You are on page 1of 42

Experimental and Numerical Investigation into the Ignition and Combustion of Aluminum Particles with TBX Applications

Eric R. Boyd, Ryan W. Houim, Dr. Kenneth K. Kuo April 28, 2009

The Department of Mechanical and Nuclear Engineering The Pennsylvania State University University Park, PA 16802

Acknowledgements
We would like to express our thanks to Jon Fox and Jan Mahar for the support and administering the DTRA-SRAP program under Contract No. DTRA01-03-D-001-0006. We would also like to thank Prof. Alon Gany and Dr. Valery Rosenband of Technion of Israel for supplying the Nickel coated Al particles for a part of this investigation.

Some Important Questions to be Answered


To what extent does the Ni-coating improve the ignition properties of two different sizes of aluminum particles?
By varying the diluent flow rate of the multi-diffusion flat-flame burner, the equilibrium flame temperature can be reduced to lower levels for determining the particles ignition behavior.

What are the effects of CO2, H2O, and O2 as oxidizing chemical species to the Ni-coated Al particles?
By adjusting the flow rates of the fuels (mixtures of H2 and CO) and the oxidizer (O2), systematic variation of product species can be achieved for studying the strength of the oxidizers and their effect on the ignition and combustion of the Ni-coated Al particles.

Does the Ni-coating inhibit any favorable combustion characteristics of Al particles?


By comparing the combustion times of the Ni-coated and uncoated Al particles.

Ni-Coated Al Particle Results


Four batches of particles: Davg = 9 m and 32 m (two batches with a 5-wt% coating of nickel) were tested. SEM images show that the aluminum particles vary in shape and are covered with nano-sized Ni particles. Not all of the particle were coated completely with some bare spots apparent.

10 m

5 m 10 m 5 m
4

Experimental Apparatus for Single Particle Ignition and Combustion Study


Particles ignite and combust in hot post-combustion zone of the flat-flame burner. Current burner fuel mixture contains both H2 and/or CO. Burner oxidizer is O2. N2 is used as a diluent. The particles are injected using a fluidized bed feeder. Quartz tube is utilized to prevent entrainment and contamination of the mixture in the post-flame zone from the ambient air.
Streak of Ignited Particle Energetic Particle Quartz Tube Flat-Flame

Oxidizing Mixture Fuel Mixture Fluidized Bed Feeder

MultiDiffusion Flat-Flame Burner

Particle Entrainment Gas

Particle Breakup Jet

Purge

N2

N2

O2

H2

CO

N2 H2 CO N2

Diagnostic System for Particle Burn Time and Temperature Measurements


Ignition Temperature were measured with a go or no go criteria. N2 diluent levels were increased until particles can no longer be ignited. The corresponding equilibrium flame temperature was considered to be the ignition temperature threshold. Photo-multiplier tube is used in conjunction with a cylindrical lens to collect data along the centerline for deducing the burning time durations (tb) Video images are taken as well for particles that stretch past the viewing range of the PMT (~5 cm above burner surface).

I t InstruNet DAQ

PMT

Burner

Cylindrical Lens

Gain Control

Camera for Streak Imaging

Measured Combustion Times of 9-m Particles


Data points lie almost on top of one another. Almost no distinction in tb between the coated and the uncoated particles can be established. Trends cannot be formed because of the large amount of data scatter due to the particle size distribution.
Ni-Coated 100% H2 Fuel Bare Al 100% H2 Ni Coated 50% H2 / 50% CO Bare Al 50% H2 / 50% CO Ni-Coated 5% H2 / 95% CO Bare Al 5% H2 / 95% CO

3
tb for DP = 25 m

2.5

Burning Time, tb (ms)

1.5

0.5
tb for DP < 5 m

0 0 0.2 0.4 0.6 Equivalence Ratio () 0.8 1 1.2

Measured Combustion Times of 32-m particles


Ni-Coated 100% H2 Bare Al 100 % H2 Ni-Coated 50% H2 / 50% CO Bare Al 50% H2 / 50% CO Ni-Coated 5% H2 / 95% CO Bare Al 5% H2 / 95% CO

8
tb for DP = 60 m

7 6 Burning time, tb (ms) 5 4 3 2 1


tb for DP = 4m

0 0 0.2 0.4 0.6 Equivalence Ratio, () 0.8 1 1.2

Data points lie almost on top of one another again. Indicating that the coating does not affect the combustion behavior of the larger particles either.
8

Comparison of tb(Dparticle) to other works


Correlation developed in Becksteads (2005) Summary of Aluminum Combustion Measured burning time for 32-m particles matches well with correlated data. Measured burning times of 9-m particle are longer due to importance of chemical kinetics for smaller sized particles.
Current Study

Consideration of Effective Oxidizer Mole Fraction


In Becksteads summary of aluminum combustion, he stated that the large variation in data was due to the relative strength of different oxidizer species. By correlating the data he found that O2 was the most effective aluminum oxidizer, H2O was about half as effective, and CO2 was about one fifth as effective. Therefore, he developed the follow effective oxidizer mole fraction:
X OX ,eff X O2 + 0.5 X H 2O + 0.22 X CO2

10

Correlated Ignition Temperature Data for the 9-m sized Al Particles


3000 Nickel-Coated Un-coated 2500

Tign 1078 X O2 + 0.5 X H 2O + 0.22 X CO2


Igition Temperature, Tign (K)
2000

-0.313

1500

1000

Tign 953 X O2 + 0.5 X H 2O + 0.22 X CO2

-0.266

500

0 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Effective Oxidizer Coefficient, XOX,eff

11

Correlated Ignition Temperature Data for the 32-m sized Al Particles


3000

Tign 1868 X O2 + 0.5 X H 2O + 0.22 X CO2


2500

-0.104

Nickel-Coated Un-coated

Ignition Temperture, Tign (K)

2000

1500

1000

Tign 969 X O2 + 0.5 X H 2O + 0.22 X CO2


500

-0.190

0 0 0.1 0.2 0.3 0.4 0.5 0.6

Effective Oxidizer Concentration, XOX,eff

12

Reasons for Lower Tign of Ni-Coated Aluminum Particles


Based upon measurements, Ni-coated aluminum particles have demonstrated lower ignition temperatures in comparison with bare aluminum particles.
The intermetallic reactions between the Ni and Al release heat to heat the particle near the Ni-Al interface. This additional heat then causes the NixAly compounds to melt which allows oxygen to diffuse to the interface. The oxidizer species can attack the Al particle surface to release significant amount of heat causing higher heterogeneous reaction rates at the Al interface leading to chain reaction that causes full ignition.

13

Conclusions from Experimental Investigation


1. Ni-coating on the aluminum particle surfaces made it possible for an ignition temperature drop of ~750 K on average for the larger (32 m) particles. The smaller (9 m) particles did not experience such a significant drop, but there was still a notable reduction in ignition temperature of ~300 K. Both sizes of the Ni-coated aluminum ignited and burned at temperatures as low as ~1100 K. The disparity is because the 9-m uncoated aluminum particle ignited at a lower temperatures than the 32-m aluminum particles. The mean combustion times for the coated and uncoated particles were almost identical. The measured tb match reasonably well with previously reported data from other experiments that tested aluminum particles of similar sizes. The considerable data scatter for tb can be attributed to the relatively large particle size variations. The primary controlling factor for aluminum particle combustion times is the size of the particles. A broad size distribution of particle sizes can result in a large variations in measured combustion times.
14

2.

3. 4.

5.

Need for a Detailed Model of Aluminum Particle Combustion


Numerical simulation is the only effective method to examine the combustion behavior of transitional Al particles (Dp from about 1m to 20m) in detail. Particles are simply too small to be observed in detail in practical experimental conditions Modern numerical methods allow the gas-liquid interface jumps to be accurately calculated, without the use of engineering correlations. There is a lack of understanding how shock waves interact with reacting droplets and particles. Fundamental understanding of shocked droplet ignition and combustion would be a great benefit to the development of advanced thermobaric explosives (TBX) as well as safety considerations. Using modern surface capturing methods, effects of shape change and particle breakup phenomena can be simulated directly A detailed understanding of the physicochemical processes on the combustion of transitional Al particles and flakes can be achievable.

SEM image of a Silberline PC-8602X aluminum flake.

10m

5 s

Calculation of a reacting Al flake.


15

Numerical Study Method of Approach (1/2)


The numerical schemes for treating the interface between different phases will be treated using techniques based on the Level Set and Sharp Interface Methods. Level Set method captures the location of the interface The Sharp Interface Method imposes the jump conditions at the interface. Chemistry is integrated using an operator-split approach by the freely available Cantera library (Developed by Dr. Dave Goodwin of CalTech) The advection and diffusion operators are evolved separately from chemical source terms This allows the choice of different solvers for fluid dynamics and chemistry, which typically have far different time scales. Transport and thermodynamic properties are calculated using routines from Cantera Mass diffusion can be changed from either mixture averaged formulation with a correction velocity for mass conservation equation or multicomponent system with thermal diffusion.

16

Numerical Study Method of Approach (2/2)


High-order numerical Weighted Essentially NonOscillatory (WENO) schemes are used to calculate the inviscid fluxes. Adaptive mesh refinement (AMR) capability using the ParaMESH library (Peter MacNeice at Drexel Univ.)
Increase computational efficiency by placing the finest cells only where they are needed.

Above: Numerical solution of the Riemann problems using high-order schemes. Left: Numerical solution of the Doubly Periodic Vortex test with AMR provided by ParaMESH.
17

Benefits of the Modeling Approach


Treatment of multi-phase reacting fluid flows with phase change while avoiding the use of empirical correlations The level set method is applicable to any particle morphology Spherical particles, oblong particles, and flake-shaped particles can be simulated by simply changing the initial conditions of the level set equation Development of a highly valuable predictive tool that can facilitate both fundamental understanding and engineering correlations in situations where detailed experimentation is difficult, expensive, or even impossible. The developed model can be utilized to systematically vary Particle geometry (size and shape), Ambient environment (mixture of gaseous species, temperature, etc.), Blast wave strength. The proposed model and numerical scheme could be extended to calculate detailed surface phenomena for many different types of multi-fluid flows. The end product can be applied to non-traditional interfaces. Some caveats are that a model of the surface phenomenon and equation of state are needed for the non-traditional material and the continuum assumption must be valid.

18

Future Work
Future areas of interest for experimental investigation include:
Test the particles under rapid heating, high-pressure environments like those that would be seen in a TBX blast Test particles in complex shock wave environments to see how the introduction of shock waves affects the particle ignition and combustion. Testing the particles as a propellant additive and in rocket motor environments would give additional useful information.

Future areas of interest for modeling investigation include:


Implementing the level set method for compressible multi-fluid flows Developing robust and accurate methods to calculate the effects of phase change at the droplet/particle interface.
19

Thank you very much for your attention

Any Questions?

20

Additional Slides Follow

21

Explanation for Large Variation of Al Particle Burning Times


Using one std. size deviation above and below mean size, burning time bounds can be found. Referring back to the burning time plots the collected data fits within the bounds.

Particle size distribution for a mean size of 9 m and a standard deviation of 16 m.

Particle size distribution for a mean size of 32 m and a standard deviation of 29 m.

22

Introduction
Aluminum releases a large amount of energy when combusted in oxygen environment. This increases the total energy content of aluminized energetic materials and hence increases explosive yield or propulsive thrust.
Al particles are very difficult to ignite, often requiring the removal of its protective oxide layer by melting at 2327 K or mechanical cracking.
Different geometries (such as flakes) may alter the stresses on the particle and aid ignition Apply coatings to the aluminum particles o Protective coatings with a lower melting temperature o Reactive coatings to initiate particle ignition.
140 Gravimetric Heat of Oxidation [kJ/gm 120 Volumetric Heat of Oxidation [kJ/cm 100 80 60 40 20 0 Boron (B) Carbon (C) Iron (Fe) Lithium (Li) Silicon (Si) Magnesium (Mg) Zirconium (Zr) Beryllium (Be) Aluminum (Al) Titanium (Ti) Tungsten (W) HTPB
3 fuel

fuel

23

Motivation
Al particle are used in thermobaric weapons as a fuel additive to fuel the destructive fireball and the resulting blast wave. The detonation event that is necessary to ignite these particles is very fast leaving a brief period for that particle ignition delay. This could lead to a large number of particle being unburned and adding nothing to the blast event. There are many other application such as rocket motors or other propellants that would benefit from a reduced particle ignition temperature and shortened ignition delay.
24

Nickel Aluminum Combustion


Several studies have been conducted showing a substantially lower ignition temperature and ignition in inert atmospheres.
Andrzejak et. al. (2006) burned 2.5 mm particles coated with different wt% of Ni. Ignition was characterized as low as ~1600 K in Ar and CO2 atmospheres. Rosenband et. al. (2007) placed bulk samples of 30m Al and Ni-Coated Al on an electrically heated metal strip in air and noted ignition at ~1350 K of the Ni-coated Al particles and no ignition of the bare Al particles. Yagodnikov et. al. (1997) completed a study on the effect of a nickel encapsulation on flame propagation in an aluminum particle aerosol. They found that Al particles encapsulated with nickel had flame propagation rates 1.5-4 times higher. Bocanegra et. al. (2007) carried out a study on coated and uncoated Al particles using laser heating. They found that, particle ignited at reduced temperatures even without a homogenous Ni coating.

25

Multi-Diffusion Flat Flame Burner


Quartz Tube PMT Assembly

Burner Surface

Cylindrical Lens

Camera

26

Prevention of Agglomerate Ignition


In order to study the single particle ignition behavior, it is important to avoid particle agglomeration. Glass beads sized 250 m and 2 mm were placed in a fluidized bed along with a mesh screen to prevent agglomerations from entering the flow stream.
Gas Outlet 400 Mesh Filter

Aluminum Particles Converging Nozzle

Streamlines

Large Glass Beads (dia.=2mm)

Small Glass Beads (dia.=250m)

Diffuser

Gas Inlet

27

Typical PMT intensity traces for a single intensely burning Al particle

1.375 ms 0.7 ms

Burning time is assumed to be when light intensity is collected until when light intensity drops off below noise levels.
28

Burning Particle Streak


Burning streak length was considered to begin when the light was emitted until when light was no long emitted The particles are completing combustion and potentially continuing to radiate light.
Visible particles are in fact burning though. Al2O3 particles were run through the burner at ~2500 K and no particles were visible. This is a potential source of error in the burn time analysis
29
End of Burning Time

Beginning of Burning Time

CHEMKIN Simulation
Oxidizer Inlet Simulation Output PSR

Quartz Tube
Equilibrium Products

Fuel Inlet

Input the surface wall conditions in conjunction with input flow rate into the CHEMKIN simulation Output the flow field temperature profiles and velocity profiles Burning times are found by dividing the recorded streak length by the gas velocity (particle velocity)
tb = Lstreak u gas
30

Instrumented Quartz Tube


The quartz tube for shielding the combustion product gases has been instrumented with five 25 m S-type thermocouples. The purpose for these measurements is to track the heat loss from the burner. The temperatures are measured at 4 locations on the outer wall of the quartz tube and 1 along the centerline on the exit plane of the quartz tube. The heat loss rate is used as a boundary condition that is needed as an input to the CHEMKIN simulation.
Quartz Tube Ceramic Insulation Square Copper Sheet Attached to Quartz Wall

25 m S-type Thermocouples

Flame

31

Velocity Results
Particles typically ignited and completely combusted within 10 cm. Typical velocities were on the order of 100 cm/s The centerline velocity increases due to the developing flow with the cylindrical tube. Temperature flow field simulations were also run to validate the model. Measure temperatures were typically with ~100 K of the calculated temperature.
Radius (cm) Exit Plane of Quartz Tube

32

Ignition Temperature Test Matrix


Composition of Fuel H2=100% CO=0% H2=75% CO=25% H2=50% CO=50% H2=40% CO=60% H2=25% CO=75% H2=5% CO=95% Equivalence Ratio () 0.25 Test 1 Test 5 Test 9 Test 13 Test 17 0.5 Test 2 Test 6 Test 10 Test 14 Test 18 1.0 Test 3 Test 7 Test 11 Test 15 Test 19 1.5 Test 4 Test 8 Test 12 Test 16 Test 20

Test conditions were selected to fully vary the product specie levels of O2, H2O, and CO2 by adjusting the fuel ratio and oxidizer content. 100% CO condition could not be studied because the OH radical is necessary to create a stable flame in CO+O2 reaction.
33

Combustion Time Test Matrix


Composition of in Fuel H2=100% CO=0% H2=50% CO=50% H2=40% CO=60% H2=5% CO=95% Equivalence Ratio 0.25 Test 1 Test 4 Test 7 0.5 Test 2 Test 5 Test 8 1 Test 3 Test 6 Test 9

Reduced in size to save in material costs and additional tests were not necessary. Captures the results that were necessary to see if the Ni-coating had any effect on the Al combustion.
34

Test Matrix
= 0.25
0.70 95% CO / 5% H2 Fuel Mixture 75% CO / 25% H2 Fuel Mixture 50% CO / 50% H2 Fuel Mixture 25% CO / 75% H2 Fuel Mixture 0% CO / 100% H2 Fuel Mixture 0.60

O2
0.50 Product Mole Fraction

0.40

0.30

H2O
0.20

0.10

CO2
0.00 0 10 20 30 40 50 60 70 80 90 100 Percent of Hydrogen in Fuel Mixture

Conditions were selected so that varying ratios of O2, H2O, and CO2 were created in the product stream.
35

9 m uncoated particle uncertainty


2600

Ignition Temperature Correlation, Tign=1078(XOX,eff)-0.313


2400

2200 Measured Ignition Temperature, Tign (K)

Measured Data Points

2000

+10% Certainty Limit


1800

1600

1400

-10% Certainty Limit

1200

1000 1000 1200 1400 1600 1800 2000 2200


-0.313

2400

2600

Correlated Ignition Temperature, Tign=1078(XOX,eff)

(K)

36

9 m Ni-coated particle uncertainty


2600

Ignition Temperature Correlation, Tign=953(XOX,eff)-0.266


2400

Measured Data Points


2200 Measured Ignition Tempearature, Tign (K)

2000

1800

+10% Certainty Limit


1600

1400

1200

-10% Certainty Limit


1000 1000 1200 1400 1600 1800 2000 2200
-0.266

2400

2600

Correlated Ignition Temperature Tign=953(XOX,eff)

(K)

37

32 m uncoated particle uncertainty


2600

Ignition Temperature Correlation, Tign=1869(XOX,eff)-0.104


2400

Measured Data Points


2200 Measured Ignition Temperature, Tign (K)

+10% Certainty Limit


2000

1800

1600

-10% Certainty Limit


1400

1200

Xox,eff > 1 so the correlation is no longer physically possible

1000 1000 1200 1400 1600 1800 2000 2200


-0.104

2400

2600

Correlated Ignition Temperature, Tign=1869(XOX,eff)

(K)

38

32 m Ni-coated particle uncertainty


2600

Ignition Temperature Correlation, Tign=969(XOX,eff)-0.313


2400

2200 Measured Ignition Tempearture Tign

2000

+10% Certainty Limit


1800

1600

Measured Data Points

1400

1200

-10% Certainty Limit

1000 1000 1200 1400 1600 1800 2000 2200


-0.190

2400

2600

Correlated Ignition Temperature, Tign=969(XOX,eff)

39

Temperature Profile
Measured exit temperature was 880 K and the calculated exit temperature was 940 K.
Exit Plane of Quartz Tube

40

Ni-Al Phase Diagram

41

Applications of PSU Model for TBX Applications


The model will zoom in on a single reactive particle. Overall domain of a TBX The particle will be impacted detonation, that is calculated using with a strong shock wave or traditional multiphase methods. contact surface Primary Shock The ignition and . Fuel Particles combustion will be calculated directly from . . . first principles. . . Modern interface capturing . . . techniques will allow the jump . . . . conditions at the gas-liquid . . surface to be calculated accurately. . .
Detonation Product Contact Surface

Some possible zoomed in domains for PSU model development

Fuel Particle

42

You might also like