ABSTRACT
Title of Thesis: DEVELOPMENT AND IMPROVEMENTS
OF THE CONTROLLED ATMOSPHERE
PYROLYSIS APPARATUS
Grayson T. Bellamy
Masters of Science, 2022
Thesis directed by: Dr. Stanislav I. Stoliarov
Department of Fire Protection Engineering
Previously developed, the Controlled Atmosphere Pyrolysis Apparatus (CAPA) was de-
signed to address the needs of a well-characterized gasification apparatus from which material
properties could be derived. Although the data from CAPA has been well-validated through mod-
eling and other various means, only a single functional and published version of the apparatus
exists which hinders widespread acceptance. Additional concerns and questions about the de-
sign remain which warrants further improvement and investigation. In this work, elements of
CAPA are revisited, redesigned, and improved to provide a more defined environment for bench-
scale material evaluation. Existing geometrical constraints were maintained to allow for a direct
comparison to previous work and confirm previous characterizations.
In pursuit of these goals, several improvements were made. Implementation of an integrally
water-cooled chamber was accomplished through additive manufacturing, allowing chamber tem-
peratures to remain steady throughout the duration of a test, reducing additional radiant heat flux
to the sample. Measurement resolution was increased with an upgraded thermal imaging cam-
era, mass balance, and data acquisition system. Full analysis of the gas temperatures, chamber
temperatures, oxygen concentration, and overall thermal environment was performed to replicate
previous results and provide quantitative information for use in a pyrolysis modeling.
Characterization experiments confirmed that heat flux profiles and gas temperatures are
within the experimental uncertainty of both apparatuses. Chamber temperatures were reduced,
providing for more clear boundary conditions and simplified modeling. Experimental results
of this improved version of CAPA were compared against previous experiments conducted on
poly(methyl methacrylate) (PMMA) and oriented strand board (OSB). Mass loss rate and surface
temperature data were comparable indicating the apparatus performs as intended. Several prob-
lems and concerns were identified in this process for further study. This work provides further
confirmation of the usefulness and accuracy of CAPA for use in analyzing the thermal decompo-
sition of materials.
DEVELOPMENT AND IMPROVEMENTS OF THE CONTROLLED
ATMOSPHERE PYROLYSIS APPARATUS
by
Grayson T. Bellamy
Thesis submitted to the Faculty of the Graduate School of the
University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Masters of Science
2022
Advisory Committee:
Dr. Stanislav I. Stoliarov, Chair/Advisor
Dr. Arnaud Trouve?, Defense Committee
Dr. Mark McKinnon, UL FSRI/Defense Committee
? Copyright by
Grayson Bellamy
2022

Acknowledgments
First, I would like to thank my family for their continuous support throughout my graduate
studies and academic career. Your words of encouragement and encouragement to always push
further have been a continuous motivation for me.
Next, I would like to thank all the staff at Underwriters Laboratories Fire Safety Research
Institute for their motivation and support in completing my thesis. Thank you to Dr. Dan
Madrzykowski and Dr. Steve Kerber for first bringing me on as an intern in 2019 and continually
supporting my education and development. Specifically to my advisor, Dr. Mark McKinnon,
thank you for all of your help in running experiments, providing insight, and helping my de-
velop as a researcher. Your passion for research and wealth of knowledge has defined my path of
inquisition and culminated in my thesis and path of future work.
To my advisor at the University of Maryland, Dr. Stanislav Stoliarov, thank you for your
support and encouragement in completing this process. Your deep insight into material flamma-
bility has proved invaluable in helping guide me when things do not appear as expected. You
have constantly pushed me to do my best and supported me through this process.
ii
Table of Contents
Preface ii
Foreword ii
Dedication ii
Acknowledgements ii
Table of Contents iii
List of Tables v
List of Figures vi
List of Abbreviations viii
Chapter 1: Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Cone Calorimeter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.2 Fire Propagation Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.3 Controlled Atmosphere Pyrolysis Apparatus . . . . . . . . . . . . . . . . 7
1.3 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Chapter 2: Apparatus Design 12
2.1 Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Heater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Gasification Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Sample Holder Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5 Data Acquisition System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5.1 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5.2 Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5.3 Mass Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.4 Infrared Camera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5.5 Gas Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.6 Pyrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5.7 Video . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
iii
Chapter 3: Characterization 32
3.1 Heater Temperature to Radiant Flux . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2 Heat Flux Uniformity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Gas Flow Uniformity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.4 Gas Flow Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Chamber Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.6 Oxygen Concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.7 Sample Thermal Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Chapter 4: Experimental Comparison 50
4.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.1.1 Poly(methyl methacrylate) . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1.2 Oriented Strand Board . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 Testing Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.1 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.2 Test Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3.1 Poly(methyl methacrylate) . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.2 Oriented Strand Board . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Chapter 5: Conclusion 66
Bibliography 77
Bibliography 77
iv
List of Tables
3.1 Gas temperature correlation coefficients . . . . . . . . . . . . . . . . . . . . . . 40
3.2 Gas temperature rise time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.1 Test durations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Filtering cutoff frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
C.1 Material properties used for CFD model . . . . . . . . . . . . . . . . . . . . . . 76
v
List of Figures
1.1 Typical setup diagram of the cone calorimeter [1] . . . . . . . . . . . . . . . . . 3
1.2 Cone calorimeter exhaust system [1] . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Controlled-atmosphere cone calorimeter sample chamber [2] . . . . . . . . . . . 6
1.4 Schematic of the Fire Propagation Apparatus [3] . . . . . . . . . . . . . . . . . . 7
1.5 Photo of CAPA I [4] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Drawing of CAPA II [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1 Rendering of CAPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Frame subassemblies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Heater control box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Diagram of gasification chamber assembly (all dimensions in mm) . . . . . . . . 17
2.5 Three water cooling designs and their resultant temperature distribution under
uniform 50 kWm?2 heat flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.6 Gasification chamber and sample cup . . . . . . . . . . . . . . . . . . . . . . . . 22
2.7 Sample holder assembly (all dimensions in mm) . . . . . . . . . . . . . . . . . . 22
2.8 Data acquisition box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.9 Gasification chamber thermocouple locations . . . . . . . . . . . . . . . . . . . 26
2.10 Medtherm GTW-10-32-485A Schmidt-Boelter heat flux gauge (all dimensions in
inches) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.11 Diagram of optical path of infrared camera and relative positioning to components 30
3.1 Heater temperature versus total heat flux at 40mm . . . . . . . . . . . . . . . . . 34
3.2 (a) Radial and (b) polar radiant heat flux distribution at h/H = 0, compared to
that found by Swann et al. [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Gas flow velocities with location, units in [ms?1] . . . . . . . . . . . . . . . . . 38
3.4 Gas temperature probe construction . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 Time-resolved gas temperature measurements in the present version of CAPA,
CAPA II, and that recorded by Swann [5] at (a) 25 kWm?2 and (b) 60 kWm?2 . 41
3.6 Time-resolved chamber temperature measurements in the present version of CAPA
and CAPA II at (a) 25 kWm?2 and (b) 60 kWm?2 . . . . . . . . . . . . . . . . 44
3.7 Oxygen concentration measuring setup . . . . . . . . . . . . . . . . . . . . . . . 46
3.8 Copper plate temperature and simulation under 25 kWm?2 radiant exposure . . . 48
3.9 Copper plate thermocouple temperature in present study and CAPA II under
25 kWm?2 radiant exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 Typical CAPA sample assembly . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Prepared (a) PMMA and (b) OSB samples . . . . . . . . . . . . . . . . . . . . . 56
vi
4.3 Pocket formation and bulging behavior of PMMA at 25 kWm?2. Picture taken
with sample cut in half and cut ends exposed. . . . . . . . . . . . . . . . . . . . 58
4.4 Post test photograph of PMMA at 25 kWm?2 . . . . . . . . . . . . . . . . . . . 59
4.5 PMMA (a) mass loss rate and (b) bottom temperature at 25 kWm?2 radiant ex-
posure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.6 PMMA (a) mass loss rate and (b) bottom temperature at 60 kWm?2 radiant ex-
posure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.7 Post test photograph of OSB at 35 kWm?2 . . . . . . . . . . . . . . . . . . . . . 63
4.8 Oriented strand board (a) mass loss rate and (b) bottom temperature at 35 kWm?2
radiant exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.9 Oriented strand board (a) mass loss rate and (b) bottom temperature at 65 kWm?2
radiant exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
B.1 CFD model overall mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
B.2 CFD model fine mesh for water channel . . . . . . . . . . . . . . . . . . . . . . 74
vii
List of Abbreviations
CAPA Controlled Atmosphere Pyrolysis Apparatus
FAA Federal Aviation Administration
FSRI UL?s Fire Safety Research Institute
FDS Fire Dynamics Simulator
MLR Mass Loss Rate
NFPA National Fire Protection Association
NIJ National Institute of Justice
NIST National Institute of Standards and Technology
OSB Oriented Strand Board
PID Proportional-Integral-Derivative
PMMA Poly(methyl methacrylate)
RH Relative Humidity
UL Underwriters Laboratories
UHMW Ultra High Molecular Weight Polyethylene
viii
Chapter 1: Introduction
1.1 Motivation
The increasing use of fire models for uses ranging from the design of fire protection sys-
tems to the investigation of fire scenarios demands adequate material property data for accurate
simulations. Although many of these thermophysical properties can be obtained through stan-
dardized methodologies, quantifying the thermal transport properties at elevated temperatures
and through the decomposition process remains a complex and uncertain task. Additionally,
properties collected through independent means at the milligram scale need to be validated as
part of a comprehensive material pyrolysis model against accurate and well-defined larger-scale
test data. The Controlled Atmosphere Pyrolysis Apparatus (CAPA) is an accepted solution to
this problem. Although the data from CAPA has been well-validated through modeling and other
various means, only a single functional and published version of the apparatus exists, which hin-
ders widespread acceptance. Previous work requires replication and improvements in pursuit of
greater confidence in boundary conditions and experimental results.
1
1.2 Background
Several bench-scale fire testing apparatuses currently exist and have become widely ac-
cepted.
1.2.1 Cone Calorimeter
The cone calorimeter is a bench-scale fire testing apparatus first developed at NIST in the
1980s [7]. Later implemented in ASTM E1354 [8] and ISO 5660-1 [9], the cone calorimeter
measures the flammability of a sample through oxygen-consumption calorimetry in which the
heat release rate (HRR) of an ignited sample is quantified through the amount of oxygen con-
sumed by the produced flame [7, 10]. The cone calorimeter can also measure the effective heat
of combustion, mass loss rate, time to ignition, and smoke production of a material [1].
The basic cone calorimeter consists of a cone-shaped electrical heater positioned to face a
sample holder that can be rotated into a horizontal or vertical configuration, an exhaust system
capable of flow rate and oxygen concentration measurements, an electric ignition system, and
a load cell for capturing sample mass evolution [8]. This typical setup is shown in Fig. 1.1.
Traditionally, tests are conducted in an open-atmosphere configuration, allowing sufficient access
to ambient air to support combustion.
Samples in the cone calorimeter are exposed to a radiant heat flux provided by a truncated
cone-shaped electrical heater that ranges from 0 kWm?2 to 100 kWm?2 [7]. The heater consists
of an electric heating coil consisting of a resistive heating element packed in a refractory material
and swaged into a protective Incoloy sheath [1]. The heating coil is wound into a conical shape
and encased in a housing packed with fibrous insulation [11]. The form of the heater delivers a
2
Figure 1.1: Typical setup diagram of the cone calorimeter [1]
nearly uniform heat flux over the sample surface at its standard height of 25mm above the sample
in the horizontal position while allowing the gaseous products produced to flow upwards and be
captured by the exhaust system [12].
The gases and products produced by the burning sample are collected by the exhaust pow-
ered by a centrifugal exhaust fan capable of producing flows of 0.012m3 s?1 to 0.035m3 s?1 [8].
Flow rate is monitored by an orifice plate flow meter equipped with a differential pressure trans-
ducer and thermocouple [1]. Oxygen consumption is continuously measured from a gas sampling
ring which permits the calculation of the heat release rate of the sample as a function of time.
This sampling system also allows other gas analyses such as CO, CO2, and HCl. An additional
ancillary measurement that can be performed is smoke obscuration through the use of a laser
photometer system that measures the attenuation of a helium-neon laser through the optical path
of the exhaust system [1]. A detailed schematic of the exhaust system in the cone calorimeter is
shown in Fig. 1.2.
3
Figure 1.2: Cone calorimeter exhaust system [1]
Tests specimens are typically 100mm by 100mm with a thickness of up to 50mm. Sam-
ples are then conditioned at a temperature of (23? 3) ?C and relative humidity of (50? 5)% [8].
They are then wrapped in a single layer of aluminum foil on the sides and bottom face and placed
in the sample holder with a layer of low-density insulation of at least 13mm thickness [8]. Other
accessories such as a wire retaining grid may be used based on the behavior of the sample un-
der test conditions. Mass of the sample holder assembly is monitored during the test using a
single-axis load cell capable of measuring a 500 g live load with an accuracy of at least 0.1 g [8].
A modification of the cone calorimeter has been developed to investigate flammability be-
havior under vitiated atmospheres. First described by Babrauskas et al. [13], the controlled-
atmosphere cone calorimeter (CACC) enables the user to feed a prescribed gas flow into an en-
closed sample chamber instead of the typical open-atmosphere configuration of the cone calorime-
ter. One implementation of this sample chamber design is shown in Fig. 1.3. To control the atmo-
sphere, a gas mixture is introduced into the sealed combustion chamber and is allowed to interact
with the sample. From this point onwards, the setup is functionally the same as a standard cone
4
calorimeter.
Proper use of this instrument requires careful sealing of the sample chamber to properly
control the atmosphere and prevent leakage from the pressurized space [13]. This is also a func-
tion of maintaining balanced supply and exhaust rates to provide this equilibrium and not over- or
under-pressurize the chamber. Additionally, no standardized design is currently formed, which
can lead to differences in results between labs, particularly under vitiated conditions [14].
The cone calorimeter has previously been used for inverse analysis and optimization exper-
iments [15?17].
1.2.2 Fire Propagation Apparatus
The Fire Propagation Apparatus (FPA) is another standardized apparatus used for measur-
ing the fire behavior of materials, initially developed by Factory Mutual Research Corporation
(FMRC), and standardized in ASTM 2058 and ISO 12136 [18, 19]. It can measure the heat re-
lease rate, mass loss rate, effective heat of combustion, and time to ignition of a sample as it
burns or thermally decomposes. Similar to the cone calorimeter, the heat release rate in the FPA
is typically quantified using the oxygen consumption method.
The design of the FPA utilizes a unique testing environment. Specimens are placed within
a quartz tube that is provided with some combination of nitrogen and oxygen gas at a flow rate of
200 SLPM. Volumetric oxygen concentrations of 0% to 40% are used [20] which can replicate
a wide range of fire conditions. Gas flows upward around the sample and out of the quartz tube
into an exhaust system. Heat to the sample is provided with four infrared tungsten lamps that
can produce a maximum heat flux of 110 kWm?2 [18]. The nature of the heater and housing
5
Figure 1.3: Controlled-atmosphere cone calorimeter sample chamber [2]
design requires them to operate at far higher temperatures than that encountered in most fires and
that produced by a cone heater which changes the spectral distribution of radiation to the sample.
When desired, an ethylene pilot burner is placed 10mm above the surface to ignite the sample.
A diagram of this setup is shown in Fig. 1.4.
Samples are usually 100mm ? 100mm square or 100mm-diameter cylindrical. Several
testing modes are commonly used based on the desired data: the ignition test, combustion test,
pyrolysis test, or fire propagation test. The pyrolysis test is conducted horizontally in an inert
atmosphere, preventing gas-phase combustion, and is of most interest here. This could be con-
sidered most similar to a cone calorimeter with a 100% nitrogen atmosphere.
The FPA provides much greater control over the ambient environment than the cone calorime-
ter. The issues with maintaining a prescribed atmosphere in the controlled atmosphere cone
calorimeter are not present in the FPA. Therefore, greater confidence is had in the results for use
6
Figure 1.4: Schematic of the Fire Propagation Apparatus [3]
in modeling purposes. The most significant discrepancy in results comes from the differing spec-
tral distribution of radiation. Depending on the spectral radiative material properties (reflectance,
absorptance, and transmittance), the results obtained can be very different and potentially not
representative of actual fire conditions [20, 21]. The FPA has also previously been used as a tool
for material property estimation of inference [22,23]. More recently, Chaffer redesigned the FPA
to improve its capabilities and ease of manufacturing [20].
1.2.3 Controlled Atmosphere Pyrolysis Apparatus
The Controlled Atmosphere Pyrolysis Apparatus (CAPA) is a custom-designed apparatus
for determining materials? thermal transport properties throughout the decomposition process.
The mass loss rate data also serves as an excellent validation of a comprehensive material property
7
model for a material given the highly-defined conditions.
The original version of CAPA was designed by Xuan Liu and underwent several design
iterations to effectively control the oxygen against the sample [4]. The final design consisted of
a square enclosure with nitrogen inlets at the bottom and a square sample holder. A door on the
front of the device allowed access to the interior and placement of the sample. A final oxygen
concentration of 3% was obtained at a flow rate of 200 SLPM, Liu noted that lower concentrations
could be achieved by increasing the flow rate. Depending on the nature of the material tested,
even this concentration can cause notable effects in the test results due to the introduction of
oxidation reactions. Heat flux uniformity at the corners of the 80mm sample dropped to 82%.
Mass measurements were taken; however, the back surface temperature was only evaluated with
a thermocouple adhered to the surface. A photo of this version, named CAPA I, is shown in
Fig. 1.5. The use of an infrared camera to measure back surface temperatures through a gold
mirror was later added [24?26] which allowed simultaneous measurement with the mass loss
rate. Problems with this version included significant heating of the chamber walls during the test
which radiated heat to the back surface of the sample in an uncontrolled manner.
This version was used as a means of material property estimation for corrugated card-
board [26] and a multilayer floor covering [27], and a variety of charring polymers and compos-
ites [28, 29].
The more present version of CAPA, named CAPA II, was first conceived by Mark McK-
innon and constructed by Swann et al. [6, 30]. This was the most significant redesign of CAPA
and a large change in the boundary conditions. It was set up as a completely separate platform
and no longer an attachment to the standard cone calorimeter. A cone heater was still used, but it
was mounted on a slide that allowed it to be slid out of the way, increasing access to the sample
8
Figure 1.5: Photo of CAPA I [4]
compartment and instantly applying the radiant heat flux. The square geometry was changed to
circular, including the chamber and sample. Water-cooling was added to the chamber walls via an
impressed copper tube that reduced reradiation to the specimen. The design is shown in Fig. 1.6.
Extensive characterization work was carried out on CAPA II to define the external boundary
conditions. Direct numerical simulation (DNS) was performed to calculate the convective heat
flux against the top and bottom surfaces. Radiation was mapped as a function of the height, radial
distance, and orientation. This allowed intumescent materials to be accurately modeled as they
grew and the surface changed position within the chamber. From this new design, the oxygen
concentration was reduced to 0.6% around the sample.
CAPA II has been used to validate and determine the thermal transport properties of various
materials, including PVC, PMMA, oriented strand board, and flame retardant polymers [5, 31?
33, 33, 34].
9
Figure 1.6: Drawing of CAPA II [5]
Compared to the cone calorimeter, CAPA has several advantages: the atmosphere can be
tightly controlled to an oxygen concentration <1%, the boundary conditions are well-characterized,
heat flux uniformity is improved, and back surface temperature can be measured simultaneously
with mass loss rate. Compared to the FPA, CAPA additionally has the advantage of using a radi-
ant source that more closely matches the emissions from the typical fire and does not drift over
time.
1.3 Approach
Given the advantages of CAPA over other bench-scale fire testing apparatuses in terms of
its defined boundary conditions and unique set of experimental data, it is desirable to iterate on
its design to improve measurement capabilities and replicate previous data. To this end, a new
version of CAPA was sought to be constructed at UL?s Fire Safety Research Institute in Columbia,
MD. Similar to previous studies, it will complement a wide range of material testing devices at
all scales to fully characterize a material?s properties in the context of fire research.
The previous iterations of CAPA identified several deficiencies and unaddressed questions,
10
such as the heater mount, water-cooling system, heat flux gauge positioning, and convective
boundary conditions. Certain limitations of the data acquisition system also limited the rate and
quality of data that could be captured. The goals here were to maintain the existing structure
of CAPA so that the existing data is comparable and to replicate previous results to prove this.
Many components are revisited, documented, and justified here to better understand and accept
the apparatus as a whole.
Boundary conditions from CAPA 2.0 are checked, and new ones are generated to define the
entire thermal environment better and provide for its ability to be represented in a comprehensive
pyrolysis model. An experimental study was performed with materials previously tested in CAPA
to identify any differences in design and prove that the results are replicable.
11
Chapter 2: Apparatus Design
2.1 Frame
The CAPA assembly is placed on top of a Speirs Robertson AMP passive vibration iso-
lation platform which utilizes pneumatic damping chambers to insulate the upper surface from
environmental disturbances. A pneumatic pump inflates each of the four chambers to 500 kPa
which separates the upper and lower portions. Vibrational frequencies above 6Hz, well below
that typically encountered in buildings, can be effectively eliminated. Isolation from external
vibration sources reduces potential noise impact on mass balance measurements.
The frame structure is built on a 762mm-by-762mm solid-aluminum optical breadboard,
referenced in Fig. 2.1. Across the plane of the breadboard, 1/4-20 UNC tapped holes are located
on a 25.4mm spacing which allows for the mounting of frame components rigidly to the surface
and along defined axes. Using tapped and predefined hole positions facilitates adding features
or mounting hardware in the future with minimal effort. The surface of the breadboard is coated
with a matte-black anodized finish to mitigate interference from heater reflections on infrared
camera measurements. The breadboard is a consistent platform on which the entire apparatus
assembly can be moved while maintaining the relative positions of components.
T-slotted extruded aluminum profiles are used to construct the primary structure of CAPA.
T-slot framing provides an easy-to-assemble, modular, and cost-effective way of building an ap-
12
Figure 2.1: Rendering of CAPA
paratus in which components are designed to be replaced in the event of a failure and added to
if new functionality is desired. This is made possible by using standard profiles, connectors, and
fasteners. Compared to steel, 6105-T5 aluminum makes the structure lightweight and machin-
able with common tools. When feasible, all profiles are black anodized to minimize infrared
reflections.
The frame comprises two separate subassemblies: the gasification chamber mount and the
heater mount, shown in Fig. 2.2. The gasification chamber mount is built out of 25.4mm by
25.4mm T-slot framing and rigidly secured to the optical breadboard via the mounting holes. Its
primary purpose is to support the gasification chamber over the mass balance and affix the gold
mirror used to provide visual access to the back of the sample for infrared measurements.
The heater mount structure provides a mechanism for the heater to be moved over the
13
B
B
(a) HeSaECtTeIOrN mount (b) Gasification chamber mountB B-BB  SECTION B-BSCALE 1 : 2
SCALE 1 : 2
Figure 2.2: Frame subassemblies
SOLIDWORKS Educational Product. For Instructional Use Only.
gasification chamber from the storage position.SOLINDWOoRKrS Emducaatiolnlaly Pro,duictn. For Ianstrucctioonanl Uese Oncly.alorimeter, the sample is
initially shrouded by a shutter that is swung out of the way at the beginning of a test. While
this method provides instantaneous exposure, the heater experiences a sudden change in radiant
feedback, which causes the heat flux to fluctuate as the heater temperature stabilizes. The choice
of the sliding-heater mechanism mitigates this effect and allows more direct access to the interior
of the gasification chamber. The vertical linear positioner attached to this portion of the assembly
allows further adjustment of the heater height above the sample.
The bottom part of the heater mount is made of 38.1mm-by-38.1mm T-slot framing, se-
cured to the optical breadboard. The top of this structure includes a V-groove railing on which the
upper portion of the assembly that holds the heater rides. The previous version of CAPA utilized a
UHMW-PE sliding mechanism that encountered difficulties moving the heater due to the buildup
of particulates on the rail, creating a high friction surface. The track required continuous mainte-
nance to ensure excessive force was not needed that could lead to vibrations that would manifest
themselves in mass measurement errors. The design of CAPA here uses a V-groove railing which
is not as susceptible to the effects of particulate buildup because the shape of the wheels cause
14
displacement of any foreign objects on the track. Sealed ball bearings provide smooth motion,
and an operator can easily replace them in the case of failure. The track is embedded into a T-slot
railing, allowing it to maintain the modularity of design and integrate with the existing structure.
The top part of the heater mount rides on the bottom portion via the aforementioned V-
groove rails. Four sealed V-groove ball bearings on each corner of the top portion distribute
the load and ensure rigidity of the assembly. A linear positioner mounted to this portion provides
50mm of vertical adjustability for the heater at a resolution of 2.54mm per turn of the handwheel.
This precise adjustment allows the heater height above the sample to be prescribed within a tight
tolerance. The heater mounts to the positioner through two mounting plates adjusted through
washers and seven bolts.
2.2 Heater
The heater utilized on CAPA is the same design as that used on a commercially available
cone calorimeter, consisting of a cone-shaped housing containing a wound electric heating coil.
Wilson et al. found this design to deliver nearly uniform radiant heat flux over a sample sur-
face [12]. In contrast to the cone calorimeter, the heater on CAPA is positioned 40mm above
the sample surface, although the functionality is present to reduce this to as little as 30mm. In
the previous version of CAPA, a significant temperature increase of the gasification chamber
walls under high heat fluxes restricted this distance. A greater height can be beneficial for certain
intumescing materials whose growth extends significantly upwards into the heater.
The cone heater is controlled through a Watlow EZ-ZONE? PM PID controller. Three
thermocouples attached to the housing of the cone heater, connected in parallel to the controller
15
input, allow the controller to sense the average temperature of the heater housing. The controller
attempts to reach and maintain a user-defined setpoint by switching an OMRON G3NA solid-
state relay that connects 240 VAC power to the heating coil. The maximum rated current of
this system is 20A which enables a heat flux of approximately 80 kWm?2 at 40mm below the
heater. Up to 100 kWm?2 is possible given a higher amperage circuit and relay. The controller
communicates through an RS-485 connection which allows control of the setpoint and recording
of the heater temperatures from an external interface. These components are packaged into a
slant-top control enclosure, shown in Fig. 2.3.
2.3 Gasification Chamber
The gasification chamber is perhaps the most critical designed part of CAPA due to its func-
tional requirements and intense operating environment. The two primary objectives of the cham-
ber are to (1) provide a means by which external gas can be fed into the system and distributed
around the sample surface and (2) provide uniform boundary conditions that can be recorded and
later used as inputs to a comprehensive pyrolysis model. Due to extensive work on characterizing
the boundary conditions and performance of the chamber used on the previous version of CAPA,
geometric specifications were maintained to allow the reuse of this data with minimal verification
experiments. The gasification chamber assembly consists of three main parts: the outer chamber
wall, the inner chamber wall, and the baseplate. A diagram of this assembly with overall basic
dimensions is shown in Fig. 2.4.
The upper half of the outer chamber wall is painted with a high-emissivity paint to reduce
reflections from the heater onto the sample. Due to the high heat flux environment in which
16
Figure 2.3: Heater control box
8 7 6 5 4 3 2 1
D 168.00 D
154.00
A Outer Wall 102.00
90.00
C C
Inner Wall
Baseplate
B B
SECTION A-A
A Figure 2.4: Diagram of gasification chamSCbALEe 2 :r 1 assembly (all dimensions in mm)
A NAME DATEDRAWN 7/11/2022 UL FSRI A
17 CHECKED TITLE:UNLESS OTHERWISE SPECIFIED:
DIMENSIONS ARE IN MILLIMETERS
TOLERANCES:
ANGULAR: MACH  1.0
1 PL  .3   2 PL  .15
PROPRIETARY AND CONFIDENTIAL
THE INFORMATION CONTAINED IN THIS DRAWING IS THE SOLE 
PROPERTY OF UNDERWRITERS LABORATORIES INC.  ANY SIZE DWG.  NO.
REPRODUCTION IN PART OR AS A WHOLE WITHOUT THE REV
WRITTEN PERMISSION OF UNDERWRITERS LABORATORIES INC. IS  
PROHIBITED. D A
MATERIAL SCALE: 1:1 WEIGHT: SHEET 1 OF 1
SO8LIDWORKS Educational Produ7ct. For Instructional Use Only.6 5 4 3 2 1
76.20
108.00
the chamber operates, the walls must be water-cooled to ensure their integrity and time-constant
boundary conditions. The previous version of CAPA accomplished this with a copper cooling coil
impressed into cutouts in the aluminum wall. Although this was adequate in protecting the wall?s
integrity, heating still took place, which could add additional radiant feedback to the sample over
time.
As a means of addressing this issue, it was sought to use integrated water cooling channels
in the walls. One method to accomplish this is the emerging technology of direct metal laser
sintering (DMLS). DMLS uses an ultra-fine metal powder sintered together progressively with a
laser, layer-by-layer. This technique enables a metal part to contain completely enclosed features
which may be difficult or impossible to accomplish via traditional machining.
Several water-cooling channel designs were considered and analyzed through CFD simu-
lations to select an optimal design, three of which are shown in Fig. 2.5. CFD simulations were
set up in ANSYS 2021 R2 using simplified geometry of the channels inside the outer chamber
wall. The mesh created is given in Appendix B and material properties in Appendix C. Water
was defined as the fluid, and the inlet pressure was prescribed according to a curve generated by
the local water supply. The outlet pressure was set as atmospheric because of the minimal extra
tubing required to reach the floor drain. The inner surface of the cylinder was exposed to a uni-
form 50 kWm?2 net heat flux. Although this does not represent the spatial distribution of actual
conditions, it allows a simplified scenario capable of testing relative cooling performance. Values
here should not be considered representative of the actual service conditions. This analysis only
investigates the outer chamber wall, while the inner wall was designed to be similar within certain
geometrical constraints.
The temperature distributions of the water-channel interface determined through the com-
18
(a) (b)
(c)
Figure 2.5: Three water cooling designs and their resultant temperature distribution under uni-
form 50 kWm?2 heat flux
19
putational model are shown in Fig. 2.5. The design choice was based on two factors: the maxi-
mum temperature of the wall and the axial temperature uniformity. A lower maximum tempera-
ture ensures minimal radiant feedback to the sample as the test progresses and the wall heats up.
Axial uniformity permits the temperature values to be accurately represented in an axisymmetric
pyrolysis model. Given this, the design shown in Fig. 2.5c was selected for its lower overall tem-
perature and axisymmetric uniformity. Actual service temperatures of the chamber are presented
in Section 3.5. The chamber is printed using AlSi10Mg, an alloy of aluminum, because of its
high thermal conductivity, promoting temperature uniformity.
A two-outlet straight-flow rectangular manifold distributes water from the supply and de-
livers it to each wall independently. Copper tubing with an internal diameter of 3.86mm for
the inner wall and 7.04mm for the outer wall transfer the water to each of the chambers. A
Hall-effect liquid flow meter monitors the flow and outputs to a seven-segment display, shown in
Fig. 2.3, allowing for at-a-glance monitoring indicating when the system is active. The output of
both chamber walls is connected to a floor drain using rubber tubing, chosen for its flexibility.
Gas is introduced to the chamber through four bulkhead fittings in the gasification chamber
baseplate. The fittings have a cap installed, and twelve holes are drilled around the threads to
provide gas flow with an angularly uniform distribution. A 25mm-tall plenum space allows the
pressure to homogenize. Above the plenum, a perforated stainless steel sheet holds a 25mm-thick
layer of glass beads with diameters decreasing vertically from 6mm to 3mm whose purpose is
to homogenize the flow further. Gas velocity uniformity is assessed in Section 3.3.
The upper half of the inside of the outer chamber wall is sprayed with a high emissivity
(? = 0.95) paint to prevent reflections of the heater from impinging on the sample surface.
This would add additional heat flux not captured by the external heat flux gauge. A cutout for
20
a 76.2mm ? 30mm quartz-glass window is also present on the front of the chamber to enable
visual inspection of the sample during a test and recording from a video camera. The quartz-glass
can withstand temperatures of up to 1400K which is necessary as it cannot be actively cooled.
A photo of the completed gasification chamber, installed window, and glass beads is shown
in Fig. 2.6
2.4 Sample Holder Assembly
The sample holder in CAPA is designed to hold a maximum 70mm diameter cylindrical
sample. The assembly comprises four components: the balance plate, the support rods, the sam-
ple holder, and the sample cup, all shown in Fig. 2.7. The balance plate is made of AISI 304
stainless steel and supports the assembly on the mass balance via three support rods inserted
into bushings. The support rods are placed such that optical access is allowed from the infrared
camera. Threads on the support rods enable it to screw into the sample holder and set its height
within the inner hole of the gasification chamber. The sample holder is made of Al 6061-T6
aluminum to reduce the amount of dead mass on the balance, increasing the sensing range. A
removable sample cup machined from AISI 316 stainless steel is used for ease of access and
recording photos and pre- and post-test mass without disturbance of the sample itself.
To function as intended, the holder must allow optical access to the back of the sample so
infrared camera recordings can be captured. It must also be mechanically isolated from other
components to ensure an accurate mass reading. One upgraded feature of this assembly over
the previous version of CAPA is the capability for multiple varying-depth sample cups, reducing
the amount of insulation required on the back side of the sample. Superfluous insulation can
21
8 7 6 5 Figure 2.6: Gas4ification chamber 3and sample cup 2 1
D D
B
A
101.60
88.90
80.00
Sample Cup 65.00
C C
Sample Holder
Support Rods
B B
Balance Plate
DETAIL B
SCALE 2 : 1
A Figure S2ECT.IO7N A:-ASCALE 1 : 1 Sample holder assembly (all dimensions in mm)
A NAME DATEDRAWN GTB 3/19/2021 UL FSRI A
CHECKED
TITLE:
UNLESS OTHERWISE SPECIFIED:
DIMENSIONS ARE IN MILLIMETERS
22 TOLERANCES: SAMPLE HOLDER ANGULAR: MACH  1.01 PL  .3   2 PL  .15 ASSY
PROPRIETARY AND CONFIDENTIAL
THE INFORMATION CONTAINED IN THIS DRAWING IS THE SOLE 
PROPERTY OF UNDERWRITERS LABORATORIES INC.  ANY SIZE DWG.  NO.
REPRODUCTION IN PART OR AS A WHOLE WITHOUT THE REV
WRITTEN PERMISSION OF UNDERWRITERS LABORATORIES INC. IS  
PROHIBITED. D A
MATERIAL SCALE: 1:2 WEIGHT: 518.95 SHEET 2 OF 2
SO8LIDWORKS Educational Produ7ct. For Instructional Use Only.6 5 4 3 2 1
228.67
50.00
23.40
contribute to additional mass loss as moisture, and residual binders are vaporized when heated.
The maximum depth of cup allowed in the current configuration is 22.4mm. Two sample cups
are used in the present study with depths of 11.7mm and 15.8mm. To prevent the entrainment
of oxygen-containing air from below the gasification chamber, a lip on the sample cup covers the
gap between the inner wall of the chamber and the sample holder.
2.5 Data Acquisition System
The data acquisition for CAPA captures all the external sensor inputs and facilitates their
sampling via LabVIEW and output to an interface. Principally, two types of inputs exist for this
system: differential analog inputs (primarily thermocouples) and digital serial devices such as the
mass balance or PID controller.
A National Instruments NI-9213 temperature-input module and a cDAQ-9171 chassis han-
dle the differential inputs. This input module can sample 16 differential inputs at a combined rate
of 75 samples per second. LabVIEW easily integrates with the chosen hardware, which makes
sampling from them simple. A large number of inputs also enables the ability to add upgrades or
additional diagnostics at a later time. A USB 2.0 port connects the chassis to a PC running the
LabVIEW software.
The digital portion of the data acquisition system communicates with the serial devices.
These devices use a mix of standards such as RS232 or RS485, thus requiring a flexible interface.
A 4-port serial hub utilizing an FTDI FT4232H UART/MPSSE IC permits using all the stan-
dard serial protocols (RS232/RS485/RS422) and full-speed acquisition from every device on the
system simultaneously. The serial hub has a USB 2.0 interface that is connectable to a standard
23
PC.
The two acquisition components are packaged into a single data acquisition box that forms
a portable system with standard connectors, pictured in Fig. 2.8. A Pelican 1400 Protector case
holds the components due to its compact size, portability, and ruggedness. Miniature K-type and
U-type panel jack connectors are used for the differential inputs and DB9 connectors for the serial
inputs. A panel was cut with a CNC router to fit into the case and use the standard connectors
from the sensors.
2.5.1 Temperature
The acquisition system logs various temperature measurements during tests for diagnostic
information and dynamic quantification of the test environment. The data can later be used to
define the boundary conditions in a comprehensive pyrolysis model for simulating experiments.
All temperature measurements use K-type thermocouples with a sampling rate of 10Hz, although
higher or lower values are possible.
The chamber walls contain three thermocouples with a wire diameter of 0.254mm. Two
are placed on the outer wall of the chamber, one on the inner surface with a high-temperature
epoxy and one on the outer surface with a self-adhesive polyimide. Both are located as close
as possible to the upper edge, nearest to the heater, where the temperature is highest. The third
wall-thermocouple is adhered to the inner surface of the inner wall, adjacent to the sample back
surface, using high-temperature epoxy. These thermocouples verify that the wall does not reach
a temperature that would lead to damage or failure of the system and account for radiant or
convective exchange with the sample.
24
Figure 2.8: Data acquisition box
The interior space of the chamber contains an additional three 0.203mm diameter thermo-
couples. A small diameter thermocouple enables the measurement of gas temperature with min-
imal error due to radiative and wire conduction effects [35]. The thermocouples come through
the bottom of the chamber and up through the glass beads. The sensing junction is placed in the
gas flow at the level of the sample surface. From the front of the apparatus, their locations are
the front, left, and back cardinal directions. The purpose of these thermocouples is to monitor
the gas flow temperature during the test, which tends to increase as the layer of beads heats up.
It should be noted that this is only for diagnostic purposes and should not be used for modeling
due to inaccuracies induced by degradation of the thermocouples over time that would require
constant replacement. As described in Section 3.4, this purpose is fulfilled with an external gas
probe. Fig. 2.9 shows the thermocouple location on/in the chamber.
Two 1.59mm diameter Inconel probe-thermocouples are placed in the water flow to moni-
tor temperatures. Fluctuations in water supply temperature can lead to erroneous sensor readings
as both a transient and absolute effect. One is placed on the inlet side of the heat flux gauge,
25
Figure 2.9: Gasification chamber thermocouple locations
taken as representative of the temperature of the overall water supply. The other is placed on the
outer chamber water-cooling outlet, which provides another means of monitoring cooling system
performance.
2.5.2 Heat Flux
A Schmidt-Boelter total heat flux transducer is used to set the heat flux from the cone
heater. This gauge outputs a differential voltage signal proportional to the incident heat flux. The
particular gauge used is the same as that typically used with a cone-calorimeter, a Medtherm
GTW-10-32-485A with a full-scale range of 100 kWm?2. A drawing of this sensor is shown in
Fig. 2.10. Water-cooling to the gauge is supplied at a flow rate of 1 Lmin?1 and can be controlled
independently and operated concurrently with the water-cooling system for the chamber. The
gauge is provided with a manufacturer NIST-traceable calibration to convert the mV output to a
heat flux value. As a further calibration, measurements were compared against a NIST-calibrated
26
gauge under a cone heater. Heat flux values were within 1.5% of the NIST-calibrated gauge at
50 kWm?2.
The heat flux transducer is mounted to the heater support frame, 40mm under the cone
heater in the storage position, the same spacing as that to the top surface of the sample. This
setup allows continuous measurement of the heat flux immediately up to and after the movement
of the heater for a test. It also eliminates uncertainty caused by the slight movement of the gauge
due to repeated installation and removal. The sensor output is connected to the data acquisition
system and used to determine the cone heater temperature corresponding to the desired heat flux,
as demonstrated in Section 3.1.
2.5.3 Mass Balance
The mass balance used is a Sartorius Cubis? MCE1203S-2S00-R which has a readability
of 1? 10?3 g and a maximum capacity of 1200 g. An RS232 interface is used, which allows con-
nection through the data acquisition box and into the LabVIEW software, where measurements
are typically captured at 20Hz. The sample holder is attached to the balance through a custom
baseplate, described in Section 2.4. The settings shown in Appendix A are used internally to the
balance to achieve the data accuracy and sampling rate. Most notably, the ambient stability is
set to Very Stable which eliminates the preprocessing algorithm that filters measurements over a
period to achieve a more stable output. This setting was disabled to allow complete control of the
processing externally via a more specific and defined methodology.
A balance shield, constructed from 3003 aluminum, encases the balance?s sides, front, and
top. There is a cutout on the top where the baseplate of the sample holder assembly sits. The
27
2-1/4 
NOMINAL 
l 1 ? 
111,50 I ?I d 1Bm?1m?11m111111z z z z z n e z z z z zC! I 
10?1 ? I 
STD, LENGTH 
w'ATER CODLING TUBES (2) 
1/8 ,D, 1/16 ?,D, STAINLESS STEEL SHEATHED I ? Mg? INSULATED LEAD 111.40 TUBE STRAIN RELIEF SPRING 
,75 SUPPORT 111, 25 L. FLEXIBLE LEAD 
1+ I 7 7 7 'i ?'IR'I  I ,,.. I y M\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ / / / / / ./_ 6 (_/_ ,__ 
I I I 
'-6,69 1-,10 
10 1 ---? 36
STD, L?ENGTH ---------- ??:88 ISTD, LENGTH 
.t:mES. 
F1i, gPIuN? GeTw'-27-.312-048:5A M(or eGdTw't-h10-e3r2-m485AG) IsT o.W wo.t-er coolecl toto.l heo.t flux tro.nsclucer which provlcles o. llneo.r EMF output 1cllr0ec-t3ly 2pr-o4po8rt5lonAo.l Schmidt-Boelter heat flux gauge (all dimensions in
to the net o.losorloecl heo.t tro.nsf'er ro.te to the sensing tip, The sto.nclo.rcl 
inoutpute sIs )10 to 20 l'llllivolts o.t the cleslgn full sea.le heo.t flux level, Ea.ch unit Is suppllecl with o.n lncllvlcluo.l co.llloro. tlon tro.ceo.lole to the 
No.tlono.l Institute of' Sto.nclo.rcls o.ncl Technology, The tro.nsclucer Is 
cleslgnecl f'or use In the cone co.lorll'leter f'lo.P1P10.lolllty test o.ppo.ro.tus, 
2, The sensing elel'lent Is o. Schl'llclt-Boelter therl'loplle, The sensor Is coo.tecl 
with optlco.l Iola.ck, 
b3a, cStko.nclio.srcl ofupll esena.le thoeo.ta fclucx ero.snsge tf'hore thpe oGTrw't-s7-3a2-n48d5A aIsl l7o Bwtu/sf't2tshece, shield to be slid into place from the front. The
o.ncl f'or the GTw'-10-32-485A Is 10 Btu/f't2 sec, Either unit l'IO.Y loe usecl to 
120 kw' /1'12 , DIMENSTOIONLERANS ARCE ESIN INCHES 
H
 EAT FLUX TRANSDUCER MEDTHE M
4, Col'lposlte type lea.cl wire Is sto.nclo.rcl, Cooling tuloes, sheo.thecl lea.cl o.ncl FRACTIONS DECIMALS ANGLES,01 1" 
 WATER COOLED I RCORPORA.TJ:ON 
shT eif'leonl dlea.pcl ro. vo.llo.lole In other lengths. Pleo.se specify cleslrecl length 
:1: 1/32 :1: : :1: GTW-7-32-485A 
In Inches, Fleexibvle lnea.tcls Is aTierflocn ucorvrereencl t24s Aaw'Gn sdtro.oncltehcl eprlo.teclx ternal dis-tu-rb-an-ces-fr-om introducing noise into PtOSTh OeFFICmE BOXa 4s12 s
copper, loro.lclecl plo.tecl copper shlelcl, Teflon overo.ll, w'hlte wire- - or GTW-10-32-485A HUNTSYlll.E, ALABAMA 3IIII04 
positive, Bio.ck wlre-nego.tlve, I MATERIAL NOTED I? 2 I DES REV
5, w'o.ter cooling Jo.cket pressure testecl to 150 pslg, ___________J ORIG. DWG 
= 
2 /21 /67 . 
measu ments. FINISH CAD DWG 10 /19 /92 CHK I B I 485 A 6, Unless requestecl otherwise, the co.llloro. tlon will loe reportecl In kw' /1'1 2 , .
DR, 61'-. IAPP. J:11:Jl SHEEI' OF 
2.5.4 Infrared Camera
An infrared camera measures the temperature of the back of the sample through non-contact
means. A FLIR E95 camera is chosen with a 464 ? 348 resolution, 30Hz frame rate, and ac-
curacy of ? 2% of reading. This camera uses a microbolometer detector with a spectral range
of 7.5 ?m to 14 ?m. The E95 is a direct and substantial upgrade over the FLIR E40 used on
the previous version of CAPA due to its higher resolution and lower noise-equivalent tempera-
ture difference (NETD). The camera is mounted rigidly to a 25.4mm-diameter, 101.6mm-tall
pedestal pillar post, secured to the optical breadboard. A 50.8mm diameter gold-coated mirror
(? ? 0.99) is mounted to the underside of the gasification chamber frame to allow for optical
access. The settings of the infrared camera are adjusted to account for this transmission loss. A
diagram of the optical path of the infrared camera with the positioning of other components is
28
shown in Fig. 2.11.
Recordings are captured via the FLIR Research Studio software at 30Hz and saved as a
*.seq file, which allows for manipulation of various measurement parameters after-the-fact such
as the emissivity or ambient environment. Connection is made directly from the USB 2.0 output
of the camera into the PC.
2.5.5 Gas Flow
An Alicat Scientific MCR mass flow controller controls the gas flow into the chamber. It
is capable of a full-scale output of 250 SLPM and mass flow accuracy of ?0.8% of reading and
?0.2% of full scale. Communication is done over RS232 using the MODBUS RTU protocol.
The controller is connected to LabVIEW to control the set point. Data logging from the mass
flow controller is not performed here, although the capability does exist.
The gas flow output from the controller enters into a four-outlet straight-flow rectangular
manifold to distribute the gas to the four inlets on the bottom of the gasification chamber base-
plate. Copper tubing with an internal diameter of 7.04mm connects the manifold to the bulkhead
fittings on the bottom of the chamber. Typically, pure nitrogen is used at a flow rate of 185 SLPM
to ensure anaerobic decomposition of the sample; however, any gas or combination of gases is
possible with minor modifications.
2.5.6 Pyrometer
An additional measurement capability of this version of CAPA is the ability to capture the
top surface temperature of a sample during a test. This is accomplished through the inclusion of
29
Figure 2.11: Diagram of optical path of infrared camera and relative positioning to components
a pyrometer above the sample. The pyrometer used is a Raytek RAYMI310LTF with a sensing
range of 273K to 1273K with an accuracy of ?1% of reading and spectral sensing region of 8 ?m
to 14 ?m. An air purge jacket is installed, which allows an external gas supply to be blown across
the sensor surface, protecting it from contamination. LabVIEW communicates and collects data
from the pyrometer using an RS485 interface.
The pyrometer is installed on a movable arm 50mm above the top of the heater, seen in
Fig. 2.1, which allows its use or removal. Careful calibration is necessary for its use due to
reflections from the cone heater and potentially dynamic emissive properties of the sample. This
sensor has not been used in the current study.
30
2.5.7 Video
An external video camera records significant observations and evolution of the sample
surface during a test. For intumescing samples, this is used to calculate the incident heat flux
on the top surface of the sample as a function of height and orientation. The camera used is a
FLIR Blackfly? S BFS-U3-120S4C-CS utilizing a Sony IMX226 sensor capable of recording at
a resolution of 4000 ? 3000 pixels and a frame rate of 31 FPS. The overall dimensions of the
camera are 29mm? 29mm? 30mm which make it ideal for use as a high-performance camera
in limited spaces. A 12mm fixed focal length lens with adjustable f/1.8 - f/16 aperture gives a
minimum working distance of 100mm.
The camera is mounted to the underside of the heater mount, so it slides with the heater.
The focal point is set at the center of the sample surface. The chosen configuration is oriented
so that the camera observes the top surface; however, a perpendicular view of the sample is
possible, which provides a more accurate measure of surface growth during the test. Ceramic
fiber insulation and reflective aluminum tape surround the camera?s lens to protect it from the
impinging heat flux from the heater. A 40mm fan cools the camera body itself to keep it within
operating temperatures. A USB 3.0 interface connects the camera to the PC, and LabVIEW
captures the video at a frame rate of 15 FPS.
31
Chapter 3: Characterization
3.1 Heater Temperature to Radiant Flux
The radiant heat flux from a cone heater is controlled by proxy through the coil?s temper-
ature, which is measured by three thermocouples in direct contact with it. The PID controller
mentioned in Section 2.2 controls the coil?s temperature by cycling power to the heater. It is,
therefore, necessary to relate this temperature to the heat flux produced by the heater at a fixed
location. This value should stay constant for a given height, although changes are expected as
the heater coil and thermocouple contact with it shift over time. Periodic calibration should be
performed to update this relation and determine when significant contact is lost, and the operator
should replace the coil.
As mentioned in Section 2.5.2, the heat flux gauge is located along the central axis of the
heater, 40mm below its bottom surface. This is the same position as the top of the sample when
the heater is over the gasification chamber. The heat flux values measured by this gauge represent
the radiant heat flux from the heater incident on the center of the sample surface during a test. It
is crucial to control this value as accurately as possible, being the most significant source of heat
flux to the sample, both for the repeatability of tests and accurately modeling them in the future.
A manual process generates a calibration table to determine the heater temperature cor-
responding to the various heat flux settings. The temperature controller setpoint is manually
32
adjusted until heat flux values of 0.5, 1, 3, 10, 20, 25, 30, 40, 50, 60, 65, 70, and 75 kWm?2 are
output from the gauge. Once the heat flux value is reached, the heater temperature stabilizes for
an hour until the value is recorded to ensure no short-term drift. Fig. 3.1 shows the generated data
and curve fit.
The trend was fitted with a third-degree polynomial as shown in Eq. 3.1:
q??(r/R = 0, h/H = 0, T 3heater) = a1Theater + a T
2
2 heater + a3Theater + a4 (3.1)
with coefficients of a1 = 9.757? 10?8, a2 = ?5.724? 10?5, a3 = 0.016 59, and a4 = ?2.424.
Although this provides an initial approximation for the heater temperature, an actual re-
finement is required to obtain an accurate value before a test. Long-term variations can occur due
to the heater coil shifting contact with the thermocouple. This methodology does not account for
additional uncertainty due to calibration of the heat flux gauge, environmental conditions, and
water temperature.
An optimization algorithm is implemented in the LabVIEW program to facilitate this pro-
cess. This algorithm first sets the temperature to the value determined by the Eq. 3.1. After
that temperature value is reached and the heater stabilizes, the heat flux value is recorded. From
there, the program iterates this process using the secant method of root-finding to find the appro-
priate heater temperature within a 0.1 kWm?2 heat flux tolerance. The operator can make further
manual adjustments if a tighter tolerance is desired.
33
80
Fit
70
60
50
40
30
20
10
0
300 400 500 600 700 800 900 1000 1100
Theater [K]
Figure 3.1: Heater temperature versus total heat flux at 40mm
3.2 Heat Flux Uniformity
Understanding the distribution of radiant heat flux incident on the sample can be crucial to
understanding the behavior of certain types of materials. In particular, intumescent materials can
expand and change relative orientation during a test which alters its position with respect to the
heater and thus the incident radiant heat flux. For other materials, it can also be beneficial to repre-
sent them axisymmetrically by accounting for the radial component of the heat flux distribution.
Swann et al. conducted extensive characterization work on the previous version of CAPA [6],
with which this apparatus shares its geometrical design. As such, a smaller characterization work
was carried out here to verify previous data and bolster its reuse.
The heat flux distribution was analyzed in terms of its radial dependence at the height
of the virgin sample surface (H = 40mm). Characterization was performed with a Medtherm
34
q??(r/R = 0, h/H = 0, T ) [kW m?2heater ]
Schmidt-Boelter total heat flux gauge (Model# 64-5T-5R(SW)-20898) with full scale output at
50 kWm?2. The sample holder and mass balance were removed to allow the gauge to be inserted
into the center hole. The water cooling in the chamber was turned on, and the gas flow was turned
off to reduce the convective component from the total heat flux gauge measurements. The heat
flux was measured at various positions relative to the sample?s axis. These radial distances, r,
were normalized to the maximum radius of the sample, R = 35mm. Distances between r = 0
and r = 0.8 were used, limited by the physical size of the heat flux gauge.
Measurements were taken in four azimuthal angles with respect to the front of the appara-
tus: 0?, 90?, 180?, and 270?. Multiple tests were conducted at three different heat flux settings:
12 tests at 35 kWm?2, 27 tests at 50 kWm?2, and 24 tests at 65 kWm?2. No differences were
observed in the distribution pattern between the three heat fluxes, and they are presented here
together. Every test exposure lasted approximately 60 seconds, and heat flux data was acquired
at 20Hz. A central axis test was conducted immediately before and after each offset position test.
To correct heater drift over time, a third-degree polynomial was used to model the time-
resolved center heat flux values from the bookended tests. The offset heat flux values were
divided by the predicted central heat flux at the corresponding time to produce a normalized
value. The average is taken over these normalized heat flux values and related to the normalized
radial distance for each test.
The radial dependence of the normalized heat fluxes is shown in Fig. 3.2a, and the angular
and radial dependence of the heat fluxes is shown in Fig. 3.2b. The radiant heat flux over the
sample radius drops to 96% of the central heat flux at the edge of the sample, compared to 95%
found by Swann et al. [6]. Given the typical dimensional tolerances of the heating coil winding,
this is considered a reasonable variation. A significant angular variation was not found.
35
1.01
This study
Swann et al.
1.00
0.99
0.98
0.97
0.96
0.95
0.94
0.0 0.2 0.4 0.6 0.8 1.0
r/R
(a)
90?
1.00
135? 45?
0.99
1.0
0.8
0.6
0.98 0.4
0.2
180? 0?
0.97
0.96
225? 315?
0.95
270?
(b)
Figure 3.2: (a) Radial and (b) polar radiant heat flux distribution at h/H = 0, compared to that
found by Swann et al. [6]
36
q??(r/R, h/H = 0)/q??(r/R = 0, h/H = 0) q??(r/R, h/H = 0)/q??(r/R = 0, h/H = 0)
The radial dependence on heat flux was captured with Eq. 3.2:
q (r/R, h/H = 0) ( r )2 ( r )
= b1 + b2 + 1 (3.2)
q (r/R = 0, h/H = 0) R R
where b1 = ?0.03692 and b2 = ?0.00259.
3.3 Gas Flow Uniformity
The uniformity of the gas flow entering the chamber verifies that the CFD models un-
derpinning the convection coefficient calculation are accurate. The gas flow velocity should be
approximately equal at any angular position within the chamber. Gas velocities in the chamber
were measured with a Kanomax 6036-0E hot wire anemometer with an accuracy of ?3% of the
reading or 0.015m s?1, whichever is greater. The sensing end of the probe was placed in the
center of the gas-flow channel and at the height of the virgin sample surface. Measurements were
taken at eight equally-spaced angular positions. The heater was placed over the chamber; how-
ever, it was not turned on. Nitrogen gas flow was turned on and set to a flow rate of 185 SLPM.
The velocities and corresponding measurement locations are shown in Fig. 3.3.
The average measured gas velocity is 0.37m s?1 with standard deviation of 0.07m s?1.
This standard deviation is less than 20% of the mean value, indicating generally low variation
angularly. Considerable variation was seen across the width of the channel, although not explic-
itly measured due to geometrical constraints of the anemometer. This observation is bolstered by
the fact that a completely uniform distribution of this gas flow through the channel should give
an average gas velocity of 0.29m s?1. Given that these measurements were conducted approx-
imately in the center of the channel, a significant boundary layer is expected near the chamber
37
Figure 3.3: Gas flow velocities with location, units in [ms?1]
walls, leading to radial deviation.
3.4 Gas Flow Temperatures
The free stream temperature of the gas flow must be quantified to account for the convective
heat flux on the top surface of a sample in CAPA. This is expected to be a function of both time,
as the glass beads increase in temperature, and the heat flux. The free stream temperature is
combined with the convective heat flux coefficient determined by Swann et al. [6] to calculate the
convective heat flux to the top surface. The free stream temperature for the bottom surface is not
measured explicitly and is assumed to be equal to the inner wall temperature.
Two external gas temperature probes were manufactured to capture accurate measurements
in the highly radiative environment of the chamber. The probe is composed of 0.127mm diam-
38
eter type-K thermocouple wires inserted in a rigid ceramic insulator with an outside diameter of
1.57mm. The thermocouple junction protrudes approximately 8mm from the end of the insula-
tor to ensure that the gas temperature is measured without conduction errors from the insulator
itself. A photo of the constructed gas temperature probe is shown in Fig. 3.4.
Measurements were conducted on the version of CAPA presented here and the previous
version of CAPA II so that a direct comparison of the two versions could be made and differences
evaluated. Each probe was placed at eight uniformly distributed angular locations, at the height
of the sample surface and in the center of the channel. The heater was placed over the chamber
and turned to a temperature corresponding to heat fluxes of 25 kWm?2 and 60 kWm?2. The gas
flow was set to 185 SLPM, and water cooling in the chamber was turned on. The water flow rate
in this version of CAPA was 4.7 Lmin?1 and CAPA II was lower at 3.9 Lmin?1. The sample cup
was filled with ceramic fiber insulation that provided an inert surface representative of a sample
being inserted into the chamber.
The results of the characterization performed in both apparatuses and the correlation devel-
oped by Swann [5] are shown in Fig. 3.5. The steady-state gas temperatures for both are within
the experimental uncertainty; however, temperatures measured for CAPA II have a small amount
of residual rise at the end of the 600 s period. The significant difference between the two lies in
the rise time.
To provide data for modeling and calculating the rise times, the gas temperatures of the
present version of CAPA and those of CAPA II were fit to Eq. 3.3:
Tgas, upper = T1e
T2t + T T4t3e + 290 (3.3)
39
Figure 3.4: Gas temperature probe construction
The coefficients determined for the present version of CAPA are shown in Table 3.1. The coeffi-
cients found for CAPA II are not presented here.
The rise time was calculated as the time it takes for the gas temperature to reach 90% of its
value at 600 s. These are presented in Table 3.2. The present version of CAPA reaches the steady-
state temperature in approximately half the time as CAPA II. This is likely due to the increased
height of the beads layer in the present study, which causes them to be exposed to higher heat
flux and therefore increase in temperature more quickly.
Compared to the data and correlation developed by Swann [5], the temperatures observed
in both versions of CAPA were notably higher and the rise times lower. It is unclear where this
Table 3.1: Gas temperature correlation coefficients
Heat Flux
[ T [K] T [Ks
?1] T [K] T [Ks?1]
kWm?2] 1 2 3 4
25 50.6 2.18? 10?5 ?4.039? 101 ?3.46? 10?2
60 110.7 3.43? 10?5 ?8.54? 101 ?3.25? 10?2
40
350
340
330
320
310
This Study, ?2?
CAPA II, ?2?
Swann (2019)
300
0 100 200 300 400 500 600
Elapsed Time [s]
(a) 25 kWm?2
420
400
380
360
340
320 This Study, ?2?
CAPA II, ?2?
Swann (2019)
300
0 100 200 300 400 500 600
Elapsed Time [s]
(b) 60 kWm?2
Figure 3.5: Time-resolved gas temperature measurements in the present version of CAPA, CAPA
II, and that recorded by Swann [5] at (a) 25 kWm?2 and (b) 60 kWm?2
41
Tgas, upper [K] Tgas, upper [K]
Table 3.2: Gas temperature rise time
Heat Flux trise [s]
[kWm?2] This Study CAPA II
25 70.6 144.6
60 78.1 159.2
discrepancy arises from, but it should be noted that Swann [5] utilized thermocouples with a di-
ameter of 0.076mm, smaller than those used in the present study. This could indicate significant
radiative effects on the temperature measurements, and further investigation is warranted.
3.5 Chamber Temperatures
Transient heating of the chamber walls can contribute additional heat flux to the sample in
the form of reradiation. As the heater is moved over the chamber, the walls will heat up from
the extra heat flux and eventually reach a steady state that is a function of the water-cooling
system?s effectiveness. It is therefore essential to characterize this temperature rise so that it can
be accounted for in any modeling of the CAPA test. It is expected that as the heat flux set point
of the heater increases, so will this steady state temperature. Accounting for the rise time is also
essential, particularly when trying to model ignition at the beginning of the test.
The thermocouples installed on the chamber walls, as described in Section 2.5.1, were
utilized to measure this effect. Tests were conducted at a 25 kWm?2 and 60 kWm?2 heater set
point with both the water-cooling and gas flow turned on to its standard setpoint of 185 SLPM.
The water flow rates were the same as is the gas temperature section, 4.7 Lmin?1 in this version
of CAPA and 3.9 Lmin?1 in CAPA II. A ceramic fiber board sample surrogate was placed in
the sample holder during the test. These tests were also repeated on CAPA II as a point of
42
comparison. As the temperature of the ambient water supplies varied slightly, the temperatures
were corrected to the steady-state temperature of the walls prior to moving the heater. Four tests
were repeated for CAPA used in the present study and six for CAPA II. The duration of the tests
was 10min. The results are shown in Fig. 3.6.
The chamber temperatures collected for the present version of CAPA are markedly lower
than those found for CAPA II, indicating increased water-cooling capacity. The outer wall, which
is more directly exposed to the heat flux from the heater, experiences an almost instantaneous rise
in temperature and then becomes a steady state at a temperature value proportional to the inci-
dent heat flux. The steady-state temperature values for the present version of CAPA are also
significantly lower than the previous version, experiencing a 70-80% reduction. The tempera-
tures experience a gradual but insignificant rise for the inner wall throughout the test. This was
primarily due to heating the sample holder itself, radiating heat to the inner wall thermocouple,
and not heating the wall from external sources. The inner wall of CAPA II experienced a much
more significant and gradual rise indicating more substantial transient effects.
From a modeling perspective, these results present a significant upgrade and confidence
in the present version of CAPA boundary conditions compared to CAPA II. Firstly, the reduced
temperature means that additional heat flux through reradiation is minimal, and uncertainty from
it is reduced. Secondly, the near-instantaneous rise and steady-state of the outer wall temperature
mean it can be represented with a constant value instead of explicitly modeling the temperature
rise time. Given this data, the integral water-cooling system of present represents an improvement
over previous versions.
43
Outer Wall
60
40 This Study, ?2?
CAPA II, ?2?
20
0
10
Inner Wall
8
6
4
2
0
0 100 200 300 400 500 600
Elapsed Time [s]
(a) 25 kWm?2
150
Outer Wall
100
This Study, ?2?
CAPA II, ?2?
50
0
Inner Wall
20
10
0
0 100 200 300 400 500 600
Elapsed Time [s]
(b) 60 kWm?2
Figure 3.6: Time-resolved chamber temperature measurements in the present version of CAPA
and CAPA II at (a) 25 kWm?2 and (b) 60 kWm?2
44
Tinner ? T [K] Touter ? Tinitial [K] Tinner ? Tinitial [K] Touter ? Tinitial [K]initial
3.6 Oxygen Concentration
Verifying that oxygen concentration within the chamber is below an acceptable value is
crucial to ensuring the anaerobic decomposition of the samples. Depending on the nature of the
material, sensitivity to certain levels of oxygen can vary; thus, it is desirable for this value to be
as minimal as possible. A RAE Systems QRAE 3 gas monitor was used to sense oxygen levels
in the chamber during operation, measuring levels from 0 to 30% with a resolution of 0.1%. A
ceramic fiber board sample surrogate was inserted into the sample holder with a 6.35mm hole
drilled in the center. A copper tube was inserted into the hole and adjusted until the end was
75mm and 20mm above the sample surface. A rubber tube was used to connect the bottom of
the copper tube to the gas monitor, where an internal pump can draw air from the chamber. The
heater was set to a heat flux of 50 kWm?2 and the gas flow to 185 SLPM. Water cooling was
turned on during the experiments. A photo of the experimental setup is shown in Fig. 3.7.
After one minute of purging, the gas concentration dropped to 0.0%, indicating that the
environment was completely eliminated of oxygen within the measurement capability of the sen-
sor. This occurred at all tested height positions in the center of the sample. Given these values,
complete anaerobic decomposition of samples is ensured with the given gas flow rate and heater
conditions. Although not explicitly tested, no significant variation is expected within the typical
operating range of the heater.
45
Figure 3.7: Oxygen concentration measuring setup
3.7 Sample Thermal Exposure
As a means of verifying the entire thermal environment, a copper plate test was conducted.
A 4mm-thick disk made of oxygen-free copper and painted on both sides with high emissivity
paint was placed in the sample holder as a typical sample would be. In this configuration, the
plate will absorb and emit all the heat fluxes associated with the apparatus. The thickness and high
thermal conductivity ensure that the plate can be treated as a lumped mass. Two thermocouples
inserted into the sides of the disk measure its temperature.
Two sets of tests were conducted here to characterize the apparatus. The first goal was to
simulate the response of the plate to the determined characterization of the chamber and with the
convection coefficients determined by Swann et al. [6]. The plate was placed in the chamber and
the radiant heat flux was set to 25 kWm?2. Gas flow was nitrogen at a flow rate of 185 SLPM.
The duration of exposure was 600 s. Simulations were conducted with a finite-difference scheme
on Eq. 3.4:
46
dT
mcp = q?heater + q?rad, top + q?rad, bottom + q?conv, top + q?conv, bottom (3.4)
dt
where,
q? ??heater = A?q?h(eater, ) (3.5a)
q? = A?? T 4rad, top ( outer ? T
4
plate) , (3.5b)
q? 4 4rad, bottom = A?? Tinner ? Tplate , (3.5c)
q?conv, top = Ahc, top (Tgas, upper ? Tplate) , (3.5d)
q?conv, bottom = Ahc, bottom (Tinner ? Tplate) (3.5e)
Infrared measurements on the bottom of the plate were captured concurrently to determine
the actual emissivity of the Medtherm Optical Black Coating used. The emissivity setting in the
camera was manually adjusted until the temperature recorded by the camera matched that of the
two thermocouples in the plate. This resulted in an emissivity value of ? = 0.94 with the fit
shown in Fig. 3.8. When the paint begins to decompose, these two temperatures deviate slightly
above 600K.
Appropriate properties of the copper were used. The top gas temperature, Tgas, upper,was
modeled using Eq. 3.3 and the coefficients in Table 3.1. The rest of the temperatures were input
from those dynamically captured during the copper plate test: Touter is the temperature of the
outer chamber wall and Tinner is the temperature of the inner wall measured with the embedded
thermocouples. The results of the simulation and the measured values are shown in Fig. 3.8.
The simulation does an excellent job at predicting the temperature of the plate, indicating
47
700
Thermocouple Measurement
Infrared Measurement
650
Simulated
600
550
500
450
400
350
300
0 100 200 300 400 500 600
Elapsed Time [s]
Figure 3.8: Copper plate temperature and simulation under 25 kWm?2 radiant exposure
that the previously defined parameters and relations are effective at capturing the thermal environ-
ment. Slight deviation above 550K is likely due to the onset of degradation of the high emissivity
paint, which would lower the effective radiation reaching the sample and thus its temperature as
well.
The second portion of this experiment was to compare the thermal environment of the
present version of CAPA against CAPA II. The same testing procedure was followed to obtain
this data. Presented in Fig. 3.9, the iteration of CAPA presented here displays a very similar but
decreased temperature profile. This difference could be due to minor deviations in the set heat
fluxes or more significant emissive losses to the chamber walls. Given this minor discrepancy, the
overall thermal environment of CAPA is confirmed, and results obtained from the two apparatuses
should be directly comparable
48
Tcopper [K]
700
This Study
CAPA II
650
600
550
500
450
400
350
300
0 100 200 300 400 500 600
Elapsed Time [s]
Figure 3.9: Copper plate thermocouple temperature in present study and CAPA II under
25 kWm?2 radiant exposure
49
Tcopper [K]
Chapter 4: Experimental Comparison
An experimental comparison of two materials previously tested in CAPA was conducted
to evaluate the present version of CAPA compared to CAPA II. Any differences between the
two versions can be explained by the differences in characterization discussed in Chap. 3 or
the material itself. All tests were performed under the same experimental conditions as that in
previous studies to allow for a direct comparison.
4.1 Materials
The two materials chosen for this study are poly(methyl methacrylate) (PMMA) and ori-
ented strand board (OSB). These materials have previously been studied in CAPA and present
both a charring and non-charring sample, which allows a range of behaviors to be explored [31,
36]. Both materials have also undergone complete characterization and implementation in a com-
prehensive pyrolysis model for validation which bolsters the collected data.
The behavior of PMMA in CAPA was studied by Fiola et al. [31] on both the clear extruded
and opaque cast forms; however, only the cast form is replicated here. Fiola et al. used CAPA
tests to extract the thermal conductivity of the decomposing polymer and to validate the suite of
other material characterization data they collected.
Gong et al. [36] looked at the behavior of OSB in CAPA. This study was a similar method-
50
ology to that employed by Fiola et al. [31]. Tests were conducted as a means of obtaining thermal
conductivity and validating additional parameters determined at the milligram scale.
4.1.1 Poly(methyl methacrylate)
The PMMA tested here was a 9.53mm-thick cell-cast acrylic manufactured by Amari Plas-
tics under the Acrycast product line. A black, opaque version was selected as it is commonly
used for standardized testing in the cone calorimeter and that used by Fiola et al. [31]. Sam-
ples were machined down to 70mm-diameter cylinders with a thickness of (5.79 ? 0.01)mm,
the same as that used by Fiola [31]. One side of the disks was lightly sanded with 120 grit
sandpaper to remove and roughen the glossy surface. The mass of the prepared samples was
(26.17? 0.05) g with a density of (1174.6? 2.2) kgm?3 compared to the density found by Fiola
et al. of (1210?30) kgm?3. Samples were placed in a desiccator with <10% RH for a minimum
of 48 h before any measurements.
4.1.2 Oriented Strand Board
The OSB tested was Georgia-Pacific Blue Ribbon, nominal 11.1mm-thick, structural pan-
els from APA Mill 451 in Brookneal, VA. Samples were cut into 70mm-diameter cylinders. The
thickness was not altered, and the samples prepared had an actual thickness of (10.87?0.02)mm
compared to a thickness of (10.8? 0.1)mm measured by Gong et al. [36]. The mass of the pre-
pared samples was (26.05 ? 0.87) g with a density of (622.5 ? 20.7) kgm?3 compared to the
density measured by Gong et al. of (664? 56) kgm?3. This large density variation is consistent
with the variability encountered by Gong et al. [36]. Samples were placed in a desiccator with
51
<10% RH for a minimum of 96 h prior to any measurements to minimize the moisture content.
The desiccator used by Gong et al. more typically produced relative humidity values of 20-30%.
This time was obtained by the sample experiencing a less than 1% change in mass over a 24 h
period.
4.2 Testing Procedure
4.2.1 Sample Preparation
Several steps are required to prepare the samples for testing in CAPA. First, copper foil is
adhered to the bottom of the samples. The copper provides a consistent substrate on the bottom
of samples to measure temperature and isolates the samples from the effects of oxygen on the
bottom surface. Using a thin copper layer minimizes any thermal gradient between the bottom
of the sample and the bottom of the foil. Copper is also the same material used for the paint
emissivity characterization in Section 3.7 so substrate effects can be neglected. Circles with
80mm-diameter were cut out of 0.0254mm-thick copper foil and their mass measured to be
(1.20? 0.01) g.
J-B Weld High Heat epoxy was used to adhere the copper foil to the samples. J-B Weld
High Heat is a two-part consumer bisphenol-A-(epichlorohydrin) epoxy that has a temperature
rating of up to 561K. The two parts of the epoxy are mixed in a 1:1 ratio and light heat is
applied to eliminate any bubbles introduced by air entrainment during mixing. A thin layer
is applied to the bottom of the samples using a putty knife to provide a smooth surface. For
the PMMA samples, the epoxy is applied on the sanded side of the sample, and for OSB, it is
applied to the smooth, finished side. These faces are chosen to maximize adhesion and reduce the
52
amount of epoxy required. The samples are placed epoxy-side-down into the center of the copper
foil circles and clamped together for a minimum of 24 h until the epoxy has completely cured.
Before remeasuring their mass, samples are then placed back in the desiccator for the minimum
time periods specified in Section 4.1. The resulting mass of epoxy on the samples was calculated
to be (0.56? 0.17) g with no significant difference between PMMA and OSB.
The exposed side of the copper foil was painted with Medtherm Optical Black Coating
that has a stated absorptance of 0.95 over the spectral range 0.3 ?m to 15 ?m. The effective
emissivity for thermal measurements was found to be 0.94 in Section 3.7 and used for the IR
camera measurement settings. The paint was allowed to dry for 30min at room temperature and
then cured for 1 h at 370K. The painted side of the copper foil was wrapped around 75mm-
diameter 1100 aluminum alloy expanded metal mesh with a thickness of 0.51mm. This mesh
had an average mass of (2.82 ? 0.02) g and was painted with the same optical paint as the foil.
The mesh was placed such that no contact was made with the inside of the cup.
Insulation rings of 80mm outer diameter and 65 and 70mm inner diameter were cut from
Kaowool M board (formerly known as Kaowool PM). The 65mm rings were placed on the bot-
tom of the sample cup, and the thickness was adjusted until the top of the sample was flush with
the top of the sample cup. The 11.7mm-deep sample cup was used for the PMMA and 15.8mm-
deep one for the OSB to allow for sufficient insulation on the backside. The 70mm rings are fitted
around the outside of the sample on the top side. The insulation is baked separately at 825K for
1 h to burn out the organic binders and moisture in the insulation, which could contribute to mass
loss during the test. Afterward, they are placed in the desiccator to control the moisture level. A
total of seven samples and associated components were prepared for both the PMMA and OSB.
The typical sample assembly is shown in Fig. 4.1.
53
Figure 4.1: Typical CAPA sample assembly
4.2.2 Test Procedure
Experiments were conducted under the same conditions as the original studies. For PMMA,
three tests were performed at a radiant heat flux of 25 kWm?2 and four tests at 60 kWm?2. Three
OSB samples were tested at 35 kWm?2 and four at 65 kWm?2. All tests were conducted with a
nitrogen flow rate of 185 SLPM.
Before each series of tests at a given heat flux, the spacing between the heat flux gauge
and heater was verified, and the initial temperature was set. Water cooling to the gauge and the
chamber was turned on at this point. The heater was allowed to stabilize at the correct heat flux
for an hour before any tests to ensure no drift.
The sample assembly and associated insulation were placed in the appropriate sample cup,
and the total mass was measured. A photo was taken before the test of the virgin sample assembly;
an example is shown in Fig. 4.2. The mass balance was tared, and the sample cup was placed in
the holder. The gas flow was set to the appropriate flow rate and allowed to stabilize and evacuate
air out of the chamber for one minute. At this point, data recording from all the devices was
turned on and continued in this idle state for one minute. After this period, the heater was swiftly
54
moved over the chamber and left for the durations given in Table 4.1. After that time, the heater
was moved back into the idle position and the recording was continued for an additional minute
to capture any residual effects.
After the test, the sample cup was allowed to cool, and post-test pictures and mass were
taken. Any residual sample char mass was measured separately.
4.3 Results
The time-derivative of the raw mass data was taken to compute the mass loss rate of burn-
ing materials. Given the noisy nature of the data induced by the sensitivity of the balance and
environment, a smoothing filter was first applied to reduce the amplification of high-frequency
noise components following differentiation [37]. This was accomplished with a Savitsky-Golay
filter coupled with a Gaussian kernel implemented in the PyNumDiff Python package [38]. The
PyNumDiff package manipulates this filter using a multi-objective optimization framework that
chooses appropriate parameters through minimization of a loss function and selection of a sin-
gle hyper-parameter [39]. This optimization scheme is run for every individual test and thus
optimized filter parameters will vary slightly between them based on the local noise charac-
teristics but remain similar for those conducted under the same experimental conditions. The
Table 4.1: Test durations
Heat Flux Test Duration
Material [kWm?2] [s]
25 1680
PMMA
60 390
35 2040
OSB
65 1020
55
(a) PMMA (b) OSB
Figure 4.2: Prepared (a) PMMA and (b) OSB samples
hyper-parameter, ?, represents a reduced filter parameter set that ultimately corresponds to the
degree of smoothness of the resulting derivative. This hyper-parameter is nearly universal across
differentiation methods and thus allows direct comparison of them if desired in the future. An
optimized hyper-parameter can be chosen from Eq. 4.1 using a chosen cutoff frequency, f , and
the sampling rate of the data, dt [39]:
? = e?1.6 ln(f)?0.71 ln(dt)?5.1 (4.1)
Cutoff frequencies were chosen through analysis of the individual power spectrums and
manual tuning. The value of the cutoff frequency corresponds to the frequency at which the power
begins to decrease at the noise of the spectra increases, determined through visual observation of
the power spectra [39]. The selected cutoff frequencies are presented in Table 4.2.
The IR temperature value at a particular time was calculated as the 90th percentile of twenty
randomly chosen points distributed across the radius of the sample from the center to the edge
56
Table 4.2: Filtering cutoff frequencies
Heat Flux Cutoff Frequency
Material [kWm?2] [Hz]
25 0.0010
PMMA
60 0.0050
35 0.0005
OSB
65 0.0010
to partially account for separation between the foil and the sample, which only decreases the
temperature. Points were carefully selected between the gaps in the mesh. The mean was taken
across replicate tests for the mass loss rate and back surface temperature data and presented in
the following figures as solid lines. The standard deviation was also taken and plotted ?2?, given
by the shaded region.
4.3.1 Poly(methyl methacrylate)
The mass loss rate and bottom surface temperature for PMMA at 25 kWm?2 are shown in
Fig. 4.5. Also shown is the corresponding data collected by Fiola et al. [31]. Several discrepancies
between the two datasets can be described by observed sample behavior. During the tests, the
sample began slightly rising in the center and moving towards the heater. To investigate this
behavior, one test was prematurely stopped once the sample began to bulge. After removing the
sample assembly from the holder, it was found that a gas pocket had formed under the PMMA and
caused a separation from the foil. Fig. 4.3 shows this sample cut in half and cut ends exposed. It is
unclear whether this pocket formed as a result of a buildup of decomposition products underneath
the sample or a manifestation of internal stresses within the PMMA. Multiple attempts were
made to eliminate this behavior, including the use of multiple different types of epoxy; however,
57
no replicable results could be captured. It is recommended that an additional study should be
conducted to investigate the efficacy of different adhesives for use in different CAPA sample
materials.
A post-test photo of the PMMA sample after 25 kWm?2 exposure is shown in Fig. 4.4.
Virtually all of the sample was decomposed at both heat fluxes with only a small residual char
layer from the epoxy.
Mass loss rate data is nearly identical up to 500 s where the data collected in this study
becomes slightly higher and begins an earlier descent. In addition, the back surface tempera-
ture starts to deviate significantly below at approximately 250 s and then later increases to nearly
match the data collected by Fiola et al. [31]. These can both be explained through the swelling
and separation behavior described above. When the pocket forms below the PMMA, the bot-
tom temperature drops because the foil is no longer in contact with the bottom of the sample.
Additionally, the top of the sample extends slightly towards the heater, which increases the heat
flux to the sample and reduces heat loss from the now decoupled back surface. The increase in
net energy to the sample leads to an increase in mass loss rate as the temperature increases and
reactions progress. Later during the test, this pocket collapses, and the PMMA regains contact
Figure 4.3: Pocket formation and bulging behavior of PMMA at 25 kWm?2. Picture taken with
sample cut in half and cut ends exposed.
58
Figure 4.4: Post test photograph of PMMA at 25 kWm?2
with the foil, which is exhibited in the sudden rise in back surface temperature and reduced mass
loss rate. A large uncertainty is associated with this process due to variation in the mechanics and
timing of the pocket formation. More representative temperatures could potentially be acquired
by careful selection of temperature sampling points however it was not performed here as the
exact locations used in prior study are not available and instead a random distribution of radial
distances was used.
The mass loss rate and bottom surface temperature for PMMA at 60 kWm?2 are shown in
Fig. 4.6 as well as the data collected by Fiola et al. [31]. Although some visual pocket formation
was still observed in the 60 kWm?2 tests, it was not as significant and occurred late in the test
after most of the mass loss had already been completed. The data shows a nearly exact replicate
of that collected by Fiola et al. [31].
59
0.010
This Study
Fiola et al.
0.008
0.006
0.004
0.002
0.000
0 200 400 600 800 1000 1200 1400 1600
Elapsed Time [s]
(a)
700
This Study
Fiola et al.
650
600
550
500
450
400
350
300
0 200 400 600 800 1000 1200 1400 1600
Elapsed Time [s]
(b)
Figure 4.5: PMMA (a) mass loss rate and (b) bottom temperature at 25 kWm?2 radiant exposure
60
T [K] MLR [kg m?2 ?1bottom s ]
0.040
This Study
Fiola et al.
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
0 50 100 150 200 250 300 350
Elapsed Time [s]
(a)
900
This Study
Fiola et al.
800
700
600
500
400
300
0 50 100 150 200 250 300
Elapsed Time [s]
(b)
Figure 4.6: PMMA (a) mass loss rate and (b) bottom temperature at 60 kWm?2 radiant exposure
61
T [K] MLR [kg m?2 ?1bottom s ]
4.3.2 Oriented Strand Board
A post-test photo of the OSB sample after 35 kWm?2 exposure is shown in Fig. 4.7. A
considerable residual char was left behind, whose mass measured (7.20? 1.73) g; a char yield of
27.7%. A notable white ?ashing? was seen on the surface, believed to be due to mineral content
in the OSB being transferred to the surface during the drying process. As the sample turns into
char, this residue layer becomes apparent. This is considered typical of any well-dried specimen.
For the purposes of this evaluation, it is considered simply a visual effect and is not believed to
have an impact on the results.
The mass loss rate and bottom surface temperature for OSB at 35 kWm?2 and the data
collected by Gong et al. [36] is shown in Fig. 4.8. The mass loss rate data shows an initial large
peak and then a slow decline with a slight second peak. This second peak is not as strong as
that seen by Gong et al. Given the large discrepancy in the density of the tested samples, and the
variability observed both here and by Gong et al. [36], this is considered acceptable typical of the
OSB sheet characteristics. A difference in the steady-state temperature of the back surface was
found, which is slightly outside the experimental uncertainty of both this study and the previous
one. This is likely due to the lower inner wall temperatures of this version of CAPA increasing the
radiative and convective losses to the back surface, and lowering its temperature. Additionally,
differences in the material itself or slightly different infrared camera/paint calibrations may play
a role.
The mass loss rate and bottom surface temperature for OSB at 65 kWm?2 and the data
collected by Gong et al. [36] is shown in Fig. 4.9. The OSB was separated from the foil at the
conclusion of the test, although it is difficult to ascertain at which point this occurred. Given the
62
Figure 4.7: Post test photograph of OSB at 35 kWm?2
similarity to previous data, this is considered typical and not thought to affect the results of this
evaluation. The mass loss rate data shows a two-peak structure and rapid decline. Compared to
the data from Gong et al. the second peak in this study is slightly less pronounced, but otherwise,
excellent agreement is found. This shifting is consistent with the observations made by Gong et
al. with regard to the movement of the second peak with variations of sample density. The bottom
surface temperature shows the same trend as that at 35 kWm?2 with a near exact match until late
in the test, where an apparent scaling difference takes place. The same inner wall temperature
discrepancies as noted above are used to describe this phenomenon. This is considered acceptable
given the experimental variability seen by Gong et al. and the density and material differences.
63
0.007 This Study (density: (621? 24) kg m
?3), ?2?
Gong et al. (density: (690? 14) kg m?3), ?2?
0.006
0.005
0.004
0.003
0.002
0.001
0.000
0 250 500 750 1000 1250 1500 1750 2000
Elapsed Time [s]
(a)
700
This Study (density: (621? 24) kg m?3), ?2?
650 Gong et al. (density: (690? 14) kg m
?3), ?2?
600
550
500
450
400
350
300
0 250 500 750 1000 1250 1500 1750 2000
Elapsed Time [s]
(b)
Figure 4.8: Oriented strand board (a) mass loss rate and (b) bottom temperature at 35 kWm?2
radiant exposure
64
Tbottom [K] MLR [kg m?2 s?1]
0.0200
This Study (density: (620? 26) kg m?3), ?2?
0.0175 Gong et al. (density: (644? 12) kg m
?3), ?2?
0.0150
0.0125
0.0100
0.0075
0.0050
0.0025
0.0000
0 200 400 600 800 1000
Elapsed Time [s]
(a)
700
650
600
550
500
450
400
350 This Study (density: (620? 26) kg m?3), ?2?
Gong et al. (density: (644? 12) kg m?3), ?2?
300
0 200 400 600 800 1000
Elapsed Time [s]
(b)
Figure 4.9: Oriented strand board (a) mass loss rate and (b) bottom temperature at 65 kWm?2
radiant exposure
65
Tbottom [K] MLR [kg m?2 s?1]
Chapter 5: Conclusion
This study presented the continued development and improvement of the Controlled At-
mosphere Pyrolysis Apparatus (CAPA). Given that only one functional and published version of
CAPA currently exists, it is essential for widespread adoption and confidence that an investigator
can replicate the design and results across versions of apparatuses and physical laboratories.
While maintaining many of the original geometrical constraints to achieve replicability,
particular design elements of CAPA were modified to improve both the data-gathering ability and
the definition of boundary conditions. The frame structure was modified to provide a more stable
and smooth operation of the heater mount by including V-groove rails and a balance shield that
isolates from ambient effects. A permanent heat flux gauge mount with a separate water supply
reduces shifting of the measurement location and accounts for inter-test heat flux variations.
Upgraded sensors, such as the mass balance and infrared camera, were used, allowing a higher
resolution, acquisition rate, and data accuracy. The use of integral water cooling for the chamber
walls managed to reduce their temperatures to near-ambient, almost wholly eliminating transient
radiative effects and greatly simplifying the process of modeling experiments.
Comprehensive but restricted characterization work also confirmed satisfactory operation
of the apparatus and compliance with previous work. Heat flux and gas flow uniformity were
within expected values and bolster previous calculations used to define the device?s operating
66
conditions. Direct measurement of gas temperatures of the apparatus presented here, and that
designed previously confirmed insignificant differences in steady-state temperatures although a
notable decrease in the rise time. A direct comparison of the overall thermal environments yielded
a minor discrepancy in a copper plate?s temperature values, further reinforcing this design?s repli-
cability.
A direct experimental comparison was made to further confirm the replicability with a
non-charring (PMMA) and charring (OSB) material previously characterized with CAPA. For
PMMA, general trends were achieved; however, the effect of sample delamination from the cop-
per substrate yielded significant impacts on both the mass loss rate and bottom surface temper-
ature. Although this only manifested in the data for the lower heat flux, further study should be
conducted to eliminate this effect, if at all possible. The use of different types or formulations of
adhesives may be required for specific materials. For OSB, the mass loss rate data gave excellent
results, particularly in light of the density and material variations of the raw sheet and trends
observed previously. Differences in the steady-state back surface temperature values warrant
an investigation of alternative well-characterized optical coatings to ensure accurate emissivity
values across a wide temperature range.
This study reinforces a growing body of literature utilizing CAPA and confidence in the
apparatus. The results found here are intended to be used as a basis for further studies and
to inform the design of future instruments. Additional work will examine the development of
a forward and inverse model utilizing Fire Dynamics Simulator (FDS) to validate and extract
materials properties. Given the issues identified in this study, ongoing studies are being conducted
to determine suitable high-temperature adhesives and thermographic paints to improve the quality
and quantity of data acquired.
67
APPENDIX A
68
7.3 Parameter List
7.3.1 Parameters in the ?Setup? Main Menu
Parameters in the ?Balance? Submenu
Parameters Setting values Explanation
AMBIENT V.STABLE Sets the ambient conditions to ?very stable?: Activates a fast change in the weight values 
in the event of a load change with a high output rate.
Recommended for the following work environment:
? Very stable table near the wall
? Closed and calm room
STABLE* Sets the ambient conditions to ?stable?.
Recommended for the following work environment:
? Stable table
? Slight movement in the room
? Slight draft
UNSTABL. Sets the ambient conditions to ?unstable?: Activates the delayed change in weight values 
with a reduced output rate.
Recommended for the following work environment: 
? Simple office desk
? Room with moving machinery or personnel
? Slight air movement
V.UNSTBL. Sets the ambient conditions to ?very unstable?: Activates a significantly delayed change 
in the weight values and long wait for stability with a further reduction in the output 
rate.
Recommended for the following work environment:
? Noticeable and slow floor vibrations
? Noticeable building vibrations
? Weighed goods moved
? Very strong air movements
APP FILT. FINAL.RD.* Activates a filter that enables a fast change in the display for very fast load changes. 
Display changes with minimal load changes (in the digit range) occur more slowly.
FILLING Activates a filter that enables a very fast change in the display with minimal load 
changes (e.g. when filling containers).
REDUC. Activates a weak but fast filter that always behaves in the same way for load changes 
(e.g. when filling automated systems).
OFF Deactivates the active application filter.
STABIL. MAX ACC. Sets the stability to ?maximum accuracy?.
V. ACC. Sets the stability to ?very accurate?.
ACC.* Sets the stability to ?accurate?.
FAST Sets the stability to ?fast?.
V. FAST Sets the stability to ?very fast?.
MAX.SPEED Sets the stability to ?maximum speed?.
* Factory setting
Parameters Setting values Explanation
ST.DEL. NONE Sets the stability delay to ?none?: The stability symbol is displayed after the stability 
criterion is reached.
SHORT* Sets the stability delay to ?short?: The stability symbol only appears after a short delay in 
order to provide a reliable result despite fluctuations.
MEDIUM Sets the stability delay to ?medium?: The stability symbol only appears after a longer 
delay in order to provide a reliable result in case of higher fluctuations.
LONG Sets the stability delay to ?long?: The stability symbol only appears after a long delay in 
order to balance out major instability.
ZERO/TAR. W/O STB. Without stability: The function of the [Zero] or [Tare] key is executed immediately once 
the key is pressed.
W/ STAB.* With stability: The function of the [Zero] or [Tare] key is only executed after stability is 
achieved.
AT STAB. At stability: The function of the [Zero] or [Tare] key is executed if stability exists when 
the key is pressed.
AUTOZER. ON* Activates automatic zeroing. The display is automatically set to zero in case of a 
deviation of 0 less than (X).
OFF Deactivates automatic zeroing. Zeroing must be triggered with the [Zero] key. 
UNIT The availability of units may depend on national legislation and is therefore country-
specific. 
GRAMS* The device displays the weight in grams.
KILOGR. The device displays the weight in kilograms (not for semi-microbalances and 
microbalances).
CARATS The device displays the weight in carats.
POUNDS The device displays the weight in pounds (not for semi-microbalances and 
microbalances).
OUNCES The device displays the weight in ounces (not for microbalances).
TROY OZ. The device displays the weight in troy ounces (not for microbalances).
HKTAEL The device displays the weight in taels ? Hong Kong (not for microbalances).
SNGTAEL The device displays the weight in taels ? Singapore (not for microbalances).
TWNTAEL The device displays the weight in taels ? Taiwan (not for microbalances).
GRAINS The device displays the weight in grains.
PENYWT. The device displays the weight in pennyweights.
MILLIGR. The device displays the weight in milligrams (not for high-capacity precision balances).
CHINATAEL The device displays the weight in taels ? China (not for microbalances).
MOMMES The device displays the weight in mommes.
TOLA The device displays the weight in tolas.
BAHT The device displays the weight in baht (not for microbalances).
MESGHAL The device displays the weight in mesghals.
NEWTON The device displays the weight in newtons (not for microbalances).
* Factory setting
Parameters Setting values Explanation
DISP.DIG. ALL* ?Show all decimal places?: All decimal places are shown in the display. Not available on 
conformity-assessed devices.
LP.ON/OFF ?Reduced by 1 decimal place for load change?:  
The last decimal place on the display is switched off until stability is achieved.
DIVIS. 1 ?Last decimal place of the 1st division?: The last decimal place always shows the 1st 
division.
MINUS 1 ?Last decimal place off?: The last decimal place is switched off.
CAL./ADJ. EXT.CAL. The [Adjust] button starts an external calibration with the preset calibration weight.
E.CAL.USR. The [Adjust] button starts an external calibration with the user-defined calibration 
weight value.
INT.CAL.* The [Adjust] button starts an internal calibration.
CAL.SEQ. ADJUST* Calibration and adjustment is one routine.
CAL.-ADJ. Adjustment must be started or exited manually after calibration with the [Adjust] 
button.
ON Z/T ON* Activates the initial taring / zeroing. The device is tared or zeroed after it is switched on.
OFF Deactivates the initial taring / zeroing: After it is switched on, the device shows the value 
before it was last switched off.
ISOCAL OFF Switches the isoCAL function off.
NOTE TO The [isoCAL] button flashes if the balance needs to be adjusted. The isoCAL function 
must be manually triggered with the [Adjust] button.
ON* Activates the isoCAL function. The device is automatically adjusted as soon as a trigger 
starts the isoCAL function.
CAL.UNIT GRAMS* Changes the calibration weight unit to grams.
KILOGR. Changes the calibration weight unit to kilograms (not for semi-microbalances and 
microbalances).
CONF.UNIT Changes the calibration weight unit to milligrams (not for precision balances and high-
capacity precision balances).
* Factory setting
Parameters in the ?General Services? Submenu
Parameters Setting values Explanation
MEN.RESET YES Resets the system settings to the factory default settings.
NO* Deactivates the option of resetting the device menu. 
* Factory setting
APPENDIX B
72
Figure B.1: CFD model overall mesh
73
Figure B.2: CFD model fine mesh for water channel
74
APPENDIX C
75
Table C.1: Material properties used for CFD model
Material
Aluminum Water
Density [kgm?3] 2719 998.2
Cp (Specific Heat) [J kg?1K?1] 871 4182
Thermal Conductivity [Wm?1K?1] 202.4 0.6
Viscosity [kgm?1 s?1] - 0.001003
76
Bibliography
[1] Vytenis Babrauskas. The Cone Calorimeter. In Morgan J. Hurley, Daniel Gottuk, John R.
Hall, Kazunori Harada, Erica Kuligowski, Milosh Puchovsky, Jose Torero, John M. Watts,
and Christopher Wieczorek, editors, SFPE Handbook of Fire Protection Engineering, chap-
ter 28, pages 952?980. Springer New York, New York, NY, 2016.
[2] John Barton, Oriol Rios, Marcus Runefors, and Patrick van Hees. The effect of oxygen con-
centration on selected industrial products in the open controlled atmosphere cone calorime-
ter. Fire and Materials, 46(4):617?627, 2022.
[3] Marc L. Janssens. Calorimetry. In Morgan J. Hurley, Daniel Gottuk, John R. Hall, Kazunori
Harada, Erica Kuligowski, Milosh Puchovsky, Jose Torero, John M. Watts, and Christopher
Wieczorek, editors, SFPE Handbook of Fire Protection Engineering, chapter 27. Springer
New York, New York, NY, 2016.
[4] Xuan Liu. Design and Analysis of New Gasification Apparatus Based on the Standard Cone
Calorimeter. Master?s thesis, University of Maryland, 2012.
[5] Joshua D. Swann. A Comprehensive Characterization of Pyrolysis and Combustion of In-
tumescent and Charring Polymers Using Two-Dimensional Modeling: A Relationship Be-
77
tween Thermal Transport and the Physical Structure of the Intumescent Char. PhD thesis,
University of Maryland, 2019.
[6] Joshua D. Swann, Yan Ding, Mark B. McKinnon, and Stanislav I. Stoliarov. Controlled at-
mosphere pyrolysis apparatus II (CAPA II): A new tool for analysis of pyrolysis of charring
and intumescent polymers. Fire Safety Journal, 91(April):130?139, 2017.
[7] Vytenis Babrauskas. Development of the Cone Calorimeter ? A Bench-Scale Heat Release
Rate Apparatus Based on Oxygen Consumption. NBS Information Report (NBSIR) 82-
2611, National Bureau of Standards, Washington, DC, 1982.
[8] ASTM E1354-22a. Standard Test Method for Heat and Visible Smoke Release Rates for
Materials and Products Using an Oxygen Consumption Calorimeter. Standard, ASTM In-
ternational, West Conshohocken, PA, 2022.
[9] ISO 5660-1:2015. Reaction-to-fire tests ? Heat release, smoke production and mass loss
rate ? Part 1: Heat release rate (cone calorimeter method) and smoke production rate
(dynamic measurement). Standard, International Organization for Standardization, Geneva,
CH, March 2015.
[10] Clayton Huggett. Estimation of rate of heat release by means of oxygen consumption mea-
surements. Fire and Materials, 4(2):61?65, 1980.
[11] William H. Twilley and Vytenis Babrauskas. User?s Guide for Cone Calorimeter. NBS
Special Publication 745, National Bureau of Standards, Gaithersburg, MD, 1988.
[12] Michael T. Wilson, Bogdan Z. Dlugogorski, and Eric M. Kennedy. Uniformity of radiant
heat fluxes in cone calorimeter. Fire Safety Science, 8:815?827, 2003.
78
[13] Vytenis Babrauskas, William H. Twilley, Marc L. Janssens, and Shyuitsu Yusa. A Cone
Calorimeter for Controlled-atmosphere Studies. Fire and Materials, 16(1):37?43, 1992.
[14] Damien Marquis, Eric Guillaume, and Damien Lesenechal. Accuracy (trueness and pre-
cision) of cone calorimeter tests with and without a vitiated air enclosure. Procedia Engi-
neering, 62:103?119, 2013.
[15] Zohreh Ghorbani, Robert Webster, Mariano La?zaro, and Arnaud Trouve?. Limitations in the
predictive capability of pyrolysis models based on a calibrated semi-empirical approach.
Fire Safety Journal, 61:274?288, 2013.
[16] Chris Lautenberger, Guillermo Rein, and Carlos Fernandez-Pello. The application of a
genetic algorithm to estimate material properties for fire modeling from bench-scale fire
test data. Fire Safety Journal, 41(3):204?214, 2006.
[17] Robert Dale Webster. Pyrolysis Model Parameter Optimization Using a Customized
Stochastic Hill-Climber Algorithm and Bench Scale Fire Test Data. Master?s thesis, Uni-
versity of Maryland, 2009.
[18] ASTM E2058-19. Standard Test Methods for Measurement of Material Flammability Using
a Fire Propagation Apparatus (FPA). Standard, ASTM International, West Conshohocken,
PA, 2019.
[19] ISO 12136:2011. Reaction to fire tests ? Measurement of material properties using a fire
propagation apparatus. Standard, International Organization for Standardization, Geneva,
CH, 2011.
79
[20] Ryan T. Chaffer. A Redesign of the Fire Propagation Apparatus to Simplify Manufacturing
and Improve Measuring Capabilities. Master?s thesis, University of Maryland, 2021.
[21] Pierre Girods, Nicolas Bal, Hubert Biteau, Guillermo Rein, and Jose? L. Torero. Compar-
ison of pyrolysis behaviour results between the cone calorimeter and the fire propagation
apparatus heat sources. Fire Safety Science, (March 2016):889?901, 2011.
[22] Marcos Chaos, Mohammed M. Khan, Niveditha Krishnamoorthy, John L. De Ris, and
Sergey B. Dorofeev. Evaluation of optimization schemes and determination of solid fuel
properties for CFD fire models using bench-scale pyrolysis tests. Proceedings of the Com-
bustion Institute, 33(2):2599?2606, 2011.
[23] Mihyun Esther Kim, Christopher William Lautenberger, and Nicholas Dembsey. Property
Estimation for Pyrolysis Modeling Applied to Polyester FRP Composites with Different
Glass Contetns. In Composites & Polycon, Tampa, FL, jan 2009.
[24] Jing Li. A Multiscale Approach to Parameterization of Burning Models for Polymeric Ma-
terials. PhD thesis, University of Maryland, 2014.
[25] Jing Li, Junhui Gong, and Stanislav I. Stoliarov. Gasification experiments for pyrolysis
model parameterization and validation. International Journal of Heat and Mass Transfer,
77:738?744, 2014.
[26] Mollie R. Semmes, Xuan Liu, Mark B. Mckinnon, Stanislav I. Stoliarov, and Artur
Witkowski. A Model for Oxidative Pyrolysis of Corrugated Cardboard. Fire Safety Sci-
ence, 11:111?123, 2014.
80
[27] Mark B. McKinnon and Stanislav I. Stoliarov. Pyrolysis model development for a multilayer
floor covering. Materials, 8(9):6117?6153, 2015.
[28] Jing Li, Junhui Gong, and Stanislav I. Stoliarov. Development of pyrolysis models for
charring polymers. Polymer Degradation and Stability, 115:138?152, 2015.
[29] Mark B. McKinnon, Yan Ding, Stanislav I. Stoliarov, Sean Crowley, and Richard E. Lyon.
Pyrolysis model for a carbon fiber/epoxy structural aerospace composite. Journal of Fire
Sciences, 35(1):36?61, 2017.
[30] Joshua D. Swann, Yan Ding, and Stanislav I. Stoliarov. Experimental study of anaer-
obic pyrolysis of poly(vinyl chloride). 10th U.S. National Combustion Meeting, 2017-
April(February 2017), 2017.
[31] Gregory J. Fiola, Dushyant Madhav Chaudhari, and Stanislav I. Stoliarov. Comparison of
Pyrolysis Properties of Extruded and Cast Poly(methyl methacrylate). Fire Safety Journal,
120(December 2019):103083, 2021.
[32] Joshua D. Swann and Stanislav I. Stoliarov. Determination of pyrolysis and combustion
properties of poly(vinylidene fluoride) using comprehensive modeling: Relating heat trans-
fer to the intumescent char?s porous structure. Fire Safety Journal, 120(April 2020):103086,
2021.
[33] Joshua D. Swann, Yan Ding, and Stanislav I. Stoliarov. A quantitative comparison of the
pyrolysis and combustion behavior of plasticized and rigid poly(vinyl chloride) using two-
dimensional modeling. Fire Safety Journal, 111(July 2019), 2020.
81
[34] Yan Ding, Stanislav I. Stoliarov, and Roland H. Kraemer. Pyrolysis model development
for a polymeric material containing multiple flame retardants: Relationship between heat
release rate and material composition. Combustion and Flame, 202:43?57, 2019.
[35] R. Falsetti, C.and Kapulla, S. Paranjape, and D. Paladino. Thermal radiation, its effect
on thermocouple measurements in the panda facility and how to compensate it. Nuclear
Engineering and Design, 375:111077, 2021.
[36] Junhui Gong, Hongen Zhou, Hong Zhu, Conor G. McCoy, and Stanislav I. Stoliarov. De-
velopment of a pyrolysis model for oriented strand board: Part II?Thermal transport pa-
rameterization and bench-scale validation. Journal of Fire Sciences, 39(6):477?494, 2021.
[37] Michael Schmid, David Rath, and Ulrike Diebold. Why and How Savitzky?Golay Filters
Should Be Replaced. ACS Measurement Science Au, 2(2):185?196, 2022.
[38] Floris van Breugel, Yuying Liu, Bingni W. Brunton, and J. Nathan Kutz. Pynumdiff: A
python package for numerical differentiation of noisy time-series data. Journal of Open
Source Software, 7(71):4078, 2022.
[39] Floris van Breugel, J. Nathan Kutz, and Bingni W. Brunton. Numerical differentiation of
noisy data: A unifying multi-objective optimization framework. IEEE Access, 8:196865?
196877, 2020.
82