A Detailed Presentation and Implementation Procedure

of Axisymmetric Method of Characteristics for Rocket

Nozzle Design
Written by
Kyril Palaveev
Sheffield Hallam University, Sheffield, UK

and

Dr. Torsten Schenkel

Sheffield Hallam University, Sheffield, UK

2nd September 2020

The goal of this report is to demonstrate and analyze the use of Method of Characteristics (MOC)
as a design tool for the geometry of the diverging section of a rocket nozzle. The purposes of such
nozzles are briefly linked to rocketry via the rocket equation and isentropic theory. Both 2D
and Axisymmetric MOC procedures are derived and implemented to demonstrate the mathematical
and physical differences and to prove the redundancy of 2D MOC. The Axisymmetric MOC
Procedure is thoroughly reviewed and convergence studies are performed to validate suitable
inputs yielding accurate nozzle geometries. Despite implementing the procedure for constructing
the kernel of the characteristic net only, the philosophy and implementation of the presented
Axisymmetric MOC procedure is sufficient to further refine a nozzle with the addition of new
boundary conditions. The produced nozzle is validated with the use of CFD software to
easily verify and discuss the outcomes of the numerical procedure.

Nomenclature
𝐴 = 𝐴𝑟𝑒𝑎 [𝑚2] 𝑚̇ = 𝑀𝑎𝑠𝑠 𝐹𝑙𝑜𝑤 𝑟𝑎𝑡𝑒 [𝑘𝑔⁄𝑠]
𝑋𝑐 = 𝐴𝑡 𝐶𝑜𝑚𝑏𝑢𝑠𝑡𝑖𝑜𝑛 𝐶ℎ𝑎𝑚𝑏𝑒𝑟 𝜈(𝑀) = 𝑃𝑟𝑎𝑛𝑑𝑡𝑙 − 𝑀𝑒𝑦𝑒𝑟 𝐴𝑛𝑔𝑙𝑒 [°]
𝑋𝑐𝑛 = 𝐴𝑡 𝐶𝑜𝑛𝑣𝑒𝑟𝑔𝑖𝑛𝑔 𝑁𝑜𝑧𝑧𝑙𝑒 𝑝 = 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒 [𝑃𝑎]
𝑋𝑑𝑛 = 𝐴𝑡 𝐷𝑖𝑣𝑒𝑟𝑔𝑖𝑛𝑔 𝑁𝑜𝑧𝑧𝑙𝑒 𝑟 = 𝑅𝑎𝑑𝑖𝑢𝑠 [𝑚]
𝑋𝑒 = 𝐴𝑡 𝐸𝑥ℎ𝑎𝑢𝑠𝑡 𝛾 = 𝑅𝑎𝑡𝑖𝑜 𝑜𝑓 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝐻𝑒𝑎𝑡 [−]
𝑋𝑡 = 𝐴𝑡 𝑇ℎ𝑟𝑜𝑎𝑡 𝑐𝑝 = 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝐻𝑒𝑎𝑡 𝑎𝑡 𝐶𝑜𝑛𝑠𝑡. 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒
𝛽 = 𝐶𝑜𝑛𝑣𝑒𝑟𝑔𝑖𝑛𝑔𝐴𝑛𝑔𝑙𝑒 [°] 𝑐𝑣 = 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝐻𝑒𝑎𝑡 𝑎𝑡 𝐶𝑜𝑛𝑠𝑡. 𝑉𝑜𝑙𝑢𝑚𝑒
𝑥, 𝑦 = 𝐶𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒𝑠 [𝑚] 𝐼𝑠𝑝 = 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝐼𝑚𝑝𝑢𝑙𝑠𝑒 [𝑠]
𝜌 = 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 [𝑘𝑔⁄𝑚3 ] 𝑎 = 𝑆𝑝𝑒𝑒𝑑 𝑜𝑓 𝑆𝑜𝑢𝑛𝑑 [𝑚⁄𝑠]
𝑢, 𝑣 = 𝐹𝑙𝑢𝑖𝑑 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝐶𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡𝑠 [𝑚⁄𝑠] 𝑇 = 𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 [𝐾]
𝑅 = 𝐺𝑎𝑠 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 [𝐽⁄𝑘𝑔𝐾] 𝐹𝑇 = 𝑇ℎ𝑟𝑢𝑠𝑡 [𝑁]
𝐿 = 𝐿𝑒𝑛𝑔𝑡ℎ [𝑚] 𝑅̂ = 𝑈𝑛𝑖𝑣𝑒𝑟𝑠𝑎𝑙 𝐺𝑎𝑠 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 [𝐽⁄𝑘𝑔𝐾 ]
𝜇 = 𝑀𝑎𝑐ℎ 𝐴𝑛𝑔𝑙𝑒 [°] 𝑣 = 𝑉𝑒ℎ𝑖𝑐𝑙𝑒 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 [𝑚⁄𝑠]
𝑀 = 𝑀𝑎𝑐ℎ 𝑁𝑢𝑚𝑏𝑒𝑟 [−] 𝜃 = 𝐼𝑛𝑓𝑙𝑒𝑐𝑡𝑖𝑜𝑛 𝐴𝑛𝑔𝑙𝑒(𝑀𝑂𝐶) [°]
𝑚 = 𝑀𝑎𝑠𝑠 [𝑘𝑔] 𝑉 = 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑉𝑒𝑐𝑡𝑜𝑟(𝑀𝑂𝐶) [𝑚⁄𝑠]
Contents
Introduction 1
Chemical Kinetics 2
Isentropic Theory 6
Method of Characteristics: Introduction 9
2D Method of Characteristics: Derivation 10

Irrotationality 10
Momentum Equation 11
Continuity Equation 12
Velocity Potential Equation 12
Summary of Equations 13

2D Method of Characteristics 14

Characteristics Equation 14
Compatibility Equation 16
Methodology 17
Implementation 20
Results 21
Discussion 21

Axisymmetric Method of Characteristics: Derivation 22

Irrotationality 22
Momentum Equation 22
Continuity Equation 23

Axisymmetric Method of Characteristics 23

Characteristic Equation 23
Compatibility Equation 25
Methodology 26
Results for Low-resolution Nozzle 31
Discussion for Low-resolution Nozzle 33
Results for High-resolution Nozzle 34
Discussion for High-resolution Nozzle 35
Implementation 35

Further Considerations 37

Converging Nozzle 38

CFD Validation 38

Mesh 39
Physics Setup 39
Convergence Validation 41
Results 42
Discussion 43

Conclusion 44
References 45
Appendices 46

Appendix A 46
Appendix B 46
Appendix C 47
 1

Introduction

The converging-diverging nozzle is, as suggested by its name, a hollow section of tubing with a
changing cross-sectional area. De Laval in 1888[1] discovered that such geometries are useful for
altering the properties of a fluid interacting internally with a nozzle. Despite its early use as a steam
accelerator for steam turbines, De Laval proved a concept of isentropic theory [2], [3] – the foundation of
compressible fluid dynamics. The isentropic relations are an essential foundation for every nozzle
however, the quasi-1D nature of the model restrict the user from producing precise and efficient
nozzle geometries. In need of more information about the flow inside a nozzle, L. Prandtl and A.
Busemann, introduced the second dimension in nozzle design theory in 1929 [1]. With this planar setup,
the ability of analyzing the radial (vertical) velocity component of the flow arose, replacing the
disadvantaged conical nozzle with more complex geometries, most popular of which is the bell
(contoured) nozzle. These complex geometries were necessary in order to produce nozzles with higher
efficiencies and to eliminate the production of shockwaves inside the nozzle. Constructing a 2D flow
field numerically, utilizing the Method of Characteristics (MOC) [1], is a great leap towards efficient
nozzles however, this more complex mathematical model is still considered detached from the
physical domain. The most basic implementation of MOC, the Planar MOC, is a dimensionless
solution which is not iterable. This eliminates the ability to introduce known information and data
from the physical domain to the procedure and the final results are usually with great error. A more
advanced, compared to the 2D MOC, procedure is the Axisymmetric MOC which follows the same
philosophy as 2D MOC however the coordinate system is cylindrical, adding more useful information
to the problem. This variant of the MOC allows to iterate some of the initial parameters which would
intern allow to control the output of the procedure. Both MOC procedures mentioned before are for
inviscid and irrotational flows – something considered inaccurate for a nozzle in the physical domain.
A procedure for rotational 3D MOC is developed however due to its complexity and infrequent
usability in industry of nozzle design it is considered outside the scope of this paper.

To familiarize the reader with the structure of this paper some brief chemical theory and calculations
are firstly presented in order to obtain the thermofluid properties of the working fluid. This section of
the report also acquaints the reader with justification for the chosen reactants and initial specifications
for the nozzle which is assumed to be part of a rocket motor assembly. The base parameters of the
nozzle are then obtained with the use of isentropic theory which automatically yields a conical nozzle.
Following the conical nozzle is the introduction of the philosophy and derivation of 2D Method of
Characteristics. The procedure is then implemented manually and a nozzle is constructed to
demonstrate its redundancy in rocket nozzle design. The already derived 2D MOC equations are then
modified for a cylindrical coordinate system and those partial differential equations are solved to an
algebraic form so that they are numerically solvable by implementing them in a computer programme
 2

written in Python 3.7. Once the nozzle contour coordinates are obtained they are discussed and the
geometry is validated using Ansys CFD.

Chemical Kinetics

Differing from De Laval’s steam nozzle, the propelling gas of a rocket is a result of combustion.
Requiring an oxidizer and fuel and also given the existence of many possible options for each of the
needed components for combustion it would be unreasonable to shift the focus from a nozzle design
procedure. Therefore this paper will adopt a seemingly popular oxidizer and fuel for demonstration
purposes. Liquid oxygen (𝑂2 ) and methane (𝐶𝐻4 ) are fluidic, non-hypergolic, and environmentally
friendly reactants with high availability and low risk factors. In chemistry, the term “chemical
[4]
equilibrium” refers to the ratio of these two compounds when they fully react. This mixture is
known as the stoichiometric ratio:

𝐶𝐻4 + 2𝑂2 →𝐶𝑂2 + 2𝐻2 𝑂 (1)

Although the expression above seems logical to implement, in practice, rocket motors burn an
[5]
optimum mixture ratio which is richer in oxidizer than the stoichiometric mixture ratio. The
optimum mixture ratio is desirable as it produces a heavier exhaust gas, increasing the inertia of the
working fluid and thus increasing thrust. Unfortunately the optimum mixture ratio cannot be
calculated accurately analytically. This is due to the generation of more complex compounds in the
product than those described in equation (1) and due to the amount of variables that differ for different
oxidizer to fuel (O/F) ratios. A numerical procedure for computing the fluid’s properties at different
mixture ratios exists. This numerical procedure is implemented into the NASA CEA[6] (Chemical
Equilibrium with Applications) software and is used for this report to determine the most efficient O/F
ratio. Prior to reviewing the data from the numerical chemical analysis, the reader is acquainted with
the derivation of a parameter known as the specific impulse that is essential to linking chemical
kinetics with motor performance.

First and foremost the Tsiolkovsky equation[7] is presented:

𝑚0
Δ𝑣 = 𝑢𝑒𝑞 ln ( ) (2)
𝑚𝑓

From a thermofluid perspective the equivalent exhaust velocity (𝑢𝑒𝑞) is defined as:

(𝑝𝑒 − 𝑝𝑎𝑡𝑚 )𝐴𝑒

𝑢𝑒𝑞 = + 𝑢𝑒 (3)
𝑚̇

From the thrust equation[1]:

 3

𝐹𝑇 = 𝑚̇ 𝑢𝑒 + (𝑝𝑒 − 𝑝𝑎𝑡𝑚)𝐴 𝑒 (4)

Dividing the right sides of equation (3) by the mass flow rate reveals:

𝑚̇𝑢𝑒𝑞 = (𝑝𝑒 − 𝑝𝑎𝑡𝑚 )𝐴𝑒 + 𝑚̇𝑢𝑒 (5)

Comparing equations (4) and (5) indicates that:

𝐹𝑇 = 𝑚̇ 𝑢𝑒 + (𝑝𝑒 − 𝑝𝑎𝑡𝑚 )𝐴 𝑒 = 𝑚̇ 𝑢𝑒𝑞 (6)

At this stage a crucial design decision is made. Rocket nozzles can operate in one of the three
following states. They can be over-expanded – when the ambient pressure is greater than the exhaust
pressure; under-expanded – when the ambient pressure is lower than the exhaust pressure; and
perfectly-expanded – when the ambient pressure is equal to the exhaust pressure. In over-expanded
nozzles an adverse pressure gradient is present thus forcing the flow to separate from the wall.
Turbulent flows are known for their chaotic behavior therefore undesirable for rocket nozzles. Having
an atmospheric pressure equal to the exhaust pressure is ideal and nozzles are usually designed around
this parameter. However, as rockets ascent the atmospheric pressure decreases putting the nozzle in
the second state of expansion. When under-expanded the efficiency of the nozzle decreases but the
overall thrust output increases as the equivalent velocity increases. A conclusion is therefore made that
nozzles are designed to be perfectly-expanded at the highest ambient pressure in which they operate.
Assuming that the rocket on which this nozzle is mounted takes-off at Mean Sea Level (MSL), the
highest ambient pressure would be 101325 [Pa]. Equating this value to the exhaust pressure would
yield:

𝐹𝑇 = 𝑚̇ 𝑢𝑒 + 0 = 𝑚̇ 𝑢𝑒𝑞 + 0 (7)
∴ 𝑢𝑒 = 𝑢𝑒𝑞 (8)

The impulse in physics is classified as the change in force over time. Known more commonly in
rocketry as the total impulse it can be mathematically expressed as:

𝐼 = ∫ 𝐹𝑇 𝑑𝑡 (9)

Substituting equation (7) into equation (9):

𝐼 = ∫ 𝑚̇ 𝑢𝑒 𝑑𝑡 (10)
 4

Since the exhaust velocity is a constant and the mass flow rate is a function of time:

𝐼 = 𝑚𝑢𝑒 (11)

The specific impulse is then defined as the total impulse over the accelerated weight. From Newton’s
2nd Law and equation (11):

𝐼
𝐼𝑠𝑝 = (12)
𝑚𝑔

Substituting equation (11) into equation (12):

𝑢𝑒
𝐼𝑠𝑝 = (13)
𝑔

The exhaust velocity, as previously mentioned, varies with the mass of the compounds. As weight
increases, velocity decreases but inertia increases. This suggests the parabolic nature of the specific
impulse when examined as a function of molar mass. Since the oxidizer and fuel have different masses
and mass fractions, altering the oxidizer to fuel ratio (O/F) will directly influence the molar mass of
the overall working fluid.

Pc = 40 [bar] Pc = 60 [bar] Pc = 40 [bar] Pc = 60 [bar]

Pc = 80 [bar] Pc = 100 [bar] Pc = 80 [bar] Pc = 100 [bar]
Ratio of Specific Heat (𝛾) [-]

330 1.260
Specific Impulse (Isp) [sec]

320 1.240
1.220
310
1.200
300
1.180
290
1.160
280 1.140
270 1.120
260 1.100
2.0 3.0 4.0 5.0 6.0 2.0 3.0 4.0 5.0 6.0
Mixture Ratio (O/F) [-] Mixture Ratio (O/F) [-]

(Graph 4.1) – Specific impulse variation over (Graph 4.2) – Ratio of specific heat variation
mixture ratio of liquid methane and liquid over mixture ratio of liquid methane and
oxygen. (Ref. [6]) liquid oxygen. (Ref. [6])
 5

Pc = 40 [bar] Pc = 60 [bar] Pc = 40 [bar] Pc = 60 [bar]

Pc = 80 [bar] Pc = 100 [bar] Pc = 80 [bar] Pc = 100 [bar]
Spec. Heat at Const. Pres. (Cp)

8000
3700.00
7000
3500.00

Temperature [K]
6000
5000 3300.00
[J/kgK]

4000 3100.00
3000
2900.00
2000
1000 2700.00
0 2500.00
2.0 3.0 4.0 5.0 6.0 2.0 3.0 4.0 5.0 6.0
Mixture Ratio (O/F) [-] Mixture Ratio (O/F) [-]

(Graph 4.3) – Specific heat at constant pressure (Graph 4.4) – Combustion temperature
variation over mixture ratio of liquid methane variation over mixture ratio of liquid methane
and liquid oxygen. (Ref. [6]) and liquid oxygen. (Ref. [6])

Strong mathematical evidence from Graph 4.1 suggests that the optimum mixture ratio is
approximately 3.2. Graphs 4.2, 4.3, and 4.4 however, peak at a mixture ratio of 4.0 and this is due to
the combustion efficiency. The ratio of specific heat is lowest at the highest combustion efficiency
which is not necessarily a desired quality since the higher the combustion efficiency the higher the
combustion temperature and from a practical standpoint the combustion chamber becomes less cost-
effective. For this nozzle design from the graphs above the presented in table 5.1 properties are
selected.

Combustion Pressure (𝒑𝒄 ) [𝑷𝒂]

8 [𝑀𝑃𝑎]

Specific Heat
Specific Gas Combustion Specific
Ratio of at Constant
Constant Chamber Impulse
O/F Ratio Specific Heat Pressure
(𝑹) Temperature (𝑰 𝒔𝒑 )
(𝜸) ( 𝒄 𝒑)
[𝑱/𝒌𝒈𝑲] [𝑲] [𝒔𝒆𝒄]
[𝑱/𝒌𝒈𝑲]
3.2 1.1877 382.08 2428.60 3538.69 313.303

(Table 5.1) – Gas properties of nozzle fluid for chosen mixture ratio

This concludes the section on chemical kinetic. The obtained parameters are essential for the next
section of the report.
 6

Isentropic Theory

Compressible flow can be algebraically expressed with the isentropic equations which demonstrate the
relationship between gas properties at two locations and the Mach number of the working fluid. To
understand the limitations, boundaries, and outcomes of isentropic theory the mass flow rate, equation
(20) and the speed of sound equation (21) are examined.

𝑚̇ = 𝜌𝐴𝑢 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (14)

𝑎 = √𝛾𝑅𝑇 (15)

Since the velocity in equation (14) can be expressed as Mach number:

𝑢
𝑀= (16)
𝑎

Equation (14) is written in the form:

𝑚̇ = 𝜌𝐴𝑀√𝛾𝑅𝑇 (17)

The ideal gas Law is used to represent the ideal gas constant as follows:

𝛾𝑝
𝑅= (18)
𝜌𝑇

The equation for continuous flow and the energy equation [3] yield:

1 2 2
𝑀 𝑎 + 𝑐𝑝 𝑇 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (19)
2

Where it is important to note that:

𝑅 = 𝑐𝑝 − 𝑐𝑣 (20)
𝑐𝑝
𝛾= (21)
𝑐𝑣
From equations (19), (20), and (21) the following equation yields:

1 2 𝑎2
𝑢 + = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (22)
2 𝛾−1
 7

Since equations (22) and/or (19) are constant along a streamline the isentropic ratios are derived:

𝛾
𝑝𝑥 𝛾 − 1 2 𝛾−1
= (1 + 𝑀𝑥 ) (23)
𝑝0 2
𝑇𝑥 𝛾 −1 2
= (1 + 𝑀𝑥 ) (24)
𝑇0 2
1
𝜌𝑥 𝛾 − 1 2 −𝛾−1 (25)
(
= 1+ 𝑀𝑥 )
𝜌0 2

(26)

From thermofluid dynamics, the term “isentropic” refers to the mathematical model of a fluid that is
adiabatic and reversible. The adiabatic process is associated with its unrealistic lack of heat and mass
transfer outside the analyzed system. This allows for the constant mass flow rate statement made
earlier. Furthermore the derivation of the isentropic relations includes the assumption that the entropy
along a streamline is constant. The term “reversible” indicates that isentropic theory does not account
for shockwaves which are usually observed in less complex nozzle geometries.

From the chosen combustion pressure and the CEA-calculated combustion temperature and gas
properties the isentropic properties of the nozzle are calculated at the chamber, throat and exhaust
(Table 8.1). It is important to note that the used isentropic process utilizes constant gas properties
which create a less accurate model than the “frozen” and “equilibrium” processes which are known for
a variable gas constant.

Combustion Chamber (𝑋𝑐 ) Throat (𝑋𝑡 ) Exhaust (𝑋𝑒 )

Pressure [𝑃𝑎] 𝟖 𝟎𝟎𝟎 𝟎𝟎𝟎.𝟎 4 536 585.5 𝟏𝟎𝟏 𝟑𝟐𝟓. 𝟎
Temperature [𝐾] 𝟑𝟓𝟑𝟖.𝟔𝟗 3 236.88 1 779.96
Density [𝑘𝑔/𝑚3 ] 5.92 3.67 0.15
Mach number [−] 𝟎. 𝟎𝟎𝟎 𝟏. 𝟎𝟎𝟎 3.2554
Velocity [𝑚/𝑠] 𝟎. 𝟎 941.4 2923.0

(Table 8.1) – Results from isentropic calculations for a converging-diverging nozzle. Note:
Values in bold are known values (boundary conditions).

The area-Mach number relationship (equation (27)) and the mass flow rate equation (equation (14))
are used to determine the areas and radii of the throat and exhaust of the nozzle (Table 9.1).
 8

𝛾+1
𝛾+1 𝛾 − 1 2 2(𝛾−1)
𝐴𝑒 𝛾+1 −
2(𝛾−1) (1 + 𝑀𝑒 ) (27)
=( ) 2
𝐴𝑡 2 𝑀𝑒

The combustion chamber area cannot be evaluated using isentropic theory due to the zero-velocity
assumption at that point. Through experimental data is has been determined that a suitable combustion
chamber area is equal to 3 times that of the throat [8].

Combustion Chamber (𝑋𝑐 ) Throat (𝑋𝑡 ) Exhaust (𝑋𝑒 )

Area [𝑚 2 ] 0.002310 0.000770 0.007841
Radius [𝑚] 0.027119 0.01566 0.049958

(Table 9.1) – Results of area and radii for critical segments of a converging-diverging nozzle

The isentropic ratios are then plotted as functions of Mach number (graph 10.1). The area ratio is
plotted in graph 10.2.

Pressure Ratio 20
1.2
Temperature Area Ratio
1
Ratio 15
Density Ratio
Area Ratio [-]

0.8
Ratio [-]

0.6 10

0.4
5
0.2

0 0
0 1 2 3 4 0 1 2 3 4
Mach Number [-] Mach Number [-]

(Graph 10.1) – Isentropic ratios versus Mach (Graph 9.2) – Area ratio versus Mach number
number for converging-diverging nozzle for converging-diverging nozzle

With the use of the isentropic process the critical areas of the nozzle are found (i.e. chamber, throat,
and exhaust area). These are crucial boundary conditions for the upcoming numerical design
procedure as all MOC procedures are detached from the area-Mach-number relationship and therefore
one cannot design a nozzle contour using MOC without having a finite value for the exhaust radius to
iteratively steer the numerical solution to.

Method of Characteristics: Introduction

 9

A technique considered most popular in computational fluid dynamics is used to numerically compute
the needed nozzle contour to optimize the behavior of the fluid [1,9,10]. Some basic conservation laws of
fluid dynamics state that flows in different regimes behave differently and these laws are in the form
of partial differential equations (PDEs), meaning that they have multiple independent variables. A
general description of the method of characteristics (MOC) is a technique for solving the PDEs in
question to obtain sets of ordinary differential equations (ODEs) which are more easily solvable.
However, a PDE cannot be reduced to an ODE for any input – the spatial locations (i.e. coordinates in
a fluid field) on which the PDEs can become ODEs are known as characteristic curves (i.e.
characteristics). In essence, the goal of the MOC procedures is to numerically compute the location of
these characteristics and where they intersect. The means of computing those locations are presented
in the following sections of this paper.

Nozzle flows go through three out of the four flow regimes that are currently explained with the aid of
partial differential equations. In the converging section of the nozzle the flow is initially in the
subsonic regime. The PDEs describing the behavior of subsonic flows are elliptical, indicating that the
solutions would be imaginary as the discriminant of the differential equations would be negative
(i.e. ⅅ = √−∞ < 𝑎 < 0). When the flow reaches the throat region of the nozzle, the flow reaches
sonic speeds – Mach number is 1. Therefore the discriminant of the PDE describing the flow would be
zero indicating a parabolic solution with two identical real solutions (i.e. ⅅ = √0 ∴ 𝑥1 = 𝑥2 ).
Increasing the Mach number even further yields hyperbolic solutions – ones that have two distinct and
real solutions. The number and type of solutions in fact indicate the number of characteristic lines
passing though one fluid parcel (vector point) with some coordinates. For subsonic flows (elliptical
PDEs) those characteristic lines are imaginary and cannot be plotted in a real 2D Cartesian coordinate
system. Furthermore the MOC is a pseudo-graphical method for obtaining a nozzle contour and
constructing a converging section using this method would be difficult and in most cases impractical
as simpler methods for constructing a subsonic contour exist. The scope of implementation of the
Method of Characteristics is from the throat to the exhaust of the nozzle since the differential
equations are hyperbolic and are also classified as first order partial differential equations – the type of
PDEs that MOC is typically applied to. More in-depth explanations of the philosophy of the Method
of Characteristics and solutions of PDEs can be found in textbooks and papers such as in reference [1,
10, and 11]

2D Method of Characteristics: Derivation

Defining a flow field begins with setting a notation for the fluid parcels’ spatial properties. Any vector
is considered to have three velocity components. Along the x-axis the velocity vector has a component
𝑢, along the y-axis the velocity vector has a velocity component 𝑣, and along the z-axis – a velocity
 10

component 𝑤. From basic vector theory the resultant of 𝑢, 𝑣, and 𝑤 yields the direction and magnitude
of the vector.

Irrotationality

Irrotationality of the vector 𝑉 is mathematically expressed as:

∇× 𝑉 = 0 (28)
𝑖̂ 𝑗̂ 𝑘̂
𝐶𝑢𝑟𝑙(𝑉) = ∇ × 𝑉 = | 𝜕 𝜕 𝜕| (29)
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑢 𝑣 𝑤

Solving the determinant using Cramer’s rule:

𝜕 𝜕 𝜕 𝜕 𝜕 𝜕
∇ × 𝑉 = 𝑖̂ ( 𝑤 − 𝑣) − 𝑗̂ ( 𝑤 − 𝑢) + 𝑘̂ ( 𝑣 − 𝑢) = 0 (30)
𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑧 𝜕𝑥 𝜕𝑦

Since equation (41) is equal to zero, the terms in the brackets (irrotationality conditions) must be
equal:

𝜕 𝜕
𝑤= 𝑣 (31)
𝜕𝑦 𝜕𝑧
𝜕 𝜕
𝑤= 𝑢 (32)
𝜕𝑥 𝜕𝑧
𝜕 𝜕
𝑣= 𝑢 (33)
𝜕𝑥 𝜕𝑦

Momentum Equation

By substituting the vector V into Euler’s equation:

ⅅ𝑉
𝜌 = −∇𝑝 (34)
ⅅ𝑡

Where the material derivative:

ⅅ 𝜕 𝜕 𝜕 𝜕 𝜕
≡ +𝑢 +𝑣 +𝑤 = + (∇ ∙ 𝑉) (35)
ⅅ𝑡 𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑡

𝜕𝑢 𝜕𝑣 𝜕𝑤 𝜕𝑝
𝜌𝑢 + 𝜌𝑣 + 𝜌𝑤 =− (36)
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥
 11

Substituting the irrotationality conditions and solving for each component (x, y, and z):

𝜕𝑝 1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
− 𝑑𝑥 = 𝜌 𝑑𝑥 + 𝜌 𝑑𝑥 + 𝜌 𝑑𝑥 (37)
𝜕𝑥 2 𝜕𝑥 2 𝜕𝑥 2 𝜕𝑥
𝜕𝑝 1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
− 𝑑𝑦 = 𝜌 𝑑𝑦 + 𝜌 𝑑𝑦 + 𝜌 𝑑𝑦 (38)
𝜕𝑦 2 𝜕𝑦 2 𝜕𝑦 2 𝜕𝑦

𝜕𝑝 1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
− 𝑑𝑧 = 𝜌 𝑑𝑧 + 𝜌 𝑑𝑧 + 𝜌 𝑑𝑧 (39)
𝜕𝑧 2 𝜕𝑧 2 𝜕𝑧 2 𝜕𝑧

Adding equations (37), (38), and (39) together yields:

𝜕𝑝 𝜕𝑝 𝜕𝑝 1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
− ( 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧) = 𝜌 𝑑𝑥 + 𝜌 𝑑𝑥 + 𝜌 𝑑𝑥 +
𝜕𝑥 𝜕𝑦 𝜕𝑧 2 𝜕𝑥 2 𝜕𝑥 2 𝜕𝑥
1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
𝜌 𝑑𝑦 + 𝜌 𝑑𝑦 + 𝜌 𝑑𝑦 + (40)
2 𝜕𝑦 2 𝜕𝑦 2 𝜕𝑦
1 𝜕𝑢2 1 𝜕𝑣 2 1 𝜕𝑤 2
𝜌 𝑑𝑧 + 𝜌 𝑑𝑧 + 𝜌 𝑑𝑧
2 𝜕𝑧 2 𝜕𝑧 2 𝜕𝑧
Knowing that:

𝑉 2 = 𝑢2 + 𝑣 2 + 𝑤 2 (41)

Equation (40) reduces to:

𝜕𝑝 𝜕𝑝 𝜕𝑝 1 𝜕𝑉 2 1 𝜕𝑉 2 1 𝜕𝑉 2
− ( 𝑑𝑥 + 𝑑𝑦 + 𝑑𝑧) = 𝜌 𝑑𝑥 + 𝜌 𝑑𝑦 + 𝜌 𝑑𝑧 (42)
𝜕𝑥 𝜕𝑦 𝜕𝑧 2 𝜕𝑥 2 𝜕𝑦 2 𝜕𝑧

Since on both sides of equation (42) all components are present the PDE is classified as an exact
differential. It can therefore be written in the form:

1
𝑑𝑝 = − 𝜌𝑑𝑉 2 = −𝜌𝑉𝑑𝑉 (43)
2

Known as the Momentum equation, equation (43) is one of the three conditions needed to define the
boundary conditions for Method of Characteristics.

Continuity Equation

For any 3-dimensional fluid parcel, its mass can be mathematically expressed as:

𝑚 = 𝜌𝑉 = ∫ 𝜌𝑑𝑉 = ∫ 𝜌 𝑑𝑥 𝑑𝑦 𝑑𝑧 (44)
 12

Typically, the continuity equation is expressed in the time domain therefore a time derivative is
introduced in the conservation of mass equation. However, in the case of this paper, steady-state flow
is assumed therefore in need of a special form of the continuity equation. This simplification allows
the assumption that through time, the mass entering or leaving the system through any of the surfaces
of the fluid parcel is constant:

𝜕 𝜕 𝜕
𝜌𝑢 + 𝜌𝑣 + 𝜌𝑤 = 0 (45)
𝜕𝑥 𝜕𝑦 𝜕𝑧

Simplifying equation (45):

∇ ∙ (𝜌𝑢 + 𝜌𝑣 + 𝜌𝑤) = 0 (46)

∇ ∙ 𝜌(𝑢 + 𝑣 + 𝑤) = 0 (47)
∇ ∙ 𝜌𝑉 = 0 (48)

Velocity Potential Equation

Defining the velocity potential function (Ф) as:

𝜕Ф 𝜕Ф 𝜕Ф
𝑢= 𝑣= 𝑤= (49)
𝜕𝑥 𝜕𝑦 𝜕𝑧

Recall and substitute equation (49) into the momentum equation (43):

Ф2𝑥 + Ф2𝑦 + Ф2𝑧

𝑑𝑝 = −𝜌𝑑 ( ) (50)
2
The differential form of the speed of sound expression is:

𝑑𝑝
𝑎2 = (51)
𝑑𝜌

Substituting equation (51) into (50) yields:

𝜌𝑑 Ф2𝑥 + Ф2𝑦 + Ф2𝑧

𝑑𝜌 = − ( ) (52)
𝑎2 2

Taking the derivative for each direction of equation (52):

𝜕𝜌 𝜌
= − 2 (Ф𝑥𝑥 Ф𝑥 + Ф𝑦𝑥 Ф𝑦 + Ф𝑧𝑥 Ф𝑧 ) (53)
𝜕𝑥 𝑎
 13

𝜕𝜌 𝜌
= − 2 (Ф𝑥𝑦Ф𝑥 + Ф𝑦𝑦 Ф𝑦 + Ф𝑧𝑦 Ф𝑧) (54)
𝜕𝑦 𝑎
𝜕𝜌 𝜌
= − 2 (Ф𝑥𝑧 Ф𝑥 + Ф𝑦𝑧 Ф𝑦 + Ф𝑧𝑧 Ф𝑧 ) (55)
𝜕𝑧 𝑎

Substituting equation (49) into the continuity equation (equation (48)):

𝜕 𝜕 𝜕
𝜌Ф𝑥 + 𝜌Ф𝑦 + 𝜌Ф𝑧 = 0 (56)
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕 𝜕 𝜕
𝜌(Ф𝑥𝑥 + Ф𝑦𝑦 + Ф𝑧𝑧) + Ф𝑥 + Ф𝑦 + Ф𝑧 = 0 (57)
𝜕𝑥 𝜕𝑦 𝜕𝑧

Equations (53), (54), and (55) are substituted into equation (57):

Ф2𝑥 Ф2𝑦 Ф2𝑧

(1 − 2
) Ф𝑥𝑥 + 1 − 2 ) Ф𝑦𝑦 + (1 − 2 ) Ф𝑧𝑧 −
(
𝑎 𝑎 𝑎
(58)
2Ф𝑥 Ф𝑦 2Ф𝑥 Ф𝑧 2Ф𝑧 Ф𝑦
2 Ф𝑥𝑦 − − 2 Ф𝑥𝑧 − − Ф𝑧𝑦 = 0
𝑎 𝑎 𝑎2

Since this paper will not address any 3-dimensional flow fields, a 2-dimensional version of the
velocity potential equation is presented:

Ф2𝑥 2Ф𝑥 Ф𝑦 Ф2𝑦

(1 − ) Ф 𝑥𝑥 − Ф 𝑥𝑦 + ( 1 − ) Ф𝑦𝑦 = 0 (59)
𝑎2 𝑎2 𝑎2

Summary of Equations

For convenience table 14.1 summarizes the previously derived equations:

Irrotationality ∇× 𝑉 = 0

Momentum Equation 𝑑𝑝 = −𝜌𝑉𝑑𝑉

Continuity Equation ∇ ∙ 𝜌𝑉 = 0
Ф2𝑥 2Ф𝑥 Ф𝑦 Ф2𝑦
Velocity Potential Equation (1 − ) Ф 𝑥𝑥 − Ф 𝑥𝑦 + ( 1 − ) Ф𝑦𝑦 = 0
𝑎2 𝑎2 𝑎2

(Table 14.1) – Summary of differential equations for 2D MOC

2D Method of Characteristics

Characteristics Equation
 14

From the previously derived equations we can treat the following terms as displayed:

Ф𝑥 = 𝑢 Ф𝑦 = 𝑣 (60)
(𝑑𝑥)Ф𝑥𝑥 + (𝑑𝑦)Ф𝑥𝑦 = 𝑑𝑢 (𝑑𝑥)Ф𝑥𝑦 + (𝑑𝑦)Ф𝑦𝑦 = 𝑑𝑣 (61)

Equations (60) and (61) are substituted into the velocity potential equation:

𝑢2 2𝑢𝑣 𝑣2
(1 − ) Ф 𝑥𝑥 − Ф + ( 1 − ) Ф𝑦𝑦 = 0 (62)
𝑎2 𝑎 2 𝑥𝑦 𝑎2

From equations (61) and (62) two 3 by 3 determinants are created to solve for Ф𝑥𝑦:

𝑢2 𝑣2
1− 2 0 1− 2
| 𝑎 𝑎 |
𝑑𝑥 𝑑𝑢 0
𝐴 0 𝑑𝑣 𝑑𝑦
Ф𝑥𝑦 = = 2 (63)[1]
𝐵 𝑢 2𝑢𝑣 𝑣2
1− 2 − 2 1 − 2
| 𝑎 𝑎 𝑎 |
𝑑𝑥 𝑑𝑦 0
0 𝑑𝑥 𝑑𝑦

[1, 9, 10, 11]

Thoroughly explained in multiple textbooks and papers , a specific condition for determinant
𝐴 and 𝐵 exist. Valid results wouldn’t yield from equation (63) if the denominator is different than zero
since different inputs for 𝑑𝑥 and 𝑑𝑦 would yield identical values for 𝑑𝑢 and 𝑑𝑣 . However, for a
denominator of 0 in equation (63), the solution would be infinite which is untrue since the physical
domain suggests the possibility of a finite solution. To force the solution of equation (63) to be finite,
determinant 𝐴 must also be 0. This indicates that the solution for Ф𝑥𝑦 must be indeterminate to yield
the location at which characteristic lines cross.

From equation (62), for 𝐵 = 0, the determinant yields a quadratic equation describing the slope of the
characteristic lines:

2
𝑢2 𝑑𝑦 2𝑢𝑣 𝑑𝑦 𝑣2
(1 − ) ( ) − ( ) + ( 1 − )=0 (64)
𝑎 2 𝑑𝑥 𝑐ℎ𝑎𝑟 𝑎 2 𝑑𝑥 𝑐ℎ𝑎𝑟 𝑎2

Using the quadratic formula on equation (64):

2𝑢𝑣 2𝑢𝑣 𝑢2 𝑣2
𝑑𝑦 − 2 ± √ 2 − 4 (1 − 2 ) (1 − 2 )
𝑎 𝑎 𝑎 𝑎
( ) = (65)
𝑑𝑥 𝑐ℎ𝑎𝑟 𝑢2
2 (1 − 2 )
𝑎
 15

Through algebraic manipulations and trigonometric functions, equation (65) is reduced to its simplest
form:

𝑑𝑦
( ) = tan(𝜃 ∓ 𝜇) (66)
𝑑𝑥 𝑐ℎ𝑎𝑟

where 𝜃 is the angle between the velocity vector and the x-axis and 𝜇 – the Mach angle – is the angle
between the velocity vector and the characteristic lines – therefore also called Mach lines (Figure 15.1).
It is obvious that two solutions exist for a set of angles. 𝜃 + 𝜇 is known as a left-running characteristic
(𝐶+) and 𝜃 − 𝜇 – a right-running characteristic (𝐶−).

(Figure 15.1) – A graphical representation of characteristic lines and a stream line in a Cartesian
coordinate system
Compatibility Equation

The characteristics equation informs the user on the slope of a characteristic line. Now the focus is on
how the slope changes with respect to space. In figure 15.1 it is shown that characteristic lines are
curved. The compatibility equation is the solution of determinant 𝐴 when it is equal to 0.

Using Cramer’s rule on the numerator in equation (63) gives:

𝑢2 𝑣2
(1 − ) 𝑑𝑢𝑑𝑦 + ( 1 − ) 𝑑𝑥𝑑𝑣 = 0 (67)
𝑎2 𝑎2
 16

Rearranging:

𝑢2
(1 − )
𝑎 2 𝑑𝑦 𝑑𝑣
− 2 ( ) = (68)
𝑣
(1 − 2 ) 𝑑𝑥 𝑐ℎ𝑎𝑟 𝑑𝑢
𝑎

Again through algebraic and trigonometric manipulations equation (68) reduces to the compatibility
equation:

√𝑀2 − 1
𝑑𝜃 = ∓ 𝑑𝑉 (69)
𝑉

As previously said, the compatibility equation describes the change in resultant angle ( 𝑑𝜃) with
respect to the change in resultant velocity ( 𝑑𝑉). Equation (69) is also identical to the equation
developed by Prandtl and Meyer used to mathematically model expansion waves. Integrating both
sides of equation (69) with some limits yields the Prandtl-Meyer function (𝜈(𝑀)) from which the
compatibility equation is reduced to:

𝜃 ∓ 𝜈 = 𝑐𝑜𝑛𝑠𝑎𝑛𝑡 (70)

For 𝜃 + 𝜈 the constant is known as 𝐾− and runs along the 𝐶− characteristics. For 𝜃 − 𝜈 the constant is
𝐾+ and runs along the 𝐶+ characteristic. It is of great importance that the reader acknowledges that
equation (69) is dimensionless. This indicates that there is no requirement for previously obtained
information and values which makes 2D MOC a rarely used method in the practical design of nozzles.
Other MOC procedures, one of which is later introduced, have different compatibility equations
eliminating the issue with dimensionless nozzles.

Methodology

To utilize the characteristic and compatibility equations boundary conditions are needed to initiate a
calculation. To demonstrate the implementation of 2D MOC, 8 characteristic lines are calculated and
plotted. The reader however, must be aware that in reality the number of characteristic lines is infinite.
This indicates that with an increase in the number of characteristic lines the accuracy of the nozzle
contour increases.

The two crucial boundary conditions for a 2D MOC minimum-length nozzle are the maximum initial
expansion angle – 𝜃𝑚𝑎𝑥 located at the throat and the exhaust angle – 𝜃𝑒 (Figure 17.1).
 17

(Figure 17.1) – Mapping of the solutions along a 2D MOC characteristic net for 8
characteristic lines

Since all last points on each characteristic line in the characteristic net (point 8, 16, 23, 29 etc.) cannot
change their flow angle from there-on due to the previously defined compatibility equations the
following statement is true:

𝜃43 − 𝜈43 = 𝜃44 (𝑒) − 𝜈44 (𝑒) (71)

Due to the perfect expansion assumption made in the “Chemical Kinetics” section of this report, the
exhaust velocity should have an axial component only (𝑣 = 0). This indicates that the velocity vector
is horizontal and that 𝜃44 (𝑒) = 0:

𝜈43 = 𝜈44 (72)

The exhaust Mach number was previously said to be a driving parameter. Using the Prandtl-Meyer
function the exhaust Prandtl-Meyer angle (𝜈44 ) can be numerically extrapolated. For the parameters
previously presented:

𝜈(3.2554) = 72.035° (73)

For that Mach number and the known 𝜃43 the compatibility constant along that characteristic line for
all points is:

𝜃43 + 𝜈43 = 𝐾− 43 = 72.035° (74)

Examining figure 17.1 shows that the point at the throat, point 43 and 44 lie on the same set of
characteristic lines therefore from equations (73) and (74):

𝜃𝑚𝑎𝑥 + 𝜈𝑡ℎ𝑟𝑜𝑎𝑡 = 𝐾− 43 (75)

 18

The expansion at the throat is a Prandtl-Meyer expansion therefore:

1
𝜃𝑚𝑎𝑥 = (𝐾− 43 ) (76)
2

Or:

𝜈(𝑀𝑒 ) 𝜈(3.2554)
𝜃𝑚𝑎𝑥 = = = 36.0175 ° (77)
2 2

At this stage a dilemma with one boundary condition arises. Since nozzles are choked and the fluid is
sonic at the throat the undefined expression of tan(90) occurs in the characteristic equation. A
workaround for this problem is setting the throat Mach number to slightly larger than 1 or setting the
throat vector angle to slightly larger than 0. In this case:

𝜃1 = 1.0175 ° (78)

The smaller 𝜃1 is the more accurate the results. Propagation of the calculation downstream is, as
mentioned, based on the constant change in the velocity vector angle. This constant change is known
as Δ𝜃 and is calculated as follows:

𝜃𝑚𝑎𝑥 − 𝜃1
Δ𝜃 = (79)
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝑙𝑖𝑛𝑒𝑠 − 1

For the demonstration nozzle the change in that angle is:

Δ𝜃 = 5° (80)

To begin extrapolating the values for the 44 points, attention is called to the compatibility constant
along the first left-running characteristic line (𝐾+ )1~8 . At point 8 it is know that the flow direction
angle (𝜃8 ) is half the exhaust Prandtl-Meyer angle. Moreover, the Prandtl-Meyer angle at this point is
also equal to half the exhaust Prandtl-Meyer angle thus equal to the flow direction angle. The
𝐾+constant along the first characteristic line is then always equal to zero for all 8 points.

𝜃1~8 = 𝜈1~8 (81)

𝐾+1~8 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 0 (82)

The remaining variables can be then calculated with the aid of the following equations:
 19

1
𝜃 = (𝐾− + 𝐾+ ) = 𝜈 (83)
2
1
𝜈 = ( 𝐾− − 𝐾+ ) = 𝜃 (84)
2
𝐾− = 𝜃 + 𝜈, 𝐾−1~8 = 2𝜃 = 2𝜈 (85)
𝐾+ = 𝜃 − 𝜈, 𝐾+1~8 = 0 (86)

For the remaining characteristic lines it is known that every point, coincident with the x-axis, has a
flow direction angle of 0° and every following 𝜃 for a point is increased by Δ𝜃. Since the right-running
compatibility constant ( 𝐾+ ) must always be equal to the Prandtl-Meyer angle at the wall of the nozzle
and it changes with a constant step of 2Δ𝜃, which was established from the first characteristic line, it is
possible to calculate all points in the characteristic net.

When discussing 𝜃𝑚𝑎𝑥 it was confirmed that beyond the points running along the last right-running
characteristic line, the flow-field does not change direction therefore:

𝜃8 = 𝜃9 𝜃34 = 𝜃35
𝜃16 = 𝜃17 𝜃38 = 𝜃39
𝜃23 = 𝜃24 𝜃41 = 𝜃42
𝜃29 = 𝜃30 𝜃43 = 𝜃44

The slope of the wall between two wall-points can therefore be calculated by taking the mean value of
their inflection angles:

𝜃9 + 𝜃17
9~17 =
2
𝜃17 + 𝜃24
17~24 =
2 (87)
𝜃24 + 𝜃30
24~30 =
2
(… )
Once all 4 variables for each point are known, from the Prandtl-Meyer angle the Mach number is
approximated numerically and the Mach angle is calculated form the Mach number. Recall that the
slope of the characteristic lines is equal to the sum of the inflection angle and Mach angle. The nozzle
can then be plotted graphically which for small numbers of characteristic lines is suitable.

Implementation

Since 2D MOC is not used for the final design of the nozzle, developing an automated, software-
driven calculation is considered unnecessary for this report. The calculation was semi-automatically
 20

implemented into an MS Excel Spreadsheet with one Macro written in Visual Basic Applications
(VBA). The defined function is the numerical extrapolation of the Prandtl-Meyer function since it
cannot be rearrange analytically (Figure 20.1).

(Figure 20.1) – A numerical procedure for calculating Mach number for given Prandtl-Meyer
angles in Visual Basic Applications

Once the slopes of the characteristic lines are obtained the Solidworks CAD Software is used to
graphically derive the nozzle contour. If the reader decides to explore an automatic implementation of
2D MOC, coordinates for each point can be derived through trigonometry. Triangulating between
three points by knowing their slopes is an elementary process. This would allow the user to use 2D
MOC to produce nozzle contours with higher accuracy.

Results

(Figure 21.1) – Graphical Results of 2D MOC for a C-D nozzle with an exhaust Mach number of
3.2554 and 8 characteristic lines

Discussion

When discussing the mathematics behind 2D MOC some of the issues were outlined. Proof from the
results section shows that the exhaust nozzle radius is 99.97% larger than the radius calculated using
 21

isentropic theory. The main reason for this error is that the compatibility equation is detached from the
area-Mach number relationship. Even though none of the MOC procedures are somehow connected to
that relationship more advanced procedures of MOC are iterable, allowing the exhaust area to be
adjusted. 2D MOC progresses downstream with some seemingly arbitrary constant which is controlled
only by the number of characteristic lines and the exhaust Mach number and is in no way “adjustable”
or iterable making it unsuitable for the design of nozzles that are required to reduce the pressure of the
flow in a controlled manner so that the exhaust pressure is some desired value. Given that 2D MOC is
dimensionless and no additional inputs are required (such as pressure distributions and temperatures)
this solution of the partial differential equations is irrelevant and not beneficial when designing nozzle
contours. An experimental nozzle with the same number of Mach lines but with a lower Mach number
of 2.688 was also made to assess the progression of the error. It is found that the error between the
isentropic exhaust radius and the 2D-MOC-obtained exhaust radius decreases to a value of 50.39%. In
Anderson’s book – “Modern Compressible Flow with Historical Perspective” a 2D MOC nozzle with
7 characteristic lines and an exhaust Mach number of 2.4 is designed. Through approximate
measurements it is discovered that that nozzle has an area ratio of 5.0625. For a Mach number of 2.4
however, isentropic theory indicates that the area ratio should in fact be between 3.01 and 2.403 for
specific heat ratios between 1.2 and 1.4. The error for Anderson’s 2D MOC nozzle is therefore
between 22.89 and 31.1%. An interesting discovery is that with a decrease of exhaust Mach number,
the error between the isentropic and 2D MOC exhaust radii decreases. Nevertheless the conclusion is
that 2D MOC is unsuitable for designing nozzles which are expected to comply with isentropic theory.
The method also does not allow for any means to alter the generated error in a meaningful and
controlled way.

Axisymmetric Method of Characteristics: Derivation

The general principle of Axisymmetric MOC is identical to that of 2D MOC. The difference arises
from the coordinate system in which Axisymmetric MOC lies. The previously defined Cartesian
system is replaced by a cylindrical coordinate system:

𝑟(𝑣) , 𝜙(𝑤), 𝑥(𝑢) (88)

To better present the change in coordinate system, the following expressions equate each of the
cylindrical components to the Cartesian:

𝑟 = √𝑥 2𝐶𝑎𝑟𝑡 + 𝑦 2 (89)

𝑦
𝜙 = tan−1 ( ) (90)
𝑥 𝐶𝑎𝑟𝑡
𝑥= 𝑧 (91)
 22

Note: caution is advised when working with cylindrical coordinates as they may mislead the reader if
compared to Cartesian coordinates.

Irrotationality

For this setup irrotational flow is expressed as:

𝑙̂ 𝑟𝑚̂ 𝑛̂
1 𝜕 𝜕 𝜕|
∇×𝑉 = | =0 (92)
𝑟 𝜕𝑟 𝜕𝜙 𝜕𝑥
𝑣 𝑟𝑤 𝑢

It is evident from equation (92) that the unit vectors are different compared to the Cartesian coordinate
system. The unit vectors for a cylindrical coordinate system change with location therefore popular ly
classified as a local type coordinate system.

Cramer’s rule to solve equation (92) indicates that:

𝜕𝑢 𝜕𝑣
= (93)
𝜕𝑟 𝜕𝑥

Momentum Equation

Recall the Momentum equation:

𝜌𝑑 2
𝑑𝑝 = 𝜌𝑉𝑑𝑉 = − (𝑢 + 𝑣 2 + 𝑤 2 ) (94)
2

Due to the type of symmetry of the nozzle the 𝑤 term is voided.

𝜌
𝑑𝜌 = − (𝑢𝑑𝑢 + 𝑣𝑑𝑣) (95)
𝑎2

where:

𝑑𝑝
𝑎2 =
𝑑𝜌

From equation (95) for each spatial term:

𝜕 𝜌 𝜕𝑢 𝜕𝑣
𝜌 = − 2 (𝑢 +𝑣 ) (96)
𝜕𝑥 𝑎 𝜕𝑥 𝜕𝑥
 23

𝜕 𝜌 𝜕𝑢 𝜕𝑣
𝜌 = − 2 (𝑢 +𝑣 ) (97)
𝜕𝑟 𝑎 𝜕𝑟 𝜕𝑟

Continuity Equation

Returning to the Continuity equation:

∇ ∙ 𝜌𝑉 = 0 (98)
𝜕 𝜕 𝜕 𝜌𝑣
(𝜌𝑢) + (𝜌𝑣) + (𝜌𝑤) + =0 (99)
𝜕𝑥 𝜕𝑟 𝜕𝜙 𝑟

Again due to the type of symmetry of the nozzle the 𝜙 term is voided.

𝜕 𝜕 𝜌𝑣
(𝜌𝑢) + (𝜌𝑣) + =0 (100)
𝜕𝑥 𝜕𝑟 𝑟

Axisymmetric Method of Characteristics

Characteristic Equation

Substituting the Momentum equation terms ((96) & (97)) into the Continuity equation – equation
(100) – yields:

𝜕𝑢 𝜌 𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜌 𝜕𝑢 𝜕𝑣 𝜌𝑣
(− 2 (𝑢 + 𝑣 )) + (− 2 (𝑢 + 𝑣 )) + =0 (101)
𝜕𝑥 𝑎 𝜕𝑥 𝜕𝑥 𝜕𝑟 𝑎 𝜕𝑟 𝜕𝑟 𝑟

From the velocity potential equation which remains the same as in 2D MOC equation (101) reduces
to:

𝑢2 𝜕𝑢 𝑣𝑢 𝜕𝑣 𝑣𝑢 𝜕𝑢 𝑣 2 𝜕𝑣 𝑣
(1 − 2
) − 2 − 2 + ( 1 − 2
) =− (102)
𝑎 𝜕𝑥 𝑎 𝜕𝑥 𝑎 𝜕𝑟 𝑎 𝜕𝑟 𝑟

The irrotationality condition gave equation (93) which is now substituted into equation (102):

𝑢2 𝜕𝑢 𝑣𝑢 𝜕𝑣 𝑣 2 𝜕𝑣 𝑣
(1 − 2
) − 2 2 + ( 1 − 2
) =− (103)
𝑎 𝜕𝑥 𝑎 𝜕𝑥 𝑎 𝜕𝑟 𝑟

The derivatives are then:

𝜕𝑢 𝜕𝑢
𝑑𝑢 = 𝑑𝑥 + 𝑑𝑟 (104)
𝜕𝑥 𝜕𝑟
 24

𝜕𝑢 𝜕𝑣
𝑑𝑢 = 𝑑𝑥 + 𝑑𝑟 (105)
𝜕𝑥 𝜕𝑥
𝜕𝑣 𝜕𝑣
𝑑𝑣 = 𝑑𝑥 + 𝑑𝑟 (106)
𝜕𝑥 𝜕𝑟

Similar to the 2D MOC philosophy the three derivatives are solved as two 3 by 3 determinants:

𝑢2 𝑣2
1− 2 0 1 − 2
| 𝑎 𝑎 |
𝑑𝑥 𝑑𝑢 0
𝜕𝑣 𝐴 0 𝑑𝑣 𝑑𝑦
= = (107)
𝜕𝑥 𝐵 𝑢2 2𝑢𝑣 𝑣2
1 − 2 − 2 1− 2
| 𝑎 𝑎 𝑎 |
𝑑𝑥 𝑑𝑟 0
0 𝑑𝑥 𝑑𝑟

The condition that 𝐴/𝐵 must equal 0/0 to obtain valid results holds. More interestingly determinant 𝐵
is identical to the 2D MOC dividing determinant with the exception that the y-coordinate is occupied
by the r-coordinate. This indicates that the slope of the characteristic lines is again governed by the
sum or difference of the flow angle and Mach angle. The characteristic equation for Axisymmetric
MOC is then:

𝑢𝑣 𝑢2 + 𝑣 2
𝑑𝑟 − 2 ∓ √( )−1
𝑎 𝑎2
( ) = (108)
𝑑𝑥 𝒄𝒉𝒂𝒓 𝑢2
1− 2
𝑎

Or:

𝑑𝑟
( ) = tan(𝜃 ∓ 𝜇) (109)
𝑑𝑥 𝒄𝒉𝒂𝒓

Compatibility Equation

For determinant 𝐴 equal to 0:

𝑢2 𝑣 𝑑𝑟
𝑑𝑣 (1 −
)−
𝑎2 𝑟 𝑑𝑢
=− 2 (110)
𝑑𝑢 𝑣 𝑑𝑟
(1 − 2 ) ( )
𝑎 𝑑𝑥 𝑐ℎ𝑎𝑟

Equation (110) can also be written in the form:

 25

𝑢2 𝑣 𝑑𝑟
𝑑𝑣 (1 −)
𝑎 2 𝑑𝑟
=− ( ) − 𝑟 𝑑𝑢2 (111)
𝑑𝑢 𝑣 2 𝑑𝑥 𝑐ℎ𝑎𝑟 𝑣
(1 − 2 ) (1 − 2 )
𝑎 𝑎

Substituting the characteristic equation (equation (109)) into equation (111):

𝑢2 𝑢𝑣
√(
𝑢2 + 𝑣 2
)−1 𝑣 𝑑𝑟
𝑑𝑣 (1 −
) − ∓
𝑎2 𝑎2 𝑎2
=− − 𝑟 𝑑𝑢2 (112)
𝑑𝑢 𝑣2 𝑢2 𝑣
(1 − 2 ) 1− 2 (1 − 2 )
𝑎 ( 𝑎 ) 𝑎

Knowing that the resultant vector is the products of the corresponding trigonometric function of the
flow angle and the velocity components and that the Mach number ratio equation (112) reduces to the
differential form of the axisymmetric compatibility equation:

𝑑𝑉 1 𝑑𝑟
𝑑𝜃 = ∓√𝑀2 − 1 ∓ (113)
𝑉 √𝑀2 − 1 ∓ 𝑐𝑜𝑡𝑔(𝜃) 𝑟

In addition to the Prandtl-Meyer flow equation, equation (113) also describes the change in flow angle
relative to the radial location of the vector. Moreover due to the Prandtl-Meyer flow equation, equation
(113) can also be written as:

1 𝑑𝑟
𝑑(𝜃 + 𝜈) = (114)
√𝑀2 − 1 ∓ 𝑐𝑜𝑡𝑔(𝜃) 𝑟
1 𝑑𝑟
𝑑(𝜃 − 𝜈) = − (115)
√𝑀2 − 1 ∓ 𝑐𝑜𝑡𝑔(𝜃) 𝑟

Methodology

The goal of this section is to reduce the differential equations to sets of algebraic equations which are
easily solvable compared to the PDEs. The characteristics equation is solved algebraically as a system
of two equations.

Note: All algebraic equations are demonstrated for points 1, 2, and 10 from figure 17.1.

𝑟2 − 𝑟1
= tan(𝜃1 + 𝜇1 )
𝑥2 − 𝑥1
{𝑟 −𝑟 (116)
2 10
= tan(𝜃10 − 𝜇10 )
𝑥 2 − 𝑥10

Solving the system for 𝑥 and 𝑟:

 26

𝑟1 − 𝑟10 + 𝑥10 tan(𝜃10 − 𝜇10 ) − 𝑥1 tan(𝜃1 + 𝜇1 )

∴ 𝑥2 = (117)
tan (𝜃10 − 𝜇10 ) − tan(𝜃1 + 𝜇1 )

∴ 𝑟2 = 𝑟1 + (𝑥 2 − 𝑥1 )tan(𝜃1 + 𝜇1 ) (118)

The compatibility differential equation is solved using forward finite differencing to reduce it to an
algebraic equation (Note: from here-on, V denotes scalar velocity, not a velocity vector):

𝑉10 − 𝑉1 tan(𝜇1 ) sin (𝜇1 ) sin (𝜃1 ) (𝑥10 − 𝑥1 )

0= − (𝜃10 − 𝜃1 ) tan (𝜇1 ) − (119)
𝑉1 cos(𝜃1 + 𝜇1 ) 𝑟1

𝑉10 − 𝑉2 tan(𝜇 2 ) sin(𝜇 2 ) sin(𝜃2 ) (𝑥10 − 𝑥 2 )

0= − (𝜃10 − 𝜃2 ) tan(𝜇 2 ) − (120)
𝑉2 cos(𝜃2 + 𝜇 2 ) 𝑟2

Equations (119) and (120) are rearranged in terms of the velocity – 𝑉 and flow angle – 𝜃.

1
𝑉2 = [𝑐𝑜𝑡𝑔 ( 𝜇1 )(1 + ℙ 1 ( 𝑥2 − 𝑥1 )) + 𝑐𝑜𝑡𝑔( 𝜇10 )(1 + ℚ10 ( 𝑥2 − 𝑥10 )) + 𝜃10 − 𝜃1 ]
𝑐𝑜𝑡𝑔 ( 𝜇1 ) 𝑐𝑜𝑡𝑔( 𝜇10 )
+
(121)
𝑉1 𝑉10

𝑉2 − 𝑉1
𝜃2 = 𝜃1 + 𝑐𝑜𝑡𝑔(𝜇1 )( ) − ℙ1 (𝑥 2 − 𝑥1 ) (122)
𝑉1

where:

tan( 𝜇1 ) sin(μ1 ) sin (θ1 )

ℙ1 =
𝑟1 cos( 𝜃1 + 𝜇1 )

tan(𝜇10 ) sin(μ10 ) sin (θ10 )

ℚ10 =
𝑟10 cos(𝜃10 − 𝜇10 )

One specific case however, cannot be solved with equations (121) and (122). That case is when the
two points from which the third is solved are on the centerline (such as 2 and 10). Since they have a
radial coordinate of 0 and a flow direction of 0, a division by zero is created. Shapiro (1954)
demonstrated a solution to this problem[11] by taking the limiting form of ℙ1 & ℚ10 :

tan(𝜇1 ) sin (μ1 ) sin( θ1 )

lim ℙ1 = lim (123)
𝑟1→0 𝑟1 →0 𝑟1 cos(𝜃1 + 𝜇1 )
𝜃1 →0 𝜃1→0
 27

tan (𝜇10 ) sin (μ10 ) sin(θ10 )

lim ℚ10 = lim (124)
𝑟10 →0 𝑟10→0 𝑟10 cos(𝜃10 − 𝜇10 )
𝜃10 →0 𝜃10 →0

Equations (123) and (124) reduce to:

𝜃3
lim ℙ1 ≅ (125)
𝑟1→0 𝑟3
𝜃1 →0

𝜃3
lim ℚ10 ≅ (126)
𝑟10→0 𝑟3
𝜃10→0

At this stage, the above equations are sufficient to provide a vector’s position, flow angle, and velocity.
To find the Mach angle for the next point along the characteristic net, the Mach number is needed. The
temperature at that point can therefore be calculated using:

𝑉22
𝑇2 = 𝑇𝑐 − (127)
2𝑐𝑝

The Mach number is therefore:

𝑉2
𝑀2 = (128)
√𝛾𝑅𝑇2

And finally the Mach angle:

1
𝜇 2 = asin ( ) (129)
𝑀2

Recall that the compatibility equations were solved with a first order accurate forward difference. The
typical engineering grade accuracy for any differential equation is second order accurate. In need to
increase the accuracy of the results, an iterative procedure is introduced which consists of the exact
same algebraic equations but with the mean values of the slopes between points 1~2 and 10~2.

Second iteration for the x-coordinate:

tan (𝜃10 − 𝜇 10 ) + tan ( 𝜃2 − 𝜇 2 ) tan ( 𝜃1 + 𝜇 1 ) + tan ( 𝜃2 + 𝜇 2 )

𝑟1 − 𝑟10 + ( ) 𝑥 10 − ( ) 𝑥1
𝑖𝑖 2 2
𝑥2 = (130)
tan (𝜃10 − 𝜇 10 ) − tan ( 𝜃1 + 𝜇 1 ) + tan ( 𝜃2 − 𝜇 2 ) − tan (𝜃2 + 𝜇 2 )
( )
2

Second iteration for the radial coordinate:

 28

tan(𝜃1 + 𝜇1 ) + tan (𝜃2 + 𝜇 2 )

𝑟2𝑖𝑖 = 𝑟1 + (𝑥 2 − 𝑥1 ) ( ) (131)
2

Second iteration for the velocity:

ℕ + ℂ − 𝜃1 + 𝜃10
𝑉2𝑖𝑖 = (132)
4((𝑉1 + 𝑉2 )−1 (tan(𝜇 1) + tan(𝜇 2))
−1 + (𝑉 + 𝑉 )−1 (tan(𝜇 ) + tan(𝜇 ))−1 )
10 2 10 2

where:

2 2𝑉1 ℙ1 + ℙ2
ℕ= [ + (𝑥 2 − 𝑥1 )]
tan (𝜇1 ) + tan(𝜇 2 ) 𝑉1 + 𝑉2 2

2 2𝑉10 ℚ10 + ℚ2
ℂ= [ + (𝑥 2 − 𝑥10 )]
tan(𝜇10 ) + tan (𝜇 2) 𝑉10 + 𝑉2 2

Second iteration for the flow angle:

2 2(𝑉𝑖𝑖2 − 𝑉1 ) ℙ1 + ℙ2
𝜃2𝑖𝑖 = 𝜃1 + ( − (𝑥2 − 𝑥1 )) (133)
tan ( 𝜇 1 ) + tan (𝜇 2 ) 𝑉𝑖𝑖2 + 𝑉1 2

Once the second iteration is complete, the values from that calculation are treated as the values that
would continue the calculation along the net:

𝑥 𝑖𝑖
2 → 𝑥2

𝑟2𝑖𝑖 → 𝑟2

𝑉2𝑖𝑖 → 𝑉2

𝜃2𝑖𝑖 → 𝜃2

The iterative process can be repeated multiple times and to determine a suitable iteration count a
convergence study is conducted by measuring the exhaust radius of the nozzle for some constant and
arbitrary inputs and a varying iteration count (Graph 29.1).
 29

Exhaust Radius
1.50%

1.00%
Error

0.50%

0.00%

-0.50%
0 2 4 6 8 10 12
Number of Iterations

(Graph 29.1) – Convergence study for iteration count of Axisymmetric MOC

Graph 29.1 indicates that after the 5th iteration, the error – difference in exhaust diameter for some
arbitrary and constant inputs and a varying iteration count – oscillates between 0.02 and -0.02% which
is a reasonable accuracy for a theoretical nozzle design. With a further increase in iterations, the
oscillations dampen and the error reduces even further.

To finalize the Axisymmetric MOC procedure the wall points and nozzle contour slopes must be
found. From planar MOC, the assumption that the flow does not accelerate or change direction
between the last point in the characteristic net and the wall can be used. The inclination or slope of the
wall contour segments, as in 2D MOC, is the average between the flow direction of the previous wall
point and the flow direction of the last point of the net on the characteristic line on which the
calculated wall point lies (e.g. the slope for line 17~24 is the average of 𝜃9 & 𝜃16 ). We can therefore
write the following equations based on the formulae for the intersection of two lines:

𝜃 + 𝜃8
−𝑥16 tan(𝜃16 + 𝜇16 ) + 𝑟16 + 𝑥9 tan ( 𝑇 ) − 𝑟8
𝑥17 = 2
𝜃 + 𝜃8 (134)
tan ( 𝑇 ) − tan(𝜃16 + 𝜇16 )
2

𝑟17 = tan(𝜃16 + 𝜇16 )(𝑥17 − 𝑥 16) + 𝑟16 (135)

The number of boundary conditions for Axisymmetric MOC is larger than that of 2D MOC. The
variability in these boundary conditions is also greater. The first and most obvious is the initial
expansion angle which for 2D MOC was derived to be half the exhaust Prandtl-Meyer angle. No such
equation exists in the world of Axisymmetric MOC therefore the initial expansion angle can
theoretically be any value between 0 and 90 degrees. This parameter heavily influences the exhaust
radius and its selection is based on complicated concepts which are later discussed.
 30

With this procedure, no information is given about any point. Furthermore recall that the algebraic
equation examples initiated from points 1 and 10 to calculate point 2. This implies that an important
boundary is the axis of symmetry or more specifically – the points that lie on the x-axis (Figure 30.1).

(Figure 30.1) – A flow chart of the Axisymmetric MOC numerical procedure

The most important step for initializing Axisymmetric MOC is calculating the gas properties and
position of the points on the x-axis. An obvious parameter is that these point have a radial coordinate
of 0. The spacing between each of the points is considered constant and a distribution length is
selected (distance from point 1 to 43). Usually the characteristic net length is larger than the isentropic
throat radius. Now the spatially defined points require an input of information that would yield gas
properties at those locations. Specifying the change in pressure as a function of the characteristic net
length is ideal. From the known pressure at each point and the known exhaust pressure, the Mach
number is calculated and therefore all other properties can also be calculated. The pressure distribution
can be either theoretical or experimental. Due to the lack of any experimental data, the theoretical
approach is chosen. Suitable for supersonic flows are two types of pressure distributions. The
parabolic and cubic pressure distribution functions are both functions of axial distance. The difference
is that the cubic distribution is more configurable due to a dimensionless factor known as the q-factor
(equations (136) & (137)).

Parabolic pressure distribution[9,10]:

ln(𝑝𝑥 ) = 𝐿−2 2
𝑛 (𝑥 − 𝐿𝑛 ) (ln(𝑝𝑡 ) − ln(𝑝𝑒 )) + ln(𝑝𝑒 ) (136)
 31

Cubic pressure distribution[9,10]:

𝑞𝐿𝑛 + 2 (ln( 𝑝𝑡 ) − ln( 𝑝𝑒 )) 2𝑞𝐿𝑛 + 3(ln (𝑝𝑡 ) − ln( 𝑝𝑒 ))

ln(𝑝𝑥 ) = ( ) 𝑥3 − ( ) 𝑥 2 + 𝑞𝑥 + ln(𝑝𝑡 ) (137)
𝐿3𝑛 𝐿2𝑛

1800
1600
1400
1200
Pressure [kPa]

1000
800
600
400
200
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Characteristic Net Length [m]

parabolic cubic, q = 0 cubic, q = -2 cubic, q = -5 cubic, q = 5

(Graph 31.1) – Graphical representation of equations 136 and 137

It is important to note that extreme q-factor values are not recommended if the distribution is not
verified against a distribution that is known to be accurate. Moreover, for a pressure distribution in the
supersonic regime, negative q-factors are used.

Results for Low-resolution Nozzle

With the methodology behind the method complete, it is of vital importance that the iterative
adjustments of the boundary conditions are outlined. It is known that the isentropic exhaust radius is
the radius that the Axisymmetric MOC nozzle exhaust must reach. For the 8-characteristic-lines
example, an arbitrary expansion angle of 0.01 [rad] is selected. It is arbitrary since the calculated
region with the procedure does not allow for analysis in the transonic region. The pressure distribution
type is selected to be parabolic and an initial characteristic net length of 0.1 [m] is selected. The
calculation is initiated and for that first iteration, the Axisymmetric MOC exhaust radius calculated to
be 0.04296 [m] – 16.3% smaller than the isentropic radius of 0.04996 [m]. At this stage it is
recommended to increase the characteristic net length which would explicitly increase the exhaust
radius. Through a steady increase in characteristic net length, an error of 0.36% is achieved for a net
length of 0.135 [m]. Typically errors below ±1% are considered valid results.
 32

(Graph 32.1) – Nozzle contour generated with Axisymmetric MOC with 8 characteristic lines

(Graph 32.2) – A graphical demonstration of the influence of the expansion angle on nozzle
length

Discussion for Low-resolution Nozzle

Note that the characteristic net crosses the nozzle wall contour. This is actually a common occurrence
in nozzles with low initial expansion angles and can still be considered a valid result for this specific
procedure. Proof supporting this statement is in Graph 32.2 – two nozzles with identical exhaust radii
 33

but different expansion angles. It is evident that the characteristic net is adjusted to arrive at the same
radius. A larger expansion angle requires a shorter net length and vice versa. It is also seen that the
larger the expansion angle the shorter the nozzle. In rocketry this is especially desirable as it reduces
the weight of the nozzle. However, excessive expansion angles cause geometrical issues. When the
nozzle turns into the flow abruptly an oblique shock is produced. An analogy can be made with
supersonic flow over a ramp. Other issues include an abrupt separation of the flow which may separate
at different times at different radial section of the nozzle which would cause an oscillatory regime.
These regimes are especially dangerous in practice as they can lead to structural failures of the nozzle.

It was mentioned that the number of characteristic lines also alter the exhaust radius. The example
nozzle with 8 Mach lines is not suitable for a final nozzle contour design. To deduce a suitable number
of characteristic lines for a final design a convergence study is presented. The process behind the study
is solving a nozzle contour for some constant set of inputs and increasing the number of characteristic
lines gradually. The exhaust radius is recorded and the difference between the previous and next
calculation is calculated.

Number of Characteristic Lines Number of Characteristic Lines

0 50 100 150 200 250 300 350 8%
135
6%
Exhaust Radius [mm]

145
Difference

155
165 4%
175
185
195 2%
205
215
225 0%
235 0 50 100 150 200 250 300 350
245

(Graph 33.1) – Number of characteristic lines (Graph 33.2) – Number of characteristic lines
convergence study measured by exhaust convergence study measured by percentile
radius. error of the difference in exhaust diameter for
previous and current number of characteristic
lines.

Examining Graphs 33.1 and 33.2 indicates that 200 characteristic lines produce a contour with an error
of 1.674%. For this paper it is considered that this error is suitable for an accurate diverging contour
thus that nozzle design is produced with 200 characteristic lines.

Results for High-resolution Nozzle

 34

(Graph 34.1) - Nozzle contour generated with Axisymmetric MOC with 200 characteristic lines

(Graph 34.2) – Close-up on the kernel of an Axisymmetric MOC-generated nozzle with 200
characteristic lines
 35

(Graph 35.1) – Close-up of the undefined transonic region in the kernel

Discussion for High-resolution Nozzle

For the chosen parameters shown in Graph 34.1, the produced contour is considered converged with
an exhaust radius error minimized to 0.1% when compared to the exhaust radius calculated using
isentropic theory. The slightly over-expanded nozzle is with a radius of 0.04996 [m] and the resolution
of the contour is generally high with the exception of the transonic region. Graph 35.1 acquaints the
reader with a region marked as undefined. This region occurs because of the nature of nozzle flows.
Typically, immediately after the throat, the flow changes its properties for a relatively short distance –
the flow expands abruptly. Due to this fact, the characteristic lines closer to the throat are curved
towards the exhaust significantly more than lines towards the end of the characteristic net. This region
can be neglected for nozzles of smaller dimensions but larger nozzles require a fix to ensure that
shockwaves and flow separation do not occur. Furthermore the expansion angle of 5 degrees is chosen
arbitrarily since at this stage there are no means of calculating a precise expansion angle. Typically
expansion angles for contoured nozzles are between 0 and 8 degrees depending on their application. A
distinctive for Axisymmetric MOC contour is the length of the nozzle caused by the boundary
conditions of the method. In rocketry, nozzles with such high efficiencies are rarely use due to their
dimensions and weight.

Implementation

The reader may have suspected that the Axisymmetric MOC procedure is more complicated and labor-
intensive than 2D MOC. Therefore the author of this paper developed a computer programme written
in Python 3.7. The software consists of equations used to calculate the isentropic relations of the
 36

wanted nozzle followed by the algebraic equations for Axisymmetric MOC. The algorithm is semi-
automated at this stage – excluding the automatic adjustment of the characteristic net length. As seen
in the results section, only the characteristic line intersections are displayed and this is solely to reduce
the usage of graphical and access memory. This is especially useful for large numbers of characteristic
lines. The software is intended to be available to the general public as a Windows executable (.exe)
and for that reason the author developed a Graphical User Interface (GUI) in order to be more easily
used by users.

The “space-marching” procedure of Axisymmetric MOC is achieved in the code by using lists which
are indexable and for-loops which call the points’ parameters in a specific order based on their index
to perform the calculation. All other implementations of various isentropic, thrust, and thermodynamic
calculations are trivial since they are not involved in any complicated repetitive loops, lists or
conditional statements. To simplify the use of the software a minimum input study was conducted to
reduce the number of inputs to a minimum and ensure that non-corresponding inputs cannot be entered.
The GUI and inputs of the software are shown in figure 36.1.

(Figure 36.1) – The Graphical User Interface of the developed Axisymmetric MOC nozzle
 37

contour designer.
Note: Instructions on how one can acquire and setup the software can be found in Appendix A.

Further Considerations

The most important part of Axisymmetric MOC is implemented and a nozzle contour is created.
However, further work is required which is classed as outside the scope of this paper to avoid
prolonging the paper to a great extent. The characteristic net demonstrated in this paper is known as
the kernel. The kernel is the most important section of the nozzle design procedure as it constructs the
greatest segment of the nozzle contour. A useful feature of Axisymmetric MOC is that for almost any
given boundary conditions a characteristic net can be constructed. Such example is the evaluation of
the undefined transonic region. Sauer’s method[9] can provide a parabolic sonic line at the throat which
can then be treated as a characteristic line and a net can be computed between the first kernel
characteristic line and the parabolic sonic line. This addition to the nozzle design procedure would
allow for a justified selection of the initial expansion angle. For the purposes of this paper the
expansion angle is selected based on common expansion angles for rocket nozzles. However, a point
where the nozzle is shortest and produces no shocks does exist and this point can be found through the
transonic region analysis procedure. A third characteristic net can also be constructed utilizing the top-
most layer of the kernel and the undefined region as a boundary condition. This net is referred to as the
boundary and it can span beyond the nozzle wall contour. The boundary region is used to determine
the location of the contour since the introduction of the transonic analysis procedure would
discontinue the currently used solution between the expansion nozzle contour and the diverging nozzle
contour (Figure 37.1).

(Figure 37.1) – Different regions of a nozzle that require separate characteristic nets for a
refined nozzle geometry
 38

Converging Nozzle

Now that the diverging section is obtained, the converging section must be designed. Due to the flow
being in the subsonic regime at that section of the nozzle, its geometry is not as critical. A conical
converging section would not satisfy our overall nozzle geometry as a sharp corner would be produced
at the throat which would separate the flow. Rao’s method[12] is a straightforward technique which
utilizes circular arcs. The radii of these arcs have been derived experimentally and have proven to
yield satisfactory converging nozzle contours. Typically Rao’s converging geometry consists of two
tangential circular arcs however, since we are not focusing on the design of the combustion chamber,
only one of the circular arcs is used and an additional tangential to that arc line is extended from the
end of circular arc to meet the desired combustion chamber radius (Figure 38.1).

(Figure 38.1) – Converging section with Rao’s radius.

The length of the circular arc is typically between 40 and 45 degrees and it can be plotted numerically
using the following equations:

𝑥 = 1.5𝑟𝑡 cos (𝛽) (138)

𝑟 = 1.5𝑟𝑡 sin(𝛽) + 2.5𝑟𝑡 (139)

for −135° < 𝛽 < −90°

CFD Validation

Despite the need of further refinement of the transonic and boundary region a CFD simulation of the
high-resolution nozzle contour is attempted to verify aspects of the nozzle, such as desired exhaust gas
 39

properties. The simulation also acquaints the reader with the behavior of nozzle flows along the axis of
symmetry.

Mesh

The generated axisymmetric nozzle contour is imported into the software and a domain is formed. A
fluid surface is created and split in multiple sections for ease of meshing.

(Figure 39.1) – Generated mesh for the computational domain of a rocket nozzle

The mesh in figure 39.1 has two important bias features one of which is towards the throat of the
nozzle and one – towards the exhaust. A finer mesh is required at these locations due to the rapid
change of fluid properties and fluid direction. The used mesh for this simulation consists of 7971
nodes and 7650 elements. The quality of the mesh can be examined in table 39.1.

Mesh Metric Average Value

Skewness 1.0984 × 10−2
Aspect Ratio 2.5564
Orthogonal Quality 0.9985

(Table 39.1) – Mesh quality parameters

Physics Setup

Since the assumptions used to generate the axisymmetric nozzle are inviscid and irrotational flow, the
simulation is run with those same assumptions. Nozzle flows however are in many cases oscillatory.
In the case of an inviscid nozzle flow simulation, a lack of boundary layer causes a zero-damping
condition for the oscillatory flow which causes the simulation to diverge. A solution to this issue is
running a transient simulation. Once set, the model is selected to be inviscid with the energy equation.
The fluid is defined per the parameters from this report and the newly created fluid is used for the
simulation. The boundary conditions are set as seen in figure 40.1 and table 40.1.
 40

(Figure 40.1) – Location of boundary conditions

Boundary Condition Input Parameters

Pressure Inlet 8 000 000 [𝑃𝑎] / 3538.69 [𝐾]
Interior 𝑁/𝐴
Axis 𝑁/𝐴
Wall 𝑁/𝐴
Pressure Outlet 101325 [𝑃𝑎] / 300 [𝐾]
Operating Conditions 0 [𝑃𝑎]

(Table 40.1) – Input parameters for all boundary conditions in figure 40.1

The residuals are set to 10−6 and additional monitors for the nozzle exhaust are set. The initialization
is hybrid and the calculation is initiated with a constant time-step of 1 × 10−5 [𝑠𝑒𝑐] and a total of
1100 time-steps. The maximum iteration-count per time-step is set to 1000.

Setup Summary
Density-Based
General Model Transient
Axisymmetric
Energy
Equations
Inviscid
Fluid
Material
Constant Ideal Gas
Reference Value Inlet
Flow: Second Order Upwind
Methods
Transient: Second Order Implicit
Residuals 1e-06 [-]
Initialization Hybrid
 41

Fixed
Time Advancement
User-Specified
Number of Time Steps 1100
Time Step Size 1e-05 [sec]
Max Iteration per Time Step 1500

(Table 41.1) – Summary of the CFD physics setup for an inviscid rocket nozzle

Convergence Validation

(Graph 41.1) – Exhaust Mach number CFD (Graph 41.2) – Exhaust pressure CFD monitor
monitor

(Graph 41.3) – Exhaust temperature CFD monitor

Graphs 41.1, 41.2, and 41.3 are monitors positioned at the nozzle exhaust. The values at that point are
important to obtain via CFD as they are already known through isentropic theory and can be directly
compared to judge the validity of the simulation. When compared to the isentropic values, the
 42

monitors indicate that the numerically calculated values are acceptable since all three parameters have
an error of less than 1%.

Property at Exhaust Analytical Numerical Error

Mach Number [-] 3.255 3.275 0.614%
Pressure [Pa] 101325 102212 0.875%
Temperature [K] 1 779.96 1767.35 0.708%

(Table 42.1) – Validation of gas properties at exhaust

Results

Throat Exhaust
4 3500
3.5 3000

Velocity Mag. [m/s]

3 2500
Mach Number

Mach
2.5
2000 Number
2
1500 Velocity
1.5
1 1000

0.5 500
0 0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Axial Length [m]

(Graph 42.1) – CFD velocity magnitude and Mach number versus nozzle axis

Throat Exhaust
8000000
7000000
6000000
Pressure [Pa]

5000000 Dynamic Pressure

4000000 Static Pressure
3000000
2000000
1000000
0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Axial Length [m]

(Graph 42.2) – CFD static and dynamic pressure versus nozzle axis
 43

Throat Exhaust
4000

3500
Temperature [K]

3000
Temperature
2500

2000

1500
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Axial Length [m]

(Graph 43.1) – CFD temperature versus nozzle axis

(Figure 43.1) – CFD contour of Mach number

Discussion

Despite the large amount of information that CFD can provide, the presented data is relatively concise
in order to focus on examining the general flow properties of the working fluid inside an axisymmetric
MOC nozzle. For each of the three graphs, the flow between the converging section and immediately
after the throat (−0.015 < 𝑥 < 0.07) the gradient is high. The reader may recall that this rapid change
 44

in properties is due to the rapid expansion for a relatively short distance associated with nozzles.
Advancing downwind, at 𝑥 = 0.09 an expected phenomenon is observed. As previously stated the
geometry needs further refinement of the transonic and boundary region and the CFD simulation now
provides strong evidence to why that is needed. Examining the velocity graph (Graph 42.1), a drop in
velocity magnitude of approximately 50 m/s is observed. This indicates the generation of a weak
shock. The shock would have been exacerbated if a larger expansion angle was selected thus this
simulation being proof that the selected expansion angle is relatively suitable and the generated shock
is weak enough to be neglected for this paper. Nevertheless the shock can be minimized even further if
the expansion angle is reduced.

Following the shock and up to the exhaust of the nozzle, the graphs show that the rate of change of the
properties of the gas decreased by a significant amount. This raises another flaw of MOC and it is
associated with mechanically inefficient nozzles. Despite the nozzle being theoretically 100%
thermodynamically efficient, in practice the nozzles are longer than desired from a practical stand
point. The graphs suggest that it is needn’t for the nozzle to exceed such lengths as the properties’ rate
of change is small for the majority of the diverging duct.

At the exhaust, the pressure graph (Graph 42.2) shows that the exhaust static gauge pressure of the gas
is approximately zero which confirms that the nozzle is perfectly expanded as designed and expected.
Some disturbance of the flow at the edges of the exhaust is observed. Caused simply by the geometry
of the sharp corner, this is expected and considered insignificant. After the exhaust, another oblique
shock is formed which is again expected and is due to the geometry of the exhaust boundary. Most
nozzle flows are expected to generate shocks once outside the nozzle. This is typical for rocket nozzles
and is also insignificant to the nozzle as they do not exert any forces on the nozzle wall – one of the
reasons for in need of designing shock-free nozzles.

Conclusion

A procedure for designing axisymmetric rocket nozzles is presented. The analytically solvable
iterative Axisymmetric MOC equations are implemented in a computer programme to increase the
speed and accuracy of the design process. To obtain valid physical results from the procedure,
isentropic calculations for nozzle flows are performed in order to iterate the MOC-generated nozzle to
the isentropic exhaust radius. It is made clear that in reality the number of characteristic lines is
infinite, however a finite amount is needed to be able to produce a nozzle contour. It is discovered that
anything above 200 characteristic lines is suitable for accurate nozzle geometries and each of the
slopes must be iterated at least 5 times to correct for the low accuracy of the compatibility equation
solution. For this specific procedure, the nozzle wall contour is generated from the kernel as it is the
only characteristic net that this paper focuses on constructing. This simplification of the model causes
 45

additional issues such as the undefined transonic region which allows for the generation of oblique
shock waves. To overcome this issue, a new boundary condition at the throat must be input to the
problem and such boundary can be obtained from Sauer’s transonic method. The length of nozzles
produced with Axisymmetric MOC is also undesirable for a practical design. Ways of artificially
truncating Axisymmetric MOC nozzles can be found to mitigate this issue. The irrotational and
inviscid compatibility equations produce nozzles that are accurate only with certain theoretical models.
For full-scale designs aiming to mimic physical nozzles corrections for variable chemical kinetics and
viscous flows must be considered. The compatibility equations can be modified to account for such
properties of the gas. Another approach to solving these issues is to use the generated inviscid contour
as a benchmark geometry and perform additional calculations and optimizations to arrive at a more
realistic model of a rocket nozzle.

References
1. Anderson, J. (2004) “Modern Compressible Flow: With Historical Perspective” 3 rd Ed. New
York: McGraw-Hill
2. Cumpsty, N. & Heyes, A. (2015) “Jet Propulsion” 3rd Ed. New York: Cambridge University
Press
3. Reid, H. DeFrance, S. & Sharp, E. (1953) “Equations, Tables, and Charts for Compressible
Flow: Report 1135”. Moffett Field, CA: NACA
4. Hyde, J. & Gill, G. (1967) “Liquid Rocket Engine Nozzles: chemical propulsion.” Cleveland:
NASA
5. Huzel, D. & Huang, D. (1967) “Design of Liquid Propellant Rocket Engines” 2 nd Ed.
Washington, D.C.: NASA
6. McBride, B. & Gordon, S. (1996) “Computer Program for Calculation of Complex Chemical
Equilibrium Compositions and Applications”. Cleveland: NASA
7. Chatwin, C (2017), “Rocket Equation & Multistaging: Lecture for Satellite and Space Systems
MSC” Sussex: University of Sussex
8. Taylor, T. (2009) “Introduction to Rocket Science and Engineering” Boca Raton: Taylor and
Francis Group
9. Young, R. (2012) “Automated Nozzle Design through Axis-Symmetric Method of
Characteristics Coupled with Chemical Kinetics” Auburn: Auburn University
10. Guentert, E. & Neumann, H. (1960) “Design of Axisymmetric Exhaust Nozzles by Method of
Characteristics Incorporating a Variable Isentropic Exponent” Cleveland: NASA
11. Shapiro, A. (1953) “The Dynamics and Thermodynamics of Compressible Fluid Flow”
Volume I. New York: John Wiley & Sons
12. Kulhanek, S. (2012) “Design, Analysis, and Simulation of Rocket Propulsion System”
Kansas: University of Kansas
 46

Appendices

Appendix A

The Axisymmetric MOC software can be found as an .exe and .py file. The .exe file can be used on a
Windows7-10 OS and does not require any coding skills to use. The .py file is the source code and
does require understanding of Python 3.7+. This file can be opened on any operating system.

The AXIMOC Nozzle Designer can be found on:

https://github.com/KyrilPalaveev/Axisymmetric-Method-of-Characteristics-Nozzle-Designer-with-
GUI

Appendix B

Tabular Data for 2D Method of Characteristics for nozzle with 8 characteristic lines. Refer to the
report for more information on this data.

deltatheta [deg] v_e first theta [deg] gamma M_e

5.0000 72.035 1.0175 1.1877 3.2554

Point K_ [deg] K+[deg] θ [deg] ν [deg] M μ [deg]

1 2.035 0.000 1.018 1.018 1.077 68.166
2 12.035 0.000 6.017 6.017 1.269 51.979
3 22.035 0.000 11.017 11.017 1.422 44.670
4 32.035 0.000 16.017 16.017 1.565 39.724
5 42.035 0.000 21.017 21.017 1.703 35.963
6 52.035 0.000 26.017 26.017 1.840 32.926
7 62.035 0.000 31.017 31.017 1.977 30.381
8 72.035 0.000 36.017 36.017 2.117 28.193
9 72.035 0.000 36.017 36.017 2.117 28.193
10 12.035 -12.035 0.000 12.035 1.452 43.529
11 22.035 -12.035 5.000 17.035 1.593 38.883
12 32.035 -12.035 10.000 22.035 1.731 35.296
13 42.035 -12.035 15.000 27.035 1.868 32.373
14 52.035 -12.035 20.000 32.035 2.005 29.910
15 62.035 -12.035 25.000 37.035 2.145 27.783
16 72.035 -12.035 30.000 42.035 2.288 25.914
17 72.035 -12.035 30.000 42.035 2.288 25.914
18 22.035 -22.035 0.000 22.035 1.731 35.296
19 32.035 -22.035 5.000 27.035 1.868 32.373
20 42.035 -22.035 10.000 32.035 2.005 29.910
21 52.035 -22.035 15.000 37.035 2.145 27.783
 47

22 62.035 -22.035 20.000 42.035 2.288 25.914

23 72.035 -22.035 25.000 47.035 2.435 24.247
24 72.035 -22.035 25.000 47.035 2.435 24.247
25 32.035 -32.035 0.000 32.035 2.005 29.910
26 42.035 -32.035 5.000 37.035 2.145 27.783
27 52.035 -32.035 10.000 42.035 2.288 25.914
28 62.035 -32.035 15.000 47.035 2.435 24.247
29 72.035 -32.035 20.000 52.035 2.586 22.745
30 72.035 -32.035 20.000 52.035 2.586 22.745
31 42.035 -42.035 0.000 42.035 2.288 25.914
32 52.035 -42.035 5.000 47.035 2.435 24.247
33 62.035 -42.035 10.000 52.035 2.586 22.745
34 72.035 -42.035 15.000 57.035 2.743 21.378
35 72.035 -42.035 15.000 57.035 2.743 21.378
36 52.035 -52.035 0.000 52.035 2.586 22.745
37 62.035 -52.035 5.000 57.035 2.743 21.378
38 72.035 -52.035 10.000 62.035 2.907 20.124
39 72.035 -52.035 10.000 62.035 2.907 20.124
40 62.035 -62.035 0.000 62.035 2.907 20.124
41 72.035 -62.035 5.000 67.035 3.077 18.966
42 72.035 -62.035 5.000 67.035 3.077 18.966
43 72.035 -72.035 0.000 72.035 3.255 17.889
44 72.035 -72.035 0.000 72.035 3.255 17.889

Appendix C

Tabular Data for Axisymmetric Method of Characteristics for nozzle with 8 characteristic lines. Refer
to the report for more information on this data.

K_ K+ θ ν M [- μ
V [m/s] x [m] r [m] p [Pa] T [K]
[deg] [deg] [deg] [deg] ] [deg]
0.00 0.00 0.00 0.00 1.00 90.00 1211.35 0.00000 0.00000 4544907.66 3236.12
22.26 -22.26 0.00 22.26 1.74 35.16 1942.68 0.00000 0.01359 2760.49
29.02 -22.56 3.23 25.79 1.83 33.06 2025.58 0.00497 0.01715 2692.65
36.94 -23.13 6.90 30.03 1.95 30.86 2121.16 0.00816 0.01955 2610.92
42.95 -23.77 9.59 33.36 2.04 29.33 2193.83 0.01071 0.02156 2546.27
47.68 -24.47 11.61 36.08 2.12 28.18 2251.55 0.01312 0.02353 2493.35
51.50 -25.28 13.11 38.39 2.18 27.26 2299.86 0.01580 0.02578 2448.01
54.62 -26.27 14.17 40.44 2.24 26.50 2342.04 0.01917 0.02866 2407.64
54.62 -26.27 14.17 40.44 2.24 26.50 2342.04 0.00406 0.01568
22.26 -22.26 0.00 22.26 1.74 35.16 1942.68 0.01929 0.00000 1656876.62 2760.49
 48

36.01 -25.03 5.49 30.52 1.96 30.63 2131.96 0.02689 0.00573 2601.45
44.24 -28.04 8.10 36.14 2.12 28.15 2252.93 0.03239 0.00975 2492.07
50.63 -30.11 10.26 40.37 2.24 26.52 2340.55 0.03686 0.01306 2409.07
55.45 -31.77 11.84 43.61 2.33 25.38 2405.81 0.04084 0.01607 2345.21
59.06 -33.26 12.90 46.16 2.41 24.54 2456.21 0.04484 0.01913 2294.68
61.66 -34.74 13.46 48.20 2.47 23.90 2495.81 0.04936 0.02258 2254.24
61.66 -34.74 13.46 48.20 2.47 23.90 2495.81 0.04771 0.02133
40.17 -40.17 0.00 40.17 2.23 26.60 2336.42 0.03857 0.00000 705465.90 2413.06
50.12 -42.23 3.94 46.17 2.41 24.53 2456.43 0.04675 0.00443 2294.45
56.14 -44.69 5.73 50.42 2.54 23.23 2538.26 0.05336 0.00805 2210.17
60.79 -46.45 7.17 53.62 2.63 22.31 2598.55 0.05927 0.01135 2146.31
64.12 -47.92 8.10 56.02 2.71 21.66 2642.79 0.06500 0.01462 2098.50
66.28 -49.24 8.52 57.76 2.76 21.20 2674.55 0.07114 0.01812 2063.68
66.28 -49.24 8.52 57.76 2.76 21.20 2674.55 0.09054 0.02920
52.86 -52.86 0.00 52.86 2.61 22.53 2584.27 0.05786 0.00000 350818.62 2161.58
59.93 -54.32 2.80 57.13 2.74 21.37 2663.01 0.06643 0.00379 2076.38
64.10 -56.22 3.94 60.16 2.84 20.60 2717.70 0.07391 0.00718 2015.69
67.13 -57.60 4.76 62.36 2.92 20.06 2756.65 0.08109 0.01048 1971.72
68.92 -58.74 5.09 63.83 2.96 19.71 2782.31 0.08850 0.01390 1942.42
68.92 -58.74 5.09 63.83 2.96 19.71 2782.31 0.13915 0.03731
61.74 -61.74 0.00 61.74 2.89 20.21 2745.62 0.07714 0.00000 203755.81 1984.25
66.50 -62.72 1.89 64.61 2.99 19.53 2795.78 0.08605 0.00344 1926.93
69.04 -64.08 2.48 66.56 3.06 19.09 2829.33 0.09432 0.00669 1888.01
70.48 -65.02 2.73 67.75 3.10 18.82 2849.65 0.10270 0.01000 1864.22
70.48 -65.02 2.73 67.75 3.10 18.82 2849.65 0.18746 0.04348
67.63 -67.63 0.00 67.63 3.10 18.85 2847.66 0.09643 0.00000 138215.94 1866.56
70.39 -68.20 1.09 69.30 3.15 18.49 2875.72 0.10564 0.00323 1833.44
71.40 -69.00 1.20 70.20 3.19 18.29 2890.82 0.11467 0.00644 1815.49
71.40 -69.00 1.20 70.20 3.19 18.29 2890.82 0.22980 0.04719
71.01 -71.01 0.00 71.01 3.22 18.12 2904.34 0.11571 0.00000 109503.24 1799.33
71.92 -71.20 0.36 71.56 3.24 18.00 2913.41 0.12522 0.00314 1788.44
71.92 -71.20 0.36 71.56 3.24 18.00 2913.41 0.26340 0.04901
72.11 -72.11 0.00 72.11 3.26 17.89 2922.54 0.13500 0.00000 101325.00 1777.46
72.11 -72.11 0.00 72.11 3.26 17.89 2922.54 0.28926 0.04979