Design of High-Area-ratio Nozzle Contours Using Circular Arcs
Design of High-Area-ratio Nozzle Contours Using Circular Arcs
Design of High-Area-ratio Nozzle Contours Using Circular Arcs
II. Design x x
θ = sin−1 1− sin θi + sin θj (6)
The major requirement for the design was that the general flow L L
characteristics were consistent with a conventional nozzle in a geo
metrically favorable design. A reduction in contour length for consis
tent area ratio may increase the thrust-to-weight ratio of the nozzle
and would imply a reduction in engine weight and cooling require
ments. The production of a contour that could be described spatially
was desirable, as it would enable greater ease of nozzle scaling both
geometrically and for varying gas compositions compared to a TOC/
TIC design.
A. Methodology
The design method uses a series of circular arcs to achieve the
contour design requirements. Similar to conventional nozzle design,
the divergence contour was separated into an expansion curve and a
turning curve, where each curve can be described by a finite series of
circular arcs. The use of arcs to generate the divergence contour
enables a full description of the curve through the manipulation of
trigonometric relations. The spatial description of the turning contour Fig. 2 Spatial description of an arc segment in the turning curve.
190 SCHOMBERG, OLSEN, AND DOIG
2.5
The nozzle contour can be described by the sum of the horizontal
and vertical or radial arc segments. The limits of these values are Exact
likely to be a design requirement of the nozzle itself; and, as such, the 2
Approximated
number of segments and angles between segments must be selected
by the designer. Calculations of these values are subject to Eq. (3) and 1.5
can be determined using Eqs. (7) and (8):
La/Lc
1
X
b
xtn = rij (sin θj − sin θi ) (7)
i=0 0.5
X
b 0
ytn = rij (cos θj − cos θi ) (8) 0 15 30 45 60 75
θm+θn
i=0
Fig. 3 Variation in exact and approximated length ratio.
B. Length Reduction
The length of any supersonic nozzle contour is often compared to
the length of an equivalent constant angle contour [10]. Assuming
that the expansion curve is consistent between both contours, a direct
comparison can be made between the turning curves. The length of a
constant angle contour is a function of the required vertical or radial
distance and contour angle, as shown in Eq. (9). Comparatively, the
turning curve length of a nozzle generated using the arc method can
be determined through manipulation of Eqs. (7) and (8) and is given
in Eq. (10). When defining a ratio between the nozzle lengths
determined through the circular arc method and that of a constant
angle contour, the required vertical of radial distance is cancelled.
The ratio becomes a function of wall angle only, as shown in Eq. (11):
ymn
Lc = (9)
tan θc Fig. 4 Contour plot of length ratio magnitude.
(sin θm − sin θn ) compare geometric parameters of the Space Shuttle Main Engine
La = (ymn ) (10) (SSME) [9], Vulcain 2 [14], and conventional Vulcain [15] nozzles
(cos θn − cos θm )
against an equivalent contour generated using the arc-based method. It
is important to note that the expansion curve was kept consistent with
the current design of each nozzle and a single curve was used for the
La (sin θm − sin θn ) 2θc
= tan θc ≈ (11) turning section to give a baseline comparison of the arc design. Tables 1
Lc (cos θn − cos θm ) θm + θn and 2 outline the current specifications of conventional nozzles and a
comparison with the arc-based contour length, respectively.
If a constant angle is selected by the designer to produce a baseline The arc-based contour returned a lower axial nozzle length when
nozzle length, the length ratio then becomes a function of the compared to all conventional designs, where the magnitude of this
expansion angle and the exit or divergence angle of the circular arc reduction increased with respect to nozzle area ratio (AR) in all cases.
contour. This length ratio can also be approximated as a simple power The associated change in thrust-to-weight ratio was calculated using
function for the given constant angle, as shown in Eq. (11). The the length ratio and nozzle length-weight coefficient as shown in
difference between the exact and approximated length ratios and a Eq. (12):
contour plots of the rate of length reduction are shown in Figs. 3 and
4, respectively, for a baseline constant angle contour of 15 deg.
1
A difference of 5% or less between the exact and approximate TW = (12)
length ratios was observed across the range of practical angles for an 1 − Cx (1 − xn ∕x2 )
axial-throat nozzle. Variation of the exact function can be seen to
increase as the difference between the combined design angle and a
value of twice the selected conical angle increases, where a maximum
Table 1 Geometric parameters in conventional nozzle
variation of 5% was observed at the upper limit of the angles con
designs
sidered here. It can be seen from Fig. 4 that a decreasing exponential
trend in length ratio was evident, and this trend was equally weighted Model yt rt1 y2 θ1 θ2 x2
between the expansion and divergence angles with respect to the Vulcain 0.1312 0.500 6.716 35.025 6.500 15.74
magnitude of length reduction. It can be seen from the contour plot Vulcain 2 0.1370 0.500 7.629 37.000 5.500 18.62
that a considerable reduction in nozzle length (i.e., less than the 80% SSME 0.1309 0.392 8.803 37.000 5.300 23.52
bell [10]) is restricted to high-angle designs. This restriction may
account for the lack of interest in an arc-based contour for use in early
nozzle design [11,12], as the low area ratios used inherently preclude
the use of high angles in contour generation. Table 2 Comparison of arc-based and conventional
nozzle contour length
C. Parametric Study Model xn ΔL, % ΔT W , %
To provide an initial estimation of the viability for the arc-based Vulcain 15.13 −3.910 0.389
design method, a preliminary analysis was conducted to verify that a Vulcain 2 17.09 −8.220 0.829
reduction in nozzle length relative to a conventional design would SSME 20.20 −14.12 1.432
be possible. A parametric study of existing contours was used to
SCHOMBERG, OLSEN, AND DOIG 191
A nozzle length-weight coefficient of 0.1 was selected for this study Inlet
based on the nozzle-to-engine weight fraction of 0.13 and 0.135 in the Outlet 15°
SSME [16] and Vulcain [17] engines, respectively. A value of 0.1
would yield a conservative estimate of the expected improvement Axis
while taking into account the difference between actual contour length 15rt 15rt
and axial distance, which was found to be less than 0.5% for the design 45° 2rt
cases considered here. Although a performance increase on the order of 4.5rt 0.25rt
0.3–1.5% may seem trivial, the increase in thrust coefficient as a result
of the Vulcain 2 upgrade was on the order of 1% [14]. rt=33.54mm 300rt
D. Design Case Fig. 5 Numerical domain including inlet and downstream exhaust
To explore the performance characteristics of the arc-based design dimensions (not to scale).
method, two scaled variants of the axisymmetric Vulcain nozzle were
selected for analysis. The nozzle variants were based on a scaling pro
cedure geared toward adaptation of the contour using a nondimen grid independence and ensuring that the selected turbulence model
sional length parameter. In these cases, the length parameter selected was capable of describing relevant flow effects. The existing [15] S3
was the radius at the nozzle exit (S1) and throat (S3). Both the S1 and nozzle was selected for the verification process because of a higher
S3 nozzles have previously been validated and shown to adequately correlation with the Vulcain nozzle than the S1 design. The
represent flow structure within the Vulcain nozzle [15]. Table 3 verification process was focused on minimizing uncertainty of the
outlines the geometric parameters for the S1 and S3 nozzles and the calculated thrust at full-flowing conditions and the wall pressure
corresponding reduction in contour length as a result of the arc-based distribution under separated flow conditions. Full-flowing conditions
design method. were replicated using a inlet-to-exit pressure ratio (PR) of 50,
The expansion contour for both nozzles was kept consistent with consistent with existing data [15]. Comparatively, separated flow
that used in [15]. The arc-based approach to the turning curve again conditions were generated using a PR of 20, where a “restricted”
resulted in a reduction in overall nozzle length when compared to the shock separation [21] would be expected to occur [15].
existing design. Although the area ratio between the S1 and S3
nozzles was similar, the magnitude of the length reduction in the S3 A. Geometric Domain
nozzle was much greater. For simplicity, the turning curve was kept to The computational domain consisted of the convergent–divergent
a single arc, allowing the wall angle to be described with respect to nozzle and a downstream exhaust region. The downstream region
the nondimensional length of the nozzle contour, as shown in Eq. (6). was sized to ensure the effect of the domain boundaries on the
With the wall angle known, both arc-based contours could be flowfield was negligible. A throat radius of 33.54 mm was selected in
described spatially throughout the domain. It is important to note that both the S1 and S3 nozzles, consistent with the existing design [15].
the design of all nozzle turning contours in the analysis was restricted As no information was provided about the convergent section of the
to a single circular arc, where a series of arcs may produce a greater nozzle, a constant angle contour at 45 deg was used from the curved
magnitude of length reduction. Based on the parametric study and throat region. The inlet-to-throat area ratio was sized to ensure the
design case, it could be concluded that the arc-based design method inlet pressure could be assumed to be equal to the total or stagnation
can produce a geometrically favorable contour for a high-area-ratio pressure with less than 0.5% error, and flow through the domain was
nozzle. controlled using a pressure inlet and pressure outlet boundary.
Figure 5 outlines the dimensions of the model and structure of the
numerical domain.
III. Numerical Model
B. Grid
A numerical approach was employed to assess flow behavior and
calculate thrust across the flow regimes expected to occur in a core The computational domain was discretized using a fully structured
stage engine. All results were generated using the commercially scheme comprising quadrilateral cells in all models. To determine the
available ANSYS Fluent 14.5 finite volume code. An axisymmetric effect of the downstream exhaust region on the solution and if
pressure-based momentum-coupled solver was selected to generate computational time could be further reduced, a “short” grid at each
all results. Although initially a subsonic flow solver, it has been density level was used that consisted of the nozzle geometry only. A
extended to cover high-speed flows with high accuracy [18,19]. The nondimensional wall distance y+ of one was maintained to ensure the
discretization of all flow parameters in space was achieved using presence of elements within the viscous region of the boundary layer.
second-order upwinding. Turbulence throughout the domain was All calculations were completed using the one-equation Spalart–
considered using a time or Reynolds-averaged approach due to the Allmaras (SA) [22] turbulence model. This model was selected due to
quasi-steady nature of the full-flowing nozzle. Air behaving in its proven capability of describing flowfields with shocks due to flow
accordance with the ideal gas law was used as the working fluid, and separation [23]. Grid independence was determined from the thrust
the inlet stagnation temperature was maintained at 450 K, consistent coefficient at full-flowing conditions and the static wall pressure at
with the conditions recorded in [15]. Viscosity was modeled using a separated flow conditions, as shown in Table 4 and Fig. 6, respectively.
three-coefficient Sutherland approximation due to the low stagnation The grid-based numerical uncertainty of the calculated thrust at
enthalpy [20]. full-flowing conditions was quantified using Roache’s grid conver
To satisfy solution convergence, mass flow rate monitors were set gence index [24] with an applied safety factor of three, and it was
on the nozzle inlet, the exit plane, and the domain outlet boundaries. found to be less than 1% in all cases. The influence of a downstream
Solution convergence was accepted when all surface monitor values
remained consistent over 500 iterations and continuity was satisfied Table 4 Numerical uncertainty of thrust coefficient at
from the recorded mass flux between the inlet and outlet boundaries. full-flowing conditions (PR = 50)
Verification of the numerical model was determined by establishing
Grid Elements CF ECS , % ECF , % ESF , %
Coarse (C) 460 × 135 1.319 0.345 0.220 ––
Table 3 Geometric parameters in the S1 and S3 nozzle Standard (S) 600 × 180 1.320 0.084 –– 0.675
Fine (F) 860 × 220 1.318 –– 0.027 0.112
Model rt1 y2 θ1 θ2 x2 rmn xn ΔL, %
C short 160 × 70 1.319 0.179 0.000 ––
S1 0.500 4.472 35.03 4.000 10.44 18.93 9.660 −7.370 S short 250 × 95 1.320 0.065 –– 0.212
S3 3.000 4.271 27.00 0.000 15.75 27.01 13.62 −13.49 F short 410 × 120 1.319 –– 0.000 0.069
192 SCHOMBERG, OLSEN, AND DOIG
0.12 0.12
Coarse Standard Fine Inviscid SA k-εrealizable k-ω SST
0.08 0.08
P/P0
P/P0
0.04 0.04
0 0
0 4 8 12 16 0 4 8 12 16
x/rt x/rt
Fig. 6 Effect of grid resolution on static wall pressure (PR = 20). Fig. 8 Effect of turbulence model on wall pressure distribution
(PR = 20).
0.16 4 0.16 4
0.12 3 0.12 3
Ma
P/P0
Ma
P/P 0
0.08 2 0.08 2
0.04 1 0.04 1
0 0 0 0
0 3 6 9 12 0 3 6 9 12
x/rt x/rt
Fig. 9 Comparison with existing data for the S1 nozzle (PR = 50). Fig. 11 Pressure and Mach number distributions in the S1 nozzle
(PR = 20).
Rao
0.12 3 Ma 3
P/P0
0.08 2
0
0.04 1
y/rt
0 3 6 9 12 15
6
0 0
0 4 8 12 16
x/rt
Arc
Fig. 10 Comparison with existing data for the S3 nozzle (PR = 50). 3
of flow turning toward the latter half of the arc nozzle was evident
0
from the static pressure and corresponding decrease in Mach number.
0 3 6 9 12 15
The greater exit pressure observed in the arc contour may reduce the x/rt
PR required for full-flowing conditions, and consequently reduce the Fig. 12 Contour plot of Mach number in the S1 nozzle (PR = 20).
B. Analysis
Rao Arc
The evaluation of both nozzle types required the observation of 0.2 5
flow parameters across a number of operating conditions. To achieve
this, the receiver pressure was varied to produce a number of pressure 0.16 4
ratios that would simulate nozzle behavior across the range of flow
conditions expected in a core stage nozzle. It is important to recall that 0.12 3
P/P0
Ma
any adverse effects observed in the arc-based design would nullify
the geometric advantage previously demonstrated. Differences 0.08 2
between the Rao and arc nozzles were compared using static pressure
distribution and contours of Mach number at separated conditions 0.04 1
and are shown in Figs. 11–14. Comparison at full-flowing conditions
was made using the thrust coefficient in the Rao-based nozzle and 0 0
weighted thrust coefficient in the arc-based nozzle, as shown in 0 4 8 12 16
Tables 6 and 7. x/rt
The onset of flow separation in the Rao nozzle was downstream of Fig. 13 Pressure and Mach number distributions in the S3 nozzle
the arc contour in both configurations under separated flow condi (PR = 20).
tions. However, when the relative length between the Rao and arc
contours is considered, the distance between the separation point and
Table 6 Variation in thrust coefficient in the S1 nozzle the nozzle exit was reduced in the arc-based design in both configu
rations. Observation of the magnitude of this offset suggested that it
PR CF Rao CF;w arc ΔCF , %
was approximately one-third of the magnitude of the total length
50 1.314 1.319 0.343 reduction in each case, which would imply a reduction in side loading
350 1.619 1.626 0.440
in the arc-based nozzle under separated flow conditions [15]. This
1000 1.653 1.660 0.428
benefit would be slightly decreased due to a small increase in the
pressure peak observed in the arc-based design. However, the
influence of this is difficult to quantify and unlikely to completely
Table 7 Variation in thrust coefficient in the S3 nozzle offset the lower moment arm evident in the arc-based design.
PR CF Rao CF;w arc ΔCF , %
The main discrepancy in flow structure observed in the attached
region between the Rao and arc contours was in the shape of the Mach
50 1.319 1.343 1.781
350 1.625 1.652 1.666
contours itself. In both cases, the lower initial rate of turning in the
1000 1.662 1.681 1.478 arc-based design reduced the severity of the compression wave
generated at the inflection between the expansion and turning curves.
194 SCHOMBERG, OLSEN, AND DOIG
Ma Ma
6 0 5 6 0 5
Rao Rao
3
3
0
0 3 6 9 12 15 18 21
y/rt
6 0
0 3 6 9 12 15
y/rt
Arc 6
3
0
Arc
0 3 6 9 12 15 18 21 3
x/rt
Fig. 14 Contour plot of Mach number in the S3 nozzle (PR = 20).
0
0 3 9 6 12 15
This effect was evident in both configurations, albeit far more x/rt
pronounced in the S3 nozzle. The general structure of the separated Fig. 16 Contour plot of Mach number in the S1 nozzle (PR = 14).
both arc-based nozzles. In both cases, this compression was not equal 0.15
Experimental Rao Arc
to the difference in nozzle length and was in agreement with the
calculated flow parameters.
The thrust coefficient was calculated at full-flowing overexpanded
(PR = 50), ideal (PR = 350), and underexpanded (PR = 1000) 0.1
flow regimes to represent the entire range of operating conditions
P/P0
expected to occur in a core stage nozzle. The thrust-to-weight scaling
factor introduced in Eq. (12) was applied to the arc-based design in
0.05
each case to ensure an accurate comparison between design methods.
The calculated thrust coefficient in the arc-based design was greater
than the equivalent Rao in all cases, where the magnitude of increase
was over 1% greater in the S3 configuration. However, the similarity 0
of the S1 nozzle geometric parameters with existing nozzle designs 0 3 6 9 12
suggests an expected increase in thrust coefficient as a result of the x/rt
arc-based design on the order of 0.4%. Fig. 17 Wall pressure distribution in the S1 nozzle (PR = 16).
C. Transition
Ma
To confirm that a reduction of nozzle length without adverse 6 0 5
separation effects was possible, an evaluation of the transition phases
in the S1 nozzle was conducted. This configuration was selected due
to the availability of experimental transition pressure information in Rao
3
[15] coupled with the slight increase in pressure peaks observed in the
separated flow condition in the arc-based design. Nozzle behavior
was evaluated at PRs of 14 and 16 to observe flow conditions at the
0
threshold between “free” and restricted shock separation during the
y/rt
0 3 6 9 12 15
transient startup condition [21]. Figures 15–18 outline the static wall 6
pressure distributions and contours of Mach number in the Rao and
arc nozzles at each PR. Arc
The correct prediction of a free or restricted separation flow regime 3
and a static pressure distribution that was consistent with the experi
mental data was observed in the Rao nozzle, providing additional
validation of the numerical model for use under separated flow 0
conditions. At a PR of 14, the difference between the numerical and 0 3 6 9 12 15
x/rt
Fig. 18 Contour plot of Mach number in the S1 nozzle (PR = 16).
0.12
Experimental Rao Arc
experimental static pressure readings was less than 4%. The main
discrepancy between the Rao and arc pressure distributions was a
0.08 predicted separation point location approximately 5% upstream in
the arc-based contour. An earlier point of separation was expected in
P/P0
the arc nozzle due to the higher initial rate of expansion shown to
occur. However, an overall decrease in the relative separation length
0.04
remained evident in the arc-based contour.
As expected, the restricted shock separation flow phenomenon
was predicted in both the Rao and arc nozzles at a PR of 16. The
0 separation location was predicted within 2% in both nozzle types,
0 3 6 9 12 where the main discrepancy of the pressure distribution was a lower
x/rt pressure peak evident in the numerical solution. This result can
Fig. 15 Wall pressure distribution in the S1 nozzle (PR = 14). largely be attributed to the use of a time-averaged solution to a highly
SCHOMBERG, OLSEN, AND DOIG 195
transient flow phenomenon [21]. Similar to the separated results at a [11] Sutton, G. P., History of Liquid Propellant Rocket Engines, AIAA,
PR of 20, a slight increase in pressure peak was observed in the arc Reston, VA, 2006, pp. 88–92, Chap. 4.
nozzle. However, in all cases, the difference in separation point [12] Rao, G. V. R., “Recent Developments in Rocket Nozzle
location and the increase in static wall pressure was minimal Configurations,” ARS Journal, Vol. 31, No. 11, 1961, pp. 1488–1494.
doi:10.2514/8.5837
compared to the reduction in nozzle length, suggesting that a [13] Davis, D. K., “Investigation of Optimization Techniques for Solid
reduction in side loads may also be possible through use of the arc- Rocket Motor Nozzle Contour,” 18th Joint Propulsion Conference,
based design method. AIAA Paper 1982-1188, 1982.
[14] Vuillermoz, P., Weiland, C., Hagemann, G., Aupoix, B., Grosdemange,
H., and Bigert, M., “Nozzle Design and Optimisation,” Progress in
V. Conclusions Astronautics and Aeronautics: Liquid Rocket Thrust Chambers, AIAA,
Two arc-based and equivalent conventional nozzle contours have Reston, VA, 2004, pp. 469–492, Chap. 13.
[15] Östlund, J., “Flow Processes in Rocket Engine Nozzles with Focus on
been evaluated numerically to compare the flow characteristics and
Flow Separation and Side-Loads,” Ph.D. Dissertation, Royal Inst. of
predicted thrust in a high-area-ratio nozzle contour. A comparison of Technology, Stockholm, 2002.
the arc-based contours to existing operational nozzles indicated that [16] Jewett, R. P., and Halchak, J. A., “The Use of Alloy 718 in the Space
an average reduction in length of 7.5% can be expected when the arc- Shuttle Main Engine,” Superalloys 718, 625 and Various Derivatives,
based method is applied to a core stage nozzle design. The calculated Minerals, Metal, and Material Soc., Warrendale, PA, 1991, pp. 749–760.
thrust coefficient in the arc-based design was 0.3–1.8% greater than [17] Kirner, E., Thelemann, D., and Wolf, D., “Development Status of the
the equivalent conventional nozzle across the range of full-flowing Vulcain Thrust Chamber,” Acta Astronautica, Vol. 29, No. 4, 1993,
conditions tested. Additionally, the flow parameters and transition pp. 271–282.
behavior under separated flow conditions suggested equivalent side doi:10.1016/0094-5765(93)90140-R
[18] Kurbatskii, K. A., and Montanari, F., “Application of Pressure-Based
loading and separated flow characteristics can be expected in the arc-
Coupled Solver to the Problem of Hypersonic Missiles with
based design. The result warrants further exploration of the arc-based Aerospikes,” 45th Aerospace Sciences Meeting, AIAA Paper 2007
method for the design of a high-area-ratio nozzle. 0462, 2007.
[19] Koutsavdis, E. K., and Stuckert, G., “A Numerical Investigation of the
Flow Characteristics of Plug Nozzles Using Fluent,” 40th Aerospace
References Sciences Meeting, AIAA Paper 2002-0511, 2002.
[1] Malina, F. J., “Characteristics of the Rocket Motor Based on the Theory [20] Sutherland, W., “The Viscosity of Gases and Molecular Force,”
of Perfect Gases,” Journal of the Franklin Institute, Vol. 230, No. 4, Philosophical Magazine, Vol. 36, No. 223, 1893, pp. 507–531.
1940, pp. 433–454. doi:10.1080/14786449308620508
doi:10.1016/S0016-0032(40)91348-5 [21] Frey, M., and Hagemann, G., “Restricted Shock Separation in Rocket
[2] Shapiro, A. H., “Nozzles for Supersonic Flow Without Shock Fronts,” Nozzles,” Journal of Propulsion and Power, Vol. 16, No. 3, 2000,
Journal of Applied Mechanics, Vol. 11, No. 2, 1944, pp. 93–100. pp. 478–484.
[3] Ahlberg, J. H., Hamilton, S., Migdal, D., and Nilson, E. N., “Truncated doi:10.2514/2.5593
Perfect Nozzles in an Optimum Nozzle Design,” ARS Journal, Vol. 31, [22] Spalart, P., and Allmaras, S., “A One-Equation Turbulence Model for
No. 5, 1961, pp. 614–620. Aerodynamic Flows,” La Recherche Aerospatiale, Vol. 1, No. 5, 1992,
doi:10.2514/8.5577 pp. 5–21.
[4] Hoffman, J. D., “Design of Compressed Truncated Perfect Nozzles,” [23] Stark, R., and Hagemann, G., “Current Status of Numerical Flow
Journal of Propulsion and Power, Vol. 3, No. 2, 1987, pp. 150–156. Prediction for Separated Nozzle Flows,” 2nd European Conference for
doi:10.2514/3.22967 Aerospace Sciences, von Karman Inst., EUCASS, Brussels, 2007,
[5] Guderley, G., and Hantsch, E., “Beste Formen fur Aschsensymmetri Paper 362.
sche Uberschallschubdusen,” Zeitschrift fur Flugwissenschaften, Vol. 3, [24] Roache, P. J., “Quantification of Uncertainty in Computational Fluid
No. 9, 1955, pp. 305–313. Dynamics,” Annual Review of Fluid Mechanics, Vol. 29, Jan. 1997,
[6] Rao, G. V. R., “Exhaust Nozzle Contour for Optimum Thrust,” Journal pp. 123–160.
of Jet Propulsion, Vol. 28, No. 6, 1958, pp. 377–382. doi:10.1146/annurev.fluid.29.1.123
doi:10.2514/8.7324 [25] Menter, F. R., “Two Equation Eddy-Viscosity Turbulence Models for
[7] Rao, G. V. R., “Approximation of Optimum Thrust Nozzle Contour,” Engineering Applications,” AIAA Journal, Vol. 32, No. 8, 1994,
ARS Journal, Vol. 30, No. 6, 1960, p. 561. pp. 1598–1605.
[8] Allman, J. G., and Hoffman, J. D., “Design of Maximum Thrust Nozzle doi:10.2514/3.12149
Contour by Direct Optimization Methods,” AIAA Journal, Vol. 19, [26] Shih, T. H., Liou, W. W., Shabbir, A., and Zhu, J., “A New k-ϵ Eddy-
No. 6, 1981, pp. 750–751. Viscosity Model for High Reynolds Number Turbulent Flows: Model
doi:10.2514/3.50999 Development and Validation,” Computers Fluids, Vol. 24, No. 3, 1995,
[9] Chen, J. S., and Freeman, J. A., “Thrust Chamber Performance Using pp. 227–238.
Navier–Stokes Solution,” NASA TR-D951729, 1984.
[10] Sutton, G. P., and Biblarz, O., Rocket Propulsion Elements, 7th ed., K. Frendi
Wiley, New York, 2001, pp. 75–81, Chaps. 2, 3. Associate Editor