Nonstop Simulating Ship Evacuation

8/17/2019 Nonstop Simulating Ship Evacuation

1/8


2/8

On 28 September 1994, the cruise ferryEstonia, en route across the Baltic Seafrom Tallinn to Stockholm, sank in heavy

weather, carr ying 989 passengers and crew. Theaccident, which claimed 852 lives and was oneof the deadliest maritime disasters in the late

20th century, had a dramatic effect on the gen-eral perception of maritime safety.

In the aftermath, the International MaritimeOrganization (IMO) fundamentally redefinedits stance on ship evacuation safety, introduc-

ing a new, performance-oriented approach.The ability to evacuate a ship efficiently in an

emergency was made a design requirementfor RoRo passenger ships. Pursuant to the newrules, RoRo passenger ships built on or after1 July 1999 must be designed to ensure success-ful evacuation within 60 minutes.

Initially, evacuation analyses relied on the so-called “Simplified Evacuation Analysis” meth-

od developed by the IMO Fire Protection Sub-Committee. From 2000 to 2002, this approach

was further developed by the committee. Sub-sequently, they proposed an alternative methodto determine evacuation times based on com-

puter simulation, the so-called “Advanced Evac-uation Analysis” method. In response to theIMO requirements for the Advanced Method asdefined in the IMO “Guidelines for Evacuation

Analysis for New and Existing Passenger Ships”

of 2007, an innovative software tool was devel-

oped to enable evacuation analyses compliant with the IMO specifications. This tool is called

AENEAS.This paper gives an overview of the devel-

opment and current status of the IMO conven-

tions, codes, regulations, and guidelines con-cerning the safety of passengers on board ofpassenger ships and their evacuation. The rel-evant IMO policies address three vessel types:

high-speed craft (HSC), RoRo passenger ships

and cruise ships.In 1914, two years after 1,502 people had lost

their lives in the Titanic disaster, the Interna-tional Safety of Life at Sea (SOLAS) Convention

was adopted by an international conferenceconvened in London by the government of the

United Kingdom. 13 nations participated in thisconference. The 1914 version was eventuallysuperseded by SOLAS 1929, SOLAS 1948, SO-LAS 1960 (under the auspices of the IMO), andSOLAS 1974, whereby the latter is still in force

today. However, SOLAS 1974 has been amendedand updated many times. A revised version ofthe regulations contained in SOLAS chapter IIIrelating to life-saving appliances and arrange-ments took effect on 1 July 1998. These rules in-tend to ensure the greatest chance for passen-gers and crew to outlive a catastrophe at sea.

When the IMO Code of Safety for High-SpeedPassenger Craft (HSC Code) was first developedin 1992, the issue of passenger evacuation was

given due consideration. Explicit requirementsincluded in section 4.8 of the code call for a

Introduction and Historical Background

DISASTER.In 1994, ferryEstonia sank

on her way toStockholm.

TITANIC.1,502 passengers

lost their lives. Twoyears after the sin-

king the SOLAS Con-vention was adopted

in London.

P h o t o / I l l u s t r a t i o n : A c c i d e n t I n v e s t i g a t i o n

B o a r d

F i n l a n d

Stockholm Tallinn

Turku

Helsinki

28 nonstop 02/2008

EXTRA


3/8

Several different simulation tools and meth-ods have been developed for the above-men-tioned aspects of evacuation analysis. One ofthese software tools, AENEAS, is the result ofa co-operative effort by TraffGo and Germa-nischer Lloyd in Germany.

Model Basics

In general, the factors influencing an evacua-

FIGURE 1. Representa-tion of the oor planas a grid of cells. Thecellular automatonis a concept well es-tablished in computerscience.

“simplified evacuation analysis” for all HSC (forguidelines on evacuation analysis, please referto MSC/Circ.1166).

RoRo passenger ships built on or after 1 July1999 must comply with SOLAS regulation II/2-

13.7.4: “For new Class B, C and D RoRo passen-

ger ships constructed on or after 1 July 1999,escape routes shall be evaluated by an evacu-ation analysis early in the design process. The

analysis shall be used to identify and eliminate,as far as practicable, congestion which may de-velop during an abandonment, due to normal

movement of passengers and crew along es-cape routes. It has to be considered that thecrew needs to move along these routes in a di-

rection opposite to the movement of the pas-sengers.

In addition, the analysis shall be used todemonstrate that escape arrangements are suf-ficiently flexible to provide for the possibility

that certain escape routes, assembly stations,embarkation stations or survival craft may notbe available as a result of a casualty.”

Among the most important aspects in thedevelopment of safety guidelines are the in-crease in size of modern cruise ships and thelarger quantity of passengers and crew for

whom they are designed and certified to carry.These facts raise a number of urgent technical

questions, such as how to lift rescued peoplefrom a lifeboat or life raft onto another ship.

The so-called “safe return to port” concept

defines three basic evacuation scenarios: (i)the ship has to be abandoned “immediately”,(ii) there is sufficient time for “safe and order-ly evacuation and abandonment”, and (iii) thevessel is able to safely return to port. For case

(ii), the aim is to provide sufficient time (three

hours) for safe and orderly abandonment.

Model Basics

FIGURE 1 B. In caseof an emergencythe passengers areassumed to movefrom cell to cell

towards the exit.

COLORMAGIC.

The new RoPax ferryof Colorline Cruises

was optimizedaccording to IMOrules using AENEAS.

P h o t o : C o l o r l i n e C r u i s e s

29nonstop 02/2008

EXTRA


4/8

tion process can be classified into the follow-ing four categories:

Geometry Population Environment Hazards

These four categories are described briefly inthe following paragraphs.

Modelling the Geometry

In AENEAS, spatial geometry is representedas a grid of cells (cellular automaton). The cel-lular automaton is a concept well establishedin computer science for its robustness, ease

of use, scalability, and speed of computation(Figures 1 and 1 b).

In a simulation, individuals (represented by“agents”) are assumed to move from cell to cell

towards the exit, which is identified by exit sig-nage information stored within the cells andaccessible to the agents.

Social and Psychological Factors

Population diversity is represented in AE-NEAS in the form of a set of individuals(microscopic simulation). Each individualis assigned a set of parameters based onstatistic (usually Gaussian) distribution(Figure 2). To account for different sub-po-pulations, an arbitrary number of groups

with parameter set s of the ir own can be de-fined, and assigned separate roles and goals(Figure 3).

Environmental Factors and Hazards

The most significant environmental factors

and hazards in passenger ship evacuationsimulations are fire, smoke and ship move-ment. Fire and smoke are accounted for by the

Available Safe Egress Time (ASET ) parameter.The ship’s motion influences the Required SafeEgress Time (RSET), which can be determinedby the simulation tool. The safe-egress timeparameter will be described in greater detaillater.

Calibration and Validation

Calibration and validation are two conceptssimilar in nature but different in purpose.

CalibrationCalibration means tuning the model param-eters until the simulation results come closeto empirical data. To be considered as wellcalibrated, a model must trace the funda-mental diagram, as shown in Figure 4. Thenew and innovative software tool AENEASprovides a safety margin by erring slightly onthe conservative side.

Validation

Valid at io n me ans ch ecki ng wh et her th esoftware is fit for the intended use. The fol-lowing validation categories can be distin-guished:

Calibration andValidation

FIGURE 3. Denition of arbitrary groups, as well as crew and passen-ger routes facilitate the modelling of complex evacuation proceduresand scenarios.

FIGURE 2. In AENEAS the passengers are dened by demo-graphic parameters.

30 nonstop 02/2008

EXTRA


5/8

Functional validation Component testing

Qualitative validation, and Quantitative validation.

For illustration purposes, Figure 5 shows anexample from the twelve test cases given inMSC.1/Circ.1238. The result to be expected would be people moving either towards themain exit or towards the secondary exit, de-pending on their assigned route (indicated by

arrows).

Simulating the Evacuationof Passenger Ships

The model of AENEAS is described in detail inreferences. We shall limit our discussion to onemajor aspect of ship evacuation: the ship’s mo-tion as it influences the orientation and walk-ing speed of agents.

Modelling Ship Motion

Existing test results accounting for the influ-ence of rolling motion are either poorly docu-mented or too small in number for meaning-ful statistical evaluation. Therefore only theresults of static heel are used for modelling theinfluence of ship motion on the movement ofhumans on board.

Slope Inuencing the Reduction Factor

The ship’s heel influences the walking speedmainly by reducing it by a certain factor. Fur-thermore, the heel adds a drift to the move-ment of the agents, increasing the space theyrequire in transverse orientation relative to themain heel angle (Figure 6).

While it has not been analysed explicit-ly, normal movement of pedestrians is assu-med to stop when the slope angle exceeds 35degrees. Due to loss of friction, loose material will usually begin to slide when angles exceed

36 to 38 degrees. This is pure mechanics butappears to apply to pedestrians on a ship, as well. At these steep angles, the geometry be-

gins to change significantly. This effect wasnot taken into account by the simulation sin-ce it would consume too much computingtime. In a conservative estimate, agents in

Simulating the Evacuationof Passenger Ships

FIGURE 6. The inu-ence of transversalslope on the speed

reduction factor,applied to a at

surface by AENEAS.The agents need to

have more space.

FIGURE 4. Flow-density relation (funda-mental diagram).

FIGURE 5. A validation case from MSC.1/Circ.1238(“Cabin Area”). People moving either towards themain exit or towards the secondary exit, dependingon their assigned route.

main exit

secondary exit

2 persons 2 persons

2 persons

3.0 m

1.2 m

1.8 m

0 . 9

m

5 . 0

m

5 . 0

m

2 persons

2 persons2 persons

2 persons1pers.

2 persons 2 persons

2 persons 2 persons

0.9 m

0,00

0,20

0,40

0,60

0,80

1,00

0 5 10 15 20 25 30 35 40 45 50

slope /°

KRISO

SSRC

Monash

TNO

SHEBA

AENEAS

31nonstop 02/2008

EXTRA


6/8

relevant area1.2 x 1.2 m2=1.44 m2

AENEAS are assumed to move at a walkingspeed reduced to only five per cent at heel an-

gles between 35 and 45 degrees, and to stop

moving altogether when the heel angle increas-es further.

Corridor – Transversal Slope

Similar figures and formulae are available forlongitudinal corridors, transversal and lon-gitudinal stairs, etc. In the AENEAS software,they are implemented as factors reducing the

walking speed and influencing orientation and walking direction.

Slope Inuencing the Drift

Since a sloping geometry increases the spaceclaimed by the agents in the simulation, aslope-dependent drift is added to each agent.This increases the probability of each agentmoving downhill, an effect that results in anincreasing demand for space in the directionof the slope.

Application Example

In order to demonstrate the effects of both stat-ic heel and dynamic roll, a number of calcula-tions were performed on an exemplary RoPaxdesign by Flensburger Schiffbau-Gesellschaft(FSG RoPax 1800).

Since the trim and pitch angles are relativelysmall, their effect is more or less negligible. The

passenger distribution and parameters werechosen according to the “Night” case specifiedby the IMO regulations. Deviating from the reg-

ulation, however, the modelling process wasextended beyond mustering time to cover the

boarding process, as well.Figure 9 reveals that small static heel an-

gles do not cause significant problems for pas-

senger movement. Once the static heel angleincreases beyond 15 degrees, however, theamount of time needed for evacuation begins

to increase sharply as passengers have to strug-gle harder to advance upwards across the in-clined decks. In addition, the evacuation time

distribution spreads significantly. When the ship performs a roll motion

(T = 12.5 s), the effects on the evacuation time are

less significant for angles up to 20 degrees. In thiscase, the slopes agents must overcome will vary,and the problems associated with climbing up a

transverse slope are less relevant.

Workow Optimization

A simulation tool will be accepted by shipyardsmore willingly if it is easy to use and deliversresults quickly. An essential requirement is theimport functionality for CAD drawings, whichis implemented in AENEAS via a filter for thecommon data exchange format (DXF).

The overall effort involved in prepar-ing this mock-up example, starting fromthe CAD drawing and including the evalu-ation of the simulation results, was four

FIGURE 8. The RoPax 1800 design by FlensburgerSchiffbau-Gesellschaft.

WorkfowOptimization

FIGURE 7.Surrounding cells of oneperson (agent).

32 nonstop 02/2008

EXTRA


7/8

and a half hours. For a typical RoRo ship withsix to ten decks and 1,500 to 2,000 persons on

board, the process (covering all calculations,

documentation and the report as required bythe Guidelines) will take one to two weeks.

The Evacuation Analysis FrameworkThe precondition for safe evacuation can beexpressed as follows:

RSET < ASET,

where RSET represents the required safeevacuationtime and ASET the available safe egress time.

RSET is determined by the simulation (taking

into account ship motion, if applicable), while ASET may be determined by fire and smokecalculations or simulations and by the simula-

tion of the sh ip’s stability (influenced by waves,flooding, etc.).

Interpretation of Results;Acceptance CriteriaThe Guidelines for Evacuation Analysis(MSC.1/Circ.1238), SOLAS (chapter II-2, Regu-lation 13.7.4), the High-Speed Craft Code (HSC2000 Code) and the Fire Safety Systems Code(FSS Code) specify the requirements for escaperoutes, signage, maximum evacuation timeand identification of congestion (as design cri-teria).

Therefore the overall evacuation time mustbe determined in two separate steps. The firststep is to establish the walking time by simu-

lating the movement of crew and passengersto the assembly station (accounting for indi-

vidual reaction times).

This time parameter i is simulated foreach individual. The overall evacuation timeis then calculated according to the following

two formulae:

FIGURE 9.The effect of staticheel or periodic rollmotion (T=12.5 s)on the signicantevacuation duration

for a given maximumangle.

FIGURE 10.An optimized

workow improvesthe usability of the

tool. An essentialrequirement is the

import functionalityfor CAD drawings.

0:10:00

0:30:00

0:50:00

0 5 10 15 20

Stat. Heeling respectively max. Angle of Roll

Angle of Roll

Stat.

33nonstop 02/2008

EXTRA


8/8

T = 1.25 x max(ti) and treact = [400 s; 700 s]based on a log-normal distribution

T + 2/3 x (E + L) < 60 minutes

For RoRo passenger ships and passengerships with no more than three main verticalzones (MVZ) the time limit is 60 minutes. Forpassenger vessels other than RoRo passenger

vessels featuring more than three MVZ, thetime limit is 80 minutes.

ConclusionThis article summarizes recent advances inpassenger safety research. It focuses on evacu-ation-simulation tools developed in recent

years, taking the simulation software AENEASas an example.

Two major developments are currently driv-

ing the evacuation simulation business: auto-mation and integration. These developmentsreflect general trends in engineering technol-

ogy, aiming to reduce the workload and accel-erate the delivery of results, especially in con-nection with design changes.

Conclusion

References• Ando, K., Ota, H. and Oki, T.: Forecasting the Flow of

People. Railway Research Review 45, 8–14 (in Japa-

nese), 1998.

• DiNenno, P. (ed.): SFPE Handbook of Fire Protection

Engineering (2nd ed.). National Fire Protection Asso-

ciation, 1995.

• International Maritime Organization (IMO), Guidelines

for Evacuation Analyses for New and Existing Passen-ger Ships, MSC.1/Circ.1238, November 2007.

• Klüpfel, H,: A Cellular Automaton Model for Crowd

Movement and Egress Simulation, Dissertation, Uni-

versität Duisburg, 2003.

• Klüpfel, H.: The Simulation of Crowds at Very Large

Events. In: Schadschneider, A., et. al.: Trafc and

Granular Flow 2005. Springer, Berlin, 2006.

• Klüpfel, H.: The Simulation of Crowd Dynamics at Very

Large Events. Calibration, Empirical Data, and Valida-

tion. In: Waldau, N. et. al.: Pedestrian and Evacuation

Dynamics 2005. Springer, Berlin, 2006.

• Klüpfel, H., Meyer-König, T. and Schreckenberg, M.:

Models for Crowd Movement and Egress Simulation.

In: Hoogendoorn, S. et. al.: Trafc and Granular Flow

2003. Springer, Berlin, 2004.

• Meyer-König, T. et al.: Assessment and Analysis of Eva-

cuation Processes on Passenger Ships by Microscopic

Simulation, Pedestrian and Evacuation Dynamics,

Springer, Berlin, 2002.

• Predtetschenski, W. and Milinski, A.: Personenströme

in Gebäuden – Berechnungsmethoden für die Model-

lierung. Köln-Braunsfeld: Müller, 1971.

FIGURE 11.

Screenshot anddensity graph.

34 nonstop 02/2008

EXTRA

Documents

Nonstop Simulating Ship Evacuation