Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract
FFI and Nammo developed a model for calculating deflagrating of particles where the
propagation of the ignition is convective. This model, the Slow Burn model, is based on
the assumption that the particles follow the gas flow, which makes it possible to
calculating deflagration without taking multiphase flow into account. This enables codes,
which is not able to generate new compounds, to compute the propagation of deflagrating
material.
ANSYS Autodyn has implemented a beta version of the Slow Burn model into its code,
which is called the Powder Burn model. The purpose of this paper is to simulate
deflagration in ANSYS Autodyn and establish Powder Burn material data for some well
known compounds within different regimes of function, for instance explosive,
pyrotechnical compound and gunpowder. Furthermore will these examples serve as a
template for how to obtain your own material data for a desired compound, and can then
be used in ANSYS Autodyns Powder Burn model.
To verify the material data which is inputted into the Powder Burn model, we use a close
bomb modelled with simply one cell model, to look at the pressure build up as a function
of time of a deflagrating material.
When establishing material data for ANSYS Autodyns Powder Burn model, a
clear view of the mathematical model of Slow Burn is of importance [1]. The
Slow Burn consists of a solid equation of state, a reaction ratio and a gaseous
equation of state for the reaction products.
1.1 Reaction ratio Given the solid mass ms (t) at time t, then the reaction
ratio F is defined by
ms t0 ms (t)
F (t) = .
ms t0
Definition
Pg = g eg e g /D , where D is constant.
Definition
1.3 Vieilles law extended Given the gas pressure Pg , then the burn rate b
is as follows
b Pg = aPg n + c, where a, c and n are constants.
Postulate
1.4 Reaction rate Given reaction ratio F and a gas pressure of Pg which
will give rise to a burn rate b, then the reaction rate will be
F = G 1 F c b Pg ,
where the growth parameter G, the growth reaction ratio exponent c and the
reaction ratio factor are constants.
Theorem
The constants G, c, and in T :1.4 are determined by the geometric shape of the
solid. For the compounds that we are going to investigate in this paper we need
G, c and for spherical and hollow cylindrical particles.
1.5 Spherical particles Given a spherical shape A with a radius of r, which
burns uniformly from the surface with the burn rate b (t) over time t. Then
the growth parameter, the growth reaction ration exponent and the reaction
ratio factor for A is given respectively by
3 2
G= , c= and = 1.
r 3
Theorem
Lastly we are going to take a quick look into ANSYS Autodyns ability to model
convective burning.
1.7 Speed of ignition front Given a gas pressure Pg and a solid density
s . Pg will give rise to a burn rate b Pg and s will give a s , then the
velocity v of the ignition front will be
v = C1 + C2 b Pg 1 + s , where C1 and C2 are constants.
Theorem
Three compounds with different application will be fitted to the Powder Burn
model: an explosive, a pyrotechnical compound and a gun powder. Our main
focus in this paper is to feed ANSYS Autodyns Powder Burn model with data
which enables Autodyn to model the gas pressure versus time in deflagration. For
this job we have used a single celled closed bomb, which means boundary
conditions with zero velocity. A step in constructing such data for Powder Burn
we computed gas pressure versus density [2], collected burn rate versus gas
pressure data from closed bomb experiments, and determined geometry and size
of the particles. This was done for each of the three compounds, see section 3, 4
and 5, and the data was fitted to D:1.2 , P:1.3 and T :1.4 in the Slow Burn
model.
When fitting the data to the Slow Burn model we used least squares, but since
least squares may give more than one solution, we derived an expressions for
finding the solution to the least square. The solution was then found by plotting
these expressions and picking the solution which is most physical.
Theorem
Theorem
In T :2.1 the internal energy per mass eg is the physical criteria for picking the
best solution if there are more than one solution. For T :2.2 the physical criteria
is that the burn rate b is always positive, since the burning process is irreversible.
Since the gas pressure Pg is positive, then a and c also needs to be positive. The
constant c was included in Vieilles Law because of the fact that just before the
burn process starts, there is no gas generated and hence zero gas pressure. Which
would then again result in zero burn rate and the burn process would never start.
By introducing a positive c as an initial value would ensure that the burning
process starts.
3.1 Measurements
a) Density g [ kg/m3 ] Pressure Pg [ GPa ] c) Pressure Pg [ MPa ]
Burn rate b [ m/s ]
0.09 0.0001 ambient 100 [ C ] 50 [ C ]
1 0.0012 0.588 0.0015 0.0017 0.00126
5 0.0065 0.588 0.0016 0.0017
10 0.0131 0.588 0.0014
50 0.0697 1.28 0.0025 0.00277 0.00239
100 0.1474 1.28 0.00264 0.00284
500 1.1053 1.28 0.003
1000 3.6387 1.62 0.00200
1500 8.937 2.66 0.0046 0.0048 0.003973
2000 19.6660 2.66 0.00488 0.0048 0.003879
2300 30.4711 2.66 0.00447
2500 40.4045 3.35 0.005077
4.04 0.0064 0.00668 0.005408
b) Size [ mm ] Quota [ % ] 4.04 0.00630 0.00665 0.005471
4.04 0.00627
4.8 100
5.41 0.00871 0.0099 0.00747
1.7 97.3
5.41 0.00744 0.00942 0.0079
0.425 3.5
5.41 0.00841
6.794 0.00975 0.0109
6.794 0.00993 0.0110
6.794 0.00996
6.794 0.0101
10.24 0.0136 0.0153
10.24 0.0136 0.0150
10.24 0.0137
Tab. 3.1.1: a) Gas density g vs. pressure Pg in the generated gas from HMX
(95%) / HTPB (5%) [2]. b) Particle size distribution for PBXN-5 [6].
c) Burn rate b vs. pressure Pg in the generated gas from PBXN-5 [4].
3.2 Fit to model
From Table 3.1.1.a we can use least squares T :2.1 to find the best fit of D and eg
in D:1.2 to describe how the gas pressure Pg is a function of gas density g for
the HMX (95%) / HTPB (5%) compound. And similarly for Table 3.1.1.c we can
use T :2.2 to find the best fit of n, a and c in P:1.3 to describe how the burn rate
b depends on gas pressure Pg for PBXN-5 which is similar to HMX (95%) /
HTPB (5%).
a) b)
eg [J/kg] 400 0.02
Deviation
60
b [m/s]
Deviation
Fit
Pg [GPa]
0 100 0.01
40 Fit
0
1e+016 n: 0.88369 0.005 Ambient
20 a: 4.0211e006 100C
100
c: 0.00027284
50C
2e+016 0 200 0
900 950 1000 1050 0 1000 2000 3000 0.2 0.4 0.6 0.8 1 0 5000 10000 15000
D [kg/m3 ] g [kg/m3 ] n Pg [kPa]
Fig. 3.2.1: a) Fitting gas pressure Pg vs. gas density g for HMX (95%) / HTPB
(5%). The plot on the left showing values for D and eg is the best fit
for Table 3.1.1.a to D:1.2 , which is plotted on the right. b) Fitting
burn rate b vs. gas pressure Pg for PBXN-5. The plot on the left
showing values for n, a and c is the best fit for Table 3.1.1.c to P:1.3 ,
which is plotted on the right.
For the single celled closed bomb we expect the terminal gas pressure Pg,T , when
g = ref , to be
2325
Pg,T = ref eg e ref /D = 2325 1341931 e 1005 [ Pa ] = 32 [ GPa ],
2000 0.8
30
s [kg/m3 ]
Pg [GPa]
1500 0.6
20
F
1000 0.4
10
500 0.2
0 0 0
0 0.2 0.4 0 0.2 0.4 0 0.2 0.4
t [ms] t [ms] t [ms]
Fig. 3.4.1: ANSYS Autodyn Lagrange 2D results for the HMX (95%) / HTPB
(5%) compound in a single celled closed bomb.
4.1 Measurements
a) Density g [ kg/m3 ] Pressure Pg [ GPa ] c) Pressure Pg [ MPa ] Burn rate b [ m/s ]
0.5 0.0001 5.41 0.01656
1 0.0005 6.794 0.029276
5 0.0026 6.794 0.01366
10 0.0054 6.794 0.02369
50 0.0284 6.794 0.02349
100 0.0601 6.794 0.034966
500 0.4275 6.794 0.033881
1000 1.0354 10.24 0.02967
1500 1.9451 10.24 0.02753
1900 3.1457 10.24 0.02644
10.24 0.07257
b) Size [ mm ] Quota [ % ]
0.125 100
0.063 6.15
b [m/s]
Pg [GPa]
0
2 0.04
5e+014 0.02
1
1e+015 0 0
1500 2000 0 1000 2000 0 5000 10000 15000
D [kg/m3 ] g [kg/m3 ] Pg [kPa]
For the single celled closed bomb we expect the terminal gas pressure Pg,T , when
g = ref , to be
1766
Pg,T = ref eg e ref /D = 1766 595291 e 1869 [ Pa ] = 2.7 [ GPa ],
0.8
1500
s [kg/m3 ]
Pg [GPa]
0.6
1000
F
0.4
1
500
0.2
0 0 0
0 1 2 0 1 2 0 1 2
t [ms] t [ms] t [ms]
Tab. 5.1: Chemical composition of 12.7mm NEXPLO double base propellant [3].
From the report by Gunnar Nevstad [3], 12.7mm NEXPLO double base
propellant is cylindrical with a hole in the center, where the length
L = 1.71 [ mm ], outer radius R2 = 0.69 [ mm ] and inner radius R1 = 0.1 [ mm ].
Since ANSYS Autodyn only supports = 1 for reaction rate T :1.4 , the only
hollow cylindrical shape possible to enter into ANSYS Autodyn is when
L = R2 R1 , see T :1.6 . By approximating to L = R2 R1 = 1.15 [ mm ] we get
the growth reaction ratio exponent c = 1/2, and the growth parameter
G = 52.17 [ 1/mm ].
5.1 Measurements
Tab. 5.1.1: 12.7mm NEXPLO double base propellant: a) Gas density g vs. gas
pressure Pg [2]. b) Burn rate b vs. gas pressure Pg [3].
The steps in finding the best fit for the Slow Burn model to Table 5.1.1 is the same
as in section 3.
a) b)
eg [J/kg] 600000 0.4
Fit
100000 1.5e+006 500000 Ambient
5e+016 20 0.3
D: 1003.2803 Data 400000
eg: 1011748.8579
Deviation
15 Fit
b [m/s]
Deviation
300000
Pg [GPa]
0 0.2
10 200000
n: 0.92359
5e+016 100000 a: 2.5094e006
0.1
5 c: 0.0070952
0
1e+017 0 100000 0
500 1000 0 1000 2000 0.2 0.4 0.6 0.8 1 0 200000 400000
D [kg/m3 ] g [kg/m3 ] n Pg [kPa]
Fig. 5.2.1: 12.7mm NEXPLO double base propellant: a) Fitting gas pressure Pg
vs. gas density g . The plot on the left showing values for D and eg is
the best fit for Table 5.1.1.a to D:1.2 , which is plotted on the right.
b) Fitting burn rate b vs. gas pressure Pg . The plot on the left showing
values for n, a and c is the best fit for Table 5.1.1.b to P:1.3 , which is
plotted on the right.
5.3 ANSYS Autodyn input
1 3
ref = 1.86 [ g / cm ]
Tab. 5.3.1: ANSYS Autodyn material: 12.7mm NEXPLO double base propellant.
For the single celled closed bomb we expect the terminal gas pressure Pg,T , when
g = ref , to be
1860
Pg,T = ref eg e ref /D = 1860 1011748 e 1003 [ Pa ] = 12 [ GPa ],
0.8
1500
s [kg/m3 ]
10
Pg [GPa]
0.6
1000
F
0.4
5
500
0.2
0 0 0
0 0.05 0.1 0 0.05 0.1 0 0.05 0.1
t [ms] t [ms] t [ms]
Fig. 5.4.1: ANSYS Autodyn Lagrange 2D results for the 12.7mm NEXPLO double
base propellant in a single celled closed bomb.
6 Further work
The last part of the Slow Burn model which we have not established data for yet
is the speed of the ignition front T :1.7 . This part makes the Slow Burn able to
1
Assumed grain compaction to 93 [ % ].
2
These values are just chosen and dont affect the single celled closed bomb model.
model convective burning. The ignition front has a constant velocity parameter
C1 , it also depends on the burn rate b Pg and a solid part s . The burn rate
we already have, so the main work will be to establish material data for s .
Since the materials we have used in this paper are powders, we would need to
look at a powder model like Compaction to find tabular values for s .
7 Conclusion
We have been able to establish material data for Slow Burn model for three
different compounds were all the particles are ignited at the same. It has been
taken into account how the pressure builds with time, how fast the material burns
and the shape of the particles. Results from ANSYS Autodyn regarding terminal
pressure in a closed bomb are as expected when we compare them with the input
data to the Slow Burn model. The time that it takes for the particles to completely
burn up is in the regime that we expect, from a few 10th of a microsecond to a
millisecond, which is much slower than a detonation which is fractions of a
microsecond.
[1] Atwood, A., Friis, E. K., Moxnes, J. F., 2003, A Mathematical Model for Combustion
of Energetic Powder Materials, 34rd International Annual Conference of ICT
[2] Sanford G., McBride B.J., (1994), Computer Program for Calculation of Complex
Chemical Equilibrium Compositions and Applications I. Analysis, vol. 1331, NASA
Reference Publication1
[3] Nevstad, Gunnar O., (2001), Karakterisering av brennegenskaper til noen krutt,
FFI/RAPPORT-2001/05202, Norwegian Defence Research Establishment
[4] D. T. Bui, A. I. Atwood, and P. O. Curran. Burning Rate Studies of Plastic Bonded
Explosive PBXN-5 and Composition A-4 Explosives, 1999 JANNAF Propulsion
Systems Hazards Subcommittee Meeting, Cocoa Beach, FL, October 1999
[5] A. I. Atwood, P. O. Curran, K. P. Ford, S. A. Baynar, J. F. Moxnes, and G.
Odegardstuen. "Convective Burning Studies of Pyrotechnic Powders," Eighth
International Symposium on Special Topics in Chemical Propulsion (8-ISICP), Cape
Town, South Africa, November 2009.
[6] Control - report, Requisition number 16398, from institution AFKL, Nammo Raufoss
AS
[7] Control - report, Requisition number 8089, from institution AFK, Nammo Raufoss
AS
1
We have corrected for the covolum with methods by FFI, John F. Moxnes and Tomas Lunde
Jensen.