Guidance on the Design of Ships for Enhanced Escape and Evacuation

Project EGO
EPSRC funded project 2005-2007


 Project Partners:
Academic Partners: FSEG University of Greenwich: GR/T22100/01
Academic Partners: DRC University College London: GR/T22117/01
Industrial Partner: Directorate of Sea Systems, Defence Equipment and Support organisation of the UK Ministry of Defence (DES-SESea)



Traditionally, when designing a ship the driving issues are seen to be powering, stability, strength and seakeeping. When the broad form of the layout has been finalised, human factors issues related to crew numbers, ship operations and evolutions, such as evacuation, are either ignored, considered as an after thought or incorporated through a set of prescriptive rules. This can result in significant operational inefficiencies and potentially hazardous environments onboard. The overall objective of this multidisciplinary research project is to show the advantages of integrating the cutting edge technologies of Escape Simulation (i.e. maritimeEXODUS developed by the FSEG of the University of Greenwich) and Ship Configurational Design (i.e. SURFCON developed by the Marine Research Group of UCL). This will enhance the guidance for all parties in the design, regulation, construction and operation of ships with regard to the main aspects of personnel movement onboard. To achieve this, the project draws on the well-established expertise of the FSEG of the University of Greenwich in the area of fire and evacuation modelling and the Design Research Centre, Marine Research Group of UCL in the area of ship architecture design. While this project addresses the design of naval vessels, the principle behind the proposed methodology and the tool set produced have direct application to the design of commercial and passenger vessels.

PROJECT OBJECTIVES 1) To explore the impact on naval ship configurational design of issues associated with crew manning numbers, function and movement.
2) To identify key performance measures for successful crew performance in normal and extreme conditions.
3) To extend the ship evacuation software maritimeEXODUS to include additional non-emergency simulation capabilities.
4) To extend the ship design software so that it can provide a modelling environment that interactively accepts maritimeEXODUS simulation output for a range of crew evolutions.
5) To demonstrate a methodology for ship design that integrates ship configuration design with modelling of a range of crewing issues through PARAMARINE-SURFCON.

FSEG were responsible for objectives (2) and (3), UCL were responsible for objective (4) and both organisations were jointly responsible for objectives (1) and (5). By clicking on this link you will be connected to the UCL web pages associated with this project.
PROJECT SUMMARY

The partnership successfully achieved all objectives, demonstrating the viability of the methodology and the benefits that can be derived through its use. Key significantly novel advances made in the project include:

a) The use of simulations to evaluate personnel movement applied in the earliest stages of ship design.
b) The evaluation of personnel movement extended to include non-emergency scenarios.
c) A transparent and reproducible methodology developed to determine overall HF performance of a given ship design and shown to be both discriminating and diagnostic.
d) Guidance provided on the level of design definition necessary to carry out personnel movement simulations in preliminary ship design.
e) The project provided insights to ship designers on specific features that enhance and restrict personnel movement onboard heavily populated vessels.
f) The project indicated those research areas that remain to be addressed for the comprehensive simulation and assessment of personnel movement in non-evacuation scenarios onboard ships.

The figure below illustrates the overall relationship between the expertise domains and software tools on which each research partner led. This shows, in flowchart form, the design and analysis activities that were carried out in each of the tools. The interface tools developed in this project were separate from the PARAMARINE-SURFCON and maritimeEXODUS software and permitted the required information to be transferred between them.
 

 

PROJECT REPORT The full project report presented to the EPSRC may be downloaded as a pdf file. Download EPSRC project report here. NOTE: This document will be made available after the project has been assessed by the EPSRC.
 
THE HPM METHODOLOGY

A novel new discriminating and diagnostic methodology for assessing the HF performance of ship design has been developed which is systematic, transparent and repeatable. The method, known as the Human Performance Metric (HPM) allows the accurate and rapid assessment of HFs issues associated with vessel layout and crew operating procedures. The HPM works by systematically evaluating one layout design against another, whether two variants of the same design or two completely different ship designs.  The analysis is performed using the maritimeEXODUS evacuation simulation software.  The HPM involves specifying and defining a range of: Evaluation Scenarios, Performance Measures, Functional Groups and Weights.  Each of these terms will be described.

The HPM requires the specification of a range of relevant Evaluation Scenarios (ES) against which the vessel will be tested.  These scenarios are intended to define the scope of the challenges the vessel will be subjected to. In order to gauge vessel performance across a range of criteria, the ES are made up of both evacuation and Normal Operations (NOP) scenarios.  The nature of the scenarios are dependent on the type of vessel.  Relevant evacuation scenarios may include those required by MSC.1/Circ 1238 and include the IMO night and day scenarios or their naval equivalent.  The NOP scenarios are dependent on the nature and class of vessel. For example, a cruise ship application may require the time to empty the cinema is minimised while a naval vessel may require watch changes to be completed within set period of time.  Each ES is then simulated using the maritimeEXODUS software.

In addition to defining the ES, a range of Performance Measures (PM) must be defined that measure various aspects of personnel performance in undertaking the tasks associated with the ES.  PM for passenger ship evacuation scenarios may include the time required to complete the assembly process while for a naval vessel NOP scenario, the total number of water tight doors (WTD) opened and closed may be relevant. The suitability of the vessel layout will be evaluated for fitness of purpose through some combination of the PM resulting from the execution of the ES.  The PM values are determined from output for the ES produced by maritimeEXODUS.

As members of the ships complement may be involved in undertaking different tasks during a particular ES, the ships complement is divided into subgroups.  Membership of each subgroup is determined by the nature of the tasks undertaken by the individuals in the particular ES, with each subgroup being made up of people undertaking a common set of tasks. These subgroups are labelled Functional Groups (FG).   The introduction of FGs allows the analysis to focus on the performance of important subgroups of the crew whose contribution may swamp that of other FGs or be swamped by other FGs when considering the overall performance of the vessel. An example of a FG is the ‘damage control and fire fighting’ group which is a prime example of a FG used in circulation ES.  In addition to the FGs defined by specific sub-populations, a special FG, identified as Ships Company, is included in all ES. 

To complete the HPM, performance scores ai,j(PMk), for a particular design variant X associated with each PMk must be determined. The PM score (ai,j(PMk))  is simply its value derived from the execution of the maritimeEXODUS simulation software for each FGj within each ESi.

To allow a meaningful comparison between performance scores, each score is normalised, producing āi,j(PMk). The normalisation is performed using the largest performance score from the competing design variants X. Using this approach, all the HPM entries will be less than or equal to 1.0.   The larger the performance score, the worse the performance of the FG for that particular PM. 

An overall score can be determined for each FG representing the performance of the FG in the particular ES.  This is calculated by taking a weighted sum of the normalised PM scores achieved by the FG across all the PMs.  As not all PMs are considered of equal importance, a weighting is introduced to differentiate between various PMs.  Thus each normalised score āi,j(PMk) will have a weight associated with it Ai,j,k, where subscript i refers to evaluation scenario ESi, the j subscript refers to the functional group FGj and the k subscript refers to the performance measure PMk.  Thus the functional group score άi,j is given by

άi,j = (Ai,j,1 x āi,j(PM1)) + (Ai,j,2 x āi,j(PM2)) + - - - + (Ai,j,n x āi,j(PMn)) + - - -

An overall scenario score (SSi) can also be determined for each ESi representing the performance of all the FGs in the particular ES.  This is calculated by introducing a FG weight Bi,j and taking a weighted sum of the FGj scores achieved in the ESi. Finally, an overall vessel performance can be determined for the design iteration X representing its performance across all the ESs.  This is calculated by introducing an ES weight Ci and taking a weighted sum of the ES scores. 

The HPM with scenario and design score along with the all the associated individual scores and weights are presented in the table below.  The overall Vessel Performance VP, for design X can then be compared against the VP score for all other designs to determine which design produced the best overall performance.  The matrix is also diagnostic in that it allows the identification of which measures contributed to the poor performance of a failed vessel design, or which PM could be improved in a winning design.

Design X

 

Evaluation

Scenario

Functional Groups

 

FG1

FG2

- - -

FGn

- -

Scenario

Score

Scenario

Weight

ES1

B11

ά1,1

B12

ά1,2

-

- -

B1n

ά1,n

-

- -

SS1

C1

ES2

B21

ά2,1

B22

ά2,2

-

- -

B2n

ά2,n

-

- -

SS2

C2

:

:

:

:

:

-

- -

:

:

-

- -

- - -

- - -

ESn

Bn1

άn,1

Bn2

άn,2

-

- -

Bnn

άn,n

-

- -

SSn

Cn

:

:

:

:

:

-

- -

:

:

-

- -

- - -

- - -

Overall Functional Group Scores

SFG1

SFG2

- - -

SFGn

- - -

 

 

Overall design performance

VPDESIGN(X)



 

MODIFICATIONS TO maritimeEXODUS

The maritimeEXODUS software forms the core of the HF evaluation component of the system.  This software was specifically designed to simulate evacuation scenarios and, as part of this project, its capabilities were extended to include the simulation of the NOP scenarios.  This capability built on an existing feature known as the Itinerary List (IL).  Using the IL it is possible to assign crew (and passengers) a list of tasks to perform.  To accommodate the range of new scenarios that needed to be simulated to complete the HPM a range of new tasks were developed including:  

  • ‘Terminate’ command, used in the NOPs scenarios allowing crew to stay at their last location once a task has been completed;
  • ‘Repeat’ command, used to allow crew to repeat predefined set of tasks a number of times as is required in the patrol task;
  • ‘Search Compartment’ command which instructs crew to enter a list of assigned compartments to undertake a search as part of the blanket search scenario;
  • Close Door’ command which instructs a crew member to check that a door has the correct status for the current ship state and if not, change the status of the door;
  • ‘Give’ and ‘Receive’ command allowing a senior member of the ship’s staff to issue tasks to lower ranked members, who ‘Receive’ the task .

 In addition to these capabilities, a range of other modifications and additional software have been developed including:

  • A separate utility program was developed (the Human Performance Metric Analyser) which automatically constructs the HPM matrix of human performance scores from maritimeEXODUS output that are used in the evaluation of the vessel design.
  • The process of building vessel geometries ready for analysis was automated. Previously, the process of preparing a geometry could take as much as two weeks to complete, with the automation the majority of this process can now be completed within half an hour (based on the Type 22 Batch III Frigate). This is achieved through linking maritimeEXODUS to Paramarine-Surfcon through appropriate filter software.
  • A new scripting language (SSF) was developed which enables third parties to easily set up a population and their itineraries within maritimeEXODUS without the need to navigate through a complex user interface, reducing a considerable amount of tedious time consuming effort.
  • Additional output files were required from maritimeEXODUS in order for it to interact with the ship configuration software SURFCON. These include images of contour maps displayed within maritimeEXODUS showing the locations of severe congestion and footfall maps. Animated output files were also implemented in order to illustrate individuals moving around the design.
DEMONSTRATION APPLICATION OF HPM

The following demonstration of the HPM concept involves evaluating the relative performance of two designs of a hypothetical naval vessel.

Vessel Geometries
The baseline vessel design (Variant 1) consists of 453 compartments spread over eight decks. Decks No1 and No 2 (deck 4 and 5 respectively) have a single central passageway connecting the aft to forward section of the deck.  The second variant design (Variant 2) consists of the 445 compartments spread over eight decks as in variant 1. The key difference between the two designs is that Variant 2 has two passageways running in parallel from the aft to the forward end of the vessel on both decks.

The Scenarios
Each vessel has a complement of 262.  The crew are initially located in the location they would be expected to be at the start of each scenario as determined by the “state” of the vessel.  Crew members not on watch are located in their cabin. In this example each variant is assessed using seven ESs. These are:

Naval evacuation scenarios;

  • ‘normal day cruising A’ ES1,

  • ‘normal day cruising B’ ES2,

  • ‘Action Stations’ES3

NOPs scenarios;

  • ‘State 1 Preps’ ES4,

  • ‘Blanket Search’ ES5,

  • ‘Family Day A’ ES6,

  • ‘Family Day B’ ES7. 

The Analysis 
The seven evaluation scenarios (i.e. ES1 - ES7) were each run 50 times using the maritimeEXODUS software and representative simulation result files were selected for each scenario to construct the HPM for each variant.   An example animation produced by maritimeEXODUS for the Action Stations ES for Variant 1 is presented below.


Click on the image to play the animation

The PMs for each variant were then determined and the final HPM constructed for each variant as shown below. 

Evaluative scenario

Scenario

Weight

Variant 1

Variant 2

% difference between

Variant 1 and Variant 2

Normal Day Cruising A

1

46.14

44.33

3.93%

Normal Day Cruising B

1

50.81

46.79

7.92%

Action Stations Evacuation

1

51.45

46.70

9.23%

State 1 Preps

1.5

67.46

75.47

-11.87%

Blanket Search

1.5

78.04

84.29

-8.01%

Family Day A

1.5

48.65

47.20

2.99%

Family Day B

1.5

56.03

55.32

1.26%

 

Overall Performance of design

523.7

531.2

 

 

Variant 1 produces an overall Vessel Performance (VP) score of 523.7 while Variant 2 produces a VP score of 531.2.  Thus we note that the overall performance of both variants is broadly similar, with Variant 1 producing a marginally better (1.4%) overall human factors performance according to the identified ES, PM and weights.  Furthermore, we note that Variant 2 outperformed Variant 1 in most of the scenarios, however Variant 1 significantly outperformed Variant 2 in two NOPs scenarios.  The worst performing scenario for Variant 1 is ‘Action Stations Evacuation’.  As Variant 1 produces the better overall performance and produces significantly better NOPs performance it is considered the superior design.  However, its performance may be improved by investigating why it did poorly in its worst performing ES.

To better understand why Variant 2 outperformed Variant 1 in the ‘Action Stations Evacuation’ scenario (ES3) and to identify potential areas in which Variant 1 can be further improved it is necessary to delve into the sub-components of the HPM (see below).   Presented below are the PM scores for Variant 1 and 2 for ES3.  We note that Variant 1 performed better than Variant 2 in five of the 18 PMs (G2, G5, M1, M14 and M16).  Of these five PMs, four show at least 10% better performance than the respective Variant 2 PM, with M14 (most times a WT door was operated) and M16 (average number of doors used per person) returning 18% better performance.  However, 12 of the PMs for Variant 1 returned poorer performance. Of these PMs nine returned values which were at least 10% worse than those in Variant 2.  The poorest performance was achieved by G4 (average time spent in congestion) which returned 50% worse performance. 

FG1 – Entire Population

Variant 1

Variant 2

 

Weight

raw

norm

raw

norm

CONGESTION CRITERIA

 

 

 

 

 

C1 – the number of locations in which the population density exceeds 4 p/m2 for more than 10% of the overall scenario time’

8

4

1

4

1

C2 – the maximum time that the population density exceeded the regulatory maximum of 4 p/m2 for 10% of the simulation time

3

75.40

1

42.14

0.56

GENERAL CRITERIA

 

 

 

 

 

G1 – average time required to complete all operations;

4

256.7

1

193.54

0.75

G2 – average time spent in transition      

3

36.61

0.80

45.76

1

G3 – time to reach final state

8

666.7

0.22

594.50

0.20

G4 – Average time spent in congestion

3

150.6

1

74.93

0.50

G5 – average distance travelled

4

47.11

0.94

50.11

1

GEOMETRIC CRITERIA:

 

 

 

 

 

M1 – the number of WTD used during the scenario.

2

24

0.89

27

1

M2 – the number of Hatches used during the scenario.

2

31

1

25

0.81

M3 – the number of ladders used during the scenario.

2

31

1

25

0.81

M5 – the number of doors used during the scenario.

1

78

1

76

0.97

M8 – the number of times the FG moved between decks

2

373

1

322

0.86

M13 – Average number of components used per member of FG during the scenario

2

4.47

1

4.36

0.98

M14 – Most times a WT door was operated

4

9

0.82

11

1

M15 – Most times a hatch was operated

3

10

1

7

0.70

M16 - Average number of doors used per person

3

1.59

0.82

1.94

1

M17 - Average number of WT doors per person

3

1.46

1

1.19

0.82

M18 - Average number of hatches used

3

0.27

1

0.23

0.83

We note that for Variant 2, the overall average time spent in congestion (as measured by G4) was some 50% less than in Variant 1.  This significant reduction in congestion results in Variant 2 being able to complete the scenario 11% quicker than Variant 1 (as measured by G3).    Indeed, we note that while both vessels easily satisfy the international set evacuation time requirements (as measured by G3) the levels of congestion experienced exceed the international set limits in four locations (as measured by C1) and Variant 1 experiences the most severe congestion (as measured by C2).  As the values for C1 and C2 are higher than the regulatory limits, neither vessel would be deemed to be acceptable.  To address this issue and to improve the overall performance of Variant 1, further investigation is required to uncover the causes of the severe congestion.

 

Variant 1: Three Primary Congestion Regions in Actions Stations Evacuation

Exploring the areas of congestion in Variant 1 (using the population density graphical function available in maritimeEXODUS) suggested that a single additional ladder connecting 01 Deck with No 1 Deck between two of the severe congestion regions may alleviate some of the congestion by providing an additional means of vertical movement.

Suggested solution place additional ladder between 01_Deck and No_01_Deck.

Further analysis using maritimeEXODUS revealed that this simple modification eliminated two of the four congestion regions in the ‘Action Stations Evacuation’ scenario.  With this modification in place the HPM was re-evaluated for the Modified Variant 1.  The Modified Variant 1 now outperforms the original Variant 1 in each scenario and produces an overall VP which is 6% more efficient than the original Variant 1 design (see Table 7) and 8% more efficient than the Variant 2 design.  We also find that the Modified Variant 1 design outperforms the Variant 2 design in all but the ‘Normal Day Cruising A’ evacuation scenario. 

Evaluative scenario

Scenario Weight

Variant 1

Modified Variant 1

% difference between Variant 1 and Modified Variant 1

Normal Day Cruising A

1

47.81

46.59

-2.61%

Normal Day Cruising B

1

51.62

44.98

-14.76%

Action Stations Evac

1

52.78

44.68

-18.11%

State 1 Preps

1.5

75.95

73.43

-3.43%

Blanket Search

1.5

86.25

85.45

-0.94%

Family Day A

1.5

52.28

49.55

-5.52%

Family Day B

1.5

57.57

53.57

-7.48%

 

Overall Performance of design

560.29

529.24

 

In summary, the HPM identified that the single passage variant produced marginally superior HF performance than the dual passageway variant.  This demonstrates that the HPM methodology is discriminating.  The HPM was also used to identify areas in which the HF performance of the single passageway variant could be further improved.  Through adding a single ladder to the single passageway variant we improved its overall performance by 6% making the single passageway variant almost 10% better than the dual passageway variant.  This demonstrates that the HPM methodology is diagnostic.  Furthermore, the HPM methodology can routinely be applied to any given design and so is systematic, the logic process by which the decisions are made are transparent and another engineer using the methodology would arrive at the same conclusions and so the methodology is repeatable.

PROJECT PUBLICATIONS AND PRESENTATIONS

The following list of papers were produced by the project partners. Some of the publications and presentations are available in pdf format. Click on the link provided to download the publication where available or to view the powerpoint presentation.

  1. Andrews, D J, “The Use of Simulation in Preliminary Ship Design, Proceedings of 12th International Conference on Computer Applications in Shipbuilding”, Busan, Korea, August 2005, pp 419 - 432.
     
  2. Andrews, D J, Pawling, R and Casarosa, L, “Integrating Ergonomics into Ship Design”, CETENA Human Factors Conference, Genoa, October 2005, Proceedings on CD; copy held by UCL.
     
  3. Andrews, D J, Pawling, R, Casarosa, L, Galea, E R, Deere, S, Lawrence, P, Gwynne, S and Boxall, P, “Integrating Ship Design and Personnel Simulation”, INEC, IMarEST WMTC, London, 6 - 10 March 2006, Proceedings on CD, ISBN 1-902536-54-1 

    A pdf version of the paper is available for download here.
     
  4. Andrews, D J, Casarosa, L and Pawling, R, "Integrating the Simulation of Operations Into Preliminary Ship Design", NAV 2006; International Conference on Ship and Shipping Research, Genoa, 21 - 23 June 2006, Vol. 1, pp 4.3.1-4.3.12.
     
  5. Andrews, D J, “Simulation and the Design Building Block approach in the design of ships and other complex systems”, 2006 Proc. R. Soc. Vol. 462, pp 3407-3433.
     
  6. Andrews, D J, Pawling, R, Casarosa, L, Galea, E R,   Deere, S and Lawrence, P, “Integrating Personnel Movement Simulation into Preliminary Ship Design”, RINA Int. Conf. on Human Factors in Ship Design, London, 21-22 March 2007, pp 117-128.

    A pdf version of the paper is available for download here.
     
  7. Andrews, D J, Pawling, R, Casarosa, L, Galea, E R, Deere, S and Lawrence, P, “Integrating Personnel Movement Simulation into Preliminary Ship Design”, International Journal of Maritime Engineering, Volume 150 Part A1 pp 19-34, ISSN 1479-8751, 2008. http://www.rina.org.uk/ijme0801.html

    A pdf version of the paper is available for download here.
     
  8. Galea, E R, “The next step in the rise of maritime human factors simulation models : Optimising vessel layout using human factors simulation”, Key Note Address, The Rise of Maritime Simulation, Ocean Innovation 2007, Halifax, Canada, 21-24 Oct 2007. (http://www.oceaninnovation.ca/OI-2007_guide_low%20res.pdf)

    A pdf version of the presentation may be viewed by clicking here.
     
  9. Deere, S J, Galea, E R and Lawrence, P, “A Systematic Methodology to Assess the Impact of Human Factors in Ship Design”, Applied Mathematical Modelling, Applied Mathematical Modelling, 33, 867-883, 2009. <http://dx.doi.org/10.1016/j.apm.2007.12.014> 

    A pdf version of the paper is available for download here
     
  10. Deere, S J, Galea, E R and Lawrence, P, “Optimising Vessel Layout using human factors simulation”, Pedestrian and Evacuation Dynamics 2008, 27-29 Feb 2008 Wuppertal Germany, To appear in conference proceedings 2009.

    A pdf version of the presentation may be viewed by clicking here
     
  11. Deere, S J, Galea, E R and Lawrence, P, “Assessing Naval Ship Design for Human Factors Issues Associated with Evacuation and Normal Operations”, COMPIT08, Ed: V.Bertram and P Rigo, 21-23 April 2008 Liege Belgium, ISBN-10 2-9600785-0-0, pp 33-47, 2008.

    A pdf version of the entire proceedings can be downloaded from here: http://www.anast.ulg.ac.be/COMPIT08/
     
  12. Andrews, D J, Pawling, R and Casarosa, L, "Interactive Computer Graphics and Simulation in Preliminary Ship Design", COMPIT08, Ed: V.Bertram and P Rigo, 21-23 April 2008 Liege Belgium, ISBN-10 2-9600785-0-0, pp 407-421, 2008.

    A pdf version of the entire proceedings can be downloaded from here: http://www.anast.ulg.ac.be/COMPIT08/
     

 

FURTHER INFORMATION

Prof. Ed Galea
Fire Safety Engineering Group
University of Greenwich
Greenwich Maritime Campus
Old Royal Naval College
Queen Mary Building
Greenwich SE10 9LS
UK

Tel: +44 (020) 8331 8730
Fax: +44(020) 8331 8925
 

e-mail: E.R.Galea@gre.ac.uk