1                                            Detailed Consideration of All Hazards

The hazard identification process is a formal review to identify all hazards for the LNG facility. The hazards identified with potential to cause loss of containment can be broadly categorised as:

·   Internal and process related hazards;

·   Natural hazards;

·   External hazards; and

·   Intentional acts

Further elaboration of the hazards under each category is included in the following paragraphs.

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

All scenarios with a frequency greater than 10-9 per year and potential to cause fatalities have the consequences of the event quantified.

Hazard scenarios are excluded from the consequence assessment if one of the following conditions is satisfied:

·       The frequency is below 1 x 10-9 per year.

·       The frequency of a particular event is significantly smaller than other causes of failure considered in the generic frequency.

·       If the generic failure frequency is judged to include events of such kind, then such events are not assessed separately.

·       If there are no consequences. If an event can be shown not to cause a loss of containment then the event is not considered further.

1.1                                      Internal and process related hazards

1.1.1                                Internal Hazards of LNG Storage Tanks

Overfilling

The nominal capacity (i.e. usable capacity) of the tank is 180,000 m3. The design unloading rate is 14,000 m3/hr and the unloading time is about 18 hours.

Overfilling of the inner tank may lead to overflow into the annular space between the inner tank and the outer tank. The bottom of the annular space is provided with 9% Ni steel up to 5m height. Furthermore leak detectors are provided to detect any LNG leak in the annular space bottom. Therefore, any overfilling event, if it ever occurs, can be detected and shutdown initiated. Also, the secondary containment provided by the outer concrete wall lined with steel will be able to contain this liquid.

There are several layers of safeguards to prevent overfilling:

a)        The tank to which the cargo is to be unloaded is identified before the arrival of the carrier, its level measured and the volume of cargo to be unloaded is pre-determined and this information is provided to the carrier. The total volume of cargo unloaded is also continuously monitored during unloading (typically, the volume of available space within the shore tanks is at least equal to the cargo volume to be discharged from the carrier, i.e. ships would not normally be required to unload cargo at multiple destinations);

b)        Continuous level measurement on tank using four separate detection systems with at least two different types of level measuring device; pre-alarm at normal maximum level in tank, corresponding to the usable capacity;

c)        Level high alarm; this is set typically with 3 to 5 minutes holding volume (between normal maximum level and high level) at design unloading rate;

d)        High high level initiates trip of shutdown valves in liquid inlet including transfer piping from jetty and re-circulation lines (the trip is initiated by a 2 out of 3 voting of separate level measuring devices of different type). This will stop further inflow of liquid into the tank. High high level trip is typically set with about 3 to 5 minutes holding volume (between high level alarm and high high level trip). The safety integrity level (SIL) of the high high level trip of liquid inlet will be determined during detailed design, however, SIL 2 classification is typical for this instrumented protective system which means that the probability of failure on demand will be less than less than 0.01;

e)        There is further holding volume of about 10 to 15 minutes at design unloading rate between the high high trip level and the inner suspended deck level before liquid can overflow through the suspended deck to the annular space (it may be noted that the actual height/volume between normal maximum level and the overflow level is determined based on sloshing height for the safe shutdown earthquake event; typical values are indicated above).

Based on the above, it can be seen that there is sufficient time (more than 10 minutes) for operator intervention in addition to the provision of high integrity instrumented protective system (i.e. high high level trip).

Rollover

Stratification, i.e. formation of two distinct layers of different density may occur in an LNG tank due to filling of cargo of different density than the liquid already in the tank or due to preferential boil-off in the tank resulting in a layer of more dense liquid at the top (due to evaporation of lighter components) as compared to the lower layers (where the boil-off is suppressed by the hydrostatic head but the liquid superheats due to heat ingress and becomes warmer and less dense) [1]. Stratification may also occur due to presence of sufficient nitrogen in LNG, typically more than 1%. Preferential boil-off of N2 results in a layer of less dense liquid at the surface.

The phenomenon of rollover occurs when the interface between the layers becomes unstable, leading to rapid mixing of contents of the two layers. As the superheated liquid from the lower layer rises to the surface, it gives off large amounts of vapour leading to potential overpressure of the tank.

There are a number of safeguards to detect and prevent stratification. These include:

a)        Temperature and density gradient measurement along the tank liquid column;

b)        Provision for circulation of tank contents through the operation of in-tank pump. Content from tank bottom is recirculated to the top, thus releasing any superheat and promote mixing;

c)        Provision for filling of tank from the top or from the bottom depending on the relative density of cargo and the tank contents;

d)        Regular sampling and analysis of boil-off gas including monitoring of boil-off gas quantity.

The tank is also protected by relief valves in the event of de-stratification leading to vapour generation. Relief valves are sized for rollover case as per EN 1473 requirements.

Inner Tank Leak

A sketch of a typical full containment tank is shown in Figure 1.1.

The main features of a full containment tank are that the liquid LNG is fully contained within a self-supporting inner 9% Nickel steel, surrounded by loose perlite insulation while the vapour is contained within a surrounding concrete outer tank (which includes the slab, the wall and the dome, all constructed of pre-stressed concrete). The concrete outer tank serves as secondary containment and is also capable of containing the liquid and of controlled venting of the vapour resulting from leakage of the inner tank, should one occur.

A 9% Ni steel plate is provided as liner along the inner surface of the outer shell at the bottom up to a height of typically 5m from the base. This protects the lower wall section of the outer tank, mainly the wall to base slab connection in the event of leakage.

A carbon steel plate lining is provided along the inner surface of the outer shell and roof (above the bottom Ni plate liner) to act as vapour barrier (i.e. to prevent vapour leakage through the concrete as well as to prevent moisture ingress from the outside).

Figure 1.1       Typical Structure of an LNG Storage Tank

The suspended deck is constructed of aluminium plates and is supported by suspension rods of stainless steel from the outer tank. Openings are provided on the suspended deck for vapour communication between the inner tank and the outer tank so as to ensure equilibrium of gas pressure on both sides. Insulation is provided over the deck to minimise heat leak from the outer shell to the liquid surface.

There are no penetrations through the outer wall or the inner tank shell. All piping to the inner tank is routed through the tank roof.

The tank will be constructed in accordance with BS 7777 [2].

For the capacity of tanks considered for this project, the concrete outer shell is typically about 0.8m thick and can withstand impact loading from projectiles.

In the case of a slab on the ground, installing an electrical heating system in the concrete base slab prevents the freezing of the base slab and the frost heave propagation in the ground beneath the foundation. Heating system control and monitoring is provided.

There are a number of design features to virtually eliminate any leakage from the full containment tank to the atmosphere:

a)   Because all piping is routed through the tank roof, there are no through penetrations in the outer shell or the inner tank. Therefore the potential for any connection failure leading to complete loss of tank inventory is eliminated;

b)   Outer tank (of pre-stressed concrete) is designed to hold the cold LNG liquid;

c)   Inner tank is constructed of low carbon 9% Nickel alloy steel which is heat treated. The plate thickness varies from about 10mm at the top to about 30mm to 40mm thick at the bottom. The material remains ductile and crack resistant at cryogenic temperatures and is also of high strength and toughness. This material is not subject to brittle failure when exposed to cold temperature.

d)   Tests performed on 9% Ni steel plates and welded assemblies show good performance against fatigue [3];

e)   In the event of any leakage from the inner tank due to crack in the weld or other defect, the leakage rate will remain small. The liquid will be contained in the annular space. Leak detectors in the form of thermocouples are provided along the annular space to detect leakage. Operator intervention is possible to empty the tank contents and isolate the tank for inspection and repair. Liquid leakage into the annular space is likely to result in additional vapour generation due to contact with warmer surfaces including the perlite insulation. The relief valves sized for the rollover case will be able to handle vapours resulting from liquid leakage into the annular space in the event of a crack on the inner tank;

f)     A 5m high 9% Ni plate at the lower wall section of the outer tank prevents any damage to the base slab to wall connection due to liquid leakage.

The engineering, construction, commissioning, and operation procedures for the inner tank have been developed to ensure the highest level of safety and reliability of the tank systems. For example, there are stringent QA/QC procedures specified for material qualification and welding qualification, material tracing and stamping, staggered vertical welds in construction, commissioning check, etc. Such measures and procedures when executed by highly experienced and competent contractors would virtually eliminate the occurrence of gross human errors in material specification and qualifications.

The long term LNG operation experience has demonstrated that there has been no reported loss of containment incidents involving full containment tanks.

It is also noted here that the full containment tank design offers significant improvements over single containment and double containment tank designs which have been prevalent earlier. More than 70 to 80% of the aboveground tanks currently in service are either single containment or double containment tanks.

Overpressure

The LNG tank is normally operated between 50 to 250mbarg. The tank is designed for a maximum pressure of 290mbarg.

Overpressure in tank may be caused by several factors:

a)   Normal boil-off due to heat leak from ambient;

b)   Vapour displacement during filling operation;

c)   Variation in atmospheric pressure (i.e. drop in atmospheric pressure);

d)   Flashing of incoming liquid if it is at a higher temperature than the bubble point of liquid at tank pressure.

Overpressure can result in failure of the tank secondary containment. However, there are a number of safeguards provided against overpressure:

a)   Normal boil-off vapours from the tank is routed to a boil-off compressor where the vapour is compressed and sent for re-liquefaction in the recondenser using the cold liquid pump-out from the tank;

b)   Vapour generated due to displacement during tank filling is returned to the ship through a blower (to provide the required head for transfer) or compressed by a separate high pressure compressor and routed to the sendout gas header;

c)   The tank pressure is continuously monitored by two sets of pressure measurements;

d)   A pressure control valve is provided on the tank to route all the excess tank vapours to a vent stack. The vent stack height and tip will be determined such that vapours discharged will disperse safely or if ignited, the radiation on the equipment and buildings adjoining the stack are within permissible limits as per EN 1473. The pressure control valve relieving to stack is typically designed for all overpressure cases under normal operations (the maximum case is typically the ship unloading case);

e)   An independent high high pressure trip is provided which will initiate shutdown of unloading operations (to stop liquid inflow);

f)     Relief valves are provided on the tanks which are sized for all the cases of overpressure. The maximum case is typically the ship unloading case for normal operations. The governing case for relief valve is however, the rollover case, which is an emergency case. The relief valve discharge is routed to the stack.

Underpressure

The LNG tank is normally operated between 50 to 250mbarg. The tank is designed for a minimum pressure of typically -5mbarg.

Underpressure may be caused by several factors:

a)   Pump-out of liquid;

b)   Increased compressor suction due to control malfunction;

c)   Variation in atmospheric pressure (i.e. rise in atmospheric pressure).

Under normal operating conditions, the boil-off generated due to heat leak is sufficient to prevent under pressure condition. Underpressure or vacuum conditions below –5mbarg due to control malfunction can cause failure of the tank containment. The tank bottom may be sensitive to vacuum and could get lifted upwards.

There are a number of safeguards in place:

a)   Continuous monitoring of tank pressure by two sets of pressure measurement;

b)   Low pressure alarm;

c)   Low-low pressure will trip the boil-off gas compressors and in-tank pumps and thus prevent further fall in tank pressure;

d)   Pressure control valve provided to inject external gas from the sendout gas header into the tank;

e)   Vacuum relief valves are provided which are typically sized for maximum vapour flow arising from compressors and pumps in operation. The operation of vacuum relief will lead to air-ingress into the tank and thereby avoid collapse of the tank. The operation of vacuum relief is envisaged as a measure of last resort. 

1.1.2                                Other Tank Related Hazards

Failure of Foundation/Ice Heave

Ice heave could occur in the event of a failure of the base slab heating system over a long period. This may lead to a crack in the base slab leading to failure of the tank base.

The base slab heating is controlled by temperature sensors. Furthermore redundant heaters with automatic switchover and redundant power supply source are provided. Provision is also made to replace the heaters if required while the tank is in service. Even if the heaters were to fail, it will be long time before the ground would freeze leading to potential failure due to ice heave. Operator intervention is possible.

Material Defect/Structural Defect/Construction Defect

Material defect may occur due to wrong materials being used in tank construction. Construction defect may result in poor welding. Structural failure of the concrete outer tank may occur, again due to poor construction or design.

In all of the above cases, quality control procedures including testing requirements during the design and construction and monitoring during operation are the main means to mitigate the hazard.

The design code EN 1473 [4] outlines a number of procedures including the following:

a)   Monitoring of concrete, every quarter of the concrete wall or every 5,000m3 of concrete;

b)   Testing of concrete outer wall by air pressure up to 125% of design;

c)   Hydrostatic test of the inner tank.

Additional precautions against failure at low temperature include gradual cooling of the tank at a rate of 50°C/hr and gradual introduction of liquid with sufficient waiting period at different levels to monitor any leakage.

Maintenance of In-Tank Pump/Dropped Object

In-tank pumps are provided to pump-out liquid from the tank. This eliminates the need for nozzle penetrations through the tank shell. Maintenance of the in-tank pump will require the lifting of the pump from the tank bottom through the pump well to the tank roof. It is then carried from the tank roof to the ground level for transport to the maintenance workshop. The removal of the pump is undertaken while the tank is in service. The liquid in the pump well is displaced by nitrogen and the pump is then lifted from its position at which point the foot valve (at the bottom of the well, which normally remains in open position due to the weight of the pump) gets closed. This prevents liquid entry into the pump well during the lifting operation.

The main hazard during this operation arises from accidental drop of the pump (which weights about 2 tonnes) due to failure of the sling. This may cause damage to the inner tank base plate leading to potential leakage from the inner tank. It is unlikely to cause damage to the concrete base. This hazard can be mitigated by appropriate lifting procedures, including the use of dual hoist. Any leakage from the inner tank will be contained by the outer tank.

1.1.3                                Tank Failure Frequency

The usual approach to estimating failure frequencies is to use historical databases of failures of similar facilities. However, there have been no failures of full containment LNG storage tanks to date. Also, the experience is limited to about 5000 tank-years. The available data are therefore insufficient to make statistical estimates of the failure frequencies. The approach in the literature has been to base an assessment on more generic data such as that from single containment tanks and then apply reduction factors and expert judgement to take credit for the additional protection offered by a full containment tank. The Purple Book [5] takes such an approach to derive a frequency for catastrophic release from a full containment tank as 1 x 10-8 per tank-year.

Other studies [6-12] into the failure frequency of single and double containment tanks provide numbers similar to the Purple Book. From these studies it becomes clear that the generic frequency data for ‘catastrophic’ failures, on which the Purple Book analysis is based, includes any failure exceeding the ‘large leak’ definition, which is about a 50mm hole. The catastrophic failure frequency of 10-8 per tank-year for full containment tanks should therefore be interpreted as inclusive of not only complete instantaneous tank failures but also large leaks.

This is consistent with comments in the Purple Book to the effect that all catastrophic ruptures leading to a release to the atmosphere are partly (50%) modelled as instantaneous release and partly (50%) as a continuous release within 10 minutes.

Based on this, a failure frequency of 1x10-8 per tank-year was adopted in the current study. The terminal may eventually have 3 LNG storage tanks and so the failure frequency was taken to be 3 x 10-8 per year.

This failure frequency includes all the causes discussed above, namely embrittlement, overfilling, rollover, overpressure, underpressure, ice heave, material/construction defects and maintenance hazards. It is also common practice [13] to take this frequency as inclusive of natural hazards since the generic failure frequencies calculated in the literature include failures caused by natural events such as earthquakes etc.

1.1.4                                Internal and Process Hazards for Piping and Equipment

The failure rate of the process areas and piping are well documented and the values used are given in Annex 13A6.

1.2                                      Natural hazards

1.2.1                                Seismic Hazard

GEO studies conducted in the last decades indicate that Hong Kong SAR is a region of low seismicity (e.g., GCO, 1991 [14]; GEO, 2002-2004 [15]). The seismicity in the vicinity of Hong Kong is considered similar to that of areas of Central Europe and the Eastern areas of the USA [16].

For this project, the full containment LNG tanks will be designed per the European Standard EN 1473 (1997) [4] and British Standard BS 7777(1993) [2]. These documents specifically exclude catastrophic failure of full containment tanks designed, fabricated, erected, inspected, and tested in accordance with the requirements contained in the codes. EN 1473 lists scenarios to be considered in the required QRA and states that “no collapse is considered for these tank types”.

In terms of seismic risk category, LNG facilities are “essential/hazardous facilities”, rather than “safety critical facilities” such as nuclear related structures. Hazardous facilities are those that contain large quantities of hazardous materials, but the release of those materials would be contained within the boundaries of the facilities and the impact to the public would be minimal. 

For seismic design in the LNG industry, a two-tier design approach is stipulated, within the same framework of the risk category seismic design philosophy used for nuclear facilities [17]. The Operating Basis Earthquake (OBE) is the maximum earthquake for which the structure sustains no permanent damage and restart and safe operation can resume after the earthquake. The OBE event has a return period of 475 years. The Safe Shutdown Earthquake (SSE) is the maximum earthquake for which the structures may sustain some permanent damage, but there is no loss of overall structural integrity and containment of contents. The SSE event has a return period of 10,000 years, which is the same as that stipulated for the highest class nuclear facilities (Class 5). In addition, the prescribed seismic design criteria for LNG tanks includes a load case for an event less frequent than the SSE basis failing the inner tank followed by an operating basis earthquake (OBE) aftershock acting on the outer tank. The potential failure of the tank due to this event is not expected to occur in Hong Kong based on the analysis below.

The full containment LNG tanks for this project will be founded on competent rock or densely piled foundation. Studies conducted by LNG tank manufacturers and constructors (e.g., Technigaz, 2003 [18]) have demonstrated that the full containment tank of similar capacity to that proposed for this project and designed by following EN 1473 and BS 7777 standards has the capacity to maintain its structural and containment integrity for seismic design motions as stipulated in Eurocode 8 [19] having up to 2.6g of peak ground acceleration (PGA) for a rock site. This level of design earthquake motions on the bed rock far exceeds the worst earthquake event ever recorded on a rock outcrop site anywhere in the world. The SSE having a return period of 10,000 years corresponds to PGA of about 0.5g based on latest seismic hazard studies conducted for Hong Kong. 

It is noted that a review of the earthquake resistance of LNG tanks to seismic demand was carried out in Japan by a national examination committee following a powerful earthquake with a PGA of 0.8g in southern Hyogo prefecture in 1995 [20]. The review concluded that the currently implemented standards are sufficient to maintain structural integrity of LNG tanks and prevent gas leaks even in an extremely rare and powerful earthquake. Note that an earthquake with a PGA of 0.8g is well within the seismic design capacity for the LNG tanks in this project which will be built to EN 1473 and BS 7777.

In consideration of the seismic design requirements and the capacity and reliability of LNG tanks to resist seismic demand, it is concluded that the proposed full-containment LNG tanks will maintain structural integrity in the Hong Kong environmental.

1.2.2                                Subsidence

Excessive subsidence or differential settlement of ground may lead to failure of the structures and ultimately potential loss of containment. This hazard is relevant to facilities built on reclamation land or poor ground without proper treatment or proper foundation design. For the Black Point terminal, the original two tanks will be founded on competent rock and hence excessive site subsidence or differential settlement is not expected. The future phase third tank will be built on densely piled foundation founded on the bed rock in the reclaimed area. Such foundation design prevents the occurrence of excessive site subsidence or differential settlement.

For the processing facilities at Black Point, a pile foundation design will also be adopted to prevent any impact of subsidence or settlement on the equipment and piping.

Based on the above, the foundation design and piping system design address the potential hazard of excessive subsidence and differential settlement of foundation. There is therefore no basis to presume that the facility will be more prone to subsidence than other facilities in the world and hence the generic failure frequencies will cover scenarios involving subsidence.

1.2.3                                Lightning

Lightning strike can ignite flammable vapour discharges from vents and stacks. Lightning strike has been the one of the causes of petroleum tank fires. However, this is applicable to cone roof tanks and floating roof tanks. In the case of cone roof tanks, the tank vent is in direct communication with the atmosphere. Breath-in and breath-out occurs during withdrawal of liquid from tank and during filling respectively. Vapours in flammable concentration may be generated which upon ignition at the vent tip due to lightning strike can flashback to the liquid inside (flame arrestor provided at the vent prevents such flame flashback). In the case of floating roof tanks, vapours generated due to seal leaks may get ignited. 

The above scenarios are not applicable to an LNG tank, which is a dome roof tank and is maintained under pressure of about 50 to 250mbarg. A lightning strike would have no impact on an LNG tank.

1.2.4                                Landslides

At the site there is up to 10m of colluvium and weathered rock material over competent rock. The colluvium is up to 2m thick and there are also occasional large boulders.

The excavations made to create the tank platforms will give rise to soil slopes at the upper part with a steeper rock slope below. Any boulders above these slopes will either be removed or secured to the slopes. The soil slopes created by the excavation will be secured by way of soil nails in accordance with Hong Kong standard practice. The existing natural soil slopes above these cut slopes will also be treated and stabilised. The rock slopes will be regularly inspected during excavation and any possible rock failure mode prevented by a combination of drainage measures and mechanical securing by dowels and rock mash as required.

Failure of the soil slopes will have no impact on the LNG storage tanks. The external concrete tank construction can readily withstand the impact of soil flow material without any compromise to the containment function of the tank.

The risk of failure of the underlying rock slopes is negligible. Rock slopes are at their most dangerous during construction. This is because the exact nature of the state of the rock strength and fissure patterns cannot be discerned with confidence until after the excavation is completed. During the excavation however the rock can be logged and fracture planes readily identified and any possible failure mode treated by mechanical support. The possibility of future adverse water conditions within the rock will also be controlled by drainage measures installed during excavation. The degradation of the rock surface with time is extremely slow and therefore this method of construction effectively guarantees that the rock slope will be permanently stable.

1.2.5                                Hill Fires

Hill fires are a fairly common phenomenon in Hong Kong, particularly in the dry season and are generally associated with fires lit accidentally by hikers or activities associated with the Chung Yeung Festival.

The terminal will have onsite landscape management to avoid the presence of combustible vegetation near the LNG storage tanks and process areas. This will prevent offsite hill fires from spreading to onsite areas. A study by Giribone [3] shows that the stability of the concrete storage tanks is not jeopardised by a fire scenario with an incident heat flux of 50kW/m2 (which is higher than the maximum allowable flux of 32 kW/m2 as per the design code EN 1473) over a period of 8 hours. There is no conceivable mechanism for a hill fire to produce such radiation fluxes given the lack of vegetation near the tanks and process areas.

1.2.6                                Storm Surges and Flooding

If the LNG storage tanks or piping become submerged under water, it is possible for buoyancy forces to lift the pipes/tanks, causing damage and possible loss of containment.

Flooding from heavy rainfall is not possible due to the coastal location of the site. The slopes of the natural terrain will drain water to the sea. The primary hazard from typhoons is the storm surge. Winds, and to a lesser extent pressure, cause a rise in sea level in coastal areas. In general, storm surges are limited to several metres unless channelling effects from the coastline exasperate the surge. Black Point’s location on the western tip of the New Territories will not create such channelling.

The terminal facilities, located 6m above sea level are therefore protected against any risk from storm surges, waves and other causes of flooding.

1.2.7                                Tsunami

Similar to storm surges, the main hazard from tsunamis is the rise in sea level and possible floatation of piping and tanks. The highest rise in sea level ever recorded in Hong Kong due to a tsunami was 0.3m high [21], and occurred as a result of the 1960 earthquake in Chile, the largest earthquake ever recorded in history at magnitude 9.5 on the Richter scale. With tanks and equipment positioned 6m above sea level, the effect of a tsunami on the terminal is considered negligible.

The reason for the low impact of tsunamis on Hong Kong may be explained by the extended continental shelf in the South China Sea which effectively dissipates the energy of a tsunami, and also the presence of the Philippine Islands and Taiwan which act as an effective barrier against seismic activity in the Pacific [22]. Secondary waves that pass through the Luzon Strait diffract and lose energy as they traverse the South China Sea.

Seismic activity within the South China Sea area may also produce tsunamis. Earthquakes on the western coast of Luzon in the Philippines have produced localised tsunamis but there is no record of any observable effects in Hong Kong. The Design Basis Report [23] considered the effects of tsunamis and demonstrated that a massive earthquake off the coast of the Philippines may produce a tsunami height in Hong Kong of up to 2.5m. Even when combined with extremes in tide level, the sea level at Black Point was shown not to exceed 5 mPD.

The terminal has been designed with due consideration of sea levels, tsunamis, waves and even rising sea levels from global warming. Being located 6 mPD, the terminal is protected against all causes of flooding.

1.2.8                                Summary of Natural Hazards

The terminal site and design of the facility are such that there will be no special risks from natural hazards. Natural hazards are therefore not treated separately in the analysis but are included in the generic failure frequencies (Annex 13A6).

1.3                                      External Hazards

1.3.1                                Aircraft Crash

The Black Point site does not lie within the flight path of Chek Lap Kok (Figure 1.2). Based on these figures, it is seen that the Black Point site is about 10km from the runways and hence 10km from the arrival and departure flight paths.

Figure 1.2       Flight Paths at Hong Kong International Airport

The frequency of aircraft crash was estimated using the methodology of the HSE [24]. The model takes into account specific factors such as the target area of the proposed hazard site and its longitudinal (x) and perpendicular (y) distances from the runway threshold (Figure 1.3). The crash frequency per unit ground area (per km2) is calculated as:

                                                                                     (1)

Where N is the number of runway movements per year and R is the probability of an accident per movement (landing or take-off). F(x,y) gives the spatial distribution of crashes and is given by:

Landings

   (2)

for  km

Take-off

    (3)

    for  km

Equations 2 and 3 are valid only for the specified range of x values. If x lies outside this range, the impact probability is zero.

Figure 1.3       Aircraft Crash Coordinate System

NTSB data [25] for fatal accidents in the U.S. involving scheduled airline flights during the period 1986-2005 are given in Table 1.1. The 10-year moving average suggests a downward trend with recent years showing a rate of about 2 x 10-7 per flight. However, only 13.5% of accidents are associated with the approach to landing, 15.8% are associated with take-off and 4.2% are related to the climb phase of the flight [26]. The accident frequency for the approach to landings hence becomes 2.7 x 10-8 per flight and for take-off/climb 4.0x10-8 per flight. The number of flights at Chep Lap Kok for year 2011 is conservatively estimated at 394,000 (a 50% increase over 2005).

Table 1.1        U.S Scheduled Airline Accident Rate [25]

Year

Accident rate per 1,000,000 flights for accidents involving fatalities

10-year moving average accident rate per 1,000,000 flights

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

0.14

0.41

0.27

1.10

0.77

0.53

0.53

0.13

0.51

0.12

0.38

0.30

0.09

0.18

0.18

0.19

0.00

0.2

0.09

0.27

-

-

-

-

-

-

-

-

0.451

0.475

0.464

0.446

0.354

0.295

0.261

0.208

0.215

0.173

0.188

 

Considering landings on runway 25R for example, the values for x and y according to Figure 1.3 are 0.4 and 10.0km respectively. Applying Equation 2 gives km-2. Substituting this into Equation 1 gives:

/year/km2

The number of plane movements has been divided by 8 to take into account that half of movements are take-offs and only a quarter of landings use runway 25R. This effectively assumes that each runway is used equally and the wind blows in each direction with equal probability.

The target area is estimated at 15,000m2 or 0.015km2 (3 tanks of 80m diameter). This gives a frequency for crashes into the tanks associated with landings on runway 25R as 2.0 x 10-9 per year. Repeating the calculation for landings and take-offs from all runways gives the results shown in Table 1.2.

Table 1.2        Aircraft Crash Frequency onto South Soko LNG Tanks

Runway

Landing (per year)

Take-off (per year)

07R

07L

25L

25R

0

0

1.5 x 10-9

2.0 x 10-9

6.9 x 10-12

3.1 x 10-11

0

0

Total

4.5 x 10-9

3.8 x 10-11

 

The crash frequencies for take-offs are well below 10-9 per year. Impacts from aircraft landing accidents have a frequency close to the 10-9 per year threshold. However, the storage tanks will be cut into the hillside at the Black Point site. There will be no line of sight from the thresholds of runways 25L and 25R to the LNG storage tanks; and with the tanks tucked in closely behind the hillside, impact from an aircraft will be much less likely than indicated by Table 1.2. Assuming a shielding factor of 0.1 gives a total impact frequency of 4.5 x 10-10. The combined frequency of all take-off and landing crashes onto the LNG tanks from activities on all runways is less than 1 x 10-9 per year. Aircraft crash is therefore neglected from the analysis.

The process units have a smaller area than the tanks and will also be shielded by the hill. Crashes into process areas are hence neglected from the analysis.

1.3.2                                Helicopter Crash

Helipad Activity

The Black Point Power Station has a helicopter landing pad although the frequency of use is low with perhaps one landing/take-off per week. The approach, landing and take-off stages of an aircraft flight are associated with the highest risk and therefore the possible impact of helicopter crashes on the LNG terminal facilities were assessed.

Historical data show that helicopter accidents during take-off and landing are confined to a small area around the helipad [24]. 93% of accidents occur within 100m of the helipad. The remaining 7% occur between 100 and 200m of the helipad. There have been no serious helipad related incidents resulting in a crash beyond 200m of the helipad.

The distance of the existing helipad from the proposed LNG terminal is approximately 900m. This is well beyond the range of accidents associated with helipad activity. It is therefore concluded that the terminal will not be exposed to any impact risks from helicopters using the helipad.

Passing Helicopters

Although take-off and landings present the greatest portion of risk from a flight, crashes during the in-flight stage of a journey may also occur. Black Point, however, does not lie near any helicopter flight paths. There are Government Flying Service helicopters in use in Hong Kong performing search and rescue, medical evacuations, surveying, fire fighting etc. type operations but these are few in number.

The CMPT Reports [27] gives a frequency of in-flight accidents of 1.2 x 10-5 per flying hour. However, only 17% of these are severe enough to cause a fatality. Assuming that incidents involving fatalities are a reasonable measure of uncontrolled crashes that may impact a facility, then the frequency becomes 2.0 x 10-6 per flying hour. The Government Flying Service conducted 4529 hours of operation in 2005 [28]. Assuming that a crash can occur anywhere within the 2922 km2 of Hong Kong, the crash rate becomes 3.1 x 10-6 /km2/year. Applying a target area of 0.015 km2 (the footprint area of 3 LNG storage tanks) gives a frequency of 4.6 x 10-8 per year. A crash into process areas would be even less likely due to the smaller footprint area. No credit has been taken for the protection offered by the hillside in this calculation and it is therefore conservative.

The frequency of helicopter crashes into the process area is small compared to the frequency of internal and process related failures. As an example, the vaporisers occupy an area of approximately 6000m2. A crash into the vaporiser area would have a frequency of 1.8 x 10-8 per year based on the above calculations. This would undoubtedly cause damage; however, the generic failure frequency of all vaporisers combined is 8.5 x 10-5 per year. The additional hazard from helicopter activities is 3 orders of magnitude smaller and may be neglected with negligible change in the risk calculations. Hazards from helicopters are therefore not treated separately but are covered by the generic failure frequency.

Consequence of Helicopter Impact

There have been several studies related to the possible impacts of aircraft collisions with hazardous facilities such as nuclear power stations. A study by the Nuclear Energy Institute [29] investigated the damage that could be caused by a fully loaded wide-bodied Boeing 767 crashing at 350 mph (560 km/h) into a nuclear reactor containment structure. These containment buildings are cylindrical shaped buildings with a dome roof, typically 140 feet high and 140 feet diameter. They are made of pre-stressed concrete with wall thicknesses from 3.5 to 4.5 feet. This is very similar to the construction of LNG storage tanks, which have a concrete wall of about 0.8m thickness. The conclusions of the report were that the concrete containment would not fail from such an impact.

In another study [13], the consequence of a Boeing 767 commercial aircraft crashing into an LNG storage tank was specifically assessed. It was concluded that aircraft are constructed from mostly soft materials and only the core of the engines is capable of penetrating an LNG full containment tank. Further, the engine would need to impact at near perpendicular incidence otherwise the engine will simply deflect off the tank.

These studies demonstrate the structural strength of the LNG storage tanks. They are designed to withstand major impacts. Naturally, a helicopter crash is very different from a 200 tonne airliner travelling at 560 km/h. Most helicopter crashes have an impact velocity below 50 m/s [24]. A fully loaded 5-tonne Sikorsky S76C with an impact velocity of 50 m/s would have a kinetic energy 390 times less than the Boeing 767 analysed by the Nuclear Energy Institute. It is concluded that a helicopter crash will not cause a failure of the LNG storage tanks.

Summary

Activities at the helipad will not impose any risks on the terminal. The frequency of passing helicopters crashing into the storage tanks was estimated at 1.2 x 10-7 per year, however, a helicopter impact would not cause a failure of a tank.

An impact into process areas or piping (including the piping on the LNG tanks) could cause damage but the frequency of such events is much lower than generic failure frequencies. The additional risk from helicopter activity is negligible. Scenarios involving helicopter activities were therefore not treated separately but are covered by the generic failure frequency.

1.3.3                                External Fire and Explosion Hazards

The LNG tank may be exposed to radiation and fire effects as well as  explosion overpressure effects including flying debris arising from ignition of flammable gas leak in process units located adjoining an LNG tank or neighbouring facilities. A study by Giribone et al [3] shows that a full containment tank with outer concrete shell (of about 0.8m thick wall and 0.5m thick roof) can withstand fire and flying debris impacts without any damage to the tank containment.

The study by Giribone shows that the stability of the concrete structure is not jeopardised by a fire scenario with an incident heat flux of 50kW/m2 (which is higher than the maximum allowable flux of 32 kW/m2 as per the design code EN 1473) over a period of 8 hours. Although through cracks in the concrete structure may occur, the structural integrity of the tank is not compromised.

The study by Giribone also shows that a flying debris of 2000kg in mass (a typical 4” valve weighs no more than 125kg) travelling at 50m/s causes no significant impact on the concrete structure, though cracks may be created by such an impact. The abovementioned studies [13, 29] suggest that the tanks can withstand considerable greater impact.

Based on the above it can be concluded that the outer concrete structure of a full containment tank provides significant resistance against external fire and explosion hazards. Failure of the tank due to such events will not occur. It may be also be noted here that the extent of process equipment and piping is very limited in a receiving terminal, as compared to an export terminal consisting of a liquefaction plant, and the hazards of fire and explosion events with potential to affect the tank is not significant. Hence external fire and explosion hazards causing damage to the tank is considered to be negligible. In any case, the frequency of tank failure considered in Section 1.1.3 is considered representative of all potential failure modes.

1.4                                      Intentional Acts

A full scope Safeguards and Security Risk Assessment (SSRA) was conducted in Hong Kong for the Project from 24 to 28 April 2006. The SSRA integrated key elements of the Critical Assets Protection (RAMCAP) method under the US Department of Homeland Security Risk Analysis and Management. Safeguards and Security Risk Assessments are conducted to assess the probability and severity of intentional undesired events to develop countermeasure recommendations that mitigate identified risks. The risk assessment was for a 10-year period from April 2006 to April 2016.

Given the sensitive and confidential nature of the analysis and in order to protect public safety, the complete SSRA report is available only on a need to know basis to GOHK and CAPCO security personnel.

The team who prepared the SSRA reviewed the history of terrorist events on LNG terminal facilities and determined that there has never been a terrorist incident at any LNG terminal or LNG carrier that has resulted in a death or injury to a member of the public. The relative risk of a terrorist induced injury or fatality to the public at a Hong Kong terminal was also compared to that for other terminals around the world, including Asia, Europe and the Americas. The qualitative conclusion of the team is that the risk of a terrorist induced incident having any consequence to public health and safety is extremely low. 

The Project was benchmarked against other LNG facilities worldwide in the United States, Europe, and the United Kingdom, where similar qualitative security risk assessments have been conducted. These other risk assessment studies were also presented to the local Authorities and their conclusions were similar to the SSRA report prepared for this project.

Based on the benchmarking comparison on the risk of terrorist threats, the Project was assessed to be at a lower security risk threat level than the other LNG terminals, indicating that the Hong Kong sites are much lower in exposure compared to the other site locations.

The conclusion reached by the SSRA team was that based on the overall low threat environment in Hong Kong and the risk analysis of the worst-case scenario of an intentional act in nature, their recommended risk mitigating measures would help to reduce the scenarios associated risk to a level that is well within levels generally accepted by industry globally, for either a Black Point or South Soko Island terminal. It was also concluded that whilst risk of a terrorist act directed at both sites is extremely low, the consequence of a terrorist event on the public would be greater for an incident at a Black Point terminal or against an LNG carrier on portions of the marine transit route to Black Point than for a South Soko terminal or its marine transit.

The SSRA team assessed the consequences that would result from the terrorist induced scenarios evaluated. Based on the available published studies, the team determined that the consequences of these events did not exceed the worst case events evaluated in the Terminal and Marine QRA studies.

1.5                                      Conclusions

Based on a detailed review of the various failure modes along with the safeguards provided, the failure frequency of the full containment LNG tank is taken to be 1x10-8 per tank-year. This failure frequency encompasses internal and process causes of failure as well as any natural/external hazards.

External hazards to the LNG storage tanks from aircraft were assessed to have a frequency below 1 x 10-9 per year. Helicopter crashes have a small but quantifiable frequency but the tanks are designed to withstand such impacts and so there would be no consequences from such an event. Hence, failure of the storage tanks due to helicopter impact was also excluded from the analysis.

Helicopter impacts into the piping and process areas are a possible scenario, but the frequency of occurrence is small compared to process risks. Helicopter crash scenarios are therefore included in the generic frequencies.

 

References

[1]     Baker, N., Creed, M., Stratification and Rollover in Gas Storage Tanks, Trans IChemE, Vol. 74, Part B, 1996

[2]     BS 7777: Flat-bottomed, vertical, cylindrical storage tanks for low temperature service, BSI, UK 1993

[3]     Giribone, R., Claude, J., Comparative Safety Assessment of Large LNG Storage Tanks, Eleventh International Conference on LNG, 1995

[4]      EN 1473: Installation and Equipment for Liquefied Natural Gas – Design of onshore installations, BSI, UK, 1993

[5]      TNO, Guidelines for Quantitative Risk Assessment (The Purple Book), Report CPR 18E, The Netherlands Organisation of Applied Scientific Research, Voorburg, 1999.

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

[7]      Taylor, J.R., Risk Analysis for Process Plant, Pipelines and Transport, London: Spon, 1994.

[8] E&P Forum, E&P Forum QRA Datasheet Directory, 15 October 1996.

[9]      Davies, T., Harding, A.B., McKay, I.P., Robinson, R.G.J., Wilkinson, A., Bund Effectiveness in Preventing Escalation of Tank Farm Fires, Trans IChem E, Vol 74, Part B, May 1996.

[10]    Christensen, R.A. and Eilbert, R.F., Aboveground Storage Tank Survey, EL RN-623, Entropy Limited, Lincoln, MA, 1989.

[11]    Batstone, R.J. and Tomi, D.T., Hazard Analysis in Planning Industrial Developments, Loss Prevention, 13, 7, 1980.

[12]  Canvey: a second report, HSE, 1991.

[13]    Hazards Analysis of a Proposed LNG Import Terminal in the Port of Long Beach, California, Quest Consultants Inc., Aug 2005.

[14]   GCO, Review of earthquake data for the Hong Kong region, GCO Publication No. 1/91, Civil Engineering Services Dept., Hong Kong Government, 1991

[15]   GEO, Seismic hazard analysis of the Hong Kong region, GEO Report No. 65, Geotechnical Engineering Office, Government of the HKSAR, 2002

[16]    Scott, D.N., Pappin, J.W., Kwok, M.K.Y., Seismic Design of Buildings in Hong Kong, Hong Kong Institution of Engineers, Transactions, Vol. 1, No. 2, p.37-50, 1994

[17]    ASCE/SEI 43-05, Seismic design criteria for structures, systems, and components in nuclear facilities, ASCE, 2005

[18]   Personal communication on tank capacity studies with Technigaz, 2003

[19]    Eurocode 8: Design of Structures for Earthquake Resistance, European Committee for Standardisation, 1998

[20]    Mizuno, T., Wadano, Y., Terada, N., Tsunomura, T., Demonstrative Safety Assessment of LNG Storage Facility Earthquake Resistance and Reinforcement of Crisis Management at LNG Receiving Terminals, Twelfth International Conference on LNG, 1998

[21]    www.hko.gov.hk

[22]    Lee, B. Y., Report of Hong Kong in the International Tsunami Seminar in the Western Pacific Region, International Tsunami Seminar in the Western Pacific Region, Tokyo, Japan, 7-12 March 1988.

[23]   LNG Receiving Terminal Project - Environmental Design Basis Report, Ove Arup & Partners, 2005.

[24]    Byrne, J. P., The Calculation of Aircraft Crash Risk in the UK, HSE\R150, 1997.

[25]  www.ntsb.gov/aviation/Table6.htm

[26]    Annual Review of Aircraft Accident Data: U.S. General Aviation, Calendar Year 2001, National Transport Safety Board.

[27]    Spouge, J., A guide to Quantitative Risk Assessment for Offshore Installations, CMPT, 1999.

[28]  www.gfs.gov.hk/perform.htm

[29]   Aircraft Crash Impact Analyses Demonstrate Nuclear Power Plant’s Structural Strength, Nuclear Energy Institute, Dec 2002.