Received: 23 May 2022 Revised: 13 July 2022 Accepted: 27 July 2022

DOI: 10.1002/fam.3101

RESEARCH ARTICLE

A probabilistic thermal radiation analysis of cooking hob

separation distance in open plan kitchens

Michael Spearpoint1 | Charlie Hopkin2 | Danny Hopkin1

1
OFR Consultants, Manchester, UK
2
Ashton Fire, Manchester, UK Abstract
Cooking oil fires present a hazard to the occupants of residential flats with open plan
Correspondence
Michael Spearpoint, OFR Consultants, kitchens where the means of escape passes fixed cooking appliances. Recent guid-
Manchester, UK. ance has proposed a separation distance of 1.8 m from a hob/stovetop, but the basis
Email: [email protected]
of this advice is unclear. This paper carries out a probabilistic thermal radiation analy-
sis using the thermal fractional effective dose (FED) methodology to assess the likeli-
hood that the specified separation distance will be exceeded. Statistical data
collected from a previous survey of cooking oil usage is used to determine reasonable
worst-case maximum heat release rate values as input to the point source model to
calculate thermal radiation exposure. Occupant exposure times are assessed using
walking speeds based on occupant demographics. The analysis suggests that the
1.8 m criterion equates to approximately a 99th percentile where there is no cap on
the heat flux that an occupant can tolerate. This falls to the 88th percentile when the
maximum tolerable heat flux is capped at 2.5 kW/m2 to account for vulnerable
occupants.

1 | I N T RO DU CT I O N These studies formed the basis for recommendations in the previous

edition of BS 9991:20111 and the current edition of BS 9991:20155
The term “flat” used in this paper refers to an apartment which is with regard to the floor area of flat arrangements with open plan
a separate and self-contained residential premises forming part of a kitchens and the provision of sprinklers. Hopkin et al.2 concluded
building, having all its rooms on one level.1 When designing flats that open plan kitchens can be considered acceptable in open plan
with open plan kitchen areas there is an expectation that the hob flats, irrespective of the flat floor area (up to a given limit), but the
of a cooking appliance is located such that a fire does not impede cooking facilities should be located remotely from both the final exit
the escape of occupants. Historically no guidance on an acceptable and occupant escape routes. This aligns with the acceptability of
separation distance has been published, although recent develop- open plan kitchens in other jurisdictions such as Australia,6 New
ments change this situation. This paper examines the separation Zealand,7 Canada8 and the US.9,10 However, by including open plan
distance of fixed cooking appliances for flats, which incorporate kitchens in a flat design, additional considerations are needed in
open plan kitchens by adopting a probabilistic methodology of terms of what is deemed an acceptable location of the cooking facili-
assessing the thermal fractional effective dose (FED) of an escaping ties, such as where they are located relative to the occupant's
occupant. escape route and where they are located relative to the final exit
Hopkin et al2 previously undertook a review of the fire safety door of the flat. With respect to the internal planning of flats, BS
design of open plan flats. Part of the review included discussion on 9991 recommends the following:
flats incorporating open plan kitchens. The review discussed how
the National House Building Council (NHBC) had previously commis- • “Cooking facilities in open plan flats or maisonettes should be
sioned a study into the risk from fires associated with open plan located in such a way that they do not prevent escape if they are
arrangements3 with follow-up work then conducted by Davis et al.4 involved in a fire” (BS 9991 clause 9.1).

352 © 2022 John Wiley & Sons Ltd. wileyonlinelibrary.com/journal/fam Fire and Materials. 2023;47:352–362.
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL. 353

• “Cooking facilities should be sited away from the flat entrance _

_ ¼ Qr cosθ
q} ð1Þ
door and the internal escape route” (BS 9991 clause 9.4.2.a). 4πR2

BS 9991 does not explicitly specify what is an acceptable distance where Q_ r is the total radiative energy from the fire (kW), θ is the angle
between cooking facilities, the entrance door and the escape route, between the normal to the target ( ) and the line of sight from the tar-
and therefore the guidance can be interpreted in different ways. In get to the point source location and R is the horizontal distance from
contrast to this, the Ireland Technical Guidance Document B the point source to the target (m).
(TGDB)11 suggests that when the kitchen is not enclosed “the flat lay- For a radiative fraction of χ rad and a fixed θ value of 0 (the most
out should be designed such that an occupant escaping the flat should conservative case for the incident heat flux), the equation can be
not have to approach within 1.8 m of the main kitchen cooking appli- amended to the following:
ance” (clause 1.6.3.e). TGDB clarifies that “a main kitchen cooking
_ _
appliance is a cooking apparatus with a fixed connection to the dwelling’s _ ¼ Qr ¼ Q  χ rad
q} ð2Þ
2
gas or electric energy infrastructure (for example, an oven and/or hob).” 4πR 4πR2
Recently a draft for public comment on a revised version of BS
999112 has been made available in which the same 1.8 m criterion is Spearpoint et al18 found that olive oil, vegetable oil and pure sun-
stated. However, it is not currently clear what the basis is for the sep- flower oil are the most frequently used types of cooking oil in the
aration distance recommendation or the safety level it corresponds to. home. Koseki et al.19 give the radiant fraction of various vegetable oils
In the absence of clarity in the guidance documents, this paper between 0.34 and 0.59 but they did not include olive oil or pure sun-
provides an assessment methodology to examine the potential ther- flower oil in their study. Chen et al.20 report the radiant fraction for
mal radiation from a developing fire scenario, which could impact corn oil during heating and auto-ignition as 0.45 ± 14%.
escaping occupants. The methodology first considers how to calculate
the radiant incident heat flux using the point source model. From this,
the tolerability resulting from radiation is considered using a thermal 2.2 | Thermal fractional effective dose
FED approach, for an occupant escaping past a cooker appliance/hob
fire. Probabilistic input distributions are determined for a cooking oil Purser and McAllister21 state that the thermal FED of an occupant
fire heat release rate (HRR) and radiant fraction using studies previ- can be determined as a function of the radiation and convection
ously undertaken by the authors, along with occupant demographics which may impact that occupant. As part of this, Purser and McAllister
from the literature. Finally, a simplified FED calculation has been itera- note that the “time to end point” (the time to reach a certain tolerable
tively undertaken using Monte Carlo sampling of the input parameters limit) is given by:
from which cumulative distribution functions (CDFs) of separation dis-
r
tances from a kitchen hob are generated. The CDFs are used to assess tIrad ¼ ð3Þ
_ 1:33
q}
the probability that specified separation distances exceed the tolera-
bility set by the FED analysis.
_ is the incident heat flux
where tIrad is the time to end point (min), q}
(kW/m2) and r is the radiant heat exposure dose ([kW/m2]4/3 min).
2 | M E TH O DO LO GY Purser and McAllister provide a range of recommendations for values
of r, as summarized in Table 1. The 1.33 (kW/m2)4/3min tolerance
2.1 | Determining incident heat flux limit has been applied when estimating the thermal FED, where this
value represents a lower bound for skin pain.
Spearpoint et al13 assessed the modelling of thermal radiation from To determine the thermal FED as a function of the radiant heat
kitchen hob fires by comparing point source, parallel plane and exposure dose and convective heat exposure dose (tIconv ), the following
cylindrical view factor hand calculations along with computational equation is applied by Purser and McAllister:
14
simulations using B-RISK and the Fire Dynamics Simulator
ð t2  
(FDS)15,16 with kitchen hob fire experiments presented in the litera- 1 1
FED ¼ þ Δt ð4Þ
ture. The experiments included cases in which the HRR achieved t1 tIrad tIconv

around 500 kW resulting in measured heat fluxes of the order of

80 kW/m2. As part of the study, it was identified that, in general, a T A B L E 1 Radiant heat exposure dose for exposed skin,
point source hand calculation gives a similar result to that given by reproduced from Purser and McAllister21
the more sophisticated computational tools. Therefore, the point
Radiant heat endpoint for exposed skin r([kW/m2]4/3.min)
source model is to be adopted as a sufficiently accurate approach
Severe skin pain 1.33–1.67
for this assessment.
Second-degree burns 4.0–12.2
For a point source model, the radiative incident radiative heat flux
_ kW/m2) is given by17 Third-degree (full-thickness) burns 16.7
(q},
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
354 SPEARPOINT ET AL.

T A B L E 2 Maximum thermocouple temperatures observed in the

42 experiments of Hamins et al.23,24

Thermocouple position Standard

below ceiling [m] Average [ C] deviation [ C]
0.03 (2.37 from floor) 214 207
0.30 (2.10 from floor) 132 180

where the FED is an integral between times t1 and t2 for time step
Δt (min).
Given that the analysis is concerned with the development of a
cooking oil fire, prior to ignition of any further combustibles within
the space, it would be expected that during this period that there will
be minimal development of an upper layer and that the lower layer
gas temperatures along the escape route will be within acceptable
levels of tenability (e.g., 100 C for an 8 min period or less than 60 C
for a water-saturated environment, as per PD 7974–622). For upper
layer temperatures, PD 7974–6 suggests that for 200 C with a 2.5 m
clear layer height “occupants are considered to be willing and able to
escape in clear air under such a layer and the downwards heat radiation FIGURE 1 Visual representation of the calculation method
is considered tolerable”.
The above position appears to be supported by experimental data
presented by Hamins et al.23,24 In experiments for cooking oil fires, a different positions of pans located on the hob. The target person is
mock kitchen was constructed with dimensions of 3.6 m long by assumed to walk in a straight line past the hob which is located equi-
3.4 m wide by 2.4 m high and temperatures were measured at differ- distant from the starting and ending point.
ent positions relative to the ceiling using a thermocouple tree. Data
for a combined 42 experiments are summarized and presented in
Table 2, where approximately 85% of these experiments considered a 2.3 | Maximum incident heat flux threshold
fuel volume of 4.5 L. The work of Hamins et al. and the practical rele-
vance of a 4.5 L oil volume are discussed further in Sections 2.4 and The thermal FED method is intended to capture the heat flux at each
2.5. Table 2 indicates that the maximum “ceiling jet” temperatures individual time step as an occupant escapes past a hob, considering
(0.03 m below ceiling) for the experimental data are on average the accumulated dose as a function of time. However, it is proposed
214 C, with maximum temperatures in the “upper layer” (0.30 m that a maximum fixed criterion for incident heat flux along any point
below ceiling) in the region of 132 C. It is important to note that the of the egress path will also be applied. This addresses a concern that
values presented in Table 2 represent brief maximum / peak values in there still could be a point where the radiative flux gets unacceptably
the timeline of the fire experiments, where temperatures subse- high, even if the duration is only for a short time such that the FED is
quently return to lower values following this peak. not exceeded.
Based on the above, it is proposed that the impact from convec- Figure 2 provides a summary of the results of experiments by
tion can be considered as negligible for the purposes of this assess- Stoll and Greene25 on pain and tissue damage resulting from thermal
ment, and the equation can be simplified to: radiation, where interpretations of these experiments have informed
many of the recommendations for radiation tenability in contempo-
ð t2  
1 rary fire safety guidance, including Purser and McAllister21 and the
FED ¼ Δt ð5Þ
t1 tIrad equations discussed in Section 2.2, PD 7974–6:201922 and
Drysdale.26
For the analysis presented herein, the FED has been calculated using The Stoll and Greene curve would suggest that the maximum
the input parameters discussed in Section 2.4 and adopting the point range of applicability is up to incident heat fluxes of 16.7 kW/m2 (the
source model for incident heat flux, summarized in Section 2.1. The highest fluxes assessed in the experiments). Despite this, it is pro-
method for the calculation is shown indicatively in Figure 1, where at posed that a lesser maximum incident heat flux limit of 2.5 kW/m2
progressing time intervals the incident heat flux is calculated, and sub- will be applied at any point along the escape route, equivalent to a
sequently the FED can be calculated as a function of the incident heat pain tolerance of approximately 30 s and a blistering threshold of
flux and travel time. For conservatism, it is assumed that the entirety around 70 s. This limit has been applied to accommodate for the pos-
of the hob represents the point source (shown in Figure 1), to accom- sibility of vulnerable occupants (not necessarily captured in the origi-
modate the possibility of different dimensions of the hob and nal dataset) who may potentially have lower tolerances for pain. In
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL. 355

F I G U R E 2 Tolerance times for pain

and blistering, adapted from Stoll and
Greene25

comparison, heat fluxes of 16.7 kW/m2 have shown to result in a tol- For these experiments, the oil mass ranged from 20 g up to 340 g
erance threshold of approximately 2 s for pain and 6 s for blistering. (~20 to 370 ml), with the maximum HRR reaching up to 111 kW for a
26 cm diameter pan with 340 g of corn oil.
Finally, a review of cooking oil use by means of information avail-
2.4 | Studies on cooking oil fires and usage able from consumer purchasing habits, studies on the health effects
of cooking with oil, and statistics related to the disposal of cooking oil
Spearpoint and Hopkin18 have previously carried out an assessment waste were undertaken by Spearpoint and Hopkin.18 In addition, an
of household cooking oil use and its bearing on fire safety. The work online survey was conducted to provide data on current cooking prac-
identified that the requirements for the testing of hood-mounted tices using oil. The survey, which amassed 291 responses, primarily
units intended to extinguish fires occurring on domestic gas and elec- included replies from residents of the UK. Analysis of the survey
tric hobs are given in UL 300A.27 The standard specifies four pans responses suggests that olive oil, vegetable oil or pure sunflower
(from c. 10 cm to c. 36 cm in diameter) using either vegetable oil or appear to be the most common types used for cooking in the home.
peanut oil in which oil depths range from ” (approx. 0.64 cm) up to
1/4 Spearpoint and Hopkin concluded that, on average, cooking with oil
400 (approx. 10 cm). Spearpoint and Hopkin were unable to establish occurs three times per month for deep frying and 11 times per month
the rationale for selecting the pan sizes or volumes of oil, but a rea- for shallow frying. Considering deep and shallow frying separately, the
sonable expectation is that they were chosen to obtain challenging 95th percentile volumes were determined to be 2.5 L and 150 ml
fires in terms of HRR as a test of the extinguishing systems. respectively. A 95th percentile design value of 250 ml was proposed
Spearpoint and Hopkin18 noted that several researchers have by Spearpoint and Hopkin for all frying activities, although further
conducted experimental work specifically on the fire safety of cooking analysis of the complete dataset suggests an upper value of 375 ml is
oil fires. Hamins et al.23,24 followed the specifications of UL 300A in more appropriate.
their experimental study of the performance of residential fire sup- The design values given by Spearpoint and Hopkin18 can be con-
pression technologies in response to cooking oil fires. In their work, sidered useful for a deterministic analysis. However, for the analysis
the heating source was allowed to remain energized after the oil had presented in this work the frequency of cooking oil use is applied
ignited. The maximum HRR was in the order of 30–115 kW. How- directly by random sampling from the original survey data.
ever, in one of the experiments using 4000 g of corn oil on a 2 kW
electric hob reached a maximum HRR of 472 ± 20 kW at ~5 min and
20 s later, at which point the fire was extinguished for safety reasons. 2.5 | Estimating a reasonable worst-case fire
Following on from the work described by Hamins et al.,23,24 Hu
et al.28 undertook a series of experiments to investigate the pre- The difficulty in considering cooking oil volume is approximating how
ignition and combustion characteristics of corn oil heated in 9.7 to it may impact the HRR of the fire. Typically, The HRR (Q, _ kW) of a
26 cm diameter pans energized on a 1.2 kW residential electric hob. pool fire can be calculated from:
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
356 SPEARPOINT ET AL.

  
Q _}
_ ¼ AQ ð6Þ _ ≈ AΔHc m
Q _ 00∞ 1  exp kβV 1=2 ð11Þ

_ } is the heat release

where A is the surface area of the fuel (m2) and Q _ 00∞ in the range of 0.01
Karlsson and Quintiere give typical values for m
2
rate per unit area (HRRPUA, kW/m ). This aligns with experiments by to 0.1 kg/m2.s and typical values for kβ between 1 to 4 m1. Assuming
28
Hu et al. where it is shown that, for a consistent mass of oil, the a fixed pan diameter of 30 cm, which is at the upper end of those
mass loss rate and maximum HRR both increase as a function of the specified by UL 300A27 and used by Hamins et al.,23,24 and adopting a
pan diameter. For the ignition of corn oil with a 7 mm oil depth, the heat of combustion of 37.4 kJ/g for corn oil from Chen et al.,20 esti-
following relationship was estimated: _ 00∞ and kβ have been determined that encompass
mated values for m
the experimental data. To be prudent, these estimated values ensure
_ ¼ 0:16D2
Q ð7Þ that the HRR is taken as an upper-level value when a given oil volume
is selected.
where D is the pan diameter (cm). This is equivalent to Equation (6) _ 00∞ are found to give
A value of kβ of 3 m1 and a value of 1 for m
2
with a value in the region of 2034 kW/m for the HRRPUA. However, the curve shown in Figure 3 when compared with the HRR for the
in using this approach with an assumed pan diameter and HRRPUA experimental data of Chen et al.,20 the open fires in Hamins et al.24
for cooking oil, the volume of the oil itself would not be estimated to and Hu et al.,28 excluding those which involved manual suppression
impact the HRR. In contrast, the quantity of oil has been shown to before the fire could reach its maximum HRR. Thus, the relationship
have a substantial impact on the potential HRR of a pan fire, particu- between HRR (Q _ in kW) and oil volume (V in m3) given by
larly in instances where the oil overflows from the pan and increases
  
the fire’s effective diameter. _ ¼ 2644 1  exp 3V 1=2
Q ð12Þ
Chen et al.20 approximated from experimental data that the
time to auto-ignition and the time to reach a maximum HRR, for a For the relationship to capture the upper HRR values for vol-
given initial fuel mass, increased in a near linear manner. That is, a umes around 200 ml and also the upper HRR of 472 kW at 4.3 L
greater quantity of oil resulted in an increased time until auto- _ 00∞ is an order of magnitude
it is recognized that the value of 1 for m
ignition occurred and an increased time to reach a maximum HRR. greater than quoted by Karlsson and Quintiere and the value for kβ is
Similarly, Chen et al. identified that a greater quantity of oil resulted at the upper end of the typical range. As such, other combinations of
in a greater maximum HRR. It was observed for larger volumes of _ 00∞ and kβ could be selected to achieve the same outcome. When
m
oil (>100 g) that “the liquid level in the pan increased until it over- considering Equation (10) there are many combinations of a depth
flowed the pan rim” and “burning liquid fuel was observed on the sides of fuel and a pan diameter for any given volume, whereas the
of the pan and in the bowl below the heating element…”. This boil- adopted correlation assumes the volume of the fuel is directly pro-
over phenomenon increased the effective diameter of the base of portional to the diameter (i.e., D / V 1=2 ), without consideration of
the fire by a factor of 1.4 to 1.8 in the experiments. This would any potential variations in the fuel depth. Furthermore, there are
indicate that the pan diameter may become a less important factor other factors that affect the development of a cooking oil fire on a
to the ultimate HRR when the oil volumes are sufficient to result in hob such as whether the oil undergoes auto-ignition or piloted igni-
boil-over. tion and whether the heating element remains energized. Hence the
29
Karlsson and Quintiere note that for diameters greater than formulation resulting in Equation (12) should be treated as being
0.2 m the burning rate of a pool fire can be expressed by semi-empirical.
Finally, radiant fraction values for olive oil and pure sunflower oil
_ 00 ¼ m
m _ 00∞ ð1  exp ðkβDÞÞ ð8Þ have not been found in the literature therefore to reflect the range of
values given by Koseki et al.19 along with the uncertainty quoted by
where _ 00∞
m is the asymptotic diameter mass loss rate and kβ is the Chen et al.20 it has been assumed that a uniform distribution from
combination of flame extinction-absorption coefficient and the mean 0.59 to 0.72 is suitably representative.
beam length corrector. Given the HRR is found from

_ ¼ Am
Q _ 00 ΔHc ð9Þ 2.6 | Occupant walking speed

and the diameter of a circular fire, for a given cylindrical container of Occupant uncongested horizontal movement speed is a function of
fuel, is related to the volume (V) and depth (δ) from age and ability. Lord et al.30 give statistical distributions, with upper
and lower limits, for the moment speed for age bands. Using the cen-
rffiffiffiffiffi
V sus data from 201131 England and Wales, the proportion of the popu-
D¼2 ð10Þ
πδ lation for the Lord et al. age bands has been determined, as shown in
Table 3.
so that D / V 1=2 when excluding depth as a variable Thus, Equation (8) The current edition of BS 9991 states that that “open-plan flat
becomes layouts should not be provided for accommodation where occupants
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL. 357

F I G U R E 3 Bounding maximum HRR

as a function of cooking oil volume.
Experimental data reproduced from Chen
et al.,20 Hamins et al.24 and Hu et al.28

TABLE 3 Assignment of walking speeds (adopted from Lord et al.30) as a function of age group from the England and Wales 2011 census31

Age group Percent of population Mean speed (m/s) Standard deviation (m/s) Lower limit (m/s) Upper limit (m/s)
Child 21.3 0.88 0.25 0.28 1.8
18–29 16.2 1.12 0.25 0.25 1.9
30–50 27.9 1.12 0.25 0.25 1.9
51+ 34.6 0.86 0.26 0.25 1.5

are not capable of independent evacuation”. Given the basis of guid- speed for “all disabled subjects” given in PD 7974–6, with a 1st quar-
ance recommendations for open plan flat design, it could be reason- tile value of 0.71 m/s.
ably argued that it does not apply to situations in traditional During the Monte Carlo simulations, the agent is first categorized
residential design where very young children may be present as disabled or not. When considering evacuation both BS EN 81–76
(e.g., families), or where elderly or mobility impaired occupants may (the design guidance for evacuation of persons with disabilities using
be present (e.g., as cohabitants). However, in its assessment of open lifts)32 and the London Plan33 recommend that, in the absence of
plan flat design, the NHBC study considered the evacuation of more detailed information, it should be assumed that 10% of the pop-
occupants with a range of population characteristics, considering ulation of the building have some form of disability and may be unable
the evacuation of adults, children and the elderly for a “typical” to use stairs. This 10% recommendation has been adopted herein. If
population. The assessment also considered the potential assisted the agent is determined as being disabled then the Lord et al. relation-
evacuation of occupants in circumstances where they are consid- ship is applied. If the agent is not disabled, then it is assigned an age
ered “dependent”. As noted in Section 1, the NHBC study directly band using the probabilities in Table 3 and the walking speed is
informed the recommendations of BS 9991, and therefore depen- derived from the corresponding Lord et al. mean and standard
dent evacuation is inherently captured when considering open plan deviation.
flat design. Given this, there needs to be consideration of the
impact of the cooking appliance location on dependent and mobility
impaired occupants when escaping. The assessment detailed herein 2.7 | Calculation input parameters and
captures the presence of these population types in the selected considerations
inputs for horizontal movement speed and selected tenability
thresholds (as noted in Section 2.3). Figure 4 illustrates the calculation procedure to find the closest dis-
For disabled occupants, Lord et al.30 give a mean of 0.77 m/s with tance between the agent travel path and the cooker hob for a given
a standard deviation of 0.24 m/s with lower and upper limits of agent demographic and fire size. The calculation procedure discretized
0.25 m/s and 1.3 m/s respectively. A value of 0.57 m/s is the worst- the travel of the agent past the hob in 0.5 s time steps. The distance
case ‘mean’ horizontal movement speed for occupants with mobility moved each time step towards and away from the closest point was
impairment provided in PD 7974–6,22 for occupants on a “walker/ derived and then the received heat flux was determined at each of
rollator”. In comparison, 1.0 m/s represents the mean horizontal travel these points and thus the thermal FED was found.
10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL.

Flow diagram for the closest separation distance between agent travel path and cooker hob calculation
FIGURE 4
358
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL. 359

The calculation was repeated by reducing the closest distance Figure 8 shows the cumulative probability functions (CDFs) for
R in Figure 1 between the cooker hob and the agent travel path was the minimum distance from the cooker hob when no maximum heat
reduced in descending steps until the thermal FED was found to reach flux is defined and where the maximum heat fluxes are capped at
a value of 1. As discussed in Section 2.3 it is possible that in some either 16.7 kW/m2 or 2.5 kW/m2. As expected, capping the maximum
cases the instantaneous heat flux received when close to the cooker heat flux leads to higher minimum distances. Since the oil volumes
hob became relatively high even though the final FED remained below used as input to the analysis are discrete values the results have been
unity. Given these higher fluxes would have exceeded the 16.7 and smoothed.
2.5 kW/m2 limits then simulations were run in which should these Examining the 1.8 m criterion in the draft of BS 9991 and
thresholds be reached then the closest distance was not reduced any TGDB,11 as discussed in Section 1, equates to approximately the 88th
further. percentile when the maximum tolerable heat flux is capped at
Applying the methodology given in PD 7974–7 34
to determine the 2.5 kW/m2 and the 98 – 99th percentile in the case of the 16.7 kW/
requisite number of Monte Carlo simulations to interrogate the 99th m2 cap or where no cap on the heat flux is considered. Although there
percentile with a coefficient of variation of ~0.05 then the analyses is no definitive measure of whether these percentiles are acceptable,
have been run for 40 000 iterations for each radiant threshold case. Hopkin et al.'s35 study on the probabilistic assessment of dwellings
found that a 95th percentile design fire broadly aligned with existing
guidance such as given by PD 7974–1.36 A 95th percentile separation
3 | RESULTS AND DISCUSSION distance with a 2.5 kW/m2 cap on the tolerable heat flux would be
c. 2.3 m, and c. 1.3 m where the cap is set to 16.7 kW/m2. Where no
Figure 5 compares the cumulative probability of the original survey cap is defined then the separation distance is 1.2 m.
data of Spearpoint and Hopkin18 and the cumulative probability curve The assessment is intended to accommodate broad design cir-
from the Monte Carlo sampling which shows a close match between cumstances and therefore applies several assumptions and restric-
the two. Figure 6 shows the cumulative probability curve from the tions. The modelling adopts a steady-state design fire for cooking oil,
Monte Carlo sampling for walking speed using the distributions as a whereas experimental data, such as from Hamins et al.,23,24 would
function of age group, given in Table 3. indicate that the maximum HRR is not sustained for a prolonged
Figure 7 shows the results from an exemplar calculation for the period and the HRR broadly follows a triangular shape. Adopting this
case when the occupant travels at the mean speed of 0.86 m/s for approach assumes that the occupant is passing by the oil fire at the
the 51+ age group, the volume of the cooking oil is 2.5 L (giving a same point in time that it is burning at its maximum HRR. On the
maximum HRR of 368 kW from Equation (12)) and the radiant frac- other hand, the sampling methodology assumes the radiant fraction is
tion is 0.65 (i.e. midway between 0.59 and 0.72). The figure shows independent of the potential maximum HRR from the type of cooking
the change in separation distance between the occupant and the hob oil although there is no specific evidence to point to this being a sig-
along with both the instantaneous and cumulative FED curves. nificant limitation.

F I G U R E 5 Cumulative probability
curve from the original survey data of
Spearpoint and Hopkin18 and the
cumulative probability from the Monte
Carlo sampling
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
360 SPEARPOINT ET AL.

F I G U R E 6 Cumulative probability of
occupant walking speeds from the Monte
Carlo sampling

F I G U R E 7 Results from an exemplar

separation distance and FED calculation

Finally, it should be noted that the analysis presented in this study this time, it is ambiguous on what basis it was included. One approach
does not account for any time to operate and manoeuvre final exit to capture the impact of door operation might be to apply the work of
door during which the occupant may still be exposed to some level of Hopkin et al.,37 which uses probabilistic calculations and experimental
thermal radiation. It is not clear whether the draft of BS 9991 includes data to estimate door open times for occupants escaping from flats.
consideration of this time, although it does provide specific recom- The door open time was estimated as a function of the door swing
mendations for the door location relative to the hob. If it does capture time, the door flow rate, and the number of occupants located in the
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SPEARPOINT ET AL. 361

F I G U R E 8 Cumulative probability
functions for minimum distance from the
cooker hob

flat. For deterministic analyses, Hopkin et al. proposed that the 95th maximum HRR of the cooking oil fire from the volume of oil in the
percentile values may be adopted, giving door open times of 10 s to pan, an assumption that the fire has attained and remains at its maxi-
19 s, depending on the number of bedrooms in the flat. Work by Son mum HRR while the occupant moves past and the application of a
38
et al. assessed the impact of door opening direction, door handle lower bound tolerance limit for skin pain. The analysis also allows a
type, and visibility conditions on occupant movement through doors maximum tolerable radiant heat flux to be considered as part of the
(for single occupants). From the experimental times, the door opening assessment of a minimum acceptable separation distance between
“delay” time ranged from 2 to 3 s for push and 4 to 5 s for pull under the hob and the evacuation path.
normal visibility conditions, and this increased to 7 to 13 s for push The analysis shows that a separation distance of 1.8 m corre-
and 13 to 16 s for pull under limited visibility conditions. These times sponds to between the 88th and 99th percentiles where there is no
could be incorporated into an analysis although it would also require a maximum tolerable heat flux or where a 2.5 kW/m2 cap is introduced.
judgement to be made on what distance the final exit door should be Should a 95th percentile separation distance be desired from the anal-
located, which is likely to be specific to a given flat arrangement. The ysis, values range from 1.2 m to 2.3 m depending on whether no max-
work of Hopkin et al. and Son et al. also does not address the time imum tolerable heat flux is used or there is a cap of 2.5 kW/m2.
required for disabled occupants to operate and manoeuvre through
the door. DATA AVAILABILITY STAT EMEN T
Data sharing not applicable - no new data generated, or the article
describes entirely theoretical research
4 | C O N CL U S I O N S
OR CID
Fixed cooking appliances in open plan living spaces should be to be Michael Spearpoint https://orcid.org/0000-0002-3632-4870
positioned in such a way that it does not prevent escape when
involved in a fire. Current and proposed guidance requires that an RE FE RE NCE S
open plan flat layout should be designed such that an escaping occu- 1. BSI. BS 9991:2011 Fire safety in the design, management and use of res-
pant should not have to approach within 1.8 m of the appliance. idential buildings. Code of practice’. BSI; 2011.
2. Hopkin C, Spearpoint M, Hopkin D, Wang Y. Fire safety design of open
This paper presents a probabilistic analysis of the thermal
plan apartments in England. Architect Eng Des Manage. 2020;16:1-20.
radiation hazard to an occupant travelling past a cooking oil fire, using 3. Fraser-Mitchell J, Williams C. Open Plan Flats: Assessing Life Safety in
statistical information on cooking oil volume usage data and occupant the Event of Fire. IHS BRE Press on Behalf of the NHBC Foundation,
walking speed as a function of population demographics as input to a 2009. https://www.nhbcfoundation.org/publication/open-plan-flats/
4. Davis C, Murphy S, Hopkin D. Open plan apartments - revisiting risks
series of FED calculations. The analysis incorporates a number of con-
in light of contemporary demands. Int Fire Profess J. 2016;18:33-35.
servatisms including a semi-empirical methodology to determine the
 10991018, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/fam.3101 by University Di Roma La Sapienza, Wiley Online Library on [02/04/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
362 SPEARPOINT ET AL.

5. BSI. BS 9991:2015 Fire Safety in the Design, Management and Use of 24. Hamins A, Kim SC, Madrzykowski D. Characterization of stovetop
Residential Buildings. Code of Practice. BSI; 2015. cooking oil fires. J Fire Sci. 2018;36(3):224-239.
6. ABCB, National Construction Code Volume one - building code of 25. Stoll AM, Greene LC. Relationship between pain and tissue damage
Australia 2019, amendment 1, Australian Building Codes Board, 2019. due to thermal radiation. J Appl Physiol. 1959;14(3):373-382.
7. Ministry of Business, Innovation & Employment C/AS2, acceptable 26. Drysdale D. An Introduction to Fire Dynamics. Third ed. John Wiley &
solution for buildings with sleeping (non institutional), for Sons, Ltd; 2011.
New Zealand building code clauses C1-C6 protection from fire. 27. UL, Outline of investigation for extinguishing system units for resi-
New Zealand Government, 2016. dential range top cooking surfaces. Underwriters Laboratories Inc.,
8. NRCC. National Building Code of Canada 2015. National Research UL 300A (Issue 3), 2006.
Council of Canada; 2015. 28. Hu Y, Chen J, Bundy M, Hamins A. The character of residential cook-
9. ICC. 2015 International Building Code. International Code Council; top fires. J Fire Sci. 2021;39:0734904121990392. doi:10.1177/
2015. 0734904121990392
10. NFPA. NFPA 101, Life Safety Code, 2018 Edition. National Fire Protec- 29. Karlsson B, Quintiere J. Enclosure Fire Dynamics. CRC Press; 2000.
tion Association; 2017. 30. Lord J, Meacham B, Moore A, Fahy R, Proulx G. Guide for evaluating
11. Government of Ireland. Building Regulations 2006, Technical Guidance the predictive capabilities of computer egress models. National Insti-
Document B (Fire Safety), Reprinted Edition 2020, Amendments & Cor- tute of Standards and Technology, Gaithersburg, MD, USA, NIST
rections Incorporated. The Department of Housing, Planning and Local GCR 06-886, 2005.
Government; 2020. 31. Office for National Statistics, Population by ethnicity - Age groups.
12. BSI, Draft British Standard BS 9991, Fire safety in the design, man- 2018. https://www.ethnicity-facts-figures.service.gov.uk/uk-
agement and use of residential buildings. 2021. population-by-ethnicity/demographics/age-groups/latest
13. Spearpoint M, Hopkin C, Hopkin D. Modelling the thermal radiation 32. BSI, BS EN 81-76. Safety rules for the construction and installation of
from kitchen hob fires. J Fire Sci. 2020;38(4):377-394. lifts. Particular applications for passengers and goods passenger lifts.
14. Wade C, Baker G, Frank K, Harrison R, Spearpoint M. B-RISK 2016 Part 76. Evacuation of persons with disabilities using lifts. BSI,
user guide and technical manual’, Building Research Association of London, 2019.
New Zealand, SR364, 2016. 33. Greater London Authority, The London Plan: The spatial development
15. McGrattan K, Hostikka S, McDermott R, Floyd J, Vanella M. Fire strategy for greater London. Greater London Authority, 2021.
dynamics simulator user’s guide. National Institute of Standards and https://www.london.gov.uk/sites/default/files/the_london_plan_
Technology, Gaithersburg, MD, NIST SP 1019, 2019. 10.6028/NIST. 2021.pdf
SP.1019 34. BSI, PD 7974-7:2019+A1:2021 Application of fire safety engineering
16. McGrattan K, Hostikka S, McDermott R, Floyd J, Vanella M. Fire principles to the design of buildings. Probabilistic risk assessment.
dynamics simulator technical reference guide volume 1: mathematical BSI, London, 2021.
model. National Institute of Standards and Technology, Gaithersburg, 35. Hopkin C, Spearpoint M, Wang Y, Hopkin D. Design fire characteris-
MD, NIST SP 1018-1, 2019. 10.6028/NIST.SP.1018 tics for probabilistic assessments of dwellings in England. Fire Technol.
17. Beyler C. Fire hazard calculations for large, open hydrocarbon fires. In 2019;56:1179-1196.
SFPE Handbook of Fire Protection Engineering, 5th Edition, Springer, 36. BSI, PD 7974–1:2019 Application of fire safety engineering principles
2016, 2591–2663. to the design of buildings. Initiation and development of fire within
18. Spearpoint MJ, Hopkin C. Household cooking oil use and its bearing the enclosure of origin (Sub-system 1). BSI, London, 2019.
on fire safety. J Fire Sci. 2021;39:265-284. 37. Hopkin C, Spearpoint M, Hopkin D, Wang Y. Estimating door open
19. Koseki H, Natsume Y, Iwata Y. Evaluation of the burning characteris- time distributions for occupants escaping from apartments. Int J High-
tics of vegetable oils in comparison with fuel and lubricating oils. Rise Build. 2021;10(1):73-83.
J Fire Sci. 2001;19:31-44. 38. Son J-Y, Bae Y-H, Kim Y-C, Oh R-S, Hong W-H, Choi J-H. Consider-
20. Chen J, Hu Y, Wang Z, et al. Why are cooktop fires so hazardous? Fire ation of the door opening process in pedestrian flow: experiments on
Saf J. 2020;120:103070. door opening direction, door handle type, and limited visibility. Sus-
21. Purser D, McAllister J. Assessment of hazards to occupants from tainability. 2020;12(20).
smoke, toxic gases, and heat. SFPE Handbook of Fire Protection Engi-
neering. 5th ed. Springer; 2016:2308-2428.
22. BSI, PD 7974-6:2019 Application of Fire Safety Engineering Princi-
ples to the Design of Buildings. Human Factors. Life Safety Strategies. How to cite this article: Spearpoint M, Hopkin C, Hopkin D. A
Occupant Evacuation, Behaviour and Condition (Sub-System 6). BSI, probabilistic thermal radiation analysis of cooking hob
London, 2019.
separation distance in open plan kitchens. Fire and Materials.
23. Hamins AP, Madrzykowski D, Kim SC, Kent J. Investigation of resi-
dential cooking fire suppression technologies. National Institute of 2023;47(3):352‐362. doi:10.1002/fam.3101
Standards and Technology, Gaithersburg, MD, NIST TN 1969, 2018.