Beruflich Dokumente
Kultur Dokumente
Ms. Alice Atwood (1), Ms. Eva K. Friis (2), Dr. John F. Moxnes (3),
(1)
(2)
Nammo Raufoss AS
P.O: Box 162,N-2831 Raufoss, NORWAY
Telephone+4761153609
E-mail: eva.friis@nammo.com
(3)
Telephone+4763807514
E-mail: john-f.moxnes@ffLno
ABSTRACT
This article presents a mathematical model of combustion which can be used in numerical
codes. We also present experimental results where a convective burning front moves behind
the pressure front during combustion in long tubes. The mathematical model is visualized in
analytical studies of burning in a closed bomb.
The detailed kinetic cherrlical reaction mechanisms which occur during combustion of
energetic materials are complicated and difficult to model from first principles. By studying
the combustion mechanism in a variety of experiments, we have constructed a mathematical
model based on average quantities, which can then be implemented (as a subroutine) into
more general numerical computer codes (Autodyn, Hi Dyna and Hi Nike). The combustion
model can be linked with granular mechanical models based on the theory of continuum
mechanics. The model enables simulation of the velocity and the pressure in a combustion
front, and the model can thus be used to study the corresponding mechanical effect on
surrounding materials.
INTRODUCTION
This article presents a mathematical model of combustion which can be used in numerical
codes. During combustion of energetic materials the velocity of the reaction front usually
moves behind the mechanical pressure front originating from the burning solid material. The
detailed kinetic chemical reaction mechanisms which occur during combustion are
complicated and difficult to model from first principles. By studying combustion both in a
closed bomb and in a long tube, we have constructed a mathematical model based on
average quantities, which can be implemented into more general numerical computer codes
(Le. Autodyn, Dyna and Nike). The combustion model can be linked with the granular M-O
model [1] and [2], which is based on the theory of continuum mechanics.
The general idea of the burning model is to facilitate simulation of burning of energetic
materials. Traditionally this is achieved by using general two-phase hydro code models. Our
approach is different. By modeling the reaction kinetics on the meso scale as a subroutine in
finite element codes, the burning of energetic materials can be simulated without using
general two-phase models. In the subroutine each numerical cell may contain both solid and
gaseous material. The subroutine calculates the different solid and gaseous masses and
densities as well as the burn rate of the solid particle surface, the burn fraction, temperature,
gas pressures and the propagation of the burning front. The average pressure in a cell as a
result of the combustion is used as input to the general code to calculate the mechanical
effect of the combustion on surrounding cells. The model can thereby be used as a tool for
studying the combustion of energetic materials and the corresponding mechanical effect on
surrounding materials, such as fragmentation.
A short outline of the paper: First the mathematical burning model is presented. Next section
presents analytical studies of burning in a closed bomb. Finally, the last section presents
theory and experimental results of the general situation where a convective burning front
moves behind the pressure front during combustion in long tubes.
Each numerical cell may contain both solid and gaseous material. This is illustrated in figure
1.
Numerical cell
with volume V
Solid
Gas'-+-WfIH.
The following equation apply for the mass of the solid and the mass of the gas:
de!
+ M g (t)
Ms(t)
= M = Ms(tO) +M /to )
(1 )
where,
to
The different volumes and densities are defined in equation (2) - (7):
de!
= Vs + Vg
de! M (t)
Ps(t)
s
: density of the solid (particles)
Vs(t)
de!
M (t)
Pg (t) =
_
de!
Ps (t) =
Pg (t)
(3)
(4)
Vg(t)
M (t)
s
(5)
(6)
Vet)
de! M
(2)
(t)
g
Vet)
"
If the cell is subjected to high external pressure the solid particles will get compacted and
thus the density Ps (t) will increase. The following relation is proposed in order to model this
phenomenon:
mod
Ps(t)
Ps(tO)+
~ (Ps (t)) _
(_
) Ps(t)
K Ps(t)
(7)
where Ps (to) is the initial density of the particles, ~ (Ps (t)) is the average solid pressure as
-
measured according to the continuum mechanics approach and K(ps (t)) is the elastic bulk
modulus.
The mathematical model is equipped with an equation of state for the gas and the solid of the
form:
mod
Pg =
f g (pg ,Tg ),
Ps
mod
fs(Ps,Ts )
(8)
Consider a cell, which is burning. The burn fraction F(t) and the burn fraction rate are defined
in equation (9). The burn 'fraction of the solid material describes the fraction of the solid
material that has reacted.
(9)
The burning velocity, b(t) , of a burning front normal to the burning particle surface is given
by a function H which typically is dependent of the gas pressure:
bet)
de!
mod
H(fJPg ),
where fJ
= 1/ Pa
(10)
de!
where
fJ = 1/ Pa
(11 )
where H (~Pg ) expresses the velocity of the burning front normal to the burning surface, and
where G(F(t expresses the size and shape of the burning surface.
This model is not obvious, although it is easily proven when the particles are identical
spheres (see later in this article).
Assume that the gas in the cell decompose, then if the heat conduction between the solid
and the gas is neglected, the following equation is descriptive during a short time interval:
~ (cvsMs (t)~ (t) ) = cvs~ (t)~ s (t) - div (r(t)~ii; (t) ) = cvs~ (t)~ s (t)
(12)
where ~ V and ~ii; (t) is zero for a non deforming closed cell. Also
~ii; (t) : Lagrangian displacement of the boundaries of a cell during a short time interval
r(t): Lagrangian stress tensor
(13)
Observe that if the initial temperature of the gas equals the decomposition temperature, the
temperature of the gas equals the decomposition temperature for all times.
Consider now, as an example, the burning of a granular press loaded explosive consisting of
particles described as identical spheres. Assume that the volume of the cell is constant
(closed bomb). The following equation is then valid if the burning rate of the outer surface is
given by bet) = -f(t) :
Ms (t)
mod
(14)
where N is the total number of particles within the total cell volume V, and b(to) = a is the
initial radius of the particles. The reaction ratio F(t) is now given from (9) and (14) as
(15)
To simplify, assume that the densities of the particles are constant, Le.
=1- r(t)3
a
::::>
(16)
F(t)
3(1- F(t))2/3
a
f(t)
(17)
Compare the relation in (17) with the more general relation in (11). It then follows that
3(1-F(t))2/3
G(F(t)) =
.
in this case when bet) = -f(t)
In this
case the time development of the burn fraction is independent of any assumed equation of
state of the solid or the gas. The position of the burning front is then given by
f(t) = -vo
when t
~ a/va
(18)
.
F(t)
2/3
3vo ( 1- F(t) )
vot 3
F(t)=l-(l--) ,
F(t)
= 1,
whent~alvo
(19)
The result of an analytical calculation of the burn fraction as a function of time for a cell with
constant volume, when a = 10-4 m, ret)
= constant = Vo = 1 ml sand
Ps (t)
= const. = Ps (to)
is shown in figure 2.
~0.6
- _._... .,
---- -- .. "
0+------+------1------+------1
O.OE+OO
3.0E-05
6.0E-05
9.0E-05
1.2E-04
Time [5]
Figure 2: Analytical calculation of the burn fraction as a function of time for a cell with
constant volume when a
= 10-4 m,
ret)
Assuming again burning of en energetic material in a closed bomb. From equation (1) - (6)
and (9) it follows that:
15s (t) =Ms (t)IV = (1- F(t))M s (t)IV = (1- F(t))(M = (1- F(t))p(t) - (1- F(t))Mg (to)IV,
M -M (t)
V s
= pet) -
M g (to)IV
M ()I
= g t Vg
M -Ms(t)
= V[l- Vs IV]
= p(t)F(t)+(1-F(t))M g (to)IV
[1- 15 (t)1 Ps (t)]
s
(20)
Ms+M g
where pet) =------:::....
V
If the initial mass of the gas in the pores is very small, Le. M g (to) : : : 0, and the particles as
such are not compacted, Le. Ps (t) = constant equation (20) gives
(21 )
P (t)
g
Ps (t)
p(t)F(t)
=
P g (t)
[1-(I-F(t))p(t)/Ps(t)] (1-Ps(t)/Ps(to))
= Ps (to)
Figure 3 shows an analytical calculation of the four densities as a function of time in a case
where pet)
2 -,----------,------.......,----.--------,
-.;
"
,""
./1s(O'
t'5' 1.4
E
Co) 1.2
-
;:1
C)
;t:::
~ 0.8
Q)
0.6
_ .. /
3.0E-OS
75
.. ,s'(t)
6.0E-OS
Time [5]
..
9.0E-OS
1.2E-04
Figure 3: Analytical calculation of the four densities as a function of time in a case where
pet)
(23)
where n' , Rand P g max are constants. The unit for n' is mol/kg.
Figure 4 shows analytical calculations of the gas pressure, Pg , as a function of time for the
same case as shown in 'figure 3. P g max is assumed to be 1.82 g/cm 3 . In addition
P g (t)(I- Ps(t)]as given in equation 28 is plotted. This expression is the effective gas
Ps (t)
pressure, as explained later in the paper.
4.00E+06 --,-------,-..------..--~-------,
3.50E+06
........ 3.00E+06
co
a..
'(; 2.50E+06
~
:::J
~ 2.00E+06
Q)
~
c-
en 1.50E+06
co
(!)
1.00E+06
5.00E+05
O.OOE+OO
"""1' - - - - - - - - i - - - - - - - - - r - - - - - - - - - . - - - - - - - l
O.OE+OO
3.0E-05
6.0E-05
9.0E-05
1.2E-04
Time [5]
the case shown in fjgure 3. P g max = 1.82 g/cm3 , nRTgPg(t) =1100 Pa. The other curve
shows the effective gas pressure as given in equation 28.
CONVECTIVE BURNING
'
Piezoelectric transducers were placed at the top and bottom of the tube as indicated in the
figure.
The pressure data were measured using Kistler model 607e3 dynamic gages, and Kistler
model 504E4 amplifiers (Type 545A1 filters). The frequency response of the pressure
transducer is reported as 50 kHz, with a 1.5-psec rise time. A Nicolet, Multipro digital
oscilloscope was used to collect the pressure time history data as well as the temporal
fiducial record. The digitized Nicolet records were imported into Igor Pro for conversion of the
signal from voltage to pressure.
Red Dot smokeless powder, shift 2, lot 733-12/94, ignited by a Reynolds, SO-80 igniter, was
used to ignite the porous bed. The Red Dot was housed in a stainless steel basket, as shown
in the Figure 5. The igniter was separated from the porous bed by a 1.27-cm (0.5-inch)
ullage. This combination of igniter aid and free volume were selected to provide minimal bed
disturbance and maintain ignition.
SO-80
HEAD
TRANSDUCER
IGNITOR BASKET
POROUS BED
High-speed motion picture photography was used to evaluate the ignition and combustion
event down the length of the tube. Two high-speed Photec IV motion picture framing
cameras viewed the experiment 1800 from each other. The cameras operated at
approximately 8000 pictures per second in the quarter framing mode resulting in an overall
framing rate range between 15,500 and 36,000 pictures per second. This framing rate
resulted in an interframe picture time of 28 to 65 psec for these experiments. Both cameras
were 'fitted with a 150-mm telephoto lens and operated with an aperture setting of f8.
The combustion rate data were determined from the high-speed motion picture film with the
aid of a Vanguard Motion Analysis System. A Reynolds header mounted at the base of the
porous bed arrangement was fired with closure of the firing circuit and provided a temporal
fiducial for data collection.
Flame propagation and pressure data for the HMX based explosive PBXN-5 at 74 percent of
theoretical maximum density (TMD) are plotted in Figure 6 a) and b). The pressure data are
appended with the flame location data. A maximum flame propagation rate of 553 m/s was
measured, while the pressurization rate, as determined 'from the two pressure signals was
1050m/s. These data indicate that the pressurization wave is moving nearly two times faster
than the luminous burn front.
200
160
150
~ 120
~
l8
1553m1~1/
o
4.3
Tube leak/break
top gage
bottom gage
100
a.
50
4.5
o-l--l------a;;;;;:;==:.--~
4.7
4.9
TIM:E-~
5.1
5.3
2.5
3.0
3.5
4.0
4.5
Time - msec
a)
b)
Figure 6: Flame propagation and pressure data for PBXN-5 at 74% TMD.
Mathematical Model
As seen in the experiment, during combustion (not detonation) of energetic materials the
velocity of the reaction front usually moves behind the mechanical pressure front originating
from the burning solid material. The ability to model this phenomenon is important if the main
physical mechanisms are to be captured in the mathematical model. The velocity of tile
burning 'front is modeled according to the following relation
5.0
(24)
where y and
Pmax and P s bulk are constants. p s bulk is the bulk density of uncompacted
00,
(25)
The first case reflects a situation for a highly compacted material with only few scattered
pores, like in a cheese. The burning will then be like a cigarette. The second case reflects the
situation for a more porous material, where the burning mechanism is convective.
Let now LU be the distance between two neighbor cells where one is ignited. This is
illustrated in figure 7.
has ignited
cell x
Figure 7: Illustration of the distance LU between two neighbor cells where one is ignited.
Based on the calculated velocity vi CPs' Pg ) of the ignited cell and the distance LU between
this cell and the not ignited neighbor cell, one can calculate an ignition time delay:
T(t)
(26)
We assume that for a given cell it is possible to track the time of ignition of the cell and keep
it as a variable tig . The ignition criterion for the neighboring cell is then given by:
(15s(t), Pg (t))
(27)
".,
The hot-spot model [1], [4] and [5] is used as an additional ignition criterion. If the hot-spot
temperature in a cell, due to rapid compaction of the cell, is higher than the ignition
temperature, the cell also gets ignited.
By assuming that the burning of the particles in a cell contribute significantly more to the gas
pressure than any gas coming from neighbor cells, the gas pressure in convective burning
may be calculated as for the closed bomb. The relation is given by equation (23).
The pressure that is used to calculation of the cell deformation is given by equation (28) as
(28)
Ps(t)
where the mechanical average solid pressure and the effective gas pressure contributes to
the cell deformation additively. The gas pressure and the effective gas pressure are
illustrated in Figure 4.
CONCLUSIONS
w~lich
hydro codes. We also present experimental result where a convective burning front moves
behind the pressure front during combustion in long tubes. The mathematical model is
visualized in analytical studies of burning in a closed bomb.
The model is currently being implemented into the explicit numerical code Hi Dyna2D.
Comparison between simulations and experimental results will be presented in future papers.
REFERENCES
[1]
Diep, Q. B., Friis, E. K., Moxnes, J. F., Str0mgard, M., 0degardstuen, G.:"Simulation
of the Compaction of Energetic Materials", 33 rd International Annual Conference of
ICT, Karlsruhe, Federal Republic of Germany, June 25 - 28, 2002
[2]
Friis, E., Moxnes, J. F.: "Establishing Material Data of Pyrotechnic Powder Materials
by use of Inverse Modeling", 28th International Pyrotechnincs Seminar, Adelaide,
Australia, November 4 -9,2001
[3]
Atwood, A., Curran, P., Moxnes, J. F., 0degardstuen, G.: "Mechanical Properties of a
Porous Material Studied in a High Speed Piston Driven Compaction experiment",
th
[4]
[5]