38th AIAA Plasmadynamics and Lasers Conference<br>In conjunction with the<br>16th 25 - 28 June 2007, Miami, FL

AIAA 2007-4600

Chemically Reacting Plasma Jet Expansion Simulation for Application to Electrothermal Chemical Guns
Andrew J. Porwitzky Leonardo C. Scalabrin Michael Keidar and Iain D. Boyd , , ,
Department of Aerospace Engineering, University of Michigan, Ann Arbor, MI 48109
A computational uid dynamics simulation of an expanding capillary plasma jet for electrothermal chemical (ETC) gun application is presented. A chemical model is developed for the plasma-air interaction that uses 26 species and 60 reactions in an implicit CFD code. Time accurate results are obtained for two plasma jet experiments, one with wall impingement with pressure sensor data from multiple locations, and another that decelerated the plasma through a plasma holding chamber, which greatly eected the chemical species present. Conclusions are drawn from the results that speak to the importance of the plasma-air interaction in actual ETC implementation. This work is part of a complete beginning-to-end model of the plasma propellant interaction currently being developed.

Nomenclature
T Te u n P Patm Xm M Temperature in Kelvin Electron temperature in eV Flow speed in m/s Mass density in kg/m3 Number density in m3 Pressure in Pa, RT Pressure in atmospheres, P/(101325 Pa) Mole fraction of species m Mach number, u/ 1.4P/

I.

Introduction

much interest years in the design and electrothermal There has beenAn ETCwith aisina recentpropellantsource. Amongimplementation oftheencountered inchemical (ETC) guns. gun solid based artillery piece in which conventional ignition system has been replaced capillary plasma the enhancements plasma based ignition systems are reduced ignition delay time, highly repeatable ignition time, and enhanced burning and combustion of the solid propellant.1, 2 Previous work by the authors has looked at the capillary plasma source3 and the interaction of the plasma with the propellant grain.4 In order to perform accurate modeling of the plasma-propellant interaction (PPI) it is necessary to know which species are reaching the propellant surface. In both experimental setups and nal implementation there is the possibility for some plasma-air interaction before the resulting species reach the propellant surface, introducing the possibility that the species coming from the plasma capillary and those interacting with the propellant are not the same. The present study develops a model to determine the eect plasma-air chemistry can have on chemical composition of the plasma. The LeMANS chemically reacting, parallelized, implicit Navier-Stokes CFD code, developed at the University of Michigan, is used in the present work. LeMANS uses a modied Steger-Warming Flux Vector
Graduate Graduate

Student, Student Member AIAA. Email: aporwitz@umich.edu Student, Student Member AIAA. Email: lscalabr@umich.edu Assistant Research Scientist, Senior Member AIAA. Email: keidar@umich.edu Professor, Associate Fellow AIAA. Email: iainboyd@umich.edu

1 of 17 American Institute of Aeronautics and Astronautics Copyright 2007 by Andrew J. Porwitzky, Leonardo C. Scalabrin, Michael Keidar, and Iain D. Boyd. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Splitting scheme for calculation of inviscid uxes between mesh volumes, is second order accurate in space, and for the purposes of this work time integration is performed using a point implicit method. Further details of the models employed in LeMANS are outlined elsewhere.5, 6 The code was modied to allow for time varying boundary conditions, namely the capillary plasma inow. This work represents the rst time LeMANS has been used for time accurate or internal ows. The temporal accuracy of LeMANS was veried by simulation of shock tube cases and comparison to an exact Riemann Solver. Previous work modeling the plasma-air interaction was conducted at the Army Research Laboratory (ARL). In one such simulation and matching experiment7 a high pressure plasma jet was red at a stagnation pressure probe 15 cm from the capillary inow. Pressure versus time data were obtained and compared to the simulation. Although accurate pressure probe results were obtained, there are some problems with these previous simulations. The ARL model simulated 38 chemical species, including 16 ionic species. However, a detailed look at the chemistry model8 shows that in the 57 reactions there are 12 species that are not included, 10 of which are ionic species. Data was provided for species concentrations at locations between the capillary inow and the pressure probe. A careful analysis of the species concentrations at the pressure probe shows that there is a strong deviation from quasi-neutrality. The most likely reason for this is that the electrons diused away at high speed in the simulation. However, since the ARL model contains no electromagnetic terms in the equations governing the uid calculations, this deviation from quasi-neutrality should not have an eect on the ion ow. The missing electrons will have an eect on the recombination rates at the pressure probe, since there will not be enough electrons to allow ions to recombine. LeMANS is designed to simulate quasi-neutral plasma ows, with an electron diusion model that maintains quasineutrality.9 In all the simulations presented here the deviation from quasi-neutrality was less than 0.01% everywhere. Since the ARL study is the previous major eort in this eld, we felt that for the above reasons that this problem warranted a second look. Recent experiments of Li et al.11, 12 have investigated the plasma-air chemistry and the plasma jet in some detail. These experiments include stagnation pressure probe results for a range of capillary-probe distances, propellant sample analysis at a range of capillary-sample distances, sequential photographs of the plasma jet interaction for these cases, and sampling the chemical species present in the jet. The conditions of these recent experiments serve as the basis for the two simulations contained herein. In Section II we present a short overview of the capillary model in use to generate the plasma inow. The chemistry model developed is outlined in Section III. Section IV contains an overview of the experiments and how they are adapted to simulation. Section V contains results and discussions, including analysis of each simulation, as well as ways the chemistry model can be reduced to save computational time, and what the simulations indicate about the plasma-air interaction in practical ETC application. We conclude in Section VI.

II.

Capillary Model Inow

The plasma inow is generated via the previously developed capillary model.3, 10 Both experiments11, 12 simulated here use the same size capillary (1.6 mm internal radius, 26 mm length) with similar current proles and peak energy supplied to the plasma generator in each case. The current prole11 represents a discharge time of 220 s with a peak current of nearly 10 kA occurring at roughly 85 s. The current prole was extracted from the publication of Li et al.11 via a data stripping code. The electric current input to the capillary model can therefore be considered exact. Figure 1 shows the data points for current obtained by the stripping code. The one dimensional capillary code solves an ablation model at the polyethylene surface and couples this to a hydrodynamic model to determine capillary exit velocity, density, pressure and temperature. The Saha equation is solved at every cell in the capillary to give the individual densities of hydrogen and carbon ions and atoms, with the electron density determined by the quasi-neutrality condition. Figure 1 shows the polynomial ts to the capillary model results that are input into LeMANS as the inow pressure and temperature. All species exit the capillary at the electron temperature, which is mostly time invariant at 1.5 eV, or approximately 12,000 Kelvin. The exit boundary condition for the capillary model is sonic ow, with the velocity inow of LeMANS determined as P/, with =1.4 assumed. The simulation is not sensitive to this condition, as the jet is a pressure driven ow. The capillary model output is t to ninth order polynomials for input into LeMANS. Some of these polynomials give inaccurate values at early times, so a 5 s oset is used. This 5 s is added back into the

2 of 17 American Institute of Aeronautics and Astronautics

CFD simulation in post processing in order for the start times to more fully agree with the experiments.

Figure 1. Capillary data.

III.

Chemistry Model

The chemistry model implemented here is built upon previous work, both theoretical and experimental, on plasma-air chemistry. LeMANS was originally designed for simulating hypersonic ow around reentry vehicles, and uses Parks Model13 for nitrogen-oxygen chemistry, the same model used by NASA.6 Due to the wide use of this model, it is used for all nitrogen-oxygen chemistry. Additional species are chosen based on the abundances from the ARL simulations,7 species with high abundances are chosen rst. Due to a desire to compare with the recent measurements of Li et al. on plasma species present in the jet,12 many of the more abundant detected species are added. As a last step, species are added that are needed for extended chemical mechanisms. For instance, to go from H to H2 O the species OH should be added to the model to allow for another reaction path. The total number of species is kept to a minimum to decrease computational expense as the needed resources and computational time can rapidly go beyond practical limits. The nal model arrived at contains 26 chemical species and 60 reactions. The rst eleven species are from Parks Model, namely N2 , O2 , NO, N, O, NO+ , N2 + , O2 + , N+ , O+ , and electrons (e). The remaining species are OH, CO, CO2 , H2 O, C2 H2 , C2 H3 , CH, CH2 , HCO, NH, H2 , C, C+ , H, and H+ . The Cx Hy species are added because the experiments detected a signicant amount of these species.12 The reaction rate data is obtained from the GRI-Mech project for all non-Parks Model and non-ion chemistry.14 The GRI-Mech data is used in several commercially available hydrocarbon chemistry software packages, including the CHEMKIN software. For the hydrogen and carbon ionization reactions the work of Chang and Ordonez15, 16 was employed. Table 1 presents a complete list of all reactions used in the simulation. Some of the third body reactions, indicated by the presence of species M, have dierent reaction rates for a certain M. These species specic rates are also incorporated into LeMANS. All the reactions used have accompanying backward reactions. Chemical data for each species, such as specic heat, enthalpy of formation, etc, are obtained from GRIMech or Field and Franklin.17 Viscosity data for each species are obtained from McBride and Gordon.18 Although LeMANS is capable of accommodating a two temperature model, representing translation and vibration, sucient information could not be found for all chemical species to use this approach. More importantly, due to the high temperatures involved in these ows the translational and vibrational temperatures are nearly equilibrated, rendering a two temperature model unnecessary.

IV.

Simulation Methodology

Two experiments are simulated in the present work. In each experiment, an identical plasma capillary with nearly identical current load was used. Identical capillaries and current loads allow us to use one capillary input for each simulation, thus the main dierence between the two simulations is the geometry.

3 of 17 American Institute of Aeronautics and Astronautics

Table 1. Reactions employed in the chemistry model.

Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Reaction O+H+MOH+M O+H2 H+OH O+CHH+CO O+CH2 H+HCO O+CO+MCO2 +M O+HCOOH+CO O+HCOH+CO2 O+C2 H2 CO+CH2 O2 +COO+CO2 H+O2 O+OH 2H+MH2 +M H+OH+MH2 O+M H+CHC+H2 H+HCOH2 +CO H+C2 H3 H2 +C2 H2 OH+H2 H+H2 O 2OHO+H2 O OH+CH+CO OH+CHH+HCO OH+CH2 CH+H2 O OH+COH+CO2 OH+HCOH2 O+CO OH+C2 H3 H2 O+C2 H2 C+O2 O+CO CH+O2 O+HCO CH+H2 H+CH2 CH+CH2 H+C2 H2 CH+CO2 HCO+CO CH2 +O2 OH+H+CO 2CH2 H2 +C2 H2

Number 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Reaction HCO+MH+CO+M N+OHNO+H NH+ONO+H NH+HN+H2 NH+OHN+H2 O NH+O2 NO+OH NH+NN2 +H NH+NON2 +OH C+NOCO+N N+CO2 NO+CO CH2 +O2 2H+CO2 2CH2 2H+C2 H2 H+e2e+H+ C+e2e+C+ O2 +M2O+M N2 +M2N+M NO+MN+O+M NO+OO2 +N N2 +ONO+N O2 + +OO2 +O+ N2 +N+N+N2 + NO+ +ONO+O+ N2 + O+N2 + +O + NO +O2 NO+O2 + NO+ +NN2 + +O N+ONO+ +e 2OO2 + +e 2NN2 + +e O+eO+ +2e N+eN+ +2e

4 of 17 American Institute of Aeronautics and Astronautics

The rst experiment11 res the plasma into an open chamber at 1 atm pressure. Five pressure sensors are implanted at 9.5 mm intervals in a wall 40 mm away, oriented parallel to the exit plane of the capillary. The rst pressure sensor is aligned with the capillary exit at the stagnation line, this sensor is labeled P0. Each subsequent sensor is labeled P1-P4, with P4s center being 38 mm above the capillary center. Before entering the chamber the capillary plasma travels through a cylindrical nozzle 1.6 mm in radius and 13 mm in length.11 Since the ow variables could change in this distance, the nozzle is also simulated. The simulation uses an axisymmetric geometry, with the outow at y=50 mm, and walls at x=13 mm and x=53 mm. The latter wall contains the ve pressure sensors. Pressure versus time data are output at the ve pressure sensors in the simulation. The P0 sensor is the rst wall cell above the symmetry plane, with P1-P4 from wall cells at position y=9.5 mm, 19 mm, 28.5 mm, and 38 mm, respectively. This simulation is labeled below as Case S1. In the second experiment12 the capillary plasma is injected into a cylindrical plasma holding chamber of radius 9.4 mm and 129.5 mm length.19 Gas escaped the plasma holding chamber through a 0.5 mm radius nozzle into a 10x10x10 cm chamber that housed a mass spectrometer probe to measure the species present in the plasma.12 The plasma holding chamber was employed in the experiment because the plasma jet washed over the probe too rapidly for any measurements to be made, thus the ow had to be slowed down. Abundances, in terms of arbitrary intensity units, of some species were measured. These arbitrary units make it dicult for exact comparisons to be made, however qualitative comparisons are possible that yield much insight. The simulation of this ow, labeled below as Case S2, has the geometry listed above, with the change that the expansion chamber after the secondary nozzle is not encumbered by the measurement apparatus. This allows for a more direct comparison of the jets in Case S1 and Case S2. It must be noted that in an attempt to decrease heat loss to the walls the holding chamber in the experiment was lined with polyethylene. However, there is evidence that this liner ablated as far as 50 mm from the primary nozzle with a total mass loss three to ve times that of the mass loss from the capillary. To see what eect this had, the experimenters replaced the polyethylene liner with one of identical dimensions made of aluminum. Their results show that only CO intensity could be accurately measured, with all other chemical species being within the noise level of the measurement.12 Essentially, by using a polyethylene liner, the same compound that is ablating in the capillary to form the plasma, the experimenters created a much larger capillary. The higher mass loss due to this wall ablation increases the abundance of carbon and hydrogen in the plasma holding chamber, allowing new reaction mechanisms to occur. This may explain the high abundance of Cx Hy detected in the experiment. The present simulation does not include this ablating holding chamber, and can be thought of as the case with the aluminum liner. These two simulations allow for some key comparisons to be made concerning the chemistry of the plasma-air interaction. Since the capillary is identical in both experiments, and hence in both simulations, we can compare the chemistry of the primary jet of Case S1 to that of the secondary jet of Case S2, to see the full eect that the holding chamber had. Also, simulating the secondary jet without wall ablation allows us to see how the polyethylene liner skewed the results of the mass spectrometer data. This comparison of the primary and secondary jet chemistry is the main result of the present study.

V.
A. Analysis of Case S1

Results & Discussion

Due to the large computational expense of running a 26 species, chemically reacting CFD simulation, grid convergence was found rst for a single species non-chemically reacting ow. This single species had the same inow temperature, velocity and pressure prole as the full simulation. Grid convergence is based on the arrival time of the pressure pulse at P0, the value of the peak pressure, and the behavior of the pressure prole just beyond the peak. Case S1 uses rectangular cells with a uniform cell spacing of x=0.1 mm, which results in 202,080 cells. Case S1 ran on thirty 2 Ghz processors with a timestep of t=2.5x109 s, having a total runtime of 296 hours to simulate 70 s of real time. Snapshots of Mach number contours for Case S1 are shown in Fig. 2, where t=0 s is at the start of the current prole in Fig. 1. For detailed examination of the chemistry we will refer to the data from t=35 s. At this time the jet is fully developed, the precursor shock can clearly be seen, and the jet has not yet begun interacting with the wall. Evaluating the chemistry at this time will allow for a much more direct comparison with the secondary jet chemistry of Case S2. Pressure versus time data are shown in Fig. 3. Experimental data indicates that the precursor shock
5 of 17 American Institute of Aeronautics and Astronautics

(a) 15 s

(b) 25 s

(c) 35 s

(d) 45 s

(e) 55 s

(f) 74 s

Figure 2. Snapshots of Mach number for Case S1.

6 of 17 American Institute of Aeronautics and Astronautics

should arrive at P0 at approximately t=60 s, and have a peak of 1.4 MPa. The width at half height for the experimental P0 is approximately 20 s.11 Case S1 gives a P0 arrival time of 45 s, a peak of 3.2 MPa and width of approximately 7 s. Experimental data exists for the chamber pressure versus time in the plasma holding chamber of Case S2.12 The experiment shows a peak pressure of around 2.5 MPa, whereas Case S2 indicates that chamber pressure can reach more than twice that. This data indicates that perhaps the capillary model is overestimating the pressure. Future work will examine which phenomenon can aect the outow pressure in the capillary model.

Figure 3. Pressure versus time at the ve pressure sensors for Case S1.

In terms of the pressure pulse arriving early in Case S1 there is a possible explanation beyond the overestimation of the capillary pressure. A change of slope in the current prole is visible in Fig. 1 at approximately 10 s. In a typical capillary, an exploding wire is used to create the initial plasma used to trigger the self-sustaining ablation, and this bump is due to the wire exploding.11 The capillary code does not model the exploding wire, and instead initializes with a plasma density typical of an exploding wire plasma. With this initial density the capillary begins ablation within the rst 3 s of the simulation due to the high current load. It is possible that the correct start time would be after the wire explodes, adding a further shift of 10-15 s. This demonstrates how dicult it can be to estimate the start time in this type of simulation. Experimentally it is logical to measure time from the initiation of the current pulse, however the initiation of the current pulse may not correspond to the start of capillary ablation on a timescale of a few microseconds. Further work may be needed to more accurately determine the delay time between initiation of the current pulse and the start of capillary ablation. Pressure and temperature contours for Case S1 are shown in Fig. 4. The pressure in the capillary nozzle can exceed 180 atm, while the temperature can exceed 17,000 K. Combined with the Mach number contours in the nozzle, it is easy to see that Case S1 is a pressure driven jet. Mole fraction contours for selected signicant species are shown in Fig. 5. The contour shape of Fig. 5a is representative of molecular compounds produced by the plasma-air interaction. The products of these reactions are contained in a thin region where the plasma comes in contact with the air, preheated by the precursor shock. Figures 5b-5f show where most of the hydrogen is going chemically. Most of the hydrogen remains atomic, and a signicant portion remains ionized in the jet. Very little hydrogen has recombined to H2 , the maximum OH mole fraction is 0.022, with the peak H2 O mole fraction below 0.01. Although chemistry is occurring in the plasma-air reaction zone, the jet remains at such a high temperature that it is still composed primarily of hydrogen and carbon atoms and ions. In the reaction zone, oxygen and nitrogen are heated to a point where atomic oxygen has a peak mole fraction of 0.25, similarly atomic nitrogen reaches a peak of 0.3. Complex air ion species, such as NO+ , remain negligible, with mole fractions below 105 . Air ionization reaches a peak mole fraction of 0.01 and 0.003 for N+ and O+ , respectively. The main conclusion we can draw from the Case S1 chemistry is that the primary result of the plasma-air reaction in the jet is not the creation of diatomic chemical species, although this does occur to some extent. The main result is ionization and recombination and the dissociation of O2 and N2 into atomic form. As we will discuss later, in practical ETC application the capillary-propellant distance will be minimized, thus we can expect this minimal chemistry to occur.

7 of 17 American Institute of Aeronautics and Astronautics

(a) Pressure in atm.

(b) Temperature in Kelvin.

Figure 4. Case S1 pressure and temperature contours at t=35 s.

8 of 17 American Institute of Aeronautics and Astronautics

(a) CO

(b) H

(c) H+

(d) H2

(e) OH

(f) H2 O

Figure 5. Case S1 mole fractions for selected species at t=35 s.

9 of 17 American Institute of Aeronautics and Astronautics

B.

Analysis of Case S2

For Case S2 a grid spacing of x=0.2 mm is used, which results in a total of 100,570 grid cells. This allows for faster computational times while still capturing the chemistry. Case S2 also ran on thirty 2 Ghz processors with a timestep of t=5x109 s, having a total runtime of 174 hours to simulate 110 s of real time. To provide a sense of scale, Mach number and temperature contours for the entire Case S2 computational domain at t=115 s are shown in Fig. 6. Mach number contours for the secondary jet at dierent ow times are shown in Fig. 7. Chemistry data is referenced to Case S1 at 115 s, as the jet is fully developed at this time. The secondary jet emerges from the 1 mm diameter nozzle 75-80 s after capillary ring. The pressure in the precursor shock behind the nozzle prior to expulsion is approximately 55 atm. As the jet emerges from the second nozzle the temperature drops several thousand Kelvin, as shown in Fig. 8b. At 85 s the pressure in the nozzle is still 50 atm, but this drops to below 7 atm in the rest of the jet. Figure 8 also shows that although the pressure in the nozzle is still roughly 50 atm, this drops to no higher than 3 atm within 10 mm of the exit. The temperature in the bulk of the jet remains below 5,000 K at all times, with the region farthest from the nozzle typically maintaining a temperature below 3,000 K. With such relatively low temperatures it is not surprising that mole fractions of all ionic species are negligible (below 1010 ) in the secondary jet. We also do not see the discrete plasma-air reaction zone that we saw in Case S1, but this is only partially due to the low jet temperature. Because the plasma holding chamber was lled with 1 atm air (0.79 N2 , 0.21 O2 by mole fraction) the plasma-air chemistry was saturated while in the holding chamber. Mach number contours in the holding chamber in Fig. 7 show pulsations, due to compression and expansion waves reecting from the holding chamber walls. This causes the reaction zone to be perturbed, increasing mixing. For these reasons we can not expect to see the same chemistry, and thus the same reaction products, in the primary jet of Case S1 and the secondary jet of Case S2. Future attempts to measure the species present in the plasma jet must take these considerations into account. As previously mentioned, in the experiment a polyethylene liner was used in the holding chamber, which drastically increased the hydrocarbon-air ratio in the experiment. This leads to the expected result that the experiment detected far more hydrocarbon and hydrocarbon-air species than are found in Case S2. Figure 9 shows that atomic hydrogen exits the nozzle but as it begins to cool it becomes depleted. Analysis shows that although there is a halo of H2 circling the region of depleted atomic hydrogen with a peak mole fraction of 0.002, most of the hydrogen goes toward the creation of OH. Almost the opposite is true for oxygen. There is so much oxygen available thanks to the holding chamber that the atomic oxygen in the jet goes toward essentially every species that contains oxygen, with the limiting factor being the presence of carbon or hydrogen. Figure 10 shows that signicant atomic oxygen reaction mechanisms are the recombination to O2 , the production of CO, and as this cools the further recombination to CO2 . As the atomic oxygen and nitrogen begin to simultaneously cool, they also recombine to form a relatively high amount of NO in the jet. It is interesting to note that in the experiment the most abundant species, both with and without the polyethylene liner, was CO. Essentially no CO2 was detected in the experiment, while a large abundance of Cx Hy species was detected. In both Case S1 and Case S2, the mole fractions remained below 107 and 1010 for C2 H2 and C2 H3 , respectively. This disagreement with experiment can not be explained by Case S2, other than to say that it seams unlikely that CO would not recombine to CO2 while carbon and hydrogen would form such a large amount of Cx Hy species given the abundance of oxygen. Perhaps the measurement method employed was not particularly sensitive to the presence of CO2 , it seams based on the results of Case S2 and the presence of Cx Hy species in the experiment that most of the CO should be depleted. The main conclusion we can draw from the results of Case S2 is that the plasma holding chamber is not a viable option for obtaining measurements of the chemical species present in the plasma jet. The changes in temperature and pressure, as well as the increased interaction time, resulting from the slowing of the ow by the use of a holding chamber means that the species emerging from the secondary nozzle are not directly relateable to the species coming from the capillary. C. Simplication of the Chemical Model

With information from Case S1 and Case S2 it is possible to rene the chemical model. Although the authors feel that the current model captures the major eects in plasma-air chemistry, renement is possible

10 of 17 American Institute of Aeronautics and Astronautics

(a) Mach number.

(b) Temperature in Kelvin. Figure 6. Full computational domain for Case S2 showing Mach number and temperature contours at t=115s.

11 of 17 American Institute of Aeronautics and Astronautics

(a) 75s

(b) 80s

(c) 85s

(d) 95s

(e) 105s

(f) 115s

Figure 7. Snapshots of Mach number for the Case S2 secondary jet.

12 of 17 American Institute of Aeronautics and Astronautics

(a) Pressure in atm.

(b) Temperature in Kelvin.

Figure 8. Case S2 pressure and temperature contours at t=115s.

(a) H

(b) H2

(c) OH Figure 9. Case S2 mole fractions for selected hydrogen species at t=115s.

13 of 17 American Institute of Aeronautics and Astronautics

(a) CO

(b) CO2

(c) O

(d) O2

(e) NO Figure 10. Case S2 mole fractions for selected oxygen species at t=115s.

14 of 17 American Institute of Aeronautics and Astronautics

by simply eliminating species that do not contribute signicantly to the plasma-air interaction, thus allowing for a great increase in computational speed. Here we present what could be considered a minimalist model for the chemistry of the plasma-air interaction. The simulations clearly indicate that all Cx Hy species can be eliminated, as they never present in abundance. Although CH does not present in the jet, it does appear in signicant quantities in the capillary nozzle (0 mm x 13 mm). However, CH2 does not present signicantly even in the nozzle, so could be eliminated. Some of Parks Model can be neglected for ETC application. Although their atomic forms are prevalent, O2 + , N2 + , and NO+ do not present signicantly. The reason for this being that when temperatures are high enough these molecular species tend to dissociate rather than ionize under these conditions. For 26 chemical species the implicit method has a 29x29 matrix to solve at every grid point, where 29 comes from 26 species plus two velocities plus energy. The elimination of six species reduces the implicit submatrix to 23x23, a reduction of 312 elements, or a reduction of 60%. Of course all chemical reactions including these species would also be eliminated, resulting in a further increase. The reactions eliminated from Table 1 are 4, 8, 15, 20, 23, 26, 27, 29, 30, 41, 42, 50, and 52-58, leaving a total of 41 reactions. This represents a signicant speedup for the code. D. Plasma-Air Chemistry in Practical ETC Application

All previous work on the expansion of the plasma jet and plasma-air chemistry for ETC application has had a major aspect in common: relatively large capillary-wall, or capillary-propellant, spacings. In the case of capillary-propellant spacing, there is typically a desire to inspect the propellant surface after it has been exposed to the plasma as this is highly informative.20 It has been found in experiments, however, that for small capillary-propellant distances the propellant is cracked or shattered by the initial pressure wave from the capillary ring,21 leading to typical experimental capillary-propellant distances greater than 20 mm. When a large capillary-wall distance is used it is typically because there are pressure sensors located in the wall, as in Case S1, which are easily damaged by the high temperature, velocity and pressure of the jet near the capillary exit,21 again leading to larger sample distances. In real ETC application, the geometry of the plasma jet expansion region is much dierent. There are a number of dierent propellant packing options, but they primarily reduce to one of two choices. The rst is to pack the chamber with dierent shape solid propellant blocks or loose propellant grain, leaving crevasses for the ignition plasma to ow through. This packing can leave a small air pocket just outside the capillary exit, or can be right up to the capillary itself.2 In fact there are results indicating that for most ecient ETC ignition the capillary-propellant distance should be minimized.22 The second method is to have a central channel for plasma to ow through, resulting in a smaller interaction region but a more uniform burn.22 This method is similar to the ignition method of solid rocket motors. With the rst method, it is easy to see that plasma-air chemistry will have little eect, as there is little air present in the regions between the tightly packed propellant blocks. The primary chemical mechanism will be the interaction between the capillary plasma and the propellant surface and ablated propellant gases and particulates. A similar conclusion is not so straight forward for the second method. As the plasma advances down the tube the plasma-air reaction zone will be pushed with it, which should severely limit the high temperature air-propellant reaction zone. Evaluation of Case S1 at t=15 s indicates that if the capillary-propellant distance is small (< 10 mm), then there will be little to no plasma-air chemistry before the plasma-propellant interaction begins. Furthermore, any ionized, or otherwise aected, air species that are present will be transient, as the reaction zone will sweep past the propellant surface followed by the capillary plasma jet. The above information for practical ETC application, along with the results for plasma-air reaction rates from Case S1, suggest that in real ETC application the plasma-air interaction is perhaps completely negligible. To test the importance of the plasma-air interaction in the second propellant packing option a simulation is conducted using the reduced chemistry model outlined above. The geometry for this simulation, labeled below as Case S3, is taken from Beyer et al.22 The capillary inow is the same used as in Case S1 and Case S2. In the experiment a tube (6.3 mm diameter, 152 mm length) was inserted into the cylindrical propellant grain. The tube had eight sets of four holes to allow hot ignition gases to escape. Gas ows into the tube from the capillary and exits out the 32 vent holes. To simulate this geometry the eight sets of holes are converted to eight axisymmetric slits of equal surface area, giving an individual slit width of 0.4 mm. A cell spacing of x=0.05 mm is used due to the small size
15 of 17 American Institute of Aeronautics and Astronautics

of the outow slits. A time step of t=5x109 s is used. In the experiment there was no capillary nozzle, thus the simulation geometry is simply an axisymmetric rectangle 3.15 mm by 152 mm with outow slits every 16.9 mm, with the capillary inow at x=0 mm. All ow properties are output at each of the outow slits over time. Figure 11 shows pressure, temperature, and signicant species exiting the rst outow slit (x=16.9 mm). Species data is presented in terms of number density. We see a strong initial spike of N2 and O2 as the precursor shock is pushed past the outow slit. As time progresses we see a steady rise in N, O, and NO as the temperature rises, while N+ and O+ remain negligible. When the temperature surpasses 10,000 K the air species have passed the slit and the capillary species have arrived. At this rst slit we see a clear distinction between the precursor shock and the plasma jet. For small capillary-sample distances the high temperature gas reaching the propellant surface will be primarily composed of the capillary species, with little plasma-air chemistry taking eect. Evaluating the outow data at the remaining seven slits yields much insight. As we approach the slit farthest from the capillary we see the distinction between the precursor shocks air species and the capillarys plasma species disappear. Outow data for the nal slit (x=135.2 mm) show that the temperature is a few hundred Kelvin lower than the inow slit due to the chemical reactions happening in the tube. There are also fewer ionic species, but a higher abundance of atomic species. Essentially we are seeing a similar eect at the later slits as we did in Case S2 with the plasma holding chamber; more plasma-air mixing alters the chemistry of the jet. This dierence in species composition could lead to a dierent ignition mechanism for the propellant located at the holes further down the tube. The importance of the chemical composition of the plasma to the PPI ignition process is not well known, and is the focus of future work by the authors.

(a) Signicant air species and temperature.

(b) Signicant hydrocarbon species and pressure.

Figure 11. Outow data for the rst slit at x=16.9 mm in Case S3.

VI.

Conclusion

Presented was a plasma-air chemistry model designed and tested for ETC application. The results of these simulations allow further renements to the chemistry model, allowing for the possible elimination of some low abundance species and a decrease in computational time. There are two major conclusions from this work. Direct comparison of Case S1 and Case S2 clearly indicate that there are signicant dierences between the primary and secondary jets in the referenced experiments. Dierences include jet pressure, temperature and chemical composition. We thus conclude that any experiments to measure the species present in the plasma jet must be very carefully designed, and that the use of a plasma holding chamber is not a viable option as it will not produce results representative of species present in the primary jet. Secondly, an analysis of Case S1 indicates that for capillary-sample distances less than approximately 10 mm the plasma-air chemistry may be entirely negligible. A simulation of a practical ETC ignition mechanism

16 of 17 American Institute of Aeronautics and Astronautics

was conducted to evaluate this hypothesis. This work is part of a larger attempt to model and study the plasma-propellant interaction. The results of these simulations and others using the model developed will allow for better estimates of the plasma composition in the plasma-propellant interaction, and thus better model results.

Acknowledgments
The authors gratefully acknowledge the nancial support by the Army Research Oce, Grant No. W911NF-04-1-0251 (Dr. Ralph Anthenien, technical monitor). This work is also supported by the FrancoisXavier Bagnound Foundation. Porwitzky would also like to thank Tom Schwartzentruber for many helpful discussions and suggestions regarding CFD techniques, as well as David Berger for the data stripping code.

References
A., Ehrhardt, W., Kelzenberg, S., and Eisenreich, N., Plasma ignition and combustion, Prop. Explos. Pyrotech., vol. 26, pp. 75-83, 2001. 2 Dyvik, J., Herbig, J., Appleton, R., OReilly, J., and Shin, J., Recent Activities in Electrothermal Chemical Launcher Technologies at BAE Systems, IEEE Trans. on Magnetics. vol. 43, no. 1, pp. 303-307, January 2007. 3 Keidar, M., and Boyd, I. D., Ablation study in the capillary discharge of an electrothermal gun, Journal of Applied Physics, vol. 99, 053301, 2005. 4 Porwitzky, A. J., Keidar, M., and Boyd, I. D., Modeling of the Plasma-Propellant Interaction, IEEE Trans. on Magnetics. vol. 43, no. 1, pp. 313-317, January 2007 5 Scalabrin, L. C., and Boyd, I. D., Development of an Unstructured Navier-Stokes Solver For Hypersonic Nonequilibrium Aerothermodynamics, 38th AIAA Thermophysics Conference, 6-9 June 2005, Toronto, Canada. 6 Scalabrin, L. C., and Boyd, I. D., Numerical Simulation of Weakly Ionized Hypersonic Flow for Reentry Conguations, 9th AIAA/ASME Joint Thermophysics and Heat Transfer Conference, 5-8 June 2006, San Francisco, California. 7 Nusca, M. J., Anderson, W. R., and McQuaid, M. J., Multispecies Reacting Flow Model for the Plasma Eux of an ETC IgnitedApplication to an Open-Air Plasma Jet Impinging on an Instrumented Probe, ARL-TR-3227, July 2004. 8 Nusca, M. J., McQuaid, M. J., and Anderson, W. R., Numerical Simulation of an Open-Air Plasma Jet Using a MultiSpecies Reacting Flow CFD Code, 31st AIAA Plasmadynamics and Lasers Conference, 19-22 June 2000, Denver, Colorado. 9 Sutton, K., and Gnoo, P. A., Multi-Component Diusion with Application to Computational Aerothermodynamics, 7th AIAA/ASME Joint Thermophysics and Heat Transfer Conference, June 15-18, 1998, Albuquerque, New Mexico. 10 Keidar, M., Boyd, I. D., Williams, A., and Beyer, R., Ablation Study in a Capillary Sustained Discharge, IEEE Transactions on Magnetics, vol. 43, no. 1, pp. 308-312, January 2007. 11 Li J., Litzinger, T. A., Das, M., and Thynell, S. T., Study of Plasma-Propellant Interaction During Normal Impingement, Journal of Propulsion and Power, Vol. 22, No. 5, September-October 2006. 12 Li J., Litzinger, T. A., Das, M., and Thynell, S. T., Recombination of Electrothermal Plasma and Decomposition of Plasma-Exposed Propellants, Journal of Propulsion and Power, Vol. 22, No. 6, November-December 2006. 13 Gupta, R. N., Yos, J. M., Thompson, R. A., Lee, K., A Review of Reaction Rates and Thermodynamic and Transport Properties for an 11-Species Air Model for Chemical and Thermal Nonequilibrium Calculations to 30000K, NASA Reference Publication 1232, 1990. 14 GRI-Mech Home Page, http://www.me.berkeley.edu/gri mech/. 15 Chang, Y., and Ordonez, C. A., Semi-Empirical Ionization and Excitation Rate Coecients for Hydrogen, Helium, and Oxygen Atoms and Ions in Plasmas, AIP Conference Proceedings, 10 June 1999, vol. 475, pp. 887-890. 16 Chang, Y., and Ordonez, C. A., Generalization of the Arrhenius relation and ionization reaction rates for carbon atoms and ions in plasmas, Chemical Physics, vol. 231, pp. 27-30, 1998. 17 Field, F. H., and Franklin, J. L., Electron Impact Phenomena and the Properties of Gaseous Ions, Academic Press, New York, Revised Edition, 1970. 18 McBride, B. J., and Gordon, S., Computer Program for Calculation of Complex Chemical Equilibrium Compositions and Applications; II. Users Manual and Program Description, NASA Reference Publication 1311, June 1996. 19 Li, J., private communication, 29 March 2007. 20 Pesce-Rodriguez, R. A., and Beyer, R. A., A Theory of Plasma-Propellant Interactions, ARL-TR-3286, September 2004. 21 Li J., Litzinger, T. A., and Thynell, S. T., Interaction of Capillary Plasma with Double-Base and Composite Propellants, Journal of Propulsion and Power, Vol. 20, No. 4, July-August 2004. 22 Beyer, R., Brant, A. L., Plasma Ignition in a 30-mm Cannon, IEEE Transactions on Magnetics, vol. 43, no. 1, pp. 294-298, January 2007.
1 Koleczko,

17 of 17 American Institute of Aeronautics and Astronautics