You are on page 1of 13

Journal of Fire Sciences http://jfs.sagepub.

com/

Using plate thermometer measurements to calculate incident heat radiation


Andreas Hggkvist, Johan Sjstrm and Ulf Wickstrm Journal of Fire Sciences published online 21 November 2012 DOI: 10.1177/0734904112459264 The online version of this article can be found at: http://jfs.sagepub.com/content/early/2012/11/20/0734904112459264 A more recent version of this article was published on - Mar 7, 2013

Published by:
http://www.sagepublications.com

Additional services and information for Journal of Fire Sciences can be found at: Email Alerts: http://jfs.sagepub.com/cgi/alerts Subscriptions: http://jfs.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

Version of Record - Mar 7, 2013 >> OnlineFirst Version of Record - Nov 21, 2012 What is This?

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Article

Using plate thermometer measurements to calculate incident heat radiation


ggkvist1, Johan Sjo stro m1 Andreas Ha m1,2 and Ulf Wickstro
Date received: 28 May 2012; accepted: 5 August 2012

Journal of Fire Sciences 0(0) 113 The Author(s) 2012 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0734904112459264 jfs.sagepub.com

Abstract The plate thermometer is a device used mainly to measure temperatures in fire resistance tests according to ISO 834-1 and EN 1363-1 and to measure the so-called adiabatic surface tempera_ i0nc ) as a simpler, more ture. However, it can also be used to measure incident radiant heat flux (q robust and less-expensive alternative to water-cooled heat flux meters. The accuracy of the mea_0 sured q inc is subject to simplifications in the heat transfer analysis model and uncertainties of parameters such as convective heat transfer coefficients, emissivities and ambient gas temperatures. This study investigates the accuracy of the model itself, isolated from the uncertainties of the physical surrounding, by comparing a simple one-dimensional model to the results of finite element modelling. The so-obtained model includes a heat transfer coefficient due to heat losses of the plate thermometer, found to be KPT = 8 W/m2 K and a heat storage lumped heat capacity CPT = 4200 J/m2 K for an ISO/EN standard plate thermometer. The model is also compared to real field experiments. Keywords Fire, plate thermometer, incident heat radiation, irradiance, heat flux meters, measuring instrument, thermal exposure

Introduction
The plate thermometer (PT), developed by Wickstro m1 for measuring high temperatures in fire resistance furnaces, is a very simple and robust instrument. It has replaced traditional
1 2

s, Sweden SP Technical Research Institute of Sweden, Bora Technical University, Lulea , Sweden Dept. of Civil, Environmental and Natural Resources Engineering, Lulea

Corresponding author: s, Sweden. stro m, SP Technical Research Institute of Sweden, Box 857, 501 15 Bora Johan Sjo Email: johan.sjostrom@sp.se

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Journal of Fire Sciences 0(0)

Figure 1. Sketch of the PT. The exposed area consists of a 100 3 100-mm metal sheet. Behind, it is a 10mm-thick insulation pad. Additional metal is folded over the insulation.
PT: plate thermometer.

thermocouples for the purpose of harmonising the thermal exposure when testing according to the European standard EN 1363-1 as well as the international standard ISO 834. According to the standards, the exposed surface of the PT consists of a 0.7-mm-thick, 100 3 100-mm nickel alloy sheet folded around a 100 3 100-mm and 10-mm-thick inorganic insulation material with a density of 280 (630) kg/m3 (see Figure 1). The temperature of the PT is measured by a type K thermocouple fixed to the geometrical centre of the sheet. During recent years, the PT has also been used to measure the so-called adiabatic surface temperature (AST), which is defined as the temperature of a perfectly insulated surface.2,3 Even though the analysis of extracting AST is related to the study presented here, this study does not consider AST. Instead, it focuses on using the PT to measure incident radiant heat _ i0nc ), which enables a simpler, more robust and less-expensive instrument compared to flux (q conventional water-cooled heat flux meters (HFMs).4,5 Thus, it offers the possibility to measure heat flux in very tough environments and in field tests where HFMs are unpractical to _0 use. The PT has successfully measured q inc in various configurations, such as the cone calorimeter,4 pool fires,6 burning goods7 and model scale fire scenarios,8 by using correction terms _ 99inc shall be for different heat losses in/from the PT. When the incident radiation heat flux q derived from PT temperatures, two parameters depending on the thermal properties of the PT have to be determined. This study presents a way of determining these parameters. One of the parameters considers the heat losses by conduction and the other the heat capacity of the PT. The magnitudes of the parameters have been obtained by comparisons against a two-dimensional finite element model (FEM) where thermal conditions are well defined and against experimental heat exposures. The results enable us to analyse the errors induced by _ i0nc the model separately and thereafter test its performance in real fires. The accuracy of q measurements using PTs is not only subject to the use of suitable correction terms but also responds to uncertainties in emissivity, convective heat transfer, gas temperatures and so on.

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Ha ggkvist et al.

One-dimensional model of the PT


_ 99tot to a surface from the gas phase is characterised by the folThe total net heat transfer q lowing three physically independent terms: (a) absorbed radiation, (b) emitted radiation and (c) convective heat transfer. These are balanced by the heat transferred by conduction into _ 99inc . the material. The absorbed radiation is a portion, defined by the emissivity, e, of the q The emitted radiation is proportional to the surface temperature (to the fourth power) and e. The magnitude of the convective heat transfer is usually simplified to being directly proportional to the temperature difference between the surface and the adjacent gas, TN, where the proportionality constant is the convective heat transfer coefficient, h. In the simplified model presented here, a lumped heat is assumed, that is, the PT is assumed to have a uniform temperature (lumped heat capacity) as indicated by the thermocouple attached to its metal surface. Heat is then transferred to the surface from the environment by radiation and convection. In addition, however, two correction terms are introduced. One is due to heat losses to the rear side of the PT and the other is due to thermal inertia, that is, heat is stored in the PT. For the loss term, the effects of conduction through the insulation as well as the in-plane conduction in the metal that overlaps the insulation, including the corresponding convective cooling, are linearised to be proportional to the temperature difference between the plate surface and the ambient gas temperature. Thus, it is assumed that the combined effects of conduction in insulation and the folded metal with its associated convective cooling can be treated via a temperature-independent parameter, KPT. The correction for the losses is thus given as
_0 q loss = KPT TPT T 1

where TPT is the temperature of the metal sheet of the PT. In addition, during transient conditions, heat is absorbed when increasing the temperature of the metal and the insulation pad. The lumped heat capacity of the metal of the PT is then added to a portion of the heat capacity of the insulation. Thus, the temperature distribution through the PT is assumed uniform in the metal and part of the insulation as shown in Figure 2. The storage correction term for the heat flux associated with this lumped specific heat is therefore
_s q 0tor = cmet rmet dmet + bcins rins dins dTPT dTPT = CPT dt dt 2

where c is specific heat capacity, r is the density and d is the thickness of the materials. The subscripts met and ins refer to metal and insulation, respectively, and b is the portion of the insulation that is considered for the lumped heat capacity. Given these two parameters, KPT and CPT, the heat balance at the PT surface can be written as
dTPT 4 _0 ePT q inc s TPT + hT TPT + KPT T TPT = CPT dt 3

The incident radiant heat flux can then be calculated on the basis of temperature measurements of PT and the adjacent gas
4 _ i0nc = sTPT + q
PT h + KPT TPT T + CPT dT dt ePT

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Journal of Fire Sciences 0(0)

Figure 2. Assumed temperature distribution through the PT to calculate (a) loss correction and (b) storage correction, consisting of the exposed metal surface (white) and the insulation (shaded). (a) The temperature is assumed uniform in the metal and linear in the insulation. (b) Lumped heat capacity is assumed in the metal and in a part of the insulation pad.
PT: plate thermometer.

As KPT is added to the heat transfer coefficient h, the accuracy of the model depends on the precision of the sum of the two. For steady-state conditions, only the loss correction term deter_ i0nc , while the storage correction term dominates when the temperature of the PT changes mines q rapidly. The combined importance of the correction terms is indicated in Figure 7. _ i0nc can be obtained by approximating the time derivative of Under transient conditions, q the measured TPT. For this study, central difference is used in order to obtain the incident flux, given the timetemperature output from the PT
_0 q inc = s
i

4 i TPT

i + 1 TPT i1 (h + KPT ) TPT i T i + CPT TPT ti + 1 ti1 ePT

where i represents the ith time step for which a new TPT and T are probed. Note that when the ambient gas temperature is close to the PT temperature as in a furnace, the influence of the conduction losses is negligible, and only the influence of the inertia needs to be considered. It is important to probe the gas temperature close to the PT. However, for measurements beyond the flames or any hot smoke layer, the gas temperatures are often very close to ambient air; h can be calculated from standard textbooks. It is estimated to be approximately 10 W/m2 K for natural convection.4 For other flow cases, it depends on gas velocities and may be difficult to estimate. This causes uncertainties, which increase with the difference between the gas temperature and the PT temperature. Similar uncertainties occur for watercooled HFM. Then, the uncertainty increases with the difference between the gas temperature and the temperature of the sensor, which is normally just over the cooling water temperature. Therefore, a HFM is expected to yield accurate temperature when placed in gases of ambient temperature while the uncertainties are much greater when placed in, for example, hot flames. In such cases, the PT is likely to yield more accurate results.

Validation to FEM
The parameters KPT and CPT of the one-dimensional model are obtained by comparing a two-dimensional FEM of the PT, using the FEM software Temperature Analysis of Structures Exposed to Fire (TASEF)9. In the first step, KPT is obtained by a steady-state

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Ha ggkvist et al.

analysis, and CPT is obtained by transient analyses. Half the PT is modelled with a line of symmetry in the middle. The insulation pad and metal sheet are divided into 35 and 17 square cells, respectively, as shown in Figure 3. In the modelling, the exposed left surface of the PT (between nodes 1 and 9) is subject to uniform incident radiant heat fluxes of 6.12, 33.0, 107 and 267 kW/m2. These represent the emitted radiation flux from a black surface with temperatures of 300C, 600C, 900C and 1200C, respectively. The surfaces of the PT are exposed to convective cooling to the ambient atmosphere with temperature TN = 20C using a heat transfer coefficient of 10 W/m2 K and emitted radiation using an emissivity e = 0.8. The material properties are chosen to represent two common materials for PTs. Inconel 600 is most commonly used as a metal sheet. Its thermal properties (conductivity, density and specific heat) as a function of temperature are given by Special Metals Corporation.10 A common material for the insulation pad is Carbowool, Unifrax. Its properties were measured at SP using the transient plane source (TPS) method,11 according to the ISO 22007-212 for determining thermal conductivity and diffusivity.5 The temperature-dependent properties used in the FEM is given by the expression k = 0.0002T + 0.0057 and rc = 2324T + 586,100, where k is given in Watt per metre Kelvin, T in Kelvin and rc in Joule per cubic metre Kelvin.

Steady-state conditions
The heating of the PT exposed to the four different radiation levels is modelled with the twodimensional FEM until steady state was reached. The steady-state temperatures of the PT (node 10 in Figure 5) are 463, 784, 1107 and 1423 K for the radiation levels of 6.12, 33, 107 _ 99inc can then be calculated for different and 267 kW/m2, respectively. Using equation (5), q values of KPT. These calculated values, compared to the actual radiation used in the FEM simulation, are shown in Figure 4. The optimum values of KPT are different depending on the incident radiation level and thereby the temperature level of the PT. Optimisation for low radiation levels yields KPT values around 7 W/m2 K, while the highest radiation level used in this study yields an optimal KPT of 15 W/m2 K. For high radiation levels, the relative error is, however, small, within 6% for any correction term between 5 and 15 W/m2 K. For the lowest radiation level, the relative error is large for high correction terms, but the absolute error is still less than 2 kW/ m2 for any correction term below 15 W/m2 K. Optimising the combined relative error for all radiation levels used here, one finds an optimum value just over 8 W/m2 K (see Figure 5). _ 99inc within 5% of the radiation used as input in the FEM for This value yields calculated q all radiation levels.

Transient conditions
To estimate the effective value of the inertia (specific heat), the transient part of the simulations is investigated. Using different portions (b) of the insulation contributing to the lumped heat capacity, the errors develop differently over the transient condition. A large value yields large error at short exposure times since the model then greatly overestimates the heat stored by the insulation. A low value, on the other hand, underestimates this contribution at times when the heat penetrates further into the insulation. At steady state, the heat capacity is irrelevant. Figure 6 shows the relative error of calculated radiation according to equation (5) using different values of CPT.

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Journal of Fire Sciences 0(0)

Figure 3. Symmetric model of the PT used in the TASEF FEM analysis (illustration is not in scale).
PT: plate thermometer; TASEF: Temperature Analysis of Structures Exposed to Fire; FEM: finite element model.

According to Eurocode 3, when analysing fire resistance of heavily insulated steel structures, the best fit is achieved when using a lumped heat capacity considering a third of the actual insulation heat capacity (see also Ref. 13). Also in this study, using b = 1/3 gives

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Ha ggkvist et al.

_0 Figure 4. Calculated q inc (solid lines) according to equation (5), given the steady-state TPT from the FEM simulation for different values of KPT. The radiation levels used in the FEM are indicated by dashed lines.
FEM: finite element model.

_0 Figure 5. Relative errors of calculated q inc according to equation (5), given the steady-state TPT from the FEM simulation for different values of KPT.
FEM: finite element model.

acceptable results with errors around 20% at the very start of the transient period but within 12% after 20 s for an incident radiation of 33 kW/m2 as shown in Figure 6. A third of the insulation thickness yields a total lumped heat capacity for the PT of CPT = 4200 J/m2 K.

Validation against experiments


To investigate the effect of the two correction terms, a PT is exposed to a radiant panel with a constant temperature. The incident radiant heat flux is measured using a HFM (indicating

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Journal of Fire Sciences 0(0)

_0 Figure 6. Relative error of calculated q inc using the simple one-dimensional model during the transient heating for various values of b. For all curve, the correction term uses KPT = 8 W/m2 K.

15 kW/m2) but also using a PT. The radiation is introduced as a step function by removing _ 99inc , withan insulation board shielding the panel. Figure 7 shows the results. Calculating q out the correction terms, only considering radiation from the exposed surface and convective cooling, yields a response of several minutes and a final level well below the result of the HFM. Introduction of the correction for the loss, using KPT = 8 W/m2 K, yields a correspondence between PT and HFM within 1 kW/m2 for the steady state, but still there is a significantly slower response of the PT at almost 5 min. However, by also introducing the storage correction term, using CPT = 4200 J/m2 K, the agreement between PT and HFM is very good over both transient and steady-state period. Ingason and Wickstro m4 describe several experiments where different sets of parameters are used. Good agreement between HFM and PTs has been demonstrated using KPT between 5 and 22 W/m2 K and CPT = 2610 J/m2 K. However, the outcome is not only dependent on the choice of KPT and CPT but also on the assumed values of h, e and TN. Using the model described here and the set of parameters that are optimised against the FEM (KPT = 8.4 W/m2 K; CPT = 4200 J/m2 K), Lo nnermark and Ingason8 compared HFM and PTs measuring incident radiation 0.5 and 2 m from a 1:10 model scale fire in an industrial building. The agreement between the two methods was mostly within 10% as shown in Figure 8. Several experiments were conducted with different fuels and different openings of the building, and all showed good agreement between HFM and PTs.8 _ i0nc has been measured in the room/corner test configuration as defined by Furthermore, q ISO 9705. Details of the test can be found in Ref. 5. Figure 9 shows comparison of HFM and PTs situated on the floor level of 0.9 and 1.2 m from the short wall on which a gas burner with a constant output of 450 kW was situated. At first, the PTs and HFM agree very well, but the difference grows during the 10-min long exposure to a maximum difference of ;2.5 kW/m2. However, it should be noted that the gas temperature in the experiment, measured close to the PT, increased to between 100C and 150C during the exposure. The effect of the convective heat transfer to the PT for this situation is corrected for according to equation (5) using continuously probed gas temperatures and h = 10 W/m2 K for natural convection as discussed in Ref. 4. For the HFM, the convective heat transfer is harder to estimate and not corrected for in the data of the HFM when plotting the heat fluxes measured with HFMs and PTs, respectively, as shown in Figure 9. This effect, which was estimated by

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Ha ggkvist et al.

_0 Figure 7. q inc from a radiant panel measured by HFM and PT using no correction term (thin dashed), only correction for the loss term (dotted) and both loss and storage correction (solid line).
HFM: heat flux meter; PT: plate thermometer.

Figure 8. Calculated radiation from PT measurements using equation (13) with KPT = 8.4 W/m2 K and CPT = 4200 J/m2 K together with HFM data: (a) 0.5 m distance and (b) 2 m distance.
Source: From Ref. 7. HFM: heat flux meter; PT: plate thermometer.

Ha ggkvist5 to be a few kilowatt per square metre, clearly influences the HFM to overestimate the incident radiation and is therefore a contribution to the differences between the HFM and PT results in Figure 9. This situation highlights that the PT can be used to measure heat flux also in situations of hot surrounding gas with a high convective heat transfer where the HFM is inappropriate to use.5 The PT results take the increasing surrounding gas temperature into account according to equation (5), but the HFM is used without any corrections.

Discussion
This study has focused on the accuracy of a simplified method of calculating incident radiant heat flux based on PT measurements. The first correction term used is an effective heat

Downloaded from jfs.sagepub.com by guest on January 14, 2014

10

Journal of Fire Sciences 0(0)

_0 Figure 9. q inc measured by HFM and PT on the floor in the room corner configuration. The indices 4 and 5 correspond to 0.9 and 1.2 m from the gas burner, respectively.
HFM: heat flux meter; PT: plate thermometer.

transfer coefficient that accounts for the losses by conduction through the insulation pad and in the plane of metal folded over the insulation. This term is added to the convective heat transfer coefficient, and since these are rarely known within accuracy of 1 W/m2 K, a more detailed optimisation of the correction term is not motivated in an experimental situation. The total error due to convection and conduction uncertainties can be studied from Figure 5. For high radiation levels, 33 kW/m2 in the example presented here, the error from the conduction is below 4% compared to the numerical model, and adding an uncertainty of the convective heat transfer coefficient of 64 W/m2 K still yields accuracy well within 10%. The better accuracy of the higher thermal exposure is due to the relative importance of radiation, which increases dramatically compared to the losses from convection and conduction at high temperatures. For radiation levels up to about 6 kW/m2 K, usually well below levels relevant to fire spread scenarios, the combined relative error can become quite large, but in absolute values, this error is very small. In general, a best fit is obtained for KPT = 8 W/m2 K. The second correction term is the lumped heat capacity of the PT, which is relevant for transient conditions. This term is deemed to yield optimum corrections when incorporating the metal plate facing the heat source plus a third of the insulation behind it, which yields CPT = 4200 J/m2 K. The present and previous studies show that the PT can be used to measure incident radiation with accuracy within 5%. This can be compared to the accuracy of a newly calibrated HFM (according to ISO 14934), which under ideal conditions has a relative error of about 3% at low radiation levels and 1%2% for the highest radiation levels. The robustness of the PT, together with its correction for convective heat losses as an integral part of its calculation routine, makes it well suited to measurements in situations with elevated ambient gas temperatures where convective heating is substantial. In such applications, the robustness of the PT is a great advantage. However, the standard PT was developed to measure temperature in fire resistance furnaces, and to really improve the measurements of heat flux in ambient air, some modifications of the PT are suggested. First, using a thicker insulation and by cutting off most of the metal folded around the insulation, only keeping enough to maintain mechanical stability, the conduction (in-plane metal and through insulation) could be considerably reduced. In

Downloaded from jfs.sagepub.com by guest on January 14, 2014

Ha ggkvist et al.

11

addition, the heat capacity of the PT could be reduced by using an insulation pad with a lower density and a thinner metal sheet.

Conclusions
The PT is able to measure incident heat radiation from fires. This study presents a model suitable for a wide range of radiation levels correcting for the convective heat transfer, the losses to the rear side and the heating of the material. This model is analysed against a FEM where external conditions and material properties are known. The correction terms of KPT = 8 W/m2 K and CPT = 4200 J/m2 K are suggested for accurate calculations of the incident radiation. For steady state, the errors induced solely by the assumptions of the onedimensional model itself are limited to within 4% for high radiation levels. For levels below those usually considered when studying flame spread and ignition, the errors from the model are expected to be limited to 10% under steady-state conditions. When placed in gases at the same temperature as the corresponding black-body radiation temperature, only corrections due to inertia need be considered. The momentary incident radiation can be obtained from PT measurements using equation (5). The corresponding accuracy of water-cooled HFMs is relatively high when placed in air of ambient temperature (temperature of cooling water), while the accuracy when placed in hot fire gases or flames is likely to be very poor. For such applications, more research is needed to be able to present quantitative values. Funding
Technical University. This project was internally funded by SP and Lulea

Acknowledgement
The authors are grateful to Dr A. Lo nnermark and Prof. H. Ingason for providing them with experimental data.

References
1. Wickstro m U. The plate thermometer a simple instrument for reaching harmonized resistance tests. Fire Technol 1994; 30: 195208. 2. Duthinh D, McGrattan K and Khaskia A. Recent advances in fire-structure analysis. Fire Safety J 2008; 43: 161167. 3. Wickstro m U. The adiabatic surface temperature and the plate thermometer. Fire Safety Sci 2011; 10: 10011011. 4. Ingason H and Wickstro m U. Measuring incident radiant heat flux using the plate thermometer. Fire Safety J 2007; 42: 161166. 5. Ha ggkvist A. The plate thermometer as a mean of calculating incident heat radiation a practical and theoretical study. NR 2009:183, 2009. Sweden: Lulea University of Technology, ISSN: 1402-1617. 6. Arvidsson M and Ingason H. Measurement of the efficiency of a water spray system against diesel oil pool and spray fires. SP report 2005:33, 2005. Sweden: SP Technical Research Institute of Sweden. 7. Lo nnermark A and Ingason H. Fire spread in large industrial premise and warehouses. SP report 2005:21, 8. 2005. Sweden: SP Technical Research Institute of Sweden. Lo nnermark A and Ingason H. Fire spread between industry premises. SP report 2010:18, 2010. Sweden: Technical Research Institute of Sweden. Sterner E and Wickstro m U. TASEF Temperature Analysis of Structures Exposed to Fire users manual. SP report 1990:05, 1990. Sweden: Technical Research Institute of Sweden, ISBN: 91-7848-210-0. Special Metals Corporation. Technical Bulletin Inconel alloy 600, http://www.specialmetals.com/products/ inconelalloy600.php (accessed 11 March 2012). Gustafsson SE. Transient plane source techniques for thermal conductivity and thermal diffusivity measurements of solid materials. Review of Scientific Instruments 1991; 62: 797804. ISO 22007-2:2008. Plastics determination of thermal conductivity and thermal diffusivity Part 2: transient plane heat source (hot disc) method. Wickstro m U. Temperature analysis of heavily insulated steel structures. Fire Safety J 1985; 9: 281285.

9.

10.

11.

12.

13.

Downloaded from jfs.sagepub.com by guest on January 14, 2014

12

Journal of Fire Sciences 0(0)

Appendix 1
Nomenclature

Roman capital letters


CPT KPT TPT TN Lumped heat capacity of plate thermometer (PT) (J/m2 K) Effective heat transfer coefficient due to conduction (W/m2 K) Temperature measured by PT (K) Ambient temperature (K)

Roman lower-case letters


cA dA h _0 q inc _0 q loss _0 q stor Specific heat capacity for material A (J/kg K) Thickness of material A (m) Convective heat transfer coefficient (W/ m2 K) Incident radiation heat flux to a surface (W/m2) Heat flux lost at the non-exposed surfaces of the PT (W/m2) Heat stored per unit area and time (W/m2)

Greek lower-case letters


ePT rA s Emissivity of PT surface () Density of material A (kg/m3) StefanBoltzmanns constant (= 5.67 3 1028) (W/m2 K4)

Author Biographies
ggkvist is presently working as a fire protection engineer at Halmstads Fire and Rescue Andreas Ho Service. His Master thesis, conducted at SP, concerned measurements of incident heat radiation using PT. stro m is a senior scientist at SP Technical Research Institute of Sweden. His work focuses on Johan Sjo thermal properties of materials exposed to fire and measurement techniques in fire sciences. m is part time professor at Lulea Technical University, Lulea , and part time senior Ulf Wickstro s. scientist at SP, Bora

Downloaded from jfs.sagepub.com by guest on January 14, 2014

You might also like