Professional Documents
Culture Documents
net/publication/340644926
CITATIONS READS
13 514
3 authors:
Jose Torero
The University of Queensland
388 PUBLICATIONS 7,674 CITATIONS
SEE PROFILE
All content following this page was uploaded by Angela Solarte on 15 April 2020.
ABSTRACT: In recent time, material developers have created novel construction products such as laminated bamboo
lumber, bamboo scrimber, and bamboo oriented strand boards. All of them have similar mechanical material properties
to those of engineered timber. Nonetheless, their fire behavior is still not fully understood. This paper presents the
results of a flammability analysis of laminated bamboo samples produced from the species Phyllostachys pubescens
“Moso”. Variables such as critical heat flux and mass loss rate for ignition, as well as time-to-ignition are taken into
account. Several methods used for calculation are presented and a reference to previous tests on similar bamboo species
is provided. Last, the ignition parameters of laminated bamboo are contrasted against commonly used timber products.
KEYWORDS: Engineered bamboo products, fire performance, flammability parameters, mass loss rates.
The primary objective of this paper is to quantify the Specimens were placed in a horizontal position at
flammability parameters that drive the fire behaviour of ambient conditions, while being subjected to a
this product. By doing so, a proper characterization of predetermined external heat flux from a cone-shaped
the behavior of bamboo in fire can be obtained, and the radiant electric heater [14]. Each sample was subjected
design of fire-safe bamboo structures can be established to a controlled spark located at 25 mm above the surface
to predict the onset of hazard due to fire. In this research of the sample until ignition was achieved. The samples
line, parameters such as critical heat flux for ignition, were exposed to radiant heat fluxes between 10-80
ignition temperature, thermal inertia, and burning rates kW/m2 (typical range for a fire); the time when ignition
will be calculated for a specific kind of laminated occurred as well as the change in mass were recorded.
bamboo product. When the sample was ignited, gases were collected into
the extraction hood, and then the heat release rate was
2 MATERIALS AND METHODS calculated with oxygen consumption calorimetry [15].
However, as this study focuses on the onset of hazard
The first stage of testing consisted of bench-scale heat release rate results will not be presented.
experiments using the Cone Calorimeter [11]. This
device is a fire testing equipment that measures the 2.1 FLAMMABILITY PARAMETERS
oxygen consumption and sample mass loss during the
The following assumptions and simplifications are used
combustion process of a material. The apparatus allows
in the study [12]:
to define parameters such as the critical heat flux for
ignition, temperature and time for ignition, thermal 1. The solid is considered inert until ignition,
inertia and burning rates, which are key to understand 2. The sample is a semi-infite solid,
the ignition behavior of solids. 3. Ignition will occur at the onset of pyrolysis, and
4. Pyrolisis will be accomplished when the surface
Two main steps determine the flamability
reaches the ignition temperature.
characterization for this methodology. First, obtaining
igntion parameters such as the critical heat flux for To validate if the sample behaves as a semi-infinite
ignition (qcr”), and time and temperature for ignition (tig, solid, four tests were previously run to determine the
Tig) as stated by Torero [12]. Second, analyzing the applicable boundary condition. Thermocouples were
burning process of the material by determining the placed inside the sample and the progression of the heat
critical mass loss rate at the firepoint (the beginning of wave was recorded. The duration of each test was long
sustained flaming combustion) as detailed by Rasbash et enough to ensure that the characteristic delay time of the
al. [13]. thermal wave to reach the back were much longer than
the time required for the thermal wave to travel through
To achieve this, 53 samples of laminated bamboo
the sample.
(Phyllostachys pubescens “Moso”) were tested. The
samples were 90 mm long, 90 mm wide and 40 mm During each experiment, detailed observations of the
thick with 6-7 lamellae. Average density was measured response of the laminated bamboo samples were made
at 570 kg/m3 with a moisture content of 6.3%. The and the time of ignition was recorded when a sustained
samples were placed in a standard sample holder and flame was observed. All other sporadic flashes were
covered with aluminium foil. As it can be seen in Figure recorded as events, but not as ignition. The information
1, a 2.5 mm thick ceramic layer was placed on the sides of time to ignition was plotted against each heat flux
while a 50 mm thick ceramic fibreboard was located at applied. The critical heat flux for ignition, defined as the
the back as insulation. The tests were stopped after 60 lowest heat flux in which ignition does not take place
minutes if no ignition was achieved. was obtained by extrapolationg the plot to find the
asymptotic value of the external heat flux as ignition
time tends to infinite [16].
Once the critical heat flux for ignition was obtained, it Rasbash et al. [13] provided a more detailed approach to
was used to determine the temperature for ignition by this theory. They stated that the mass loss measured is
means of equation 1 [12]. initially due to the pyrolysis process of the sample, and
the production of volatile gases. This will eventually
" 𝛼
𝑞𝑐𝑟 create a critical mass flow of fuel volatiles at the
𝑇𝑖𝑔 = 𝑇∞ + (1) firepoint, which combustion can produce enough heat to
ℎ𝑇 sustain the generation of volatiles from the solid to
sustain the flame. According to Rasbash et al.,
where Tig is the ignition surface temperature [K], T∞ is determining the critical mass loss rate with pilot ignition
the ambient temperature [K], qcr” is the critical heat flux is a method for evaluating ignition in terms of the
for ignition [kW/m2], is the surface absorptivity [-], products generated by material decomposition [13].
and hT is the total heat transfer coeficient for heat losses
[W/m2 K]. For each of the experiment performed in the Cone
Calorimeter, the mass loss was recorded in a 1-second
Once the material has reached the ignition surface interval. With the use of a load cell the weight of the
temperature (temperature for ignition), if the sample is sample was recorded while being exposed to a known
exposed to an external heat flux, then the time measured radiant heat flux. The mass loss rate per unit area was
between the exposure and ignition will be the ignition obtained by differentiation of the mass over the time step
delay time. Equation 2 shows a commonly used model to and divided by the exposed surface area of the sample.
obtain the time to ignition for themally thick materials
The mass loss rate per unit area (g/m2s) calculated over
[12]. The thermal thickness is a parameter used to
each second was graphed vs time (s). A typical mass loss
charactarize a material when the temperature across the
rate profile is one were the mass loss rate increases
material has a temperature gradient and is not constant.
gradually before achieving a steep increment around the
2 time in which the sample ignites [13]. The critical mass
𝜋 𝑇𝑖𝑔 − 𝑇∞ loss rate for ignition (MLRc) in the laminated bamboo
𝑡𝑖𝑔 = 𝑘𝜌𝐶𝑝 (2) samples was obtained at the inflection point where
4 𝑞𝑒" ignition takes place.
From this equation tig is the ignition delay time [s], Tig is One of the major factors of uncertainty was due to the
the ignition surface temperature [K], T∞ is the ambient imprecision of the load cell. It was observed that
temperature [k], qe” is the external heat flux for ignition recordings had major reading fluctuations in the outcome
[kW/m2], k is the themal conductivity [W/m K], Cp is the data, and finding the critical point was quite inconsistent.
specific heat [J/kg K] and ρ is the density [kg/m3]. The direct results from the readings of the recorded time
to ignition were discarded because the data signal
The time to ignition depends on the thermal inertia, i.e. included significant noise. Initial analyses deemed the
the product of the thermal conductivity (k), the density data unreliable.
(ρ) and the specific heat (Cp). It characterizes the ability
of a material to resist the change in its own temperature. Another factor of uncertainty found in this analysis was
This parameter is calculated by obtaining the inverse of that the values of the mass loss near ignition tend to be
the square root of the time to ignition (tig) as a function small, especially at high heat fluxes [19], so the
of the external heat flux by a linear relationship. The calculation of these values can produce errors in the
value of the slope is related to the thermal inertia. exact definition of the ignition time. To minimize these
uncertainties, three different procedures were followed
1 2 1 1 to determine the best and most accurate way to obtain
= 𝑞𝑒" (3) the critical mass loss rate for ignition.
𝑡𝑖𝑔 𝜋 𝑘𝜌𝐶𝑝 𝑇𝑖𝑔 − 𝑇∞
2.2.1 Limiting smooth slope intercept
Finally, with Equation 2 and 3 the thermal inertia is
obtained where tig is the ignition delay time [s], Tig is the For each of the plots, the smoothed trend of the data was
ignition surface temperature [K], T∞ is the ambient obtained with the use of Locally Weighted Regression
temperature [k], qe” is the external heat flux for ignition (LOESS). According to Cleveland et al., LOESS is a
[kW/m2], k is the thermal conductivity [W/m K]. procedure used for fitting a regression surface through
smoothing in order to carry out data prediction [20, 21].
2.2 MASS LOSS RATES AT THE FIREPOINT Figure 2 shows an example of this limiting slope
The second step to characterize ignition was through the intercept approach. Once the fitted/smoothed curve was
analysis of the rate of mass loss of the samples while obtained (the solid line following the dotted trend), two
being heated. This criterion was first proposed by straight lines were used to identify each trend (solid red
Bamford et al. in 1946 [17]. Kanury stated that in the lines) and in their intersection was marked as the mass
threshold of ignition, the pyrolysis rate is expected to be loss rate for ignition (red dotted line). In this particular
increased so much that the resulting mixture of gases case the MLRc obtained by this method is 2 g/m2s.
reaches concentrations above the lower flammability
limits [18].
Figure 2: Limiting smooth slope intercept example to find the Figure 3: Ignition times (tig) and external heat flux for ignition
critical mass loss rate (g/m2s). (qe) for laminated bamboo samples. Vertical red line indicates
the critical heat flux for ignition.
2.2.2 Average smoothing data
With the value of the critical heat flux for ignition
In this procedure, the critical mass loss rate is calculated known, the temperature for ignition was calculated with
by taking the average of the data points around the Equation 1. The value used for total heat transfer was
change in slope of the “smoothing data” when the obtained as 37 W/m2K following the regression proposed
increment of the curve starts. by Hidalgo [22] based on a convective coefficient for
horizontal hot plates and an emissivity of 0.8-1.0. The
2.2.3 Average noise data
value calculated for the temperature of ignition for these
The third approach consisted in calculating the average tests was 365ºC.
of the data points around the changed tendency and slope
increment of the fluctuating “noise data”, taking into
consideration the unstable values.
Results from the three methods for each heat flux tested
are reported and a comparison between the values
obtained is provided.