Cooperative Authentication in Underwater Acoustic Sensor ...
Acoustic Optimization of an Underwater Vehicle
-
Upload
jossy-abgon -
Category
Documents
-
view
222 -
download
0
Transcript of Acoustic Optimization of an Underwater Vehicle
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
1/83
ACOUSTIC OPTIMIZATION OF AN UNDERWATER
VEHICLE
A thesis submitted in partial fulfillment
of the requirements for the degree of
Master of Science in Engineering
By
RAHUL KHAMBASWADKAR
B.E., University of Mumbai, India 2001
2005Wright State University
WRIGHT STATE UNIVERSITY
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
2/83
SCHOOL OF GRADUATE STUDIES
June 6, 2005
I HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MYSUPERVISION BY Rahul Khambaswadkar ENTITLED Acoustic Optimization of anUnderwater Vehicle BE ACCEPTED IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF Master of Science in Engineering.
Ravi C. Penmetsa, Ph.D.
Thesis Director
Richard J. Bethke, Ph.D.Department Chair
Committee on
Final Examination
Ravi C. Penmetsa, Ph. D.
Ramana V. Grandhi, Ph.D.
Nathan W. Klingbeil, Ph.D.
Joseph F. Thomas, Jr., Ph.D.Dean, School of Graduate Studies
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
3/83
ABSTRACT
Khambaswadkar, Rahul, M. S. Engineering., Department of Mechanical and Materials
Engineering, Wright State University, 2005. Acoustic Optimization of an UnderwaterVehicle Involving Fluid-Structure Interaction
A torpedo is a guided missile that travels underwater and detonates when it comes
in proximity of the target. Its speed and accuracy make it one of the most lethal weapons
in navy munitions. The torpedo is a complex system comprising various subsystems:
propulsion, weapon, guidance and control, and many other complicated auxiliary
equipment important for proper operation of the torpedo. The structural design and
optimization of a lightweight torpedo involves multiple disciplines, such as structures,
fluids, and controls, of which acoustic analysis is a critical part.
In underwater warfare, sophisticated active and passive SONAR (SOund NAvigation and
Ranging) techniques are used by the enemy submarines to detect approaching torpedoes.
Therefore, it is very important for a torpedo to be acoustically silent in order to increase
its effectiveness. Each torpedo emits a specific acoustic signature depending on its
propulsion, hydrodynamics, and other auxiliary noise-producing sources. In this thesis,
experimental data available for the gear noise is simulated using computational sound
sources that are then used to determine the acoustic signature of a torpedo. Furthermore,
the Finite Element Method (FEM) is used to quantify acoustic behavior for the
computational model of a lightweight torpedo. A framework for computational modeling
of experimental data from various sources, incorporation of this information into the
acoustic analysis, and multidisciplinary optimization of a lightweight torpedo are the
main focal points of this thesis.
iii
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
4/83
TABLE OF CONTENTS
1. Introduction.... 1
1.1 Literature Review.. 2
1.2 Project Approach... 4
2. Finite Element Modeling of Torpedo, Fluid, and Noise Source... 8
2.1 Modeling of Lightweight Torpedo 8
2.2 Modeling of Fluid. 11
2.3 Fluid-Structure Interaction 19
2.4 Normal Mode Analysis Results 23
3. Torpedo Noise Modeling...... 28
3.1 Sources of Noise Generation. 29
3.2 Experimental Setup and Noise Profile 31
3.3 Optimization Formulation for Noise Source Modeling...... 33
3.4 Results and Discussion.. 37
4. Multidisciplinary Design Optimization (MDO) of Lightweight Torpedo 41
4.1 Optimization Formulation......... 46
4.2 Optimization in NASTRAN. 47
4.3 Optimization Results and Discussion 49
5. Concluding Remarks ... 54
Appendix........ 56
Appendix A: C++ Program to generate torpedo model input file.... 56
Appendix B: MATLAB file to read NASTRAN output and calculate error 63
Appendix C: NASTRAN input deck for Optimization Run. 65
iv
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
5/83
References 70
v
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
6/83
LIST OF FIGURES
Figure 1.1 Flowchart of Research Approach 5
Figure 2.1 Dimensions of Radial and Longitudinal Stiffeners 9
Figure 2.2 Finite Element Model of the Lightweight Torpedo 10
Figure 2.3 Fluid and Structural Finite Element Models 15
Figure 2.4 Tolerances for Fluid-Structure Interaction 16
Figure 2.5 First Bending Mode 24
Figure 2.6 Second Bending Mode 25
Figure 2.7 Breathing Mode 25
Figure 3.1 Important Sources of Noise Generation in a Torpedo 29
Figure 3.2 Experimental Setup Used for Gear Noise 31
Figure 3.3 Noise Levels on the Meridian of Hemisphere
about the MK-40 Torpedo 32
Figure 3.4 Finite Element Model of Air Chamber 34
Figure 3.5 Noise Recovery Points in the Air Chamber 34
Figure 3.6 Optimization Algorithm to Determine Source Strength 36
Figure 3.7 Intensity of a Pulsating Point Source 38
Figure 3.8 Results for a Constant Profile Case 39
Figure 3.9 Variation Between Experimental and NASTRAN Results 40
Figure 4.1 Air Mesh Inside Torpedos Transmission Section 43
Figure 4.2 Torpedo and Node Locations 43
Figure 4.3 Design Variables for the Problem 46
Figure 4.4 Method of feasible direction 48
vi
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
7/83
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
8/83
LIST OF TABLES
Table 2.1 Dimensions of MK-48 Lightweight Torpedo 9
Table 2.2 Structural Frequencies 24
Table 2.3 Coupled Structural Frequencies 26
Table 2.4 Frequencies of the Fluid Model 26
Table 4.1 Noise Levels with Infinite Boundary Condition 44
Table 4.2 Noise Levels without Infinite Boundary Condition 45
Table 4.3 Torpedo Optimal Configuration 52
viii
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
9/83
Acknowledgement
I would like to thank my advisor Dr. Ravi Penmetsa for his guidance and support and for
giving me opportunity to work on this project. He was instrumental in directing me
towards successful completion of this thesis.
My sincere thanks go to Dr. Ramana Grandhi for his constructive comments and
encouragement. Also, a special mention has to be made about efforts taken by Dr.
Vipperla Venkayya throughout this research work. His eagerness to guide me in proper
direction is highly appreciated. I would like to express my gratitude to Dr Klingbeil for
being a part of thesis committee.
I was well supported by all the members at Computational Design and Optimization
Center (CDOC), Wright State University. These are wonderful people to work with.
Apart from that, my friends, especially, Mayur, Nikhil, Ajay, Savio, Arun, Justin, Milind
and Prithvi made my stay at Dayton enjoyable and deserve an appreciation. I would like
to thank Brandy Foster for her help in making this document grammatically correct and
readable.
Finally, I would like to take this opportunity to thank my parents and my elder sister
without whose efforts it would have been impossible for me to come to USA for studies.
ix
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
10/83
-
I would like to dedicate this thesis to,
My parents
Rekha and Bhagwan Khambaswadkar
x
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
11/83
1. Introduct ion
A torpedo is an underwater missile that can be launched from a submarine, a ship, or an
aircraft. It is a highly sophisticated weapon whose optimal design requires satisfying
multiple conflicting criteria that are equally important. Since every torpedo has numerous
subsystems that produce easily detectable noise, the acoustic signature of a torpedo
becomes one of the critical design criteria. The early detection of a torpedo gives the
target time to take the necessary countermeasures to avoid the assault and reduces the
effectiveness of the torpedo as a weapon. Therefore, when performing design
optimization of a torpedo, it is important to ensure that the torpedos acoustic signatures
are below the detectable range of certain SONAR systems.
The self-generated noise typically increases with the speed of the torpedo and is
extremely undesirable. Noise produced by the torpedo can have other detrimental
implications as well: It may damage or interfere with the smooth operation of different
electronic sensors inside the torpedo itself, which in turn will have an effect on the
guidance and control of the torpedo. Furthermore, the amplification of structural
vibrations due to this sound might result in fatigue of the panels. Finally, noise produced
by torpedoes that is within a specific frequency can be of concern to the sea life. Due to
all of these reasons, structural design of a torpedo subject to acoustic constraints is
required for improved stealth characteristics.
1
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
12/83
In this research, a computational finite element model of a lightweight torpedo is
developed that has longitudinal and radial stiffeners to provide additional strength to the
shell. Since these components are absent in most of the conventional torpedoes, the
current model needs to be analyzed and optimized to meet various design requirements.
Due to these structural modifications, it is also important to investigate how structure-
born noise is transmitted to the fluid, which involves solving a fluid-structure interaction
problem.
1.1 Literature Review:
The problem involving the interaction of an elastic structure with fluid has been of
primary interest to many researchers due to its wide applicability. The problems that can
be associated with this phenomenon can be categorized into exterior and interior
applications. The exterior problems are those in which the sound propagation is exterior
to the structure, such as the sound produced by a vibrating cylinder placed in fluid, which
involves the determination of radiated and scattered noise. The interior problems are
associated with acoustic cavities, piping systems, and other applications in which the
sound is propagated within the structure. These problems have applications in ship noise
reduction, acoustic analysis of a car interior, vibration response of underwater structures,
blast analysis, etc.
Many different formulations were proposed to solve these problems and studies have
been conducted to see the relative trade-offs between these formulations by many
researchers. The methods available in the literature are: boundary element [1-2], finite
element [3-14], coupled boundary-finite elements [2], energy finite elements [15-16], and
2
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
13/83
various decoupling approximations [17-18], to name a few. The numerical modeling
schemes used to model the fluid differentiate these methods from one another. The
usefulness of each of these formulations is highly problem-dependent, and their
availability or the users experience with a particular kind of tool become significant
factors. Methods involving boundary elements generally use the dynamic response of the
structure as input to a boundary element code, which is used to obtain the far field
acoustic response in the fluid domain [1]. The decoupling approximation methods
decouple the structural response from the fluid response and can reduce computational
complications involved with solving coupled equations [18]. Many researchers, including
Zienkiewicz and Newton [3], were instrumental in initiating efforts towards the
successful use of finite elements to solve structural acoustic problems. Everstine, Marcus
et al, [4, 6], continued on the same research and formulated methods that use the
capabilities of the finite element code NASTRAN to solve the fluid-structure interaction
problems. Everstine summarized different finite element-based formulations to solve
structural acoustics problems [7].
Finite element-based methods have the advantage that they can use the matrix capabilities
of sophisticated commercial codes, which are easily accessible and have sophisticated
visualization capabilities. In the current version of NASTRAN, the pressure analog
method developed by Everstine [4, 9] is implemented in the acoustic module. This
method uses solid finite elements to represent scalar fluid fields by modifying the
material properties so that they represent fluid. This method uses an analogy between
equations of elasticity for structure and acoustic wave equations, which will be discussed
in detail in later chapters. In this thesis, the finite element method is used to solve an
3
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
14/83
acoustic radiation problem for a lightweight torpedo structural model. This method is also
used to model the noise source that would result in gear noise characteristics similar to
the experimental data.
1.2 Project Approach:
The main objective of this study is to minimize this structure-born noise in a lightweight
torpedo through the modification of structural parameters. In order to study the sound
radiation from the lightweight torpedo structure, a noise source needs to be modeled that
has the same characteristics as the experimental data available for the gear machinery.
4
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
15/83
Experimental DataFluid
FEA Modeling
Optimization BasedSourceModeling
Structure
Modal Analysis Noise Modeling
Acoustic Analysis
The design methodology is divided into different units that are identified in Figure 1.1.
The first important step in the process is the modeling phase. The proposed
computational model of the torpedo has a shell structure that is supported with ring and
longitudinal stiffeners. These stiffeners provide additional stiffness to the structure with
minimal increase in weight. The fluid surrounding the structure is also modeled using
Minimize: Mass
Subject to: Constraints
Is Design Optimum?
ModifiedStructure
Frequency
Sound
Optimum structure
Figure 1.1 Flowchart of Research Approach
5
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
16/83
finite elements by using an analogy between equations of elasticity and the acoustic wave
equation. In the fluid-structure interaction, structural displacements cause variations in
fluid pressure and these variations in turn affect the structural behavior. The coupling
effect becomes more significant when modal frequencies for the structure and fluid are
similar. Frequency analysis is performed to ensure that the structure-borne noise is not
amplified due to matching of the fluid and structural frequencies. The results of this
eigenvalue analysis provide information about which structural frequencies to avoid
while redesigning the torpedo.
The next task is the modeling of the source that produces structural vibrations, which
result in pressure variations in the fluid. This pressure distribution around the torpedo is
its acoustic signature subject to the noise that is modeled. There are many sources of
noise that excite the torpedo; however, the current research effort is targeted towards
modeling noise due to speed reduction machinery, such as gears, in the torpedo. Even
though this is not the most critical noise source, it is selected because only its
experimental data is available in the public literature [22]. The goal of this research is to
develop a methodology to obtain computational noise sources that produce similar
characteristics as the experimental data.
An optimization-based problem formulation is used to generate a computational model of
the experimental noise data for gears. The fluid model represents the sea, which is
modeled as an infinite domain. To represent the infinite nature of the water model, the
finite element model of the fluid is truncated at a certain distance from the structure and a
doubly asymptotic approximations-based radiation boundary condition is applied on the
6
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
17/83
outer surface of the fluid to ensure that there are no reflections back into the fluid from
the boundaries.
Once the modeling of the structure, fluid, and source is completed, the frequency
response analysis, with the modeled source as excitation is performed. Finally, a
multidisciplinary design optimization problem is formulated to reduce structural mass
with the frequency and acoustics response as constraints. A constraint that reduces noise
generated by the torpedo will result in increased weight. And the amount of this
increment depends on the desired reduction of sound. Therefore, a multi-objective
optimization problem is solved to obtain the Pareto frontier that clearly shows the
tradeoff between the weight and sound produced.
The following chapters will discuss these individual tasks in detail. Chapter 2 discusses
the finite element models used to model the fluid and structure that was used in this
research. Chapter 3 discusses the noise sources and the methodology developed to
determine the computational source that would simulate experimental data. Chapter 4
discusses the multidisciplinary design optimization problem formulation and the Pareto
frontier results obtained.
7
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
18/83
2. Finite Element Modeling of Torpedo, Fluid, and Noise Source
2.1 Modeling of Lightweight Torpedo
A torpedo structure is essentially a cylindrical shell. The dimensions for the torpedo
configuration used in this research are based on the data available in the public literature
about the lightweight torpedo. The diameter of 0.32 m and a length of 2.42 m are selected
for the structure. The nose and tail sections of the torpedo are tapered cones with lengths
of 0.12 m and 0.35 m, respectively. Longitudinal and ring stiffeners are placed along the
torpedo shell to improve structural rigidity. The width and thickness of these stiffeners is
taken as 0.015 m and 0.01 m, respectively. The thickness for the torpedo shell is assumed
to be 0.00635 m. Table 2.1 shows the dimensional parameters of the torpedo structure.
Figure 2.1 shows the solid model of the torpedo with the longitudinal and radial
stiffeners. The stiffeners in this research are modeled without specifying any offset from
the nodes that are used to model the shell elements. The fundamental frequency of the
torpedo structure has a one percent variation from the offset stiffeners model, at the
configuration shown in Table 2.1. This variation is considered to be negligible compared
to the complexities that will be introduced in the optimization problem in which the
offsets need to be adjusted in each iteration, based on the shell thickness.
8
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
19/83
Overall Length 2.42 m
Body Diameter 0.32 m
Nose Length 0.12 m
Tail Length 0.35 m
Shell Thickness 0.00635 m
Stiffener Width 0.015 m
Stiffener Thickness 0.01 m
Table 2.1: Dimensions of MK-48 Lightweight Torpedo
Longitudinal
Radial
Figure 2.1 Dimensions of Radial and Longitudinal Stiffeners
The shell surface is modeled using quadrilateral and triangular plate elements and the
circular rings and longitudinal stiffeners are modeled with bar/beam elements. The
surface plate elements are called QUAD4 or TRIA3 elements for quadrilateral or
triangular elements. The torpedos auxiliary equipments, warhead, and fuel contribute to
9
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
20/83
its total mass. These non-structural components are modeled using mass elements
distributed along the torpedo body at nodal locations. The overall mass of the torpedo
structure is assumed to be 254 Kg. The material chosen for the torpedo structure is
aluminum-2024.
Initial analysis suggests that the finite element model of the nose was introducing local
modes that restricted the determination of the actual axial, bending, and breathing modes
of the torpedo structure. These local modes were clearly dominated by the nose model
rather than the torpedo structure. Therefore, in order to capture the behavior of the entire
structure, rigid elements were introduced in the torpedo nose to connect the tip of the
torpedo to the first section of the main body of the torpedo. These rigid elements provide
enough strength to the otherwise flimsy nose structure to capture the global vibration
characteristics of the torpedo.
Longitudinal Stiffeners
Ring Stiffeners
Figure 2.2 Finite Element Model of Lightweight Torpedo
10
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
21/83
Figure 2.2 shows the finite element model of the torpedo structure used in this research.
The box-like structures distributed all along the model are the mass elements that
represent the non-structural components of the torpedo. This non-structural mass
becomes critical when optimizing the torpedo structure for frequency and frequency-
dependent responses.
2.2 Modeling of Fluid
The three-dimensional fluid is modeled using the conventional solid finite elements
available in NASTRAN. These solid finite elements are used to represent the fluid using
an analogy between the equations of elasticity, which are generally used for solving
structural problems, and the wave equation, which represents fluid acoustics [21, 24, 26].
The pressure analog method [4, 9] developed by Everstine is used in the current version
on NASTRAN. The structural-acoustic analogy is discussed in detail here.
Equation of state:
The equation of state for a fluid relates the internal restoring forces to the corresponding
deformations. Therefore, in fluids terminology it relates pressure to condensation as:
sBp = (2.1)
where,
p = acoustic pressure or excess pressure at any point = -ip 0p
ip = instantaneous pressure at any point
0p = constant equilibrium pressure in the field
11
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
22/83
B = adiabatic bulk modulus =
0
)/(0 iip
0 = constant equilibrium density of the fluid
i = instantaneous density at any point
s = condensation at any point = ( i - 0 )/ 0
Continuity Equation:
This equation gives the functional relationship between particle velocity and the
instantaneous density. The equation is as follows:
0)( =+
u
t
r
(2.2)
Eulers Equation:
This is a simple nonlinear inviscid force equation that is obtained using Newtons second
law:
ipuut
u
=
+
rrr
)( (2.3)
This equation becomes linear if it is assumed that changes in momentum are negligible,
meaning that the second term inside the bracket vanishes.
ipt
u=
r
0 (2.4)
The Homogeneous Linearized Wave Equation:
The above three equations are combined to obtain a single differential equation with one
dependent variable [21, 24]. Using equations (2.1), (2.2), (2.3) we get:
12
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
23/83
2
2
2
2 1
t
p
cp ii
= (2.5)
Where,
c = phase speed for acoustics waves in the fluid =0
B
Equation 2.5 represents a compressible, inviscid fluid whose pressure satisfies the above
equation. In Cartesian co-ordinates the above equation can be written as:
2
21111
dt
pd
Bz
p
zy
p
yx
p
x=
+
+
(2.6)
Elastic-Acoustic Analogy:
In classical elasticity, the equation for equilibrium of stresses in the x direction is given as
follows [27].
2
2
dt
ud
zyx
x
s
xzxyxx
=
+
+
(2.7)
Where,
xu = structural displacement in x-direction
xzxyxx ,, = stress components
s = structural mass density
Now stress-strain relation by Hooks law is given by:
, (2.8)
=
xz
xy
xx
xz
xy
xx
GGG
GGG
GGG
664616
464414
161411
where is an element of a 6 by 6 anisotrpic elastic material matrix.ijG
13
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
24/83
Since displacement components in y and z direction are zero the strain-displacement
relationships becomes,
z
u
y
ux
u
x
xz
x
xy
xxx
=
=
=
Since represents scalar pressure the strains are analogues to pressure gradients.xu
Comparing equations (2.6) and (2.7) and taking,
0
1
1
1
1
1
461614
664411
===
===
=
=
=
=
=
GGG
GGG
z
p
y
p
x
p
B
pu
s
xz
xy
xx
s
x
It can be shown that, with the above assumptions, Equation 2.6 and 2.7 are analogous. It
is clear that the equations of elasticity represent the acoustic fluid wave equation with
certain modifications in the properties of the elements. In NASTRAN, fluid properties
entered on a MAT10 card are internally modified to obtain corresponding analogous
structural properties that are used in matrix manipulations.
14
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
25/83
Fluid is modeled as a cylinder surrounding the torpedo structure. The sphere diameter of
the cylinder is taken as 2 m and the overall length of the fluid domain is taken as 4m.
Solid elements CTETRA are used to model the fluid with an element edge length of
0.004 m. Figure 2.3 shows the fluid domain modeled surrounding the torpedo structure.
Tetrahedral Solid Elements
2.42
4 m
Torpedo Structure
Figure 2.3 Fluid and Structural Finite Element Models
The interface between the fluid and structure may be modeled so that the grid points of
the fluid are coincident with those of the structure. This is called a matching grid. But,
because of the large size of the fluid domain and the irregular shape of the structure, it is
difficult to obtain the matching grid. Therefore, free meshing is used to create the non-
matching fluid mesh around the structure. NASTRAN uses a method called Body in
15
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
26/83
White (BW) to determine wetted elements for fluid-structure interaction coupling.
According to this method, a fluid node is coupled to a structural node for the load transfer
based on the tolerances specified by the user. The method used for determining wetted
nodes is called boxing technique.
First step in the BW algorithm is to determine related structural and fluid nodes. Figure
shows the fluid face and a box around in based on the tolerances provided on ACMODL
card.
Figure 2.4 Tolerances for Fluid-Structure Interaction
Here,
L - The smallest edge length for the particular fluid element
D - The distance from centre of the fluid face to the fluid node
NORMAL, SKNEPS, INTOL The tolerances specified on ACMODL card.
The structural nodes in the above mentioned box will be coupled to the fluid nodes of that
specific fluid element. Once the fluid and structural faces have been determined, a face
co-ordinate system is established for each fluid face. The resultant pressure force on each
node on the fluid element is determined by the relation:
16
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
27/83
{ } Grid/Elem,10;1 NippdSNR jis
fi ====
A virtual work expression used to resolve this resultant pressure force for a unit grid
pressure to grids of the fluid element is as follows:
{ } { } { }ifT
s
fipdSNNF =
The centre of pressure for the fluid face can be obtained using relations:
( )0XXR
FX j
Grids
j i
i
pi=
( )0YYR
FY j
Grids
j i
i
pi=
The resulting load distribution at the grids of the each of the structural element is
calculated using rigid relations such that there is unit motion normal to fluid face with
appropriate moment relationships. The area of each of the structural element is projected
normal to fluid element face is used as weighing factor. The expression for load is as
follows:
{ } ( )
=
0
0])][[][]][[1
i
TT
j
R
RWRRWF
Here,
jF - Vector of resulting load distribution
W- Diagonal weighting matrix
R - Rigid transformation matrix
17
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
28/83
The forces at the structural grids are looped over at each of the fluid element grids. The
same procedure is repeated for all the wetted elements.
It was important to ensure that the fluid modeled is good enough to capture the fluid-
structure interaction effect based on frequency of excitation. NASTRAN provides
guidelines on how far the fluid should be modeled in order to be able to capture all the
required effects. The rule states that six elements per wavelength are required in all
directions from the structure for approximately 10% accuracy. So for 99% accuracy,
approximately 60 elements are required per wavelength. In the current problem the
excitation frequencies for the load is less than 100 Hz. So, the wavelength for this
frequency is given by:
Hz.inexcitationofFrequency
in water,soundofspeed
=
=
=
f
c
where
f
cwavelength
The wavelength is obtained as 14.5 m, which requires that 60 elements should be used in
14.5 m of fluid extent. For 2 m extent, approximately 8 elements are required in all the
directions from the structure. Since the element edge length used is 0.04 m, the used
mesh density is sufficient to capture the effect of the fluid-structure interaction.
18
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
29/83
2.3 Fluid-Structure Interaction (FSI):
In general engineering problems, the effect of fluid-structure coupling is investigated
under the following assumptions of the fluid state:
1. Compressible
2. Inviscid
3. Irrotational
The compressibility assumption is required because the acoustic wave equation involves
the speed of sound, which directly depends upon the bulk modulus of the fluid. For
incompressible fluids, the bulk modulus is infinite, so the assumption of compressibility
is required in the current analysis. Moreover, when deriving the boundary conditions for
the wave equation, the viscous terms are neglected, so the assumption of inviscous fluid
is justified. Since the small motion theory is assumed for the fluid particles, the rotational
effects have very little chance of playing a role in the analysis, so irrotational assumption
is required.
NASTRAN uses a potential-based formulation to represent fluid finite elements. In these
formulations, displacement remains the primary unknown for structural grid points
whereas pressure is the primary unknown for fluid grid points.
The standard equation of motion for the structure is given by:
sssss FuKuM =+ }]{[}]{[ && (2.9)
19
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
30/83
where
sM and are the structural mass and stiffness matrices.sK
Since fluid pressure is applied to the structure as load, the force term on the right hand
side can be divided into external forces on the structure and the fluid pressure on the
interface nodes
}]{[ pAPF ss = . (2.10)
Here, the second term represents the load applied on the structure due to fluid pressure.
The matrix [A] is called a coupling matrix, which is defined as follows:
dSNNAS
sfT }{][ = . (2.11)
Where,
fN and are fluid and structural shape functions, respectively,sN
and is a vector of fluid pressure values at the interface grid points.}{p
So the equation of motion for the structure becomes
}]{[}{}]{[}]{[ pAPuKuM sssss =+&& . (2.12)
Also, the equation of motion for the fluid can be written as
fff FpKpM =+ }]{[}]{[ && . (2.13)
20
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
31/83
Again, the force term on the right hand side can be divided into a fluid force, or acoustic
disturbance in the fluid, and the force applied on the fluid grid points due to structural
displacement.
}]{[}{ sT
ff uAPF &&+= (2.14)
Here, the second term represents the effect of structural displacement on the fluid grids at
the interface. And the vector is the vector of accelerations of the structural grid
points at the interface.
}{ su&&
So, the fluid equation of motion can be written as
}]{[}]{[}]{[ sT
fff uAPpKpM &&&& +=+ . (2.15)
By combining Equations (2.12) and (2.15), the coupled system of equations is obtained as
=
+
f
ss
f
ss
f
T
s
P
P
p
u
K
AK
p
u
MA
M
0
0
&&
&&. (2.16)
The above equations are unsymmetric and difficult to solve. But, Everstine developed a
symmetric version of these equations by using the potential formulation as follows [23].
Let the velocity potential be defined as
qp &= .
21
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
32/83
And by substituting this in the structure Equation (2.12),
}]{[}{}]{[}]{[ qAPuKuM sssss &&& =+ . (2.17)
By integrating the fluid equation with time and multiplying it by -1,
}]{[}]{[}]{[ sT
fff uAdtPqKqM &&& = . (2.18)
Now let = dtPG f .
Now by combing Equations (2.17) and (2.18) the symmetric formulation of the fluid-
structure coupled system is obtained as
=
+
G
P
q
u
K
K
q
u
M
M ss
f
ss
f
s
0
0
0
0
&&
&&. (2.19)
NASTRAN uses this equation by default to solve a coupled system. The output from this
equation is expressed in terms of structural displacement and fluid pressure. It is to be
noted that in all the above equations the damping terms are not included and they can be
added to the equations, if there is damping present in the system. Therefore, the current
approach will be a conservative model that can be improved by incorporating the
structural damping.
22
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
33/83
2.4 Normal Mode Analysis Results:
One of the most significant analyses required in designing a torpedo is the modal analysis
for determining the natural frequency of vibration. This is important not only to maintain
the structural integrity of the torpedo, but also to ensure that the torpedo does not operate
near certain critical frequencies. These frequencies could be those that would interfere
with the onboard electronics, or those that would match with the fluid frequencies and
amplify the noise generated, or others that are undesirable for safe operation of the
torpedo. A modal complex eigenvalue analysis is performed to obtain structural
frequencies, fluid frequencies, and coupled structural frequencies. Table 2.2 shows the
first eight structural frequencies. Some of the frequencies are grouped due to the two axis
of symmetry present in the current model. This table shows the structural frequencies
with no added mass effects from the surrounding fluid. Table 2.3 shows the coupled
structural frequencies where the added mass effect from the surrounding fluid can clearly
be noticed. The frequencies are reduced compared to the structural model that was
analyzed without the surrounding fluid.
23
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
34/83
Mode number Frequency (Hz)
1 22.114
2 22.114
3 127.24
4 127.97
5 136.64
6 136.64
7 142.47
8 178.38
Table 2.2 Structural Frequencies
Figure 2.5 First Bending Mode
24
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
35/83
Figure 2.6 Second Bending Mode
Figure 2.7 Breathing Mode
25
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
36/83
Mode number Frequency (Hz)
1 21.52
2 21.62
3 126.92
4 127.71
5 133.91
6 134.31
7 142.47
8 177.97
Table 2.3 Coupled Structural Frequencies
Mode number Frequency (Hz)
1 26.39
2 52.84
3 58.74
4 63.00
5 63.00
6 67.73
7 68.31
8 79.42
Table 2.4 Frequencies of the Fluid Model
26
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
37/83
Table 2.4 shows the fluid frequencies that represent the vibration characteristics of the
fluid model surrounding the torpedo. Since this model has two symmetric planes XY and
XZ, the modes 4 and 5 are repeating modes. Similar trend was observed in NASTRAN
verification problems [28]. Since the fluid elements have only pressure as degree of
freedom, the mode shape information models the pressure distribution characteristics in
the fluid. It can clearly be seen that the fluid and structural frequencies do not match in
the current configuration. However, the first fluid frequency is close to the structural
frequency and must be avoided during structural re-design.
27
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
38/83
3. Torpedo Noise Modeling
In order to determine the acoustic signature of a torpedo, a frequency response analysis is
performed. A frequency-dependent force is used to excite the structure, which interacts
with the fluid surrounding it to produce noise. In the literature, one can find numerous
acoustic simulations in which the noise is modeled as a simple pulsating force with a
wide range of frequencies. By using this forcing function, the sound produced by the
torpedo is analyzed and minimized using structural sizing algorithms. The drawback of
all of these techniques is the failure to realize that the response is entirely dependent on
the spatial distribution of the forcing function and the frequency of excitation that is
determined by the noise source used.
Therefore, this chapter is dedicated to the accurate modeling of the noise source based on
actual experimental data available. In the literature, no emphasis is placed on the
modeling of the noise sources in this manner. In this research, an optimization-based
formulation is used to model the noise source that will mimic the experimental data
available through the literature for the lightweight torpedoes. The noise source thus
modeled will be placed inside the torpedo structure for the multidisciplinary optimization
of the torpedo.
28
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
39/83
3.1 Sources of Noise Generation:
Figure 3.1 lists some of the major noise sources in a torpedo based upon the information
available in literature. These noises induce structural vibrations at different frequencies
that are then transmitted to the fluid to generate noise.
Exhaust-wake interaction Fans, Converters, Pumps etc.
Guidance & Control
Propulsor
Among these noise sources, the propulsor is the most critical source that is of interest to
the U.S. Navy. However, due to a lack of experimental data available in the public
literature, a less significant but important noise source - engine assembly noise is selected
in this research. Experimental data for this noise source is available along with the details
of the experimental setup. This comprehensive information about the data enabled the
modeling of a computational setup that would mimic the experimental setup.
In a torpedo, the transmission gears or the engine assembly is used in the speed reduction
machinery to control the propeller angular velocity. These gears produce significant
noise, despite their high precision manufacturing [22]. Gear noise comes from a variety
of sources, as follows:
Boundary Layer
Engine Assembly Noise
Figure 3.1 Important Sources of Noise Generation in a Torpedo
29
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
40/83
Roughness of tooth surface
Deflection of gear tooth
Eccentricity of tooth rotor
Misalignment of tooth rotor
Unbalance of tooth rotor
Noise of rotor bearing
Noise of flexible couplings
Seal rub or squeal
Vibrations transmitted by driving or driven equipments
Most of these noise sources can be avoided, but during severe operating conditions some
of these might result in noise generation. Moreover, in some instances the air trapped
between the gear tooth will result in an amplified noise that would then be propagated
through the structure.
A program of experimental research was undertaken at the U. S. Naval Ordnance Test
Station (NOTS) to improve the basic knowledge of gear noise transmission in torpedoes.
The gears were considered as non-uniform point sources radiating into a sphere, and the
total noise output was calculated by integrating data obtained at numerous locations. The
results of these experiments give the noise profile generated by the transmission gears of
a MK-40 lightweight torpedo. This MK-40 torpedo is the precursor to the currently
operating lightweight torpedoes. In this research, an acoustic source that will emit a noise
pattern similar to that of an experimental setup described below will be determined. An
optimization-based problem formulation is used for designing a computational noise
source model that will represent the experimental data.
30
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
41/83
3.2 Experimental Setup and Noise Profile:
The experimental test setup involves a gear assembly placed in an acoustically quiet
chamber, with sensitive microphones placed at fixed radial distances from the
transmission [22]. Figure 3.2 shows the details of the experimental setup.
Noise sensors Transmission
TurbineDynamometer
Acoustic cavity
Acoustically silent room
Figure 3.2 Experimental Setup Used for Gear Noise
A steam turbine is used to rotate the transmission gears and is connected by long shafts so
that the turbine noise does not influence the experiment results. The dynamometer is used
to absorb the load and to measure the torque and speed. Here, the concept of spherical
measurement is used to measure noise. Sensitive microphones placed at a distances of
0.32 m from the transmission are used as measuring points to collect information about
the sound produced by the gear mechanism. The transmission-noise profile generated by
the MK-40 lightweight torpedo captured by the above mentioned experimental setup is
shown in Figure 3.3 [22].
31
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
42/83
110 dB
110 dB
110 dB
114 dB
116 dB
113 dB
110 dB
109 dB
106 dB
103 dB
Figure 3.3 Noise Levels on the Meridian of Hemisphere about the MK-40 Torpedo
Figure 3.3 clearly shows the nonlinear nature of the sound generated by the machinery
noise. The spatial nonlinearity exhibited by the experimental data indicates that the
previous attempts by the researchers to model the noise as a pulsating force at varying
frequencies is inaccurate. This data is used to model a source on the axis of the torpedo
model that can result in a pressure distribution similar to the experimental acoustic data.
The pressure distribution obtained from NASTRAN is converted into the appropriate
decibel level by using the analytical equations shown in the following section.
32
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
43/83
3.3 Optimization Formulation for Noise SourceModeling:
A finite element model of the air representing the hemisphere on which sensors are
placed is modeled using solid finite elements. The use of these solid finite elements to
represent air is possible because of an acoustic-elastic analogy discussed in the previous
chapter. An acoustic load is applied at the center of the cavity to act as a simple noise
generating source. This simple noise source can be imagined to generate a pulsating
sphere in an infinite space. This source will emit noise in a spherical direction, the
magnitude of which will depend on the strength and frequency of the source. This noise
source is used as excitation in a frequency response analysis. The noise emission at
certain key locations (Figure 3.5) in the air model that match the sensor locations in the
experimental setup is monitored. Therefore, this analysis identifies the source strength
that will give the exact same sound profile as the experimental result. The following
figure shows the air model with the source placed at its center. The current air model has
a diameter of 0.64 m whereas the torpedo has a diameter of 0.32 m. This difference
occurs because the sensor locations in the experimental are at 0.64 m. Therefore, once the
source strength and the frequency that would match the sound levels of the experimental
setup is determined, this source would be placed in a similar air model within the torpedo
internal cavity.
33
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
44/83
Figure 3.4 Finite Element Model of Air Chamber
Figure 3.5 Noise Recovery Points in the Air Chamber
Z
Y
X
Recovery pointsNoise Source
56
7
12
34
8
34
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
45/83
The objective of the optimization problem is to minimize the squared error between the
noise levels at particular locations in the air model and the experimental values. The
design variables in the problem are the source strength of the acoustic load and the
frequency of the source. The upper and lower bounds on the source strength and
frequency are also applied as side bound constraints in the optimization problem. The
optimization problem can be summarized as follows:
Minimize
( )2ii BA , (3.1)
where
iA = Noise value obtained from acoustic analysis at location i and
iB = Noise value from the literature at location i,
subject to:
0.100001.0 S and (3.2)
100010 f , (3.3)
where is the source strength that is the design variable for the problem and is the
frequency of the source. The flowchart in Figure 3.6 explains the flow of the optimization
algorithm. The Design Optimization Tool is used for the optimization [25].
S f
35
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
46/83
Nastraninput.cpp
Generate the NASTRAN inputM
odifiedSourceStrengthandFre
quency
NASTRAN
Since gradient-based search methods were used in the optimization iterations, the
sensitivity of the objective to the source strength and frequency were required in this
algorithm [25]. These sensitivities are calculated using the finite difference method
executed using a series of function calls between MATLAB and a C++ program that is
used to generate the NASTRAN model.
Is Error
Minimum
?
DOT
MATLAB
Read NASTRAN Output and Calculates
Error
No
Yes
Final Source
Figure 3.6 Optimization Algorithm to Determine Source Strength
36
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
47/83
3.4 Results and Discussion
Analytical Verification:
Before optimizing the torpedo for the minimum acoustic signature, it was important to
verify the accuracy of the finite element simulation results for the noise levels in the air
model. For the noise emitted at a certain distance by a simple noise source such as a
pulsating sphere, approximate equations are available that give the source strength
needed for a particular decibel level at a specified location [21]. Since this analytical
equation is applicable for a constant sound profile at a distance from the source, the
profile that is considered is constant strength at all the key locations. Once the
NASTRAN and analytical results are verified, the experimental data can be matched
using a similar approach.
The NASTRAN acoustic source is a simple monopole source; therefore, the acoustic
intensity radiated from this simple point source is given by the following equation:
2
4 r
WI
= , (3.4)
where r is the radius of the sphere in which the source radiates energy and W is source
strength in watts. From the available experimental profile, it can be seen that 116 dB is
the maximum sound emitted by the gear assembly. These decibels can be converted into
intensity by using the relation
( )10)10(
dB
oII = . (3.5)
37
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
48/83
24 r
WI
=
r
Figure 3.7 Intensity of a Pulsating Point Source
And, by substituting this intensity into the above equation, the analytical source strength
needed to produce 116 dB at 0.32 m from the source is obtained as follows
24 rIWatts = (3.6)
W5122.0
W/cm5981.3
dB,116For2
=
=
Watts
EI
This source strength is given as input to NASTRAN and the noise generated by this
source measured at 0.32 m from the center is analyzed. The finite element model results
were compared to these approximate equations, and the deviation was 4% from the
expected values. This validated the finite element setup to within the required accuracy.
In this case, the optimization problem is solved such that the noise levels at all the desired
locations are expected to be 116 dB. Figure 3.8 shows the difference between the
38
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
49/83
analytical and the NASTRAN results at various key locations. These key locations are the
same as the sensor locations in the experimental setup.
90
95
100
105
110
115
120
125
1 2 3 4 5 6 7 8
NASTRAN Analytical
Deviation4%
SoundindB
Recovery Points
Figure 3.8 Results for a Constant Profile Case
Matching the Exact Sound Profile:
The experimental noise profile that was shown earlier is highly nonlinear. In this case, the
goal is to match the nonlinear profile as close as possible and then to use the obtained
sources to determine the acoustic signature of the torpedo. Initial attempts to match the
experimental data showed that it is not possible to match the nonlinear profile with only
one design variable; i. e., only one acoustic source. Therefore, in order to match the
profile exactly, more sources are distributed in the transmission section, which increases
the number of design variables for the problem. By using many different combinations of
source distributions in the transmission area and varying the frequency of the sources, the
best fit for the data is obtained. From the optimization results it was clear that we needed
two sources with source strengths 0.9 watts and 0.15 watts at 77.85 Hz. Figure 3.9 shows
39
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
50/83
the deviation between NASTRAN and the experimental results. The maximum deviation
at a given key location is 4%. The current noise source model with a 4% deviation from
the experimental data is more realistic than the traditional approaches that use pulsating
forces to model the noise source.
The optimization formulation that is used in this research is generic, and can be used for
any experimental data that is available in the future. The general idea behind this whole
effort is to use the source obtained from the optimization problem as a load in the
proposed computational model of the lightweight torpedo for acoustic analysis. This will
ensure that realistic data is used to model the source instead of applying random forces to
excite the structure.
90
95
100
105
110
115
120
125
1 2 3 4 5 6 7 8
NASTRAN Experimental Deviation4%
Sound
indB
Recovery Points
Figure 3.9 Variation Between Experimental and NASTRAN Results
40
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
51/83
4. Multidisciplinary Design Optimization (MDO) of L ightweight
Torpedo
The torpedo body can be broadly divided in three sections: transmission, fuel and
warhead, and guidance and control. The modeled noise source will be placed in the
transmission section of the torpedo body, and will act as an excitation force in
determining the frequency response of the torpedo. Figure 4.1 shows the source placed in
the transmission section. An air chamber is modeled inside the transmission section using
solid models and material properties that reflect air density and bulk modulus. This model
has half the diameter as the air model used in the source modeling section, because of the
available cavity size inside the torpedo. The source determined from the earlier analysis
is placed in the appropriate location. The boundaries of the air model transverse to the
axis of the torpedo are left free. This condition assumes no transmission of noise along
the torpedo length through the rest of the cavity. The only transmission is through the
structure. Therefore, the fluid-structure interaction conditions are very significant in an
air-torpedo interface and a torpedo-water interface.
The source inside the air cavity produces a pressure variation in the transmission section
of the torpedo that will result in the displacement of the torpedo structure. This
displacement will be transmitted into the water model resulting in a pressure distribution,
which is the acoustic response of the torpedo. The fluid-structure model is analyzed to
41
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
52/83
verify the effect of the fluid-structure interaction on the results obtained. If the fluid-
structure interaction effect is turned off in the analysis, then the sound intensity in the
water is found to be zero, which indicates that the structural displacements are not
transferred to the fluid model.
As discussed in the modeling section of this document, the outer surface of the fluid has a
radiation boundary condition that simulates the infinite nature of the fluid. In order to
verify the validity of this boundary condition, two analyses, one with the radiation
boundary condition and one without, were performed. The results from the two analyses
can be seen in Figure 4.2 and Tables 4.1 and 4.2. It can be seen from the tables that the
sound is reflecting back from the surface in the case in which there is no absorbing
boundary condition. Also, with an increase in the distance from the source, the noise
should reduce. This is clearly seen from the radiation boundary condition case, but this
trend not very evident in the case without a radiation boundary condition. Therefore, the
infinite boundary condition is a critical component of any underwater acoustic analysis.
When one has to simulate reflections from the ocean floor, one side of the fluid can be
modeled without the infinite boundary conditions.
42
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
53/83
Air Mesh Transmission Fuel warhead and uidance and control
Acoustic Source
Figure 4.1 Air Mesh inside Torpedos Transmission Section.
14
13
12
11321 4
Vibrating Torpedo
5 6
7
8
9Water Medium
10
Figure 4.2 Torpedo and Node Locations
43
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
54/83
Node Co-ordinates in mNode
Location
Node
N
Sound in dB
umber X Y Z
1 1514 -1.09 38.470 0
2 5158 -0.82 0.09 0 49.39
3 4592 -0.37 0 0 59.38
4 5095 2.23 0 0 74.81
5 4716 2.77 0 0 64.15
6 1501 2.91 0 0 53.02
7 5771 0.32 -0.621 0 66.56
8 5018 0.32 -0.735 0 65.05
9 4642 0.32 -0.86 0 61.44
10 1808 0.32 -0.995 0 50.77
11 5786 0.32 0.621 0 66.94
12 5101 0.32 0.735 0 64.54
13 4741 0.32 0.86 0 60.25
14 1763 0.32 0.995 0 49.36
Table 4.1 Noise Levels with Infinite Boundary Condition
44
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
55/83
Node Co-ordinatesNode
Location
Node
N X Y Z
Sound in dB
umber
1 1514 -1.09 88.910 0
2 5158 -0.82 0.09 0 88.99
3 4592 -0.37 0 0 89.23
4 5095 2.23 0 0 81.41
5 4716 2.77 0 0 81.37
6 1501 2.91 0 0 81.38
7 5771 0.32 -0.621 0 87.36
8 5018 0.32 -0.735 0 87.22
9 4642 0.32 -0.86 0 87.13
10 1808 0.32 -0.995 0 87.09
11 5786 0.32 0.621 0 88.941
12 5101 0.32 0.735 0 88.944
13 4741 0.32 0.86 0 88.945
14 1763 0.32 0.995 0 88.945
Table 4.2 Noise Levels without Infinite Boundary Condition
45
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
56/83
4.1 Optimization Formulation:
t is to determine the optimum configuration of the
inimize:
s of the structure
Subjec
d level at certain location < 70 dB
Hz
The str f different sections of the shell, the
The optimum design is one t mean that the
sound signatures from the structure are increased to meet the requirements. This is
The final objective of this projec
torpedo that would have minimum noise propagated into the surrounding water. To
achieve this, an optimization problem is formulated as follows:
M
Mas
t to:
Soun
Natural frequency of the torpedo >= 23
uctural parameters, such as the thickness o
cross-sectional width, and the height of the ring and longitudinal stiffeners, were used as
design variables for the optimization problem. Figure 4.3 shows these design variables.
Figure 4.3 Design
Radial Stiffeners
Longitudinal Stiffeners
RW
RH
LW
LT
Torpedo Shell
Variables for the Problem
hat has minimum mass; however, this would
46
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
57/83
obvious because as the mass is reduced, the shell thickness and dimensions of the
stiffeners decreases, which results in increased noise. Therefore, a realistic solution for
this problem will provide a trade-off analysis between the weight and sound levels
produced by the source.
4.2Optimization in NASTRAN:
ASTRAN has in built design optimization capabilities and most of it comes from
) which is customized to run with NASTRAN. The
directions which is very
strained optimization problems. The basic idea of the
N
Design Optimization Tools (DOT
optimization algorithms that NASTRAN uses are gradient-based methods. The finite
difference method is used for gradient calculation here. In this thesis, for the acoustic
optimization problem method of feasible directions, which is a default method for
NASTRAN, is used. Next section briefly explains this method.
Feasible Directions Method:
Design Optimization Tools (DOT) uses method of feasible
popular method used for con
method is to move from one feasible design to another improved feasible design by
taking small steps. The method tries to keep the design away from boundaries as much as
possible.
0gs jT > gj critical constraints
0gsfsTT
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
58/83
The first condition is the feasible direction condition. According to this condition, the
arch direction S sh all step in along it
- Push-off factors
feasible usable search direction is obtained from above optimization problem, a
line search is performed to determine how far to proceed along the obtained search
se own in figure 4.6 should be such a way that a sm
should not make the design infeasible. The second condition is called usability condition.
According to this condition the search direction should be such way that the objective
should be reduced if it a minimization problem.
Based on the above two conditions a compromise is defined by following maximization
problem:
g2
g1
s
g1=0g2=0
x
-f
1s
s
00gs-Subject to
i
T
jjj
+
+
Figure 4.4 Method of feasible direction
Maximize
T
0f
Where,
j
Once the
48
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
59/83
direction. This leads to another feasible design and the process is repeated till
convergence.
4.3 Optimization Results and Discussion:
240
260
280
300
320
62 64 66 68 70 72 74 76
360
340
Sound in dB
O
ptimizedMassinKg
Figure 4.5 Pareto Optimization Curve
This trade-off analysis can be seen in the Pareto frontier shown in Figure 4.4. This figure
shows how the reduction the structure. Based onin sound level increases the weight of
the weight requirements of the torpedo, an appropriate sound level can be determined
from this plot along with the corresponding configuration for the thickness and cross-
section of the stiffeners, which are available from previous optimization solutions.
49
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
60/83
68.50
69.00
69.50
70.00
70.50
71.00
71.50
72.00
72.50
73.00
73.50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
SoundindB
Iteration Number
Figure 4.6 Iteration History for Sound
21.00
22.00
23.00
24.00
25.00
26.00
27.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
FreuencinHz
Figure 4.7 Iteration History for Frequency
Iteration Number
50
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
61/83
250.00
255.00
260.00
265.00
270.00
275.00
280.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
MassinK
g
Figure 4.8 Iteration History for Mass
Iteration Number
0
5
10
15
20
25
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Ring Width Ring Height
Shell Thickness Long. width
Long. Height
DesignVariablesinmm
Iteration Number
Figure 4.9 Changes in Design Variables with Iterations
51
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
62/83
Shell
Thickness (m)
Ring
Width (m)
Ring Height
(m)
Longitudinal
Stiffener
Width (m)
Longitudinal
Stiffener
Height (m)
0.0091 0.005 0.005 0.022 0.013
Mass (Kg) Sound (dB) Frequency (Hz)
274.47 70.00 25.76
Table 4.3 Torpedo Optimal Configuration
Figures 4.5, 4.6, and 4.7 show the histories of the objective and constraints as the
optimization iteration progresses. The initial values for ring and longitudinal stiffener
widths are given as 0.015 m, and their thicknesses are taken as 0.01 m, respectively. The
initial shell thickness is taken as 0.0635 m. An inverse relation between the sound and
mass of the structure is evident from these plots. Figure 4.8 shows the variation in all the
design variations with optimization iterations. From this figure, it is clear that the
optimizer is pushing the ring stiffener dimensions to the lower limits and the shell
thickness and longitudinal stiffener dimensions are increased to reduce noise. From
Figure 4.8, it can be observed that shell thickness is the most important design variable
here. Table 4.3 shows the optimal configuration of the torpedo structure from one of the
several optimization runs required to get the Pareto frontier. The table also shows the
weight of the structure and the corresponding sound level at a critical location. This
critical location was determined for one particular configuration, and kept constant in
order to have a continuous function definition for all of the iterations in the optimization
52
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
63/83
problem. In reality, as the structural model is changed, the location of maximum sound
intensity changes. However, it is assumed in this research that if the intensity at the fixed
critical location is reduced, then the intensity at other locations is also reduced which was
verified to be true.
53
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
64/83
5. Concluding Remarks
In this research, an acoustic optimization methodology is presented for a computational
model of a lightweight torpedo using the finite element method to model both the fluid
and the structure. Fluid and structural models are coupled to incorporate the effect of the
fluid-structure interaction. As it can be seen from the numerical results, the fluid-structure
interaction and the infinite boundary conditions are critical for the acoustic analysis of
underwater structures. This research has shown that the noise profile generated by the
gear machinery demonstrates spatial nonlinearity, which cannot be represented by the
pulsating force models used by many researchers. Therefore, experimental results and the
corresponding computational noise source models are very important for determining the
acoustic signature of torpedo structures. The optimization problem solved in this work
gives the relative trade-off between the mass of the structure and the sound emitted by it
due to gear noise.
Future work in this area can be directed towards acoustic analysis and optimization of
composite structures based on the source model developed in this research. With the
introduction of composite models, active and passive damping techniques can be
explored through embedded systems in the torpedo shell. However, before the
development of damping technologies is possible, all other sources need to be modeled
and incorporated into the torpedo model. This can only be done through experimental
54
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
65/83
data collection for noise from the various sources, as discussed earlier. Finally, this thesis
outlines the steps involved in the acoustic design of an underwater vehicle with a
realistically modeled noise source.
55
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
66/83
APPENDIX
Appendix A
This appendix has C++ code to generate structural model of lightweight torpedo in the
NASTRAN format.
#include
#include
#include
#include
void main()
{
charoutfile[]="Torpedowithoffset.dat";int j=0,k=0;
FILE *stream2;
stream2=fopen(outfile,"w");//Executive Control
fprintf(stream2,"%s\n","ID");
fprintf(stream2,"%s\n","SOL 103");fprintf(stream2,"%s\n","CEND");
fprintf(stream2,"%s\n","SPC=20");
fprintf(stream2,"%s\n","METHOD = 15");
fprintf(stream2,"%s\n","BEGIN BULK");fprintf(stream2,"%s%11d%24d\n","EIGRL",15,20);
int n_x, n_y,n_t, n_r, x,y,r, N1;
double Dia, Len, Nose, th, xc,ti,h,ofs;
Dia=0.32; //Diameter of body
Len=2.0; //Length of bodyNose=0.16; //Length of nose
n_x=40; //Number of rings(node rings) on body
n_y=3; //Number of rings(node rings) on nose
n_t=7; //Number of rings(node rings) on tailn_r=24; //Number of nodes on each ring
ti=0.00635; //Thickness of shell
h=0.010; //Height of ring stiffeners
ofs=(ti/2)+(h/2); //offset for ring stiffnersdouble GridCoordX[10000],GridCoordY[10000],GridCoordZ[10000], theta, X_coord;
theta=360.0/24.0;N1=0;
X_coord=2.00/40.0;
double p = 3.1416;
charGrid[]="GRID";int GridNum=1;
xc=1.0;
// rear tip node
GridCoordX[N1]=-0.30;
GridCoordY[N1]=0.0;GridCoordZ[N1]=0.0;
fprintf(stream2,"%s%12d%16.2f%8.2f%8.2f\n",Grid,GridNum,GridCoordX[N1],GridCoordY[N1],GridCoordZ[N1]);
GridNum=GridNum+1;
N1=N1+1;// Tail section
//-30
th=1.0;
for(r=1;r
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
67/83
GridCoordZ[N1]=0.5*0.18*sin(th*theta*p/180.0);
fprintf(stream2,"%s%12d%16.4f%8.4f%8.4f\n",Grid,GridNum,GridCoordX[N1],GridCoordY[N1],GridCoordZ[N1]);
GridNum=GridNum+1;
N1=N1+1;
th=th+1.0;}
//-25th=1.0;
for(r=1;r
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
68/83
GridCoordZ[N1]=0.5*0.28*sin(th*theta*p/180.0);
fprintf(stream2,"%s%12d%16.4f%8.4f%8.4f\n",Grid,GridNum,GridCoordX[N1],GridCoordY[N1],GridCoordZ[N1]);
GridNum=GridNum+1;
N1=N1+1;
th=th+1.0;}
//0th=1.0;
for(r=1;r
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
69/83
// Nodes for the orientation of the rings(GO entry in the nastran card)
int nor;double R;
R=0.1;
for(nor=1;nor
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
70/83
fprintf(stream2,"%s%10d%8d%8d%8d%8d%8d\n",ELEM,ElmNum,3,j,j+1,j+n_r+1,j+n_r);ElmNum=ElmNum+1;
j=j+1;}
fprintf(stream2,"%s%10d%8d%8d%8d%8d%8d\n",ELEM,ElmNum,3,j,j-n_r+1,j+1,j+n_r);
ElmNum=ElmNum+1;
j=j+1;}
//Nose tip elements
for(r=1;r
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
71/83
x=x+3.0;//3X=X+0.15;//15
a1=a1+1;}
//longitudinal bars
double J[30];int i;
J[0]=196.0;J[1]=199.0;
J[2]=202.0;
J[3]=205.0;J[4]=208.0;
J[5]=211.0;
J[6]=214.0;
J[7]=217.0;Y=0;
Z=0;
for(i=0;i
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
72/83
int el,mp;
int intergrid[200];el=0;
j=196;
for(x=1;x
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
73/83
Appendix B:
This appendix gives the MATLAB file to read the NASTRAN output file and to calculate
error.
%This code reads a .pch file and stores the decibel values at ten
%different locations and calcualtes error in the sound
%levels relative to actual value.
clc
clear all
%opens air_shell_nastran.pch in read format
fid=fopen('noisefile1.pch','r');
i=1;
for i=1:90
tline=fgets(fid);
end
aa=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);tline=fgets(fid);
tline=fgets(fid);
bb=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
cc=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
dd=sscanf(tline,'%*60c %12c %*8c');tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
ee=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
ff=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);gg=sscanf(tline,'%*60c %12c %*8c');
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
tline=fgets(fid);
hh=sscanf(tline,'%*60c %12c %*8c');
a=str2num(aa)
b=str2num(bb)
63
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
74/83
c=str2num(cc)
d=str2num(dd)
e=str2num(ee)
f=str2num(ff)
g=str2num(gg)
h=str2num(hh)
S1=[a b c d e f g h]
fclose(fid);
no_exp=116;
% % This line calculates the error value.
Errorsquare=sqrt((a-103)^2+(b-106)^2+(c-109)^2+(d-110)^2+(e-110)^2+(f-
110)^2+(g-114)^2+(h-116)^2);
fid1=fopen('ErrorSquare.dat','w');
fprintf(fid1,'%10.6f\n',Errorsquare);
fclose(fid1);
64
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
75/83
Appendix C:
This appendix gives the input file for the NASTRAN Optimization Run.
NASTRAN REAL = 63000000
NASTRAN SYSTEM(151)=1
INIT MASTER(NORAM)INIT DBALL LOGICAL=(DBALL(500000))
INIT SCRATCH(NOMEM) LOGICAL=(SCRATCH(1800000)),SCR300=(SCR300(1800000))
ASSIGN OUTPUT2='final_optimization.op2',UNIT=12
ID NoiseModelling
SOL 200
DIAG 8,12
ECHOOFF
CEND
$$$$$$$$$$$$$$$$$$$ CASE CONTROL BEGINS HERE
ECHO=NONE
set 44 = 5951
TITLE=NOISE MODELLING ANALYSIS
SPC=10$DESGLB = 5
DESOBJ(MIN) = 33
subcase 1
ANALYSIS = MODES
METHOD(STRUCTURE)=10
DESSUB = 55
subcase 2
ANALYSIS = MFREQ
METHOD(STRUCTURE)=20
METHOD(FLUID)=20
DLOAD=70
FREQUENCY=15
DISPLACEMENT(SORT1,PRINT,PUNCH)=44
FORCE(SORT1,PRINT,PUNCH)=44DESSUB=117
$ DESOBJ(MIN) = 100
$$$$$$$$$$$$$$$$$$$ BULK DATA BEGINS HERE
BEGIN BULK
ECHO=NONE
EIGRL,10,,,50
EIGRL,20,,,50
$ Define the design variables
DESVAR,1,ringwid,0.015,0.005,0.03
DESVAR,2,ringthic,0.01,0.005,0.03
DESVAR,3,shelthic,0.00635,0.002,0.02
DESVAR,4,longwid,0.015,0.005,0.03
DESVAR,5,longthic,0.01,0.005,0.03$ Relate the Design Varibles to change in stiffener and shell
thicknesses
DVPREL1,1,PSHELL,1,4,0.002,,,,+
+,3,1.0
DVPREL1,2,PSHELL,4,4,0.002,,,,+
+,3,1.0
DVPREL1,3,PSHELL,3,4,0.002,,,,+
+,3,1.0
DVPREL1,4,PBARL,1,12,0.005,,,,+
65
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
76/83
+,1,1.0
DVPREL1,5,PBARL,2,12,0.005,,,,+
+,4,1.0
DVPREL1,6,PBARL,1,13,0.005,,,,+
+,2,1.0
DVPREL1,7,PBARL,2,13,0.005,,,,+
+,5,1.0
$Define Objective
DRESP1,1,DRUCK,FRDISP,,,1,77.8586,5951
DRESP1,2,DRU,FRDISP,,,7,77.8586,5951
DRESP2,100,BETA,100,
,DRESP1,1,2
$1234567$1234567$1234567$1234567$1234567$1234567$1234567$1234567$123456
7
DEQATN 100 OBJ(R,I) = 20.0 * LOG10(SQRT((R * * 2) + (I * * 2))
/(2.0E-5) )
DCONSTR,117,100,,70.0
$ Define Constraint on weight
DRESP1,33,WEIGHT,WEIGHT,
$DCONSTR,5,22,100.0,2500.0
$ Define Constraint on frequencyDRESP1,3,frequ,FREQ,,,1,
DCONSTR,55,3,23.0
$ Override default optimization parameters
DOPTPRM,DESMAX,20,p1,1,p2,15,CONV1,1E-6
,IPRINT,3
$ Grid points of the Torpedo
GRID 1 -0.30 0.00 0.00
GRID 2 -0.3000 0.0869 0.0233
GRID 3 -0.3000 0.0779 0.0450
GRID 4 -0.3000 0.0636 0.0636
GRID 5 -0.3000 0.0450 0.0779
. . . . .
. . . . .
. . . . .
GRID 1213 1.6000 0.0000 0.0000
GRID 1214 1.7500 0.0000 0.0000
GRID 1215 1.9000 0.0000 0.0000
$
$Grid Points of the surrounding water
$
GRID* 1501 2.9100000000
0.0000000000
* 0.0000000000 -1
GRID* 1502 1.9996380000
0.9959744000
* 0.0000000000 -1
GRID* 1503 1.9996380000 -0.9959744000
* 0.0000000000 -1
. . . . .
. . . . .
. . . . .
GRID* 24964 -0.2567208000 -
0.3616001000
* -0.1620268000 -1
66
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
77/83
GRID* 24965 1.7692490000 -
0.7309163000
* -0.2201162000 -1
$
$ Grid points of the Air
$
GRID 25001 0.05000 0.11314 0.11314-1
GRID 25002 0.05000-0.11314 0.11314-1
GRID 25003 0.05000 0.09868 0.12595-1
. . . . .
. . . . .
. . . . .
GRID 31663 0.66061-0.10952-0.10952-1
GRID 31664 0.68030-0.10952-0.10952-1
$ This Completes Grid Data
$Now Elements for the Torpedo
CTRIA3 1 1 2 3 1
CTRIA3 2 1 3 4 1
. . . . . .
. . . . . .
. . . . . .
CTRIA3 23 1 24 25 1
CTRIA3 24 1 25 2 1
CQUAD4 25 1 2 3 27 26
CQUAD4 26 1 3 4 28 27
. . . . . . .
. . . . . . .
. . . . . . .
CQUAD4 1199 3 1176 1177 1201 1200
CQUAD4 1200 3 1177 1154 1178 1201
CTRIA3 1201 3 1178 1179 1202
CTRIA3 1202 3 1179 1180 1202
. . . . . .
. . . . . .
. . . . . .
CTRIA3 1223 3 1200 1201 1202
CTRIA3 1224 3 1201 1178 1202
CBAR 1225 1 194 195 1203
0.00000-0.00790-0.00212 0.00000-0.00708-
0.00409
CBAR 1226 1 195 196 1203
0.00000-0.00708-0.00409 0.00000-0.00578-
0.00578
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
CBAR 1839 2 1081 1105 1209-0.00000-0.00817-0.00000-0.00000-0.00817-
0.00000
CBAR 1840 2 1105 1129 1209
-0.00000-0.00817-0.00000-0.00000-0.00817-
0.00000
CONM2 1841 196 1.9321
CONM2 1842 199 1.9321
. . . . .
. . . . .
67
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
78/83
. . . . .
CONM2 1943 1078 1.9321
CONM2 1944 1081 1.9321
RBE2 2000 1202 1 1106 1107 1108 1109
1110
1111 1112 1113 1114 1115 1116 1117
1118
1119 1120 1121 1122 1123 1124 1125
1126
1127 1128 1129
RBE2 3000 1 1 170 171 172 173
174
175 176 177 178 179 180 181
182
183 184 185 186 187 188 189
190
191 192 193
$Now Elements for water
CTETRA 5001 400 1728 4392 3106 4391
CTETRA 5002 400 1728 3106 1729 4391. . . . . . .
. . . . . . .
. . . . . . .
CTETRA 127228 400 20508 22572 21832 21161
CTETRA 127229 400 20508 22572 24965 21832
$ Elements for Air
CHEXA 130001700 25001 25003 25053 25052 25425
25457
27057 27025
CHEXA 130002700 25003 25004 25065 25053 25457
25489
27441 27057
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
CHEXA 135576700 31280 31664 26320 26352 25380
25392
25225 25226
CHEXA 135577700 31664 26256 25872 26320 25392
25223
25211 25225
$Infinite Boundary Condition
CAABSF 150001 4000 1501
CELAS1 140001 5000 1501 1
CAABSF 150002 4000 1502CELAS1 140002 5000 1502 1
. . . . .
. . . . .
. . . . .
CAABSF 154372 4000 9637
CELAS1 144372 5000 9637 1
CAABSF 154373 4000 9638
CELAS1 144373 5000 9638 1
$End of infinite boundary condition data
68
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
79/83
PAABSF,4000,,,,,2.2572e12
PELAS,5000,9.7169e-5
PSHELL 1 101 0.00635 101
PSHELL 4 102 0.00635 102
PSHELL 3 103 0.00635 103
PBARL,1,300,,BAR
,1.5e-2,1.0e-2
PBARL,2,300,,BAR
,1.5e-2,1.0e-2
MAT1 1017.0E+010 0.33 2780.00
MAT1 1027.0E+010 0.33 2780.00
MAT1 1037.0E+010 0.33 2780.00
MAT1 3007.0E+010 0.33 2780.00
SPC1 10 123 1
SPC1 10 1 2 THRU 25
SPC1 10 2 175 187
SPC1 10 3 181 193
MAT10,800,2.2+9,1026.0
PSOLID,400,800,,,,,PFLUID
$Acoustic Source Dafinition for source one
ACSRCE,70,500,,,2000,1.21,142355.3TABLED1,2000,,,,,,,,+T1
+T1,0.0,0.0,77.76,0.0,77.86,1.0,77.96,0.0,+T2
+T2,1000.0,0.0,ENDT
DAREA,500,25151,1,0.904778,29150,1,0.140919
FREQ,15,77.858604
$Material properties for the fluid
MAT10,900,,1.21,343.0
PARAM,GRDPNT,0
PSOLID,700,900,,,,,PFLUID
PARAM,POST,-2
PARAM,PREFDB,2.-5
PARAM RMS,YES
ACMODL,diff,,,,0.04
PARAM,AUTOSPC,NO
Param,Prgpst,No
ENDDATA
69
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
80/83
References
1. Allen, M. J., Vlahopoulos, N., Integration of Finite Element and Boundary
Element Methods for Calculating the Radiated Sound from a Randomly Excited
Structure, Computers and Structures, Vol. 77, No. 2, pp. 155-169, 2000.
2. Everstine, G. C., Henderson, F. M., Coupled Finite Element/Boundary Element
Approach for Fluid-Structure Interaction, Journal of Acoustical Society of
America, Vol. 87, 1990.
3. Zienkiewicz, O. C., Newton, R. E., Coupled Vibrations of a Structure
Submerged in a Compressible Fluid, Proceedings of International Symposium on
Finite Element Techniques, Stuttgart, pp. 359-379, 1969.
4. Everstine, G. C., Schroeder, E. A., Marcus, M. S., The Dynamic Analysis of
Submerged Structures, Nastran Users Experience, NASA TM X-3278,
Washington, 1975.
5. Pinsky, P. M., Abboud, N. N., Transient Finite Element Analysis of the Exterior
Structural Acoustics Problem, In: Numerical Techniques in Acoustic Radiation,
American Society of Mechanical Engineers, New York, pp. 35-47, 1989.
6. Marcus, M. S., A Finite Element Method Applied to the Vibration of Submerged
Plates,Journal of Ship Research, Vol. 22, pp. 94-99, 1978.
7. Everstine, G. C., Finite Element Formulations of Structural Acoustic Problems,
Computers and Structures, Vol. 65, pp. 307-321, 1997.
8. Ruzzene, M., Baz, A. Finite Element Modeling of Vibration and Sound
Radiation from Fluid-Loaded Shells,Journal of Thin-Walled Structures, Vol. 36,
pp. 21-46, 2002.
70
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
81/83
9. Everstine, G. C., A Symmetric Potential Formulation for Fluid-Structure
Interaction,Journal of Sound and Vibration, Vol. 79, pp. 157-160, 1981.
10.Everstine, G. C., Finite Element Solution of Transient Fluid-Structure Interaction
Problems, 19th
NASTRAN Users Colloquium, NASA CP-3111, Washington,
DC, pp. 162-173, 1991.
11.Everstine, G. C., Structural Analogies for a Scalar Field Problem, International
Journal of Numerical Methods in Engineering, Vol. 17, pp. 419-429, 1981.
12.Hunt, J. T., Knittel, M. R., Barach, D., Finite Element Approach to Acoustic
Radiation from Elastic Structures,Journal of Acoustical Society of America ,Vol.
55, pp. 269-280, 1974.
13.Hunt, J. T., Knittel, M. R., Nichols, C. S., Barach, D., Finite Element Approach
to Acoustic Scattering from Elastic Structures, Journal of Acoustical Society of
America, Vol. 57, pp. 287-299, 1975.
14.Everstine, G. C., Henderson, F. M., Lipman, R. R., Finite Element Prediction of
Acoustic Scattering and Radiation from Submerged Elastic Structures,
Proceeding of 12thNASTRAN Users Colloquium,pp. 194-209, 1984.
15.Zienkiwicz, O. C., Taylor, R. L., The Finite Element Method, McGraw-Hill, Vol.
2, 1989.
16.Choi, K. K., Dong, J., Design Sensitivity Analysis and Optimization of High
Frequency Radiation Problems Using Energy Finite Method and Energy
Boundary Element Method, 10th AIAA/ISSMO Multidisciplinary Analysis and
Optimization Conference, AIAA 2004-4615, 30th
August-1st
September, Albany,
New York, 2004.
71
-
8/4/2019 Acoustic Optimization of an Underwater Vehicle
82/83
17.Zhang, W., Wang, A., Vlahopoulos, N., An Alternative Energy Finite Element
Formulation Based on Incoherent Orthogonal Waves and its Validation for
Marine Structures, Finite Elements in Analysis and Design, Vol. 38, pp. 1095-
1113, 2002.
18.Geers, T. L., Felippa, C. A., Doubly Asymptotic Approximations for Vibration
Analysis of Submerged Structures, Journal of Acoustical Society of America,
Vol. 57, Issue. 4, pp. 1152-1159, 1975.
19.Everstine, G. C., Taylor, D. W., A NASTRAN Implementation of the Doubly
Asymptotic Approximation for Underwater Shock Response, NASTRAN Users
Experiences, pp. 207-228, 1976.
20.Rosen, M. W., Acoustic Studies on Power Transmission in Underwater Missile
Propulsion, 1st
ed., Arlington: Compass Publications, pp. 357379, 1967.
21.Kinsler, L. E., Frey, A. R., Coppens, A. B., Sanders, J. V., Fundamentals of
Acoustics, Wiley, NY, 1982.
22.Dudley, D. W., Introduction to Acoustical Engineering of Gear Transmission,
in Underwater Missile Propulsion, 1st
ed., Arlington: Compass Publications, pp.
349357, 1967.
23.NASTRAN Users Manual, Mac-Neal Schwendler Corporation, Los Angeles, CA,
2001.
24.Pierce, A. D., Acoustics: An Introduction to its Principles and Applications,
1st
ed., Acoustical Society of America, 1989.
25. DOT Users Manual, Vanderplaats Research & Development, Inc., Colorado
Springs, CO, 2001.
72
-
8/4/2019 Acoustic Optimi