Professional Documents
Culture Documents
A THESIS SUBMITTED TO
THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF
MIDDLE EAST TECHNICAL UNIVERSITY
BY
SEVDA AÇIK
MAY 2010
Approval of the thesis:
Date: ________________________
I hereby declare that all information in this document has been obtained and
presented in accordance with academic rules and ethical conduct. I also
declare that, as required by these rules and conduct, I have fully cited and
referenced all material and results that are not original to this work.
Signature :
iii
ABSTRACT
Açık, Sevda
M. Sc., Department of Mechanical Engineering
Supervisor: Prof. Dr. Zafer Dursunkaya
Co-Supervisor: Dr. K.Atılgan Toker
Design process of a solid rocket motor with the objective of meeting certain
mission requirements can be specified as a search for a best set of design
parameters within the overall design constraints. In order to ensure that the best
possible design amongst all achievable designs is being achieved, optimization is
required during the design process.
In this thesis, an optimization tool for internal ballistic design of solid rocket
motors was developed. A direct search method Complex algorithm is used in this
study. The optimization algorithm changes the grain geometric parameters and
nozzle throat diameter within the specified bounds, finally achieving the optimum
results.
iv
prediction analysis of rocket motor and the mathematical optimization algorithm.
The code developed is verified against pretested rocket motor performance.
v
ÖZ
Açık, Sevda
Yüksek Lisans, Makina Mühendisliği Bölümü
Tez Yöneticisi: Prof. Dr. Zafer Dursunkaya
Ortak Tez Yöneticisi: Dr. K.Atılgan Toker
Katı yakıtlı roket motoru tasarım süreci; belirlenmiş misyon isteklerini sağlamak
için, sistem kısıtları içinde en iyi tasarım parametrelerinin aranması olarak
tanımlanabilir. Bulunabilecek olası çözümlerin içinden en iyisinin seçildiğinden
emin olmak için tasarım sürecinde optimizasyon gereklidir.
vii
ACKNOWLEDGEMENT
I would like to express my deepest thanks and gratitude to Prof. Dr. Zafer
DURSUNKAYA for his supervision and constant guidance through this study.
I would like to thank to my co-supervisor Dr. Kemal Atılgan TOKER, for his
crucial advises and suggestions, he encouraged me to overcome the problems.
I would like to thank to Uğur ARKUN, for sharing his experience and for his great
suggestions and aids throughout this study which help combining the mathematical
work to real physics of the rocket propulsion.
I would also like to thank to my friends Umut AKAY and Özlem SARCAN for
their friendship and support.
Special thanks go to my mother Ayşe AÇIK and my entire family for always being
there for me and giving me the motivation.
This thesis is dedicated to my family, without them nothing would have been made.
viii
TABLE OF CONTENTS
ABSTRACT...................................................................................................................................... IV
ÖZ ..................................................................................................................................................... VI
CHAPTERS ........................................................................................................................................ 1
1 INTRODUCTION ......................................................................................................................... 1
3.1 GENERAL.......................................................................................................................... 26
ix
3.2 OPTIMIZER ....................................................................................................................... 29
3.2.1 Complex Method ....................................................................................................... 29
3.2.2 BCPOL Subroutine of IMSL...................................................................................... 31
3.3 BURNBACK ...................................................................................................................... 32
3.3.1 Burnback of End-Burning Grain............................................................................... 32
3.3.2 Burnback of Internal Burning Grain......................................................................... 32
3.3.3 Burnback of Star and Star-Tube Grain ..................................................................... 33
3.3.4 Burnback of Slot and Slot-Tube Grain ...................................................................... 33
3.4 BALLISTICS SOLVER ..................................................................................................... 34
3.4.1 Assumptions............................................................................................................... 34
3.4.2 Governing Equations................................................................................................. 35
3.5 OBJECT.............................................................................................................................. 38
4 VALIDATION ............................................................................................................................ 40
REFERENCES ................................................................................................................................. 63
x
LIST OF TABLES
TABLES
Table 3.1 Discrete and Continuous Parameters of Different Grain Types ............. 27
Table 3.2 Continuous Parameters of Different Grain Types .................................. 28
Table 4.1 Geometric Parameters of Tubular Motor, Nondimensionalized Using
the Throat Diameter ................................................................................ 41
Table 4.2 Dimensionless Geometric Parameters of Star Motor (SM) .................... 43
Table 4.3 Initial Guess and Bounds of Parameters of Optimization,
SM Grain Dimensions and Optimized Solution ...................................... 45
Table 4.4 Dimensionless Geometric Parameters of Slot-Tube Motor (STM) ........ 47
Table 4.5 Initial Guess and Bounds of Parameters of Optimization, STM
Grain Dimensions and Optimized Solution ............................................. 48
Table 4.6 Dimensionless Initial Guess Parameters and Bounds of Star
Grain Optimization .................................................................................. 52
Table 4.7 Dimensionless Optimized Results for Star Grain ................................... 54
Table 4.8 Dimensionless Initial Guess Parameters and Bounds of Slot-Tube
Grain Optimization .................................................................................. 57
Table 4.9 Dimensionless Optimized Results for Slot-Tube Grain ......................... 59
xi
LIST OF FIGURES
FIGURES
Figure 2.1 Main Parts of a Solid Propellant Rocket Motor .................................... 10
Figure 2.2 Burnback Analysis of a Slotted Grain Geometry .................................. 20
Figure 2.3 End-Burning Configuration................................................................... 22
Figure 2.4 Internal-Burning Tube Configuration.................................................... 22
Figure 2.5 Star Grain Geometry and Burnback Analysis [7].................................. 23
Figure 2.6 Slot Grain Configuration ....................................................................... 24
Figure 2.7 Slot-Tube Configuration [15] ................................................................ 24
Figure 3.1 Optimization Tool Architecture ............................................................ 26
Figure 3.2 Illustration of a Simplex in Two and Three Dimensions [12]............... 30
Figure 3.3 Zero-dimensional SRM Conservative Relations [26] .......................... 35
Figure 4.1 Tubular Motor....................................................................................... 41
Figure 4.2 Comparison of Thrust-Time Histories of 0-D Ballistic Solver and
Firing Test for TM ..................................................................................... 42
Figure 4.3 Comparison of Thrust-Time Histories of 0-D Ballistic Solver and
Firing Test of SM....................................................................................... 43
Figure 4.4 Comparison of Thrust-Time histories of Optimized Solution,
Initial Guess and Objective for SM ........................................................... 46
Figure 4.5 Comparison of Thrust-Time histories of Optimized Result,
Initial Guess and Objective for STM .......................................................... 49
Figure 4.6 Typical Thrust-Time Requirements of Solid Rocket Motors................ 50
Figure 4.7 Objective Thrust-Time Prepared for Optimization and SM
Thrust-Time Curve..................................................................................... 51
Figure 4.8 Comparison of Thrust-Time histories of Initial Guess for
Different Number of Stars and Objective for Star Optimization................ 53
Figure 4.9 Comparison of Thrust-Time Histories of Initial Guesses and
xii
Optimized Results for Star Grain Optimization......................................... 54
Figure 4.10 Objective Thrust-Time Prepared for Optimization and Original
STM’s Thrust-Time Curve.......................................................................... 56
Figure 4.11 Comparison of Thrust-Time histories of Initial Guess for
Different Number of Slots and Objective .................................................. 58
Figure 4.12 Comparison of Thrust-Time histories of Initial Guesses and
Optimized Results for Slot-Tube Grain ...................................................... 59
xiii
LIST OF SYMBOLS
xiv
mpconstraint Propellant mass constraint
M Mass stored in the chamber
Me Exit Mach number
N Number of parameters
nrb Burning rate pressure exponent
P Pressure
Pamb Ambient pressure
Pc Chamber pressure
Pcmax Maximum chamber pressure
Pcmax_constraint Maximum chamber pressure constraint
Pe Nozzle exit pressure
r1 Fillet radius of star grain
r2 Cusp radius of star grain
rb Propellant burning rate
R Gas constant
t Time
tb Burning time
T Temperature
ve Velocity at nozzle exit
Greek Symbols:
ρ density
γ specific heat ratio
∂ partial derivative operator
ε Expansion ratio
εf Convergence tolerance for complex method
σp Temperature sensitivity of burning rate, % /° K
Subscripts:
xvi
CHAPTER 1
CHAPTERS
1 INTRODUCTION
INTRODUCTION
Design process of a solid rocket motor with the objective of meeting certain
mission requirements can be specified as a search for a set of design parameters
within the overall design constraints. A wide variation in the design parameters
and the large number of possible combinations of these parameters turn this design
process into a search with a very large number of possible answers.
In order to ensure that best possible design amongst all achievable designs is being
acquired, optimization is essentially required during the design process.
Finding the optimal design of a solid rocket motor to meet certain criteria has been
subject of search since the 1960’s. In 1980, Sforzini [1] commented in his own
solid rocket optimization paper that despite the tractable mathematics involved in
designing a solid rocket motor, “limited treatment of this subject appears in the
literature”. Today, Sforzini's statement is still true to a great extent [2]. Limited
number of work can be found in open literature.
1
Woltosz, [4] in 1977, proposes a pattern search technique (developed by Hook and
Jeeves [5] ) to find an optimal solid grain geometry. He determines five critical
design dimensions which maximize the total impulse - to motor weight ratio while
meeting a specified minimum achieved velocity for a specific vehicle. [2]
Hook and Jeeves pattern search technique was also used by Foster and Sforzini [6]
to minimize the differences between computed and desired solid rocket motor
ignition characteristics based upon several igniter design parameters. Also In
Sforzini's [1] 1980 paper, the pattern search technique was used to manipulate
fifteen geometric variables governing the solid rocket motor design. Specified
constraints were enforced by penalty functions in order to discourage unrealistic
designs [2].
Swaminathan and Madhavan [7] tried to find the optimum propellant composition,
which gives the highest possible specific impulse, using a direct random search
technique, similar to simulated annealing.
In 1993 Clergen [8] developed a computerized expert system with which the
designer can define design criteria such as minimum motor mass and obtain desired
motor parameters for a certain mission. The system has a user-friendly, hypertext
interface and the system basicly uses past experience while selecting the motor
design parameters. This system is built around a data base of known systems. [2]
McCain's [9] method is also an expert system in the sense that it is heuristic. This
method combines a pattern searsh method with a heuristic system to develop rocket
performance characteristics. The heuristic performs an independent design
variable selection, and these design variables are then passed to the pattern search
optimization package. The heuristic method then analyzes the effect of altering
each independent SRM design variable and selects, based on gradient information,
the variables to alter for the next design attempt. This technique, with its heuristic
2
sense, is an advancement of the pattern search technique of Sforzini, but still has
some of the inherent weakness of any gradient-based method [2]
In 2001 Anderson et al.[2] used genetic algorithms to design solid rocket motors as
a component within an overall missile system. In this study multiple goals, such as
maximized range, minimized g-loading, minimized takeoff weight, and maximized
fuel volume, are used to test the ability of genetic algorithms to work efficiently
within a multidisciplinary framework.
Nisar and Guozhu presents a methodology for design optimization of wagon wheel
grain [10] and for design optimization of SRM finocyl grain [11]. In both studies,
they utilize a Hybrid Optimization (HO) technique by using Genetic Algorithm
(GA) for global convergence integrated with Sequential Quadratic Programming
for further local convergence of the solution thus attaining the final optimal
solution.
As it can be seen in above examples, the methods used for optimization of solid
rocket motors design fall into three broad categories: gradient based methods
heuristic methods and hybrid methods
3
Other well-known gradient method, Newton’s Method, is a second order one. This
method is similar to the Gradient Descent method, but it adds second order
information to its calculations in the form of the Hessian [12].
Heuristic methods are the optimization methods that use no gradent information
and they are sometimes called gradient-free algorithms. “A heuristic method
applies a simple rule of thumb, often derived from natural processes, combined
with some amount of stochasticity to an optimization problem” [14]. In order to
escape local extrema, most heuristic methods incorporate randomness. They can
perform relatively well in non-convex, complex and noisy problems with both
continuous and discrete design problems where gradient based methods have
difficulty. But heuristic methods are still not guaranteed to find the global
optimum [14].
The simplest derivative-free method is the one referred to as Direct Search. This
method could be named random search, in that it merely checks objective function
values, and accepts good points and rejects bad points, ending when a maximum
iteration number has been achieved. Through the years more and more
sophisticated logic has been developed to allow these types of algorithms to
intelligently search through the design space. These may be as simple as
4
distributing the search, such as in Parallel Direct Search [15], or using a simplicial
method, as in Box’s Complex Method [16]. One of the more interesting methods
developed recently, is Jones’ Direct method, which employs a bounding technique
performing Lipshitz optimization without the Lipshitz constant [17].
5
1.2 SCOPE OF THE THESIS
The purpose of this study is to develop an internal ballistic design optimization tool
for solid rocket motors. For a given objective, in this case objective thrust-time
curve of the rocket motor, optimization tool yields the optimum propellant grain
geometry and nozzle geometry. This optimization process will aid the solid rocket
motor design engineer in making the best initial design selections and thereby
reducing the overall "design cycle time" of a project.
Six types of propellant grain geometries are involved in this study; end burning,
internal burning tube, slot, slot-tube, star and star-tube grain geometries. Burnback
analysis is conducted by using analytical methods. For the performance prediction
of a rocket motor, a 0-D internal ballistic solver is developed and used.
Optimization is obtained using a direct search technique “complex method”. The
objective function to be optimized or minimized is the summation of the squares of
the differences between the desired and computed thrust values of the SRM at
specified times during motor operation divided by the average desired thrust and
total number of data. Specified constraints on propellant weight and chamber
pressure are enforced by penalty functions in order to discourage unrealistic and/or
undesired designs.
The developed optimization tool, which is described in this thesis, is validated with
the results of previously designed rocket motor data and actual firing test data .
6
1.3 CONTENTS OF THE THESIS REPORT
Chapter 3 contains the detailed description of the optimization tool and its
subprograms. Optimization tool is composed of subprograms OPTIMIZER,
BURNBACK, BALLISTIC SOLVER and OBJECT. Every subprogram has a
different function. The methodology used and their governing equations are
presented in this chapter.
Validation of the tool and the results are presented in Chapter 4. Test cases
involving pretested rocket motor results are implemented.
7
CHAPTER 2
2 FUNDEMENTALS OF SOLID ROCKET MOTORSATION TOOL
Rocket motors are widely used to impart a desired velocity to a flight vehicle
which requires high thrust in order to transport its payload. A rocket motor is a
typical energy transfer system. The chemical energy inside the fuel is converted to
the thermal energy by a combustion process. High pressure and high temperature
combustion product gases are expanded through a converging-diverging nozzle
[18]. By this process “internal energy of the gas is converted into kinetic energy of
the exhaust flow and the thrust is produced by the gas pressure on the surfaces
exposed to the gas” [19].
Rocket motors are classified in many ways. The most common way is the
classification according to the physical state of the propellant [18]. These are as
follows:
a) Solid Propellant Rocket Motors (SRM): As its name implies, the propellant of
the motor is in the solid state. The oxidizer and the fuel is premixed and is
contained and stored directly in the combustion chamber. Since the solid
propellant both includes fuel and oxidizer, solid propellant rocket motors can
operate in all environmental conditions. In comparison to other types of rockets,
solid propellant rocket motors have simple design, , are easy to apply and require
little or no maintenance [20].
8
b) Liquid Propellant Rocket Motors: This type of rocket motors use liquid
propellants that are fed under pressure into the combustion chamber. In the
chamber, the liquid fuel and oxidizer are mixed and burned to form hot gaseous
products. Some liquid propellant rocket motors are capable of repetitive operation,
that is they can be started and shut off at will. Also the thrust level can be adjusted
during operation. The main disadvantage of a liquid rocket propulsion system is
that; it requires several precision valves and a complex feed mechanism which
includes, a relatively complex combustion chamber and propellant pumps, turbines,
or a propellant-pressurizing device [18], [20].
d) Hybrid Rocket Motors: In hybrid rocket motors usually the fuel is in solid state
and the oxidizer is in liquid phase. In this hybrid motor concept, oxidizer is
injected onto the solid fuel grain inside the combustion chamber. Like the liquid
propellant rocket motors hybrid motors can be started and shut off at will. Hybrid
motors have lower density-specific impulse than solid propellant systems [18].
When compared to other types of rocket motors, solid propellant rocket motor
(SRM) is the most commonly used one due to its relatively simple design, high
reliability, ease of manufacture and cheapness. SRM can be used for a wide
variety of applications requiring wide range of magnitude (few Newtons to several
million Newtons) and duration of thrust [21].
9
The schematic diagram of a solid propellant rocket motor is shown in Figure 2.1
As shown in the figure above, SRMs are mainly composed of a motor case, an
igniter, a nozzle, propellant grain and insulation.
Generally motor case is a cylindrical cover containing the solid propellant, igniter
and insulator. The combustion takes place in the motor case; therefore, sometimes
it is referred to as combustion chamber.
The case must be capable of withstanding the internal pressure resulting from the
motor operation, approximately 3-30 MPa, with a sufficient safety factor.
Therefore motor case is usually made either from metal (high-resistance steels or
high strength aluminum alloys) or from composite materials (glass, kevlar, carbon)
[22]. In addition to the stresses due to the pressure in the chamber, thermal stresses
10
may sometimes be critical and, when the case also serves as flight vehicle body,
bending loads and inertial forces also play an important role in determining the
thickness and the material of the motor case.
2.2.1.2 Insulation
High temperature of the combustion gases, ranging from approximately 2000 to
3500 K, requires the protection of the motor case or other structural subcomponents
of the rocket motor. Typical insulator materials have low thermal conductivity,
high heat capacity and usually they are capable of ablative cooling. Most
commonly used insulation materials are EPDM (Ethylene Propylene Diene
Monomer) with addition of reinforcing materials.
2.2.1.3 Igniter
The ignition system gives the energy to the propellant surface necessary to initiate
combustion. Ignition usually starts with an electrical signal. The ignition charge
have a high specific energy, they are designed to release either gases or solid
particles. Conventional heat releasing compounds are usually pyrotechnic
materials, black powder, metal-oxidant formulations and conventional solid rocket
propellant [20].
2.2.1.4 Nozzle
High temperature, high pressure combustion gases are discharged through the
converging-diverging nozzle. By this way, chemical energy of the propellant is
converted to kinetic energy and thrust is obtained. The geometry of the nozzle
directly determines how much of the total energy is converted to kinetic energy.
Therefore nozzle design has a very important role on the performance of a rocket
motor [21].
11
Nozzles are usually classified according to their structural mounting technique or
the shape of the contour; such as submerged nozzle, movable nozzle and bell-
shaped nozzle.
Combustion product gases have an erosive effect with their high temperature and
high velocity and also with a high concentration of liquid and solid particles like
metal oxides inside them. Material selection of the nozzle is a very important step
of nozzle design, especially for the throat region where erosive effects are more
dominant. Refractory metal, carbon containing composites or graphite and
reinforced plastic that will withstand erosive effects are commonly used as throat
material.
Solid propellant is cast in a certain configuration and geometry that is called the
propellant grain.
The propellant grains can be subcategorized into two main configurations; case-
bonded grain and free-standing grain. Case-bonded grains are directly cast into the
motor case already provided with thermal insulation. After the curing operation
the propellant grain is completed, this motor case with the propellant grain is ready
to be mounted with the other components of the motor. Free standing grains are
not directly cast into the motor case, instead the propellant is cast in some special
mold. When the cure process of the propellant is completed, the grain is extracted
from this mold. Free standing grains are loaded to the insulated motor case on the
assembly line, that is why they are sometimes called as cartridge-loaded grains.
Burning surface of the grain changes during motor operation as the propellant
burns. Burnback analysis determines this change in the grain geometry. The
12
geometric design of the grain ultimately defines the performance characteristics
that can be obtained with a given propellant type and nozzle.
The mission requirements of a specific flight vehicle system are usually the desired
range, time of flight and velocity of the system in operation. For the rocket motor
side, all these requirements can be simplified to the thrust time history of the rocket
motor with some geometrical constraints. The thrust time curve characteristics
depend on the combustion features propellant properties, grain geometry, and
nozzle design. The branch of applied science describing these is known as internal
ballistics [1].
The objective of the internal ballisticians is to provide the rocket motor a propellant
grain that will evolve combustion products consistent with the thrust-time schedule
required for the mission. In order to achieve this objective, the designer deals with
some parameters of the rocket motor, called ballistic parameters. Some of these
ballistic parameters are explained in the following sections.
The flow area at the nozzle throat At is a very important design parameter of
nozzle. It is evaluated in conjunction with the variables associated with thrust
coefficient, nozzle exit diameter, ambient pressure, chamber pressure and nozzle
inefficiencies.
13
Nozzle expansion ratio is defined as the ratio of nozzle exit area to the nozzle
throat area as given below [1]. The expansion ratio for the optimum expansion can
be calculated by using this isentropic formula as given below.
γ −1
1 1
At γ + 1 γ −1 Pe γ γ + 1 Pe γ
= 1− (2.1)
Ae 2 Pc γ − 1 Pc
Ae
ε= (2.2)
At
Thrust coefficient CF is defined as the thrust divided by the chamber pressure and
the throat area At. Physically CF is an expression for efficiency of the nozzle for a
fixed propellant configuration.
γ −1
γ +1
2γ 2 2 γ −1 Pe γ Pe − Pamb
CF = 1− + ε (2.3)
γ − 1 γ + 1 Pc Pc
The relation between CF, Pc, At and thrust F is given by the following equation;
F
CF = (2.4)
Pc At
14
2.3.1.3 Burning Rate
Propellant grain burns in a direction perpendicular to the grain surface. The rate, at
which a propellant burns, usually described by a reference value at a specific
pressure [23]. Such value is called burning rate and its unit is meters per second.
As an independent parameter, the burning rate is one of the propellant properties.
The relation between burning rate and the chamber pressure is governed by the
following empirical equation, also known as Saint Robert's burn rate law:
n rb
rb = a.Pc (2.5)
This empirical expression defines the burning rate of the propellant; values a and n
usually derived from strand burner tests or small subscale burning rate test motor
firings at different operating pressures
1 ∂rb
σp = ∂T (2.6)
rb P
15
1 ∂P
πK = (2.7)
P ∂T K
Pc At
C* = (2.8)
m&
2.3.1.5 Thrust
The thrust of a SRM is the force produced by a rocket propulsion system acting
upon a vehicle. In a simplified way, it is the reaction experienced by its structure
due to the ejection of matter at high velocity. The trust is the main design
constraint of a propulsion system.
Thrust can be calculated from momentum equation applied on the overall rocket
system ;
16
F=m
& v e + (Pe − Pamb )A e (2.9)
F = C F Pc A t (2.10)
Specific impulse, Isp, is a measure of the impulse or momentum change that can be
produced per unit mass of the propellant consumed. Specific impulse, on the other
hand, can be described as the ratio of the motor thrust to mass flow rate and hence
its value is very important in the determination of the propellant weight necessary
to meet the ballistic requirements [23]. Specific impulse is defined as
c * CF F
I sp = = (2.11)
g0 m
& g0
The total impulse is the thrust force F integrated over the burning time t. It is
directly proportional to the total energy released by all the propellant of the
propulsion system [18].
t
I t = ∫ Fdt (2.12)
0
For constant thrust and negligible start and stop transients this reduces to
I t = Ft (2.13)
17
2.3.1.7 Chamber Pressure and MEOP
Chamber pressure is the gas pressure inside the combustion chamber during motor
operation. In grain design process, usually a limit on maximum pressure is
established at the time grain design activity commences. Concurrent with grain
design, the motor case and other structural components are being designed and
analyzed with this maximum pressure [18]. This constraint on chamber pressure is
usually named as “Maximum Expected Operating Pressure” (MEOP).
Burning surface of the grain changes during motor operation as the propellant
burns. Burnback analysis determines this change in the grain geometry. As the
burning surface changes chamber pressure and the thrust of the rocket motor
changes, therefore performance of the motor is directly related to burnback steps of
the propellant.
18
Analytical methods, numerical methods and drafting techniques are commonly
used in grain burnback analysis.
In these methods, usually the dimensional parameters are adapted for every burn
step and the burning surface is calculated analytically at each burn step [21].
Several analytical methods for 2-D grain geometries in literature [24], [25], [26]
have been reported.
In these methods, numerical algorithms are used in order to evaluate the propellant
grain surface regression. These methods do not need to divide the grain geometry
into simple solids, complex geometries can be modeled and burnback analysis can
easily be performed. Main disadvantages of such methods are the numerical errors
involved and high computation time required for analysis [21]. Several different
applications of numerical methods in burnback analysis can be found in literature
[28], [29], [30], [31].
19
2.4.1.3 Drafting Techniques
Drafting tools can be used for both two-dimensional and three-dimensional grain
burnback analysis.
20
2.4.2 Typical Grain Configurations
For different system missions, different thrust-time profiles are required for rocket
motors such as progressive, neutral or boost and sustain By changing the propellant
grain configuration, different thrust time profiles can be obtained from a rocket
motor. Grain configurations can be categorized in several ways [23];
• Inner geometry of the grain (Star, wagon, internal burning tube etc.)
• Outer shape of the grain (Tubular, spherical any other unconventional
shape)
• The propellant used (single propellant grain, dual propellant grain)
• the dimensional analysis (two-dimensional grain, three-dimensional grain)
Details of the most commonly used grain configurations in SRM applications are
given in the subsequent sections.
21
Figure 2.3 End-Burning Configuration
The internal-burning tube is one of the most practical and preferred configurations.
It is a radially burning grain with ends usually unrestricted, otherwise it burns
progressively [23]. It is typically case-bonded which inhibits the outer surface The
internal-burning tube is defined by a length L and two diameters D and Dport.
This parameters determine the trend (progressive, neutral or regressive) of thrust-
time curve of internal-burning tube grain
22
2.4.2.3 Star Grain
The trust time profile of a star shaped grain can be regressive, neutral or
progressive depending on the parameters defining the geometry.
23
Figure 2.6 Slot Grain Configuration
24
CHAPTER 3
3 SRM OPTIMIZATION TOOL
In this chapter, internal ballistics design optimization tool developed for solid
rocket motors will be presented.
Six types of propellant grain geometries are involved in this study; end burning,
internal burning tube, slot, slot-tube, star and star-tube grain geometries. Burnback
analysis is conducted by using analytical methods. For the performance prediction
of rocket motor, a 0-Dimensional internal ballistics solver is developed and used.
Optimization is obtained using a direct search technique “complex method”. The
object function to be optimized or minimized is the summation of the squares of
the differences between the desired and computed thrust values of the SRM at
specified times during motor operation divided by the average desired thrust and
total number of data. Specified constraints on propellant weight and chamber
pressure are enforced by penalty functions in order to discourage unrealistic and/or
undesired designs.
25
3.1 GENERAL
26
Variable parameters of solid rocket motor design involved in this study can be
divided into two groups; continuous parameters and discrete parameters.
Continuous parameters such as grain diameter or grain length can change
continuously in the solution domain. Discrete parameters such as “grain type” and
“number of slot or star points(N)” can not change continuously. OPTIMIZER
subprogram does not deal with discrete parameters. For a set of discrete
parameters optimum values of continuous parameters are found by OPTIMIZER,
then iteration starts again for other set of discrete parameters. For each set of
discrete parameters, optimization is performed separately and then all the results
are compared to find the optimum solution.
Six different grain geometries are involved in this study. Number of continuous (n)
and discrete parameters changes for each grain type as shown in
Table 3.1 . These parameters were depicted in figures of each grain type in section
2.4.2. Also definition of each parameter is given in Table 3.1
4. Dt 4. r1 4. r1 4. Dtip-center 4. Dtip-center
5. r2 5. r2 5. Dtip 5. Dtip
6. η 6. η 6. Dt 6. Dt
7. ξ 7. ξ 7. Lslot
8. Dt 8. Dt
9. Lstar
27
Table 3.2 Continuous Parameters of Different Grain Types
End Burning, n=3
Grain length L
Outer diameter Dout
Nozzle throat diameter Dt
Internal Burning Tube, n=4
Grain length L
Outer diameter Dout
Port diameter Dport
Nozzle throat diameter Dt
Star, n=8
Grain length L
Outer diameter Dout
Web thickness w
Fillet radius r1
Cusp radius r2
Star point semi angle η
Star angle ξ
Nozzle throat diameter Dt
Slot, n=6
Grain length L
Outer diameter Dout
Port diameter Dport
Tip-center diameter Dtip-center
Tip diameter Dtip
Nozzle throat diameter Dt
User defines which grain type will be used and if slot or star grain is selected which
number of slot or star points (N) will be tried to the program. For each
combination of selected grain types and N, optimization tool runs and finds an
28
optimum. Then the optimum points are compared and the best one is output as the
optimized solution.
An initial guess (grain type, parameters of grain geometry and nozzle throat
diameter), is supplied to the program by the user. Upper and lower bounds of the
continuous parameters (geometric constraints, such as maximum and minimum
outer diameter that the optimization program can try) are also supplied by the user
3.2 OPTIMIZER
29
r
min
r n f (x) (3.1)
x∈ R
r r r
subject to l ≤ x ≤ u
r
where l is a vector of length n (number of parameters) containing the lower bounds
r
on variables and u is a vector of length n containing the upper bounds on variables
30
equations, or its function value is still the largest, suggesting that the new node is
not better than the old one, a midway node will be located to replace the new node
with a reflection factor of 0.5, as suggested by Box” [35]. Since the reflection
factor is less than one, this procedure is called contraction. The simplex keeps
contracting until the new node fulfills all constraint equations and the new node has
a smaller function value than the old one. At the end of this iteration attempt a new
simplex, with one node improved, is generated. [35]
The iteration procedure continues until some convergence criteria defined at the
beginning of the iteration is reached. And the best node of the last simplex of
iteration will be the optimum..
DBCPOL repeats the iteration of complex algorithm until the maximum number of
allowed function evaluations, defined by the user, is reached or one of the
following stopping criteria is satisfied:
Criterion 1:
Criterion 2:
2n
2n ∑ fj
∑
j =1
( f ( xi) − )2 ≤ ε f (3.3)
i =1 2n
31
where ε f is a user defined tolerance
3.3 BURNBACK
At each iteration OPTIMIZER outputs a new iteration point that is a set of design
variables; grain geometry parameters and nozzle throat diameter. For these grain
geometry parameters, burnback analysis of the grain is performed by the
BURNBACK subprogram. Output of these subprogram is web vs. burn area
information; web burnt vs. burn area at each burn time.
For end-burning grain type burnback analysis is very straightforward. Initial burn
area is calculated analytically as follows;
πDout 2
Ab = (3.4)
4
By its very nature, burn area of the end burning grain does not change during the
burn time. Therefore initial burn area calculated by using equation 3.4 is constant
during the motor operation.
Initial burn area of the internal burning tube grain is calculated analytically as
fallows;
2 2
Dout − D port
Ab = 2π + πD port L (3.5)
4
32
where head and aft ends of the grain are assumed to be unrestricted. If the end
areas are defined to be restricted then the first term of the above equation vanishes.
For every burn step, the dimensional parameters are adapted with web burnt and
the burning surface is calculated again by using the above formula.
Since the slot grain is essentially a special case of the concave star grain, same
methodology and the computer code used in star grain is valid for slot grain
burnback analysis.
33
Slot-tube grain geometry is a combination of slot and internal burning tube
geometries. Therefore burnback analysis of slot-tube geometry is performed as the
summation of burn areas evaluated for slot and internal burning tube geometries.
3.4.1 Assumptions
7. The properties of the gases in the chamber can be found by using weighted
averages.
34
3.4.2 Governing Equations
For a radial burning solid rocket motor with subsonic flow (M<1.0), it can be
assumed that the properties of gases are constant (d( )/dx = 0) along the grain
length. The total pressure at the throat of the nozzle is also assumed to be equal to
the chamber pressure.
dM
m& g = + m& n (3.7)
dt
where m& g is the rate of mass addition by burning propellant, dM is the rate of
dt
change of stored mass in the combustion chamber and m& n is the rate of mass flow
through the nozzle. Since in almost all applications nozzle closures are used,
m& n = 0 until the pressure in the chamber reaches the closure blowout pressure
[37].
& g = ρ p A b rb = ρ p A b aPc n rb
m (3.8)
35
where ρ p is the propellant density and Ab is the burn area of the grain at that
instant. The mass flow through the nozzle is calculated by the relation
Pc At
m& n = (3.9)
C∗
where Pc is the chamber pressure, At is the throat area and C* is the characteristic
speed of the propellant. The throat area may change due to erosion of the nozzle
insulation material and the characteristic speed is a function of chamber pressure.
The rate of change of mass stored in the chamber is given by the relation
dM d ( ρυ ) dυ dρ
= =ρ +υ (3.10)
dt dt dt dt
dPc 1 PAg dυ
= RTc ρ p Ab (t )aPc rb − c t∗ 0 − Pc
n
(3.11)
dt υ (t ) C dt
dυ
where = rb Ab . Equation (3.5) is integrated with infinitesimal time steps as the
dt
propellant burns in order to obtain chamber pressure at these infinitesimal time
steps.
Using the chamber pressure, nozzle exit pressure is evaluated by the equation 2.1.
And then the thrust coefficient Cf is calculated by using the equation 2.3.
36
The thrust is then calculated by the following relation,
F (t ) = C f Pc (t ) At (t )C d (3.12)
where C d is the divergence loss coefficient. As seen in the above equation thrust
is directly proportional to Cf and therefore proportional to nozzle exit pressure Pe.
Nozzle exit area is the primary parameter that defines the exit pressure (thrust has
been defined in terms of nozzle exit conditions in equation 2.9). Therefore thrust
of the rocket motor is highly dependent on nozzle exit area or the expansion ratio
During the design process essentially in optimization, designers deal with a wide
variety of chamber pressure and nozzle throat diameter. For each design attempt
the nozzle exit area that gives the optimum expansion must be found. Therefore an
automated procedure for finding the best nozzle exit area for each design is needed.
A subroutine that calculates desired nozzle exit area depending on the user
preference is added to BALLISTIC SOLVER. After the chamber pressure at
infinitesimal time steps is calculated by equation 3.5, this subroutine evaluates the
average chamber pressure during the burn time of motor. For this average chamber
pressure, nozzle exit area that gives the desired exit pressure is calculated by using
the below equations.
k −1
2 Pc k
Me = ⋅ − 1 (3.13)
k − 1 Pe
37
k +1
Ae 1 2 k − 1 2 2( k −1)
= 1+ M (3.14)
At Me k + 1 2
User may prefer to have constant nozzle exit area or prefer that the program
calculate the best nozzle exit area for each design.
3.5 OBJECT
For the rocket motors, mission requirements can usually be simplified to the thrust
time history of the rocket motor with some geometrical constraints. Rocket motor
designer tries to find the best design in order to achieve the desired thrust-time
curve. For this reason, the Optimization tool developed in this study tries to find
best designs with thrust-time curve that fit the desired thrust-time history.
Therefore the object function to be minimized is defined as how much the current
iteration thrust-time curve different than the desired one. It is typically summation
of the squares of differences between the desired and computed thrust values at
specified times during motor operation divided by average desired thrust and total
number of data. It can be formulated as follows,
∑ (F
i =0
desired − Fiteration ) 2
Object.Function =
j ⋅ Fdesired _ avg
Where j depends on the time increment of above calculation and maximum of the
burn-times of desired and iterated thrust-time curves.
38
In order to discourage unrealistic designs maximum chamber pressure and
propellant mass constraints are enforced by penalty functions. Main idea of the
penalty function method is adding a term to the objective function that prescribes a
high cost for violation of the constraints. Constraints are specified by the user as
maximum chamber pressure and maximum propellant weight.
If the iterated solution has a higher chamber pressure that violates the predefined
pressure constraint, the following penalty term is added to the object function;
10
Pc max − Pc max_ constra int
10 + 1 (3.15)
Pc max_ constra int
If the iterated solution has a higher propellant mass that violates the predefined
mass constraint, the following penalty term is added to the object function;
10
mp − mpconstra int
10 + 1 (3.16)
mpconstra int
This program calculates and outputs the objective function. This information is
then passed to the OPTIMIZER.
39
CHAPTER 4
4 VALIDATION
VALIDATION
In this section, results obtained for the validation of the optimization tool will be
presented. Known solutions, test motor firings and predesigned motor data are
used for comparison.
Burnback analysis methods are validated previously. Therefore test cases present
in this section are used in order to validate the zero-dimensional ballistic solver.
Static firing test results of a test motor having tubular propellant grain geometry
with two burning ends, named here after “Tubular Motor” TM is used for
validation. The grain geometry which is given in Figure 4.1 produces a neutral
burning profile (little change in thrust during burning time). Nondimensionalized
geometric parameters with respect to the nozzle throat diameter of TM are given in
Table 4.1.
40
Figure 4.1 Tubular Motor
41
1
0.8
Firing Test
0.4 0-D Ballistic Solver
0.2
0
0 0.2 0.4 0.6 0.8 1
DimensionlessTime
Figure 4.2 Comparison of Thrust-Time Histories of 0-D Ballistic Solver and Firing
Test for TM
In order to understand how close ballistic solver estimates the thrust of the motor,
objectivet function is calculated. Here firing test result is the objective thrust-time
history and ballistic solver result is the estimated one. Object function is calculated
to be 6.07 %.
Static firing test results of a motor having star grain geometry with 6 star points,
named here after “Star Motor” SM is used for validation. Dimensionless
geometric parameters of SM are given in Table 4.2
42
Table 4.2 Dimensionless Geometric Parameters of Star Motor (SM)
Grain length, L/Dt 18.8
Outer diameter, Dout/Dt 3.55
Web thickness, w/w 1.00
Fillet radius, r1/w 0.27
Cusp radius, r2/w 0.18
Star point semi angle, η/w 3.27
Star angle, ξ/w 2.27
Nozzle throat diameter, Dt/Dt 1.00
0.8
Dimensionless Thrust
0.6
0.4
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
Figure 4.3 Comparison of Thrust-Time Histories of 0-D Ballistic Solver and Firing
Test of SM
43
In order to understand how close ballistic solver estimates the thrust of the motor,
objective function is calculated. Here firing test result is the objective thrust-time
history and ballistic solver result is the estimated one. Object function is calculated
as 5.91 %.
Star Motor (SM) grain and nozzle properties, given in section 4.1.2, are defined to
the zero-dimensional ballistic solver and its thrust-time curve is obtained. This
thrust-time history is defined to the optimization tool as the objective (desired)
thrust-time.
An initial guess and bounds of the variables are defined as shown in Table 4.3.
Only 6 star point geometries are chosen for the optimization. Convergence
tolerance for optimization algorithm is defined as 5x10-7. In this test case program
converges after 1081 function evaluations and outputs the optimum solution shown
in Table 4.3. Values in Table 4.3 are nondimensionalized with SM grain
dimensions. Difference between optimized solution and SM grain dimensions is
also given in Table 4.3. All the parameters except filet radius is optimized to the
actual solution (SM) with a difference less than 1 percent.
44
Table 4.3 Initial Guess and Bounds of Parameters of Optimization, SM Grain
Dimensions and Optimized Solution
Initial Lower Upper SM Optimized Difference
Parameter
Guess Bound Bound Grain Values (% )
Grain length, L / LSM 0.90 0.90 1.14 1.00 1.00 0.22
Outer diameter, 1.33 0.84 1.45 1.00 0.99 0.41
Dout / Dout,SM
Web thickness, w / wSM 0.73 0.73 1.82 1.00 1.00 0.91
Fillet radius, r1 / r1,SM 0.83 0.50 1.17 1.00 0.97 3.00
Cusp radius, r2 / r2,SM 1.25 0.80 1.50 1.00 1.00 0.00
Star point semi angle, 0.89 0.83 1.11 1.00 1.00 0.03
η / ηSM
Star angle, ξ / ξSM 1.12 0.80 1.20 1.00 1.00 0.12
Throat diameter, 1.07 0.64 1.07 1.00 1.00 0.09
Dt / Dt,SM
Dimensionless initial guess and optimized solution are also shown in Figure 4.4
(thrust is normalized with respect to Fmax and time is normalized with respect to
maximum firing time in Figure 4.2). As seen in Table 4.3 and in Figure 4.4
optimization tool starts with an initial guess far away from the objective thrust-time
curve, but the optimized solution is very close to the objective with less than 1 %
difference.
45
1
Objective
0.8 Initial Guess
Optimized
Dimensionless Thrust
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
Thrust-time history of a motor having a certain slot-tube grain geometry with 5 slot
points, named here after “Slot-Tube Motor” STM is used for validation.
Dimensionless geometric parameters of STM are given in Table 4.4
46
Table 4.4 Dimensionless Geometric Parameters of Slot-Tube Motor (STM)
Grain length, L/Dt 44.1
Outer diameter, Dout/Dt 2.8
Port diameter, Dport/Dtip 9.2
Tip-center diameter, 15.8
Dtip-center/Dtip
Tip diameter, Dtip/Dtip 1.0
Nozzle throat diameter, 1.0
Dt/Dt
Slot length, Lslot / L 0.2
Slot-Tube Motor (STM) grain and nozzle properties, are defined to the zero-
dimensional ballistic solver and its thrust-time curve is obtained. This thrust-time
history is defined to the optimization tool as the objective (desired) thrust-time.
An initial guess and bounds of the variables are defined as shown in Table 4.3.
Only 5 slot point geometries are chosen for the optimization. Convergence
tolerance for optimization algorithm is defined as 5x10-7. For this case program
converges after 791 function evaluations and outputs the optimum solution shown
in Table 4.5. Convergence is assumed when the objective function value is 0.003.
47
Table 4.5 Initial Guess and Bounds of Parameters of Optimization, STM Grain
Dimensions and Optimized Solution
Initial Lower Upper STM Optimized Difference
Parameter
Guess Bound Bound Grain Values (% )
Grain length, 1.05 0.98 1.05 1.00 0.99 0.68
L / LSTM
Outer diameter, 1.11 0.90 1.13 1.00 0.96 0.47
Dout / Dout,STM
Port diameter, 1.49 0.96 1.53 1.00 0.97 3.32
Dport / Dport,STM
Tip-center diameter, 1.19 0.89 1.19 1.00 1.00 0.05
Dtip-center / Dtip-center,STM
Tip diameter, 1.40 0.60 1.60 1.00 0.80 19.8
Dtip / Dtip,STM
Throat diameter, 1.05 0.72 1.05 1.00 0.98 2.06
Dt / Dt,STM
Slot length, 1.42 0.81 1.45 1.00 1.02 1.45
Lslot / Lslot,STM
Difference between optimized solution and STM grain dimensions is also given in
Table 4.3 as percent difference. Nondimensionalized initial guess and optimized
solution are shown in Figure 4.5 (thrust is normalized with respect to Fmax and time
is normalized with respect to maximum firing time in Figure 4.5).
48
F 55500, t 2.06
0.8
Dimensionless Thrust
0.6
0.4
Objective
Initial Guess
Optimized
0.2
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
As seen in Table 4.5 and in Figure 4.5 optimization tool starts with an initial guess
far away from the objective thrust-time curve, but the optimized solution is very
close to the objective.
49
In this part of the study an objective similar to these curves is given to the
optimization program. Objectives of predesigned rocket motors are implemented
and optimized solutions are compared with these rocket motors to see whether the
optimization tool gives better designs then the present designs.
In order to define objective thrust-time curve, total impulse and average thrust of
the actual STM is calculated. And a new thrust time curve that meets these total
impulse and average thrust requirements is prepared. as the objective function of
the optimization.
A Star Motor (SM) was previously designed without the aid of any optimization
tool, by trial and error and knowledge of experience. The motor has a known star
grain configuration with 6 star points, and a known nozzle geometry. Grain and
nozzle parameters of this motor are given in section 4.1.2.
In order to define objective thrust-time curve, total impulse and average thrust of
the actual SM is calculated, and a new thrust time curve that meets these total
50
impulse and average thrust requirements is prepared as the objective of the
optimization.
As seen in Figure 4.7, SM thrust has a slow decrease after the time is 0.45. This
part of the curve, called the sliver region, is almost useless due to the inefficient
burning of the propellant. Sliver is an inherent characteristic of star geometry but
the amount depends on the specific design. Any design having relatively small
sliver region is more acceptable. Therefore average thrust of SM is calculated up
to time 0.45 and the remaining part of the curve is discarded. Implementing this
average thrust, the objective thrust-time curve of this test case is defined as in
Figure 4.7 (thrust is normalized with respect to Fmax and time is normalized with
respect to maximum firing time in Figure 4.7).
1
SM
Objective
0.8
Dimensionless Thrust
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
51
For this test case an initial guess and bounds on the variables are defined as shown
in Table 4.6. Values in Table 4.6 are nondimensionalized with the SM grain
parameters. Grain length and outer diameter bounds come from the SM design.
Designer of the SM had these constraints on grain length and outer diameter.
Therefore these constraints are given to the optimization tool without any change.
But the bounds of the other parameters are now chosen arbitrarily considering
producibility issues.
Table 4.6 Dimensionless Initial Guess Parameters and Bounds of Star Grain
Optimization
Initial Lower Upper
Parameter
Guess Bound Bound
Grain length, L / LSM 1.05 0.96 1.05
Outer diameter, Dout / Dout,SM 0.96 0.96 1.02
Web thickness, w / wSM 0.73 0.73 1.27
Fillet radius, r1 / r1,SM 1.50 0.50 1.50
Cusp radius, r2 / r2,SM 1.25 0.75 2.25
Star point semi angle, η / ηSM 0.83 0.83 1.11
Star angle, ξ / ξSM 1.12 0.80 1.20
Nozzle throat diameter, Dt / Dt,SM 0.94 0.77 1.54
Initial guess thrust-time curves compared to objective thrust-time curve are shown
in Figure 4.8 (thrust is nondimensionalized with respect to Fmax of initial guesses
and time is nondimensionalized with respect to maximum firing time of initial
guesses in Figure 4.8).
52
1
Objective
Guess N= 4
0.8 Guess N= 5
Guess N= 6
Dimensionless Thrust
Guess N= 7
Guess N= 8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
53
1
0.8
Dimensionless Thrust
0.6
Objective
Optimized N= 4
0.4 Optimized N= 5
Optimized N= 6
Optimized N= 7
Optimized N= 8
0.2 SM
0
0.2 0.4 0.6 0.8 1
Dimensionless Time
54
Objective function value that is minimized at the end of optimization (as described
in section 3.5) is also given in Table 4.7 for each star geometry. For comparison
objective function is calculated for the actual SM geometry and it is found to be
11.2 % for this objective thrust-time curve.
All of the optimized star grain geometries have a better curve fit with the
objective thrust-time curve than the actual motor SM. The minimum function
value is obtained for the star geometry with 5 star points. Therefore this 5 star
grain geometry has a thrust-time curve more close to the objective thrust-time
curve then all the other star grains within the specified bounds and constraints.
Slot-Tube Motor (STM) was previously designed without the aid of any
optimization tool, by trial and error and based on experience. The motor has a
known slot-tube grain configuration with 5 slot points, and a known nozzle
geometry. Grain and nozzle parameters of this motor are given in section 4.2.2.
In order to define objective thrust-time curve, total impulse and average thrust of
the actual STM is calculated, and a new thrust time curve that meets these total
impulse and burn time requirements is prepared as the objective of the
optimization. Comparison of the objective and STM thrust-time curves is shown in
Figure 4.10 (thrust is nondimensionalized with respect to Fmax and time is
nondimensionalized with respect to maximum firing time in Figure 4.10).
55
F 47100, Time 2.05 normalize
0.8
Dimensionless Thrust
0.6
0.4
Objective
STM
0.2
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
Figure 4.10 Objective Thrust-Time Prepared for Optimization and Original STM’s
Thrust-Time Curve
For this test case an initial guess and bounds on the variables are defined as shown
in Table 4.8. Values in Table 4.8 are nondimensionalized with the STM grain
parameters. Grain length and outer diameter were fixed for the STM design.
Therefore grain length and outer diameter is not variable for this optimization test
case. Also the nozzle exit diameter must be lower than a certain value. All these
constraints are defined in the optimization process. But the bounds of the other
parameters are now chosen arbitrarily considering producibility issues. Also
maximum chamber pressure is constrained to be 3000 Psi using the penalty
functions explained in Chapter 3.5. Slot geometries of N=4-8 are chosen for the
optimization.
56
Table 4.8 Dimensionless Initial Guess Parameters and Bounds of Slot-Tube Grain
Optimization
Initial Lower Upper
Parameter
Guess Bound Bound
Grain length, 1.00 1.00 1.00
L / LSTM
Outer diameter, 1.00 1.00 1.00
Dout / Dout,STM
Port diameter, 1.09 0.83 1.53
Dport / Dport,STM
Tip-center diameter, 0.92 0.88 1.17
Dtip-center / Dtip-center,STM
Tip diameter, 1.20 0.80 1.80
Dtip / Dtip,STM
Throat diameter, 0.86 0.72 1.08
Dt / Dt,SM
Slot length, 1.61 0.81 2.10
Lslot / Lslot,STM
Initial guess thrust-time curves compared to objective thrust-time curve are shown
in Figure 4.11 (thrust is nondimensionalized with respect to Fmax of initial guesses
and time is nondimensionalized with respect to maximum firing time in Figure
4.11).
57
1
0.8
Dmensionless Thrust
0.6
0.4 Objective
Guess N= 4
Guess N= 5
Guess N= 6
Guess N= 7
0.2 Guess N= 8
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
58
time 2.05 F 47000 normalize
0.8
Dimensionless Thrust
0.6
Objective
0.4
Optimized N= 4
Optimized N= 5
Optimized N= 6
Optimized N= 7
0.2 Optimized N= 8
STM
0
0 0.2 0.4 0.6 0.8 1
Dimensionless Time
59
Objective function value that is minimized at the end of optimization (as described
in section 3.5) is also given in Table 4.7 for each slot-tube geometry. The
minimum function value is obtained for slot-tube geometry with 4 slot points. For
comparison objective function is calculated for the STM geometry and it is found
to be 19.8 % for this objective thrust-time curve.
All of the optimized grain geometries have a better curve fit with the objective
thrust-time curve than the actual motor STM. The minimum function value is
obtained for slot-tube geometry with 4 slot points. Therefore this 4 slot grain
geometry has a thrust-time curve more close to the objective thrust-time curve then
all the other grains within the specified bounds and constraints.
60
CHAPTER 5
5 CONCLUSION AND DISCUSSION
The purpose of this study is to develop an internal ballistic design optimization tool
for solid rocket motors
An optimization tool for an internal ballistic design of solid rocket motors has been
developed and presented. A direct search method Complex algorithm is used in
this study. Optimization algorithm changes the grain geometric parameters and
nozzle throat diameter finally achieving the optimum results. The main features of
the developed code are given below.
61
For this study;
1. implementation of dual- pulse operating motor or double
motor solvers may increase the capability of the optimization
tool,
2. implementation of structural design parameters into the
optimization algorithms,
3. implementation of stability parameters into the optimization
algorithms,
would be the future areas of interest for further research.
62
REFERENCES
63
[10] Nisar, K., Guozhu, L., “A Hybrid Approach for Design Optimization
of Wagon Wheel Grain for SRM”, AIAA 2008-4893
[11] Nisar, K., Guozhu, L., “A Hybrid Optimization Approach for SRM
FINOCYL Grain Design”, Chinese Journal of Aeronautics 21(2008)
481-487
[12] Ryan A. Fellini., "Derivative-Free and Global Search Optimization
Algorithms in an Object-Oriented Design Framework", MS. Thesis
Department of Mechanical Engineering and Applied Mechanics,
University of Michigan, 1998.
[13] Gill, P.E, Murray, W., and Wright M.H., “Practical Optimization”,
London, Academic Press, 1981
[14] Bairstow, B.K., "Effectiveness of integration of System-Level
Optimization in Concurrent Engineering for Rocket Design", MS.
Thesis Department of Aeronautics and Astronautics, Massachusetts
Institute of Technology, 2004
[15] Dennis, J.E., Jr., Torczon, V., “Direct Search Methods on Parallel
Machines,” SIAM Journal on Optimization, Volume 1, No. 4, pp. 448-
474, 1991.
[16] Box, M.J., “A New Method of Constrained Optimization and a
Comparison with other Methods,” Imperial Chemical Industries
Limited, Central Instrument Research Laboratory, Bozedown House,
WhitChurch Hill, Nr. Reading, Berks, 1965.
[17] Jones, D.R., Perttunen, C.D., and Stuckman, B.E., “Lipschitzian
Optimization Without the Lipschitz Constant,” Journal of Optimization
Theory and Applications, Volume 79, No.1, October, 1993.
[18] Sutton, G., P., Biblarz, O., “Rocket Propulsion Elements”, John Wiley
& Sons, 7th edition, 2001,
[19] [1] Barrere. M, Joumatte. A, Vandenkerckhove. J, “Rocket
Propulsion”, The Advisory Group for Aeronautical Research and
Development of NATO, New York, 1960
64
[20] Yıldırım C., “Analysis of Grain Burnback and Internal Flow In Solıd
Propellant Rocket Motors in 3- Dımensıons”, Ph. D. Dissertation,
Dept. of Mechanical Engineering, METU, 2007.
[21] Püskülcü G., “Analysis of 3-D Grain Burnback of Solid Propellant
Rocket Motors and Verification with Rocket Motor Tests”, MS. Thesis,
Dept. of Mechanical Engineering, METU 2004
[22] Davenas A., “Solid Rocket propulsion Technology”, 1993
[23] “Solid Propellant Grain Design and Internal Ballistics”, NASA, SP-
8076, 1972
[24] Hartfield, R., Jenkins, R., Burkhalter, J., Foster, W., “A Review of
Analytical Methods for Solid Rocket Motor Grain Analysis” AIAA
2003-4506, July, 2003
[25] Barrere, M., Jaumotte, A., Veubeke, B.,and Vandenkerckhove, J.,
“Rocket Propulsion”, Elsevier Publishing Company, Amsterdam,
1960.
[26] Ricciardi, A., “Generalized Geometric Analysis of Right Circular
Cylindrical Star Perforated and Tapered Grains”, J.Propulsion Vol. 8,
No. 1, Jan-Feb 1992
[27] U.S. Air Force Astronautics Lab. “The Solid Propellant Rocket Motor
Performance Prediction Computer Program (SPP)”, AFAL-TR-87-
078, 1987
[28] Toker K.A., "Three-Dimensional Retarding Walls and Flow in Their
Vicinity", Ph.D. Thesis Submitted to the Graduate School of Natural
and Applied Sciences of the Middle East Technical University, Dec
2004.
[29] Willcox M. A., Brewster M. Q., Tang K. C., Stewart D. S, “Solid
Propellant Grain Design and Burnback Simulation Using a Minimum
Distance Function”, AIAA-2005-4350, 41st AIAA/ASME/SAE/ASEE
Joint Propulsion Conference, Tucson, Arizona, 2005.
65
[30] Hejl R. J., Heister S. D., “Solid Rocket Motor Grain Burnback
Analysis Using Adaptive Grids”, Journal of Propulsion and Power, Vol
11, No 5, pp1006-1012, 1995
[31] Mashayek F., Ashgriz N., “Solid Propellant Grain Design by an
Interface Reconstruction Technique”, Journal of Spacecraft and
Rockets, Vol 31, No5, pp908-910, 1994
[32] Zarchan P., Editor-in-chief, “Tactical Missile Propulsion”, AIAA,
Virginia, 1996
[33] Spendley W, Hext GR, Himsworth FR. “Sequential applications of
simplex designs in optimization and evolutionary operation”.
Technometrics 1962;4:441–61.
[34] Nelder JA, Mead R. “A simplex method for function minimization.”
The Comput J 1965;7:308–13.
[35] Ssu-yuan Hu, Jung-Ho Cheng, “Development of the unlocking
mechanisms for the complex method” Computers and Structures
83(2005) 1991–2002
[36] Bunday BD. Basic optimisation methods. Edward Arnolds 1984:98–
102.
[37] Netzer, D.W, “Propusion Analysis for Tactical Solid Propellant Rocket
Motors”, Lecture Notes, 1990
66