Fire Tests of FRP Composite Ship Structures

(1)

SP Report 2016:35

S

P

T

ec

hni

ca

l R

es

ear

ch I

ns

tit

ut

e of

S

w

ed

en

(2)

(3)

FRP composite structures instead of steel could significantly reduce weight but could also potentially increase fire risks. This report documents a number of fire tests performed with the objective to quantify and ensure safety when lightweight FRP composite are used as ship structures. Two test series were carried out with focus on reaction to fire performance of external combustible FRP surfaces. Five test series consisted in fire resistance to evaluate FRP the structural fire integrity of different FRP composite structures. Three feasible safety measures were found suitable to protect external FRP surfaces: a drencher system (3 mm/min), a fire protective coating (LEO) and a certified balcony sprinkler. The fire resistance tests showed that an insulated FRP composite bulkhead concept can be conservatively evaluated by testing the bulkhead with the highest design load. Loadbearing double FRP composite bulkheads can provide sufficient fire resistance when traditional insulation is not applicable (e.g. for external surfaces). Key words: Fire, FRP composite, fire test, fire resistance, structural integrity, fire integrity.

SP Sveriges Tekniska Forskningsinstitut

SP Technical Research Institute of Sweden SP Report 2016:35

ISBN 978-91-88349-37-8 ISSN 0284-5172

(4)

3 Summary of fire tests 10

3.1 Fire growth tests of protected external FRP 10

3.2 Balcony sprinkler tests with external FRP 12

3.3 Fire resistance tests of loadbearing FRP composite bulkheads 13 3.4 Fire resistance test of loadbearing double FRP composite

bulkhead (non-insulated) 16

3.5 Fire resistance test of loadbearing FRP composite bulkhead

with steel joint 16

3.6 Fire resistance test of sandwich panel bulkhead with

installations 17

3.7 Fire resistance test of novel loadbearing FRP composite deck 18

4 Summarized conclusions 20

5 References 20

Annex A. Fire growth tests of protected external FRP Annex B. Balcony sprinkler tests with external FRP

Annex C. Fire resistance test of loadbearing FRP composite bulkhead with 12.4 kN design load

Annex D. Fire resistance test of loadbearing FRP composite bulkhead with 20.7 kN design load and reduced safety factor

Annex E. Fire resistance test of loadbearing FRP composite bulkhead with 7 kN design load

Annex F. Fire resistance test of loadbearing thick FRP composite bulkhead with 39,2 kN design load

Annex G. Fire resistance test of loadbearing stiffened FRP composite bulkhead with 31 kN design load

Annex H. Fire resistance test of loadbearing double FRP composite bulkhead (non-insulated)

Annex I. Fire resistance test of loadbearing FRP composite bulkhead with steel joint Annex J. Fire resistance test of sandwich panel bulkhead with installations

(5)

grant agreement n.233980. The authors are grateful to the BESST, WP06 consortium and to the BESST steering committee for allowing the publication of this information in the form of an SP Report. The BESST WP06 consortium included the following partners:

- Center of Maritime Technologies - Chalmers - DAMEN - DNV-GL - Flensburger Schiffbau-Gesellschaft - Kockums - Meyer Werft - Rhebergen Composites

- SP Technical Research Institute of Sweden - Swerea SICOMP

(6)

and Classification Society requirements. Using combustible FRP composite structures instead of steel could significantly reduce weight but could also potentially increase fire risks. This report documents a number of fire tests performed with the objective to quantify and ensure safety when lightweight FRP composite are used as ship structures. Two test series were carried out with focus on reaction to fire performance of external combustible FRP surfaces. Five test series consisted in fire resistance to evaluate FRP the structural fire integrity of different FRP composite structures. The large-scale tests documented in this report are:

− Fire growth tests of protected external FRP − Balcony sprinkler tests with external FRP

− Fire resistance tests of loadbearing FRP composite bulkheads with varying loads − Fire resistance test of loadbearing double FRP composite bulkhead

(non-insulated)

− Fire resistance test of loadbearing FRP composite bulkhead with steel joint − Fire resistance test of non-combustible sandwich bulkhead with installations − Fire resistance test of novel loadbearing FRP composite deck

Three feasible safety measures were found suitable to protect external FRP surfaces: • A drencher with a design discharge density of at least 3 mm/min;

• A fire protective coating called the LEO system; and • A balcony sprinkler certified according to MSC/Circ.1268.

A tested non-loadbearing sandwich panel bulkhead concept from the building sector was found to have future potential for deck house structures. To conservatively evaluate fire resistance of an insulated FRP composite bulkhead concept, the bulkhead designed for the highest applicable load level should be tested with a realistic design load. The time to failure for loadbearing FRP composite bulkheads was found quite in-sensitive to the design load. Structural redundancy concepts, like loadbearing double FRP composite bulkheads, can provide sufficient fire resistance when traditional insulation is not applicable (e.g. for external surfaces).

(7)

and Classification Society requirements. Using combustible FRP composite structures instead of steel could significantly reduce weight but could also potentially increase fire risks and is a deviation from prescriptive fire safety requirements. Alternative solutions are although allowed if they can be demonstrated to provide at least the same degree of fire safety, in accordance with SOLAS chapter II-2, regulation 17 [1]. Engineering analyses were therefore performed to find sufficiently safe designs and arrangements for the application cases in the BESST project [e.g. 2]. The analyses were based on statistics, regulation investigations, simulations and not least fire tests. Several fire tests were performed within the BESST project in order to bridge knowledge gaps and quantify safety in the engineering analyses. This report documents the large-scale fire tests performed in WP06:

− Fire growth tests of protected external FRP − Balcony sprinkler tests with external FRP

− Fire resistance tests of loadbearing FRP composite bulkheads with varying loads − Fire resistance test of loadbearing double FRP composite bulkhead

(non-insulated)

− Fire resistance test of loadbearing FRP composite bulkhead with steel joint − Fire resistance test of non-combustible sandwich bulkhead with installations − Fire resistance test of novel loadbearing FRP composite deck

(8)

shipbuilding industry’s competitive advantage on the global market with the keytargets competitiveness, environmentally friendliness and safety. The main focus of the project was holistic life cycle performance assessment on ship level, which is meant to guide the technical developments on system level. The results were then integrated in virtual show cases (ship concepts) demonstrating the technical solutions as well as the life cycle impact compared to current designs of passenger ships, ferries and mega-yachts, even if the results to a large extent will be applicable also to other ships.

Load-bearing structures on large ships are traditionally built in steel, which is the most cost-efficient shipbuilding material in the construction phase. Life cycle cost assessments have although shown that shipping companies can increase profits by investing in a lightweight ship design, since the lower fuel consumption per ton-km payload may make additional manufacturing costs pay off in short time of operation [1]. Furthermore, environmental life cycle assessments have shown that usage of fossil fuel has the greatest impact to surroundings throughout the ship life cycle, which could be lowered by a lightweight ship design [1].

Fibre reinforced polymer (FRP) composite consists of a polymer matrix reinforced with fibres. This forms laminates which can be used together with core material to make up lightweight sandwich panels and stiffeners, often referred to as FRP composite structures. These structures can be used to achieve lightweight ship design, which was the objective in WP06 of the BESST project. The main introduced difference in fire safety is that the material is combustible, as opposed to steel which by definition is non-combustible. This report documents a number of fire tests performed with the objective to quantify and ensure safety when lightweight FRP composite are used as ship structures. The particular questions raised and the technical approaches for addressing these questions are described in further detail subsequently.

2.1 Problem definition

In the progress of conducting engineering analyses according to SOLAS II-2/17 [1], a number of questions were identified:

1. What water discharge density is necessary for a drencher system to protect external FRP surfaces?

2. Is it possible to use passive fire protection for external FRP surfaces?

3. Is a certified balcony sprinkler sufficient to prevent fire spread from a cabin to external FRP surfaces on a cruise ship?

4. Are results from tests with 7 kN/m loading valid for 5-10 times higher load levels?

5. Will the novel loadbearing bulkhead and deck designs developed in WP06 of BESST perform satisfactory fire resistance?

2.2 Technical approach

To address questions 1 and 2 in section 2.1 Problem definition, a series of six fire tests was conducted studying fire growth on vertical FRP composite panels, as documented by Evegren, Rahm [3]. A drencher system was used in four tests to determine the flow rate required to extinguish and prevent fire growth on the FRP composite panels. In addition, one test was performed with a fire protective coating and one test was performed with a

(9)

To address question 3 in section 2.1 Problem definition, a series of four fire tests was conducted studying how effective a balcony sprinkler is to control fire on a balcony with FRP surfaces. The sprinkler was certified to control a balcony fire when furniture and furnishings other than those of restricted fire risk are used, in accordance with

MSC/Circ.1268 [5]. The tests were carried out based on this procedure but with some modifications to cover the two main risks associated with external FRP surfaces: the added fire load and the potential for fire spread from a cabin, via the balcony, to external surfaces. In the tests, the fire load was therefore increased by adding combustible FRP composite surfaces on the balcony mock-up. Furthermore, an FRP composite panel was fitted above the balcony, to determine whether fire spread occurs.

To address question 4 in section 2.1 Problem definition, a series of five fire resistance tests was conducted studying time until collapse for loadbearing FRP composite bulkheads. The tests were performed with different loads and with different bulkhead scantlings, resulting in varying safety factors against buckling. This was done to determine whether fire resistance is sensitive to the design load, design methods and safety factors against buckling.

With regards to question 5 in section 2.1 Problem definition, four novel designs were developed in WP06:

• A double FRP composite bulkhead, achieving fire resistance without insulation. • An FRP composite bulkhead joined to a steel bulkhead.

• A steel/mineral wool/steel sandwich panel concept with outfitting, developed for deck houses.

• An FRP composite deck structure concept, achieving stiffness criteria relevant for cruise ships.

All of the novel designs were tested with the ambition to reach 60 minutes of fire resistance (structural fire integrity).

(10)

the first stages of a fire. The focus in these tests was external combustible FRP surfaces, which pose a new fire hazard in comparison with conventional steel structures. The other test series consisted in “fire resistance” tests, i.e. evaluations of the FRP composite structures’ abilities to withstand long-lasting fire exposure. Each of the test series is described below, with the used method and materials, the test results and conclusions.

3.1 Fire growth tests of protected external FRP

A series of six fire tests was conducted to study fire growth on vertical FRP composite panels. This has already been documented in summary by Evegren, Rahm [3]. However, the complete test documentation is included in this report as Annex A. Fire growth tests of protected external FRP and summarized below.

3.1.1 Method

The fire tests were conducted based on SP FIRE 105 [4]. The method specifies a

procedure to determine reaction to fire properties of external wall assemblies and facade claddings. A fire source is used which exposes the wall materials to a simulated

apartment fire with flames emerging through a large window opening. The test rig consists of a lightweight concrete wall, 4 000 mm (W) by 6 000 mm (H) and a thickness of 150 mm. At the bottom edge of the wall there is a fire compartment with a 3 000 mm (W) by 710 mm (H) front opening. The compartment has a horizontal air intake opening at the floor, close to the back wall, that measures 3 150 mm (W) by 300 mm (B). The fire source consists of two fire trays, positioned adjacent to each other, filled with a total of 60 litres of heptane. Each fire tray measures 1 000 mm (L) by 500 mm (W) by 100 mm (H). A flame suppressing lattice is installed on top of the trays and after each tray has been filled with 30 litres of heptane, water is added such that the fuel level is touching the underside of the lattice.

In the test series, three different panels were used:

1. Standard composite panels (sandwich panels with vinyl ester/glass fibre laminates surrounding a PVC foam core).

2. Non-combustible boards (calcium silica boards by Promatect®)

3. Composite panels with fire protective coating (laminates with improved reaction to fire performance properties by the LEO® system).

The non-combustible Promatect® boards were used for a reference test, to determine the fire source and represent a non-combustible (steel) external ship surface. The boards and different composite panels were used together with a drencher system in a series of six tests, described in Table 1.

Table 1. Program for fire growth tests of protected external FRP

Test Type of panels Design discharge density [mm/min]

1 Standard composite panels 3

4 Promatect® boards No water application

5 Standard composite panels 3*

(11)

minutes. The fire growth rate was thereafter very rapid. The application of water suppressed the fire in the panels. However, the temperature exposure at the bottommost panel was high despite the flowing water. Post-fire damages were severe, but it is likely that the primary damages occurred prior the start of water application. This suggests that a premature application of water, i.e. initiated application of water prior to fire exposure of the panels, is essential for the protection of untreated standard panels. This was investigated in one of the tests and it could be concluded that a premature application of water prevented fire ignition and fire involvement of a large portion of the panels. Based on this test it could be concluded that a design discharge density of 3 mm/min

(corresponding in this particular case to an actual density of 1.77 mm/min) is sufficient to prevent fire ignition and fire involvement of untreated composite panels. The results also revealed that it is important that water is applied uniformly.

A practical technique for applying a uniform and sufficiently high discharge density of water needs to be developed for actual installations. It is imperative that a water film covers all vertical surface areas exposed to fire in order to achieve proper fire protection. This will not only put high demands on the nozzle design and the nozzle installation guidelines but also on the design of the outer surfaces of a superstructure or a hull, e.g. that it allows an unobstructed flow of water. It appears that thin obstructions (e.g. the thin steel plates for the measurement of the surface temperatures used in the tests) can

influence the formation of a water film.

The LEO system delayed fire ignition and fire involvement significantly compared to the standard untreated panels. Fire ignition occurred after approximately nine minutes (on a relative time-adjusted scale) whereas the standard untreated panels ignited after between approximately 30 seconds and 1 minute. Once burning, the energy contribution from the LEO system panels was about 10% higher compared to the non-combustible boards. However, the fire did not escalate in the way experienced with the standard untreated panels. The heat release rate decreased in relation to the reduced fire source heat release rate and the remaining fire in the composite panel self-extinguished a few minutes after the fuel in the fire trays had burnt out. The fire post-fire damages were excessive, with relatively severe damages to the core. Hence, even though the reaction to fire properties by the LEO system were satisfying, measures must be taken to ensure that a local collapse is avoided.

3.1.3 Conclusion

In conclusion, for standard FRP composite panels used as external ship surfaces: • exposure to a large fire can cause ignition within minutes and a following rapid

fire growth;

• a drencher system with a discharge density of 3 mm/min is sufficient to extinguish fire established in the material (during continued fire exposure); • a pre-activated drencher system with a discharge density of 3 mm/min is

sufficient to prevent ignition and fire involvement;

• considering the short time until ignition, pre-activation of a drencher system must be swift; and

• passive fire protection solutions can be used to prevent ignition and significantly reduce fire involvement.

(12)

3.2 Balcony sprinkler tests with external FRP

A series of four fire tests was conducted to study how effective a balcony sprinkler is to control fire on a balcony with FRP surfaces. The complete test documentation is included in this report as Annex B. Balcony sprinkler tests with external FRP and summarized below.

3.2.1 Method

The fire tests were conducted based on MSC/Circ.1268 [5]. A balcony structure was built according to MSC/Circ.1268 [5] with the general dimensions 3 m x 2 m x 2.5 m (with x depth x height). In the test procedures, the main fuel source is represented by combustible chairs, simulated by polyether foam fitted on steel frames. A wood crib above a heptane fuel pan is also placed under a steel table in the corner and used as ignition source. The tests were carried out to determine if an approved balcony sprinkler certified can (1) control a balcony fire with the added fire load from FRP composite surfaces and (2) prevent fire spread to external FRP surfaces above the balcony. Two experimental setups were used to represent these scenarios:

1. Balcony fire

2. Cabin fire (spreading through balcony).

In both setups, a 0.5 m high FRP panel was mounted above the balcony and used as a target to indicate fire spread to external surfaces. Furthermore, FRP composite panels were mounted on the balcony surfaces to simulate the added fire load from external FRP composite. In the balcony fire scenario, two FRP composite panels, with the dimensions 2 m x 2 m, were mounted on the two walls by the ignition. In the cabin fire scenario, a cabin with the dimensions 4.3 m x 3 m x 2.4 m (L x W x H) was built adjacent to the balcony and a 2 m x 1.8 m (H x W) opening connected the cabin to the balcony. The surfaces surrounding the opening and under the air duct were covered with FRP composite panels.

In the balcony fire scenario, a fire source was provided in accordance with the procedures (with the potential addition from combustible FRP composite balcony surfaces). In the cabin fire scenario the fire source consisted of two fuel trays positioned adjacent to each other and filled with a total of 60 litters of heptane. Each fuel tray measured 1000 mm (L) by 500 mm (W) by 100 mm (H). A flame suppressing lattice was installed on top of the tray and after each tray had been filled with 30 litres of heptane, water was added such that the fuel level was touching the underside of the lattice.

A fan and a duct were installed in line with MSC/Circ.1268 [5] to generate a side wind of 5 m/s. Each scenario was performed with an without side wind, making up four scenarios, summarized in Table 2.

Table 2. Program for balcony sprinkler tests with external FRP

Test Fire Scenario Side wind

1 Balcony fire - 2 Balcony fire 5 m/s

3 Cabin fire -

(13)

prevented the fire from spreading to the target panel. Furthermore, the balcony sprinkler prevented flames and hot smoke from a flashover cabin fire to ignite the target FRP composite panel, simulating external FRP structures above the balcony. Side wind delayed sprinkler activation but it also cooled the gases exposing the target panel. It is therefore likely that a higher wind speed than 5 m/s would not increase the risk of fire spread to external surfaces.

3.2.3 Conclusion

In conclusion, a certified balcony sprinkler can be used to:

• prevent fire involvement of FRP surfaces in a balcony fire • prevent fire spread to other external FRP surfaces.

A certified balcony sprinkler is thereby an effective safety measure to manage fire risks associated with FRP composite used for balcony structures.

The full test report from these tests is found in Annex B. Balcony sprinkler tests with external FRP.

3.3 Fire resistance tests of loadbearing FRP

composite bulkheads

A series of five loaded fire resistance tests was conducted to investigate how the

structural fire integrity of a loadbearing FRP composite bulkhead depends on the design method, design load and safety factor against buckling. The complete documentation from the tests is included in this report as:

• Annex C. Fire resistance test of loadbearing FRP composite bulkhead with 12.4 kN design load;

• Annex D. Fire resistance test of loadbearing FRP composite bulkhead with 20.7 kN design load and reduced safety factor;

• Annex E. Fire resistance test of loadbearing FRP composite bulkhead with 7 kN design load;

• Annex F. Fire resistance test of loadbearing thick FRP composite bulkhead with 39,2 kN design load; and

• Annex G. Fire resistance test of loadbearing stiffened FRP composite bulkhead with 31 kN design load.

The tests are summarized below.

3.3.1 Method

The fire resistance tests were conducted based on the test procedure for fire-resisting divisions (FRD) of high-speed craft (HSC) [6], as described in Part 11 in the FTP Code [7]. Structural fire integrity is tested by mounting the bulkhead vertically on a furnace with controlled heat exposure [7]. The temperature in the furnace increases over time in a well-defined manner, in accordance with the standard time-temperature curve [8] illustrated in Figure 1. A condition evaluated in the test is the specimen insulation property, i.e. its ability to withstand heat while keeping the temperature low at the unexposed side of the specimen. The average temperature increase should be below 140°C for a time depending on the type of test and classification. An FRD30 division

(14)

Figure 1. The standard temperature-time curve (blue) used in fire resistance testing for HSC [8]. For SOLAS ships in general it is not included in the test procedures to evaluate the divisions’ load-bearing capabilities [9]. This property is defined by the prescribed scantlings of the steel divisions. The test for HSC is almost identical, with the exception that a load is applied to the specimen during the fire test. This condition was added to account for deterioration effects when using divisions of other materials than steel, which is allowed on HSC [10]. Hence, bulkheads for HSC shall withstand the standard fire test while uniformly subjected to in-plane loading, applied along the top edge of the vertical specimen, 7.0 kN/m of the specimen width [7]. A bulkhead structure must thus sustain the specified loading and exposure to fire for 30 or 60 minutes in order to be certified as an FRD30 or FRD60 division, respectively.

In order to investigate how the fire resistance of a loadbearing FRP composite bulkhead depends on the design method, design load and safety factor against buckling, a series of fire tests was arranged. The test program is described in Table 3.

Table 3. FRP composite structures and results from fire resistance bulkhead tests [6] Test 1 Test 2 Test 3 Test 4 Test 5,

stiffened tcore [mm] 50 50 37 50 50 Core quality H80 H80 H80 H80 H80 tlaminate [mm] 1,3 1,3 1,3 3,9 1,3 Pcritical [kN/m] 31 31 17,4 98 78 Pdesign [kN/m] 12,4 12,4 7 39,2 31 Ptest [kN/m] 12,4 20.7 7 39,2 31 Safety factor against buckling 2,5 1,5 2,5 2,5 2,5

The starting point of the test series was Test 1, where a typical FRP composite sandwich panel was used, consisting of 1.3 mm glass fiber reinforced polyester laminates

surrounding a cross linked PVC foam core called Divinycell H80 (80 kg/m3_{). The}

philosophy was to use a typical quite weak bulkhead as starting point in order to provide 0 200 400 600 800 1000 0 20 40 60 80 100 Te mp era tu re [° C] Time [min] Standard temperature-… Fire exposure 0 200 400 600 800 1000 1200 1400 0 30 60 90 120 150 180 210 240 270 300 Time (minutes) T em per at ur e ( C ) X2000 HC-curve Standard curve

(15)

with the standardized test for HSC [7], i.e. 7 kN/m. The safety factor against buckling was 2.5 in Test 3, as conventional. In Test 4 the laminate thickness was instead tripled and tested with a safety factor against buckling of 2.5. In Test 5 the original sandwich panel (used in Test 1) was stiffened and tested with a safety factor against buckling of 2.5.

3.3.2 Results

Firstly it can be noted in Table 4 that 60 minutes of fire resistance was not achieved in most of the tests, even though the FRP composite panel was previously certified. This was a consequence of an alteration in the IMO test procedures. The used FRD-60 structure was certified before the 2010 edition of the FTP Code was ratified. This harmonized the test procedure between laboratories and gave a slightly tougher temperature development at SP Fire Research than the test in which the panel was certified.

Table 4. FRP composite structures and results from fire resistance bulkhead tests [6] Test 1 Test 2 Test 3 Test 4 Test 5,

stiffened Time to failure [min] 56 53 58.5 51 55 T laminate at failure

[°C] 260 206 255 161 223

Temperature between exposed laminate and core at failure [°C]

148 149 154 87 136

The relative comparison is still valid and showed that the time to failure did not differ much in the tests with single sandwich structures, varying between 51 and 58.5 minutes. Changing the safety factor from 2.5 to 1.5 resulted in a relatively small difference in time to failure of 3 minutes. The stiffened test showed that structural resistance is better achieved by use of stiffeners than by thick laminates. Furthermore, applying this as a design principle and using a safety factor of 2.5 gives a test variation between 55 and 58.5 minutes.

The temperature at the exposed laminate-core interface was quite similar at test failure. This excepts the test when the laminate thickness was increased as a measure for structural improvement.

3.3.3 Conclusion

In conclusion, the test series shows that fire resistance bulkhead testing of insulated FRP composite panels can be simplified and does not have to be performed with varying design loads. A conservative way to test an insulated FRP composite bulkhead concept is not to test a weak panel at 7 kN/m (in accordance with the standard test procedure [7]) but to test the bulkhead designed for the highest applicable load level. Furthermore, the following design rules could be established for non-stiffened solutions:

- Heavy loadbearing bulkheads should be constructed with stiffeners rather than with thick laminates; and

(16)

A loaded fire resistance tests was conducted to investigate the structural fire integrity of a loadbearing double FRP composite bulkhead developed by Kockums AB within the scope of WP06. The philosophy was a structural redundancy concept where half of the structure was sufficient to carry the design load. The idea was to thereby provide passive fire resistance for a certain time without traditional fire insulation, by sacrificing the exposed half of the structure. The double sandwich panel consisted of three laminates (1.3 mm vinyl ester/glass fibre laminates) and two cores (50 mm end grain balsa wood), of which two laminates and one core were designed to carry the design load. The starting point was hence the same structure as in Test 1 in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads above, but with balsa wood instead of PVC core. In addition, a fire protective coating (the LEO® system, by Saertex) was used for the exposed laminate in the test. The design would thus allow the exposed LEO laminate and a 50 mm balsa core to burn off while the unaffected parts of the structure has sufficient strength to prevent collapse.

3.4.1 Method

To verify the fire resistance of the double FRP composite bulkhead, it was tested based on the same test procedure as above in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads. This is the test procedure for fire-resisting divisions (FRD) on high-speed craft (HSC) [6], described in Part 11 in the FTP Code [7]. However, the applied load in the test was not 7 kN/m in accordance with the code, but 12.4 kN/m, to represent the design load of the structure (with a safety factor against buckling of 2.5).

3.4.2 Results

The loadbearing double FRP composite bulkhead maintained loadbearing capacity, fire integrity and acceptable temperature levels on the unexposed side for 90.5 minutes of testing. The centre laminate temperature was then 453°C and at the interface between the middle laminate and the unexposed core the temperature was 425°C.

3.4.3 Conclusion

The loadbearing double FRP composite bulkhead passed the 60 minute fire resistance test (FRD-60) with a large margin and provides possibilities for achieving structural integrity with uninsulated lightweight structures.

The full test report from this test is found in Annex H. Fire resistance test of loadbearing double FRP composite bulkhead (non-insulated).

3.5 Fire resistance test of loadbearing FRP composite

bulkhead with steel joint

A loaded fire resistance tests was conducted to investigate the structural fire integrity of a loadbearing FRP composite bulkhead with a steel joint, developed by Kockums AB within the scope of WP06. The philosophy was to develop a joint between the FRP composite and a steel bulkhead to provide for joining of a composite superstructure to a steel hull in a steel shipyard environment. The joint is a so called crutch joint where the steel bulkhead plate is ended with an “open four” (4) profile, a crutch. The FRP composite panel is fitted and glued in this crutch, as illustrated in Figure 2.

(17)

Figure 2. Illustrations of the crutch joint between steel and FRP composite, with its location, actual appearance and technical description with insulation [11].

The starting point for the FRP composite structure was the same as in Test 1 in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads above, i.e. 1.3 mm vinyl ester/glass fibre laminates surrounding a 50 mm PVC core. The steel profile had a nominal thickness of 5 mm. The whole structure was insulated on the inside to provide 60 minutes of fire resistance with the used FRP composite structure. More information on the joint and its mechanical properties are found in [11].

3.5.1 Method

To verify the fire resistance of the FRP composite bulkhead with a steel joint, it was tested based on the same test procedure as above in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads. This is the test procedure for fire-resisting divisions (FRD) on high-speed craft (HSC) [6], described in Part 11 in the FTP Code [7]. However, the applied load in the test was not 7 kN/m in accordance with the code, but 12.4 kN/m, to represent the design load of the FRP composite structure (with a safety factor against buckling of 2.5).

3.5.2 Results

The loadbearing FRP composite bulkhead with a steel joint failed in loadbearing capacity after 49 minutes of testing. The integrity was maintained for 49 minutes of the test and the temperatures on the unexposed side did not exceed acceptable levels during 49 minutes of the test.

3.5.3 Conclusion

The loadbearing FRP composite bulkhead with a steel joint did not pass the 60 minute fire resistance (FRD-60) test due to insufficient insulation.

The full test report from this test is found in Annex I. Fire resistance test of loadbearing FRP composite bulkhead with steel joint.

3.6 Fire resistance test of sandwich panel bulkhead

with installations

A loaded fire resistance tests was conducted to investigate the structural fire integrity of a sandwich panel bulkhead with installations, developed by FSG within the scope of WP06. The non-loadbearing sandwich panel concept was based on light weight steel-mineral wool sandwich panels currently available in the building sector. The panelling system

(18)

To verify the fire resistance of the sandwich panel bulkhead, it was tested based on the test procedure for A-class divisions on SOLAS vessels [9], as described in Part 3 of the FTP Code [7]. This procedure differs from that for fire-resisting divisions (FRD) only in the way that the tests are not loaded, as further discussed in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads.

3.6.2 Results

The sandwich panel bulkhead maintained integrity during the test but the temperatures on the unexposed side exceeded 180°C after 38 minutes of the test. Pipe and ventilation penetrations maintained integrity and temperatures on the unexposed side did not exceed acceptable levels during the test. The fire integrity of the door was maintained for

49 minutes of the test and during this time the temperatures on the unexposed side did not exceed acceptable levels.

3.6.3 Conclusion

The sandwich panel bulkhead did not pass the 60 minute fire resistance test (A-60) due to insufficient insulation. Of the installed maritime outfitting (certified for steel bulkheads), the pipe and ventilation penetrations passed the 60 minute fire resistance test but the fire door (A-60) failed in fire integrity.

The full test report from this test is found in Annex J. Fire resistance test of sandwich panel bulkhead with installations.

3.7 Fire resistance test of novel loadbearing FRP

composite deck

A loaded fire resistance tests was conducted to investigate the structural fire integrity of a novel loadbearing FRP composite deck, developed by Kockums AB within the scope of WP06. The deck structure concept was designed to get sufficient width between load-bearing elements with minimized deck height (including stiffeners) whilst achieving stiffness criteria relevant for cruise ships. To achieve this the structure was designed with an FRP composite sandwich with an unconventionally thick core (200 mm), fitted between 700 mm high transversal steel beams. The beams also work as stiffeners and would be supported by pillars in a ship design. The FRP composite panels were further supported by steel rods at the beams and the whole structure was insulated, as illustrated in Figure 3.

(19)

Figure 3. Insulated novel loadbearing FRP composite deck solution.

3.7.1 Method

To verify the fire resistance of the novel loadbearing FRP composite deck, it was tested based on the test procedure for decks corresponding to that described for bulkheads above in section 3.3 Fire resistance tests of loadbearing FRP composite bulkheads. This is a test procedure for fire-resisting divisions (FRD) on high-speed craft (HSC) [6], described in Part 11 in the FTP Code [7]. Structural fire integrity is tested by mounting the deck horizontally on a furnace with controlled heat exposure [7]. As for bulkheads, the temperature in the furnace increases over time in accordance with the standard time-temperature curve [8], illustrated in Figure 1. However, instead of applying a load of 7 kN/m as in the bulkhead tests, it is prescribed to apply a universal transversal load corresponding to 3.5 kN/m2_{of the deck area [7].}

3.7.2 Results

The novel loadbearing FRP composite deck failed in loadbearing capacity after 47 minutes of testing. The integrity was maintained for 47 minutes of the test and the

temperatures on the unexposed side did not exceed acceptable levels during 47 minutes of the test.

3.7.3 Conclusion

The novel loadbearing FRP composite deck failed did not pass the 60 minute fire resistance test (A-60) due to loss of loadbearing capacity (insufficient insulation). The full test report from this test is found in Annex K. Fire resistance test of novel loadbearing FRP composite deck.

(20)

fire resistance testing. This means that further development, e.g. added insulation, is necessary if 60 minutes of fire resistance is the be achieved.

Three feasible safety measures were found suitable to protect external FRP surfaces: • A drencher with a design discharge density of at least 3 mm/min, preferably in

combination with a possibility for pre-activation upon cabin fire; this could prevent fire involvement of untreated composite panels;

• The LEO system delayed fire ignition and fire involvement of the FRP surface and the energy contribution from panels was limited; and

• A balcony sprinkler certified according to MSC/Circ.1268 controlled a balcony fire and prevented fire spread from a cabin to external FRP surfaces, despite exposed FRP balcony surfaces.

The non-loadbearing sandwich panel bulkhead concept from the building sector could be a safe alternative to traditional deck house design after some modifications. An improved method for installing an A-60 door must be developed and validated.

To conservatively evaluate fire resistance of an insulated FRP composite bulkhead concept, the bulkhead designed for the highest applicable load level should be tested with a realistic design load. The time to failure for loadbearing FRP composite bulkheads does not seem to be very sensitive to the design load. The following design rules seem to reduce this sensitivity even further:

- heavily loaded bulkheads should be constructed with stiffeners rather than with thick laminates; and

- a safety factor against buckling of 2.5 should be used.

Structural redundancy concepts, like the loadbearing double FRP composite bulkhead, can provide sufficient fire resistance when traditional insulation is not applicable (e.g. for external surfaces).

5 References

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

2. Evegren, F., Engineering analysis report – Norwegian Future. 2013, SP Technical Research Institute of Sweden: Borås.

3. Evegren, F., et al. Fire testing of external combustible ship surfaces. in 11th International Symposium on Fire Safety Science. 2014. Christchurch, NZ: IAFSS. 4. SP Fire Technology, External wall assemblies and facade claddings: Reaction to

fire, in SP FIRE 105. 1985, SP Technical Research Institute of Sweden - Fire Research.

5. IMO, Guidelines for the approval of fixed pressure spraying and water-based fire-extinguishing systems for cabin balconies, in MSC.1/Circ.1268. 2008, International Maritime Organization: London.

6. IMO, Test Procedures for Fire-resisting Divisions of High Speed Craft. 1995, London: International Maritime Organization.

7. IMO, FTP Code: International Code for Application of Fire Test Procedures, 2010. 2012 ed. 2012, London: International Maritime Organization.

(21)

International Maritime Organization.

11. Lantz, V., Joining of FRP Sandwich and Steel, in SP Technical Research Institute of Sweden, S.T.R.I.o. Sweden, Editor. 2011, SP Technical Research Institute of Sweden: Borås.

(22)

(23)

Fire growth tests of protected external FRP

(3 appendices)

SP Technical Research Institute of Sweden

Postal address Office location Phone / Fax / E-mail This document may not be reproduced other than in full, except with the

prior written approval of SP. SP Box 857 SE-501 15 BORÅS Sweden Västeråsen Brinellgatan 4 SE-504 62 BORÅS +46 10 516 50 00 +46 33 13 55 02 info@sp.se

Introduction

The department of Fire Technology at SP Technical Research Institute of Sweden has, as a part of the FP7 project BESST, conducted a series of six fire tests with the objective of studying fire spread on vertical composite panels and the water application flow rates required to prevent fire spread or to protect the panels from burning.

The test rig

The fire tests were conducted basically in accordance with SP FIRE 105 [1]. The method specifies a procedure to determine the reaction to fire of materials and construction of external wall assemblies or facade claddings, when exposed to fire from a simulated apartment fire with flames emerging out through a window opening. The test rig consists of a lightweight concrete wall, 4000 mm (W) by 6000 mm (H) having a thickness of 150 mm. At the bottom edge of the wall there is a fire compartment with a 3000 mm (W) by 710 mm (H) front opening. The compartment has a horizontal air intake opening that measure 3150 mm (W) by 300 mm (B) at the floor, close to its back wall.

The fire source consists of two fire trays positioned adjacent to each other that are filled with a total of 60 liters of heptane. Each fire tray measures 1000 mm (L) by 500 mm (W) by

100 mm (H). A flame suppressing lattice is installed on top of the tray edges and after each tray has been filled with 30 liters of heptane, water is added such that the fuel level is touching the underside of the lattice. Figure 1 shows the fire trays inside the fire compartment.

1 SP FIRE 105, “EXTERNAL WALL ASSEMBLIES AND FACADE CLADDINGS REACTION TO FIRE”, Approved 1985-07-18, Edition 5, dated 1994-09-09

(24)

Figure 1 The fire trays inside the fire compartment.

The test specimens (see description below) were mounted on the outside of the lightweight concrete substrate structure using horizontal steel (“hat”) profiles, such that there was a gap between the backside of the specimens and the lightweight concrete of approximately 50 mm. In order to prevent fire spreading on the backside of the specimens, a nominally 10 mm thick non-combustible Promatect® board covered the edge of the concrete wall, the gap and the edge of the test specimens.

The composite panels

Two types of composite panels provided by Kockums AB were tested:

 Untreated standard panels.

 LEO system panels.

The standard panels had a sandwich structure with glass fibre reinforced polyester face laminates and a cross linked PVC foam core. The thickness of the laminate was 1.7 mm and the thickness of the core 50 mm. The density of the core was 80 kg/m3_{. The sandwich panels}

were vacuum infused in one shot with resin distribution by grooves in the core. The fibre fraction was about 50% by volume and about 1500 g/m2_{of resin was absorbed in the surface}

of the core including the resin in the grooves. This type of panels was used in Tests 1, 2, 3 and 5. Figure 2 shows the test set-up prior depictured prior Test 5.

(25)

Figure 2 The test set-up when using the untreated standard composite panels, depictured prior Test 5.

The LEO system consists of a special glass fibre reinforcement, infusion resin and top coat. The sandwich panels were manufactured in the same way as the standard panels resulting in the same thicknesses, fibre fractions and resin absorption. The top coat was applied manually by roller and was about 1 mm thick and was weighing about 1.3 kg/m2_{. The infusion resin}

behaves like standard polyester but with a slightly higher viscosity than normal infusion resins. The top coat was heavily loaded with fire retarding additives and was very thick, which made it difficult to achieve uniform layer thickness and good surface finish. This type of panel was used in Test 6.

Three panels, independent of the type described above, were used in each test. The individual panel measured 2150 mm (H) by 4000 mm (W), making the overall size 6450 mm (H) by 4000 mm (W). The panels were attached to the lightweight concrete substrate structure using horizontal (“hat”) steel profiles) by screws. The horizontal steel profiles were installed at the bottom of the bottommost panel, at the top of the topmost panel and at the intersection between the individual panels. A drawing of the test set-up is provided in Appendix 1. In order to handle the test set-up with a transverse crane, a piece measuring 500 mm (W) by 350 mm (H) was cut out around the vertical centreline of the top of the topmost panel. This had no or insignificant influence on the test results.

(26)

The Promatect® boards

Test 4 comprised a free-burn fire test with non-combustible, nominally 10 mm thick Promatect® boards instead of composite panels. The intent of the test was to provide benchmark data for the comparison of the heat release rate, the gas temperatures and the surface temperatures. Each board measured 3000 mm (H) by 1250 mm (W), i.e. the overall area was 6000 mm (H) by 3750 mm (W). The boards were attached to the lightweight concrete substrate structure in a similar fashion as the composite panels. The pipe with the flat spray nozzles was not installed in this particular test. Figure 2 shows the test set-up for this particular test and a drawing of the test set-up is provided in Appendix 2.

Figure 3 The test set-up when using non-combustible Promatect® boards, depictured prior the free-burn fire test in Test 4.

(27)

Instrumentation and measurements

The surface temperatures along the centreline of the outside face of the panels were measured using nominally 0.7 mm thick, 150 mm by 150 mm Inconel steel plates. A wire thermocouple was spot-welded to the backside of the steel plate. The steel plates covered a Ø=100 mm circular hole drilled through the panels and were fastened flush with the surface with screws. The gap between the plate and the panel was filled with heat resistant silicone. The backside of the plate covering the circular hole was insulated with Thermal Ceramics Kaowool ceramic blanket insulation and the opening at the backside of the panel was secured with duct tape. Figure 4 shows the front face of the steel plate.

Figure 4 The front face of one of the steel plates that was used to measure the surface temperature along the centreline of the outside face of the panels.

The gas temperature along the centreline of the panels was measured using Ø=0.50 mm, sheathed, type K thermocouples. The bead of each thermocouple was positioned 50 mm from the front surface of the panels and 50 mm offset the surface temperature measurement steel plates.

In total, six surface temperature measurement steel plates and six gas measurement thermocouples were installed. The bottommost position equalled the quarter-height of the bottommost composite panel i.e. 2150 mm divided by 4 = 537 mm. The individual distance between the other measurement positions equalled half-height the panel, i.e. 2150 mm divided by 2 = 1075 mm.

For the test with the Promatect® boards, the Inconel steel plates were installed at the same position as used in the tests with the composite panels. The positions of the sheathed thermocouples were also identical.

(28)

Table 1 shows the surface and gas temperature measurement points and the associated channels. All measurement points are also shown in the drawings of Appendices 1 and 2. The total water flow rate (measurement channel C33) was measured with a Krohne 0 – 200 L/min water flow meter and the water pressure (measurement channel C35) at the end of the pipe was measured using a 0 – 16 bar pressure transducer.

Table 1 The surface and gas temperature measurement points and the associated channels.

Channel Description Channel Description Position

C21 Surface temp. C27 Gas temperature 537 mm above bottom edge C22 Surface temp. C28 Gas temperature 1612 mm above bottom edge C23 Surface temp. C29 Gas temperature 2687 mm above bottom edge C24 Surface temp. C30 Gas temperature 3762 mm above bottom edge C25 Surface temp. C31 Gas temperature 4837 mm above bottom edge C26 Surface temp. C32 Gas temperature 5912 mm above bottom edge The tests were conducted under the Industrial Calorimeter, a large hood connected to an evacuation system capable of collecting all the combustion gases produced by the fire. In the duct connecting the hood to the evacuation system, measurements of gas temperature, velocity and the generation of gaseous species such as CO2 and CO and depletion of O2, can be made.

Based on these measurements, both the convective and the total heat release rate can be calculated. The heat release rate histories and the parameters described below were used for the evaluation of the test results.

HRRconv: The convective heat release rate measured during a test is calculated on the basis

of the gas temperature and mass flow rate in the calorimeter system. The convective fraction of the total heat release varies with the fuel and other factors, but usually approximately

two-thirds of the energy generated by a fire is released through convection. Additionally, convection has a strong impact on the velocities and temperatures within the fire plume.

HRRtot: The total heat release rate measured during a test is calculated on the basis of the

oxygen depletion of the fire, as measured in the calorimeter system. HRRtot is comprised of both the convective and radiative heat release rate, as well as the heat being conducted away and absorbed within the test set-up.

The total convective energy: The energy convected upwards is largely responsible for heating

of structures and building components above a fire, for example the underside of a deck. The total convective energy, calculated from fire ignition until the termination of a test helps to characterise fire severity; as in the case of heat transfer, duration is as important as magnitude. Some fires are very intense but short-lived and their thermal impact may be less severe than a fire of lower intensity with a longer duration. The total convective energy is an important measure of a fire’s maximum potential for causing thermal damage.

The total energy: The total energy, calculated from fire ignition until the termination of a test,

is a measure of the amount of combustibles being consumed.

The tests were documented by digital still photos and video recording.

(29)

The water application system consisted of a 1¼" (nominally 32 mm) steel pipe with outlets every 1000 mm. The pipe was fed from one end and a pressure transducer was positioned at the opposite end, i.e. at the hydraulically most remote position. The pipe was installed at the top of the set-up with two consoles that was attached directly to the topmost composite panel. Flat spray nozzles manufactured by BETE Fog Nozzle, Inc. was used to distribute water at the top of the set-up. These nozzles produce a thin, flat sheet of water spray which covers a narrow area. The spray angle of the nozzles was 145°. The nozzles were installed in a pendent

orientation, with the outlet facing the composite panels. Five nozzles were installed at a 1000 mm horizontal spacing. This resulted in a configuration with one nozzle positioned on front of the vertical centreline of the set-up. The two outermost nozzles were positioned in front of the vertical edges of the panels, i.e. the actual water flow that reached the panels were from “four” nozzles. The pipe was installed such that the water spray from the nozzles hit an imaginary, horizontal line 450 mm below the top of the set-up. The area below this line, which corresponds to the total coverage area of the nozzles, therefore equalled 6000 mm (H) by 4000 mm (W), i.e. 24 m2_{. This area was used for the calculation of the design discharge}

density. Figure 5 shows the position of the pipe with the flat spray nozzles.

Figure 5 The pipe with the flat spray nozzles depictured when setting the flow rate prior Test 3.

Table 2 shows the specific nozzles used, their K-factor, the specific design discharge density and the corresponding water flow rates. The design discharge density is expressed in mm/min which equals liter/min per m2_{. The nominal, water pressure was calculated based on the}

(30)

Table 2 The flat spray nozzles used in the tests. Note: Tests 4 and 6 were conducted without the application of water.

Test Nozzle

designation ((l/min)/barK-factor 1/2₎ _dischargeDesign

density (mm/min) Design flow rate (L/min)* Total water flow rate (L/min)** Flow rate per nozzle (L/min) Calculated water pressure (bar) 3 FF187 13.7 2 48 60 12 0.76 1 and 5 FF209 18.2 3 72 90 18 0.98 2 FF250 23.9 4 96 120 24 1.00

*) Based on the design discharge density times the flow from “four” nozzles. **) Based on the flow rate from all five nozzles.

The water that reached the vertical surface of the composite panels were quantified prior each test by the collection of water in a channel positioned below the bottom edge of the composite panels, refer to figure 6. The collected water was lead to a tray positioned on a load cell and the amount of water was weighted over a time period of at least three minutes dependent on the water flow rate.

Figure 6 The channel used to quantify the water that was sprayed on the vertical surface of the panels depictured prior Test 3.

Although it was intended that all water should flow down the vertical surface of the panels, some portion was reflected away. This water fell down as water droplets in front of the composite panels and provided cooling of the flames and the hot gases flowing up the vertical surface. Table 3 summarises the actual measured discharge density and the ratio of water reaching the vertical surface.

(31)

Table 3 The actual measured discharge density and the ratio of water reaching the vertical surface. Note: Tests 4 and 6 were conducted without the application of water. Test Design discharge density (mm/min) Design flow rate (L/min)* Total water flow rate (L/min)** Collected flow rate (L/min) Actual measured discharge density (mm/min)*** Percentage of water reaching the vertical surface (%) 3 2 48 60 31.8 1.32 66% 1 3 72 90 34.7 1.45 48% 5 3 72 90 42.5 1.77 59% 2 4 96 120 57.3 2.39 60%

*) Based on the design discharge density times the flow from “four” nozzles. **) Based on the flow rate from all five nozzles.

***) Based on the collected flow rate divided by the total coverage area, i.e. 24 m2_.

The difference in the collected amount of water between Test 1 and Test 5 reveal the problem associated with nozzle alignment and nozzle angle to obtain identical flow rates flowing down the vertical surface. For Test 1, only 48% of the design flow rate flowed down the surface, for Test 5, 59% flowed down the surface.

The fire test programme and the fire test procedures

In total, six fire tests were conducted per the following test programme, refer to Table 4.

Table 4 The fire test programme.

Test Date Type of panels Design discharge density [mm/min]

1 October 3, 2012 Standard composite panels 3 2 October 4, 2012 Standard composite panels 4 3 October 8, 2012 Standard composite panels 2

4 October 9, 2012 Promatect® boards No water application 5 October 10, 2012 Standard composite panels 3*

6 October 12, 2012 Panels with FP coating (LEO) No water application *) The application of water was started 90 seconds prior to fire ignition.

The composite panels and the temperature measurement equipment were installed with the lightweight concrete substrate laid horizontal on the floor inside the fire test hall. The consoles and pipe for the water application system was installed with the test set-up in its intended position and the desired water flow rate was thereafter adjusted.

A few minutes prior to each test, heptane and water was filled in the fire trays. The

measurement system was started and the fuel was ignited with a torch two minutes thereafter. The fire was allowed to burn until all fuel had been consumed and the measurements were terminated some minutes later to document the natural decline of surface temperatures of the steel plates.

The ventilation rate in the Industrial calorimeter was 110 000 m3_{/ throughout the tests and an}

additional ventilation rate of 50 000 m3_{/h inside the fire test hall was initiated at}

10:00 [min:sec].

The fire damages were documented with still photos with the test set-up in place.

(32)

panels were dismantled and cut along their vertical centreline in order to document the fire damages to the core.

Summary of the fire test results

Fire test chronologies, measurement graphs and photos for all tests are given in Appendix 3. The measurement graphs shows the surface and gas temperatures plotted against the total heat release rate. The gas temperature data is averaged over 60 seconds in order filter the

temperature fluctuations and to improve the readability.

Given below is a summary of the essential sequence of events along with the recorded heat release rates.

Test 1

The first test was conducted using a design discharge density of 3 mm/min. The application of water was started when it was judged that the majority of the vertical surface of the test set-up was involved in fire. This equalled a convective heat release rate of approximately 2800 kW and a total heat release rate of approximately 4400 kW. The corresponding time from fire ignition was 05:28 [min:sec]. The application of water suppressed the fire in the composite panels almost immediately. Due to the characteristics of the fire test source (refer to the discussions for Test 4), the fire reached a second peak. However, the actual involvement of the panels was marginal at this phase of the fire.

Figure 7 The convective and total heat release rate histories of Test 1.

Test 2

For the second test, the design discharge density was increased to 4 mm/min. The application of water was started at an earlier stage of the fire as compared to Test 1. The convective heat release rate was around 1100 kW and the total heat release rate was approximately 2500 kW at the start of water application. The corresponding time from fire ignition was 03:37 [min:sec].

0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 1 (2012-10-03) HRRtot [kW] HRRconv [kW] H e a t R e le a se R a te (kW ) Time (min)

(33)

The fire in the composite panels was immediately suppressed. Due to the characteristics of the fire test source (refer to the discussions for Test 4), the fire reached a second peak. However, the actual involvement of the composite panels was marginal at this phase of the fire. It was visually observed that the composite panels caught fire at an earlier stage compared to Test 1 and the initial fire growth rate was therefore faster.

Test 3

For the third test, the design discharge density was decreased to 2 mm/min and an attempt was made to start the application of water at a heat release rate similar to that of Test 1. For this test, the initial fire growth rate was very similar compared to Test 1, indicating that the composite panels were involved in the fire in a similar fashion. The convective heat release rate was around 2300 kW and the total heat release rate was approximately 4000 kW at the start of water application. The corresponding time from fire ignition was 05:10 [min:sec]. These figures were slightly lower compared to Test 1. Due to the characteristics of the fire test source (refer to the discussions for Test 4), the fire reached a second peak. However, the actual involvement of the composite panels during this particular phase of the fire was marginal.

0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 2 (2012-10-04) HRRtot [kW] HRRconv [kW] He a t Re le a se R a te ( kW ) Time (min)

(34)

Test 4

Test 4 comprised a free-burn fire test, i.e. without the application of water, with

non-combustible Promatect® boards instead of composite panels. The intent of the test was to provide benchmark data for the comparison of the heat release rate as well as the surface and gas measurements.

The fire growth rate was significantly slower compared to Tests 1 through 3 and the fire peaked at approximately 1700 kW after about eight minutes and declined to a level around 1400 kW. After approximately ten minutes, the fire developed rapidly to a peak of around 3000 KW at about 12 minutes. The rapid development can probably be explained by the fact that the heptane in the fire trays was very hot and started to boil. After this second peak, the fire gradually declined as the heptane fuel in the fire trays was consumed. The fire was more or less out 17 minutes after fire ignition.

0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 3 (2012-10-08) HRRtot [kW] HRRconv [kW] He a t Re le a se Ra te (kW ) Time (min)

(35)

Test 5

This test was similar to Test 1, except that the water flow was initiated prior (90 seconds) to fire ignition. The intent was to simulate a premature activation of an exterior deluge system and to investigate whether the design discharge density of 3 mm/min was sufficient to prevent fire ignition and fire spread in the composite panels.

The initial fire growth rate was slower compared to the free-burn fire test in Test 4, i.e. the fire in the fire trays developed slower. It is, however, unlikely the premature application of water had an influence on the initial fire growth rate as water droplets did not reach the fire trays and the droplets had probably no influence on the air flowing to the fire. The second heat release rate peak did also occur at a later stage. The convective heat release rate was significantly reduced over the entire test duration time due to the cooling of the water spray.

0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 4 (2012-10-09) HRRtot [kW] HRRconv [kW] H e a t R e le a se R a te (kW ) Time (min)

(36)

Test 6

Test 6 involved LEO system composite panels. The intent of the test was to compare the initial fire growth rate as compared to both the untreated standard composite panels and the free-burn fire test involving non-combustible Promatect® boards. The pipe with the flat spray nozzles was installed at the top of the test set-up, however, water was not applied.

0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 5 (2012-10-10) HRRtot [kW] HRRconv [kW] He a t Re le a se R a te ( kW ) Time (min) 0 1000 2000 3000 4000 5000 0 5 10 15 20 Test 6 (2012-10-12) HRRtot [kW] HRRconv [kW] He a t Re le a se R a te ( kW ) Time (min)

(37)

An analysis of the fire test results

The following section contains an analysis of the fire test results based on the heat release rate and temperature measurements.

An analysis based on the heat release rate measurements

Figure 13 and 14, respectively, show the total and convective heat release rate histories for all the tests.

Figure 13 The total heat release rate histories for all tests.

0 1000 2000 3000 4000 5000 0 5 10 15 20

Total heat release rate

Test 1 Test 2 Test 3 Test 4 Test 5 Test 6 T o ta l h e a t rel e a se r a te (kW ) Time (min)

(38)

Figure 14 The convective heat release rate histories for all tests.

Table 5 shows the calculated convective and total energy, respectively, during the most intense 10-minute part of the second heat release rate peak. This part of the tests, were the fire reached a second heat release rate peak, is the most rational for a test-to-test comparison that includes all tests, as it is independent of the time when water application was initiated (if applicable). The calculation interval for the convective and total energy is dissimilar for different tests, but always ten minutes, as the fire growth occurred at different times. The calculation interval for the convective energy as compared to the total energy may also be slightly different for a specific test, which is due to a small delay time in the measurement system. Coincidently, the total energy during Tests 1 and 3 is identical, however, it may be noticed that the actual measured discharge density (refer to Table 3) were quite similar for these two tests.

Table 5 The calculated convective and total energy, respectively, during the most intense 10-minute part of the second heat release rate peak.

Test Convective energy

[MJ] interval [sec] – Calculation [sec]

Total energy

[MJ] interval [sec] – Calculation [sec] 1 150 560 - 1160 935 583 - 1183 2 95 511 - 1111 1004 542 - 1142 3 205 528 – 1128 935 571 – 1171 4 (Promatect®) 561 600 - 1200 1077 600 - 1200 5 128 690 - 1290 1103 705 - 1305 6 (LEO) 740 637 – 1237 1258 622 - 1222 0 1000 2000 3000 4000 5000 0 5 10 15 20

Convective heat release rate

Test 1 Test 2 Test 3 Test 4 Test 5 Test 6 C o n v e c ti ve h e a t re le a se rat e ( k W) Time (min)