Simulation of fires in a RoPax vessel

Heimo Tuovinen, Tommy Hertzberg

Fire Technology SP Report 2009:02

Simulation of fires in a RoPax vessel

Abstract

Simulation of fires in a RoPax vessel

Simulations of a public space on a RoPax ferry have been conducted as part of an effort to

demonstrate possible means to use SOLAS regulation 17 for approval of new light weight materials in ship construction.

Upper and lower decks and outer walls consisted of insulated combustible polymer materials. Four different ventilation scenarios were simulated. The laminate temperatures in the construction were simulated in order to assess the fire risks.

The simulations were validated by simulating a previous large scale cabin fire experiment. Key words: Fire simulation, composites, Regulation 17, FSE, RoPax vessel

SP Sveriges Tekniska Forskningsinstitut SP Technical Research Institute of Sweden SP Report 2009:02

ISBN 978-91-85829-85-9 ISSN 0284-5172

Table of contents

Abstract 3

Table of contents

4

Preface 5

Sammanfattning 6

1

Introduction 7

1.1 Simulation and Risk analysis 7

2

The simulated space

8

2.1 Instrument locations 10

2.2 Fire scenarios 11

3

Simulation set-up

12

3.1 Material data 12

3.2 The design fire 12

3.3 Calculation of fire load 13

4

Simulations 14

4.1 Scenario 1 Closed room, inert walls 14

4.2 Scenario 2 – Closed window; door open 15

4.2.1 Mesh refinement 15

4.2.2 Laminate temperatures 16

4.2.3 Gas temperatures 18

4.3 Scenario 3 – Window and door open 18

4.3.1 Laminate and gas temperatures 19

4.4 Scenario 4 – Stairway and door open 21

4.4.1 Laminate and gas temperatures 21

5

Simulation comparison to a large scale cabin fire test

23

5.1.1 Results 25

6

Conclusions and discussion

29

6.1 Estimation of fuel input levels, HRR 30

6.2 Gas temperatures 30

6.3 Laminate temperatures 31

Preface

This work was financially supported by VINNOVA, the Swedish Governmental Agency for Innovation Systems. Further was much information and support obtained from the Swedish LASS-project and the subgroup within the EU-LASS-project SAFEDOR (Dag Mcgeorge from DNV, Bjørn Højning from Fireco AS and Henrik Nordhammar from STENA ) that have been working on the design of a RoPax with composite superstructure. Their support and assistance are gratefully acknowledged.

Sammanfattning

Denna rapport är en del av forskningsprojektet DIBS, ”Dimensionerande bränder till sjöss”, vilket stöds ekonomiskt av VINNOVA. Projektet hade bl a som mål att ta fram dimensionerande bränder för användande av SOLAS nya regel 17 rörande alternativ fartygsdesign, vilken möjliggör att frångå det tidigare kravet på stål som konstruktionsmaterial. Dock kräver den nya regeln att en

säkerhetsbedömning görs och där är just den dimensionerande branden ett nyckelbegrepp. Projektet har samarbetat med ett annat större VINNOVA finansierat projekt, LÄSS,

”Lättviktskonstruktioner till sjöss” (www.lass.nu) samt EU-projektet SAFEDOR (www.safedor.org) där dessa i sin tur samarbetat kring hanteringen av regel 17 för en passagerarfärja med överbyggnad i plastkomposit. Central del i detta har varit en riskanalys där speciellt bedömning av brandfaran varit viktig. För detta har bl a krävts brandsimuleringar för att kunna uppskatta energiproduktion och gas/material temperaturer för konstruktionen. De simuleringar som beskrivs i rapporten har använts för detta.

Verifierande simulering har också gjorts där resultat från ett tidigare genomfört storskaligt experiment på SP Brandteknik jämförts mot simulerade data. Simulering av brand är alltid vanskligt med stora osäkerheter på grund av att bränder omfattar kraftigt icke-linjära fenomen varför små variationer i ingångsdata kan ge kraftiga variationer av resultat. Trots detta visade den verifierande simuleringen på god överrensstämmelse, vilket bl a beror just på tillgång till väl definierade ingångsdata. Verifieringen skänker i sin tur viss trygghet åt tolkningen av övriga simuleringsresultat.

1

Introduction

Transport is a sector in which lightweight products and lightweight designs are becoming increasingly important. This applies also to shipbuilding where composites or aluminium materials are being used to reduce overall weight or superstructure weight, and to build high-speed crafts (HSC). The lighter a vessel, the more it can carry, or the less energy it needs for propulsion. However, developments and use of lightweight materials are impeded not only by technical difficulties, but also by international regulations.

HSC vessels are restricted by a displacement/speed requirement but also by the requirement to stay close to land and hence close to rescue services. Standard vessels or vessels travelling between continents have for a long time had limited access to new materials due to the IMOi _{requirement for} constructions made of “steel or equivalent material”. However, in July 2002 a new regulation

appeared in the document regulating fire safety at sea, SOLAS (Safety Of Life At Sea), permitting use of any construction material provided that the same level of fire safety as for conventional steel constructions can be demonstrated.

In the reported project, a RoPax ferry owned by Stena Line was re-designed using a superstructure in FRP (Fibre Reinforced Plastic) composite as shown in Figure 1 and Figure 2. The work was carried out in co-operation between the Swedish research project LASS (www.lass.nu) and the EU-project SAFEDOR (www.safedor.org). Fire safety of the new design was demonstrated through the use of Risk analysis and many explicit fire tests and certifications of fire safe constructions but also through fire simulations, some of which are described in this report.

1.1

Simulation and Risk analysis

The risk analysis made showed that a major threat was a fire in one of the public spaces such as a café or a restaurant. Simulations were therefore performed in order to estimate the Heat Release Rate (HRR) development of such a fire in different situations. Important information for the risk analysis was gas temperatures and temperatures in the composite structure. The temperatures in the structure are especially important in this context as the FRP softens at a certain temperature and thus loses its load bearing capacity. The gas temperatures obtained in the simulations were also compared to the standard heating curve used in fire testing and certification of e.g. composite constructions, in accordance to the IMO test MSC.45(65)1_.

Figure 1. Possible RoPax ferry construction with composite superstructure

Fibre

High-strength composite material

Core material

Polymer

Figure 2 Schematic picture of FRP composite.

Four different scenarios were chosen to be simulated representing different ventilation situations (closed/open doors and windows). To validate the calculations a simulation of a real cabin fire was conducted where experimental data were obtained from a large scale fire experiment performed at SP in December 20072_.

The simulation results were provided to the SAFEDOR team for incorporation into the Risk analysis and for the design considerations.

2

The simulated space

A public space including café, restaurant, etc. with a total floor area of 1350 m2 was chosen for the simulations. The surrounding decks and bulkheads were assumed to consist of combustible light-weight composite materials instead of steel.

A CAD drawing is shown in Figure 3. The geometry was simplified for the computer model shown in Figure 4. The middle part represents the staff service location, which was assumed closed during the fire, i.e. the locations were not involved in the fire. The fire source was located in the back of the room (a red rectangle in the figure). A 4.8 m wide door was located in the front part of the room. In one scenario a broken window, represented by a 1.8 m x 1.8 m “hole”, was opened in the centreline of the back wall. The green dots denote the locations of simulated “instrument” locations. The temperature graphs presented in this report refer to these instrument locations with their x-y coordinates (exact positions described in the next section). The positive x-coordinate goes from the front part of the room and backwards and the y coordinate goes from right to left when “looking” towards the windows. For all simulations, a standard fire growth curve (HRR) described as a “fast fire” curve (Eq (1)) was used. The growth rate of this curve was very similar to what was measured in a the large scale cabin fire2 _{(simulated in scenario 1) and it was judged to be “severe enough” for the purpose of estimating} fire risks in the RoPax public space.

Figure 3 A CAD drawing of the public spaces used in the investigation.

2.1

Instrument locations

A schematic picture of the calculation domain is shown in Figure 5. The dots in the figure are the instrument locations. In the figure the x-axes in oriented from left to right and y–axes upwards in the plane. The z-axes is perpendicular to the plane of the page (positive direction towards the reader). These instruments location were used in scenarios 3 and 4. In scenario 5 the instrument locations were somewhat different, but are given in x-y coordinates in the time-temperature diagrams.

The following x-y coordinates were used for registration of gas temperatures: • Lowest x-row, from left: (10,7); (23,7) and (39, 7)

• Middle x-row, from left: (3,19) and (39,19)

• Upper x-row, from left: (10,32); (23,32) and (39,32)

Gas temperatures were registered at various heights (z coordinates): 0.8 m, 1.3 m, 1.8 m, 2.4 m and 2.79 m. The highest point was 5 cm below the upper deck insulation.

The following x-y coordinates were used for registration of upper deck laminate temperature: • Lowest x-row, from left: (10,7); (23,7); (34,7) and (39,7)

• Middle x-row, from left: (3,19); (34,19) and (39,19)

• Upper x-row, from left: (10,32); (23,32),(34,32) and (39,32)

The following x-y coordinates were used for registration of lower deck laminate temperature: • Lowest x-row, from left: (10,7); (23,7); (34,7) and (39,7)

• Middle x-row, from left: (3,19); (31,19) and (44,19)

Figure 5 “Instrument” locations in scenario 3 and 4. The fuel was injected uniformly through the fire source (red rectangle). The yellow rectangle represents the staff service location (kitchen, dish room, galley, stores, etc.), which was assumed only as blockage, i.e. the contribution to fire load was assumed negligible.

2.2

Fire scenarios

Four different scenarios were chosen for the simulations representing different ventilation situations 1) RoPax public space simulation. Closed room simulations in order to test level of HRR and fire

duration.

2) RoPax public space simulation. Open door (4.8 x 2.6 m2_{) and closed window.}

3) RoPax public space simulation. Open door (4.8 x 2.6 m2_{) and open (broken) windows (2} windows, 0.9 ‘ 1.8 m2 _{= 1.8 x 1.8 m}2 _{total area).}

3

Simulation set-up

In all simulations the calculation domain were divided in about 200 000 control volumes, which gives a cell size of 30 cm. Scenario 2 was also simulated with a finer mesh in order to check cell size dependence on the calculation results. The finer mesh was divided to about 700 000 control volumes, thus a cell size of about 20 cm throughout the whole calculation domain.

The calculations were performed on a Sun X4600 X2 Shared Memory Multi-core Windows Server with 8 CPU using a parallelized version of FDS (version 5.1.6). The calculation domain was divided into 8 meshes in order to be able to make use of all 8 processors efficiently.

3.1

Material data

Composite and insulation materials were characterised at SP, using the TPS-method3 _{and a summary} of the results obtained is given in Table 1. The data was used as input to the simulations.

Table 1 Material parameters of the components in composite walls

Material Density [kg/m3] Temp [°C] Conductivity [W/(mK)] Temp [°C] Specific heat [kJ/(kg K) 0 0.040 0 0.74 80 0.050 20 0.75 150 0.051 80 0.96 180 1.11 Laminate 1870 1200 0.052 1200 1.30 0 0.028 0 1.25 20 0.029 80 2.12 80 0.030 150 2.38 150 0.045 Core H80 80 1200 0.050 1200 2.38 10 0.033 50 0.035 100 0.045 150 0.053 200 0.063 300 0.088 400 0.119 500 0.198 Insulation material 100 1000 0.250 All Temp 0.80 20 60 800 27 Steel 7850 1500 27 All Temp 0.5

3.2

The design fire

The most critical part of a fire simulation is to define the design fire, i.e. the time-HRR curve that describes the fire evolution. There are flame spread models available but for most cases, these are highly unreliable and rely heavily on correct input data, data which is difficult to find in most cases. It was therefore deemed better to simply decide on a standard type of fire that is “severe enough” to provide conservative data when the purpose is to estimate fire consequences. It was decided to use a standard fire growth curve frequently used in fire safety engineering for describing the HRR (heat release rate, kW) of a “fast fire growth”:

2

t

Q

=

α

where α is a fire growth parameter (for fast fire growth α = 0.0469 kW/s2) and t is the time in seconds.

Depending on the α-value, a slow, medium, fast or ultra fast fire growth curve is obtained4. The fire

growth parameter depends on the actual fire situations involving the type of burnable commodities and geometric arrangement of the fuel. The fast curve used in the simulations is close to the fire growth rate obtained in the large scale cabin fire test run at SP2_{. In the experiment, four easily ignitable} mattresses together with a high degree of ventilation and radiation from a hot gas layer provided “good” conditions for a fast fire growth.

Since the simulations were focused on finding the HRR at flash-over and the temperatures during a prolonged over fire, the choice of fire curve was only important for defining the time to flash-over. Once flash-over is obtained, the HRR is dependent on the amount of oxygen available, i.e. on the ventilation. This flash-over HRR determines the maximum severity of a fire of long duration and the fire growth curve chosen merely defines the relatively short time until this level is reached. How long the fire will last depends on the amount of combustible material in the room.

The investigation made used several simulations to stepwise approach a realistic fire situation as the level of maximum HRR was unknown prior to the simulations. The investigation therefore started by running a number of a t2

– fires in a simpler (and much quicker) two-zone model simulation tool

Branzfire5 _{and the HRR levels obtained were then confirmed in the CFD (Computational Fluid} Dynamics) based simulation tool FDS6 _{where also the main simulation work was done in order to} estimate fire behaviour and construction temperatures.

3.3

Calculation of fire load

The fire load in the room was assumed to be approximately 115 GJ based on the following estimates:: • 500 upholstered seats, at 10 MJ/seat according to measurements made at SP fire laboratory

(Figure 6). Total fire load 5 GJ

Figure 6. Fire test of chair from the Stena SAGA RoPax vessel

• Number of wood tables 80, size 1.5 m2. Assumed table sheet thickness of one inch (2.54 cm) and total area of table sheet 120 m2, which gives approximately 3 m3 wood material. Total fire load of 47 GJ.

• Panel on wall and ceiling surfaces, total area 250+1350 m2 = 1600 m2 at 3.35 MJ/m2. Total fire load 5.4 GJ

• PVC flooring material and total floor area of 1350 m2. Fire load 46.2 MJ/m2. Total fire load 62.4 GJ.

All materials user were accepted in accordance to regulations. Material data (except for literature data of the wood material) were obtained by direct measurements in relation to the large scale cabin fire tests performed at SP in 20072_.

4

Simulations

Four different simulation were conducted representing 4 different ventilation situations in the public space.

4.1

Scenario 1 Closed room, inert walls

The aim of this scenario was to calculate the maximum duration and HRR of a fire if all the openings in the public space were closed. Decks and bulkheads were assumed to consist of inactive material in this scenario, i.e. the heat transfer through wall materials was neglected as this was of no importance for the simulation due to the expected short time of fire before self-extinguishment due to lack of oxygen.

The simulations in this scenario resulted in a maximum HRR of approximately 20 MW before all oxygen in the room was consumed. This occurred approximately 8 minutes after the ignition. Both FDS and Branzfire gave similar results as shown in Figure 7.

0 5000 10000 15000 20000 25000 0 100 200 300 400 500 600 700 800 Time [s] HRR [ k W] kW FDS Branzfire

Figure 7 Calculated heat release rate in the closed room scenario from simulation tools FDS and

Branzfire. Both programs calculate HRR according to fast fire growth curve. The lack of oxygen causes extinction ~ 8 minutes after ignition.

4.2

Scenario 2 – Closed window; door open

In this simulation the deck and bulkheads were assumed to be real materials and the heat transfer through them were taken into account. Both top and lower decks were considered as 5 cm thick composite panels with 1 mm laminate on each side. The lower face of the top deck was covered by 100 mm thick insulation and the upper face of the lower deck was covered by a 30 mm thick

insulation sheet. Side bulkheads were assumed to consist of exactly the same composite material with 100 mm insulation on the inner faces. Vertical B-class panels were placed 20 cm inside the composite bulkheads. The B-class panels consisted of a 5 cm insulation board between two 0.7 mm steel sheets. Material parameters for the components in the composite walls are presented in Table 1.

A steady state HRR value of approximately 25 MW was calculated using the two-zone model Branzfire. This HRR value was used to define a maximum input HRR to the CFD-simulation. The reason for doing this was that the CFD-code calculated the produced energy based on an inflow of combustible gas. If, for instance, a 50 MW maximum value would have been used instead, then the HRR curve used (equation 1) would have “forced” the simulation tool to include the ~25 MW of surplus gas in the calculation domain, which would have an important impact on the result due to cooling and lowering of the oxygen partial pressure.

The initial fire was simulated using the fast αt2_{-curve for approximately 100s before the full HRR level} was reached. The mean value of the HRR reached in the simulation was about 23 MW. The Heat Release curve is shown in Figure 8.

0 5000 10000 15000 20000 25000 30000 35000 40000 45000 0 30 60 90 120 150 180 Time [min] HRR [kW]

Figure 8 HRR in the scenario with closed window and 4.8 m wide opening.

It should be noted that although the simulation was run for a 3-hour period, the amount of available combustible materials in the fire scenario, estimated in section 3.3, would only permit a fire duration of this intensity for 1 hour and 25 minutes.

4.2.1

Mesh refinement

Scenario 2 was also run using a finer computational mesh to test the mesh independence. The coarse mesh consisted of cubic control volumes of side length about 33 cm (for optimal results in calculation FDS needs cubic control volumes). For the finer mesh, control volumes of side length about 20 cm were used. The calculation results for both meshes were similar and the coarser grid was therefore

judged to be “good enough” and hence, was used for all other simulations. The simulated temperature fields in three slices through the calculation domain are shown in figure 9. Due to strong under-ventilated conditions the flaming combustion occurred on the fuel source (red rectangle in the figure) only in the beginning of the simulation. After one hour simulation the flaming combustion was moved near the regions were the oxygen contents for combustion was sufficient, i.e. nearer the opening. Because the opening was located to the right of the symmetry line on the room when entering from the left the burning occurred mostly in the right part of the room and at the entrance, i.e. in the lower left corner in figure 9. A slight difference in temperatures is noticed. Especially, in the left part of the room i.e. the upper part in Figure 9, the simulation using fine mesh generates slightly higher temperatures than the simulation with the coarse mesh. The finer mesh resolves the smaller local eddies better which in turn influences the mixing of fuel and air and hence influences the combustion and the local gas temperature. This does not have any impact on the total heat release rate in the room in this case as the HRR is prescribed and the oxygen available is sufficient. The combustion zones are, however, possible widened with the coarse mesh. The coarse mesh might also provide a more conservative estimate in this case as the combustion is concentrated to the opening area and thus higher temperatures is encountered there in this case.

Figure 9 Temperature slices through the calculation domain after one hours simulation using fine

mesh (left) and coarse mesh (right), respectively.

4.2.2

Laminate temperatures

Laminate surface temperatures were calculated at selected locations in both upper and lower decks. Both decks were covered by insulating boards. The thicknesses on upper and lower decks were 100 mm and 30 mm, respectively. The simulation tool FDS calculated the laminate temperatures, underneath the insulation material simply by solving the one-directional heat transfer equation. In Figure 10 and Figure 11 the laminate surface temperatures in upper and lower decks and at various points are presented as a function of time.

Generally the laminate temperatures at the lower deck are higher than on the upper deck. This is due to the fact that the insulating layer in the lower deck was only 30 % of the upper deck insulation and that the hot smoke layer after a short while reached the floor level in the simulation.

Laminate temperatures upper deck 0 50 100 150 200 250 300 350 400 450 500 0 30 60 90 120 150 180 Time [min] Temp erature [C]

C Laminate temperarure x-y 10-32 C Laminate temperature x-y 23-7 C Laminate temperature x-y 10-7 C Laminate temperature x-y 23-32 C Laminate temperature x-y 3-19 C Laminate temperature x-y 34-19 C Laminate temperature x-y 34-32 C Laminate temperature x-y 34-7 C Laminate temperature x-y 39-19 C Laminate temperature x-y 39-32 C Laminate temperature x-y 39-7

Figure 10 Laminate temperature in upper deck in selected locations as a function of time. Scenario

with closed window.

Laminate temperatures lower deck

0 100 200 300 400 500 600 700 800 900 1000 0 30 60 90 120 150 180 Time [min] Te mp erat ure [ C

] C Lamtemp floor x-y 10-32C Lamtemp floor x-y 10-7 C Lamtemp floor x-y 23-32 C Lamtemp floor x-y 23-7 C Lamtemp floor x-y 3-19 C Lamtemp floor x-y 31-19 C Lamtemp floor x-y 34-32 C Lamtemp floor x-y 34-7 C Lamtemp floor x-y 39-32 C Lamtemp floor x-y 39-7 C Lamtemp floor x-y 44-19

Figure 11. Laminate temperature in lower deck in selected locations as a function of time. Scenario with

4.2.3

Gas temperatures

Since the fire quite quickly becomes strongly under-ventilated, the location of the flaming combustion moves towards the opening where the oxygen is in the simulation. This phenomenon can readily be observed in a real fire situation. The fuel source in this scenario was located at an x-coordinate between 35 m and 40 m and the y-coordinate between 15 m and 23 m. During the first 20 minutes of fire the oxygen concentration in the room is sufficiently high for flaming combustion to take place in the vicinity of the fuel source. After that the flames moved towards the oxygen richer locations. Because the opening is located to the right hand side of the centre line (seen in the positive

x-direction) the flaming combustion occur mostly in this part of the room. This also means that we will find the highest temperatures here. Figure 12 shows the temperature at 23 m from the opening and 7 m from the right hand side bulkhead. The gas temperature at that location is generally 800-1000 °C. Also near the door the temperatures are close to these values, see Figure 12. The standard ISO 834 fire curve used when testing deck and bulkhead for 60 minutes fire resistance is included for comparison in the figure.

In the left hand side of the room, the gas temperatures are generally below 600 °C. Also, in the back part of the room, the steady state gas temperatures are quite low, except for a peak value of ~900 °C at about 15 minutes, due to the initial flaming combustion above the fuel source.

Temperatures in more locations are shown in Appendix A. Temperatures at x23,y7 0 200 400 600 800 1000 1200 0 30 60 90 120 150 180 Time [min] Te mperat ure [C] C Temp23-7-1 C Temp23-7-2 C Temp23-7-3 C Temp23-7-4 C Temp23-7-5 ISO 834

Figure 12. Gas temperatures at x = 23 m, y = 7 m at heights 0.8, 1.3, 1.8 and 2.4m in scenario with

closed window. Standard ISO 834 fire curve is shown as a comparison.

4.3

Scenario 3 – Window and door open

In this scenario a 4.8 m wide opening and an additional opening of 1.8 m x 1.8 m, corresponding to two 1.8 m x 0.9 m broken windows, were introduced. Except for the broken window the calculation domain was exactly the same as in scenario 2. Prior to the simulation, the steady state HRR was calculated by Branzfire5 _{to be approximately 28.5 MW, which was the used as input value to the CFD}

simulation. The HRR was started smoothly, by increasing as t2_{-fire during the first 1000s up to} 28.5 MW. After time 1000s the HRR was held constant. The location and the size of fuel source was the same as in scenario 2. The mean value of the simulated HRR became just below 25 MW. The HRR history is shown in Figure 13.

0 5000 10000 15000 20000 25000 30000 35000 40000 0 15 30 45 60 75 90 105 120 Time [min] H RR [kW ]

Figure 13 HRR in the scenario with an open window and a 4.8 m wide opening to other spaces.

The laminate temperatures were monitored at several points in both the upper and lower decks. On the lower deck the laminate temperatures reaches generally higher values than on the upper deck as in scenario 2. The results are shown in Figure 14 and Figure 15.

The open window changes the fire behaviour so that the flaming combustion is moved more towards the back of the room, i.e. nearer the location of the fuel source than in scenario 2. However, it is still burning mostly on the right hand side, when looking in the positive x-direction, i.e. towards the windows at the back of the room. The maximum (and minimum) temperatures are at about the same level as in the case with closed window, but the locations of the hottest and coldest regions are somewhat different. In this case the hottest region is at x = 39 m and y = 7, see Figure 16. Also the laminate temperatures are of the same order as in the scenario with closed window.

Since the simulated HRR is somewhat higher in this scenario than in scenario 2, the estimated amount of available combustible materials would allow a ~10 minute shorter fire duration, i.e. a fire of this intensity could only last for about 1 hour and 15 minutes.

4.3.1

Laminate and gas temperatures

Laminate temperatures are shown in Figure 1 and Figure 15. As expected, the laminate temperatures on the lower deck are considerably higher than on the upper deck due to the thinner insulation layer on the lower deck. The maximum temperatures on the lower deck are generally more than 300 degrees above those calculated on the upper deck.

The gas temperatures in the hottest location are shown in Figure 16. The maximum temperature in the upper part of the gas layer reaches 1000 °C in about 1 hour after the ignition.

Laminate temperatures upper deck 0 50 100 150 200 250 300 350 0 15 30 45 60 75 90 105 120 Time [min] Temper atur e [C]

C Laminate temperarure x-y 10-32 C Laminate temperature x-y 23-7 C Laminate temperature x-y 10-7 C Laminate temperature x-y 23-32 C Laminate temperature x-y 3-19 C Laminate temperature x-y 34-19 C Laminate temperature x-y 34-32 C Laminate temperature x-y 34-7 C Laminate temperature x-y 39-19 C Laminate temperature x-y 39-32 C Laminate temperature x-y 39-7

Figure 14 Laminate temperature at selected location on the upper deck in the scenario with open

(broken) window.

Laminate temperatures lower deck

0 100 200 300 400 500 600 700 800 900 0 15 30 45 60 75 90 105 120 Time [min Temperature [C]

C Lamtemp floor x-y 10-32 C Lamtemp floor x-y 10-7 C Lamtemp floor x-y 23-32 C Lamtemp floor x-y 23-7 C Lamtemp floor x-y 3-19 C Lamtemp floor x-y 31-19 C Lamtemp floor x-y 34-32 C Lamtemp floor x-y 34-7 C Lamtemp floor x-y 39-32 C Lamtemp floor x-y 39-7 C Lamtemp floor x-y 44-19

Figure 15 Laminate temperature at selected location on the lower deck in the scenario with open

Gas temperatures at x39,y7 0 200 400 600 800 1000 1200 0 15 30 45 60 75 90 105 120 Time [min] Temperature [C] C Temp39-7-1 C Temp39-7-2 C Temp39-7-3 C Temp39-7-4 ISO 834

Figure 16 Calculated gas temperatures at x = 39 m and y 7 m at heights 0.8, 1.3, 1.8 and 2.4m in the

scenario with an open window. Standard ISO 834 fire curve is shown for comparison.

Gas temperatures in other locations are shown in Appendix A.

4.4

Scenario 4 – Stairway and door open

The HRR data used as input to the simulation was assumed to follow the t2_{-history up to 55 MW}ii _in the first 1000 s, and the rest of the simulation time the HRR was assumed constant at that level. The fire source in this scenario was assumed to be a 90 m2 _{large area in the front part of the room, about 30} m from the openings. Due to the large openings the calculated HRR reaches the high value of 50 MW after 30 minutes of fire as shown in Figure 17.

4.4.1

Laminate and gas temperatures

Temperatures at laminate-insulation interface are shown in Figure 18. In most locations the laminate temperature reaches about 100 °C after one hour of fire, except in one location (yellow line in the figure) where the laminate temperature reached 137 °C. The reason for this is that the flaming combustion occurred in this location later in the simulation.

0 10000 20000 30000 40000 50000 60000 70000 80000 0 10 20 30 40 50 60 Time [s] HR R [kW]

Figure 17 HRR in the scenario with large openings.

This very intense fire effect level actually only permits a 43 minute fire before all combustible materials in the enclosure, as estimated in section 3.3, would be consumed.

Laminate temperature at selected locations

0 20 40 60 80 100 120 140 160 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Time [min] Temperature [C] x-y 10-30 x-y 23-8 x-y 10-8 x-y 23-30 x-y 39-10 x-y 39-19 x-y 39-28

Figure 18 Temperatures on the laminate-insulation interface at selected locations in the top deck.

The steady-state gas temperatures calculated in this scenario are considerably lower than in scenarios 2 and 3, despite the double HRR. The reason for this is that the flow rate of cold fresh air into the room was considerably larger and more unburned fuel was ventilated out through the larger openings.

Figure 19 shows the gas temperatures in the right hand side of the room, approximately 25 m from the openings. As in scenarios 3 and 4 the temperatures are lower in the left and back part of the room.

Gas temperature at x=26 m, y=4.8 m

0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 Time [min] Tem p eratu re [C] h=0.5 m h=1.0 m h=1.5 m h=1.8 m h= 2.3 m

Figure 19 Gas temperatures in several heights as a function of time. This location is 4.8 m from the

right bulkhead, 25 m from the door openings and 10 m from the fire source, which is placed in the front part (opposite to the doors) of the room.

More gas temperature graphs are shown in Appendix A.

5

Simulation comparison to a large scale cabin fire test

In order to validate the simulations a comparison was made between a simulated fire in a cabin and a real scale fire test. The cabin was modelled using the same enclosure geometry as for the cabin in a real fire test. Material characteristics of deck and bulkheads were also modelled based on the real case. The top deck was considered to consist of 5 cm thick composite panel with 1 mm laminate on both sides. The lower face was insulated with 10 cm thick FireMaster insulation board. A suspended ceiling, which consisted of 5 cm FireMaster insulation covered by 0.7 mm steel sheets on both sides, was placed 40 cm below the top deck insulation.

The lower deck consisted of several layers of different materials. The bottommost part 10 cm concrete, above it a 5 cm composite panel covered by 1 mm laminate on both sides (as in the top deck). On the composite panel, 2 cm insulation sheet covered by 2 mm aluminium plate on the top face.

The neighbouring cabin was assumed to be inactive in the simulations, i.e. it was unaffected by heat transfer.

The cabin used in the fire test is shown in Figure 20. Figure 21 shows a computer model of the test cabin. Cabin dimensions were 4.3 m x 3.0 m x 2.7 m (length x width x height). A more detailed description concerning all materials in the test cabin is given in SP Report 2008:332.

All materials in the figure were treated as non-burning items in the calculation model. Instead, all fuel was ‘lumped’ together and injected to the cabin in the simulation through an assumed 2 m2 _{burner in} the floor. The reason for this approach was that the measured HRR curve from the experiment was used as input to the simulations as we wanted to validate the Laminate and gas temperature calculations. The HRR curve is shown in Figure 22.

Figure 20. A view inside the four-bed test cabin. Two additional beds were located at the opposite side

of the cabin.

p 0 200 400 600 800 1000 1200 1400 1600 1800 0 15 30 45 60 75 90 Time [min] HRR [kW/m2]

Figure 22 The experimentally measured HRR used as input in the scenario 1 simulations.

5.1.1

Results

The simulation was run for 90 minutes. The suspended ceiling was assumed to collapse 8 minutes after ignition. Gas temperatures were monitored at four locations in the gas layer in the cabin marked with 1-4 in Figure 5. Points 5 and 6 in the figure refer to aluminium plate temperatures on the floor. The laminate temperatures and surface temperatures of the lower face of the suspended ceiling were monitored at locations 1-3.

The calculated and measured gas temperatures 5 cm below the insulation at location 2 are shown in Figure 24. The suspended ceiling was modelled to fail 8 min after the fire started in the simulation. In reality, the collapse happened after approximately 7 minutes and the graph is therefore shifted one minute backwards in time to make the comparison with measured values easier.

1 1 2 3 4 5 6

Figure 23 Locations of temperature monitoring in the cabin.

Gas temperature 5 cm below insulation

0 100 200 300 400 500 600 700 800 900 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 Time [min] Tem p erat ure [C] Measured Calculated

Gas temperatures 0 100 200 300 400 500 600 700 800 0 15 30 45 60 75 90 Time [min] Tem p eratu re (C) C Gastemp_1 C Gastemp_1_low C Gastemp_2 C Gastemp_2_low C Gastemp_3 C Gastemp_3_low C Gastemp_door C Gastemp_door_low

Figure 25 Calculated gas temperatures in the cabin. Numbers 1, 2 and 3, respectively, in the legend

refers to position denoted in the Figure 5.

Panel surface temperature

0 50 100 150 200 250 300 350 400 450 500 0 1 2 3 4 5 6 7 8 9 10 Time [min] Tem p eratu re [C]

C panel surf temp_1 C panel surf temp_2 C panel surf temp_3

Figure 26 Calculated panel temperatures (suspended ceiling) at locations 1-3 (Figure 5). The suspended

ceiling was modelled to collapse at 8 minutes.

Figure 27 shows the comparison of calculated laminate temperature with measured ditto at location 2 as denoted in Figure 5. The reason for the short measurement curve is technical problems during the

experimentiii_{. As can be seen from the figure, the agreement between simulated and measured data is} excellent and varies less than 1 % between 60 and 90 minutes. Additional calculated temperatures at laminate-insulation interface are shown in Figure 27 shows the calculated aluminium plate

temperatures at two positions (5 and 6). In the real experiment, the aluminium melted, which occurs at ~660°C. Laminate temperature 0 20 40 60 80 100 120 140 0 15 30 45 60 75 90 Time [min] Temperature [ C ] Calculated Measured

Figure 27 Comparison of calculated laminate temperatures with the measured at location 2

(see Figure 5) Laminate-insulation interface 0 20 40 60 80 100 120 140 0 15 30 45 60 75 90 Time [min] Temperature [ C ] C Ins-lam_2 C Ins-lam_1 C Ins-lam_3

Figure 28 Calculated temperatures at laminate-insulation interface at locations 1, 2 and 3

(see Figure 5).

Aluminium plate temperature 0 100 200 300 400 500 600 700 800 0 15 30 45 60 75 90 Time [min] Tem p eratu re [C] Al-plate temp_5 Al-plate temp_6

Figure 29 Calculated temperatures of aluminium plate at locations 5 and 6 (see Figure 5) as a function

of time.

6

Conclusions and discussion

A fast t2_{- fire, i.e. a HRR curve following a quadratic growth in time, has been used to simulate fires in} a RoPax public space with various ventilation conditions. The reason for using this type of curve was that it was considered to be “severe enough” for the purpose of defining a design fire that could be considered to be conservative when used in a analysis for estimating fire risks of using lightweight construction materials on the vessel.

It was found that the laminate temperatures in the deck above the enclosure after one hour in a ventilated scenario began to reach levels where a standard PVC-core based FRP sandwich composite (see Figure 2) would begin to loose strength, i.e. >100°C . After two hours of simulation, the laminate temperatures would, according to the simulations, be well above this temperature, however, still not so high (~350°C) as to produce pyrolysis gases and thereby contribute to the fire.

On the lower deck, however, temperatures were sufficiently high after only half an hour to start pyrolysing. The difference in behaviour was that the upper deck was insulated using a 100 mm insulation whereas the lower deck used only 30 mm. These results are also well in line with the large scale cabin fire experiment performed at SP in 20072 where the above deck managed the intense fire very well but the lower deck, having a more limited insulation, were somewhat damaged.

This large-scale cabin fire experiment was also simulated in the same way as the larger public space and the result showed a very good agreement between simulated and measured data. Input to the simulation was the cabin geometry, the material characteristics and the HRR-curve. The only difference in input data between this simulation and the public space simulation was the fire growth which for this case was available from the experiment while in the public space fire, a t2 _{fire was} defined as input. However, the experimental data was very close to a fast t2 _{curve. The results} obviously strengthen the reliability of the simulations made for the larger public space.

The experiments at SP in 2007 showed that a well-insulated composite construction material could sustain an intense cabin flash-over fire. The fire simulations made for the public space indicate that even though the fire intensity from such a large space with large openings would provide much higher total HRR, similar thermal loads would be obtained on the construction as in the SP experiment. However, having said that, it must be clearly stated that any simulation of a fire is uncertain due to the very non-linear nature of a fire event and that only real world experiments could be really relied upon. The work reported on here has tried to circumvent the difficulty and cost of running such very large experiments by comparing simulations with another large experiment and by using empirically measured material data as input to the simulation. Advanced computing instruments have been used where an 8-CPU work station made it possible to run simulations at high speeds. Also, mesh-independence of the simulations has been demonstrated.

6.1

Estimation of fuel input levels, HRR

These simulations show that the openings have large influence on the fire size. It was also found that in large fire simulations such as these, it is important for the modeller to define appropriate maximum heat release rate (fuel inflow from the fire source) when simulating fires in a large space with

comparably small openings and a large fire load. If too much fuel is input, this might have an important impact on the result due to cooling and lowering of the oxygen partial pressure.

Depending of the geometry of the compartment (i.e. small openings, low compartment height and large distance from opening to fuel bed), it was difficult, prior to these simulations, to know what the steady state HRR would be when the fire becomes ventilation controlled. The transfer of oxygen to fire in this kind of compartment, i.e. with the height less than 1/10 of the room length and width, resembles that of oxygen transfer to the fire in a tunnel with one end closed. The fresh air and

combustion products must travel a long distance to and from the fire through the same small opening. The existing general equations and hand calculation methods on expected HRR levels, which are valid in normal shaped rooms, are not applicable or are uncertain in flat rooms like this.

The estimation of HRR levels was made using two different computer programs, the 2-zone model program, Branzfire, and the CFD-based fire simulation tool FDS. When a fire in a closed room was tested with the two different programs the results obtained were virtually identical: extinction at about 8 minutes of fire and a maximum HRR of approximately 20 MW (see Figure 7). For each fire scenario simulated, the maximum level of HRR was therefore first calculated using Branzfire prior to the CFD calculations.

6.2

Gas temperatures

In scenarios with closed and open window (scenarios 2 and 3) the gas temperatures reached 600-1000 °C in the gas layer where the flaming combustion occurred. In gas layers without flames, typical values were lower; 500-600 °C. In scenario 2 the highest gas temperatures were obtained near the door opening where the oxygen content was highest. As flaming combustion occurred near the door the combustion products were then transported further inside the room, making further combustion impossible or less intense in the back of the room. Generally the combustion took place more towards the right hand side of the room as this was where the door opening was located. The open window simulation case (scenario 3) changed the fire behaviour so that flaming combustion occurred also in the back part of the room and near the open window, but also in this scenario mostly in the left side of the room. Worth noting is that this extra opening compared to scenario 2 did not provoke very much higher HRR (relatively speaking). This might be understood from the idea that the enclosure is too large and the distance between the door and window openings too long for the extra opening of a small window to really influence the ventilation situation.

The gas temperatures are generally lower, below 600 °C in scenario 4, i.e. stairway and door open, despite of the almost doubled HRR. The explanation to this is that the volume flow rate of cold fresh air was considerably larger due to much larger opening. Obviously, a 50 MW fire would be a more severe incident to handle due to the risk of fire spread to other parts of the vessel but the simulations indicate that the gas temperatures in the enclosure would not be as severe as in simulation scenarios 2 and 3.

6.3

Laminate temperatures

Laminate temperatures become quite high after some time, especially on the lower deck, where the insulation thickness was merely 30 mm.

A comparison between scenario 2 and 3 show that the levels of laminate temperatures were similar except that the hottest laminate regions were found mostly in the back part of the room in scenario 3 rather than near the door as in scenario 2. Similar pattern was seen on both upper and lower deck. After the first hour of simulated fire duration, the laminate temperatures were generally between 105 and 125 °C on the upper deck where the insulation thickness was 100 mm. After 90 minutes laminate temperatures reached a highest temperature of 230 °C but in some locations it was still as low as 150 °C. After two hours the laminate temperatures generally reached 200-300 °C but in some places it was still below 200 °C. The laminate temperatures locally reached 350-400 °C after 3 hours of fire. On the lower deck the highest laminate temperatures reached almost the floor gas temperature, in some locations over 700 °C. The reasons were that the insulation layer was thin on the lower deck and that the hot gas layer was close to the floor level after a short while.

The scenario with open staircase was simulated only for 60 minutes of fire. The laminate temperatures at the upper deck reached about the same levels as in scenarios 3 and 4 after the first hour of fire duration, i.e. approximately 100 °C.

1 _{FTP Code, International Code for Application of Fire Test Procedures, International Maritime Organisation, London, 1998} 2 _{Arvidson M., Axelsson J., Hertzberg T., Large-scale fire tests in a passenger cabin, SP Technical Research Institute of}

Sweden, SP Report 2008:33

3 _{Simultaneous Determination of thermal conductivity and thermal properties by Transient Plane Source method.}

An important input to the understanding of rock material properties, Construction Technology in Europe (ENBRI), Issue 28, 2005

4 _{Evans, David D., Ceiling jet flows, The SPPE Handbook of Fire Protection Engineering,SPPE/NFPA,1995,Chapter 4.} 5 _{Wade C.A. LeBlanc D. Ierardi J. and Barnett J.R., A Room-Corner Fire Growth & Zone Model for Lining Materials,}

ICFRE2 Conference, Maryland August 1997

Appendix A

Gas temperatures in Scenario 2

Gas temperatures in additional locations calculated in Scenario 2 are shown in Figures A1 through A7. Temperatures at x10,y32 0 200 400 600 800 1000 1200 0 30 60 90 120 150 180 Time [min] Te m p er at u re [C ] C Temp10-32-1 C Temp10-32-2 C Temp10-32-3 C Temp10-32-4 C Temp10-32-5

Figure A 1 Gas temperatures at x = 10 m, y = 32 m at heights 0.8, 1.3, 1.8, and 2.4 m. Closed window. Temperatures at x10,y7 0 100 200 300 400 500 600 700 800 900 0 30 60 90 120 150 180 Time [min] Te mperat ure [C] C Temp10-7-1 C Temp10-7-2 C Temp10-7-3 C Temp10-7-4 C Temp10-7-5

Figure A 2. Gas temperatures at x = 10 m, y = 7 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window.

Temperatures at x23,y32 0 100 200 300 400 500 600 700 800 0 30 60 90 120 150 180 Time [min] T e m p er at u re [ C ] C Temp23-32-1 C Temp23-32-2 C Temp23-32-3 C Temp23-32-4 C Temp23-32-5

Figure A 3. Gas temperatures at x = 32 m, y = 32 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window. Temperatures at x3,y19 0 200 400 600 800 1000 1200 0 30 60 90 120 150 180 Time [min] Te mperat ure [C] C Temp3-19-1 C Temp3-19-2 C Temp3-19-3 C Temp3-19-4 C Temp3-19-5

Figure A 4. Gas temperatures at x = 3 m, y = 19 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window.

Temperatures at x 39,y19 0 100 200 300 400 500 600 700 800 900 1000 0 30 60 90 120 150 180 Time [min] Temperat ure [C] C Temp39-19-1 C Temp39-19-2 C Temp39-19-3 C Temp39-19-4 C Temp39-19-5

Figure A 5. Gas temperatures at x = 39 m, y = 19 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window. Temperatures at x39,y32 0 100 200 300 400 500 600 700 800 900 1000 0 30 60 90 120 150 180 Time [min] Te m p er at u re [C ] C Temp39-32-1 C Temp39-32-2 C Temp39-32-3 C Temp39-32-4 C Temp39-32-5

Figure A 6. Gas temperatures at x = 39 m, y = 32 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window.

Temperatures at x39,y7 0 100 200 300 400 500 600 700 800 900 1000 0 30 60 90 120 150 180 Time [min] Te m p er at u re [C ] C Temp39-7-1 C Temp39-7-2 C Temp39-7-3 C Temp39-7-4 C Temp39-7-5

Figure A 7 Gas temperatures at x = 39 m, y = 7 m at heights 0.8, 1.3, 1.8 and 2.4m. Closed window.

Gas temperatures in Scenario 3

Temperatures at x10,y32 0 200 400 600 800 1000 1200 0 15 30 45 60 75 90 105 120 Time [min] T e mpe ratu re [c] C Temp10-32-1 C Temp10-32-2 C Temp10-32-3 C Temp10-32-4

Figure A 8 Gas temperatures at x = 10 m, y = 32 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures at x10,y7 0 100 200 300 400 500 600 0 15 30 45 60 75 90 105 120 Time [min] Te m p er at u re [C] C Temp10-7-1 C Temp10-7-2 C Temp10-7-3 C Temp10-7-4

Figure A 9 Gas temperatures at x = 10 m, y = 7 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures at x23,y32

0 100 200 300 400 500 600 700 800 0 15 30 45 60 75 90 105 120 Time [min] Te mp er at u re [ C ] C Temp23-32-1 C Temp23-32-2 C Temp23-32-3 C Temp23-32-4

Figure A 10 Gas temperatures at x = 23 m, y = 32 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures at x3,y19 0 100 200 300 400 500 600 700 800 900 0 15 30 45 60 75 90 105 120 Time [min] T e mpe ratu re [C] C Temp3-19-1 C Temp3-19-2 C Temp3-19-3 C Temp3-19-4

Figure A 11 Gas temperatures at x = 3 m, y = 19 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures at x39,y19

0 100 200 300 400 500 600 700 800 900 1000 0 15 30 45 60 75 90 105 120 Time [min] T e mpe ratu re [C] C Temp39-19-1 C Temp39-19-2 C Temp39-19-3 C Temp39-19-4

Figure A 12 Gas temperatures at x = 39 m, y = 19 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures at x 39,y32 0 200 400 600 800 1000 1200 0 15 30 45 60 75 90 105 120 Time [min] Temperat ure [C] C Temp39-32-1 C Temp39-32-2 C Temp39-32-3 C Temp39-32-4

Figure A 13 Gas temperatures at x = 39 m, y = 32 m at heights 0.8, 1.3, 1.8 and 2.4m. Open window.

Gas temperatures in Scenario 4

Gas temperature at x=26m,y=33.2 m

0 100 200 300 400 500 600 700 800 0 10 20 30 40 50 60 Time [min] Temperature [ C ] h=0.5 m h=1.0 m h=1.5 m h=1.8 m h= 2.3 m

Figure A 14 Gas temperatures in several heights as a function of time. This location is 4.8 m from the left bulkhead, about 25 m from the door openings and 10 m from the fire source, which is placed in the front part (opposite to the doors) of the room.

Gas temperature at x=41 m,y =33.2 m 0 100 200 300 400 500 600 0 10 20 30 40 50 60 Time [min] Tem p eratu re [C] h=0.5 m h=1.0 m h=1.5 m h=1.8 m h= 2.3 m

Figure A 15 Gas temperatures in several heights as a function of time. This location is 4.8 m from the right bulkhead, about 5 m from the front window, and about 10 to the right from the fire source.

Gas temperature at x=41 m,y =4.8 m

0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 Time [s] T e m p er at u re [ C ] h=0.5 m h=1.0 m h=1.5 m h=1.8 m h= 2.3 m

Figure A 16 Gas temperatures in several heights as a function of time. This location is 4.8 m from the left bulkhead, about 5 m from the front window, and about 10 m to the left from the fire source.

SP Technical Research Institute of Sweden

Box 857, SE-501 15 BORÅS, SWEDEN

Telephone: +46 10 516 50 00, Telefax: +46 33 13 55 02 E-mail: info@sp.se, Internet: www.sp.se

www.sp.se Fire Technology SP Report 2009:02 ISBN 91-7848-978-91-85829-85-9 ISSN 0284-5172 SP Technical Research Institute of Sweden develops and transfers technology for improving competitiveness and quality in industry, and for safety, conservation of resources and good environment in society as a whole. With Sweden’s widest and most sophisticated range of equipment and expertise for technical investigation, measurement, testing and certification, we perform research and development in close liaison with universities, institutes of technology and international partners.

SP is a EU-notified body and accredited test laboratory. Our headquarters are in Borås, in the west part of Sweden.