Comparison of fire effluent composition between large-scale and small-scale tests with sandwich panels

(1)

with sandwich panels

Per Blomqvist and Patrik Johansson

Fire Research SP Report 2014:25

S

P

T

ec

hn

ic

al

R

es

ea

rc

h

In

st

itu

te

o

f S

w

ed

en

(2)

Comparison of fire effluent composition

between large-scale and small-scale

tests with sandwich panels

(3)

Abstract

The composition and toxicity of the fire effluent spread in a building during a fire is an important factor in a fire safety assessment. Large-scale tests to investigate the conse-quences of specific fire scenarios are expensive to conduct and are consequently infrequently made. Modelling of smoke spread in buildings is an alternative method which is often used, but input data on smoke production and effluent composition from specific building products are most often not available. As a result, contribution to smoke toxicity from building products is often overlooked despite that these products can often be part of fire partitioning elements. An approach to determine such data is to test the building product using a small-scale laboratory method. It is, however, important that the use of small-scale data on fire effluent composition for extrapolation to large-scale fire scenarios is validated and based on sound principles. The work presented in this report concerns small-scale characterization of the effluent content from burning sandwich panel products and the comparison of these results with the fire effluent composition measured during large-scale fire tests.

Multiple fire resistance tests were conducted with sandwich panels according to EN 1364. In addition to standard measurements, the fire effluents on the cold side of the panels were collected and analysed in detail. Fire effluents were collected and analysed both before and after integrity failure. Materials from the same batch of sandwich panels were tested using ISO/TS 19700:2007, the steady-state tube furnace method. The effluent compositions from three fire stages as defined in ISO 19706 (pyrolysis, well-ventilated, and under-ventilated combustion) were investigated in the small-scale tests. This is a unique feature of the ISO/TS 19700 which allows the extrapolation of small-scale test results and correlations to other tests and scales. As sandwich panels are composite products, consisting of metal sheets on both sides of a core of insulation material, this had to be considered in the sample preparation for the small-scale tests.

The sandwich panels investigated were of two common types, with a core of either mineral wool or PIR insulation material, respectively. These are products which are interesting to study as both types of products can produce, e.g., hydrogen cyanide (HCN) and nitrogen oxides (NOX) in addition to carbon monoxide (CO). Further, there is a distinct difference between the two types of sandwich panels, as a mineral wool panel contains two combustible components in small quantities only: adhesive connecting the insulation core and steel sheets, and the binder in the mineral wool core. While the PIR core material is combustible with its combustion performance depending on the combustion conditions and presence or absence of additives.

The study of correlation of the fire effluent composition between the large-scale tests and the small-scale characterization was challenging. Both the complexity of the tested products and the nature of the large-scale tests presented complications. However, the high level of complexity has been useful to exemplify the considerations that have to be made when using small-scale fire effluent composition data.

Key words: Large-scale fire resistance test, tube furnace, toxic gases, correlations

SP Sveriges Tekniska Forskningsinstitut

SP Technical Research Institute of Sweden SP Report 2014:25

ISBN 978-91-87461-73-6 ISSN 0284-5172

(4)

Acknowledgments

Rockwool International A/S has sponsored the experimental work reported here. The assessment of the effluent data from the different physical scales of tests was co-sponsored between VINNOVA – Swedish Governmental Agency for Innovation Systems and Rockwool International A/S. The funding from VINNOVA was in support of the development of a virtual test bed to couple experimental resources and computational modelling. The work presented here concerns the use of small-scale laboratory test data for the prediction of large scale performance.

(7)

Sammanfattning (in Swedish)

Kunskap om innehållet i den rök som sprids i en byggnad vid brand är en viktigt del i en funktionsbaserad brandteknisk riskutvärdering. Att genomföra storskaliga brandförsök för att utreda konsekvenserna av en specifik brand är svårt och kostsamt och utförs därför bara i undantagsfall. Ett vanligt metod är istället att beräkna spridningen av rök och brandgaser, men det är sällan att det finns inputdata för specifika material utan man använder sig istället av schablonvärden. Som ett resultat av bristen på relevant information överser man i många fall från risken med rökens toxicitet för specifika byggnadsprodukter, även när dessa ingår i brandavskiljande konstruktioner. En teknik för att bestämma produktionsparametrar för toxiska komponenter är att utföra provningar med en småskalig laboratoriemetod. Det är här mycket viktigt att den småskaliga metoden som tillämpas är validerad för att kunna representera det storskaliga scenariot som skall modelleras samt att dataextrapolering görs enligt accepterade brandtekniskt principer. I denna rapport beskrivs hur brandgaserna från sandwichpanel produkter har karakteriserats med hjälp av en småskalig metod och hur denna information sedan har använts vid en jämförelse med brandgassammansättningen uppmätt från storskaliga brandförsök.

Brandmotståndsprovningar utfördes med sandwichpaneler enligt EN 1364. I tillägg till de normala mätningarna vid ett sådant försök gjordes detaljerade mätningar av rökgas-sammansättningen på läckage på den kalla sidan av provkroppen. Dessa analyser utfördes både innan och efter det att den provade konstruktionen förlorat sin brandavskiljande förmåga. Material från identiska paneler provades sedan med ISO/TS 19700:2007, en rörugnsmetod med titeln ”Controlled equivalence ratio method for the determination of hazardous components of fire effluents”. Rökgassammansättningen vid tre olika brandförhållanden undersöktes; oxidativ pyrolys, välventilerad brand och underventilerad brand (definierade i ISO 19706). Att kunna prova med definierade brandförhållanden är en unik egenskap för ISO/TS 19700, vilket ger möjlighet att kunna extrapolera resultat från en småskalig provning till andra tester och skalor. Då sandwichpaneler är kompositmaterial som består av en kärna av isoleringsmaterial omgiven av metalplåtar på båda sidor, var man tvungen att beakta detta vid den småskaliga provningen.

Två vanligt förekommande typer av sandwichpaneler undersöktes, den ena med en kärna av minerallull och den andra med en kärna av PIR. Detta är produkter som är intressanta att studera, då båda kan producera, t.ex., vätecyanid (HCN) och kväveoxider (NOX) vid en brand, förutom kolmonoxid (CO). Det finns också en distinkt skillnad mellan de två typerna av sandwichpaneler. Panelen med mineralullskärna innehåller två olika brännbara komponenter, ett lim för sammanfogning av kärnan med metallplåtarna och ett bindemedel i kärnmaterialet, båda i små kvantiteter. Kärnmaterialet i PIR panelen däremot, är brännbart i sin helhet med en förbränningseffektivitet beroende av det specifika brandförhållandet samt förekonsten av additiv i materialet.

Den genomförda undersökningen av korrelationen av rökgassammansättning mellan småskaliga och storskaliga brandtester var utmanande. Både sammansättningen av produkterna och utformningen av den storskaliga provningsmetoden gav svårigheter. Men komplexiteten var värdefull för att exemplifiera de hänsynstaganden man måste göra vid tillämpningar av produktionsdata från en småskalig provningsmetod.

(8)

1 Introduction

A series of fire resistance furnace tests with two types of sandwich panel wall products was commissioned by Rockwool International A/S to be conducted at SP Fire Research. The two products investigated included a PIR sandwich wall panel product and a mineral wool product. These were not comparable products, the PIR product was nominally designed for 15 minutes fire resistance while the mineral wool product was designed for 120 minutes. These tests were not conducted to investigate any implications concerning safe use of the tested products specifically. The tests were made to investigate the phenomenon of smoke leakage to the non-exposed side of a fire separating construction. Triplicate tests were conducted with the PIR product to investigate the repeatability in this case and a single test was conducted with the mineral wool product. The tests were conducted as standard tests of fire resistance performance with the deviation that the edges of the wall panels were fixed to the frame of the furnace to avoid any smoke effluent leakage from these points. Addition to the standard test method was that fire gases on the cold side of the panel specimen were collected and analysed. The tests with the PIR product were run up until the point of failure of the structural integrity of the product. Any measurements taken after failure of structural integrity were outside of the standard methodology as this represents the cut off point for measurements during a standard test.

The tests presented in this report are unique in that they have been supplemented by detailed quantitative analysis of the fire effluent leakage to the cold side of the wall, with the purpose to investigate the quality and magnitude of a possible leakage and to serve as validation experiments for further studies. The subject of this report is the comparison of the smoke gas effluent information achievable from a small-scale physical fire model with that from the large-scale tests. The primary focus of this study has been on the period of fire effluent leakage before integrity failure of the studied wall panel constructions.

The ISO/TS 19700 steady-state tube furnace method [1] was selected as the most relevant small-scale physical fire model for the purpose of studying the emissions from the wall panel materials.

It is important to keep in mind that it is very challenging to make representative small-scale tests with a commercial composite product. A sandwich panel consists of an insulating core material surrounded by metal sheets which can be glued to the core and have a combustible surface coating. It is not feasible to directly replicate the burning behaviour, and thus the smoke gas effluent production, of such a product with a small-scale test. The purpose of the study presented here was therefore to investigate whether useful data could be provided by the steady-state tube furnace and to what extent this data could be used for predictions of qualitative and quantitative nature of the fire effluent production from sandwich panels.

The comparisons presented here include the production of the major fire gases, total production of organics and particles. The tests also included the analyses of individual organic combustion products but these are not reported on here.

(9)

2 Sandwich wall panel products

Two types of commercial wall panel products were selected for the study by Rockwool International A/S and delivered to SP. The two products included a PIR wall panel and a mineral wool panel. These were not comparable products in terms of their design fire resistance, i.e. the PIR product was nominally dimensioned for 15 minutes of fire resistance and the mineral wool product was designed for 120 minutes.

The identification of the panels was made using information on the labels on the products and from specification tests (see below). The commercial identification of the products is, however, not given here. The designations used throughout in this report are MW core panel (MW = Mineral Wool) and PIR core panel.

The products were tested for specifications of significant properties and material content. The sponsor delivered several panels of each kind. SP randomly picked which ones would be used for the fire test while using some of the extra material for the specification tests.

A summary of the results are given here focusing on specifications important for burning behavior and fire effluent content. The detailed results of the specification tests are given in Appendix A.

The MW core panels were constructed of a core material of mineral wool board with one steel sheet on each side of the core. The steel sheets were glued to the core. The mineral wool lamellas had the nominal thickness 173 mm and a nominal density of 135 kg/m3. The core material was to a great extent non-combustible, containing only 3.3 weight-% combustible material. The core material contained 0.5 weight-% of nitrogen (presumably from a binder additive) and the glue additionally contained 1.3 weight-% of nitrogen. The PIR core panels were constructed of a core material of PIR with one steel sheet on each side of the core. The PIR core had a nominal thickness of 95 mm and a nominal density of 37 kg/m3. The core material was nearly completely combustible (99 weight-% combustible material). The core material had a high content of nitrogen (7.4 weight-%) and is assumed to have been flame retarded, as both chlorine and phosphorous were detected in the core material. The metal plate had a coating which contained high concentrations of chlorine and titanium, which indicate that the coating was made of PVC.

(10)

3 Analysis of fire effluent leakage in vertical

furnace fire resistance tests

3.1 Fire resistance tests

The large-scale tests were in most parts conducted as standard vertical furnace fire resistance tests. The collection and analysis of the fire effluent leakage from the joints of the panels were additions to the standard test method. The tests were further run for longer time after insulation and integrity failure of the tested constructions compared to normal tests.

3.1.1 Test set-up

The tests were following a test procedure based on the methods EN 1364-1:1999 [2] and EN 81-58:2003, Annex A [3], where the smoke gases on the cold side of the wall construction were collected. In addition to what is prescribed in these two standards the measurement effort was increased in order to enable the analysis of contents of the leaked fire effluents. The vertical fire resistance furnace used was constructed according to EN 1363-1:2012 [4] and had the outlet for the combustion gases in the bottom of the furnace. Due to the additional measurements the tests were performed with some deviations from the EN 1364-1 standard. The mounting was performed without a vertical free edge to give focus to the vertical panel-to-panel joint in the specimens. The tested walls were fixed and sealed at all four edges to avoid smoke leakage from the furnace and to only allow leakage from the joints between the panelsi. Further, the evaluation of integrity failure by the cotton wool pad was not performed to give the possibility for the gases escaping through the panel-to-panel joint to ignite spontaneously. After the burners were turned off, the air flow through the outlet for the combustion gases in the bottom of the furnace was increased to cool the burners and the furnace.

Measurements were made of temperatures, smoke gas collection volume flow, smoke production, and fire effluent contents. The fire effluents leaked from the joints of the panels were collected using a hood and duct system which made it possible to quantitatively analyse the contents of the fire effluents. Details on the test set-up are given in Appendix B.1.

To examine repeatability three identical tests were conducted with the PIR product. A single test was conducted with the MW product.

3.1.2 Summary of events in the tests

The points in time in the experiments for failure of the construction according to criteria in EN 1364-1 are summarized in Table 1. Here is also given the length of time of active burners in the furnace, and the total time for monitoring the measurement devices. Note that the starting time in the experiments was the time for burner activation.

i

This installation was different from the requirements in EN 1364-1, and resulted in a different mechanical response of the tested wall panels that influenced the tightness of the panel-to-panel joints and thus the smoke leakage. The influence on the tightness was probably positive as seen from earlier tests [12] with similar products.

(11)

The results of the test with the MW core panel differed significantly from the tests conducted with the PIR core panel. This difference was expected as these two products have different nominal fire resistance ratings as explained above.

The MW core panel did not fail any of the test criteria, which means that there was no visible flames seen from the cold side of the wall construction (integrity criteria) and that the temperature rise on the cold surface never exceeded 140 °C in average, and 180 °C as maximum (insulation criteria). It was decided to turn off the burners and end this test after slightly more than one hour, as it was decided at this point that enough information from the measurement of smoke gas constituents had been collected.

The PIR core panel had lower nominal fire resistance properties and showed integrity failure (visible flames) after ~16 minutes in average and insulation failure somewhat earlier in general. In these cases the burners were left active until heavy fire was observed in both joints of the construction, after which they were turned off. These tests were continued with full measurements for about 30 minutes. As shown later, this length of time for the tests was sufficient for most of the PIR core material to be combusted.

Table 1 Summary of fire resistance test results and length of time for the tests.

Test specimen Integrity failure* (min:sec) Insulation failure average/maximum temp (min:sec) Burners active (min:sec) Test monitored (min:sec) MW No failure No failure 68:40 68:40 PIR 1 18:15 18:25 / 15:50 18:30 32:00 PIR 2 15:53 16:11 / 14:56 19:55 33:00 PIR 3 17:05 19:30 / 17:11 18:45 32:10

* Integrity failure was determined by the occurrence of visible flames.

Leakage of smoke from the wall constructions was detected by visual observations of the tests and by on-line measurement of visual obscuration (expressed as smoke production rate, SPR) in the duct of the smoke gas collection system. There was no smoke leakage seen visually in the test with the MW core panel and only a very low visual obscuration could be measured (~0.01 m2/s). In the tests with the PIR core panel smoke leakage was detected both by visual observation and by measurement of visual obscuration. The measurements of visual obscuration in the tests with the PIR core panel are shown in Figure 1.

(12)

Figure 1 Smoke production rate (SPR) measured in the duct of the smoke collection system in the tests with PIR core panel.

It can be seen from Figure 1, that leakage of smoke occurred first around two minutes into the tests and that the major leakage started somewhere between 11 and 15 minutes, varying between tests.

In order to evaluate the composition of the leaked fire effluents during specific comparable periods in the tests with the PIR core panel the following time periods have been defined:

1) before significant smoke leakage,

2) during significant smoke leakage before integrity failure, 3) after integrity failure until burners were turned off, and 4) from when burners were turned off until flaming stopped.

Significant smoke production was defined as the time the smoke production rate (SPR) reached over a value of 0.1 m2/s in the beginning of the large peak in smoke production (leakage) that was seen in all three tests. Period 2-4 have been used for the evaluation of the smoke gas composition and the length of these periods in the individual tests are given in Table 2.

Table 2 Length of periods used for evaluation .

Test Period 2 (min:sec) Period 3 (min:sec) Period 4 (min:sec) PIR 1 6:50 0:15* 13:30 PIR 2 3:18 4:02 13:05 PIR 3 2:10 1:40 13:25

* This period was too short and was not included in the evaluation.

Notes from observations in the experiments and time lines with important events in the experiments can be found in Appendix B.2.

(13)

3.2 Results on gas components

The results of the measurements of fire gases from leakages to the cold side in the large-scale tests are summarized in the sections below. Complete test data is available in Appendix B.4. The test data includes graphs of time resolved measurements, graphs of calculated leakage rates from time resolved measurements, calculated average leakage rates for specific time periods, and the total leakages for these time periods.

3.2.1 MW core panel

In the single test with the MW core panel wall construction there was no visible leakage of fire gases. Instead a continuous, initially increasing, leakage occurred during the test. The gases detected with the on-line FTIR measurement included CO2, CO, HCN and NH3, which are shown in Figure 2. The measurement results are presented as leakage rate in g/s, which have been calculated from the concentration measured in the collection duct where all leakage was collected. The presented leakage rate is therefore based on the total leakage from the tested wall.

Figure 2 Leakage rates of fire gases measured in the single test with the MW core panel

wall construction.

Note that CO2 increases very rapidly in the beginning of the test, while e.g. CO increases much later. A possible explanation is that while the degradation of the combustible material in the panel progresses into the panel, the available oxygen is more limited and the production of less oxidised species is favoured. However, it is not possible to completely determine the origin of CO2 for the assessment of the production behaviour, as small contributions of CO2 from the burners could not be ruled out in the total leakage. Calculation of the amount of CO2 expected from the burners indicates that emissions in the furnace are nominally in the order of 67 kg of which the vast majority is extracted by the furnace ventilation system. The total maximum theoretical production from the MW core panel material is calculated to be in the order of 13 kg and the leakage would have been significantly less (see Section 6.2.2). In total a leakage of 2.4 kg of CO2 was measured. It is therefore probable that all of the CO2 measured originated from the

0.00 0.20 0.40 0.60 0.80 1.00 0 10 20 30 40 50 60 70 Le a k a g e r a te ( g /s ) Time (min)

Carbon dioxide (CO2) Carbon monoxide (CO) Hydrogen cyanide (HCN) Ammonia (NH3)

(14)

product. However, it cannot be ruled out that a small portion of the CO2 measured originated from the burners.

3.2.2 PIR core panel

The tests with the PIR core panel wall construction all had similar outcome, starting with a small leakage in the beginning of the tests and major leakages later in the tests in connection with the time of integrity failure of the structure. The gases detected with the on-line FTIR measurement are shown in Figure 3 - Figure 5 for the individual tests. Measurable quantities of CO2, CO, HCN, NO, NH3 and HCl could be detected in all tests.

Figure 3 Leakage rates of fire gases measured in Test 1 with the PIR core panel wall

construction.

construction. 0.0 0.5 1.0 1.5 2.0 0 5 10 15 20 25 30 35 Le a k a g e r a te ( g /s ) Time (min) CO2 CO HCN NO NH3 HCl 0.0 0.5 1.0 1.5 2.0 0 5 10 15 20 25 30 35 Le a k a g e r a te ( g /s ) Time (min) CO2 CO HCN NO NH3 HCl

(15)

construction.

The initial small gas leakage seen in all tests can be explained by thermo-mechanical movements of the panels creating a temporary leakage in the panel joints. This leakage is rather quickly sealed by the swelling of the PIR core material. The swelling behaviour of the PIR core material was clearly seen in the tube furnace tests which are presented in more detail in the next chapter.

Note the increased CO2 base level in all three tests after the initial short period of leakage. It is not possible to determine whether this is a result of continued leakage from the joints between the panels or from leakage of CO2 from the burners between the frame and the panels. However, the CO2/CO ratio for this period is significantly higher compared to after the time for start of significant smoke leakage, which indicates that at least some portion of the CO2 measured originates from the burners.

3.3 Leakage rates and mass consumption

The measurements of fire effluents in the tests were made on the cold unexposed side of the sandwich panel wall. This means that the leakages in the joints between the panels and any leakages between the panels and the frame of the furnace were included in the measurements. The last mentioned potential source of leakages, was limited but could not be separated from the overall measurements. However, only leakages from the joints were visually observed.

The leakages from the PIR core panels could be clearly seen by visual observation. This is shown in Figure 8 with photos from different times in Test 3 with the PIR core panel. What is clear is that only part of the produced fire effluents leaked out on the unexposed side where the measurements were made. The remaining part of the effluents went into the exhaust system in the bottom of the furnace. The extent of leakage varied significantly during the time of the test as can be seen from the photos in Figure 8. It is not possible to

0.0 0.5 1.0 1.5 2.0 0 5 10 15 20 25 30 35 Le a k a g e r a te ( g /s ) Time (min) CO2 CO HCN NO NH3 HCl

(16)

quantify the extent of leakage but the measured pressure inside the furnace can give some guidance.

A neutral pressure plane with the height of 670 mm was maintained within the furnace during the tests, which is in according with the test procedure in EN 1364-1. The pressure gradient in the furnace shall be 8.5 Pa/m height, resulting in a positive difference pressure at the top of the specimen (at 3000 mm) of max 20 Pa. Graphs of the pressure measurements at the height of 670 mm are provided in Appendix B.3 for all tests. It can be seen here that the pressure situation changes drastically when the burners are turned off, and thereafter the neutral plane is located at a significantly higher position in the furnace.

Calculations are given in Appendix B.5 where the surface area of the wall constructions that experienced a positive difference pressure (relative to the outside of the furnace) during the tests have been investigated. The assumption is that all combustion products from the wall within areas of positive pressure are leaking out to the unexposed side. This is clearly an extreme assumption. Further, assuming that the consumption of combustible material started directly in the tests, and continued at a constant rate, the maximum possible leakages have been calculated for each test. The results of these calculations shows maximum possible leakages of 77 % in the test with the MW core panel and an average of 58 % in the tests with the PIR core panels. However, if we assume that there was no leakage at all in the first (low smoke production) period in the tests with PIR, the average maximum leakage rates is 34 %.

Time: 13:35 min:s Smoke from joint.

(17)

Time: 17:00 min:s Smoke level is increasing. Time: 17:18 min:s Ignition. Time: 18:06 min:s Ignition in second joint.

Figure 6 Photos from Test 3 with the PIR core panel showing visible leakages of fire effluents on the unexposed side of the wall.

The amount of material consumed in the tests determined the total production of fire effluents from the sandwich wall panels. The mass lost from the sample species is not

(18)

known since the walls were extinguished with water in the end of the tests and post-test weighing was therefore not possible.

However, inspections of the wall panels after the test showed that significant amounts of the combustible materials in the walls had been consumed. For the MW core panel, the mineral wool core material essentially is non-combustible and the consumption of binder in the core could only be assessed from the depth of blackened (pyrolysed) material. Figure 7 shows pictures of the MW core panel after the tests, and it is clear that the heat wave from the furnace resulted in pyrolysis to a significant depth into the core.

Time: After the test

The fire exposed side of the MW test specimen having had one of its metal sheets removed.

Time: After the test Pieces of mineral wool cut out from the core. The depth of pyrolysis can be seen to extend 50-75 % into the material.

Figure 7 Photos from the test with the MW core panel showing the extent of burning of the

core material.

Inspection of the PIR core panels after the tests showed that much of the core material had been consumed but that patches with thick layers of charred material remained. Figure 8 shows photos of the wall panels after Test 3 with PIR where the metal sheets have been removed. Only a small fraction of non-charred material was found.

(19)

Time: After the test The fire exposed side of the test specimen having had its metal sheets removed.

Time: After the test A closer look on the fire exposed side in the centre of a panel showing a small fraction of non-charred material.

Figure 8 Photos from Test 3 with the PIR core panel showing the extent of burning of the

(20)

4 Steady-state tube furnace tests

4.1 Description of tests

4.1.1 Steady-state tube furnace

The steady-state tube furnace tests were conducted with a test set-up according to ISO/TS 19700 [1]. The test set-up is shown schematically in Figure 9. This method has proven repeatable in establishing controlled combustion conditions with solid polymeric materials [5]. The combustion conditions are established by feeding the sample material into the furnace together with primary air for combustion. The combination of the specific material feeding rate, the primary air flow rate, and the furnace temperature determines the combustion conditions.

Figure 9 Schematic drawing of the ISO/TS 19700 steady-state tube furnace.

The fire stages [1] modelled in the tests reported here included (nominal furnace temperature in parenthesis):

Fire stage 1b – oxidative pyrolysis (350 °C). Fire stage 2 – well-ventilated flaming fire (650 °C).

Fire stage 3b – post-flashover fire, under-ventilated (825 °C).

ISO/TS 19700 specifies a nominal loading of 25 mg combustible material/mm with a feeding rate of 40 mm/min. The primary air flow is nominally 10 l/min for well-ventilated combustion. This results in a material/air ratio of 100 mg combustible material/ l air. The feeding rate and the primary airflow rate can, however, be adjusted to allow tests with, e.g., low-density materials such as insulation materials.

Flaming combustion is required for fire stages 2 and 3b. If the nominal temperature does not give flaming combustion the furnace temperature shall be raised in steps of 25 °C until flaming combustion occurs.

(21)

4.1.2 Sample preparation

Two different types of samples were made from the tested products, i.e.: • Core material - Sample species made from core material of the panel.

• Metal sheet - Sample species made from the metal plate of the sandwich panel with a certain amount of core material to make flaming combustion possible (Note: this type of composite sample species is not covered by ISO/TS 19700). The size of the sample boat and the diameter of the quarts tube of the furnace sets restrictions on the maximum sample dimensions. It was decided to use the same sample width for all types of samples/materials and that dimension was set to 22 mm. The total length of the sample species was in all cases 790 mm. The height of the sample specimen was then adjusted to give an optimal combustible loading for a well-ventilated test (Fire stage 2). This was possible for PIR/metal sheet, PIR/core material and MW/metal sheet. It was, however, not possible to attain a high enough material-air ratio for the MW/core material although the maximal sample dimension was used in this case.

Information on the sample species and tube furnace settings used for nominally well-ventilated tests are summarized in Table 3.

Table 3 Sample specimen data and calculated material-air ratio for tube furnace settings used for nominally well-ventilated tests.

Specimen type Height* (mm) Combustible loading (mg/mm) Advance rate (mm/min) Primary air flow rate (l/min) Material-air ratio (mg combustible/l air) PIR/metal sheet 15 16 60 10 97 PIR/core material 15 12 60 7 102 MW/metal sheet 15 11 60 7 95 MW/core material 22 2.5 60 7 22

* All specimen types had a length of 790 mm and a width of 22 mm.

The combustible loading for the different sample species was calculated by determining the combustible content of the individual materials by combustion in a furnace at 550 °C. The metal plates were removed of insulation material before determining their combustible contents (surface coating and, if applicable, glue).

More information is available in Appendix C.1. Pictures of typical sample specimen can be found in Appendix C.3.

4.1.3 Testing scheme

The tube furnace tests conducted, the test conditions, and the analyses conducted in each test are given in Table 4 for the tests with PIR and in Table 5 for the tests with MW. The analyses are described in more detail in Appendix C.1.

(22)

The test series was started with well-ventilated tests with PIR/core material. A number of preparatory tests were conducted, and it was found that steady flaming combustion was not attained at the nominal 650 °C. The material showed intermittent flashing combustion at this temperature. At 675 °C uneven flaming was attained, and it was decided to use 700 °C for the following well-ventilated tests. It was also found in the preparatory tests that the PIR material increased significantly in volume and also lifted from the bottom of the sample boat due to the effect of the heat. The sample specimen was in subsequent tests fixed to the sample boat with a thin steel-wire each 10 cm along the sample, to keep the sample in place. The mounting of the sample specimens is shown in Appendix C.3. The primary flow rates for the under-ventilated tests with the PIR samples were calculated from the oxygen consumption in the well-ventilated tests as described in ISO/TS 19700.

The test programme with MW sample specimen was limited in terms of the number of tests. Only single tests were conducted and under-ventilated tests were not included in the programme. Under-ventilated tests (825 °C) were considered to be less important for the MW core panel, as these conditions were not assumed to take place in the large-scale furnace tests. In the well-ventilated tests with MW it was decided to use the same temperature (700 °C) as was used in the tests with PIR.

The boat speed in the pyrolysis tests at 350 °C was 40 mm/min, which is the standard advance rate given in ISO/TS 19700. The fact that the combustible loading was lower that the nominal loading of 25 mg/mm is not critical in a pyrolysis tests as flaming combustion shall not occur.

Table 4 Tests with PIR: test conditions and analyses.

Sample Furnace temperature (°C) No. of tests Boat speed (mm/min) Primary flow (l/min) Secondary flow (l/min)

Fire effluent analysis

PIR/core material 675 700 1 2 60 60 7.0 7.0 43.0 43.0 O2, FTIR Complete analysis* 825 825 1 2 60 60 2.4 2.4 47.6 47.6 O2, FTIR Complete analysis* 350 1 40 2.0 48.0 Complete analysis* PIR/metal sheet 700 700 1 2 60 60 10.0 10.0 40.0 40.0 Complete analysis* Complete analysis* 825 825 1 2 60 60 3.3 3.3 46.7 46.7 Complete analysis* Complete analysis* 350

350 1 ₂ 40 ₄₀ 2.0 _2.0 48.0 _48.0 _{Complete analysis*}O2, FTIR

(23)

Table 5 Tests with MW: test conditions and analyses. Sample Furnace temperature (°C) Boat speed (mm/min) Primary flow (l/min) Secondary flow (l/min)

Fire effluent analysis

MW/core material 700 60 7.0 43.0 Complete analysis* 350 40 2.0 48.0 Complete analysis* MW/metal sheet 700 60 7.0 43.0 Complete analysis* 350 40 2.0 48.0 Complete analysis*

* Complete analysis included O2, FTIR, total particulates, FID and individual organics.

4.2 Results of the steady-state tube furnace tests

4.2.1 Burning behaviour of PIR samples

The PIR/core material tests with 700 °C furnace temperature and high ventilation resulted in continuous but uneven flaming combustion. The uneven flaming behaviour indicates that the material was flame retardant treated. The material analysis reported on in Appendix A, shows the presence of phosphorous and chlorine in the core material, which confirms the presence of a flame retardant. The mass-loss was relatively high in these tests, on average around 80 %.

The PIR/core material tests with 825 °C furnace temperature and low ventilation resulted in a more steady flaming combustion. The mass-loss was lower compared with the tests with high ventilation (average around 65 %). A voluminous residue was left in the sample boat after these tests (see Appendix C.4).

The PIR/core material tests with 350 °C furnace temperature did not show flaming combustion and can be regarded as regular oxidative pyrolysis tests. The sample increased in volume and the mass-loss was limited (35 %).

The PIR/metal sheet tests with 700 °C furnace temperature and high ventilation showed steady flaming behaviour.

The PIR/metal sheet test with 825 °C furnace temperature and low ventilation showed steady flaming behaviour. A voluminous residue remained after these tests (see Appendix C.4).

The PIR/metal sheet test with 350 °C furnace temperature did not show flaming combustion and can be regarded as regular oxidative pyrolysis tests. The sample increased in volume.

(24)

4.2.2 Combustion products from PIR

The major combustion products found from the tube furnace tests with the PIR/core material are summarized in Figure 10. Detailed results for all tests conducted are provided in Appendix C.4.

Figure 10 Averaged yield data from steady-state tube furnace tests with PIR /core material samples.

The tests with the PIR /core material show the expected increase in the production of CO and HCN for the under-ventilated tests. Also the increased yields of particles in the under-ventilated and the pyrolysis tests are expected. The yield of HCl is approximately equal for the two flaming combustion conditions investigated, while HCl was not detected from pyrolysis at 350 °C. Note that HCN was not detected in the pyrolysis test and that NH3 was not detected for any of the combustion conditions.

Figure 11 Averaged yield data from steady-state tube furnace tests with PIR /metal sheet samples.

The results from the tests with the PIR/metal sheet samples are shown in Figure 11. The results do not differ significantly from those from the PIR/core material. This could be

(25)

explained by the fact that the majority of the combustible material of the sample specie was from the core material; only 32 % of the combustible material was from the coating on the metal sheet. The major differences seen is in the yield of particles and that NO is detected also from under-ventilated combustion conditions. One should remember, at this point, that the proportion of combustible material from the metal sheet and from the core material was selected to create the desired combustion conditions in the tube furnace tests. The proportion was not the same as that for the PIR core panel product. The reason for this choice, was that focus was put on the investigation of the emissions from the metal sheet with this type of sample specimen.

4.2.3 Burning behavior of MW samples

The MW/core material test with 700 °C furnace temperature and high ventilation did not result in flaming combustion which was a result of the low combustible content of the mineral wool. This test should be regarded as an oxidative pyrolysis tests at a high temperature.

The MW/core material test with 350 °C furnace temperature did not show flaming combustion and can be regarded as a regular oxidative pyrolysis tests.

The MW/metal sheet test with 700 °C furnace temperature and high ventilation showed intermittent flashing combustion but did not burn steadily. This tests must be regarded as a mixture of oxidative pyrolysis and well-ventilated flashing combustion.

The MW/metal sheet test with 350 °C furnace temperature did not show flaming combustion and can be regarded as a regular oxidative pyrolysis tests.

More detailed information and observations from the tests are given in Appendix C.2.

4.2.4 Combustion products from MW

The major combustion products found from the tube furnace tests with the MW/core material are summarized in Figure 12. Detailed results for all tests conducted are provided in Appendix C.4.

(26)

The tests with the MW/core material at 350 °C and 700 °C could both be regarded as pyrolysis tests. But the effect of the higher temperature and ventilation, in the 700 °C test, is quite significant in the increased production of CO2 and also CO. Additionally, NO is produced in the 700 °C test.

Figure 13 Yield data from steady-state tube furnace tests with MW /metal sheet samples.

The results from the tube furnace tests with the MW/metal sheet sample are shown in Figure 13. These results differ quite significantly from the results of the tests with the core material. There are two major reasons for this. First, the proportion of combustible material from the metal sheet was as high as 85 %. Secondly, the sample tested at 700 °C flashed intermittently, which was different from the tests with the core material where there was pure pyrolysis.

From the tests with both types of sample specimen it is clear that NH3 is produced principally during low temperature pyrolysis. It is also clear that the yield of HCN decreases with temperature, and that NO is produced instead during well-ventilated conditions.

4.3 Discussion of applicability

The products investigated here with the steady-state tube furnace were especially challenging as they both consisted of multi-layered materials. Also other characteristics of these products increased the difficulties in applying the ISO/TS 19700 test standard. The PIR core panel gave additional problems in testing due to the flame retardance of the core material which resulted in unsteady burning. The furnace temperature was raised from the standard 650 °C to 700 °C for well-ventilated tests. The temperature could, possibly, have been raised even more as the combustion was still somewhat unsteady. Also the swelling with large volume of residue from especially the under-ventilated and the pyrolysis tests gave complications in testing, e.g., the length of sample to use in the calculation of mass-loss was not evident to determine on.

The MW core panel could not be tested according to ISO/TS 19700 as the core material contains too low combustible content. It will not burn with flames irrespective of combustion conditions. Two different pyrolysis conditions were investigated instead; 350 °C using the primary air-flow prescribed in ISO/TS 19700 for pyrolysis conditions,

(27)

and 700 °C using the primary air-flow prescribed in ISO/TS 19700 for well-ventilated conditions. Both flows gave most possibly very high ventilation, considering the low combustible content of the core material.

The method used for testing multi-layered samples focused on including as large a surface area of the metal sheet as possible, while maintaining sample loading that complies with the requirements in ISO/TS 19700 for creating the desired combustion conditions. The proportions of combustibles on the different layers did not, however, represent the composition of the complete product. The yields from the metal sheet samples should, therefore, not be used directly without special considerations. The information gained from this sample type was quite limited in the work reported here, but could be useful for other types of composite samples.

In spite of the different complications, useful data was obtained from the steady-state tube furnace tests, which is demonstrated in Sections 5 and Section 6.

(28)

5 Qualitative comparison of gas composition

between test scales

5.1 Comparison principles

A meaningful comparison of the gas composition in the fire effluents from different fire tests and different scales of tests requires knowledge of the combustion conditions in each case. The production of fire gases depends mainly on the fuel, the thermal environment and most importantly the ventilation conditions. In-depth information on the consider-ations needed in such a comparison is given in ISO 29903 [6].

The tests reported here were conducted with sandwich panel wall products which are complicated types of products from the perspective of comparison of fire gas production between test scales. These products are composed of a core material surrounded with coated metal sheets on each side of the core and the metal sheet was further glued to the core in one of the products studied.

In the large-scale fire resistance tests the wall panels were heated from one side, the exposed side in the furnace, and were therefore gradually consumed in the direction towards the unexposed cold side of the panel. In the tests with the PIR core panel, joints between individual panels opened up as the test progressed, and the combustion conditions changed significantly due to flaming combustion that also included the outside coating of the panels. Thus the combustion conditions in these tests changed strongly with time and could also have varied at different positions on the test object at different times. Further, the different materials in the panels were involved in the combustion to a varying degree over the time of the test.

In the test with the MW core panel, no visible opening was seen in the joints between the individual panels during the tests which makes interpretation of this test less complex. The intention here is to investigate whether the gas composition data from the steady-state tube furnace tests show similarities with the gas composition found during the different periods with leakage of fire effluents that have been identified in the large-scale tests with the PIR core panel and for the single test with the MW core panel. It is also interesting to investigate whether the small-scale data can be used to identify the general combustion conditions in these different periods of the large-scale tests.

5.2 PIR core panel

The comparison of gas composition between the different scales of tests will by necessity be qualitative. Yield data is available for the steady-state tube furnace tests and there is information on the combustion conditions for these tests. For the large-scale tests there are only measured concentrations and production rates available, and there is limited information on the combustion conditions, mainly based on observations. Further, only a part of the produced combustion gases leaked out to the unexposed side of the panels and was measured.

The parameter chosen for the comparison is production rate. Production rates have been calculated in mol/s for selected periods in the two scales of tests. In the tube furnace tests averaged production rates were calculated for the steady-state periods which are also used for calculating yields. In the large-scale tests the averaged production rates were

(29)

calculated for some specific periods identified (significant smoke, flaming combustion when the burners were on, and flaming combustion with the burners turned off).

The advantage of using the molar unit in the qualitative comparison instead of mass is that mole is proportional to volume concentration which mass is not. It would have been possible to use the actual measured concentrations in each scale, but that would have included dilution factors and molar productions is therefore more straightforward.

5.2.1 Production rates

Average leakage rates (in mol/s) of gas species for the three tests with the PIR core panel are given in Figure 14 for the specific selected burning periods. The first test with the PIR panel (PIR 1) had a very short period of flaming combustion when the burners were still turned on, and this data was therefore excluded in the comparison.

Note that the relative distribution between gases for each studied period is of the most interest here, as absolute values for the different periods are dependent on the size of the actual leakage.

Quite naturally, CO2 is the dominating gas component, especially during the “Flaming” period, but one must consider that part of the CO2 measured in the leaked fire effluents could have originated from the burners. Other gas components were found in different proportions depending on the burning period, and included CO, HCN, NO, HCl, NH3 and unburned hydrocarbons (THC). The presence of NH3 was not very consistent between the tests. It was only for the “Smoke” period that NH3 was found in all tests. In the “Flaming” period NH3 was only found in PIR 3, while for the “Flaming-burner off” period NH3 was only found in the PIR 1 test.

Figure 14 Average leakage rates of gas components to non-exposed side during periods with different burning conditions in the large-scale tests with PIR-panels.

The extent of leakage of the different gas components could be better seen using a logarithmic scale on the y-axis, as shown in Figure 15. Disregarding CO2, one can see that the gas composition during the “Smoke” period is dominated by THC, contains a significantly lower part of CO, and even lower parts of HCN, NH3 and HCl (HCl was

(30)

actually only found in low concentrations during a short period in the PIR 1 test). NO was not detected at all during this period. The gas composition indicates that the leaked fire effluents were produced mainly from pyrolysis, alternatively in combination with under-ventilated combustion. The high proportion of unburned hydrocarbons, the presence of HCN and NH3, and the absence of NO, are clear evidence corroborating this conclusion. The gas composition during the “Flaming” period contained, apart from CO2, also CO, THC, HCN, NH3 (in one test only) and NO. This composition indicates a mixture of combustion conditions. In addition, HCl is produced in all three tests during this period, which could originate from the PVC coating on the metal sheeting that was exposed to the flames here, in addition to production of HCl from the chlorine in the core material. The gas composition during the “Flaming – burners off” period, which was an extended period of similar length in all tests, is similar to the composition during the previous discussed period. The main difference is that the proportion of HCN is lower, which indicates a more well-ventilated combustion. This is logical as the furnace was still supplied with fresh air during this period, although the burners were not active. The presence of NH3 here is not conclusive as NH3 was only found in one of the tests during this period.

Figure 15 Average leakage rates of gas components to non-exposed side during periods with different burning conditions in the large-scale tests with PIR-panels (logarithmic x-axis).

The averaged production rates (in mmol/s) for the steady-state tube furnace tests with the PIR core material are given in Figure 16 and Figure 17 (logarithmic y-axis in the last figure).

The gas composition for the “Non-flaming” test conditions is similar regarding CO and THC compared to that for the “Smoke”-period in the large-scale tests. A major difference, however, is that neither HCN, NH3 nor HCl are found in the tube furnace test. The “Flaming, high ventilation” and the “Flaming, low ventilation” test conditions with the tube furnace tests show a somewhat similar gas composition. The main differences between these test conditions are that there is not any NO produced under the low ventilation conditions and that the production of CO, THC and HCN are 5-10 times

(31)

higher for the low ventilation conditions. Notably NH3 is not found in any of the tube furnace tests with the core material. It would have been expected to find NH3 in the non-flaming test and perhaps in the test with low ventilation conditions, as NH3 is an early pyrolysis product. However, the fact that the production of NH3 is difficult to predict was seen also in the large scale tests where the production of NH3 was very varied between tests.

Averaged production rates (in mmol/s) for the steady-state tube furnace tests with the PIR /metal sheet samples (contained metal sheet and core material, see Section 4.1.2) are given in Figure 18 and Figure 19. The gas composition from this type of samples can be seen to be quite similar to that from the core material sample species (Figure 16 and Figure 17). That is not unexpected, as 68 % of the combustible content in this type of sample species was in fact PIR core material. The major difference is that the proportion of NO for the well-ventilated conditions is higher from the metal sheet sample and that NO is found here also from the low ventilation condition. The explanation for the increased production of NO must either be that the production was affected by the presence of the metal sheet component (but how is not clear), or alternatively that the combustion conditions unintentionally were less under-ventilated than expected.

Figure 16 Production rates from tube furnace tests for steady-state burning of PIR core material. 0 0.1 0.2 0.3 0.4 0.5 0.6 Non-flaming, 350 °C

Flaming, high ventilation, 700 °C

Flaming, low ventilation, 825 °C P ro d u ct io n r a te s (m m o l/ s)

Tube furnace: PIR core material

CO2 CO THC HCN NO HCl

(32)

Figure 17 Production rates from tube furnace tests for steady-state burning of PIR core material (logarithmic x-axis).

Figure 18 Production rates from tube furnace tests for steady-state burning of PIR metal sheet with core material.

0.0001 0.001 0.01 0.1 1 Non-flaming, 350 °C

Tube furnace: PIR core material

CO2 CO THC HCN NO HCl 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Non-flaming, 350 °C

Tube furnace: PIR metal sheet + core material

(33)

Figure 19 Production rates from tube furnace tests for steady-state burning of PIR metal sheet with core material (logarithmic x-axis).

A simplified method for trying to identifying the combustion conditions during the different burning periods in the large-scale tests, is to make a relative ranking of the individual gas components for the burning periods in the large-scale tests and compare this with ranking of gas compositions from the different fire stages in the steady-state tube furnace tests.

Such a ranking is presented for the large-scale tests in Table 6 and for the tube furnace tests in Table 7. Note that products only found in one of the three large-scale tests during a specific period are not included in Table 6.

Table 6 Ranking of production rates for specific periods in the large-scale tests.

Period in large-scale tests

PIR core panel tests Ranking of gas components

1 2 3 4 5 6 Smoke CO2 THC CO HCN NH3 Flaming CO2 CO THC HCl HCN NO Flaming – burners off CO2 CO THC HCl NO HCN 0.0001 0.001 0.01 0.1 1 Non-flaming, 350 °C

Tube furnace: PIR metal sheet + core material

(34)

Table 7 Ranking of production rates for specific combustion conditions in steady-state tube furnace tests.

Combustion condition

PIR core material

Ranking of gas components

1 2 3 4 5 6 Non-flaming, 350 °C CO2 THC CO Flaming, high ventilation, 700 °C CO2 CO HCN THC HCl NO Flaming, low ventilation, 825 °C CO2 CO HCN THC HCl Combustion condition

PIR metal sheet with core material Ranking of gas components

1 2 3 4 5 6 Non-flaming, 350 °C CO2 THC CO Flaming, high ventilation, 700 °C CO2 CO HCl HCN NO THC Flaming, low ventilation, 825 °C CO2 CO THC HCN HCl NO

The information from the ranking indicates that the gas composition during the “Smoke” period in the large scale tests mainly originates from non-flaming combustion, i.e., pyrolysis, as THC is the dominating (group of) gaseous species and CO is the second most abundant species. The presence of HCN indicates that flaming combustion contributes, most certainly under-ventilated combustion as there is not any NO present. Another possible explanation of the presence of HCN could be the production of HCN from high temperature pyrolysis. There was no HCN found in the oxidative pyrolysis tests at 350 °C, but it is quite possible that HCN is formed from pyrolysis at higher temperatures [7].

The ranking of the gaseous components for the two flaming periods in the large-scale tests is very similar. The only difference is the higher ranking of NO for the period when the burner was off. This indicates a more well-ventilated combustion for this period, as was concluded earlier when directly comparing the relative production rates.

Using ratios of production rates for comparison could be helpful for identifying the combustion conditions during the different burning periods in the large-scale tests. Some possible combinations of ratios were calculated from molar production rates and are presented in Table 8 and Table 9 below. We have avoided the use of ratios including CO2, as the extent of contribution from the burners is unknown.

(35)

Table 8 Ratios of gas components for specific periods in the large-scale tests.

PIR core panel tests

CO/THC CO/HCN CO/NO CO/HCl

Smoke 0.23 6.2 - 25

Flaming 1.2 11 22 2.0

Flaming –

burners off 1.5 13 6.6 2.1

Table 9 Ratios of gas components for specific combustion conditions in steady-state tube furnace tests.

PIR core material

Non-flaming, 350 °C 0.44 - - - Flaming, high ventilation, 700 °C 11 8.7 34 12 Flaming, low ventilation, 825 °C 12 10 - 73 Combustion condition

PIR metal sheet with core material

Non-flaming, 350 °C 0.26 - - - Flaming, high ventilation, 700 °C 24 8.1 13 5.6 Flaming, low ventilation, 825 °C 8.6 8.7 377 80

The ratios for the large-scale tests are quite informative in that there are differences in the smoke gas composition between the different burning periods. A significant difference between the “Smoke” period and the two flaming periods is shown by the CO/THC ratio, which is quite similar for the two flaming periods and much lower for the “Smoke” period. The two flaming periods are different in that the CO/NO ratio is significantly lower in the “Flaming-burners off” period. This again is indicating a somewhat more under-ventilated combustion for the “Flaming” period. One can note that there is no difference for the CO/HCl ratio for the two flaming periods.

(36)

For the PIR core material samples tested with the steady-state tube furnace there is a clear difference in the CO/THC ratios between the non-flaming test conditions (350 °C) and the two flaming test conditions (Flaming, high ventilation,700 °C and Flaming, low ventilation, 825 °C). This shows similarities with the CO/THC ratios from the large-scale tests. However, the sizes of these ratios for the tube furnace tests are higher compared with the ratios from the large-scale tests, especially for the flaming combustion conditions. A possible explanation is that the flaming periods in the large-scale tests include the products of pyrolysis and thus a higher proportion of unburned hydrocarbons (THC).

The CO/HCN ratios from the tube furnace tests were all similar for the flaming tests conditions and for the two types of sample species. Compared to the CO/HCN ratios found from the large-scale tests these ratios were of the same magnitude. Again, it is clear that the smoke gas composition in the “Smoke” period in the large-scale tests does not match the composition found from the non-flaming (350 °C) tube furnace tests as HCN is found in the large-scale tests, and not from the tube furnace.

The CO/NO ratio could provide some additional information on the ventilation conditions for the combustion. In the tube furnace tests, NO was only produced in significant amounts in the well-ventilated tests. The ratio was lower in the case of the metal sheet sample specimen. The CO/NO ratio in the “Flaming” period of the large-scale tests was of the same magnitude as the ratios from the well-ventilated flaming tests with the two types of sample specimen tested with the tube furnace, indicating generally well-ventilated combustion conditions. The ratio was lower in the “Flaming – burner off” period which again indicates a more well-ventilated combustion for this period.

The last ratio presented in the tables above is the CO/HCl ratio. In many cases the production of HCl is rather insensitive to the ventilation conditions, and yields of similar magnitude are seen from both well-ventilated and under-ventilated combustion. That was also seen here for the PIR material. The significantly higher CO/HCl ratios seen in the tube furnace tests for low ventilation conditions are thus a result of the increase in the CO-yields. In the large-scale tests there is no significant difference between the CO/HCl ratios between the two flaming periods in the tests. This indicates that the CO-yield would have been rather constant between the flaming periods in the large-scale tests.

5.2.2 Toxicity weighting of gas components

An alternative objective for a comparison is to investigate whether the information on toxicity from a small-scale test predicts the nature of the toxicity in the effluents from a large scale fire test. A qualitative comparison could be made between the toxicity weighted relative importance of toxic fire gases produced between the two scales.

A relative ranking of the toxicity weighted molar production rates is given below for the different periods in the large-scale tests and for the different combustion conditions simulated in the steady-state tube furnace tests. In this case the weighting factors used are taken from two different sources. The first type of weighting factors used are LC50-values for rat mortality taken from ISO 13344 [8] where the results are presented in Table 10 and Table 11. The second type of weighting factors used are IDLH-valuesii which describe severe effects from exposure on human [9]. In this case the results of the

ii

IDLH is an acronym for Immediately Dangerous to Life or Health, and is defined by the US National Institute for Occupational Safety and Health (NIOSH) as exposure to airborne contaminants that is "likely to cause death or immediate or delayed permanent adverse health effects or prevent escape from such an environment”.

(37)

rankings are presented in Table 12 and Table 13. Both sets of reference values are for an exposure time of 30 minutes. Note that toxic products only found in one of the three large-scale tests, during a specific period, are not included in Table 10 and Table 12.

Table 10 Ranking of toxicity weighted production rates for specific periods in the large-scale tests. Toxicity weighting based on LC50 for rats (ISO 13344).

PIR core panel tests

Ranking of Toxicity weighted gas components (ISO 13344) 1 2 3 4 Smoke HCN CO - - Flaming HCN NO CO HCl Flaming – burners off NO HCN CO HCl

Table 11 Ranking of toxicity weighted production rates for specific combustion conditions in steady-state tube furnace tests. Toxicity weighting based on LC50 for rats

(ISO 13344).

PIR core material

Ranking of Toxicity weighted gas components (ISO 13344)

1 2 3 4 Non-flaming, 350 °C CO - - - Flaming, high ventilation, 700 °C HCN CO NO HCl Flaming, low ventilation, 825 °C HCN CO HCl - Combustion condition

PIR metal sheet with core material

Ranking of Toxicity weighted gas components (ISO 13344)

1 2 3 4 Non-flaming, 350 °C CO - - - Flaming, high ventilation, 700 °C HCN NO CO HCl Flaming, low ventilation, 825 °C HCN CO NO HCl

Comparison of fire effluent composition between large-scale and small-scale tests with sandwich panels

with sandwich panels

Per Blomqvist and Patrik Johansson

S

P

T

ec

hn

ic

al

R

es

ea

rc

h

In

st

itu

te

o

f S

w

ed

en

Comparison of fire effluent composition

between large-scale and small-scale

tests with sandwich panels

Abstract

Contents

Acknowledgments

Sammanfattning (in Swedish)

1

Introduction

2

Sandwich wall panel products

3

Analysis of fire effluent leakage in vertical

furnace fire resistance tests

3.1

Fire resistance tests

3.1.1

Test set-up

3.1.2

Summary of events in the tests

3.2

Results on gas components

3.2.1

MW core panel

3.2.2

PIR core panel

3.3

Leakage rates and mass consumption

4

Steady-state tube furnace tests

4.1

Description of tests

4.1.1

Steady-state tube furnace

4.1.2

Sample preparation

4.1.3

Testing scheme

4.2

Results of the steady-state tube furnace tests

4.2.1

Burning behaviour of PIR samples

4.2.2

Combustion products from PIR

4.2.3

Burning behavior of MW samples

4.2.4

Combustion products from MW

4.3

Discussion of applicability

5

Qualitative comparison of gas composition

between test scales

5.1

Comparison principles

5.2