unloading are shown in Column 6 of Table 10.13 and are taken from the HA of Silvermine Bay WTW (DNV, 1999).  A delivery is assumed every 200 days.  The multipliers, shown in the last column of Table 10.13 for each accident case are calculated as follows:

¨              Dropped cylinder racks: in approximately 20% of dropped drums, there is a leak. 1% of these cases are assumed to result in large leaks from a single cylinder.  0.01% of cases are assumed to result in leaks from a whole rack.  The remainder are small leaks, which will not have an offsite effect and are not considered further. Calculation of multipliers is as follows:

-         UnLV:        365 / 8 / 5 x 0.01 x 0.2

-         Un5LV:      365 / 8 / 5 x 0.0001 x 0.2

 

¨              Accident while positioning to unload: in 1% of cases there is a leak, 9% of these are large leaks, 1% leaks from all cylinders in a rack and the remainder are small leaks, which will not have an offsite effect and are not considered further. Calculation of multipliers is as follows:

-         UnLV:        365 / 8 / 25 x 0.01 x 0.09

-         Un5LV:      365 / 8 / 25 x 0.01 x 0.01

 

¨              Drive-away while still unloading: in 1% of cases there is a leak, 9% of these are large leaks, 1% leaks from all cylinders in a rack and the remainder are small leaks, which will not have an offsite effect and are not considered further. Calculation of multipliers is as follows:

-         UnLV:        365 / 8 / 25 x 0.01 x 0.09

-         Un5LV:      365 / 8 / 25 x 0.01 x 0.01

 

¨              Truck fire: the approach road is 155m long and the truck delivers 5 racks of 5 cylinders (25 cylinders per delivery).  The truck may leave the site carrying a part load.  Thus the truck travels 0.57km per year on site when loaded.  Half of truck fires are considered to result during unloading and 10% are considered to result in 50% of cylinders with large leaks. Calculation of multipliers is as follows:

-         UN22.5:     0.57 x 0.5 x 0.1

 

10.6.2.4   A summary of the calculation of the frequency of events during unloading is given in Table 10.13.

Table 10.13     Summary of Event Frequency during Unloading

Ref.

Failure Case

Released Mass (kg)

Hole Diameter (mm)

Phase

Base frequency (per year)

Multiplier (probability)

Frequency of release to atmosphere (per year)

 

Dropped Cylinder Rack

 

 

 

5.00E-06

1.83E-02

9.13E-08

 

Accident while positioning to unload

 

 

 

5.00E-05

1.64E-03

8.21E-08

 

Driveaway while still unloading

 

 

 

2.00E-05

1.64E-03

3.29E-08

UnLV

Total

50

8.00

Gas

-

-

2.06E-07

 

Dropped Cylinder Rack

 

 

 

5.00E-06

1.83E-04

9.13E-10

 

Accident while positioning to unload

 

 

 

5.00E-05

1.83E-04

9.13E-09

 

Driveaway while still unloading

 

 

 

2.00E-05

1.83E-04

3.65E-09

Un5LV

Total

250

5 x 8mm

Gas

-

-

1.37E-08

UN22.5

Truck Fire, 50% of cylinders leak

1125

22.5 x 8mm

Gas

4.00E-09

0.03

1.13E-10

Earthquake Failure Cases

 

10.6.2.5   The frequency of earthquake, Modified Mercali VIII, taken to be 1 in 100,000 years, has been apportioned between the failure cases as shown in Tables 10.14 and 10.15:

 

Table 10.14     Earthquake Event Tree

 

 

Damage Item

Damage Quantity

Outcome

 

 

 

 

 

 

One Cylinder, valve leak

One Chlorination Cylinder Leaks

 

 

0.25

0.125

 

 

 

 

 

 

Piping from valves to vacuum regulator (4.53mm dia)

Two Chlorination Cylinders Leak

 

 

0.25

0.125

 

Chlorination Cylinder

 

 

 

0.5

2 Cylinder valves leak

2 Cylinder valves leak

 

 

0.25

0.125

 

 

 

 

 

 

2 pipe coils from valves to manifold (2 x 4.53mm dia)

2 pipe coils from valves to manifold (2 x 4.53mm dia)

Earthquake

 

0.25

0.125

 

 

 

 

 

 

One Cylinder Rupture

One Cylinder Rupture

 

 

0.9

0.45

 

Storage Cylinder

 

 

 

0.5

2 Cylinder Rupture

2 Cylinder Rupture

 

 

0.09

0.045

 

 

 

 

 

 

34 Cylinders Rupture (Local Supply Store)

34 Cylinders Rupture (Local Supply Store)

 

 

0.005

0.0025

 

 

 

 

 

 

8 Cylinders Rupture (Hydrants Store)

8 Cylinders Rupture (Hydrants Store)

 

 

0.005

0.0025

 

Table 10.15     Earthquake Failure Frequencies

Reference

Failure Case

Released Mass (kg)

Hole Diameter (mm)

Phase

Release Rate to Atmosphere (kg/s)

Base Frequency (per year)

Base Prob- ability

Frequency of Release to Atmosphere (per year)

One Chlorination Set

Eq1SVP

Cylinder valve

50

8

gas

0.13

1.00E-05

0.125

1.25E-06

Eq1SPVM

Piping from valves to vacuum regulator (4.53mm dia)

50

4.53

gas

0.035

1.00E-05

0.125

1.25E-06

Two Chlorination Sets

Eq2SVP

2 Cylinder valves

100

11.28

gas

0.58

1.00E-05

0.125

1.25E-06

Eq2SPVM

2 pipe coils from valves to manifold (4.53mm dia)

100

6.4

gas

0.07

1.00E-05

0.125

1.25E-06

One Cylinder Rupture

Eq1Rup

Instantaneous Failure of one cylinder

50

Rupture

liquid

Instant

1.00E-05

0.45

4.50E-06

2 Cylinder Rupture

Eq2Rup

Instantaneous Failure of 2 cylinders

100

Rupture

liquid

Instant

1.00E-05

0.045

4.50E-07

34 Cylinders Rupture (Local Supply Store)

Eq34Rup

Instantaneous Failure of 34 cylinders

1700

Rupture

liquid

Instant

1.00E-05

0.0025

2.50E-08

8 Cylinders Rupture (Hydrants Store)

Eq8Rup

Instantaneous Failure of 8 cylinders

400

Rupture

liquid

Instant

1.00E-05

0.0025

2.50E-08

 

10.6.3         Summary of Failure Frequencies

 

10.6.3.1   A summary of the failure frequencies used in the risk model is given in Table 10.16.


Table 10.16     Summary of Failure Frequencies

Internal Releases

 

 

 

 

 

 

 

Reference

Failure Case

Released mass
(kg)

Hole diameter (mm)

Phase

Indoor release rate (kg/s)

Release rate to atmosphere (kg/s)

Concentation of release to atmosphere

Frequency of release to atmosphere
(per year)

SVSF

Release, forced ventilation

50

4.53

gas

0.035

0.177

3.15%

1.22E-05

RupSF

Release, forced ventilation

50

rupture

liquid

instant

Instant (16.9m^3)

15.08%

6.44E-07

LVSF

Release, forced ventilation

100

6.39

gas

0.069

0.177

15.64%

6.44E-06

SVNV

Release, natural ventilation

50

4.53

gas

0.035

0.012

10.20%

1.21E-03

RupNV

Release, natural ventilation

50

rupture

liquid

instant

Instant (16.9m^3)

15.08%

6.38E-05

LVVF

Release, natural ventilation

100

6.39

gas

0.07

0.024

13.20%

6.38E-04

External Releases while unloading

 

 

 

 

 

 

 

UnLV

Total

50

8.00

gas

-

0.130

-

2.06E-07

Un5LV

Total

250

5 x 8mm

gas

-

0.650

-

1.37E-08

UN22.5

Truck Fire, 50% of cylinders leak

1125

22.5 x 8mm

gas

-

2.925

-

1.13E-10

External Releases on Road

 

 

 

 

 

 

 

SVCS

Small gas leak, emergency response suceeds

5

2.5

gas

-

0.0122

-

5.04E-10

SVCF

Small gas leak, emergency response fails

50

2.5

gas

-

0.0122

-

5.04E-10

SLC

small liquid leak

50

2.5

liquid

-

0.0444

-

1.64E-07

MVC

medium gas leak

50

7.5

gas

-

0.11

-

4.64E-09

MLC

medium liquid leak

50

7.5

liquid

-

0.4

-

4.13E-08

RC

cylinder rupture

50

rupture

liquid

-

instant

-

9.05E-09

RCM2

multiple rupture

500

rupture

liquid

-

instant

-

2.89E-09

Earthquake Failure Cases*

 

 

 

 

 

 

 

Eq1SVP

Cylinder valve

50

8

gas

-

0.13

-

1.25E-06

Eq1SPVM

Piping from valves to vacuum regulator (4.53mm dia)

50

4.53

gas

-

0.035

-

1.25E-06

Eq2SVP

2 Cylinder valves

100

11.28

gas

-

0.58

-

1.25E-06

Eq2SPVM

2 pipe coils from valves to manifold (4.53mm dia)

100

6.4

gas

-

0.07

-

1.25E-06

Eq1Rup

Instantaneous Failure of one cylinder

50

rupture

liquid

-

instant

-

4.50E-06

Eq2Rup

Instantaneous Failure of 2 cylinders

100

rupture

liquid

-

instant

-

4.50E-07

Eq34Rup

Instantaneous Failure of 34 cylinders (Local Supply Store)

1700

rupture

liquid

-

instant

-

2.50E-08

Eq8Rup

Instantaneous Failure of 8 cylinders (Hydrants Store)

400

rupture

liquid

-

instant

-

2.50E-08

 

 


10.6.4         Variation of Release Frequency With Time Period

 

10.6.3.2   The internal process related releases and the earthquake releases can occur at any time.  The releases due to accidents on the approach road or unloading inside the chlorine store can only occur during working hours (on the days when the DG ferry runs).  However, it has been found in previous studies, that the risk picture is usually dominated by the earthquake events and so the variation of release frequency with time period has not been modelled in detail.

 

10.7               Consequence and Impact Analysis

10.7.1         Models Used

 

10.7.1.1   The models used to calculate the consequences of each of the failure cases are contained within the SAFETI Expert software package.  This is considered to be a standard software tool for use in Hong Kong.  The modelling approach used has been a flat terrain dense cloud model.  Further detailed information on the modelling calculations is contained in the SAFETI Theory Manual.

 

10.7.1.2   The wind tunnel model in the 8 Water Treatment Works Reassessment Study (ERM 2000) showed that chlorine could travel uphill for a distance of up to 150m since the dilution of chlorine cloud with air reduces the density of the plume which tends towards passive dispersion. Based on the results of the 8 WTWs reassessment study, 150m height cut-off should be used in this study. The new route section and construction site located on ground up to 150m above the water treatment works are shown in Figure 10.2.

 

10.7.2         Risk Assessment Parameters

 

10.7.2.1   The assessment has used the standard DNV parameters for risk assessment of this type.  Of particular significance are the assumptions relating the population distributions and the fraction of people killed in an event.  In this assessment it has been assumed:

 

¨              90% of outdoor population (e.g. construction workers and people in vehicles on the road) are modelled as exposed to the full outdoor toxic dose (the remaining 10% outdoor population successfully escapes, described in the 3rd bullet below) while no indoor population was modelled.

¨              transient populations (i.e. people in vehicles on the road) have been modelled as an average population density spread along the road and present at all times.  Since the real life situation is groups of people passing along the road and only present for part of the time, the method of modelling will over-estimate the time related dose and hence the probability of death of those within the effect zone of a release, but may underestimate the actual maximum number of people who could be affected (e.g. light traffic composed mainly of busses).  Overall the level of risk is considered to be an over-estimate and as the risk is found to be well within acceptable limits, this approximation does not change the conclusions of the study.

¨              escape and evacuation from the effected area have been modelled by the assumption above; namely the precept is that 90% of the outdoors population receiving the fatal toxic doses would be counted as fatality with the remaining 10% successfully escaping.

¨              the likelihood of becoming a fatality due to exposure of chlorine have been estimated using the standard Hong Kong probit equation:

P=-2.82 + 0.53 ln (C2t).

              The time of exposure modelled by SAFETI in the Probit expression is exactly the time calculated for the cloud to pass by the person at risk.

 

10.7.2.2   The consequence results calculated by SAFETI are used directly within the integrated software. The intermediate consequence results are produced in the form of the concentration and dimensions of the cloud versus time. The consequence results for the two key large events involving a greater than 1 tonne chlorine release has been tabulated for various weather classes in Table 10.17 below.

 

Table 10.17     Consequence Results for > 1 tonne Chlorine Release 

Failure Case Reference

Probit Value

Cloud Dimension

Size of Chlorine Cloud at Probit Value (m)

B

D

F

B

E

D

D

D

0.5 m/s

0.5 m/s

1.0 m/s

3.0 m/s

3.0 m/s

4.0 m/s

7.0 m/s

15.0 m/s

UN22.5

3.11

Length

480

1,100