Kristiansen Svein. Maritime Transportation: Safety Management and Risk Analysis

Подождите немного. Документ загружается.

lifejackets and immersion suits, proper use of lifeboats and liferafts, behaviour in these

evacuation crafts, use of first aid, etc.

14 . 7 EVACUATION S IMUL ATION

SOLAS specifies the following maximum times for key evacuation phases on passenger

ships:

. The maximum time from when abandon ship signal is given to when all surviva l craft

are ready for evacuation is 30 minutes (Chapter III, Regulation 11).

. The maximum time for abandonment of mustered people is 30 minutes (Chapter III,

Regulation 21.1.4).

The emerging computer simulation technology prompted IMO to develop standards

for the adoption of such techniques in the assessment of evacuation effectiveness. In 2002

the Maritime Safety Committee (MSC) of IMO formally adopted the ‘Interim guidelines

for evacuation analysis of new and existing passenger ships including Ro-Ro’ (IMO,

2002b). These guidelines only address the assembly/mustering part of the evacuation

process and two scenarios are defined, namely day and night conditions. The SOLAS

performance requirements are based on calm weather conditions, no list, and no effect of

fire. It is evident that in harsh weather conditions, with list an d/or the effects of fire, it will

be much more difficult to achieve evacuation within the given performance requirements.

This section of the chapter will take a closer look at evacuation simulations. First,

however, a brief introduction to crowd behaviour is given. Not considering crowd

behaviour is considered to be a major shortcoming of many evacuation simulation models.

14.7.1 Crowd Behaviour

Jørgensen and May (2002) have discussed a number of important issues related to cro wd

behaviour. The authors point to the fact that evacuation simulation models basically

estimate individual behaviour and more or less neglect so-called crowd behaviour, which

according to them is a major shortcoming of these models. Jørgensen and May have

defined the concept of group-binding, which expresses the fact that people both rationally

and emotionally have an interest in finding their relatives before being evacuated. The

crew ideally manages the mustering of crowds in an emergency situation, but due to

group-binding people will often be non-compliant to the instructions given by the crew

and instead focus on finding their relatives. The degree of group-binding will be a function

of the so cial composition of the passenger s: singles, couples, families, and groups of

friends. The effect of group-binding will also be related to the size and arrangement of the

vessel as well as the time of the day. Jørgensen and May studied the social composition of

the passengers on different Danish vessels, and interviewed people about their willingness

to disobey crew instructions. An average of 30% of the passengers would disobey

crew instructions in order to find family members and other people they felt closely

connected/related to. With an estimated probability of actually being separated of 0.2,

446 CHAPTER14 EME RG EN CY PREPARE D N ESS

the group-binding problem would affect 6% (i.e. 0.3 0.2 ¼0.06) of the passengers in a

given situation.

Jørgensen and May (2002) also discuss panic in relation to emergency situations.

As cited earlier (see Table 14.5), it has been estimated that 1–3% of a group will panic

and/or lose mental control. Jørgensen and May challenge this view based on psychiatric

generalizations and the fact that crowd behaviour is not taken into consideration. Panic

behaviour should also be seen as a sociological phenomenon, and they refer to work by

Berlonghi (1993) who makes a dist inction between different crowd phenomena:

. Passive crowd (e.g. spectators)

. Active crowds

– Hostile crowd (e.g. mobs)

– Escape crowds (often characterized by panic)

– Acquisitive crowds (often characterized by craze)

– Expressive crowds (often characterized by mass hysteria)

Other aspects of the realism of evacuation simulations are also discussed by Jørgensen

and May (2002). First, to verify a numeric al simulation model it is necessary to run large

and full-scale evacuation exercises with people in actual ships. This, however, is impossible

in practice, mainly for economic reasons. Secondly, in terms of arranging such evacuation

exercises, it is problematic to make the exercise fully realistic, as people will not be

influenced by the perception of danger and feeling of urgency that characterizes real

emergency situations. It may also be dangerous to arrange such realistic evacuation

exercises as real panic may arise.

14.7. 2 Modelling the Evacuation Process

The evacuation of crew and passengers is a process involving a number of phases. In terms

of modelling evacuation processes the following phases can be used:

1. Detection and acknowledgement of an emergency.

2. Sound the alarm.

3. Recognition (by people) of the alarm.

4. Collection of lifejacket, orientate oneself about the situation.

5. Search for and unite with family and friends.

6. Evacuate to safe place or mustering station.

7. Prepare and deploy survival craft or escape system.

8. Board survival craft.

9. Launch craft or leave the vessel.

10. Rescue by external resource.

The simulation approach replicates the evacuation process described above in the time

domain. In a time-stepping mode the behaviour of each individual is estimated, taking into

consideration the physical conditions and interaction with other people. The vessel is

1 4.7 EV ACU A TION SIMU LA TION 447

normally incorporated into the simulation model by a two-dimensional space grid. The

result of the simulation is an estimation of the total time of evacuation. As pointed out

by Galea et al. (2002), the simulation must address a number of aspects:

. Configurational: The physical layout and arrangement of the vessel with dimensions of

rooms, corridors and stairways.

. Environmental: Environmental factors that affect people under the evacuation, such as

list, ship motion, presence of debris, heat, smoke, toxic substances, etc.

. Proce dural: Basic rules for the phases in the evacuat ion process, for example related to

the guidance of passengers by crew, the organization at mustering stations, etc.

. Behavioural: Characteristics of how individuals behave and perform. The group of

people onboard should reflect a realistic composition in terms of sex, age, walking

speed and ability to respond adequately. Som e of these attributes may be dynamic and

change value during the evacuation.

The EXODUS numerical simu lation tool (Galea et al., 2002) con sists of a number of

interacting program modules as illustrated in Figure 14.10. The model considers the

interaction of people relative to other people, physical arrangement, the state of the vessel

and fire threat. The models are rule-based and the behaviour of individuals is based on

heuristic rules .

The ‘behaviour’ module in EXODUS is critical for the realism of the simulation. It

controls how people respond to the changing situation and controls the ‘move ment’

module. It functions on two levels, globally and locally, where the former addresses the

decision on escape strategy and the latter determines behavioural responses and decisions

made locally by individuals. For exampl e, through the ‘behaviour’ module EXODUS

reflects reactions to such phenomena as congestion and group ties.

The EXODUS model has a number of output formats that visualize the development

of the evacuation. A so-called footfall contour map indicates the most heavily used routes

Figur e 14.1 0. Module inte ractions in the EX O DU S numerical simulation tool. ( Source: Galea et al ., 2002. )

448 CHAPTER14 EME RG EN CY PREPARE D N ESS

that passengers use during the evacuation. Another format presents the final assembly

and density of people at the designated mustering stations. This format is presented in

Figure 14.11. The simulation package also offers an option where parts of the evacuation

can be viewed ‘live’ in 3D as indicated in Figure 14.12, which makes it possible to study the

effects of the ship’s arrangement and potential ‘bottlenecks’.

The effectiven ess of an evacuation may be exp ressed by a number of variables:

. Times:

– For individuals to muster

– Total time us ed to muster and evacuate

– Time wasted in congestions

– Time to clear particular compartments or decks

Figure 1 4.12. Live 3D view/presentation of the EXODUS simulation. (Source: Galea et al., 2002.)

Figur e 14.11. Population density at mustering stations. (Source: Galea et al., 2002.)

1 4.7 EV ACU A TION SIMU LA TION 449

. Distance travelled by individuals

. Flow rate through doors or openings

An experimental evacuation exercise has been cond ucted for a so-called Thames

pleasure boat with two decks (Galea et al., 2002). A total of 111 volunteers, from 16 to 65

years of age, were engaged in the exercise. 49 of these were located on the lower deck and

62 on the upper deck. For each deck there were four exits, two forward and two aft, and

a twin set of stai rcases connected the two decks. During the experi ment the vessel was

moored with its starboard side to the jetty. Several evacuation tests were performed with

different restrictions on access to the exits. The results from these evacuation exercises

were then compared to the EXODUS simulation model where the predicted evacuation

times were within 7% of the exp eriments.

14.7.3 A Simulation Case

The evacuation time for a large passenger ship with a total of 650 passengers has been

estimated using the EXODUS simulation tool (Galea et al., 2002), and some interesting

aspects of this simulation case are presented below. The vessel in question had ten decks

divided into three vertical fire zones. The muster deck (i.e. deck 8) and the deck below

are shown in Figure 14.13. The initial distribution of passengers within the ship before

evacuation was as shown in Table 14.11. Passengers in fire zones 1 and 3 are assumed to be

in their cabins. Fire zones 1 and 3 have four staircases each, with each staircase located in

the far corners and only allowing a single lane of passengers. Fire zone 2 has a single

centrally located staircase allowing two lanes of people to move.

Figur e 14.13. Deck arrangement. (Source: Galea et al., 2002.)

450 CH APTER14 EME RG ENCY PREPARE DNESS

The simulation was done for night-time conditions with a response tim e from 7 to 13

minutes, allowing for people sleeping in their cabins to wake up and get dressed (an IMO

requirement). Travel speeds are also specified by IMO, and these are summarized in

Table 14.12. Depending on gender, age and degree of impairment, the travel speed in flat

terrain varies by a factor of 3. The speed of walking downstairs is 30% lower than that on

flat terrain, and less than 50% lower than walking upstairs. The IMO regulation specifies

that the simulation should be run 50 times with random values. MSC Circular 1033 (IMO,

2002b) gives ranges of variation for the travel speeds cited in Table 14.12, and the range of

variation is in the order of 25%.

The estimated mustering times for an even keel (i.e. no heel) vessel condition are

given in Table 14.13. The fact that fire zone 1 has the longest mustering time can be

explained by the relatively high number of passengers, which may result in congestion, and

that the evacuation includes walking up the stairs from deck 6 and 7 to the muster deck

(i.e. deck 8).

IMO specifies that the dimensioning evacuation time should be taken as the highest

value of four scenarios and an extra 10 minutes added to account for the assumptions and

Table 14.11. Initial location of passengers

Deck Fire zone 1 Fire zone 2 Fire zone 3

6 172 28

7 176 24

9 150

10 100

Source: Galea et al. (2002).

Ta b l e 1 4 .1 2 . Pa ssenger travel speed (m/s) as specified by MSC Circular1033 (IMO, 2002b)

Age (years)/

impairment

Walking on flat

terrain (m/s)

Walking downstairs

(m/s)

Walking upstairs

(m/s)

Female <30 1.24 0.75 0.63

30–50 0.95 0.65 0.59

50þ 0.75 0.60 0.49

Impaired 1 0.57 0.45 0.37

Impaired 2 0.49 0.39 0.31

Male <30 1.48 1.01 0.67

30–50 1.30 0.86 0.63

50þ 1.12 0.67 0.51

Impaired 1 0.85 0.51 0.39

Impaired 2 0.73 0.44 0.33

1 4.7 EV ACU A TION SIMU LA TION 451

uncertainties in the model. The highest value in each scenario is to be taken as the 95th

percentile. For the given scenario (see Table 14.13) the estimated maximum mustering

time was 15 minutes and 58 seconds, and given an added safety margin of 10 minutes the

regulation says that the vessel needs 25 minutes and 58 seconds to evacuate/must er.

Congestion (of people) in specific areas of the ship can be studied in the

EXODUS model, and for the case described above there was congestion in the range of

2.2–3.5 persons/m

at the base of the staircases for 19 seconds. IMO defines a congested

area as an area where there is a passenger density of 4 persons/m

The effect of heel on the mustering time is not very significant in the present version of

the EXODUS simulation model. Muster times for fire zone 1 are shown in Tabl e 14.14

for 0



,10



and 20



heel, respectively. With 10



heel the mustering time increa ses by only

2 seconds, and for a heel of 20



the increase in time is still marginal with 26 seconds. 20



heel is quite dramatic and makes it considerably more difficult to move around the vessel.

The heel itself may, in addition, result in increased stress and even panic. The increase in

mustering time will therefore most certainly be much larger than 26 seconds. Evacuation

from partly capsized passenger vessels is discussed in more detail below. Table 14.14

confirms the inability of the EXODUS model to take factors such as change in human

behaviour, physical chaos and potential loss of electricity because of heel into

consideration. When using such simulation tools it is important always to have a clear

understanding of the inherent limitations.

14.7.4 Evacuation from Partly Capsized Vessels

Planning and training for evacuation of passenger vessels normally assumes that the vessel

is in an upright or only moderately heeled condition. However, experience shows that this

Ta b l e 1 4 .13 . Range of mustering times with even keel vessel condition

Estimate Fire zone 1 Fire zone 2 Fire zone 3

Minimum 14

Average 15

Maximum 15

Source: Galea et al. (2002).

Ta b l e 1 4 .1 4 . Muster time in the EXODUS model for fire zone 1 at different

degrees of heel

Heel 0



Assembly time 15

Source: Galea et al. (2002).

452 CHAPTER14 EM ERG EN CY PREPAR ED NESS

is not necessa rily the case in real emergency situations:

. European Gateway (1982): Collided with another ship off Felixstowe (England) as a

result of confusion at a bend in the channel. The collision resulted in puncturing of the

main vehicle deck and the generator room below the waterline. Because of asymmetric

flooding the vessel star ted to heel, reachi ng 40



in only 3 minutes, at which point the

bilge grounded. During the next 10–20 minutes the ship rolled on to its side. Six of

the 70 people on board drowned.

. Herald of Free Enterprise (1987): Uncontrolled flooding through the open bow door

resulted in rapid heel to 30



only minutes after leaving port at Zeebrugge (Belgium).

Within 90 seconds the vessel heeled/capsized to 90



, at which point the side of

the vessel was resting on the seabed in the shallow water. At least 193 passengers and

crew died.

. Estonia (1994): Failure of the bow door and ramp in a severe storm in the north

Baltic Sea led to rapid water ingress onto the vehicl e deck. Because of free-surface

effect the ship heeled to 30



within minutes, increasing to 90



only 20 minutes afte r the

bow ramp opened. About 10 minutes later the ship sank completely, resulting in

852 fatalities.

According to Spouge (1996), the difficulty of evacuation increases dramatically when a

vessel heels beyond 45



. The main causes of death in such situations are as follows:

. Falling headlong with extreme heel

. Shock of water immersion results in heart diseases or other paralysing illnesses

. Drowning due to rising water in compartments and inability to swim or escape

(primarily to higher level)

After capsizing a vessel will usually come to rest in a stable position for a period of

time. This will give some time for evacuation from inside the ship as well as rescue away

from the ship. After a while, depending on the vessel’s construction and the extent of

damage, further water ingress will result in heel to 180



or sinking. Spouge (1996) has

assessed the fatality risk for this accident scenario. The consequences of an evacuation of a

passenger ferry, prima rily consisting of large public spaces (i.e. type A), after capsize to 90



are summarized in Table 14.15. Of the estimated 45% fatality ratio, most people perished

inside the vessel. The data in Table 14.15 also emphasize the importance of dry compared

to wet evacuation. For a ship with cabins (i.e. type B) the fatality rate during night

conditions will typically be 56%, considerably more than for type A vessels with mainly

large open public spaces.

Spouge (1996) also proposed technical measures to improve the evacuation success

rate. It was estimated that the survival rate could be improved by 3–7% for capsize

scenarios beyond 45



if additional escape equipment and arrangement features were

implemented. The following types of equipment/features were proposed: ladders, bridges,

ropes, escape windows and elimination of full height partitions in public areas. In

addition, limiting heel is considered very important in terms of saving lives.

1 4.8 EV ACU A TION SIMU LA TION 453

14. 8 POLLUTION EMERGENCY PLANNING

1 4.8.1 Contingency Planning

MARPOL, the International Convention for the Preven tion of Pollution from Ships

(IMO, 1997), includes requirements on pollution emergency pla nning. MARPOL

Regulation 26 of Annex I requires that all oil tankers of 150 grt and above, and

all other ships greater than 400 grt, shall carry a Shipboard Oil Pollution Emergency

Plan (SOPEP) approved by the Administration (IMO, 2001b). Regulation 16 of

Annex II requires all ships of 150 grt and above, certified to carry noxious liquid

substances in bulk, to have onboard a pollution emergency plan . Ships to which both

regulations apply may have a combined plan called a Shipboard Marine Pollution

Emergency Plan (SMPEP). Such a pollution emergency plan should cover the following

four elements:

. Procedures for reporting pollution incidents.

. A listing of authorities to be notified.

. A detailed description of actions to be taken by the ship’s crew to reduce or control

a discharge of oil or a noxious liquid substance.

. Procedures for co-ord inating shipboard activities w ith national and local

authorities.

Some countries (coastal states) define addition al measures to be taken agains t marine

pollution (ICS, 2002). All ships operating in the territorial waters of these co untries must

Table 14.15. Fatalities and survivors on a passenger ferry for short crossings in

the case of 90



capsize

Outcome Relative number

of people (%)

Killed by fall to side of compartment 5

Killed by shock of immersion 4

Drowned in rising water 26

Drowned awaiting rescue 10

Total fatalities 45

Escaped on own 3

Rescued from dry by survivors 23

Rescued from water by survivors 1

Rescued from dry by rescuers 26

Rescued from water by rescuers 2

Total surviving 55

Source: Spouge (1996).

454 CH APTER14 EME RG ENCY PREPARE DNESS

adhere to these requirements. For example, the US requires the following additional

measures:

. The ship must identify and ensure, through contract or other approved means, the

availability of private firefighting, salvage, lightering and clean-up resources.

. A qualified individual with full authority to implement the response plan, including the

activation and funding of contracted clean-up resources, must be identified.

. Training and dril l procedures shall be described.

IMO (2001b) and the International Chamber of Shipping (ICS, 2002) have published

guidelines that may assist companies in setting up a SOPEP or SMPEP. The ICS guideline

gives flowcharts that support the decision-making process during an emergency.

Figure 14.14 shows the guideline for the reporting of a polluting discharge. MARPOL

specifies in detail what kind of information has to be reported: ship description, position,

nature of dischar ge, vessel damage, etc. The guidelines further list whom to contact in case

of a polluting discharge:

. Coastal state

. Port: terminal master, agent, port authority, etc.

. Ship interest contacts:

– Head office

– Charterer

– Classification society

– P & I Club

Another important step in the pollution emergency plan is to mobilize the vessel’s

pollution prevention team, which involves all key officers onboard the vessel. The

emergency plan should have detailed job descriptions for each team member. For a

specific spill scenario the pollution emergency plan, which gives a description of measures

to be taken, should be given in both plain text and as a checklist. The plan is to be

categorized according to the source of the spill and the causes. Examples of an emergency

plan from the ICS guideline are given in Tables 14.16 and 14.17 in plain text and as a

checklist, respectively.

14.8.2 O rganization

Maritime pollution accidents may directly involve a number of parties:

. Master and crew

. Shipping company

. Salvage vessel

. Port administration

. Firefighting brigade

. Pollution prevention agency, etc.

14.8 POL LUTIO N EM ERGEN CY PLAN N I NG 455