ASTM E3057-19

Standard Test Method for Measuring Heat Flux Using Directional Flame Thermometers with Advanced Data Analysis Techniques

SIGNIFICANCE AND USE
5.1 Need for Heat Flux Measurements:  
5.1.1 Independent measurements of temperature and heat flux support the development and validation of engineering models of fires and other high environments, such as furnaces. For tests of fire protection materials and structural assemblies, temperature and heat flux are necessary to fully specify the boundary conditions, also known as the thermal exposure. Temperature measurements alone cannot provide a complete set of boundary conditions.  
5.1.2 Temperature is a scalar variable and a primary variable. Heat Flux is a vector quantity, and it is a derived variable. As a result, they should be measured separately just as current and voltage are in electrical systems. For steady-state or quasi-steady state conditions, analysis basically uses a thermal analog of Ohm's Law. The thermal circuit uses the temperature difference instead of voltage drop, the heat flux in place of the current and thermal resistance in place of electrical resistance. As with electrical systems, the thermal performance is not fully specified without knowing at least two of these three parameters (temperature drop, heat flux, or thermal resistance). For dynamic thermal experiments like fires or fire safety tests, the electrical capacitance is replaced by the volumetric heat capacity.  
5.1.3 The net heat flux, which is measured by a DFT, is likely different than the heat flux into the test item of interest because of different surface temperatures. An alternative measurement is the total cold wall heat flux which is measured by water-cooled Gardon or S-B gauges. The incident radiative flux can be estimated from either measurement by use of an energy balance [Keltner, 2007 and 2008 (16, 17)]. The convective flux can be estimated from gas temperatures and the convective heat transfer coefficient, h [Janssens, 2007 (18)]. Assuming the sensor is physically close to the test item of interest; one can use the incident radiative and convective fluxes from the...
SCOPE
1.1 This test method describes the continuous measurement of the hemispherical heat flux to one or both surfaces of an uncooled sensor called a “Directional Flame Thermometer” (DFT).  
1.2 DFTs consist of two heavily oxidized, Inconel 600 plates with mineral insulated, metal-sheathed (MIMS) thermocouples (TCs, type K) attached to the unexposed faces and a layer of ceramic fiber insulation placed between the plates.  
1.3 Post-test calculations of the net heat flux can be made using several methods. The most accurate method uses an inverse heat conduction code. Nonlinear inverse heat conduction analysis uses a thermal model of the DFT with temperature dependent thermal properties along with the two plate temperature measurement histories. The code provides transient heat flux on both exposed faces, temperature histories within the DFT as well as statistical information on the quality of the analysis.  
1.4 A second method uses a transient energy balance on the DFT sensing surface and insulation, which uses the same temperature measurements as in the inverse calculations to estimate the net heat flux.  
1.5 A third method uses Inverse Filter Functions (IFFs) to provide a near real time estimate of the net flux. The heat flux history for the “front face” (either surface exposed to the heat source) of a DFT can be calculated in real-time using a convolution type of digital filter algorithm.  
1.6 Although developed for use in fires and fire safety testing, this measurement method is quite broad in potential fields of application because of the size of the DFTs and their construction. It has been used to measure heat flux levels above 300 kW/m2 in high temperature environments, up to about 1250 °C, which is the generally accepted upper limit of Type K or N thermocouples.  
1.7 The transient response of the DFTs is limited by the response of the MIMS TCs. The larger the thermocouple the slower the transient response. Res...

General Information

• Status: Published
• Publication Date: 31-May-2019
• ICS: 17.200.10 - Heat. Calorimetry
• Technical Committee: E21 - Space Simulation and Applications of Space Technology
• Drafting Committee: E21.08 - Thermal Protection

Relations

• Replaces: ASTM E3057-16 - Standard Test Method for Measuring Heat Flux Using Directional Flame Thermometers with Advanced Data Analysis Techniques
• Effective Date: 01-Jun-2019
• Refers: ASTM E176-24 - Standard Terminology of Fire Standards
• Effective Date: 01-Jan-2024
• Refers: ASTM E119-19 - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Oct-2019
• Refers: ASTM E176-18a - Standard Terminology of Fire Standards
• Effective Date: 15-Dec-2018
• Refers: ASTM E119-18c - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Nov-2018
• Refers: ASTM E119-18ce1 - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Nov-2018
• Refers: ASTM E119-18b - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Sep-2018
• Refers: ASTM E119-18a - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Jun-2018
• Refers: ASTM E176-18 - Standard Terminology of Fire Standards
• Effective Date: 01-Mar-2018
• Refers: ASTM E119-18 - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Mar-2018
• Refers: ASTM E2683-17 - Standard Test Method for Measuring Heat Flux Using Flush-Mounted Insert Temperature-Gradient Gages
• Effective Date: 01-Sep-2017
• Refers: ASTM E1529-16 - Standard Test Methods for Determining Effects of Large Hydrocarbon Pool Fires on Structural Members and Assemblies
• Effective Date: 01-Nov-2016
• Refers: ASTM E119-16a - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Jul-2016
• Refers: ASTM E119-16 - Standard Test Methods for Fire Tests of Building Construction and Materials
• Effective Date: 01-Apr-2016
• Refers: ASTM E176-15ae1 - Standard Terminology of Fire Standards
• Effective Date: 01-Aug-2015

Overview

ASTM E3057-19 outlines the Standard Test Method for Measuring Heat Flux Using Directional Flame Thermometers (DFTs) with Advanced Data Analysis Techniques. Developed by ASTM International, this standard provides comprehensive procedures for the continuous measurement of hemispherical heat flux using uncooled sensors, known as Directional Flame Thermometers, in demanding environments such as fires, furnaces, and industrial heat exposure scenarios. Accurate heat flux measurements are essential for fire safety engineering, validation of computational models, and assessment of fire protection materials and structural assemblies.

Unlike conventional water-cooled Gardon or Schmidt-Boelter gauges, DFTs are robust, require no cooling, and are less susceptible to fouling by soot or condensation, making them suitable for harsh and high-temperature environments. The standard describes multiple advanced data analysis techniques for deriving precise heat flux results from sensor data, supporting both research and real-world applications.

Key Topics

• Directional Flame Thermometer (DFT) Design

  • Two heavily oxidized Inconel 600 plates
  • Mineral insulated, metal-sheathed (MIMS) type K thermocouples
  • Ceramic fiber insulation between plates
  • No calibration required due to advanced data analysis

• Measurement Principles

  • Simultaneous measurement of temperature and derived heat flux
  • Heat flux distinguished as a vector (directional) quantity
  • Data reflects transient and steady-state conditions

• Advanced Data Analysis Techniques

  • Nonlinear inverse heat conduction analysis for transient measurements
  • Energy balance methods utilizing temperature readings
  • Real-time estimation using inverse filter functions (IFFs) and digital filtering

• Practical Implementation

  • Guidance on mounting, thermocouple installation, and thermal modeling
  • Recommendations for ensuring minimal disturbance to the environment under test
  • Procedures for data acquisition system validation and thermocouple accuracy

• Uncertainty Considerations

  • Emphasis on independent measurement of temperature and heat flux
  • Recommendations for reporting uncertainty and quality of thermal analysis

Applications

• Fire Testing and Fire Safety Engineering
  Directional Flame Thermometers are especially useful in fire safety testing of materials, large-scale pool fires, building construction, and fire-resistance-rated assemblies. Their design makes them resistant to failure in environments where soot and condensation present major challenges to traditional water-cooled gauges.

• Validation of Engineering Models
  By providing independent temperature and heat flux data, DFTs support accurate boundary condition definitions for computational fluid dynamics (CFD) and other thermal performance models.

• Industrial High Temperature Environments
  Suitable for process industries (e.g., furnaces, kilns) where real-time monitoring of heat flux above 300 kW/m² and temperatures up to about 1250°C is necessary.

• Research and Development
  Used in diverse studies, including wildfire research, marine fire safety, automobile fire scenarios, and hazardous materials transportation simulations.

• Limitations and Recommendations

  • DFTs are best suited where Gardon and Schmidt-Boelter gauges are impractical due to temperature extremes, fouling, or the need for rugged, maintenance-free measurement devices.
  • For environments with predominantly radiative transfer and manageable conditions, traditional gauges may still be preferred.

Related Standards

Numerous related ASTM and international standards complement ASTM E3057-19. Key references include:

• ASTM C177: Steady-State Heat Flux Measurement by Guarded-Hot-Plate Apparatus
• ASTM E119: Fire Tests of Building Construction and Materials
• ASTM E1529: Large Hydrocarbon Pool Fire Effects on Structural Members and Assemblies
• ASTM E511, E457, E459, E2683: Various heat flux measurement methods and terminology
• ISO 834-11:2014: Fire Resistance Tests-Elements of Building Construction
• MNL12-4th Manual on Thermocouple Use: Recommended practices for temperature measurement with thermocouples

Utilizing ASTM E3057-19 and related standards enhances the reliability of fire testing, industrial heat flux measurements, and supports the development of performance-based codes and safety assessments across a wide range of high-temperature applications.

Standards Content (Sample)

This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
Designation: E3057 − 19
Standard Test Method for
Measuring Heat Flux Using Directional Flame Thermometers
with Advanced Data Analysis Techniques
This standard is issued under the fixed designation E3057; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
INTRODUCTION
This test method describes a technique for measuring the net heat flux to one or both surfaces of
a sensor called a Directional Flame Thermometer. The sensor covered by this standard uses
measurements of the temperature response of two metal plates along with a thermal model of the
sensor to determine the net heat flux. These measurements can be used to estimate the total heat flux
(also known as thermal exposure) and bi-directional heat fluxes for use in CFD thermal models.
The development of Directional Flame Thermometers (DFTs) as a device for measuring heat flux
originated because commercially available, water-cooled heat flux gauges (for example, Gardon and
Schmidt-Boelter gauges) did not work as desired in large fire tests. Because the Gardon and
Schmidt-Boelter (S-B) gauges are water cooled, condensation and soot deposition can occur during
fire testing or in furnaces. Both foul the sensing surface which in turn changes the sensitivity
(calibration) of the gauge. This results in an error during data reduction. Therefore, a different type of
sensorwasneeded;onesuchsensorisaDFT.DFTsarenotcooledsocondensationandsootdeposition
are minimized or eliminated.
Additionally, a body of work has shown that for both Gardon and Schmidt-Boelter gauges the
sensitivity coefficients determined through the calibration process, which uses a radiative heat source,
are not the same as the sensitivity coefficients determined if a purely convective source is used for
calibration [Test Method E511-07; Keltner and Wildin, 1975 (1, 2); Borell, G. J., and Diller, T. E.,
1987 (3); Gifford,A., et al., 2010 (4); Gritzo, L.A., et al., 1995 (5); Young, M. F., 1984 (6); Sobolik,
et al., 1987 (7); Kuo and Kulkarni, 1991 (8); Keltner, 1995 (9); Gifford, et al., 2010 (10); Nakos, J.
T., and Brown,A. L., 2011 (11)]. As a result, one can incur significant bias errors when reducing data
in tests where there may be a non-negligible convective component because the only sensitivity
coefficient available is for a radiation calibration. It was desired to reduce/eliminate these potential
sources of error by designing a gauge that does not depend on a radiation only calibration. DFTs have
this characteristic.
A sensor, also called a Directional Flame Thermometer, was developed to help estimate flame
thickness in pool fire tests of hazardous material shipping containers [Burgess, M. H., 1986 (12); Fry,
C. J., 1989 (13); Burgess, M. H., et al., 1990 (14); and Fry, C. J., 1992 (15)].As originally designed,
DFTs were quasi-equilibrium sensors that used a thin metal plate with a single thermocouple attached
and backed by multiple radiation shields. To make a sensor suitable for continuous transient heat flux
measurements, this basic design was modified to use two instrumented plates, with a layer of
insulation in between.
For the Directional Flame Thermometers described in this standard, the net heat flux is calculated
using transient temperature measurements of the two plates and temperature dependent material
properties for the plates and the insulation. Three methods are described in this standard to calculate
the net heat flux. The most accurate method for calculating the net heat flux is believed to be the
1-dimensional,nonlinearinverseheatconductionanalysis,whichusestheIHCP1Dcode.Thisisbased
on uncertainty analyses and comparisons with measurements made with Schmidt-Boelter and Gardon
gauges,whichhaveNISTtraceablecalibrations.Thesecondmethodusestransientenergybalanceson
the DFT. As will be shown below, the energy balance method compares very well with the inverse
method, again based on uncertainty analyses. The third method uses sets of linearized, convolution
digital filters based on IHCP1D.These allow a near real-time calculation of the net heat flux [Keltner,
N.R.,2007 (16);Keltner,N.R.,etal.,2010 (17)].SeeSection1formoredetailedinformationoneach
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
E3057 − 19
analysis technique. Additional information is given in the Annexes and Appendices.
Various DFT designs have been used in a variety of applications including very large pool fires,
LNG spill fires, marine fire safety testing, automobile fires, to study rocket launch accident fires, and
in research of forest and wild-land fires. Appendix X1 provides a comprehensive list of applications
where DFTs have been successfully used.
Advantages of DFTs are their relatively low cost, ease of construction, they require no calibration
(see later), and require no cooling. They are robust and can survive intense fire environments without
failure. Disadvantages include most are large compared with Gardon and S-B heat flux gauges and
because they are not calibrated, one cannot reference the measurements to a NIST standard. Because
no calibration is required, one must quantify the uncertainties present in the temperature measure-
ments and the data reduction methods used to calculate the heat flux. Also, DFTs measure net heat
flux; for a direct comparison with Gardon and S-B gauges, which are calibrated to incident (or “cold
wall”) flux, one must use a thermal model to estimate the incident flux.
The best applications for DFTs are where Gardon and S-B gauges cannot be used (for example, due
to high temperatures, lack of cooling, soot deposition, fouling, and so forth), or when long life and
overall costs are a consideration. Gardon and Schmidt-Boelter gauges are recommended in non-sooty
environments, when it is possible to mount the gauges and cooling lines, and in predominantly
radiative environments with a small convective contribution.
This test method was jointly developed by ASTM Committee E21 on Space Simulation and Applications of Space Technology and is the direct responsibility of
Subcommittee E21.08 on Thermal Protection.
CurrenteditionapprovedJune1,2019.PublishedJuly2019.Originallyapprovedin2016.Lastpreviouseditionapprovedin2016asE3057 – 16.DOI:10.1520/E3057-19.
The boldface numbers in parentheses refer to the list of references at the end of this standard.
1. Scope 300 kW/m in high temperature environments, up to about
1250 °C,whichisthegenerallyacceptedupperlimitofTypeK
1.1 This test method describes the continuous measurement
or N thermocouples.
of the hemispherical heat flux to one or both surfaces of an
1.7 The transient response of the DFTs is limited by the
uncooled sensor called a “Directional Flame Thermometer”
response of the MIMS TCs. The larger the thermocouple the
(DFT).
slower the transient response. Response times of approxi-
1.2 DFTs consist of two heavily oxidized, Inconel 600
mately 1 to 2 s are typical for 1.6 mm diameter MIMS TCs
plates with mineral insulated, metal-sheathed (MIMS) thermo-
attached to 1.6 mm thick plates. The response time can be
couples (TCs, type K) attached to the unexposed faces and a
improved by using a differential compensator.
layer of ceramic fiber insulation placed between the plates.
1.8 The values stated in SI units are to be regarded as
1.3 Post-test calculations of the net heat flux can be made standard. The values given in parentheses after SI units are
provided for information only and are not considered standard.
using several methods. The most accurate method uses an
inverse heat conduction code. Nonlinear inverse heat conduc-
1.9 This standard does not purport to address all of the
tionanalysisusesathermalmodeloftheDFTwithtemperature
safety concerns, if any, associated with its use. It is the
dependentthermalpropertiesalongwiththetwoplatetempera-
responsibility of the user of this standard to establish appro-
ture measurement histories. The code provides transient heat
priate safety, health, and environmental practices and deter-
flux on both exposed faces, temperature histories within the
mine the applicability of regulatory limitations prior to use.
DFT as well as statistical information on the quality of the
1.10 This international standard was developed in accor-
analysis.
dance with internationally recognized principles on standard-
ization established in the Decision on Principles for the
1.4 Asecond method uses a transient energy balance on the
Development of International Standards, Guides and Recom-
DFT sensing surface and insulation, which uses the same
mendations issued by the World Trade Organization Technical
temperature measurements as in the inverse calculations to
Barriers to Trade (TBT) Committee.
estimate the net heat flux.
1.5 A third method uses Inverse Filter Functions (IFFs) to 2. Referenced Documents
provide a near real time estimate of the net flux. The heat flux
2.1 ASTM Standards:
history for the “front face” (either surface exposed to the heat
C177 Test Method for Steady-State Heat Flux Measure-
source) of a DFT can be calculated in real-time using a
ments and Thermal Transmission Properties by Means of
convolution type of digital filter algorithm.
1.6 Although developed for use in fires and fire safety
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
testing, this measurement method is quite broad in potential
contact ASTM Customer Service at [e-mail]. For Annual Book of ASTM
fields of application because of the size of the DFTs and their
Standards volume information, refer to the standard’s Document Summary page on
construction.Ithasbeenusedtomeasureheatfluxlevelsabove the ASTM website.
E3057 − 19
the Guarded-Hot-Plate Apparatus or sensitivity coefficients are not required because alternate
E119 Test Methods for Fire Tests of Building Construction methods of data reduction are used. DFTs are simple to
and Materials fabricate and use, but are more complicated when reducing the
E176 Terminology of Fire Standards data. Gardon and Schmidt-Boelter gauges have relatively
E457 Test Method for Measuring Heat-Transfer Rate Using linear outputs with heat flux and only require a single sensi-
a Thermal Capacitance (Slug) Calorimeter tivity coefficient (for example, xx mv/unit of flux) to convert
E459 Test Method for Measuring Heat Transfer Rate Using the output to an incident heat flux. DFTs have two thermo-
a Thin-Skin Calorimeter couple outputs as a function of time. Those outputs along with
E511 TestMethodforMeasuringHeatFluxUsingaCopper- temperature dependent thermal properties and advanced analy-
Constantan Circular Foil, Heat-Flux Transducer sis techniques are used with a thermal model to calculate the
E1529 Test Methods for Determining Effects of Large Hy- net heat flux.The net heat flux (with an energy balance) can be
drocarbon Pool Fires on Structural Members and Assem- used to estimate the total cold wall heat flux, which is same as
blies the measurement made by Gardon or S-B gauges [Janssens,
E2683 Test Method for Measuring Heat Flux Using Flush- 2007 (18)].
Mounted Insert Temperature-Gradient Gages
2.2 Other Standards: 5. Significance and Use
ISO 834-11:2014 Fire ResistanceTests—Elements of Build-
5.1 Need for Heat Flux Measurements:
ing Construction—Part 11: Specific Requirements for the
5.1.1 Independent measurements of temperature and heat
Assessment of Fire Protection to Structural Steel Ele-
flux support the development and validation of engineering
ments
models of fires and other high environments, such as furnaces.
IMO A754 Fire Resistance Tests: Fire Safety Onboard
For tests of fire protection materials and structural assemblies,
Ships
temperature and heat flux are necessary to fully specify the
2.3 Other ASTM Document:
boundary conditions, also known as the thermal exposure.
MNL12-4th Manual on the Use of Thermocouples in Tem-
Temperature measurements alone cannot provide a complete
perature Measurement, Fourth Edition, 1993, Sponsored
set of boundary conditions.
by ASTM Committee E20 on Temperature Measurement
5.1.2 Temperature is a scalar variable and a primary vari-
able.HeatFluxisavectorquantity,anditisaderivedvariable.
3. Terminology
As a result, they should be measured separately just as current
3.1 Definitions—Refer to Terminology E176 for definitions
and voltage are in electrical systems. For steady-state or
of some terms used in these test methods.
quasi-steady state conditions, analysis basically uses a thermal
3.2 Definitions of Terms Specific to This Standard:
analogofOhm’sLaw.Thethermalcircuitusesthetemperature
3.2.1 incident radiative heat flux (irradiance; q ),
inc,r
difference instead of voltage drop, the heat flux in place of the
n—radiative heat flux impinging on the surface of the DFT or
current and thermal resistance in place of electrical resistance.
the unit under test.
Aswithelectricalsystems,thethermalperformanceisnotfully
3.2.2 net heat flux, n—storage in the DFT front plate + specified without knowing at least two of these three param-
eters (temperature drop, heat flux, or thermal resistance). For
transmission (in other words, loss) to insulation layer. It is
equal to the [absorbed radiative heat flux + convective heat dynamic thermal experiments like fires or fire safety tests, the
electrical capacitance is replaced by the volumetric heat
flux] – [re-radiation from the exposed surface].
capacity.
3.2.3 total absorbed heat flux, n—absorbed radiative heat
5.1.3 The net heat flux, which is measured by a DFT, is
flux + convective flux.
likely different than the heat flux into the test item of interest
3.2.4 totalcoldwallheatflux,n—theheatfluxthatwouldbe
because of different surface temperatures. An alternative mea-
transferred by means of convection and radiation to an object
surement is the total cold wall heat flux which is measured by
whose temperature is 21 °C (70 °F).
water-cooledGardonorS-Bgauges.Theincidentradiativeflux
3.2.5 total heat flux (thermal exposure), n—incident radia-
can be estimated from either measurement by use of an energy
tive heat flux + convective heat flux.
balance [Keltner, 2007 and 2008 (16, 17)].The convective flux
can be estimated from gas temperatures and the convective
4. Summary of Test Method
heat transfer coefficient, h [Janssens, 2007 (18)].Assuming the
4.1 This test method provides techniques for measurement
sensor is physically close to the test item of interest; one can
of the net heat flux to a surface. Because Directional Flame
usetheincidentradiativeandconvectivefluxesfromthesensor
Thermometers are un-cooled devices, they are minimally
as boundary conditions into the test item of interest.
affectedbysootdepositionorcondensation.Calibrationfactors
5.1.4 In standardized fire resistance tests such as Test
Methods E119 and E1529, or ISO 834 or IMO A754, the
furnace temperature is controlled to a standard time-
Available fromAmerican National Standards Institute (ANSI), 25 W. 43rd St.,
4th Floor, New York, NY 10036, http://www.ansi.org.
temperature curve. In all but Test Methods E1529, implicit
Available from International Maritime Organization (IMO), 4, Albert
assumptions have been made that the thermal exposure can be
Embankment, London SE1 7SR, United Kingdom, http://www.imo.org.
described solely by the measured furnace temperature history
Available from theASTM website, www.astm.org, or contactASTM Customer
Service at [e-mail]. and that it will be repeatable from time to time and place to
E3057 − 19
place. However, these tests provide very different thermal 5.1.10 Reports by Sultan, M., (2006 and 2008) (22, 23) and
exposures due to the use of temperature sensors with very Janssens, M., (2008) (18) have shown it is difficult to measure
differentdesignsforfurnacecontrol.Asaresult,thesedifferent
one parameter in a fire resistance test (such as the furnace
thermal exposure histories produce different fire ratings for the temperature) and calculate the other (heat flux or thermal
same item. Historical variations of up to 50 % or more in the
exposure).
qualitative fire protection ratings (for example, 1 h) between
5.1.11 From the discussions in 5.1, it is highly recom-
different furnaces or laboratories indicate that the assumptions
mended that both temperature and heat flux be measured
for time-temperature control are not well founded.Also, due to
independently in fire tests.
different sensors, thermal exposure in a vertical furnace is
5.2 Use for DFTs:
generally higher than in a horizontal furnace, and thermal
exposureonthefloorofahorizontalfurnaceisgenerallyhigher 5.2.1 Although both cooled and non-cooled sensors can be
thanontheceiling.Thesereasonsprovidesupportforwhyboth used to measure heat flux, the results are generally quite
temperature and heat flux measurements are needed to provide
different. Water-cooled sensors are the direct reading Schmidt-
consistent test results. Boelter or Gardon gauge designs that are used in some
Committee E5 Methods (Test Methods E2683 and E511,
5.1.5 In the mid-90’s, the U. S. Coast Guard authorized a
study of the problems in marine fire resistance tests, such as respectively, have been developed for these sensors by Sub-
committee E21.08 ).
large variations in the ratings obtained in different furnaces.
One important conclusion was that the thermal exposure in
5.2.2 There are three types of passive or un-cooled sensors
furnaces could not be predicted solely from furnace tempera-
that can be used to measure net heat flux. One is the hybrid
ture measurements without large static and dynamic uncertain-
sensor (so-called High Temperature Heat Flux Sensor,
ties [Wittasek, N. A., 1996 (19)].
HTHFS) developed by Diller, et al., at Virginia Tech. It is
5.1.6 One of the recommendations that resulted from
designed to measure heat transfer to a surface without water
NIST’s investigation of the World Trade Center disaster was cooling [Gifford, A., Hubble, D., Pullins, C., and Diller, T.,
the need to move towards performance based codes and
2010 (4)]. The HTHFS requires a calibration factor that is a
standards.Areport developed forThe Fire Protection Research
function of sensor temperature [Pullins and Diller, 2010 (24)].
Foundation expanded on this recommendation [Beyler, C., et
Another is the so-called “direct write heat flux sensor” which
al., 2008 (20)]. Part of this effort involves making a more
can be used at temperatures from 25 to 860 °C [Trelewicz,
comprehensive set of measurements in fire resistance tests
Longtin, Hubble, and Greenlaw, 2015 (25)]; this gauge re-
including quantitative heat flux measurements. It also involves
quires a calibration coefficient. The third is the Directional
the development and use of “design fires” and defining their
Flame Thermometer (DFT), which was developed at Sandia
relationship with standardized test methods.
National Laboratories (based on work in the UK) and else-
5.1.7 Work at Sandia National Laboratories on transporta- where for measuring heat transfer in large sooty pool fires.
tion accidents involving hazardous materials compares the
DFTs do not require a calibration factor, which may be viewed
Prescriptive and Performance based approaches [Tieszen, et
as a mixed benefit. The passive sensors typically have higher
al., 2010 (21)].
temperature capability, based mainly on the Type K or N TC
limit of about 1250 °C. Even though they are water cooled,
5.1.8 Work by the National Research Council of Canada
used four (4) different temperature sensors to control a hori- quite often Gardon and Schmidt-Boelter gauges do not survive
in temperatures due to fouling of the sensing surface, and other
zontal furnace. Differences in the thermal exposure (see
definition in 3.2.5) were as high as 100 % during the first effects. DFTs usually survive up to about 1100 °C. They are
10 min [Sultan, M., 2006 and 2008 (22, 23)]. Assuming the very rugged, require no cooling, and are not susceptible to
temperature measurements from the different sensors or differ-
fouling of the sensing surface. These characteristics simplify
ent installations of the same sensor are actually the furnace installation in a wide range of fire and other applications. This
temperature, one can predict very different thermal exposures
standard will only address DFTs. See 10.2.2 for a more
depending on which temperature measurement method is used.
thorough discussion of heat flux gauge calibrations.
5.1.9 In another series of horizontal furnace tests, the
5.2.3 Early work on DFTs (and the data analysis techniques
National Research Council of Canada (NRCC) studied the
for them) focused on acquiring quantitative heat flux data to
effect of six (6) different temperature sensor designs on fire
help define the thermal conditions in large, liquid hydrocarbon
resistance tests in a large, horizontal furnace [Sultan, 2008
pool or spill fires. Large pool fires can reach quasi-steady
(23)]. NRCC used six different temperature sensors for furnace
conditions in times as short as a minute.As a result, Pool Fire
control: Test Methods E119 Shielded Thermocouple, ISO 834
DFTs were designed with 1.6 mm thick plates to provide rapid
Plate Thermometer, 6 mm MIMS TC from Test Methods
equilibration with the fire (the maximum heating rate in these
E1529, Directional Flame Thermometers, and 1.6 mm MIMS
fires was approximately 30 °C⁄s).
TCs with grounded and ungrounded junctions. Total heat flux
at the ceiling was measured using a Gardon gauge. Results
6. Apparatus
showed that very different thermal exposures are possible
6.1 DFT Construction:
depending on the measurement method used. During the first
10 min of a fire resistance test, the integrated heat flux varies 6.1.1 DFT apparatus consists of the DFT, mounting
by a factor of two. hardware, and a data acquisition system.
E3057 − 19
6.1.2 The DFT consists of two heavily oxidized Inconel facing into the furnace with a 1.6 mm back plate. Different
plates with a ceramic fiber insulation layer sandwiched be- plate thicknesses are used for different applications. Some
tween the plates. Alternately, to obtain a high emissivity special designs have used a third plate and thermocouple.
surface one can apply high emissivity paint to the exposed Some used in automotive fires were small and used intrinsic
surface. If paint is used, one must be careful as at high
thermocouples to provide very fast response. Fig. 1 shows the
temperatures some paints do not remain in place. A 1.6 mm construction of a typical DFT, and Fig. 2 shows a photo of a
OD, mineral-insulated, metal-sheathed (MIMS) thermocouple
typical DFT.
(TC) is attached to each unexposed face. Typically the sheath
6.1.3 Platethicknessesvarydependingontheapplication.If
materialisInconel.Tooptimizetheresponseinavarietyoffire
faster response is desired a thinner plate is used (for example,
scenarios, there are three basic DFT designs. The original
1.6mm),orifslowerresponseisacceptable,athickerplatecan
furnace DFT uses two 3 mm (nominal) thick plates; the
be used (for example, 3 mm). It is advisable to never have the
original pool fire DFTuses two 1.6 mm (nominal) thick plates.
plate thickness less than the TC sheath diameter, so the effect
Both Inconel and SS have been used; Inconel 600 is recom-
of the TC on the plate temperature measurement is minimized
mended because 304SS can sometimes form a scale that falls
[see Figueroa, 2005 (26-28) for a detailed analysis]. Due to
off the surface. The modified furnace DFT usesa3mm plate
Intrinsic thermocouples use bare wires welded to the metal surface of the DFT.
This forms an “intrinsic” junction using the metal of the DFT. Intrinsic TCs have
small dynamic errors compared with ungrounded junction (sheathed) TCs but are
not very robust and fail more often. MIMS TCs are fully sheathed and encase the
TC junction, and can be grounded or ungrounded.
FIG. 1 Basic Design of a Directional Flame Thermometer (Using 3 mm Thick Plates)
E3057 − 19
FIG. 2 Photo of Typical DFT
manufacturers recommended limits on MIMS thermocouples, Inconel plates.A12 mm wide by 25 mm long strip of nickel or
TC sheath diameters less than 1.6 mm are not recommended. Nichrome foil (for example, 0.08 mm thick) is formed over the
6.1.4 TheInconelplatesaremountedparallelwithalayerof tip of the thermocouple and spot welded to the unexposed
ceramicfiberinsulationmateriallightlycompressedinbetween surface of each plate (see Fig. 3). This technique provides a
the plates. The plates are held together with four bolts. One good thermo-mechanical attachment of the thermocouples,
thermocouple is mounted on the inside surface of each of the which is critical for good dynamic response In general the
FIG. 3 Photo of Typical TC Installation
E3057 − 19
nichrome strip should be as small as possible while still spacers (for example, made of 304 stainless steel) to compress
ensuring good mechanical contact with the surface (see the insulation layer to a thickness of 19 mm (0.75 in.). This
MNL12, page 183). Fig. 3 shows a typical TC installation. compression is important because the insulation thermal prop-
6.1.5 Apparatus to mount the DFT near the test unit should erties depend on thickness. See Annex A1.
be as small as possible to disturb the environment as little as
7.1.10 Route the two TCs together out of the heated region.
possible. The DFT should be mounted so that one of the
It is recommended that the TC sheaths be insulated until they
Inconel plates is facing the environment one wants to measure.
reach a room temperature location.
The DFT has a 180° field of view, so the DFT should be
7.2 Fabrication of Mounting Hardware—Mounting hard-
oriented so that the entire environment is captured within that
ware is not unique.Any mounting design that holds the DFTin
field of view.
place but does not affect the environment is suitable. Any
6.1.6 The data acquisition system needs to be able to
material that can withstand the temperatures in the environ-
accuratelyrecordTypeKorTypeNthermocouples.Manysuch
ment of interest can be used. Mild steel can be used if the melt
systems exist and we will not discuss them further here.
temperatureisnotexceeded.Butrecallthatthestrengthofmild
steel at high temperatures is reduced to approximately that of
7. Preparation of Apparatus
aluminum, so strength is much reduced. Stainless steel is the
7.1 Fabrication of Directional Flame Thermometers:
better, but more expensive option.
7.1.1 See Fig. 1 for a sketch of a DFT.
7.1.2 See Fig. 2 for a photograph of a DFT.
8. Hazards
7.1.3 Cut or shear two 1.6 or 3 mm ( ⁄16 or 0.12 in.) thick
2 2
Inconel plates, 120 mm (4.75 in. ).
8.1 This standard does not purport to address all of the
7.1.4 Drill 6.75 mm (letter drill H, ⁄64 in.) holes in four safety concerns, if any, associated with the use of DFTs. It is
corners, leaving approximately one hole diameter from each
the responsibility of the user of this standard to establish
edge. appropriate safety and health practices and determine the
7.1.5 Heat the plates in a furnace at approximately 1000 °C
applicability of regulatory limitations prior to use.
for 24 h to develop a stable, high absorptivity oxide layer. If
8.2 Warning—The only known potential hazard is related
this is not possible, one can substitute a high emissivity paint
to the insulation in the DFT. Long durations in unventilated
that adheres to the plate at high temperatures.
areas with used insulation may be cause for concern because
7.1.6 Use 1.6 mm ( ⁄16 in.) OD Inconel sheathed Type K
some of the ceramic fibers may become airborne. The user
(Chromel/Alumel) or Type N thermocouples (TCs) with an
should contact the insulation manufacturer for information
ungrounded junction. Sand the oxide off the plate over a 3 by
about proper safety procedures related to the insulation.
1.5–3.9 cm (1.2 by 0.6–1.3 in.) area in the center of each plate.
7.1.7 Using 0.08 mm (0.003 in.) thick by 6.4 mm (0.25 in.)
9. Procedure
wide Nickel or Nichrome foil, form the foil strips tightly over
last 25 mm (1 in.) of the TC and completely cover the TC tip.
9.1 Fabricate DFT in accordance with 7.1 and the mounting
Then, spot weld the foil to the sanded area of the plate (do not
hardware in accordance with 7.2.
spotweldtotheTCsheath).Providealoopforstressrelief.Do
9.2 Mount the DFT so that the field of view of the DFT
not weld the TC because the welding process might penetrate
encompasses the entire heat source.
through the sheath. See Fig. 3 (in the photo the strap is shorter
than recommended).
9.3 Route the two thermocouple leads to a room tempera-
7.1.8 Cut a 120 by 120 by 25 mm (4.75 by 4.75 by 1 in.)
turelocation.TheTCsheathsshouldbeprotectedfromtheheat
3 3
piece of 128 kg/m (8 lb⁄ft ) ceramic fiber insulation and place
source by wrapping them with the same type of insulation used
between the plates. Temperature dependent thermal properties
in the DFT. This protection can reduce the chance of “shunt-
of a Thermal Ceramics insulation called “Cerablanket” have
ing” occurring. See Appendix X3.2.
beenmeasured,andthosepropertiesareprovidedinAnnexA1.
9.4 Calibrate the DAS by using a NIST traceable source to
If a different insulation is used, it is important to measure the
place a known input into each channel at a select number of
properties of that material. There are other brands (for
temperatures to ensure each DAS channel is reading properly.
example, Kaowool by Morgan Thermal Ceramics, http://
www.morganthermalceramics.com/products/refractory-
9.5 Connect the TCs to a data acquisition system (DAS).
ceramic-fibre-rcf/blanket)thatcanbeused,butthetemperature
9.6 Do not calibrate the TCs used in the DFT, because for
dependent, thermal properties would need to be measured.
Type K TCs the calibration process can change the output of
7.1.9 Assemble the DFT using four 9.53 mm (~ ⁄8 in.
the TC and therefore change the calibration (30, 31).
diameter) Inconel 600 or silver plated SS bolts and tubular
9.7 Set up DAS to scan at a rate of about 1 Hz. Because the
time constant of the TCs can be several seconds, there is no
Work at Sandia National Laboratories has shown the Inconel emissivity can
reason to sample at a faster rate.
vary considerably depending on the extent of the oxide layer. Values of about 0.85
have been measured [for example, Figueroa, 2006 (26-28)], but others [Brundage,
9.8 Measure pre- and post-test emissivity of the exposed
A., et al (29)] measured emissivity between 0.67 and 0.90. For highest accuracy the
DFT plate surfaces. These measurements can be used to
user should either measure the emissivity measurements or apply high emissivity
black paint with known emissivity. estimate the incident heat flux.
E3057 − 19
9.9 Carefully review thermocouple results to ensure anoma- 10.1.3.3 Verify insulation (8 lb/ft ) is compressed to
lies are not present (for example, a spike in temperature that 1.91 cm thickness and is not forced out the sides of the plates.
has no basis for occurring). If the insulation is different from Cerablanket, measure the
temperature dependent thermal properties. See Annex A1.
9.10 Reduce data by means of one of the methods described
10.1.3.4 Verify the TCs are mounted according to the
below.
procedure in 7.1.
10.1.3.5 Verify the plate surfaces have a stable oxidation
10. Calibration and Standardization
layer or stable paint layer (if one cannot oxidize the plates at
10.1 Apparatus Calibration—Two items should be cali-
1000 °C, an alternate is to use a high absorptivity paint).
brated when using DFTs, the data acquisition system and the
10.1.3.6 Measure and record the hemispherical total surface
thermocouples. However, the thermocouples should be “cali-
emissivityofexposedsurfacesofeachplate.Thisisusedifone
brated” in a manner different than is typically done.
desires to convert the net heat flux to incident heat flux.
10.1.1 Thermocouple “Calibration”—TheASTM standards
Estimate the emissivity uncertainty by making multiple mea-
for accuracy are sufficiently good for use in DFTs. It has been
surements and using the manufacturer’s reported accuracy.
shown that Type K thermocouples are actually affected by the
10.2 Reference Standards and Calibration Curves and
calibration process. As a result, the chemistry of the Chromel
Tables:
andAlumel wires change when calibrated above about 320 °C
10.2.1 Refer to MNL12 for Thermocouple accuracies.
(30, 31). TCs used in DFTs should be calibrated up to the
10.2.1.1 For type K thermocouples, MNL12 specifies the
maximum temperature expected (for example, 1100 °C).
range to be from 0 to 1250 °C (32 to 2300 °F). The “standard
Therefore, one should not use the TC after calibration. This
tolerance” is 62.2 °C or 0.75 % of the reading in °C, which-
apparent conundrum is resolved in the following manner. One
ever is greater. The “special tolerance” is 61.1 °C or 0.4 % of
can obtain additional TCs fabricated from the same spool of
the reading in °C, whichever is greater.
wire from the manufacturer (preferably wire before and after
10.2.2 Discussion on Calibration of DFTs:
the TCs used on the DFTs). Those additional TCs are sent
through the calibration process. If those calibratedTCs meet or 10.2.2.1 Most heat flux gauges (for example, thin film,
exceed the ASTM standard for Type K TCs (in other words, Gardon, Schmidt-Boelter) are designed to have a linear output
62.2 °Cor 60.75 %ofreadingin°C,seeMNL12,1993),then with heat flux. Data reduction is easy because the gauge comes
one assumes the remainingTCs from the same spool also meet
with a calibration in the form of a sensitivity coefficient (in
the same ASTM standard. Experience has shown that in all other words, xx mV/unit of heat flux), and these sensitivity
cases the calibrated TCs were more accurate than the ASTM
coefficients are made with reference to a NIST standard.
standard. But the more accurate values are not used; one uses
10.2.2.2 Traditional heat flux gauge calibrations use a ra-
the accuracy specifications from the MNL12, so as not to
diative only heat source and seek to minimize convection
assume accuracy better than the standard.
effects(forexample,NIST,Medtherm,andsoforth).Detailsof
10.1.2 Data Acquisition System Calibration—The recom- those calibration procedures will not be discussed in detail
mended method to assure that the data acquisition system here. Typical accuracies reported are 63 %. An effort was
uncertaintyisknownistocalibrateeachchannelovertherange initiated at NIST to develop a convective heat transfer calibra-
of temperatures expected. For example, if one expects tem- tion capability, but the effort was not completed and no such
peratures ranging from 20 to 1100 °C, one can calibrate at facility exists at NIST.Aconvective calibration capability does
several set points over the range (for example, 20, 200, 400, exist at Virginia Tech University under the guidance of Prof.
600, 800, 1100 °C). The calibration is performed using a NIST Tom Diller.
traceable thermocouple simulator (for example, Fluke and
10.2.2.3 Adetailed analysis of heat flux gauges has led to a
Ectronmakesuchcalibrators).Astatisticallysignificantsample
better understanding of under what conditions one can assume
(for example, 10s of samples) is taken at each set point, then
the linear sensitivity coefficients are an accurate representation
the mean and standard deviation of each set point on each
of the behavior of the heat flux gauges.Abody of evidence has
channel can be estimated. Because this can be a very large data
shown that in fact the sensitivity coefficients developed for
set, one can average the data for all channels at all set points to
radiation only calibrations are not accurate for mixed heat
provide a single estimate of the accuracy of all DAS channels.
transfer applications where convection is non-negligible. (See
Typically this source of uncertainty is small, but on occasion
references in the Introduction.)
one finds a bad channel so the exercise is worthwhile.
10.2.2.4 For both Gardon and Schmidt-Boelter gauges what
10.1.3 Detailed Measurements of DFT—Detailed measure-
has been discovered is that the gauge sensitivity for an
ments of the DFT materials should be made before assembly,
equivalent level of radiant heat flux is different than for the
because those measurements will be used in the data reduction
same level of convective heat flux. One might reasonably ask:
process.
“So what?” How much does this affect readings? These are
10.1.3.1 Measure and record the thickness of both Inconel good questions that will be discussed below. First, a simple
plates as close to the center as possible. Estimate the accuracy example will be discussed to show possible effects of the
of those measurements. problem.
10.1.3.2 Measure and record the spacer thicknesses to 10.2.2.5 For a radiation only heat flux measurement, one
confirm they are 1.91 cm (0.75 in.). records the voltage output and multiplies the output by the
E3057 − 19
sensitivity coefficient provided by the manufacturer to get an 10.2.2.13 Similarly, for S-B gauges, one assumes a uniform
estimate of heat flux. This is expressed in Eq 1: exposure of radiative flux over the sensing element. In shear
flow this is not the case so again one might expect different
q 5 v·S (1)
sensitivities for radiative and convective fluxes.
where:
10.2.2.14 Therefore, because sensitivity coefficients in ra-
q = the heat flux, diative and convective heat transfer environments are different
v = the voltage output, and when using Gardon and S-B gauges, and there is no NIST
S = the sensitivity coefficient (unit of heat flux/volt).
traceable convective heat flux calibration capability, and be-
cause making an estimate (for example, using Eq 2) of the heat
10.2.2.6 What is normally done is the user assumes the
flux in mixed heat transfer environments has a number of
sensitivity coefficient is the same for all modes of heat transfer,
uncertain parameters, it is difficult to fully understand the
thereforeoneestimatestheheatfluxusingEq1.Butithasbeen
uncertainty of these types of gauges when used in a mixed
shown that one should not assume the sensitivity coefficients
mode heat transfer environment. Therefore, a different method
are the same, therefore, a different data reduction method is
to estimate heat flux was developed.
appropriate. How then does the user reduce his data?
10.2.2.15 Characteristics of this “different” method were as
10.2.2.7 One method to reduce the data is to use a linear
follows:
combination of the fluxes as shown in Eq 2:
(1) The gauge had to be rugged and survive temperatures
q 5 v· F · S 1 F · S (2) up to about 1100 °C.
~ !
rad rad conv conv
(2) The gauge should not be actively cooled.
where:
(3) The gauge does not use a single sensitivity coefficient,
F and F = the fractions of the total heat flux attributed
rad conv
so one does not suffer from the issues discussed above
to radiation and convection, and
(different radiative and convective sensitivity coefficients).
S and S = the radiative and convective sensitivity
rad conv
(4) The gauge is simple so can be analyzed by means of a
coefficients.
thermal model.
(5) Thegaugerespondstobothradiationandconvectionso
10.2.2.8 What makes Eq 2 difficult to use is that most of the
one measures the total heat flux to a surface.
terms are not known to high accuracy. Only S is known to
rad
10.2.2.16 DFTs satisfy all of the desired characteristics
good accuracy (from the manufacturer); none of the others are
listed above. But the downside is the complication of data
known except the output voltage.
reduction and a more complicated uncertainty analysis. The
10.2.2.9 Eq2assumesthatboththeradiativeandconvective
uncertainty analysis for DFTs is more complicated, and de-
sensitivity coefficients are linear with heat flux. This is true for
pends on the data reduction method used (in other words,
radiativefluxbutithasnotbeenshownthecaseforconvection.
energy storage method, inverse heat conduction, inverse filter
But Eq 2 serves to make the point, as follows.
function).
10.2.2.10 In Eq 2, assuming the radiative and convective
10.2.2.17 With DFTs, one trades the convenience of having
sensitivity coefficients are equal, and the sum of the radiative
a linear sensitivity coefficient with known and traceable accu-
and convective fractions equals 1.0, then Eq 2 is reduced to Eq
racy and a relatively complicated gauge design (S-B and
1. This is in fact what every user of heat flux gauges assumes,
Gardon)foramuchsimplerdesign(inotherwords,DFTs)with
whether known or not, when using a calibration performed in
a more complicated data reduction and uncertainty analysis.
radiation only.
10.2.2.18 The discussion above sheds light on the advan-
10.2.2.11 For Gardon and S-B gauges the convective sen-
tages of using a gauge that does not require a calibration,
sitivity coefficients can be quite different from radiative sensi-
assuming one has the tools to reduce the data and analyze the
tivity coefficients. For example, Gifford, et al., 2010 (4, 10),
uncertainties when using DFTs.
showed that for S-B gauges the convection sensitivity coeffi-
11. Calculation or Interpretation of Results
cients can be up to about 20 % different than the radiation
sensitivity coefficients. Further complicating matters is that the
11.1 General:
convective sensitivity coefficients are different for shear and
11.1.1 The data analysis techniques in this section use the
stagnation flows. Similar results have been shown for Gardon
DFT plate temperature histories and material properties to
gauges by Kuo and Kulkarni, 1991 (8).
provide quantitative estimates of net heat flux data over the
10.2.2.12 Why would Gardon and S-B gauges have differ- entire test duration. The inverse heat conduction analysis and
ent sensitivity coefficients in radiative and convective environ- energy storage methods both calculate the net heat flux
ments? A qualitative understanding is possible by understand- post-test. The inverse filter function method provides near
inghowthegaugesareconstructed.Gardongaugeshaveavery real-time estimates of net heat flux during a test.
thin sensing element that has a parabolic temperature profile 11.1.2 Implicit in the energy storage method analyses is that
from the center of the element to the edge when exposed to a the temperature measurements, made on the unexposed side of
uniformradiativeheatflux.Butduringaconvectiveshearflow, theplate,aresufficientlyclosetotheexposedsidetemperature.
the temperature profile can “tilt” to the downstream side of the This is due to the relatively high conductivity of the Inconel
sensing element. There is good reason to expect that the plate.TheTCs are mounted on the unexposed side because the
sensitivity coefficient for a Gardon gauge in shear flow might bias errors are lower and survivability is higher. This assump-
be different than for the same magnitude of radiative heat flux. tion can be confirmed with an inverse heat conduction analysis
E3057 − 19
which provides an estimate of the exposed side plate tempera- h
q 5 ~q ⁄ ε !1~σ · T !1 ·~T 2 T ! (6)
FS D G
inc,r net DFT DFT DFT gas
ture. The measurements are typically very close to that ε
DFT
estimated from inverse heat conduction calculations (see Ap-
11.2.8 In Eq 6, one can measure or estimate ε and h.
DFT
pendix X5 for an example). This approximation is more
T is measured. T can be assumed equal to the fire or
DFT gas
accurate for the 1.6 mm plate. The inverse heat conduction
flame temperature. If CFD simulations are available, the
method does not suffer from this assumption.
temperature of the fluid near the DFT can be used for T .
gas
11.1.3 Before the heat flux estimation techniques are
Because‘h’andT areassumedconstant,oneshouldconsider
gas
described, an energy balance on the sensing surface will be
this a quasi-steady energy balance. q can be estimated in
net
developed, and how one should use the measurement will be
three ways as discussed next.
discussed.
11.3 Inverse Heat Conduction Analysis Method:
11.2 Energy Balance on DFT:
11.3.1 The inverse heat conduction analysis uses a one
11.2.1 An
...

This document is not an ASTM standard and is intended only to provide the user of an ASTM standard an indication of what changes have been made to the previous version. Because
it may not be technically possible to adequately depict all changes accurately, ASTM recommends that users consult prior editions as appropriate. In all cases only the current version
of the standard as published by ASTM is to be considered the official document.
Designation: E3057 − 16 E3057 − 19
Standard Test Method for
Measuring Heat Flux Using Directional Flame Thermometers
with Advanced Data Analysis Techniques
This standard is issued under the fixed designation E3057; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
INTRODUCTION
This test method describes a technique for measuring the net heat flux to one or both surfaces of
a sensor called a Directional Flame Thermometer. The sensor covered by this standard uses
measurements of the temperature response of two metal plates along with a thermal model of the
sensor to determine the net heat flux. These measurements can be used to estimate the total heat flux
(aka (also known as thermal exposure) and bi-directional heat fluxes for use in CFD thermal models.
The development of Directional Flame Thermometers (DFTs) as a device for measuring heat flux
originated because commercially available, water-cooled heat flux gauges (for example, Gardon and
Schmidt-Boelter gauges) did not work as desired in large fire tests. Because the Gardon and
Schmidt-Boelter (S-B) gauges are water cooled, condensation and soot deposition can occur during
fire testing or in furnaces. Both foul the sensing surface which in turn changes the sensitivity
(calibration) of the gauge. This results in an error during data reduction. Therefore, a different type of
sensor was needed; one such sensor is a DFT. DFTs are not cooled so condensation and soot deposition
are minimized or eliminated.
Additionally, a body of work has shown that for both Gardon and Schmidt-Boelter gauges the
sensitivity coefficients determined through the calibration process, which uses a radiative heat source,
are not the same as the sensitivity coefficients determined if a purely convective source is used for
calibration [Test Method E511-07; Keltner and Wildin, 1975 (1, 2); Borell, G. J., and Diller, T. E.,
1987 (3); Gifford, A., et al., 2010 (4); Gritzo, L. A., et al., 1995 (5); Young, M. F., 1984 (6); Sobolik,
et al., 1987 (7); Kuo and Kulkarni, 1991 (8); Keltner, 1995 (9); Gifford, et al., 2010 (10); Nakos, J.
T., and Brown, A. L., 2011 (11)]. As a result, one can incur significant bias errors when reducing data
in tests where there may be a non-negligible convective component because the only sensitivity
coefficient available is for a radiation calibration. It was desired to reduce/eliminate these potential
sourcesources of error by designing a gauge that does not depend on a radiation only calibration. DFTs
have this characteristic.
A sensor, also called a Directional Flame Thermometer, was developed to help estimate flame
thickness in pool fire tests of hazardous material shipping containers [Burgess, M. H., 1986 (12); Fry,
C. J., 1989 (13); Burgess, M. H., et al., 1990 (14); and Fry, C. J., 1992 (15)]. As originally designed,
DFTs were quasi-equilibrium sensors that used a thin metal plate with a single thermocouple attached
and backed by multiple radiation shields. To make a sensor suitable for continuous transient heat flux
measurements, this basic design was modified to use two instrumented plates, with a layer of
insulation in between.
For the Directional Flame Thermometers described in this standard, the net heat flux is calculated
using transient temperature measurements of the two plates and temperature dependent material
properties for the plates and the insulation. Three methods are described in this standard to calculate
the net heat flux. The most accurate method for calculating the net heat flux is believed to be the
1-dimensional, nonlinear inverse heat conduction analysis, which uses the IHCP1D code. This is based
This test method was jointly developed by ASTM Committee E21 on Space Simulation and Applications of Space Technology and is the direct responsibility of
Subcommittee E21.08 on Thermal Protection.
Current edition approved April 1, 2016June 1, 2019. Published May 2016July 2019. Originally approved in 2016. Last previous edition approved in 2016 as E3057 – 16.
DOI: 10.1520/E3057-16,10.1520/E3057-19.
The boldface numbers in parentheses refer to the list of references at the end of this standard.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
E3057 − 19
on uncertainty analyses and comparisons with measurements made with Schmidt-Boelter and Gardon
gauges, which have NIST traceable calibrations. The second method uses a transient energy balances
on the DFT. As will be shown below, the energy balance method compares very well with the inverse
method, again based on uncertainty analyses. The third method uses sets of linearized, convolution
digital filters based on IHCP1D. These allow a near real-time calculation of the net heat flux [Keltner,
N. R., 2007 (16); Keltner, N. R., et al., 2010 (17)]. See Section 1 for more detailed information on each
analysis technique. Additional information is given in the Annexes and Appendices.
Various DFT designs have been used in a variety of applications including very large pool fires,
LNG spill fires, marine fire safety testing, automobile fires, to study rocket launch accident fires, and
in research of forest and wild-land fires. Appendix X1 provides a comprehensive list of applications
where DFTs have been successfully used.
Advantages of DFTs are their relatively low cost, ease of construction, they require no calibration
(see later), and require no cooling. They are robust and can survive intense fire environments without
failure. Disadvantages include most are large compared with Gardon and S-B heat flux gauges and
because they are not calibrated, one cannot reference the measurements to a NIST standard. Because
no calibration is required, one must quantify the uncertainties present in the temperature measure-
ments and the data reduction methods used to calculate the heat flux. Also, DFTs measure net heat
flux; for a direct comparison with Gardon and S-B gauges, which are calibrated to incident (or “cold
wall”) flux, one must use a thermal model to estimate the incident flux.
The best applications for DFTs are where Gardon and S-B gauges cannot be used (for example, due
to high temperatures, lack of cooling, soot deposition, fouling, and so forth), or when long life and
overall costs are a consideration. Gardon and Schmidt-Boelter gauges are recommended in non-sooty
environments, when it is possible to mount the gauges and cooling lines, and in predominantly
radiative environments with a small convective contribution.
1. Scope
1.1 This test method describes the continuous measurement of the hemispherical heat flux to one or both surfaces of an uncooled
sensor called a “Directional Flame Thermometer” (DFT).
1.2 DFTs consist of two heavily oxidized, Inconel 600 plates with mineral insulated, metal-sheathed (MIMS) thermocouples
(TCs, type K) attached to the unexposed faces and a layer of ceramic fiber insulation placed between the plates.
1.3 Post-test calculations of the net heat flux can be made using several methodsmethods. The most accurate method uses an
inverse heat conduction code. Nonlinear inverse heat conduction analysis uses a thermal model of the DFT with temperature
dependent thermal properties along with the two plate temperature measurement histories. The code provides transient heat flux
on both exposed faces, temperature histories within the DFT as well as statistical information on the quality of the analysis.
1.4 A second method uses a transient energy balance on the DFT sensing surface and insulation, which uses the same
temperature measurements as in the inverse calculations to estimate the net heat flux.
1.5 A third method uses Inverse Filter Functions (IFFs) to provide a near real time estimate of the net flux. The heat flux history
for the “front face” (either surface exposed to the heat source) of a DFT can be calculated in real-time using a convolution type
of digital filter algorithm.
1.6 Although developed for use in fires and fire safety testing, this measurement method is quite broad in potential fields of
application because of the size of the DFTs and their construction. It has been used to measure heat flux levels above 300 kW/m
in high temperature environments, up to about 1250°C,1250 °C, which is the generally accepted upper limit of Type K or N
thermocouples.
1.7 The transient response of the DFTs is limited by the response of the MIMS TCs. The larger the thermocouple the slower
the transient response. Response times of approximately 1 to 2 s are typical for 1.6 mm diameter MIMS TCs attached to 1.6 mm
thick plates. The response time can be improved by using a differential compensator.
1.8 The values stated in SI units are used in this to be regarded as standard. The values statedgiven in parentheses after SI units
are provided for information only.only and are not considered standard.
1.9 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility
of the user of this standard to establish appropriate safety safety, health, and healthenvironmental practices and determine the
applicability of regulatory limitations prior to use.
1.10 This international standard was developed in accordance with internationally recognized principles on standardization
established in the Decision on Principles for the Development of International Standards, Guides and Recommendations issued
by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
E3057 − 19
2. Referenced Documents
2.1 ASTM Standards:
C177 Test Method for Steady-State Heat Flux Measurements and Thermal Transmission Properties by Means of the
Guarded-Hot-Plate Apparatus
E119 Test Methods for Fire Tests of Building Construction and Materials
E176 Terminology of Fire Standards
E457 Test Method for Measuring Heat-Transfer Rate Using a Thermal Capacitance (Slug) Calorimeter
E459 Test Method for Measuring Heat Transfer Rate Using a Thin-Skin Calorimeter
E511 Test Method for Measuring Heat Flux Using a Copper-Constantan Circular Foil, Heat-Flux Transducer
E1529 Test Methods for Determining Effects of Large Hydrocarbon Pool Fires on Structural Members and Assemblies
E2683 Test Method for Measuring Heat Flux Using Flush-Mounted Insert Temperature-Gradient Gages
2.2 Other Standards:
ISO 834-11:2014 Fire Resistance Tests—Elements of Building Construction—Part 11: Specific Requirements for the
Assessment of Fire Protection to Structural Steel Elements
MNL12-4thIMO A754 Manual on the Use of Thermocouples in Temperature Measurement, Fourth Edition, 1993, Sponsored by
ASTM Committee E20 on Temperature MeasurementFire Resistance Tests: Fire Safety Onboard Ships
2.3 Other ASTM Document:
MNL12-4th Manual on the Use of Thermocouples in Temperature Measurement, Fourth Edition, 1993, Sponsored by ASTM
Committee E20 on Temperature Measurement
3. Terminology
3.1 Definitions—Refer to Terminology in ASTM Standard E176 for definitions of some terms used in these test methods.
3.2 Definitions of Terms Specific to This Standard:
3.2.1 incident radiative heat flux (irradiance; q ), n—radiative heat flux impinging on the surface of the DFT or the unit under
inc,r
test.
3.2.2 net heat flux, n—storage in the DFT front plate + transmission (in other words, loss) to insulation layer. It is equal to the
[absorbed radiative heat flux + convective heat flux] – [re-radiation from the exposed surface].
3.2.3 total absorbed heat flux, n—absorbed radiative heat flux + convective flux.
3.2.4 total cold wall heat flux, n—the heat flux that would be transferred by means of convection and radiation to an object
whose temperature is 21°C (70°F).21 °C (70 °F).
3.2.5 total heat flux (thermal exposure), n—incident radiative heat flux + convective heat flux.
4. Summary of Test Method
4.1 This test method provides techniques for measurement of the net heat flux to a surface. Because Directional Flame
Thermometers are un-cooled devices, they are minimally affected by soot deposition or condensation. Calibration factors or
sensitivity coefficients are not required because alternate methods of data reduction are used. DFTs are simple to fabricate and use,
but are more complicated when reducing the data. Gardon and Schmidt-Boelter gauges have relatively linear outputs with heat flux
and only require a single sensitivity coefficient (for example, xx mv/unit of flux) to convert the output to an incident heat flux. DFTs
have two thermocouple outputs as a function of time. Those outputs along with temperature dependent thermal properties and
advanced analysis techniques are used with a thermal model to calculate the net heat flux. The net heat flux (with an energy
balance) can be used to estimate the total cold wall heat flux, which is same as the measurement made by Gardon or S-B gauges
[Janssens, 2007 (18)].
5. Significance and Use
5.1 Need for Heat Flux Measurements:
5.1.1 Independent measurements of temperature and heat flux support the development and validation of engineering models
of fires and other high environments, such as furnaces. For tests of fire protection materials and structural assemblies, temperature
and heat flux are necessary to fully specify the boundary conditions, also known as the thermal exposure. Temperature
measurements alone cannot provide a complete set of boundary conditions.
For referenced ASTM standards, visit the ASTM website, www.astm.org, or contact ASTM Customer Service at [e-mail]. For Annual Book of ASTM Standards
volume information, refer to the standard’s Document Summary page on the ASTM website.
Available from American National Standards Institute (ANSI), 25 W. 43rd St., 4th Floor, New York, NY 10036, http://www.ansi.org.
Available from International Maritime Organization (IMO), 4, Albert Embankment, London SE1 7SR, United Kingdom, http://www.imo.org.
Available from the ASTM website, www.astm.org, or contact ASTM Customer Service at [e-mail].
E3057 − 19
5.1.2 Temperature is a scalar variable and a primary variable. Heat Flux is a vector quantity, and it is a derived variable. As
a result, they should be measured separately just as current and voltage are in electrical systems. For steady-state or quasi-steady
state conditions, analysis basically uses a thermal analog of Ohm’s Law. The thermal circuit uses the temperature difference instead
of voltage drop, the heat flux in place of the current and thermal resistance in place of electrical resistance. As with electrical
systems, the thermal performance is not fully specified without knowing at least two of these three parameters (temperature drop,
heat flux, or thermal resistance). For dynamic thermal experiments like fires or fire safety tests, the electrical capacitance is
replaced by the volumetric heat capacity.
5.1.3 The net heat flux, which is measured by a DFT, is likely different than the heat flux into the test item of interest because
of different surface temperatures. An alternative measurement is the total cold wall heat flux which is measured by water-cooled
Gardon or S-B gauges. The incident radiative flux can be estimated from either measurement by use of an energy balance [Keltner,
2007 and 2008 (16, 17)]. The convective flux can be estimated from gas temperatures and the convective heat transfer coefficient,
h [Janssens, 2007 (18)]. Assuming the sensor is physically close to the test item of interest; one can use the incident radiative and
convective fluxes from the sensor as boundary conditions into the test item of interest.
5.1.4 In standardized fire resistance tests such as Test Methods E119 and E1529, or ISO 834 or IMO A754, the furnace
temperature is controlled to a standard time-temperature curve. In all but Test Methods E1529, implicit assumptions have been
made that the thermal exposure can be described solely by the measured furnace temperature history and that it will be repeatable
from time to time and place to place. However, these tests provide very different thermal exposures due to the use of temperature
sensors with very different designs for furnace control. As a result, these different thermal exposure histories produce different fire
ratings for the same item. Historical variations of up to 50%50 % or more in the qualitative fire protection ratings (for example,
one hour) 1 h) between different furnaces or laboratories indicate that the assumptions for time-temperature control are not well
founded. Also, due to different sensors, thermal exposure in a vertical furnace is generally higher than in a horizontal furnace, and
thermal exposure on the floor of a horizontal furnace is generally higher than on the ceiling. These reasons provide support for
why both temperature and heat flux measurements are needed to provide consistent test results.
5.1.5 In the mid-90’s, the U. S. Coast Guard authorized a study of the problems in marine fire resistance tests, such as large
variations in the ratings obtained in different furnaces. One important conclusion was that the thermal exposure in furnaces could
not be predicted solely from furnace temperature measurements without large static and dynamic uncertainties (Wittasek,[Wittasek,
N. A., 1996 (19)).].
5.1.6 One of the recommendations that resulted from NIST’s investigation of the World Trade Center disaster was the need to
move towards performance based codes and standards. A report developed for The Fire Protection Research Foundation expanded
on this recommendation [Beyler, C., et al., 2008 (20)]. Part of this effort involves making a more comprehensive set of
measurements in fire resistance tests including quantitative heat flux measurements. It also involves the development and use of
“design fires” and defining their relationship with standardized test methods.
5.1.7 Work at Sandia National Laboratories on transportation accidents involving hazardous materials compares the Prescriptive
and Performance based approaches [Tieszen, et al., 2010 (21)].
5.1.8 Work by the National Research Council of Canada used four (4) different temperature sensors to control a horizontal
furnace. Differences in the thermal exposure (see definition in 3.2.5) were as high as 100%100 % during the first ten minutes
10 min [Sultan, M., 2006 and 2008 (22, 23)]. Assuming the temperature measurements from the different sensors or different
installations of the same sensor are actually the furnace temperature, one can predict very different thermal exposures depending
on which temperature measurement method is used.
5.1.9 In another series of horizontal furnace tests, the National Research Council of Canada (NRCC) studied the effect of six
(6) different temperature sensor designs on fire resistance tests in a large, horizontal furnace [Sultan, 2008 (23)]. NRCC used six
different temperature sensors for furnace control: Test Methods E119 Shielded Thermocouple, ISO 834 Plate Thermometer, 6 mm
MIMS TC from Test Methods E1529, Directional Flame Thermometers, and 1.6 mm MIMS TCs with grounded and ungrounded
junctions. Total heat flux at the ceiling was measured using a Gardon gauge. Results showed that very different thermal exposures
are possible depending on the measurement method used. During the first ten minutes 10 min of a fire resistance test, the integrated
heat flux varies by a factor of two.
5.1.10 Reports by Sultan, M., (2006 and 2008) (22, 23) and Janssens, M., (2008) (18) have shown it is difficult to measure one
parameter in a fire resistance test (such as the furnace temperature) and calculate the other (heat flux or thermal exposure).
5.1.11 From the discussions in 5.1, it is highly recommended that both temperature and heat flux be measured independently
in fire tests.
5.2 Use for DFTs:
5.2.1 Although both cooled and non-cooled sensors can be used to measure heat flux, the results are generally quite different.
Water-cooled sensors are the direct reading Schmidt-Boelter or Gardon gauge designs that are used in some Committee E5
Methods [Test(Test Methods E2683 and E511, respectively, have been developed for these sensors by Subcommittee E21.08 ].).
5.2.2 There are three types of passive or un-cooled sensors that can be used to measure net heat flux. One is the hybrid sensor
(so-called High Temperature Heat Flux Sensor, HTHFS) developed by Diller, et al., at Virginia Tech. It is designed to measure heat
transfer to a surface without water cooling [Gifford, A., Hubble, D., Pullins, C., and Diller, T., 2010 (4)]. The HTHFS requires a
calibration factor that is a function of sensor temperature [Pullins and Diller, 2010 (24)]. Another is the so-called “direct write heat
E3057 − 19
flux sensor” which can be used at temperatures from 25 to 860°C860 °C [Trelewicz, Longtin, Hubble, and Greenlaw, 2015 (25)];
this gauge requires a calibration coefficient. The third is the Directional Flame Thermometer (DFT), which was developed at
Sandia National Laboratories (based on work in the UK) and elsewhere for measuring heat transfer in large sooty pool fires. DFTs
do not require a calibration factor, which may be viewed as a mixed benefit. The passive sensors typically have higher temperature
capability, based mainly on the Type K or N TC limit of about 1250°C.1250 °C. Even though they are water cooled, quite often
Gardon and Schmidt-Boelter gauges do not survive in temperatures due to fouling of the sensing surface, and other effects. DFTs
usually survive up to about 1100°C.1100 °C. They are very rugged, require no cooling, and are not susceptible to fouling of the
sensing surface. These characteristics simplify installation in a wide range of fire and other applications. This standard will only
address DFTs. See 10.2.2 for a more thorough discussion of heat flux gauge calibrations.
5.2.3 Early work on DFTs (and the data analysis techniques for them) focused on acquiring quantitative heat flux data to help
define the thermal conditions in large, liquid hydrocarbon pool or spill fires. Large pool fires can reach quasi-steady conditions in
times as short as a minute. As a result, Pool Fire DFTs were designed with 1.6 mm thick plates to provide rapid equilibration with
the fire (the maximum heating rate in these fires was approximately 30°C/s).30 °C ⁄s).
6. Apparatus
6.1 DFT Construction:
6.1.1 DFT apparatus consists of the DFT, mounting hardware, and a data acquisition system.
6.1.2 The DFT consists of two heavily oxidized Inconel plates with a ceramic fiber insulation layer sandwiched between the
plates. Alternately, to obtain a high emissivity surface one can apply high emissivity paint to the exposed surface. If paint is used,
one must be careful as at high temperatures some paints do not remain in place. A 1.6 mm OD, mineral-insulated, metal-sheathed
(MIMS) thermocouple (TC) is attached to each unexposed face. Typically the sheath material is Inconel. To optimize the response
in a variety of fire scenarios, there are three basic DFT designs. The original furnace DFT uses two 3 mm (nominal) thick plates;
the original pool fire DFT uses two 1.6 mm (nominal) thick plates. Both Inconel and SS have been used; Inconel 600 is
recommended because 304SS can sometimes form a scale that falls off the surface. The modified furnace DFT uses a 3 mm plate
facing into the furnace with a 1.6 mm back plate. Different plate thicknesses are used for different applications. Some special
designs have used a third plate and thermocouple. Some used in automotive fires were small and used intrinsic thermocouples
to provide very fast response. Fig. 1 shows the construction of a typical DFT, and Fig. 2 shows a photo of a typical DFT.
6.1.3 Plate thicknesses vary depending on the application. If faster response is desired a thinner plate is used (for example, 1.6
mm), or if slower response is acceptable, a thicker plate can be used (for example, 3 mm). It is advisable to never have the plate
thickness less than the TC sheath diameter, so the effect of the TC on the plate temperature measurement is minimized,
(seeminimized [see Figueroa, 2005 (26-28) for a detailed analysis).analysis]. Due to manufacturers recommended limits on MIMS
thermocouples, TC sheath diameters less than 1.6 mm are not recommended.
6.1.4 The Inconel plates are mounted parallel with a layer of ceramic fiber insulation material lightly compressed in between
the plates. The plates are held together with four bolts. One thermocouple is mounted on the inside surface of each of the Inconel
plates. A 12 mm wide by 25 mm long strip of nickel or Nichrome foil (for example, 0.08 mm thick) is formed over the tip of the
thermocouple and spot welded to the unexposed surface of each plate (see Fig. 3). This technique provides a good
thermo-mechanical attachment of the thermocouples, which is critical for good dynamic response In general the nichrome strip
should be as small as possible while still ensuring good mechanical contact with the surface (see ASTM Manual on Use of
Thermocouples, MNL12, page 183). Fig. 3 shows a typical TC installation.
6.1.5 Apparatus to mount the DFT near the test unit should be as small as possible to disturb the environment as little as
possible. The DFT should be mounted so that one of the Inconel plates is facing the environment one wants to measure. The DFT
has a 180° field of view, so the DFT should be oriented so that the entire environment is captured within that field of view.
6.1.6 The data acquisition system needs to be able to accurately record Type K or Type N thermocouples. Many such systems
exist and we will not discuss them further here.
7. Preparation of Apparatus
7.1 Fabrication of Directional Flame Thermometers:
7.1.1 See Fig. 1 for a sketch of a DFT.
7.1.2 See Fig. 2 for a photograph of a DFT.
2 2
7.1.3 Cut or shear two 1.6 or 3 mm ( ⁄16 or 0.12 in.) thick Inconel plates, 120 mm (4.75 in. ).
7.1.4 Drill 6.75 mm (letter drill H, ⁄64 in.) holes in four corners, leaving approximately one hole diameter from each edge.
7.1.5 Heat the plates in a furnace at approximately 1000°C1000 °C for 24 h to develop a stable, high absorptivity oxide layer.
If this is not possible, one can substitute a high emissivity paint that adheres to the plate at high temperatures.
Intrinsic thermocouples use bare wires welded to the metal surface of the DFT. This forms an “intrinsic” junction using the metal of the DFT. Intrinsic TCs have small
dynamic errors compared with ungrounded junction (sheathed) TCs but are not very robust and fail more often. MIMS TCs are fully sheathed and,and encase the TC junction,
and can be grounded or ungrounded.
Work at Sandia National Laboratories has shown the Inconel emissivity can vary considerably depending on the extent of the oxide layer. Values of about 0.85 have been
measured [for example, Figueroa, 2006 (26-28)], but others [Brundage, A., et al (29)] measured emissivity between 0.67 and 0.90. For highest accuracy the user should either
measure the emissivity measurements or apply high emissivity black paint with known emissivity.
E3057 − 19
FIG. 1 Basic Design of a Directional Flame Thermometer (Using 3 mm Thick Plates)
7.1.6 Use 1.6 mm ( ⁄16 in.) OD Inconel sheathed Type K (Chromel/Alumel) or Type N thermocouples (TCs) with an ungrounded
junction. Sand the oxide off the plate over a 3 by 1.5–3.9 cm (1.2 by 0.6–1.3 in.) area in the center of each plate.
7.1.7 Using 0.08 mm (0.003 in.) thick by 6.4 mm (0.25 in.) wide Nickel or Nichrome foil, form the foil strips tightly over last
25 mm (1 in.) of the TC and completely cover the TC tip. Then, spot weld the foil to the sanded area of the plate (don’t (do not
spot weld to the TC sheath). Provide a loop for stress relief. Do not weld the TC because the welding process might penetrate
through the sheath. See Fig. 3 (in the photo the strap is shorter than recommended).
3 3
7.1.8 Cut a 120 by 120 by 25 mm (4.75 by 4.75 by 1 in.) piece of 128 kg/m (8 lb ⁄ft ) ceramic fiber insulation and place between
the plates. Temperature dependent thermal properties of a Thermal Ceramics insulation called “Cerablanket” have been measured,
and those properties are provided in Annex A1. If a different insulation is used, it is important to measure the properties of that
material. There are other brands (for example, Kaowool by Morgan Thermal Ceramics, http://www.morganthermalceramics.com/
products/refractory-ceramic-fibre-rcf/blanket) that can be used, but the temperature dependent, thermal properties would need to
be measured.
7.1.9 Assemble the DFT using 4,four 9.53 mm (~ ⁄8 in. diameter) Inconel 600 or silver plated SS bolts and tubular spacers (for
example, made of 304 stainless steel) to compress the insulation layer to a thickness of 19 mm (0.75 in.). This compression is
important because the insulation thermal properties depend on thickness. See Annex A1.
7.1.10 Route the two TCs together out of the heated region. It is recommended that the TC sheaths be insulated until they reach
a room temperature location.
7.2 Fabrication of Mounting Hardware—Mounting hardware is not unique. Any mounting design that holds the DFT in place
but does not affect the environment is suitable. Any material that can withstand the temperatures in the environment of interest can
E3057 − 19
FIG. 2 Photo of Typical DFT
FIG. 3 Photo of Typical TC Installation
be used. Mild steel can be used if the melt temperature is not exceeded. But recall that the strength of mild steel at high
temperatures is reduced to approximately that of aluminum, so strength is much reduced. Stainless steel is the better, but more
expensive option.
8. Hazards
8.1 This standard does not purport to address all of the safety concerns, if any, associated with the use of DFTs. It is the
responsibility of the user of this standard to establish appropriate safety and health practices and determine the applicability of
regulatory limitations prior to use.
E3057 − 19
8.2 TheWarning—The only known potential hazard is related to the insulation in the DFT. Long durations in unventilated areas
with used insulation may be cause for concern because some of the ceramic fibers may become airborne. The user should contact
the insulation manufacturer for information about proper safety procedures related to the insulation. only known potential hazard
is related to the insulation in the DFT. Long durations in unventilated areas with used insulation may be cause for concern because
some of the ceramic fibers may become airborne. The user should contact the insulation manufacturer for information about proper
safety procedures related to the insulation.
9. Procedure
9.1 Fabricate DFT in accordance with 7.1 and the mounting hardware in accordance with 7.2.
9.2 Mount the DFT so that the field of view of the DFT encompasses the entire heat source.
9.3 Route the two thermocouple leads to a room temperature location. The TC sheaths should be protected from the heat source
by wrapping them with the same type of insulation used in the DFT. This protection can reduce the chance of “shunting” occurring.
See Appendix X3.2.
9.4 Calibrate the DAS by using a NIST traceable source to place a known input into each channel at a select number of
temperatures to ensure each DAS channel is reading properly.
9.5 Connect the TCs to a data acquisition system (DAS).
9.6 Do not calibrate the TCs used in the DFT, because for Type K TCs the calibration process can change the output of the TC
and therefore change the calibration (Reed,(30, 31Oct. ).1996).
9.7 Set up DAS to scan at a rate of about 1 Hz. Because the time constant of the TCs can be several seconds, there is no reason
to sample at a faster rate.
9.8 Measure pre- and post-test emissivity of the exposed DFT plate surfaces. These measurements can be used to estimate the
incident hetheat flux.
9.9 Carefully review thermocouple results to ensure anomalies are not present (for example, a spike in temperature that has no
basis for occurring).
9.10 Reduce data by means of one of the methods described below.
10. Calibration and Standardization
10.1 Apparatus Calibration—Two items should be calibrated when using DFTs, the data acquisition system and the
thermocouples. However, the thermocouples should be “calibrated” in a manner different than is typically done.
10.1.1 Thermocouple “Calibration”—The ASTM standards for accuracy are sufficiently good for use in DFTs. It has been
shown that Type K thermocouples are actually affected by the calibration process. As a result, the chemistry of the Chromel and
Alumel wires change when calibrated above about 320°C320 °C (Reed,(30, 31Oct.). 1996). TCs used in DFTs should be calibrated
up to the maximum temperature expected (for example, 1100°C).1100 °C). Therefore, one should not use the TC after calibration.
This apparent conundrum is resolved in the following manner. One can obtain additional TCs fabricated from the same spool of
wire from the manufacturer (preferably wire before and after the TCs used on the DFTs). Those additional TCs are sent through
the calibration process. If those calibrated TCs meet or exceed the ASTM standard for Type K TCs (in other words,
62.2°C62.2 °C or 60.75%60.75 % of reading in °C, see ASTM Manual on the Use of Thermocouples, MNL12, 1993), then one
assumes the remaining TCs from the same spool also meet the same ASTM standard. Experience has shown that in all cases the
calibrated TCs were more accurate than the ASTM standard. But the more accurate values are not used; one uses the accuracy
specifications from the Manual on the Use of Thermocouples, MNL12, so as not to assume accuracy better than the standard.
10.1.2 Data Acquisition System Calibration—The recommended method to assure that the data acquisition system uncertainty
is known is to calibrate each channel over the range of temperatures expected. For example, if one expects temperatures ranging
from 20 to 1100°C,1100 °C, one can calibrate at several set points over the range (for example, 20, 200, 400, 600, 800,
1100°C).1100 °C). The calibration is performed using a NIST traceable thermocouple simulator (for example, Fluke and Ectron
make such calibrators). A statistically significant sample (for example, 10s of samples) is taken at each set point, then the mean
and standard deviation of each set point on each channel can be estimated. Because this can be a very large data set, one can
average the data for all channels at all set points to provide a single estimate of the accuracy of all DAS channels. Typically this
source of uncertainty is small, but on occasion one finds a bad channel so the exercise is worthwhile.
10.1.3 Detailed Measurements of DFT—Detailed measurements of the DFT materials should be made before assembly, because
those measurements will be used in the data reduction process.
10.1.3.1 Measure and record the thickness of both Inconel plates as close to the center as possible. Estimate the accuracy of
those measurements.
10.1.3.2 Measure and record the spacer thicknesses to confirm they are 1.91 cm (0.75 in.).
10.1.3.3 Verify insulation (8 lb/ft ) is compressed to 1.91 cm thickness and is not forced out the sides of the plates. If the
insulation is different from Cerablanket, measure the temperature dependent thermal properties. See Annex A1.
E3057 − 19
10.1.3.4 Verify the TCs are mounted according to the procedure in 7.1.
10.1.3.5 Verify the plate surfaces have a stable oxidation layer or stable paint layer (if one cannot oxidize the plates at
1000°C,1000 °C, an alternate is to use a high absorptivity paint).
10.1.3.6 Measure and record the hemispherical total surface emissivity of exposed surfaces of each plate. This is used if one
desires to convert the net heat flux to incident heat flux. Estimate the emissivity uncertainty by making multiple measurements and
using the manufacturer’s reported accuracy.
10.2 Reference Standards and Calibration Curves and Tables:
10.2.1 Refer to ASTM Manual on the Use of Thermocouples MNL12 for Thermocouple accuracies.
10.2.1.1 For type K thermocouples, the ASTM Manual on the Use of Thermocouples MNL12 specifies the range to be from
0 to 1250°C1250 °C (32 to 2300°F).2300 °F). The “standard tolerance” is 62.2°C62.2 °C or 0.75%0.75 % of the reading in °C,
whichever is greater. The “special tolerance” is 61.1°C61.1 °C or 0.4%0.4 % of the reading in °C, whichever is greater.
10.2.2 Discussion on Calibration of DFTs:
10.2.2.1 Most heat flux gauges (for example, thin film, Gardon, Schmidt-Boelter) are designed to have a linear output with heat
flux. Data reduction is easy because the gauge comes with a calibration in the form of a sensitivity coefficient (in other words, xx
mV/unit of heat flux), and these sensitivity coefficients are made with reference to a NIST standard.
10.2.2.2 Traditional heat flux gauge calibrations use a radiative only heat source and seek to minimize convection effects (for
example, NIST, Medtherm, and so forth). Details of those calibration procedures will not be discussed in detail here. Typical
accuracies reported are 63%.63 %. An effort was initiated at NIST to develop a convective heat transfer calibration capability,
but the effort was not completed and no such facility exists at NIST. A convective calibration capability does exist at Virginia Tech
University under the guidance of Prof. Tom Diller.
10.2.2.3 A detailed analysis of heat flux gauges has led to a better understanding of under what conditions one can assume the
linear sensitivity coefficients are an accurate representation of the behavior of the heat flux gauges. A body of evidence has shown
that in fact the sensitivity coefficients developed for radiation only calibrations are not accurate for mixed heat transfer applications
where convection is non-negligible. (See references in the Introduction.)
10.2.2.4 For both Gardon and Schmidt-Boelter gauges what has been discovered is that the gauge sensitivity for an equivalent
level of radiant heat flux is different than for the same level of convective heat flux. One might reasonably ask: “So what?” How
much does this affect readings? These are good questions that will be discussed below. First, a simple example will be discussed
to show possible effects of the problem.
10.2.2.5 For a radiation only heat flux measurement, one records the voltage output and multiplies the output by the sensitivity
coefficient provided by the manufacturer to get an estimate of heat flux. This is expressed in Eq 1:
q 5 v·s (1)
q 5 v·S (1)
where:
q = the heat flux,
v = the voltage output, and
S = the sensitivity coefficient (unit of heat flux/volt).
10.2.2.6 What is normally done is the user assumes the sensitivity coefficient is the same for all modes of heat transfer, therefore
one estimates the heat flux using Eq 1. But it has been shown that one should not assume the sensitivity coefficients are the same,
therefore, a different data reduction method is appropriate. How then does the user reduce his data?
10.2.2.7 One method to reduce the data is to use a linear combination of the fluxes as shown in Eq 2:
q 5 v· F · S 1 F · S (2)
~ !
rad rad conv conv
where:
F and F = the fractions of the total heat flux attributed to radiation and convection, and
rad conv
S and S = the radiative and convective sensitivity coefficients.
rad conv
10.2.2.8 What makes Eq 2 difficult to use is that most of the terms are not known to high accuracy. Only S is known to good
rad
accuracy (from the manufacturer); none of the others are known except the output voltage.
10.2.2.9 Eq 2 assumes that both the radiative and convective sensitivity coefficients are linear with heat flux. This is true for
radiative flux but it has not been shown the case for convection. But Eq 2 serves to make the point, as follows.
10.2.2.10 In Eq 2, assuming the radiative and convective sensitivity coefficients are equal, and the sum of the radiative and
convective fractions equals 1.0, then Eq 2 is reduced to Eq 1. This is in fact what every user of heat flux gauges assumes, whether
known or not, when using a calibration performed in radiation only.
10.2.2.11 For Gardon and S-B gauges the convective sensitivity coefficients can be quite different from radiative sensitivity
coefficients. For example, Gifford, et al., 2010 (4, 10), showed that for S-B gauges the convection sensitivity coefficients can be
E3057 − 19
up to about 20%20 % different than the radiation sensitivity coefficients. Further complicating matters is that the convective
sensitivity coefficients are different for shear and stagnation flows. Similar results have been shown for Gardon gauges by Kuo and
Kulkarni, 1991 (8).
10.2.2.12 Why would Gardon and S-B gauges have different sensitivity coefficients in radiative and convective environments?
A qualitative understanding is possible by understanding how the gauges are constructed. Gardon gauges have a very thin sensing
element that has a parabolic temperature profile from the center of the element to the edge when exposed to a uniform radiative
heat flux. But during a convective shear flow, the temperature profile can “tilt” to the downstream side of the sensing element.
There is good reason to expect that the sensitivity coefficient for a Gardon gauge in shear flow might be different than for the same
magnitude of radiative heat flux.
10.2.2.13 Similarly, for S-B gauges, one assumes a uniform exposure of radiative flux over the sensing element. In shear flow
this is not the case so again one might expect different sensitivities for radiative and convective fluxes.
10.2.2.14 Therefore, because sensitivity coefficients in radiative and convective heat transfer environments are different when
using Gardon and S-B gauges, and there is no NIST traceable convective heat flux calibration capability, and because making an
estimate (for example, using Eq 2) of the heat flux in mixed heat transfer environments has a number of uncertain parameters, it
is difficult to fully understand the uncertainty of these types of gauges when used in a mixed mode heat transfer environment.
Therefore, a different method to estimate heat flux was developed.
10.2.2.15 Characteristics of this “different” method were as follows:
(1) The gauge had to be rugged and survive temperatures up to about 1100°C.1100 °C.
(2) The gauge should not be actively cooled.
(3) The gauge does not use a single sensitivity coefficient, so one does not suffer from the issues discussed above (different
radiative and convective sensitivity coefficients).
(4) The gauge is simple so can be analyzed by means of a thermal model.
(5) The gauge responds to both radiation and convection so one measures the total heat flux to a surface.
10.2.2.16 DFTs satisfy all of the desired characteristics listed above. But the downside is the complication of data reduction and
a more complicated uncertainty analysis. The uncertainty analysis for DFTs is more complicated, and depends on the data reduction
method used (in other words, energy storage method, inverse heat conduction, inverse filter function).
10.2.2.17 With DFTs, one trades the convenience of having a linear sensitivity coefficient with known and traceable accuracy
and a relatively complicated gauge design (S-B and Gardon) tofor a much simpler design (in other words, DFTs) with a more
complicated data reduction and uncertainty analysis.
10.2.2.18 The discussion above sheds light on the advantages of using a gauge that does not require a calibration, assuming one
has the tools to reduce the data and analyze the uncertainties when using DFTs.
11. Calculation or Interpretation of Results
11.1 General:
11.1.1 The data analysis techniques in this section use the DFT plate temperature histories and material properties to provide
quantitative estimates of net heat flux data over the entire test duration. The inverse heat conduction analysis and energy storage
methods both calculate the net heat flux post-test. The inverse filter function method provides near real-time estimates
...