Fire Safety Journal 63 (2014) 1–8

Contents lists available at ScienceDirect

Fire Safety Journal

journal homepage: www.elsevier.com/locate/firesaf

Modeling and analysis of water discharge trajectory with large

capacity monitor$
Tatsuya Miyashita a,n, Osami Sugawa b, Tomohiko Imamura b, Kyoko Kamiya c,
Yasuo Kawaguchi d
a
Graduate School of Science and Technology, Department of Mechanical Engineering, Tokyo University of Science, 2641 Yamazaki, Noda 278-8510, Japan
b
Department of Mechanics Systems Engineering, Tokyo University of Science, Suwa, 5000-1 Toyohira, Chino 391-0292, Japan
c
Graduate School of Environmental and Information Sciences, Yokohama National University, 79-9 Tokiwadai, Hodogaya-ku, Yokohama 240-8501, Japan
d
Department of Mechanical Engineering, Tokyo University of Science, 2641 Yamazaki, Noda 278-8510, Japan

art ic l e i nf o a b s t r a c t

Article history: In the event that a full-blown fire occurs in a large fuel oil storage tank of over 34 to 100 m in diameter
Received 19 November 2012 and of over 10 to 20 m in height, the firefighting will be carried out using a set of large capacity monitors.
Received in revised form It is important to estimate the discharge trajectory of water and/or fire-foam in order to avoid the
29 May 2013
thermal updrafts from the burning oil surface, and utilizing the entrainment around the peripheral zone
Accepted 3 September 2013
of a tank edge. A large-scale water discharge with flow of 10 to 40 kL/min, pressure of 0.6 to 0.9 MPa was
simulated using a 3D simulation model based on the Moving Particle Semi-implicit (MPS) method, and it
Keywords: was confirmed that the flying behavior and the water discharge trajectory correlates with the discharge
Full-blown fire flow and pressure. The water discharge which was described using this 3D simulation model gave good
Large capacity monitor
agreement with the recommended curved lines from the Disaster Management Committee. It is
Water discharge trajectory
suggested that this model can estimate with precision the percentage of the delivered water mass into
MPS method
Spreadsheet model a tank, and can be a useful support tool for fighting tank fires.
& 2013 The Authors. Published by Elsevier Ltd. All rights reserved.

1. Introduction
On September 2003, in Tomakomai City, Hokkaido, Japan,
full-blown fire occurred in a large fuel oil storage tank with floating
roof (42.7 m in diameter and 24.4 m in height) caused by the
sloshing phenomenon due to the long-term oscillation of the
Tokachi offshore earthquake and static electricity. The fire con-
tinued burning for about 44 h and the 26,000 kL naphtha burned
out. The fire triggered partial revisions of the Fire Service Act and
the Act on the Prevention of Disaster in Petroleum Industrial
Complexes and Other Petroleum Facilities which has obligated to
install a set of large capacity monitors to any petroleum tank of
over 34 to 100 m in diameter since November 2008 [1]. The large
capacity monitor can discharge fire-foam over 10 to 40 kL/min,
which is ten times bigger capability than that of an aerial platform,
and can deliver fire-foam into a tank located 50 to 100 m away.
Also, the monitor have a proven record of extinguishing in 65 min
a full-blown tank fire of 82 m in diameter and 10 m in height in
☆
This is an open-access article distributed under the terms of the Creative
Commons Attribution-NonCommercial-No Derivative Works License, which per-
mits non-commercial use, distribution, and reproduction in any medium, provided
the original author and source are credited.
n
Corresponding author. Tel.: þ 81 266 73 1201; fax: þ 81 266 73 1230.
E-mail address: tatsuya-mi@hotmail.co.jp (T. Miyashita).
the U.S. in 2001 [2]. However, Japan's petroleum tanks are
generally built taller than 20 m in height, and half of the fire-
foam discharged from monitor is blocked by the tank walls and
drops to the ground. Moreover, the monitor cannot shift easily
from its first installed position due to its large-sized discharge
system. It is necessary to install the monitor in a suitable position
in order to deliver the fire-foam into the tank efficiently by
predicting the discharge trajectory—that is, the maximum and/or
minimum range and height and also the width of the area where
the water is most concentrated, called the “footprint” or “landing
zone”. It is difficult to predict discharge trajectories with accuracy,
because they depend not only on the discharge flow, pressure and
angle (discharge conditions), but also on the wind velocity and
direction (wind conditions), due to the discharged water and/or
fire-foam being easily blown around by the wind. Therefore,
development of support tools which can estimate the discharge
characteristics taking account of the effects of wind has been
demanded in order to conduct efficient firefighting of tank fires.
To predict the delivery of the fire-foam onto the burning oil
surface considering the buoyancy and entrainment at the around
the tank edge which are induced by the tank fire is very important
for efficient firefighting. Therefore, we divided the analysis of the
problem into two regions, that's are (a) discharge performance
of large capacity foam monitor, and (b) the spread behavior of
fire-foam on the oil tank surface. This paper deals with the water

0379-7112/$ - see front matter & 2013 The Authors. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.firesaf.2013.09.028
2 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8

Nomenclature P discharge pressure [MPa]

P′ dimensionless discharge pressure [  ]
A amplitude Q discharge flow [kL/min]
CD coefficient of air resistance [ ] Q′ dimensionless discharge flow [  ]
Cw reduction coefficient of air resistance [  ] r distance between particles [m]
d particle diameter [m] re interaction radius [m]
dm mean particle diameter [m] Rm mass fraction of particles [  ]
f frequency [Hz] S projected sectional area [m2]
- t time [s]
F external force [m/s2] -
u velocity of particle [m/s]
g gravitational acceleration [m/s2] -
h height of water discharge trajectory [m] U wind velocity [m/s]
h0 nozzle height [m] U′ dimensionless wind velocity [  ]
m mass of particle [kg] x distance from nozzle [m]
M proportion (percentage) of water mass delivered into θ discharge angle [rad]
a tank [  ] ρ density [kg/m3]
n uniformity number [  ] ρair density of air [kg/m3]
Ni particle number density [  ]

and/or fire-foam discharge trajectory as the primary problem, but
the second item (b) including the effect due to the buoyancy is not
concerned. The authors constructed previously a three-dimensional
simulation model [3,4] based on the Moving Particle Semi-implicit
(MPS) method [5], and verified the validity of the simulation results
through comparison with medium-scale water and fire-foam dis-
charge experiments with flow of 1.5 kL/min [6,7]. The purpose of this
study is to simulate a large scale water discharge with flow of more
than 10 kL/min as a case study used for the 3D simulation model,
and to make clear the flying behavior and water discharge trajectory
correlate with the discharge conditions. Furthermore, this paper
presents a new model which can predict simply the trajectory of
straight water discharge based on 3D simulation results, called the
“spreadsheet model”. In order to serve as a utility tool which can
support the planning of disaster prevention, the discharge trajectory
described using the spreadsheet model is to be evaluated by
comparing it with a recommended curved line.
2. Simulation
2.1. Simulation model
The 3D simulation model of water discharge was constructed
based on the MPS method, which was developed for incompres-
sible fluid dynamics. MPS method is one of the particle methods or
meshless methods, as well as Smoothed Particle Hydrodynamics
(SPH) method. The MPS method has been applied to free surface
flows and multi-physics problems: for example, wave breaking,
sloshing, shipping water, micro flow and solid dynamics. In the
MPS method, the governing equations are the equation of con-
tinuity and the Navier–Stokes equation, and particle interaction
models are prepared with respect to differential operators: gra-
dient, divergence and Laplacian. All the interactions are limited
among the neighboring particles within a finite distance. Mass,
momentum and energy conservation equations, which involve
such differential operators, are transformed to particle interaction
equations. The particles move in Lagrangian description, so that
the convection terms are not calculated. Thus, numerical diffusion
does not occur. This is preferable to keep the interfaces clear.
Incompressibility is calculated by a semi-implicit algorithm where
the pressure field is implicitly solved using the Poisson equation,
while the other terms are explicitly calculated [8,9].
For analysis of the flying behavior in a straight water discharge,
surface tension is calculated as force that operates among neighboring
particles in the MPS method. Droplet break-up behavior is classified by
the Weber number W e , and viscosity effects on the droplet breakup
are correlated with the Ohnesorge number On [9]. However, these
calculations need to employ the high computation cost with the large
size calculation domain such as water discharge of large capacity
monitor. Therefore, we constructed new models – a water stream
model, a water mass break-up model and a wind turbulence model –
instead of the previous MPS method. This model focused on the water
particle movement and interaction between them but not on the air
flow in order to reduce the process of numerical simulation, and
considering only fluid flow. Thus, the water particles are affected by
-
the air resistance as external force F of Eq. (1). However, the particles
in the central portion of the water mass (assembly of particles) are
only slightly affected by the air resistance, because only the particles of
surface portions of water mass are affected by air resistance. Therefore,
we defined a reduction coefficient of air resistance C w which is
determined as Eq. (2). N i of Eq. (3) is the particle number density,
can be calculated by the position and velocity of other particles
(neighboring particle) in the hemispherical interaction radius of each
fluid particles, shown in Fig. 1. This model is applied the weight
function (function for searching neighboring particle) of MPS method
where the mass of particle m, coefficient of air resistance C D ð ¼ 0:5Þ,
-
density of air ρair , projected sectional area S, velocity of particle u ,
-
wind velocity U , distance between particles r and interaction radius r e
(is 12 times as large as the diameter of particle).
! 1 ! ! ! !
F ¼ C w C D ρair Sð u  U Þj u  U j ð1Þ
2m
C w ¼ 1=Ni ðN i a 0Þ ð2Þ
Fig. 1. Definition of a particle interaction model for a calculation of air resistance
considering to the other particles.
 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8 3

!
re uj
Ni ¼ ∑ ! ðr o r e Þ ð3Þ 2
j a i10r ui

Wind velocity [m/sec]

1

0
When the particle number density of the surrounding fluid
particle became zero (N i ¼ 0), the water mass is counted that -1
broke up to water droplets. Then the diameter of water droplets
broke from water mass is applied the Rosin–Rammler distribution -2
[10,11], the breakup behavior of water mass is approximated by
this model. The Rosin–Rammler distribution shown in Eq. (4) is
influenced by two coefficients of mean particle diameter dm and 60

Wind direction [deg.]

uniformity number n. Also, these coefficients were obtained by
the equation for the balance between the pressure due to air 30
resistance and the reactive force due to the surface tension, as
0
dm ¼ 2:89 and n ¼ 2:14.

 1=n -30
lnð1 Rm Þ
d¼   dm ð4Þ
0:693 -60
0 30 60 90 120
where the particle diameter d, mass fraction of particles Rm Time [sec]
(random numbers of 0.01 to 0.99). Fig. 2. Comparison between the composite function of several sine functions based
The wind model is that reproduced the change of wind velocity on FFT and the measured data with (a) wind velocity and (b) wind direction.
and direction which have been measured in the former field
experiments of medium-scale water discharge experiments with
flow of 1.5 kL/min [6,7]. In the field experiments, the wind velocity 2.2. Condition of simulation
and direction were measured for 2 min at a height of 10 m from
the ground. The measurement data of wind velocity and direction The straight water discharge trajectories used with a large-
include the variations in turbulent wind velocity (  2.0 to 2.0 m/s) capacity monitor vary extremely depending on discharge flow,
and wind direction (  601 to 601). Thus, the composite sine pressure, angle and wind. In order to analyze the flying behavior
functions for wind velocity and direction were created using Fast of the discharged water and to construct a spreadsheet model for
Fourier Transforms (FFT) respectively. The composite function is the water discharge trajectory, large-scale water discharges in 3D
composed of several sine functions for the parameters of ampli- simulation were conducted under the following conditions: dis-
tude and frequency, as shown in Eq. (5). charge flow set to 10, 20, 30 and 40 kL/min, pressure set to 0.6, 0.7,
0.8 and 0.9 MPa (initial velocity being 34.6, 37.4, 40.0 and 42.4 m/s
f ðtÞ ¼ A1 sin ð2π f 1 tÞ þA2 sin ð2π f 2 tÞ þ ⋯ þ An sin ð2π f n tÞ ð5Þ respectively), angle set to 301, 351, 401, 451 and 501. The maximum
range, the maximum height, the footprint width and the distribu-
tion of the water particles that dropped on the ground were
The amplitude and frequency are determined based on FFT of the measured for each set of discharge conditions. Regarding the wind
wind velocity and direction. Table 1 shows representative combina- velocity and direction, a 2.0 m/s cross wind with the turbulence
tions of amplitude and frequency for the composite function, and was applied to the respective conditions. This was because the
Fig. 2(a) and (b) shows the comparison of the composite function and mean wind value in the large-scale water discharge experiments
the measured data, with the wind velocity and direction respectively. was 2.0 m/s [13] and also in order to analyze the way in which the
The maximum frequency and the cycle are simulated well, with less water distribution was affected by the cross wind. The turbulence
than 10% difference from the measured data. Note that the model of the wind velocity and direction applied for the composite
takes account of the flow of the boundary layer under the effects of function shown in Fig. 1. In addition, simulation of water discharge
frictional drag with the ground surface [12], however, did not in following and adverse winds (adverse wind being represented
consider the effect of the tank itself with the calculation of the wind by a negative quantity), with wind velocity set to 0.0, 2.0, 4.0,
velocity and direction and wind turbulence since the purpose of this 6.0 and 8.0 m/s, was conducted in order to construct the spread-
study was to analyze only the water discharge characteristics. The sheet model. In this case, wind turbulence was not considered,
turbulence of wind velocity and direction shows a cyclical wind flow since the purpose was to analyze how the water discharge
at the open space in this paper. trajectory was affected by a uniform lateral wind.
Fig. 3 presents graphics illustrating the 3D simulation of water
discharge with flow of 20 kL/min, pressure of 0.7 MPa, angle of
451 and a 2.0 m/s cross wind. The columnar structures in figure
Table 1
Combinations of amplitude: A and frequency: f used as coefficients for the
represent a virtual tank of diameter 80 m and height 25 m which
composite function. was created using wall particles. The monitor's nozzle was used in
the same way as the non-aspirating nozzle for the experiments,
Wind velocity Wind direction and the nozzle height was set to 2.0 m. The simulated discharge
duration was set to 30 s and the particle size (mesh size in the
A1 0.50 f1 0.02 A1 10.76 f1 0.02
A2 0.23 f2 0.03 A2 11.96 f2 0.05 Finite Element Method, FEM) was set to 0.04 m. Note that this
A3 0.52 f3 0.05 A3 7.68 f3 0.07 simulation did not consider the thermal updrafts from the burning
A4 0.27 f4 0.08 A4 6.19 f4 0.10 oil surface, the spread behavior of the fire-foam on the oil surface
A5 0.17 f5 0.09 A5 7.81 f5 0.13 or the foaming phenomenon of the foam solution, since the
A6 0.13 f6 0.14 A6 5.14 f6 0.15
A7 5.74 f7 0.20
purpose of this study was to analyze only the water discharge
characteristics.
4 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8

3. Results and discussions the measured data from the large-scale water discharge experi-
ments that have been conducted in 2009 at Kiire, Kagoshima,
3.1. Validity of simulation results Japan [14]. Fig. 4 shows a comparison of images of the experimental
and simulated water discharges with flow of 20 kL/min, pressure
To evaluate the validity of the 3D simulations of water of 0.7 MPa, angle of 351 and 2.0 m/s following wind. Fig. 4(b)
discharge trajectory, the simulation results were compared with shows a projection of the 3D simulated water discharge results
onto a 2D plane, giving an instantaneous image of the state ten
second after discharge. Since the fine water droplets are hardly
captured by a video camera such as used for Fig. 4(a), the simu-
Height [m]

lated water droplets are shown depending on the diameter of the

particles. That is, the simulated trajectory was expressed only by
the particles which have a diameter of more than 0.007 mφ. The
flying behavior, in which the water droplets repeatedly combine
and break up, forming a creased pattern of stripes, and falling
Width [m]

cascade-like (resembling a curtain of water) at close to the max-

imum height, was simulated well. The curved lines for the simulated
discharge trajectory approximate to the experiments, and the
20 40 60 80 100 120
maximum range and maximum height are simulated well, with less
Distance from a nozzle [m] than 10% difference.
Fig. 3. Graphics showing the side view of the 3D water discharge simulation using Table 2 shows the measured data and simulation data for the
the MPS method. maximum range, the maximum height and the footprint width
under each set of discharge conditions, along with the deviations,
which were calculated by comparing the measured data to the simu-
lation data. The scatter of the experimental data in the maximum
range and maximum height are 10 to 20%. In the standard deviation,
maximum range is 10 to 20 m, maximum height is 5 to 10 m. Also,
the reproducibility of width was no good. It is assumed that these
reasons are difference of wind condition, discharge condition (flow
rate, pressure and angle) and measurement due to the difficulty
of large scale experiments. For this reason, this simulation model
is verified if the difference between simulated data and measured
data is less than 20%. The maximum range and height are simu-
lated well with less than 0.5 to 10% difference, and so the validity
of simulation results is verified. However, the deviation in the
width shows both 5 to 7% and 30 to 50% under different discharge
conditions. It is assumed that large deviation in the width is
caused by the difference in wind conditions – wind velocity,
Fig. 4. Comparison of side views of (a) experimental and (b) simulated water direction and turbulence – between simulation and experiment.
discharge trajectories with flow of 20 kL/min, pressure of 0.7 MPa, angle of 351 and The width is easily influenced by the wind, and the waves depend
2.0 m/s following wind. on the wind turbulence. Therefore, in order to estimate the width

Table 2
Comparison of data for measured and simulated water discharge trajectories.

Flow (kL/min) Pressure (MPa) Angle (deg.) Range (m) Dev.(%) Height (m) Dev. (%) Width (m) Dev. (%)

20 0.7 35 Exp. 102 6.7 26 7.7 9 32.2

Sim. 108.8 24 6.1
40 Exp. 101 8.5 32 8.6 11 32.6
Sim. 109.6 29.2 7.4
45 Exp. 112 3.5 34 1.8 6 44.6
Sim. 108 34.6 8.7
0.8 40 Exp. 119 0.5 32 2 8 7.3
Sim. 118.3 32.6 8.6
0.9 40 Exp. 123 2.2 32 11.9 9 5.9
Sim. 125.7 35.8 9.5
45 Exp. 125 3 40 5.8 7 57.1
Sim. 121.3 42.3 11

30 0.7 40 Exp. 119 2.6 29 1.9 9 15.1

Sim. 115.9 29.5 7.6
45 Exp. 124 6.4 35 0 9 4.8
Sim. 116.1 35 9.4
0.9 40 Exp. 141 3.5 37 1.9 8 19.4
Sim. 136.1 36.3 9.5

40 0.7 45 Exp. 140 14.7 39 9.9 7 30.3

Sim. 119.4 35.1 9.1
0.8 45 Exp. 140 6 41 3.9 9 11.3
Sim. 131.7 39.4 10
0.9 45 Exp. 150 5.4 48 9.5 8 38.6
Sim. 142 43.4 11.1
 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8 5

30 30

Height [m]

Height [m]
20 20
10 10

20 40 60 80 100 120 140 20 40 60 80 100 120 140

Distance from a nozzle [m] Distance from a nozzle [m]

30

Height [m]
30
20
Height [m]

20
10
10

20 40 60 80 100 120 140 20 40 60 80 100 120 140

Distance from a nozzle [m] Distance from a nozzle [m]

Fig. 5. Water discharge trajectories (left side) and water distribution contours (right side) for 3D simulation results with angle of 351 and 2.0 m/s cross wind under each set
of discharge conditions. (a) Flow 20,000 L/min, pressure 0.7 MPa, angle 351, 2.0 m/s cross wind, (b) Flow 40,000 L/min, pressure 0.7 MPa, angle 351, 2.0 m/s cross wind,
(c) Flow 20,000 L/min, pressure 0.9 MPa, angle 351, 2.0 m/s cross wind. (d) Flow 40,000 L/min, pressure 0.9 MPa, angle 351, 2.0 m/s cross wind.

with accuracy, the simulation conditions should be set to give the
same wind data and turbulence as the experiments.
3.2. Analysis of discharge characteristics
The left side in Fig. 5 shows 2D projections of the water
discharge trajectories, and the right side shows contour graphs
of the distribution of the water particles that dropped on the
ground. The angle in all these graphics is set at 351 and they are
representative of the various simulation results. In order to discuss
the discharge flow and/or pressure dependence of the discharge
trajectory and water distribution, the simulation results for flow
of 20 kL/min and pressure of 0.7 MPa as shown in Fig. 5(a) were
established as a comparative standard. Fig. 5(b) is the case where
only the flow was increased, to 40 kL/min, Fig. 5(c) is the case
where only the pressure was increased, to 0.9 MPa, and Fig. 5(d) is
the case where both flow and pressure were increased from the
standard values.
Let us discuss the analysis of the discharge trajectories in the
2D projection. In the case where only the flow was increased, the
maximum range becomes longer by 7% from the standard because
the water mass dose not readily disperse and the air resistance
does not readily affect the discharge trajectory. Meanwhile, in the
case where only the pressure was increased, the maximum range
becomes longer by 16% from the standard, but the water mass
disperses readily due to the air resistance. In the case where both
parameters are increased, the maximum range becomes longer by
29% from the standard and the dispersion of the water mass is
smaller than in the case where only the pressure was increased.
In the water distribution contour graphs, it is confirmed that the
footprint is an almost elliptical shape under each set of conditions.
It is offset to one side of the center line due to the effect of the
2.0 m/s cross wind. The distortion of the water distribution
becomes small in Fig. 5(b), because the water mass, concentrated
by the increased flow, is less susceptible to wind. In Fig. 5(c),
however, the distortion is large and the contour interval is large.
This suggests that the dispersion of the water mass becomes large
with increasing pressure, and the water mass is susceptible to
the wind. The above analysis results revealed that the discharge
characteristics develop the following tendencies depending on
flow and pressure. First, increased flow results in the range staying
unchanged and the discharge trajectory being less susceptible to
the wind. Second, increased pressure results in the range becom-
ing long and the discharge trajectory being susceptible to wind.
4. Proposal of spreadsheet model
4.1. Spreadsheet model
This section proposes a new model which can predict simply
the trajectory of straight water discharge. A histogram of the
particles' height coordinates was constructed at intervals of 5 m
along discharge axis using the simulated particles' x-coordinates
(discharge direction) and z-coordinates (height direction). The y-
coordinates, which represent the width direction, were projected
onto a 2D plane in order to construct a 2D spreadsheet model.
A frequency distribution table classified the particle coordinates
as abundance ratios of fluid particles at each distance from the
nozzle, and the trajectories for each mass delivery percentage
were described by plotting the abundance ratios using the poly-
nomial approximation technique, as shown in Fig. 5 The trajectories
6 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8

for the mass delivery percentage represent the proportion of particles Eqs. (8)–(16).
projected along the discharge axis per discharge flow, with the
h ¼ a0 G3 ðUÞG3 ðQ ÞG3 ðθÞM  n x3 þb0 G2 ðUÞG2 ðQ ÞG2 ðθÞG2 ðPÞx2
trajectory of 100% representing the maximum range and maximum
height. The rectangle in Fig. 6 is the virtual tank image with diameter þc0 G1 ðθÞx þ d0 ð7Þ
50 m and height 20 m.
Assuming that the water discharge trajectories, which vary n ¼ 1:2 þ 1:9  10  2 U′ þ2:1  10  3 U′2 ð8Þ
due to the effects of the parameters flow, pressure, angle, wind
and mass delivery percentage, can be approximate by a third- G3 ðUÞ ¼ 1:0 0:15 U′þ 4:9  10  3 U′2 ð9Þ
order function, the discharge trajectory will depend on variation in
the coefficients in the various terms of the third-order function. G2 ðUÞ ¼ 1:0 þ5:0  10  2 U′ ð10Þ
Therefore, the variation in the coefficients with the effects of the 0
 0:8
parameters was functionalized, based on the coefficients of the G3 ðQ Þ ¼ Q ð11Þ
various terms, into an inclined projection equation ignoring air
G2 ðQ Þ ¼ 1:0  expð  ðQ ′ þ 0:1Þ=5Þ ð12Þ
resistance, shown in Eq. (6). These coefficients are representative
values in order to describe a parabola, and it is assumed that
G3 ðθÞ ¼ 1:0  2:86 θ þ 3:0 θ
2
coefficients of trajectory affected by the parameters would be ð13Þ
derivable from the representative coefficients.
G2 ðθÞ ¼ ð cos θÞ  2 ð14Þ
ρg
h¼  x2 þ tan θx þ h0 ð6Þ G1 ðθÞ ¼ tan θ ð15Þ
4 P cos 2 θ

where ρ is the density of water (1000 kg/m3) and g is the G2 ðPÞ ¼ P′  1 ð16Þ
gravitational acceleration (9.81 m/s2).
(coefficients of the third-order function are:)
In the coefficient of the x2 term in Eq. (6), pressure P was used
as a variable because the initial velocity u is estimated from a0 ¼  7:0  10  4 ½1=m2 
the discharge pressure using Bernoulli's principle: u ¼ ð2 P=ρÞ1=2 . b0 ¼ 2:0  10  3 ½1=m
When the representative coefficients of the term x3 , x2 , x1 and x0 c0 ¼ 1:0 ½  
are set to a0 , b0 , c0 and d0 respectively, it is suggested that b0 and
d0 ¼ h0 ½m
c0 haves pressure P and/or angle θ as variable. There is no x3 term
among the representative values in Eq. (6), so coefficient a0 is set where M is the proportion (percentage) of mass delivered (0.1 to 1,
to zero and will increase depending on the effects of the para- in other words 10 to 100%), U is the wind velocity (  8 to 8 m/s),
meters. In view of the foregoing, we present the spreadsheet θ is the discharge angle (0.52 to 0.87 rad, in other words 30
model for water discharge trajectory shown in Eq. (7), which to 50 deg.), P is the discharge pressure (0.6 to 0.9 MPa), Q is the
has the functions for the various parameters that are expressed as discharge flow (10 to 40 kL/min), h is the height of the water

60%
40% 80% 30
100%
20% Height [m]

20

10

0 20 40 60 80 100 120
Distance from a nozzle [m]

Fig. 6. Trajectories of discharged water for each delivery percentage with flow of 20,000 L/min, pressure of 0.7 MPa and angle of 401, without wind. The rectangle is the
virtual tank with diameter of 50 m and height of 20 m.

Users input parameters

Trajectory is
described a graph

Virtual tank image

Fig. 7. Example of spreadsheet model used in Microsoft Office Excel.

 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8 7

into the tank, because the definition of the lower line is not clear
50
Flow: 25 kL/min
physically.
40 Pressure: 0.7 MPa In order to serve as a utility tool which can support the planning
Angle: 45 deg.
Height [m]

30 2.1 m/sec following wind of disaster prevention, the discharge trajectory described using
the spreadsheet model is to be evaluated by comparing it with
20 a recommended curved line. Fig. 8 shows these water discharge
100%
10
60% trajectories: (a) with flow of 25 kL/min, pressure of 0.7 MPa, angle of
70% 451 and 2.1 m/s following wind, and (b) with flow of 30 kL/min,
0
0 20 40 60 80 100 120 140 pressure of 0.7 MPa, angle of 401 and 1.5 m/s adverse wind. The
Distance from a nozzle [m] trajectory for mass delivery percentage 100% in the spreadsheet
model is similar to the recommended upper lines, and the trajec-
tories for 60% and 70% are similar to the recommended lower lines.
50 With other discharge conditions, both derived discharge trajectories
Mass delivery percentage 60% Mass delivery
Flow: percentage
30 kL/min 70% showed good agreement. Therefore, the validity of the spreadsheet
40 Mass delivery percentage 100% Pressure: 0.7 MPa
Angle: 40 deg. model was verified. This means that the water discharge trajectory
Height [m]

30 1.5 m/sec adverse wind can be obtained with precision when the parameters are set within
20
the simulation conditions. Furthermore, it is suggested that the foot-
100%
print is the area where the water is concentrated at 30 to 40% of the
60%
10
70%
discharge flow, in view of the similarity between the recommended
0 lower lines and the trajectories for mass delivery percentage 60% and
0 20 40 60 80 100 120 140
70%. In the other results of comparison, there is a slight apparent
Distance from a nozzle [m] difference in the maximum ranges, but this is due to a difference
Recommended upper line Recommended lower line between the measured mean wind data and the wind data captured
Mass delivery percentage 60% Mass delivery percentage 70% at the moment when the photograph was taken. If a 0.5 to 1.0 m/s
Mass delivery percentage 100%
following wind is set in the wind conditions, the trajectory with the
Fig. 8. Good agreement between the recommended curved lines and the spreadsheet spreadsheet model will approximate to the recommended lines.
model lines for water discharge trajectories with (a) flow of 25,000 L/min, pressure of
0.7 MPa, angle of 451 and 2.1 m/s following wind, (b) flow of 30,000 L/min, pressure
of 0.7 MPa, angle of 401 and 1.5 m/s adverse wind.
5. Conclusion

discharge trajectory (m), x is the distance from the nozzle, and h0
is the nozzle height (usual nozzle height is set to 2.0 m).
The mass flow Q , pressure P, and wind velocity U were set to
the dimensionless quantities of Q ′, P′, U′, which were divided by
the each unit of Q 0 ¼1.0 kL/min, P 0 ¼1.0 MPa and U 0 ¼1.0 m/s,
respectively. This model is such that the height h for the distance
from the nozzle x is calculated in proportion to the variation in the
parameters, and the discharge trajectory can be readily described
using a spreadsheet such as Microsoft Office Excel.
Fig. 7 shows an example of spreadsheet model used in Micro-
soft Office Excel. First, the users input parameters to describe the
discharge trajectory, and then the discharge trajectory is described
in the right graph automatically. Second, the users input a tank
position (distance from nozzle to tank), diameter and height, and
then virtual tank image is described in the graph automatically.
Finally, the users calculate the percentage of mass delivered into
the tank. In this example, it can be estimated that the percentage
of the mass that is delivered into the virtual tank is 60%, because
the trajectories for the mass delivery percentages of 40% and 100%
intersect at the upper tank surface.
A large-scale water discharge with flow rate of more than
10 kL/min was simulated using a 3D simulation model based on
the MPS method. It was confirmed that the detailed characteristics
of the flying behavior and water discharge trajectory depend
on the discharge flow and/or pressure. We present a spreadsheet
model for water discharge trajectory, which uses as parameters
the discharge flow, pressure, angle, wind velocity, and mass
delivery percentage. Because the trajectory with the spreadsheet
model is similar to the recommended curved lines, the water
discharge trajectory can be obtained with precision when the
parameters are set within the simulation conditions. Furthermore,
it is suggested that the footprint is the area where the water is
concentrated at 30 to 40% of the discharge flow, for the reason that
the recommended lower line is defined as a trajectory with mass
delivery percentage 60 to 70%. Using this spreadsheet model as a
support tool will enable disaster prevention planning that takes
account of the effects of wind, etc., so that optimal discharge
conditions for fighting fuel oil tank fires can be proposed.
Acknowledgements

4.2. Evaluation of spreadsheet model
An empirical formula, based on full-sized water discharge
experiments, for the water discharge trajectory of a large-
capacity monitor has been proposed at the committee on disaster
management of petrochemical complexes [15]. This formula is
constructed by approximating a side-view photograph of a water
discharge, using a higher-order function, and the curves described
by the formula are called recommended curved lines. These lines
are of two types: one is an upper line along the maximum range
and height and the other one is a lower line along the minimum
range and height. And, this empirical formula is restricted to the
experimental conditions shown in a photograph. Moreover, it
cannot estimate the percentage of the water mass that is delivered
This study was facilitated by the cooperation of the FUKADA
KOGYO Co., Ltd and the Hazardous Materials Safety Techniques
Association (Kikenbutsu hoan gijutsu kyokai, khk).
References
[1] Act on the Prevention of Disaster in Petroleum Industrial Complexes and Other
Petroleum Facilities, Capture 3 Disaster Prevention of the Business Operator,
Article 8 to 13.
[2] Henry Person, Anders Lonnermark, Tank Fires Review of the Incidents 1951–
2003, SP Technical Research Institute of Sweden Fire Technology, Fire Protec-
tion (2004) B14.
[3] T. Miyashita, O. Sugawa, Y. Wada, Y. Kawaguchi, Three-dimensional simulation
model for water and fire-foam discharge using MPS method, J. Jpn. Soc. Saf.
Eng., Jpn. Soc. Saf.Eng. (JSSE) 51 (2) (2012) 96–105.
8 T. Miyashita et al. / Fire Safety Journal 63 (2014) 1–8
[4] T. Miyashita, O. Sugawa, Y. Wada, R. Ishikawa, Y. Kawaguchi, Development of
two-dimensional simple simulation model and evaluation of discharge ability
for water discharge of firefighting, Bull. Jpn. Assoc. Fire Sci. Eng. (JAFSE) 62 (1)
(2012) 13–19.
[5] S. Koshizuka, MPS Method, JSCES, MARUZEN Publishing Corp (2005) 63–66
(pp. 1–28).
[6] T. Miyashita, R. Ishikawa, O. Sugawa, T. Imamura, K. Kamiya, Y. Kawaguchi,
2011. Development of simulation model for water and fire-foam discharge
using MPS method, in: Proceedings of Asia Pacific Symposium on Safety
(APSS2011), The Korean Society of Safety, pp. 254–257.
[7] R. Ishikawa, T. Miyashita, O. Sugawa, T. Imamura, K. Kamiya, 2011. Experi-
mental study of the water discharge behavior for firefighting, in: Proceedings
of Asia Pacific Symposium on Safety (APSS2011), The Korean Society of Safety,
pp. 340–343.
[8] S. Koshizuka, Moving particle semi-implicit (MPS) method—a particle method
for fluid and solid dynamics, IACM Expressions 18 (2005) 4–9.
[9] K. Nomura, S. Koshizuka, Y. Oka, H. Obata, Numerical analysis of droplet breakup
behavior using particle method, J. Nucl. Sci. Technol. 38 (12) (2001) 1057–1064.
[10] D.T. Sheppard, Spray Characteristics of Fire Sprinkler, NIST (2002) 33–35.
[11] Y. Mizutani, Combustion Engineering (Written in Japanese), MORIKITA
Publishing Co., Ltd (1989) 139–140.
[12] T. Miyashita, O. Sugawa, R. Ishikawa, Y. Wada, Y. Kawaguchi, 2010. Simulation
of fire foam flying behavior using MPS method, in: Proceedings of Eighth Asia-
Oceania Symposium on Fire Safety Technology (AOSFST), Paper 3–3.
[13] M. Fujimoto, H. Hagura, Modern Architecture Engineering Fire Disaster
(Written in Japanese), Ohmsha, Ltd (1981) 101.
[14] Institute for Fire Safety and Disaster Preparedness, 2009. Interim Report of
Risk Assessment for Oil Industrial Complex in Chiba (written in Japanese),
Weather Conditions of Target Area, Reference Materials 2–2.
[15] Disaster Management Committee of a Petrochemical Complex in Localities of
Japan, 2009. Performance Test Report of a Large Capacity Foam Monitor
(Written in Japanese), Image Analytical Results, Materials 2–2.