RoBound – Ro-ro space boundary fire protection – Fire integrity between ro-ro space and accommodation space

(1)

pierrick.mindykowski@ri.se

RISE Research Institutes of Sweden AB

Postadress Besöksadress Tfn / Fax / E-post Detta dokument får endast återges i sin helhet, om inte RISE i förväg skriftligen godkänt annat.

Box 857

501 15 BORÅS Brinellgatan 4 504 62 BORÅS 010-516 50 00 033-13 55 02 info@ri.se

RoBound – Ro-ro space boundary fire

protection – Fire integrity between ro-ro space

and accommodation space

Pierrick Mindykowski

RISE Report 2021:69 ISBN 978-91-89385-59-7

(2)

Summary

The International Maritime Organization, through its correspondence group on fire safety, has underlined the need for more scientific studies regarding the performance of A-60 boundaries in case of a ro-ro space fire, especially to prevent fire spread to accommodation spaces. RISE has carried out the RoBound project in order to answer to this need. The goal of the project was to clarify the performance of “state-of-the-art” fire boundaries between ro-ro spaces and accommodation spaces or other ro-ro spaces, and to give recommendations on how sufficient fire containment is ensured.

In order to obtain realistic exposure reached during a fire within a ro-ro space, simulations were performed using Computational Fluid Dynamics (Fire Dynamics Simulator).The first step was to model representative ro-ro spaces as well as representative cargo. Two

representative ro-ro spaces were then defined: closed and open ro-ro spaces with open ends. Concerning the cargo, the ro-ro spaces were assumed fully loaded with trucks or fully loaded with cars. Moreover, two types of thermal insulation were chosen, A-60 and A-30. The highest temperature given for each simulated case was then compared with time-temperature curves for designing fire safety.

Almost all comparisons showed that the hydrocarbon time-temperature curve fits better to the highest temperature reached in the simulations. The hydrocarbon time-temperature curve is more severe than the standard (cellulosic) time-temperature curve according to ISO 834, used for type approval of thermal insulation. Experimental tests were then carried out to observe the performance of A class insulation when exposed to the more representative hydrocarbon time-temperature curve in a cubic furnace. The fire insulations were mounted on steel plates with different thicknesses (5 mm, 6 mm and 12 mm).

Tests results showed a significantly reduced fire integrity when exposed to the hydrocarbon time-temperature curve, meaning that it took less time to reach the maximum temperature elevations required by the FTP Code (140 °C for the average temperature elevation and 180 °C for the highest temperature elevation). The reduction was about 50%, depending on the

thickness of the steel plate. These results apply for stone wool. Glass wool fire insulation was also used in the tests but it was deteriorated when exposed to the high heat exposure in accordance with the hydrocarbon time-temperature curve.

(4)

1 Introduction

A ro-ro ship is a vessel onto which cargo, generally trucks and personal vehicles, can be rolled on and rolled off, and it is the most common type of vessel in Swedish waters. In the past decades, there have been many fires on ro-ro ships and a major challenge is smoke and fire spread. RoBound was a research project which aims to clarify how "state-of-the-art" in passive fire protection between ro-ro spaces and adjacent spaces affects fire and smoke spread, and on this basis make recommendations on appropriate improvements. This was achieved through fire experiments, simulations, literature study of incident reports and regulations, and a workshop for identification of weaknesses and in collaboration with vessel operators, industry and authorities. Proposals for measures that contribute to a satisfactory and harmonized level of safety regarding the spread of smoke and fire was prepared and communicated to the Swedish Transport Agency, for consideration to be forwarded to the IMO (International Maritime Organization).

1.1 Background

Ro-ro ships have been an important component of the commercial maritime industry since their introduction in the 1940’s. The ships have a large longitudinal space where cars, trucks and other cargo can be rolled on and rolled off. Despite improved fire protection regulations, many fire accidents have occurred on ro-ro ships and there are no signs of them diminishing in number or magnitude. This was a conclusion at the IMO [1] based on a statistical study of ship fires, which has led to an ongoing update of the international fire safety regulations for ro-ro ships in SOLAS Chapter II-2 [2] and associated codes. During a review of the fire safety regulations [1], the IMO correspondence group has particularly pinpointed the need for additional experimental data or results of scientific studies regarding:

- The performance of A-60 boundaries in case of a ro-ro space fire, especially to prevent fire spread to accommodation spaces; and

- The performance of A-0 boundaries in case of a ro-ro space fire, especially to prevent fire spread between ro-ro spaces.

In this process, Sweden has moreover underlined the issue of the smoke tightness of A class divisions with doors. While smoke tightness is a requirement for A class divisions, the fire resistance test method (for doors same as for bulkheads and decks) in the Fire Test Procedures (FTP) Code [3] (Annex 1, part 3 – test for “A”, “B” and “F” class divisions, sections 1-6) is not designed to evaluate hazards associated with smoke spread. At the same time, prevention of smoke spread from a ro-ro space fire to the accommodation part of the ship is a difficult and significant problem on ro-ro ships.

1.2 Purpose and methodology of the current study

The purpose of the RoBound project was to clarify the performance of “state-of-the-art” fire boundaries between ro-ro spaces and accommodation spaces or other ro-ro spaces, and to give recommendations on how sufficient fire containment is ensured. The technical basis provided will support the revision of international IMO regulations and thus also the overall purpose to improve the independent management of fires on ro-ro ships.

RoBound aims to strengthen competence and influence regulation development regarding fire divisions of ro-ro ships.

The RoBound study covered two weaknesses that had been identified regarding the fire integrity of ro-ro spaces. The first weakness concerns a gap between the regulations and the

(5)

test method to approve doors in A class divisions, where the smoke tightness requirement is not represented in the standard fire test.

The second weakness, which is the one studied in this report, concerns the fire integrity

between ro-ro spaces and accommodation. It has been questioned whether A class fire integrity is sufficient when approved by using the standard time-temperature curve, which might not represent a real fire in a ro-ro space.

A literature study precedes this report and includes a review of regulations concerning fire integrity of ro-ro spaces and a review of accident investigations as basis to identify weaknesses in fire integrity [44]. The study documented in this report aimed to establish safe design regarding thermal insulation between ro-ro spaces and accommodation spaces. In practice, this was achieved by simulating, within a ro-ro space, the growth and propagation of fires that were considered as realistic based on real cargo representations. Simulations were performed using a Computational Fluid Dynamics software called FDS (Fire Dynamics Simulator [5]) in order to monitor the highest temperature reached in the ceiling of the ro-ro space when insulated by different thermal insulation. This highest temperature was then compared with

time-temperature curves for designing fire safety. From this comparison, the best fitting fire curve was selected and used to perform small scale fire resistance tests. These tests were used to compare the actual performance of thermal insulation when exposed to what was considered to be a realistic worst-case heat exposure in a ro-ro space fire.

(6)

2 Fire simulations to determine a realistic heat exposure

In order to define what is a realistic fire in a ro-ro space, the following questions need to be answered: What are realistic cargos representations in ro-ro spaces and how can they be modelled?

2.1 Representation and modelling of realistic cargo in ro-ro spaces

The representation of realistic cargo in ro-ro spaces was based on a previous study

(FIRESAFE II [6]) where a Standard RoPax owned by Stena Rederi has been used to perform fire simulations.

2.1.1 Geometry of ro-ro spaces and cargo representation

In the present study, it was decided to use two types of ro-ro spaces what are representatives of ro-ro spaces. From one hand, an open ro-ro space and from other hand a closed ro-ro space. Both spaces had the following size:

- 65 meters long, and - 26 meters wide.

A deviation from the Standard RoPax concerns the height of ro-ro spaces. It was decided to fit it in function of the nature of the cargo. Two types of cargo was chosen:

- Ro-ro space full of trucks with a length of 15 meters, width of 2.5 meters and a height of 4.2 meters.

- Ro-ro space full of cars with a length of 4.4 meters, width of 1.6 meters and a height of 2 meters.

Based on the cargo representation, the height of the ro-ro spaces was:

- Trucks: 6 meters height, giving a free space between the top of trucks and ceiling of 1.8 meters, but only 0.6 meter clearance taking into account the presence of transversal stiffeners having a height of 1.2 meters.

- Cars: 3 meters height, giving a free space between the top of trucks and ceiling of 1.5 meters, but only 0.3 meter taking into account the presence of transversal stiffeners having a height of 1.2 meters.

The geometry of the ro-ro spaces and the defined above allow for the following cargo arrangement:

- Trucks: 8 lanes of 4 trucks and a total of 32 trucks - Cars: 10 lanes of 12 cars and a total of 120 cars.

It should be noted here that a row of cars was added in front of the fully open end in order to see if the fire could propagate from the cargo inside the ro-ro space to the cargo outside the space. This row of cars was the same for both types of cargo representation and consisted of 10 cars with the same geometry as the cars inside the ro-ro space.

Both representative spaces were designed with one fully open end. The decision to have a fully open end for the closed ro-ro space was based on the RO5 project [7] stating that a fire in a fully enclosed ro-ro space will self-extinguish, and thus representing a less critical scenario for the thermal insulation.

Concerning the open ro-ro space, permanent side openings were assumed and taken from the existing Standard RoPax ship. They were designed in 3 groups of openings on each side of the ship:

(7)

- 2 groups of 4 openings - 1 group of 2 openings.

The dimensions of the openings depend on the type of cargo in the ro-ro space. In case of trucks, the dimensions were 2 meters by 3.2 meters. With cars the dimensions were 2 meters by 1.3 meters.

Concerning the thermal insulation of the ro-ro space, it was selected based on what was installed on the Standard RoPax ship. Only the ceiling was therefore assumed to be insulated, with a standard thermal insulation certified for A-30 or A-60 performance.

2.1.1.1 Modelling of geometry and cargo

The geometries of the ro-ro spaces and their cargo representations defined above were modelled in the FDS software as illustrated in Figure 1- Figure 4 below.

Figure 1. FDS representation of the open ro-ro space fully loaded with trucks.

Figure 2. FDS representation of the closed ro-ro space fully loaded with trucks.

(8)

Figure 4. FDS representation of the closed ro-ro space fully loaded with cars.

In order to perform simulations with FDS, a set of parameters need to be established.

The domain of calculation was defined as 70 meters (x axis) by 30 meters (y axis) by 7 meters (z axis), with a discretization of 210 points of calculation in the x axis, 75 points in the y axis and 28 in the z axis. The set was based on a rectilinear numerical grid.

Moreover, all surfaces constituting the ro-ro space were assumed to be steel with a thickness of 10 mm. Steel properties were taken from the Eurocode 3 [8] and are presented in Table 1. Table 1. Steel properties

Emissivity (-) 0.7

Density (kg/m3₎ ₇₈₅₀

Specific heat (J/kg.K) 450 + 0.28 ∙ T_s - 2.91 ∙ 10-4∙ T_s2 + 1.34 ∙ 10-7∙ Ts3 Conductivity (W/m.K) 14.6 + 1.27 ∙ 10-2∙ Ts

Ts represents the temperature of the steel in Celsius.

Regarding the properties of the thermal insulation, they were taken from the manufacturers and are presented in Table 2.

Table 2. Thermal insulation properties Material

property A-30 A-60

Thickness (cm) 5 7 Emissivity (-) 0.7 0.7 Density (kg/m3₎ ₂₄ ₅₆ Specific heat (kJ/kg._K) Ti =100 0.84 0.84 Ti=400 1.154 1.154 Ti=600 1.212 1.212 Ti=700 1.586 1.586 Conductivity (W/m._K) Ti=10 0.032 0.031 Ti=100 0.045 0.041

(9)

Ti=200 0.069 0.057

Ti=550 0.18 0.18

Ti=650 0.22 0.22

Ti=700 0.23 0.23

Ti represent the temperature of the thermal insulation in Celsius.

2.2 Modelling of fire within ro-ro spaces

In order to describe what can be a realistic fire within a ro-ro space, it was decided to use experimental data from literature to be able to build a design fire curve for simulations of 60 minutes of fire, as further elaborated below

2.2.1 Design fire

The design fire selected was based on experiments performed during the project FIRETUN-EUREKA 499 [9] where a Heavy Good Vehicle (HGV) had burned in a tunnel and the Heat Release Rate (HRR) was monitored. In Figure 5, the result of the HRR of the burning HGV is presented.

Figure 5. Heat Release Rate of a burning Heavy Goods Vehicles [9].

To be able to produce a design fire curve closer to reality, the capacity of FDS should be taken into account. According to Wahlqvist and van Hees [10], the influence of the environment (e.g. insulation capacity of the walls, opening dimensions, height and surface material of the room) can be high on the fire growth rate when simulating with FDS. It is explained by the fact that FDS does not take into account the re-radiation of the hot smoke layer to the fire itself. Wahlqvist and van Hees have found that the fire growth rate can be multiplied by 4 when a fire is enclosed. Therefore, in the current project it was decided to modify the HRR curve from the

0 20 40 60 80 100 120 140 160 0 10 20 30 40 50 60 H RR (MW) Time (min)

(10)

experiments (Figure 5) and increase the fire growth rate by a factor of 4. The resulting fire curve is presented in the Figure 6.

Figure 6. Heat Release Rate of the design fire compared to the experimental Heavy Goods Vehicles curve [9].

It can be noted that the maximal Heat Release Rate is around 120 MW which corresponds to a full Heavy Goods Vehicle burning.

In FDS, several solutions can be used to model the area of a vehicle fire. It can be the top surface of a vehicle, but in such case the fire becomes very concentrated and does not represent the reality. Indeed, a fire on a truck comes from all the surfaces (top and side surfaces) due to the fact that a trailer is practically always made of tarpaulin. Therefore it was decided that the possible areas of the fire of a truck would consist of the top area as well as the side areas. This led to the use of a Heat Release Rate Per Unit Area (HRRPUA) calculated by the total Heat Release Rate divided by the total possible burning area. The calculation of the HRRPUA is presented in Table 3. Concerning the Heat Release Rate for cars, it was decided to take the value of the HRRPUA for a truck and to calculate the total Heat Release Rate based on its total possible burning area. For the cars cargo representation, the possible burning area consists only of the top surface of the car, in contrary to the calculation for trucks. Indeed, the fire load of a car (seats, dashboard and engine) is mainly situated at the same height. The total HRR calculated was found close to data found in literature [11].

Table 3. Values of Heat Release Rate and Heat Release Per Unit Area used for trucks and cars Type of

vehicle Length (m) Width (m) Height (m)

Total possible burning area (m2₎ HRRPUA (MW/m2₎ HRR total (MW) Truck 15 2.5 3.2 149.5 800 120 Car 4.4 1.6 1.5 7 800 5.6 0 20 40 60 80 100 120 140 160 0 10 20 30 40 50 60 H RR (MW) Time (min) Design fire HGV fire from EUREKA 499

(11)

Based on Table 3, design fire curves for trucks and cars were determined, as presented in Figure 7.

Figure 7. Design fire curves for one car and one truck used in this study. 2.2.2 Propagation of fire within a ro-ro space

With the design fire curves defined, the propagation of fire to involve further vehicles should also be taken into account. FDS proposes an easy way for how to model fire propagation. It is based on a definition of the thermal properties of target and a heat transfer model, to calculate the internal increase of temperature of the target. When this temperature reaches the ignition temperature of the material constituting the target, ignition of the target is triggered.

In order to keep the approach realistic but simplified, the material as target for ignition was taken to be the same for cars and trucks i.e. natural rubber. This material is used for the fabrication of tyres and windows joints, which have proved to be the first materials that ignite in vehicles fires [12]. The thermal properties and ignition temperature of natural rubber used in the simulations are described in Table 4.

Table 4. Thermal properties of natural rubber [12]

Material Thickness (cm) Ignition Temperature (°C) Conductivity (W/m/K) Specific Heat (J/kg/K) Density (kg/m3₎ Natural rubber 1 250 0.13 1880 910 2.2.3 Virtual sensors

The aim of the FDS simulations was to monitor the maximal temperatures reached during a fire in a ro-ro space. Therefore, a grid of virtual sensors registering the temperatures was created. It consisted of 64 thermocouples placed 10 cm below the ceiling. The longitudinal and transversal positions of sensors are described in Table 5.

0 20 40 60 80 100 120 140 160 0 5 10 15 H RR (MW) Time (min)

Design fire for truck

(12)

RISE Research Institutes of Sweden AB Table 5. Positions of thermocouples.

Longitudinal position (m)

Position 0 was at the fully open end

Transversal position (m)

Position 0 was at the left wall, looking from the fully open end.

6 2.85 11 5.75 21.8 8.65 26.8 11.55 37.4 14.45 42.4 17.35 53 20.25 58 23.15

2.2.4 Position of the fire origin

With the propagation of the fire defined, the origin of the fire should be described. For the four different geometry cases (two types of ro-ro spaces and two cargo

representations), the position of the origin of the fire was the same. It corresponded to a vehicle (truck or car) positioned in the right forward part of the space, at the following coordinates:

- Longitudinal: 48 meters - Transversal: 24.4 meters.

The vehicles at this position are shown in their respective configuration in Figure 8 - Figure 11 below.

(13)

Figure 8. Position of the fire origin (vehicle marked red) for the open ro-ro space fully loaded with trucks.

Figure 9. Position of the fire origin (vehicle marked red) for the closed ro-ro space fully loaded with trucks.

(14)

Figure 10. Position of the fire origin (vehicle marked red) for the open ro-ro space fully loaded of cars.

Figure 11. Position of the fire origin (vehicle marked red) for the closed ro-ro space fully loaded of cars.

2.3 Results of fire simulations

The results of the simulations were represented by the temperatures which were monitored by the virtual sensors.

2.3.1 Temperatures within the ro-ro spaces

As stated previously, a total of 64 sensors were used in the simulations. For a better overview, the sensor showing the highest temperature at every given time gave the maximum

(15)

Figure 12. Maximum temperatures during fire simulated in a ro-ro space loaded with trucks.

Figure 13. Maximum temperatures during fire simulated in a ro-ro space loaded with cars.

It should be noted here that the row of sensors closest to the end opening (longitudinal of 6 meters) and the sensors adjacent to the fire source were discarded. At this position, non-combusted gases reached oxygen from the open end resulting in flaming combustion and high

0 200 400 600 800 1000 1200 1400 0 500 1000 1500 2000 2500 3000 3500 Te m p era tu re ( °C) Time (s) TRUCKS OPEN A60 TRUCKS OPEN A30 TRUCKS CLOSED A60 TRUCKS CLOSED A30 0 200 400 600 800 1000 1200 1400 0 500 1000 1500 2000 2500 3000 3500 Te m p era tu re ( °C) Time (s) CARS OPEN A60 CARS OPEN A30 CARS CLOSED A60 CARS CLOSED A30

(16)

temperatures. These flame temperatures do not represent the heat exposure within a ro-ro space.

In each scenario, the maximum temperatures presented in Figure 12 and Figure 13 corresponded to one sensor, presented in Table 6.

Table 6. Positions of the thermocouples showing the highest temperatures.

Type of cargo Type of ro-ro space Type of insulation Longitudinal position of the sensor (m) Transversal position of the sensor (m)

Cars Open A-60 22.2 24.4

Cars Open A-30 22.2 24.4

Cars Closed A-60 22.2 20.25

Cars Closed A-30 22.2 20.25

Trucks Open A-60 58 14.45

Trucks Open A-30 58 17.35

Trucks Closed A-60 58 14.45

Trucks Closed A-30 58 17.35

For a better presentation, the areas of the sensors having the highest temperatures are presented in Figure 14 and Figure 15 for the trucks and the cars representation, respectively.

Figure 14. Area (in blue) of the maximum temperatures during simulated fire in ro-ro space loaded with trucks.

(17)

Figure 15. Area (in blue) of the maximum temperatures during simulated fire in ro-ro spaces loaded with cars.

2.4 Analysis of results

The very first conclusion is that the type of thermal insulation (A-30 or A-60) seems not to have an influence on the maximum temperatures reached during simulated fire. This result is not surprising and can be explained by the fact that FDS does not take into account the

feedback of hot smoke to the source of the fire, which will in reality accelerate the heat release rate as explained in paragraph 2.2.1.

Concerning the influence of the type of ro-ro space, a difference can be noted between whether the cargo consisted of trucks or cars. With cars, the ro-ro space had a big influence, with temperatures differing up to 500 °C between an open and a closed ro-ro space. This result is in line with the results of the project RO5 [7] where it was demonstrated that an open ro-ro space gives a more severe fire due to the contribution of oxygen from side and end openings. On the other hand, there was no big influence of the space when the cargo consisted of trucks. In fact, there is no influence until about 25-30 minutes (1 500 - 2 000 seconds). This might be

explained by the difference of free height between the spaces fully loaded with trucks and cars. With trucks, the free height under the transversal stiffeners was 0.6 meters, which was twice that for the space with cars, i.e. 0.3 meters of clearance. This difference allows a higher movement of air transporting oxygen during the fire. A higher free height gives a higher oxygen flow and thus a more severe fire.

2.4.1 Comparison of simulated temperatures with time-temperature curves

In fire safety engineering, several time-temperature curves are used. For instance, the fire integrity which is defined by fire testing according to the FTP Code is tested using a time-temperature curve described in ISO 834 (for more explanations about fire integrity tests, please refer to paragraph 2.3.2 of the Literature study report from the RoBound project [4]). This time-temperature curve has been developed in order to represent a typical enclosure fire encountered in building, based on a typical cellulosic-based fire load. Another

time-temperature curve that should be mentioned here is one representing a hydrocarbon-based fire load (named the HC curve), commonly used in offshore applications. Both curves are

(18)

Figure 16. Illustration of the ISO 834 and Hydrocarbon time-temperature curves.

According to Figure 16, it is clear that the HC curve is more severe than the ISO 834 curve. The two curves were compared with results of the simulations in the Figure 17 and Figure 18, to evaluate how well they representative a fire in a ro-ro space.

0 200 400 600 800 1000 1200 0 500 1000 1500 2000 2500 3000 3500 Te m p era tu re ( °C) Time (s) ISO 834 fire curve HC fire curve

(19)

Figure 17. Comparison of the highest temperatures during a simulated fire within a ro-ro space loaded with trucks and time-temperature curves.

Figure 18. Comparison of the highest temperatures during a simulated fire within a ro-ro space loaded with cars and time-temperature curves.

For the scenario with a fire in a ro-ro space fully loaded with trucks, it is clear that the HC time-temperature curve is more representative of the worst-case heat exposure within a ro-ro space loaded by trucks. Indeed, after 5 minutes of fire, the temperature encountered is higher

0 200 400 600 800 1000 1200 1400 0 500 1000 1500 2000 2500 3000 3500 Te m p era tu re ( °C) Time (s) TRUCKS OPEN A60 TRUCKS OPEN A30 TRUCKS CLOSED A60 TRUCKS CLOSED A30 ISO 834 fire curve HC fire curve 0 200 400 600 800 1000 1200 1400 0 500 1000 1500 2000 2500 3000 3500 Te m p era tu re ( °C) Time (s) CARS OPEN A60 CARS OPEN A30 CARS CLOSED A60 CARS CLOSED A30 ISO 834 fire curve HC fire curve

(20)

than the ISO 834 time-temperature curve temperature. The heat exposure in case of a fire in a ro-ro space loaded with trucks would therefore be better represented by the HC curve.

For the scenario with a fire in a ro-ro space with cars, a differentiation between a closed and an open ro-ro space should be made. For a closed ro-ro space fully loaded with cars, the ISO 834 time-temperature curve is adequate to represent a worst-case heat exposure. However, for an open ro-ro space fully loaded by cars, the Hydrocarbon curve is more representative to the worst-case heat exposure. The conclusion of the simulation comparison is summarized in the Table 7.

Table 7. Time-temperature curves best representing simulated fires within ro-ro spaces Type of cargo Type of ro-ro

space

Type of insulation

Representative standard fire curve

Cars Open A-60 Hydrocarbon curve

Cars Open A-30 Hydrocarbon curve

Cars Closed A-60 ISO 834 curve

Cars Closed A-30 ISO 834 curve

Trucks Open A-60 Hydrocarbon curve

Trucks Open A-30 Hydrocarbon curve

Trucks Closed A-60 Hydrocarbon curve

(21)

3 A class thermal insulation vs. realistic fire

It was demonstrated above that a realistic fire within a ro-ro space is better represented by the Hydrocarbon time-temperature curve than by the ISO 834 curve, today used to type approve thermal insulation on ships. Using the same criteria for the approval of A class thermal insulation defined by the SOLAS, a set of experiments were performed to judge the likely fire integrity of an A class thermal insulation when exposed to the hydrocarbon time-temperature curve in a small cubic furnace.

3.1 Definition of A class thermal insulation

SOLAS II-2/3 defines the following for A class divisions in §2:

"A" class divisions" are those divisions formed by bulkheads and decks which comply with the following criteria:

1. they are constructed of steel or other equivalent material; 2. they are suitably stiffened;

3. they are insulated with approved non-combustible materials such that the average temperature of the unexposed side will not rise more than 140°C above the original temperature, nor will the temperature, at any one point, including any joint, rise more than 180°C above the original temperature, within the time listed below:

class "A-60" 60 min class "A-30" 30 min class "A-15" 15 min class "A-0" 0 min

4. they are constructed as to be capable of preventing the passage of smoke and flame to the end of the one-hour standard fire test; and

5. the Administration has required a test of a prototype bulkhead or deck in accordance with the Fire Test Procedures Code to ensure that it meets the above requirements for integrity and temperature rise.

As stated previously, the FTP Code evaluates the fire integrity of an A class division using the ISO 834 standard time-temperature curve. It is hence with this thermal exposure that the temperature of the unexposed side of thermal insulation should not rise more than 140 °C above the original temperature.

In order to judge the likely fire integrity of an A class thermal insulation when exposed to more realistic heat exposure, a set of experiments were performed, as presented in section 3.2.

3.2 Thermal insulation experiments

The experiments conducted for this part of the study were based on standard tests done according to the FTP Code. A thermal insulation which had already been tested and type approved was mounted on a steel plate (different thicknesses of the steel plate were tested) and placed horizontally on a cubic furnace with the dimensions: 1.2 m x 1.2 m x 1.2 m. This furnace is smaller than a standard furnace used for type approving material but was used with consideration to the project budget. The furnace is shown on Figure 19 and Figure 20.

(22)

Figure 19. Picture of the furnace used for the testing (top view).

(23)

Concerning the thermal insulation, Figure 21 and Figure 22 present the arrangement of the thermal insulation mounted on the steel plate as well as the positioning on the furnace.

Figure 21. Presentation of the mounting of the thermal insulation on the steel plate.

Figure 22. Position of the thermal insulation on the furnace.

The testing procedure consisted in exposing the thermal insulation, mounted on a steel plate, to a heat exposure following the hydrocarbon time-temperature curve for a time of 30 minutes or 60 minutes. The increase in temperature on the unexposed side was monitored with the help of thermocouples mounted as indicated in Figure 21, in similarity with the FTP Code, Part 3. Concerning the type of thermal insulation, it should be noted that currently several types of materials are used on ships. For the current study, two types of materials were tested: glass wool and stone wool.

(24)

Three different thickneesses of the steel plate were used: 5 mm (as required by the FTP Code), 6 mm (used on real ships when the ro-ro space is designed for cars) and 12 mm (used on real ships whenthe ro-ro space is designed for trucks). The matrix of tests is represented in the Table 8.

Table 8. Matrix of the tests.

Test number Class of thermal insulation Type of thermal insulation material Thickness of the steel plate (mm) Duration of test (minutes)

1 A-30 Glass Wool 6 30

2 A-60 Stone wool 6 60

3.3 Results of thermal insulation experiments

In this section, the elevation of temperatures monitored by the five thermocouples and their average are presented for each performed tests.

3.3.1 Test 1: A-30 glass wool on 6 mm steel plate

Test 1 was carried out with A-30 thermal insulation made of glass wool with a steel plate of 6 mm. The temperature results from test 1 are presented in Figure 23.

Figure 23. Elevation of temperature in test 1: A-30 glass wool on 6 mm steel plate. Test 1 had to be stopped before the planned end time. At 8 minutes, the elevation of temperatures has already reached around 500 °C. This unexpected temperature can be

0 50 100 150 200 250 300 350 400 450 500 550 600 0 1 2 3 4 5 6 7 8 Ele vat ion o f Te m p era tu re ( °C) Time (min)

Test 1

Thermocouple 1 Thermocouple 2 Thermocouple 3

(25)

explained by the fact that the insulation (glass wool) melted during the test. The final state of the insulation is shown in Figure 24.

Figure 24. Picture of the glass wool insulation after 8 minutes of heat exposure according to the Hydrocarbon time-temperature curve.

Surprised by the results, an additional test was performed with the A-30 glass wool thermal insulation but with heat exposure according to the cellulosic time-temperature curve. Figure 25 shows this thermal insulation after 30 minutes.

(26)

Figure 25. Picture of the glass wool insulation after 30 minutes of heat exposure according to the cellulosic (ISO 834) standard time-temperature curve.

Figure 25 demonstrates that the glass wool thermal insulation can sustain the cellulosic standard time-temperature curve, while it deteriorated when exposed to the more severe Hydrocarbon time-temperature curve.

3.3.2 Test 2: A-60 stone wool on 6 mm steel plate

(27)

Figure 26. Elevation of temperature in test 2: A-60 stone wool on 6 mm steel plate. A quite homogenous range of temperatures was measured in test 2.

Test 3 consists on A-30 thermal insulation made of stone wool with a steel plate of 6 mm. 0 50 100 150 200 250 300 350 400 0 10 20 30 40 50 60 Ele vat ion o f tem p era tu re ( °C) Time (min)

Test 2

Thermocouple 1 Thermocouple 2 Thermocouple 3 Thermocouple 4 Thermocouple 5 Average

(28)

Figure 27. Elevation of temperature in test 2: A-30 stone wool on 6 mm steel plate. Regarding test 3, two thermocouples (1 and 5, see Figure 21) seemed to show erroneous behaviour compared to the other three thermocouples. This difference might be explained by the fact that the insulation might have lost its contact with the steel plate, or the insulation joints (as shown in Figure 21) might have opened up during the test. As no visual observation were done after this tests, the previous explanations are only hypothetical.

Test 4 consists on A-30 thermal insulation made of stone wool with a steel plate of 5 mm.

0 100 200 300 400 500 600 0 5 10 15 20 25 30 Elev atio n o f te m p erature (° C) Time (min)

Test 3

(29)

Figure 28. Elevation of temperature in test 2: A-30 stone wool on 5 mm steel plate. In the test 4, the thermocouples 3 and 5 showed a steep increase in temperature around 20 minutes. This behaviour did not influence the final result (as shown in Table 9) and can be explained by the fact that the insulation did not fully resist the intensity of the heat exposure and started to deteriorate, as shown in the Figure 29. It should be noted here that the mounting of the thermal insulation did not follow the requirement as shown in Figure 21 due to a shortage of insulation. 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 Elev atio n o f te m p erature (° C) Time (min)

Test 4

(30)

Figure 29. Picture of the stone wool thermal insulation after test 4.

(31)

Figure 30. Elevation of temperature in test 2: A-30 stone wool on 12 mm steel plate.

In this test, thermocouple number 4 showed a different behaviour than the others. It resulted in a relatively high average temperature compared with thermocouples 1, 2, 3 and 5.

Test 6 consists on A-60 thermal insulation made of stone wool with a steel plate of 12 mm.

0 50 100 150 200 250 300 350 0 5 10 15 20 25 30 Elev atio n o f te m p erature (° C) Time (min)

Test 5

(32)

Figure 31. Elevation of temperature in test 2: A-60 stone wool on 12 mm steel plate. The results of test 6 showed a normal behaviour in elevation of temperatures, without any apparent issue concerning measurements.

3.4 Analysis of the thermal insulation experiment results

An analysis of the tests consisted in tracking the highest and average temperature elevations reached at the end of the tests. All temperatures were used, even those showing a different behaviour from others during the same test. These temperatures were compared with the requirements in the FTP Code:

• Maximum 140 °C average temperature elevation (referred to as criterion 1); and • Maximum 180 °C peak temperature elevation (referred to as criterion 2). The comparison of temperature elevations is summarized in Table 9.

Table 9. FTP Code requirements compared with temperature elevations in tests with heat exposure according to the hydrocarbon time-temperature curve

Test Number Class of thermal insulation Steel plate thickness (mm) Peak temperature elevation in tests (°C) Highest average temperature elevation in tests (°C) Fire integrity based on the FTP Code using the ISO 834 standard curve (min) Fire integrity based on HC curve Time (min) Criterion

1 A-30 6 Test stopped at 8 minutes

0 50 100 150 200 250 300 350 0 5 10 15 20 25 30 35 40 45 50 55 60 Elev atio n o f te m p erature (° C) Time (min)

Test 6

(33)

2 A-60 6 346 286 60 20 2

3 A-30 6 503 354 30 11 1

4 A-30 5 782 687 30 11 1

5 A-30 12 324 232 30 17 2

6 A-60 12 297 236 60 31 1

Without any surprise, the fire integrity based on the HC curve is lower than the fire integrity based on the ISO 834 standard curve.

A result which does not require much analysis is related to the test with glass wool thermal insulation. A-30 or A-60 class cannot resist to a heat exposure according to the hydrocarbon time-temperature curve. The material melted and shrunk when exposed to high temperature in the tests. Based on this result, thermal insulation of glass wool should be avoided when protecting an accommodation space from a fire in a ro-ro space (unless it is a closed ro-ro space designed for cars).

Concerning thermal insulation of stone wool, the level of protection was significantly reduced when exposed to the hydrocarbon time-temperature curve, but it still presented a certain degree of protection. Table 10 shows the reduction in thermal integrity capacity when the A class stone wool insulation was exposed to the hydrocarbon curve.

Table 10. Reduction in fire integrity of A class stone wool thermal insulation exposed to the hydrocarbon time-temperature curve

Test Number Class of thermal insulation Steel plate thickness (mm) Fire integrity based on ISO 834 (min) Fire integrity based on HC tests* (min) Reduction in fire integrity (HC vs ISO834) 2 A-60 6 60 20 66% 3 A-30 6 30 11 60% 4 A-30 5 30 11 63% 5 A-30 12 30 17 43% 6 A-60 12 60 31 48%

*The tests were performed in a reduced scale cubic furnace of 1.2 m, which is a scale of about 1:2.5. The reductions presented in Table 10 are influenced by the thickness of the steel plate on which the insulation was mounted. When the steel plate thickness was around 5-6 mm, the reduction was around 65%. As expected, when the steel plate was thicker, the reduction was less. With a steel thickness of 12 mm, the reduction was decreased to about 45%.

(34)

4 Conclusion

The goal of this part of the RoBound project was to clarify the fire integrity performance of fire boundaries between ro-ro spaces and accommodation spaces and to give recommendations on how sufficient fire integrity can be ensured regarding thermal insulation. This was studied by judging if the currently required fire integrity between ro-ro spaces and accommodation spaces can sustain a realistic fire encountered in a ro-ro space. In order to define such a fire, simulations were performed with representative ro-ro spaces (a closed and an open ro-ro space) and representative cargo (full load of trucks or cars). The main result of the simulations was that a fire in ro-ro spaces designed for trucks and in open ro-ro spaces designed for cars (lower height) present a worst-case heat exposure which is better modelled by the hydrocarbon time-temperature curve than the standard (cellulosic) time-temperature curve (ISO 834). The hydrocarbon curve is more severe than the cellulosic curve, which is used to test and approve thermal insulation according to the FTP Code.

A series of experiments was carried out to understand how the performance of standard A class thermal insulation is affected by exposure to the more severe hydrocarbon curve. Different types of certified thermal insulation were used: glass and stone wool achieving A-30 or A-60 class. The thermal insulation was installed on steel plates with varying thickness and exposed to the hydrocarbon time-temperature curve in a reduced scale furnace (scale 1:2.5 compared to the FTP Code). The results showed that thermal insulation based on glass wool should be avoided in well-ventilated ro-ro spaces, i.e. ro-ro spaces with openings or if the height is designed for trucks and an end is open. Temperature measurements on the unexposed side showed that stone wool gave a reduced level of protection when exposed to the hydrocarbon fire.

Based on the investigations made, the maritime community should be made aware that the current requirements do not provide a protection of 60 minutes (with A-60 thermal insulation) or 30 minutes (with A-30 thermal insulation). Depending on the steel plate thickness, the reduction in thermal insulation capacity was 45-60% of the values indicated by regulations, when exposed to a hydrocarbon heat exposure (realistic for most ro-ro spaces). If ship owners or shipbuilders really want to achieve 30 or 60 minutes of thermal insulation for a ro-ro space, they are recommended to assume a 50% reduction and use thermal insulation of a higher standard (e.g. A-60 instead of A-30) or use insulation type approved with the hydrocarbon curve (e.g. HC-30).

(35)

5 References

1. S.-C. o. F. s. Implementation, “FSI 21/5 Casuality Statistics and investigation,” International Maritime Organization, London, 2012.

2. IMO, International Convention for the Safety of Life at Sea (SOLAS), 1974, London: International Maritime Organization, 1974.

3. International Maritime Organization, International Code for Application of Fire Test Procedures, 2010, London: IMO, 2010.

4. A. Olofsson, F. Evegren, P. Mindykowski, RoBound project – Literature Study. 5. K. McGrattan, S. Hostikka, J. Floyd, R. McDermott, and M. Vanella. (2020). Fire

Dynamics Simulator, User’s Guide (NIST Special Publication 1019). National Institute of Standards and Technology.

6. J. Leroux et al. (2018). FIRESAFE II, Containment and Evacuation, 2018.

7. A. Olofsson et al., Model scale tests of a ro-ro space fire ventilation (RO5 project), 2020. 8. European Committee for Standardization. (2005). Eurocode 3: Design of steel structures.

Part 1-2: General rules. Structural fire design. (EN Standard EN 1993-1-2:2005). 9. A. Haack, Fire protection in traffic tunnels: General aspects and rsults of the EUREKA

project

10. Wahlqvist, J., & Van Hees, P.. Validation of FDS for large-scale well-confined mechanically ventilated fire scenarios with emphasis on predicting ventilation system behavior. Fire Safety Journal, 62, 102-114, 2013.

11. A. Lönnermark, H. Ingason, Recent acheivements regarding heat release and temperatures during fires in tunnels, Safety in Infrastructure – Budapest 2004.

12. T. Marton, A. Dederichs, L. Giuliani, Modelling of fire in an open car park, Applications of structural fire engineering, 2016.