Copyright 2009 by ASME 1

Proceedings of IDETC/CIE 2009

ASME 2009 International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference
August 30-September 2, 2009, San Diego, CA, USA
DETC2009/DAC-87276
MICROSTRUCTURE-MEDIATED INTEGRATION OF MATERIAL AND
PRODUCT DESIGN UNDERSEA SUBMERSIBLE

Ayan Sinha
Undergraduate student
Madhusudan Chakraborty
Professor
Sudipto Ghosh, C.S. Kumar
Associate Professors
Indian Institute of Technology, Kharagpur, India

Jitesh H. Panchal
Assistant Professor
School of Mechanical and Materials Engineering
Washington State University, Pullman, WA, USA
Janet K. Allen, David L. McDowell, Farrokh Mistree
Professors
Woodruff School of Mechanical Engineering
Georgia Institute of Technology, Atlanta, GA, USA

Contributors
S. Bagchi, S. Lenka, A.Patra, M. K. Singh, A. K. Srivastava T. K. Kundu
Undergraduate students Assistant Professor
Indian Institute of Technology, Kharagpur, India

ABSTRACT
In this paper, we introduce the construct of microstructure-
mediated design of material and product. The microstructure
of the material is controlled within feasible bounds to achieve
the performance targets of the product. We illustrate the
efficacy of this construct via the integrated robust design of a
submersible and Al-based matrix composites. The integrated
design is carried out using an Inductive Design Exploration
Method (IDEM) that facilitates robust design in the presence
of model structure uncertainty (MSU).

Model structural uncertainty (MSU), originating from
assumptions and idealizations in modeling processes, is a form
of uncertainty that is often virtually impossible to quantify. In
this paper, we demonstrate a method, the Inductive Design
Exploration Method (IDEM), that facilitates robust design in
the presence of model structural uncertainty. We achieve
robustness by trading off the degree of system performance
and the degree of reliability based on structural uncertainty
associated with system models (i.e., models for performances
and constraints). IDEM is demonstrated in the design of a
shell of a robotic submersible. The material considered is in-
situ Al metal matrix composites (MMCs) due to the
advantages that the in-situ MMCs have over the conventional
MMCs. This design task is a representative example of
integrated materials and product design problems.

Keywords: Microstructure mediated design, robust design,
inductive design exploration

Nomenclature
B Buoyant weight of the submersible
d Grain diameter
d
p
Reinforcement size
eff Efficiency of the battery
g Gravity
h Depth of the submersible below water
ID Inner diameter of the shell
k
y
Strengthening coefficient in the Hall-Petch
relation
L Length of the submersible
OD Outer diameter of the shell
P External pressure
T Semisolid processing temperature
T
opr
Endurance time of the submersible
t Thickness of the shell
W Weight of the cylindrical shell
x
Cu
Volume fraction of Cu
2
TiB
x
Volume fraction of TiB
2

Copyright 2009 by ASME 2
Y Output of a response surface model

ij
Coefficients in a response surface model
Density of the composite

TiB2
,
Cu
,
Al
Densities of TiB
2
, copper and aluminum
respectively

w
Density of water
Overall yield stress incorporating Orowan
particle bypass

o
Material constant related to lattice
resistance

y
Yield stress calculated from the Hall-Petch
relation
1. FRAMING THE PROBLEM
Traditionally materials are selected from databases of
experimentally determined materials properties. However, the
paradigm is shifting towards the concurrent design of
materials and products. This entails tailoring materials for
specific performance required in specific products or
processes.

In order to tailor materials, the approach taken by materials
scientists is sequential deductive analysis, with a bottom-up
mapping from processing path to nano- and micro-structure,
material properties and performance. This corresponds to
Olsons materials design hierarchy [22] shown in Figure 1.

The microstructure of a material strongly influences physical,
mechanical and chemical properties such as strength,
toughness, ductility, corrosion resistance, high/low
temperature behavior, etc., which in turn govern the
application of these materials. The microstructure represents
the interface between structure-property-performance relations
including systems design and process-structure relations. A
microstructure-mediated design-centered approach has been
adopted for concurrent design of materials and product.

A systems-based approach has been adopted. This combines
inductive (top-down) engineering with deductive (bottom-up)
science; see Figure 1. Fundamental to this design approach is
an interconnected system of modules (a design process chain)
expressed in terms of variables, constraints, and models that
embed relevant aspects of the material microstructures through
overall system configuration.

Figure 1 Hierarchical Materials Design [22]

In this paper, the method is illustrated through the design of
the shell and design of the material from which the shell of an
submersible is made. The shell is characterized by both
geometrical and material features; see Figure 2 and Refs. 14
and 15. The objective is to design the shell of a robotic
submersible for deep sea exploration with the multifunctional
requirements of minimizing the mass in walls (wall thickness)
for given support superstructure for given maximum depth and
associated pressure differential. Other design requirements
include a) suitable factor of safety with respect to collapse at
target maximum operating depth, b) a large endurance time
satisfying the time of operation constraints under water
without resurfacing/refueling/battery changes, c) satisfying
geometric and weight constraints. The preferred design must
have a) high strength to weight ratio and b) resistance against
environmental factors such as corrosion. Recent advances in
material processing allow designing the material to attain
specific desired properties.



Figure 2 Pressure Shell of a Submersible Robot

Al-based metal matrix composite is used to illustrate the
proposed method. Metal matrix composites (MMCs), in
general, and Al-based MMCs in particular, have been the
subject of intense research for the past two to three decades
and are being exploited for a range of commercial applications
related to aerospace and automotive industries. Al-based metal
matrix composites can be divided into two classes,, namely,
ex-situ and in-situ. In ex-situ composites the reinforcements
are added externally [16, 21, 24] whereas in in-situ composites
the reinforcing particulates are formed by chemical reaction
within the liquid melt. One of the important drawbacks during
the processing of ex-situ MMCs is the presence of interfacial
impurities and oxides between reinforcement and matrix
resulting in poor wettability and bonding. This has led to the
development of in-situ composites, wherein the
reinforcements are generated in a metallic matrix via chemical
reactions between elements and/or compounds with Al alloy
melt during the composite fabrication. The advantages that in-
situ MMCs have over conventional MMCs include
thermodynamically stable reinforcements in the matrix, clean
reinforcement-matrix interfaces resulting in a strong
interfacial bonding, finer particle size yielding better
mechanical properties and potential for lower cost of
production. These advantages make it a strong candidate for
the design task at hand. On the other hand, the reinforcement
particles in in-situ composites are subject to strong segregation
effects and therefore post solidification process strategies are
necessary to more uniformly mix the particles.

Processing
Structure
Properties
Performance


Processing
Structure
Properties
Performance
Copyright 2009 by ASME 3
2. MICROSTRUCTURE-MEDIATED DESIGN
The design approach is based on systems-based integrated top-
down (inductive) and bottom-up (deductive) multilevel design
as illustrated in Figure 3. Multilevel design for the shell design
problem involves two activities, namely, process path -
structure relationships and structure-property-performance
relationships. These two design objectives interact via the
microstructure. While on one hand the processing conditions
influence the obtained microstructure, the performance of the
product depends on the mechanical properties which in-turn
are mapped from the microstructure.

In the present study, two major aspects of the design problem,
namely, the materials design (rather than just materials
selection) and structural design, are combined. The materials
design aspect has been divided into three parts based on the
different processing steps of the material. The interface
between materials design and structural design is the mapping
of the processed microstructure to the required mechanical
properties.

The Inductive Design Exploration Method (IDEM) is used to
effect solution. The design process chain for this application
constitutes of six interconnected modules. Five modules
account for the modeling of the behavior of the material and
the structure. The sixth is used to address uncertainty
embodied in the simulation models, the management of
uncertainty propagation and tools for design exploration in the
presence of propagated uncertainty in the design process
chain. Based on the materials processing steps involved and
mechanical design requirements, the interconnected modules
that constitute the design process chain for this application are
(see Figure 3):
MODULE 1: Precipitation modeling in liquid Al.
MODULE 2: Modeling of microstructure evolution in MMCs.
MODULE 3: Evolution of microstructure during semisolid
processing of MMCs.
MODULE 4: Structure - property correlations of MMCs.
MODULE 5: Requirement list, microstructure mapping and
system-level design.
MODULE 6: Robust design strategy using IDEM to address
model structure uncertainty and propagated
uncertainty among levels of models.

MODULEs 1, 2 and 3 provide the simulated microstructure
after processing. The resulting mechanical properties are
estimated in MODULE 4, whereas MODULE 5 maps the
required mechanical properties based on the system design
considerations.

Given the complexity inherent in the design process chain, it is
important to define the variables, the interface and the design
constraints between the different modules. In Figures 4, 5 and
6 we show the analysis, interface and the respective integrated
flow diagrams for this design process chain. In the analysis
diagram [Figure 4] we show the various independent and
dependent variables in the six modules of the design process
chain. In the interface diagram [Figure 5] we map the
connectivity and flow of information between the modules.
















Figure 3 Microstructure Mediated Design of Material and
Structure
















Figure 4 Analysis Diagram









Figure 5 Interface Diagram
Geometric
Parameters
MODULE 4
Structure property
correlation of MMCs
MODULE 5
Requirement list,
microstructure
mapping and design
Mechanical
Properties
MODULE 6
Robust design
using IDEM
Perfor-
mance
Constraints
1.Stress conditions
2.Heat transfer
3.Shock response
Constraints
Range of mechanical
properties
Phases,
Ppt size
Init microstructure
& ppt distribution
MODULE 1
Pptn modeling in
liquid aluminum
MODULE 2
Modeling of
microstructure
evolution in MMCs
MODULE 3
Semisolid
processing of
MMCs
Constraints
Max. volume fraction of
reinforcement
1. Temperature field
2. Solutal field
Constraints
1.Range of working
temperature
2.Shear stress
1.Rolling parameters
2.Temperature
Constraints
Mass transfer phenomenon
(convection)
1. Composition
2. Processing temp
3. Rxn time
Geometric
Parameters
MODULE 4
Structure property
correlation of MMCs
MODULE 5
Requirement list,
microstructure
mapping and design
Mechanical
Properties
MODULE 6
Robust design
using IDEM
Perfor-
mance
Constraints
1.Stress conditions
2.Heat transfer
3.Shock response
Constraints
Range of mechanical
properties
Phases,
Ppt size
Init microstructure
& ppt distribution
MODULE 1
Pptn modeling in
liquid aluminum
MODULE 2
Modeling of
microstructure
evolution in MMCs
MODULE 3
Semisolid
processing of
MMCs
Constraints
Max. volume fraction of
reinforcement
1. Temperature field
2. Solutal field
Constraints
1.Range of working
temperature
2.Shear stress
Constraints
1.Range of working
temperature
2.Shear stress
1.Rolling parameters
2.Temperature
Constraints
Mass transfer phenomenon
(convection)
1. Composition
2. Processing temp
3. Rxn time
Init. Micro-
structure, ppt.
distribution
[Templates]
MODULE 1
Precipitation
modeling in
liquid aluminum
1.Phases
formed
2. Ppt size
MODULE 2
Modeling
microstructure
evolution in
MMCs
MODULE 3
Semisolid
processing of
MMCs
Final microstructure after
semisolid processing
[Templates]
MODULE 4
Structure - Property
correlation of MMCs
1. Composite composition
2. Temp. of processing
3. Time of reaction
[Templates]
Reqd mech. properties
[Templates]
Obtained mech. properties [Templates]
MODULE 6
Robust design
using IDEM
Interfacevariables of
Projects 1, 2, 3, 4
[Templates]
Design and uncertainty parameters
[Text and Abaqus Output Files]
Modification parameters [Templates]
Ppt. info.
MATERIALS
DESIGN
MECHANICAL
DESIGN
MODULE 5
Requirement list,
microstructure
mapping & design
INTERFACE
Init. Micro-
structure, ppt.
distribution
[Templates]
MODULE 1
Precipitation
modeling in
liquid aluminum
1.Phases
formed
2. Ppt size
MODULE 2
Modeling
microstructure
evolution in
MMCs
MODULE 3
Semisolid
processing of
MMCs
Final microstructure after
semisolid processing
[Templates]
MODULE 4
Structure - Property
correlation of MMCs
1. Composite composition
2. Temp. of processing
3. Time of reaction
[Templates]
Reqd mech. properties
[Templates]
Obtained mech. properties [Templates]
MODULE 6
Robust design
using IDEM
Interfacevariables of
Projects 1, 2, 3, 4
[Templates]
Design and uncertainty parameters
[Text and Abaqus Output Files]
Modification parameters [Templates]
Ppt. info.
MATERIALS
DESIGN
MECHANICAL
DESIGN
MODULE 5
Requirement list,
microstructure
mapping & design
INTERFACE
G
O
A
L
S
/
M
E
A
N
S

(
I
N
D
U
C
T
I
V
E

D
E
S
IG
N
-
I
D
E
M
)
C
A
U
S
E

A
N
D

E
F
F
E
C
T


(
D
E
D
U
C
T
I
V
E
)
Performance
Properties
Processing
Module
5
Module
4
Module
3
Module
2
Module
1
Parameter to be determined
Goal
Given Value or Parameter
Output Response
Parameter to be determined
Goal
Given Value or Parameter
Output Response
Goal
Given Value or Parameter
Output Response
Microstructure
Copyright 2009 by ASME 4
MODULE 1 involves the prediction of the precipitation of
liquid aluminum based on the composition and processing
temperature. The output of MODULE 1 is the information
about different phases formed, the size of precipitates and the
time required to complete the reaction. This information is
used in MODULE 2, which embodies the process of
microstructure evolution and the effect of temperature and
solutal fields on the resulting microstructure. The next step is
the semi-solid processing of the Al-MMCs through a rolling
operation which modifies the materials microstructure. In
MODULE 3 the effect of the rolling parameters on the
resulting microstructure is predicted. In MODULE 4, this
microstructure is used to predict the mechanical properties
inherent in the material. These mechanical properties are used
in the system-level MODULE 5 to predict the effects of
different AUV geometries on overall system performance. As
can be seen from the integrated flow diagram [Figure 6], the
microstructure is the essential link between the design of the
material and the design of the undersea submersible.
Init. Micro-
structure, ppt.
distribution
[Templates]
MODULE 1
Precipitation
modeling in
liquid aluminum
1.Phases
formed
2. Ppt size
[Templates]
MODULE 2
Modeling
microstructure
evolution in
MMCs
MODULE 3
Semisolid
processing of
MMCs
Final microstructure after
semisolid processing
[Templates]
MODULE 4
Structure - Property
correlation of MMCs
1. Composite composition
2. Temp. of processing
3. Time of reaction
[Templates]
Reqd mech. Prop. [Templates]
Obtained mech. properties [Templates]
MODULE 6
Robust design
using IDEM
Interface variables of
Projects 1, 2, 3, 4
[Templates]
Design and uncertainty parameters
[Text and Abaqus Output Files]
Modification parameters [Templates]
Ppt. info.
MATERIALS
DESIGN
MECHANICAL
DESIGN
MODULE 5
Requirement list,
microstructure
mapping & design
INTERFACE
Range of Mech.
Properties
Constraints
Convection
Stress
Heat transfer
Shock response
Max. vol. frac.
TiB
2
Analysis Flow
Variables
Indep. Parameters
Rolling Parameters
Temperature
Syn. Flow Variables

Figure 6 Integrated Flow Diagram
In this application, the strength is principally determined by
the sizes, shapes and distribution of TiB2 precipitates in
other words the microstructure of the material. The
microstructure is determined by processing methods in this
case, it is initially created by precipitation and followed by the
evolution of the precipitate size and distribution during the
semi-solid rolling. The structural design can be modified in
two ways, namely, 1) by changing the processing conditions to
modify the microstructure, which has an effect on the overall
system performance and 2) by changing the geometry of the
shell, which in turn not only affects structural performance,
but also puts constraints on required mechanical properties of
the material. Hence, the material microstructure needs to be
designed in such a way that the constraints on the material
properties, imposed by the structure, are satisfied. Since the
material microstructure acts as the interface between the
material and structure, we have adopted the phrase
microstructure mediated design. Having defined the design
variables and the connectivity within the design process chain,
the modules described in the sections that follow.

2.1 MODULE 1 (Precipitation Modeling in Liquid
Aluminum)
A suitable route (Mixed-Salt route) for the in situ Al / TiB
2

composite manufacturing process utilizes the reduction of
K
2
TiF
6
and KBF
4
with aluminum, generally known as the
halide salt process. Yang and coauthors [31] proposed a
diffusion mechanism wherein Al
3
Ti is formed in the melt
initially by a very fast reaction. Boron then diffuses into Al
3
Ti
particles in the melt, thus forming TiB
2
particles according to
the reaction, Al
3
Ti + 2B = 3Al + TiB
2
.

The liquid-state processing techniques to produce in-situ
composites include self propagating high temperature
synthesis (SHS), exothermic dispersion (XD), reactive hot
pressing (RHP), flux assisted synthesis (FAS) and rapid
solidification processing (RSP). Any of these processes could
be used. K
2
TiF
6
and KBF
4
are other precursors that dissolve in
the aluminum melt to form intermediate phases Al
3
Ti and
AlB
2
. The reaction between these intermediate phases has
been studied to predict the particle size distribution of TiB
2

phase thus formed in the matrix.

A model proposed by Anestiev and coauthors [1] has been
used to investigate the diffusion reactions taking place
between the intermediate phases. In this model, Al
3
Ti and
AlB
2
are allowed to react in liquid Al to form TiB
2

particulates. A coordinate system dividing a 2-dimensional
space into strips of equal length has been used, half of which
contains Al
3
Ti and the other half AlB
2
dissolved in the Al
melt, shown in Figure 7. When these intermediate phases
react, random nucleation of TiB
2
particulates is assumed. The
kinetics of the formation of TiB
2
particles is governed by
unsteady state diffusion equations (Solute redistribution
theory), which in turn depends on the concentration profile of
the intermediate solute phases in the region. The solute
consumption rate due to TiB
2
formation is described by
volume fraction of the region transformed per unit time.
Johnson-Mehl-Avrami analysis [2,12] is used to find the
transformed volume fraction from the nucleation and growth
rates of the particles as: = 1 exp(-ktn) where is is the
volume fraction transformed, k = _ N G
3
/3 and n = 4, N and G
are Nucleation and Growth Rate respectively.
The Nucleation rate is primarily a function of the Gibbs
energy change associated with the formation of the particle,
while the Growth rate also depends on its surface energy. The
thermodynamic models predicting the Gibbs free energies of
the involved phases in the current system are described in
Refs. [19, 26-28]. The kinetics of reinforcement particles can
be mathematically described by the following set of partial
differential equations:
X
1
/t= D(
2
X
1
/x
2
) - X
1
S
(/t),
Copyright 2009 by ASME 5
X
2
/t= D(
2
X
2
/x
2
) X
2
S
(/t)
where, X
1
and X
2
are the mol fractions of the dissolved Ti and
B in the Al matrix respectively, t is the time, D is the diffusion
coefficient, X
1
S
and X
2
S
are the mol fractions of Ti and B in
the solid phase (TiB
2
). The complex diffusion equations are
solved numerically to compute the TiB
2
particle size
distribution across the matrix.






Figure 7 Schematic of Coordinate System
Used in MODULE 1

2.2 MODULE 2 (Modeling Microstructure Evolution)
Microstructural evolution of materials during various material
processes relates key properties such as mechanical strength
and electrical properties to the average grain size and the grain
size distribution, which are direct consequences of the
microstructure evolution. In MODULE 2, microstructure
evolution during solidification depends on the thermal and the
solutal fields. The mathematical description of the dendritic
solidification process of a three component alloy in two
dimensional square solidification domain () is:


The S/L interface evolves in time and has to be found as part
of solution. The solidification of a three component alloy is
governed by the evolution of temperature T(t,x,y) and
concentration field C

(t,x,y) ,where = 1,2 which satisfies

several boundary conditions at the moving S/L interface as
well as the initial and the boundary conditions. The equations
that describe the physics of solidification process follow.

Temperature T in (heat transfer equation):

C
p
( T/ t) = KT + L( f
s
/ t)

where t is time, (x,y) is the domain co-ordinates, is the
density, C
p
is the specific heat, K is the thermal conductivity,
L is the latent heat of solidification and f
s
is solid fraction. For
simplicity the notation f
L
= 1 - f
s
, denotes the liquid fraction.
The concentration (C_) for the solute (solute diffusion
equation)

C
L

/ t = D
L

C
L

For liquid phase

C
S

/ t = D
S

C
S

...For solid phase

where = 1,2, D
L

and D
S

are liquid and solid diffusion

coefficients of solute , respectively. The cross diffusion is
neglected and zero flux boundary conditions are applied to
four wall of simulation domain .

Fluid flow due to forced or natural convection also influences
the microstructure evolution. The present module involves the
numerical solution of continuum equations for thermal fields
and coupling it with a cellular automata model that computes
the evolution of grain structure with solidification time. The
measured flux values are used to derive the evolution of the
thermal fields with solidification time. Using measured
temperature values at the specific points along the metal-mold
interface, realistic flux values at the metal-mold interface can
be derived which can be fed into a Computation Fluid
Dynamics (CFD) modeling tool to obtain accurate thermal
fields across the casting domain. These fields are used in the
cellular automata model to predict the microstructure
evolution as the solidification proceeds.

2.3 MODULE 3 (Semi-solid Processing in MMCs)
The present module deals with the simulation of the semi-solid
processing [9] of metal matrix composites. The actual process
[10, 11 and 17] consists of passing slabs of as-cast composite
material through rollers [Figure 8] at such a temperature that
part of it is in semi-solid or mushy state. Two-high mill
rollers of diameter 120 mm and 125 mm barrel width are used
in this process. The sample is heated to temperatures between
610 to 633C to obtain 10 to 30% liquid in the material. When
the slab is passed through the rollers, the grains deform and
rearrange and a nearly homogeneous distribution of TiB
2

particles is obtained. Multiple passes are performed to refine
the grain size. Such a process enhances the properties of the
MMC and homogenizes its composition.

Figure 8 Schematic of Semi-Solid Processing
Since this is a novel process and its physics are not yet fully
understood, an empirical model is used, based on data taken
from a large number of experiments. The model takes as input
the processing conditions of semi-solid processing, including
ratio, and then predicts the final average grain size and also
Interface
Al + AlB
2
Al + Al
3
Ti
Copyright 2009 by ASME 6
gives an approximate microstructure. To predict the final grain
size it takes in the experimental details and interpolates the
grain size. After processing, the TiB
2
particulates rearrange
themselves to achieve a more uniform spatial distribution,
which is also reflected in the model. Using a genetic algorithm
based Voronoi and Monte Carlo code [8], equiaxed globular
grains are created. It forms in 100 x 100 matrix grains
differentiated by different color codes which can be then be
interpreted to render the final microstructure after semi-solid
processing.

2.4 MODULE 4 (Structure-Property Correlation of
MMCs)
2.4.1 Yield Stress: The matrix yield stress is assumed to obey
the Hall-Petch relation, i.e.,


y
=
0
+ k
y
(d)
-0.5
(1)

where k
y
is the strengthening coefficient (a constant unique to
each material; for pure Al, k
y
= 3.4 MPa-mm),
o
is a material
constant related to lattice resistance (for pure Al,
o
= 2.95
MPa), d is the grain diameter, and
y
is the yield stress. The
constants corresponding to matrix properties are assumed to
be that of pure Al. The calculation of overall yield stress ()
also incorporates Orowan particle bypass via dislocation
looping [30], i.e.,

=
y
(1 +f
1
) (1 +f
2
) (1 +f
orowan
) (2)

where f
1
takes the effect of volume fraction of particles, f
2

takes into account the thermal expansion coefficient mismatch
between matrix and reinforcement, and f
orowan
takes into
account the effect of particle size (d) and spacing. It receives
input from outputs of MODULEs 1 and 3, specifically
reinforcement size (d
p
, grain size (d), semisolid processing
temperature (T) and volume fraction of TiB
2
particles.

2.4.2 Density: The determination of density is based on the
average property of each of the constituent phases, i.e.,

=
TiB2
x
TiB2
+
Cu
x
Cu
+
Al
(1-x
TiB2
x
Cu
) (3)

where ,
TiB2
,
Cu
,
Al
are the densities of the composite, TiB
2
,
copper and aluminum respectively. Also, x
TiB2
is the volume
fraction of TiB
2
and x
Cu
is the volume fraction of copper
(typically 6%).

2.5 MODULE 5 (Property-Performance Correlation of
MMCs)
MODULE 5 acts as an interface between the materials design
aspect and the design of the structure of the submersible. The
performance parameters considered are depth, time of
operation and weight of the outer shell of submersible. The
objective is to maximize the depth and time of operation while
minimizing the weight of the outer shell of the submersible.
The formulas used for the calculation of these performance
parameters are stated in what follows [13, 25 and 29].

2.5.1 Model for Depth (h): We use Roarks formula [29] for
thickness (t) to outer diameter (OD) ratio.

|
|

\
|
=

P 2
1 1
2
1
OD
t
(4)

where t is the thickness of the shell, OD is the outer diameter
of the shell; P is the external pressure and (from eq. 2) is the
yield stress of the metal matrix composite. Substituting for P
as
w
gh where
w
is the density of water (1025 kg/m
3
), g is the
gravitational attraction (9.81 m/sec
2
) and h is the depth of
submersible below water. Solving for h we get:

2
w
2t
h 1 1
2 g OD

| |
| |
=
| |
\
\


(5)

2.5.2 Model for Weight (W): The weight of a cylindrical shell
with spherical end caps is calculated.

W = L (OD
2
ID
2
) + (4/3) (OD
3
ID
3
) (6)

where in eq. (3) is the density of the composite, L is the
length of the submersible, OD is the outer diameter and ID is
the inner diameter of the cylindrical shell with spherical end-
caps [Figure 2]. We shall fix the outer diameter (OD) at 260
mm and the length (L) at 1.6 meter. Thickness (t) can vary
from 5 mm to 15 mm as representative parameters of a typical
Autonomous Underwater Vehicle as described in [14, 15].

2.5.3 Model for Endurance Time (T
opr
)


Load propulsion FixedLoad
Density Energy eff W B
T
opr
+

=
) ( 8 . 0
(7)

where B is the buoyant weight of the submersible, W (eq.6) is
the weight of the cylindrical shell, eff is the efficiency of the
battery. The efficiency of a Lithium-Ion battery is typically
60% and its energy density is 128 Watt-Hour/Kg. For the
initial design, assuming a slow moving submersible and
submergence/surfacing rates, we shall ignore propulsion load
in our calculations and assume a fixed electrical load of 400
Watt-Hour which is typical of the control computers and
electronics payloads in a small underwater robotic submersible
[14, 15].

2.6 MODULE 6 (Robust Design using IDEM)
We employ IDEM to achieve a robust multi-level design that
traverses process-structure, structure-property and property-
performance relationships; see Figure 1. IDEM includes
Copyright 2009 by ASME 7
parallel discrete function evaluation, Inductive Discrete
Constraints Evaluation (IDCE) based on Hyper-Dimensional
Error Margin Indices (HD-EMIs), and the Compromise
Decision Support Problem (cDSP) for finding the best solution
under MSU [3-7 and 23]. The overall procedure for the IDEM
is schematically illustrated in Figure 9.

Figure 9 Schematic of IDEM [3]

IDEM is exercised to identify adjustable ranges of control
factor (design variable) values in a system with uncertainty
propagation in a design/analyses process chain and to account
for uncertainty in downstream activities and uncertainty
propagation. With IDEM, a designer can maximize or
maintain ranges of values for design variables or performance
parameters that are shared or linked with another designers
robust design process. Thereby, design freedom is preserved
for another collaborating designer who can make changes to
the designwithin specified rangeswithout compromising
design requirements.

IDEM facilitates multi-level design, the management of
uncertainty inherent in the models and the propagation of
uncertainty through the design process chain shown in Figure
4. In IDEM, we deal with the propagation of uncertainty in
design and analysis modules that constitute the design process
chain for a particular application. We start with performance
and traverse sequentially to process; see Figure 1. At each
level we identify a ranged set of feasible specifications.

IDEM, embodies the concept of Error Margin Indices (EMIs).
EMIs are indicators of the degree of reliability of a decision
that it will satisfy the prescribed system constraints or bounds.
The procedure for obtaining the EMI is as follows: (a) obtain
the upper and/or lower deviation of a response (URL and
LRL) and (b) calculate the EMI from this deviation. The EMI
is calculated by including the response mean (
y
) and
upper/lower deviations (Y
upper
and Y
lower
) from a combined
distribution of a system model and error bounds. The EMI
includes the response deviations of a system model due to
variations in design variables and the response deviations of
error bounds as well as the system model. The mathematical
formulations of EMI corresponding to a goal i are:

( ( )) /
i i i upper
EMI URL f x Y =
for minimization problems;
( ( ) ) /
i i i lower
EMI f x LRL Y = for maximization problems;
i i i i
i
i i
| f URL | | f LRL |
EMI Min{ , }
Y Y
( i 1,2,..., Number of the goals )

=

=

Y
Y
upper upper y
lower y lower
Y
Y

=
=


As shown in Figure 9, the objective is to find the best ranged
set of design specifications in the space x considering
uncertainty in mapping functions (f) and propagated
uncertainty through a design process. IDEM involves finding
ranged sets of design specifications by passing feasible
solution spaces from performance requirements by way of an
interdependent response space to the design space while
preserving the feasible solution space as much as possible. The
procedure includes the following steps [3].
Step 1: Conduct parallel discrete function evaluation:
Define rough design and performance spaces (hyper-
dimensional x, y, and z spaces) and generate discrete
points in each of these spaces.
Evaluate the generated discrete points using the
mapping models (f and g in Figure 9) that include all
quantified amount of uncertainty.
Store the evaluated data sets, including discrete input
points and output ranges, in a database.
Step 2: Inductive Discrete Constraints Evaluation (IDCE)
process: Using information from Step 1, sequentially
identify feasible regions in y and x spaces with a given
initial requirement range in z space
Step 3: Solve the Compromise Decision Support Problem
(cDSP): Find the best robust solution under MSU by
performing Step 2 with adjusted HD-EMIs.
As HD-EMI increases for a particular model, the output range
moves farther from the constraint boundary. This means the
decision becomes more reliable under potential shift of the
output range due to MSU. In the IDCE process, the
specifications, the performance ranges and the initial HD-EMI
values for the discrete constraint evaluation are specified by
the designer. To determine the best solution among feasible
sets of solutions the required HD-EMIs for each space should
be anchored in statistics. Values of HD-EMIs are important in
determining the most desirable robust solution against model
structural uncertainty, because HD-EMIs represent the amount
of margin for potential errors in the mapping models for
estimating output range. A designer may leave wider margins
between an output range and constraint boundaries by
increasing the HD-EMI for the mapping model whose MSU is
larger than others. An additional constraint is that all HD-
EMIs should be greater than or equal to one so that the entire
output range can satisfy the constraints Depending on the
value of required HD-EMI, the identified feasible range may
be large, small, or unattainable. The solution strategy for this
application is outlined in the next section.

Copyright 2009 by ASME 8
3. SOLUTION STRATEGY USING IDEM
The solution strategy for this application is illustrated in
Figure 9. The modeling in MODULE 2 has presented many
challenges and these have yet to be resolved. Hence, it is
bypassed in illustrating our method via this application.















Figure 10 Modules Used in This Application

In Figure 10, f1, f3, f4, f5, f7, f8 and f9 represent the theoretical
or empirical models considered at the different levels of
design. The inputs to MODULE 1 are the volume fraction of
TiB
2
(x
TiB2
) and temperature of processing in degree K (T).The
output of MODULE 1 (f1) is the average TiB
2
particle size (d
p
)
which is one of the inputs to MODULE 4. The independent
inputs to MODULE 3 are volume fraction of TiB
2
(x
TiB2
) and
percentage of liquid in processing (%L) and the output of
MODULE 3 (f3) is the average grain size (d) of
microstructure. MODULE 4 receives inputs from the outputs
of MODULE 1 and 3 along with the independent inputs of
volume fraction of TiB
2
(x
TiB2
) and temperature of semi-solid
processing (temp). MODULE 4 deals with the structure-
property relationships and f4 gives the density () [eq. 3]and f5
gives yield stress ( [eq. 2]) as outputs. Finally, MODULE 5
deals with the property-performance relationship of the
developed microstructure and f7 evaluates the performance
variable of depth of operation (h), f8 evaluates the weight of
the outer shell (W) and f9 evaluates the time of operation (T
opr
)
of the submersible. The independent parameter in this level of
design is the thickness of the shell (t) and the dependent
parameters are density () and yield stress ().

The solution scheme for this application is illustrated in Figure
11. We observe that the that the feasible design spaces are
inductively passed from MODULE 5 to MODULE 4 and
subsequently to MODULES 3 and 1 of design.

We note that the volume fraction of TiB
2
is an input to
MODULE 1, MODULE 3 and MODULE 4 of design. The
responses of MODULE 1 and MODULE 3 are influenced by
multiple variables and hence we use response surface
methodology for modeling and analysis of the design task at
these levels. The Response Surface Methodology employed
embodies second order models [20]:
k k
2
0 i i ii i ij i j
i 1 i 1 i k
Y x x x x
= = <
= + + + +


(8)

where,
ij
,i 1,2,...,k; j 0,1,2,...,k = = are the regression
coefficients and x
j
are the regression variables, Y is the
response. The Response Surfaces for MODULE 1 and
MODULE 3 are generated using MINITAB.TABLE 1 gives
the data set of the variables used to generate the response
surface of MODULE 1.


















Figure 11 Solution Strategy Using IDEM


Table 1 Data Set for MODULE 1
Volume fraction
(xTiB
2
, %)
Temperature
(T, K)
Average particle
radius (r, m)
2.5 1073 0.96
5.0 1073 1.25
7.5 1073 1.22
10.0 1073 1.11
10.0 1173 1.57
10.0 1273 1.74
10.0 1373 1.80


The response surface generated for MODULE 1 is represented
by the equation:
2 2
TiB2 tiB2
Y 17.3246 0.2290x 27.7783T' 0.0167x 10.4230T' = + + (9)
where Y is the response i.e. the average TiB
2
particle grain
radius (d
p/2)
, x
TiB2
is the volume fraction of TiB
2
and T is
T/1000 where T is the temperature in Kelvin.

The data set of the variables used to generate the response
surface of MODULE 3 is shown in TABLE 2.
HD-EMI_5
Geometric
Parameters
Range of vol.
fraction TiB
2
HD-EMI_3
Design space of:
% liquid reduction
Volume fraction TiB
2
HD-EMI_1
Design space of:
% TiB
2
T
Vol. fraction
TiB
2
T
Required Range
Design Space
HD-EMIs
Solution
f3
f1
Range of grain size distribution
HD-EMI_4
f4 f5
Range of depth (D) Range of weight (W)
Range of
time of operation (T)
HD-EMI_7 HD-EMI_8 HD-EMI_9
Range of
f7 f9 f8
Design Space of:
Geometric
Parameters
ID, t, , and rho
Range of
Design space of:
TiB
2
size distribution
Volume fraction TiB
2
Grain size distribution
% liquid reduction
Average grain size, d
HD-EMI_5
Geometric
Parameters
Range of vol.
fraction TiB
2
HD-EMI_3
Design space of:
% liquid reduction
Volume fraction TiB
2
HD-EMI_1
Design space of:
% TiB
2
T
Vol. fraction
TiB
2
T
Required Range
Design Space
HD-EMIs
Solution
f3 f3
f1 f1
Range of grain size distribution
HD-EMI_4
f4 f4 f5 f5
Range of depth (D) Range of weight (W)
Range of
time of operation (T)
HD-EMI_7 HD-EMI_8 HD-EMI_9
Range of
f7 f9 f8 f7 f7 f9 f9 f8 f8
Design Space of:
Geometric
Parameters
ID, t, , and rho
Range of
Design space of:
TiB
2
size distribution
Volume fraction TiB
2
Grain size distribution
% liquid reduction
Average grain size, d
Depth
Weight
Geometric Parameters
OD and t
MODULE 1
Average TiB
2
grain size
Grain size
after
semisolid
processing
f3
% Liquid Reduction xTiB
2
f1
Dependent Variables
Independent Variables
Time of
Operation
T

Temp
f5
f4 f7
f8
f9

MODULE 3 MODULE 4 MODULE 5

Depth
Weight
Geometric Parameters
OD and t
MODULE 1
Average TiB
2
grain size
Grain size
after
semisolid
processing
f3
% Liquid Reduction xTiB
2
f1
Dependent Variables
Independent Variables
Time of
Operation
T

Temp
f5
f4 f7
f8
f9

MODULE 3 MODULE 4 MODULE 5

Copyright 2009 by ASME 9

Table 2 Data Set for MODULE 3
% Volume
fraction TiB
2

% Liquid Grain Size
(m)
2.5 10 62
2.5 20 58
2.5 30 30
5.0 10 54
5.0 20 51
5.0 30 55
7.5 10 62
7.5 20 48
7.5 30 53
10.0 10 49
10.0 20 47
10.0 30 54

The response surface generated for MODULE 3 is represented
by the equation:
2 2
2
2
80.67 0.167 2.25 0.3067
2
0.01375 0.202
TiB ToB
TiB
Y x L x
L x L
=
+ +

(10)
where Y is the response i.e., the average grain size(d), x
TiB2
is
the volume fraction of TiB
2
and L is the % liquid.

4. DISCUSSION OF RESULTS
The primary design variables in the present case for the
submersibles shell [Figure 2] are thickness of the shell (t) and
volume fraction of TiB
2.
(xTiB
2
). Target requirements include:
The safe depth of operation of the submersible with a
small shell thickness should be as large as possible
preferably exceeding 2500 meters and greater depth is
better.
The submersible must have a good endurance with a large
time of operation of at least 12 hours without resurfacing
or recharging and greater duration of submersion is better.
Given a weight of the vessel of 76 Kgs and allowing as large
a payload as feasible, a representative limit the weight of
the outer shell of the submersible may not exceed 18 kgf,
and a lighter shell is preferred.

IDEM has been implemented in MATLAB for this
application. The resolutions for discrete points are fixed as
1(mm), 30 (kg/m
3
), 10 (MPa), and 0.5(%) for thickness (t),
density of composite (), yield stress (
y
), and volume fraction
of TiB
2
(x
TiB2
) respectively. The range for discrete variables are
fixed as 5-15(mm), 2700-3300 (kg/m
3
), 300-500 (MPa), and
2-10(%) for thickness (t), density of composite (), yield stress
(), and volume fraction of TiB
2 (
xTiB
2)
, respectively. These
resolutions are reasonably small in order to be able to ignore
discretization errors. We assume a 5% performance variability
for each of the mapping models (f1, f3, f4, f5, f7, f8 and f9). This
is the sum of all quantifiable uncertainty, including natural
uncertainty and model parameter uncertainty, of each mapping
model. First, we search the entire feasible ranges in property
space (the spaces of t, and ) with given performance
requirements. The required HD-EMIs (HD-EMI7, HD-EMI8,
and HD-EMI9) for mapping models (f7,f8, and f9) are set as
greater than or equal to unity, which means all quantified
uncertainty is accounted for and the performance output range
satisfies the boundary constraints. Among the obtained
feasible spaces of t, , and , we select the value of t
(thickness of shell) that has the largest feasible space for the
rest of the properties ( and ) because we want to maintain
the feasible design domain to be as large as possible until the
end of the design process to achieve robustness under model
structure uncertainty (MSU).

Figure 12 Color Graph for HD-EMI Values

Different colors are used to indicate the variation in the HD-
EMI values for different values of variables [Figure 12]. Red
indicates a HD-EMI value of 5 and the progressively lighter
shades denote lower HD-EMI values. The blue diamond
points in the figures are the boundary points.

When HD-EMI7, HD-EMI8, and HD-EMI9, are assumed as
unity, the largest feasible range in y space is achieved at t = 10
(mm). Satisfactory discrete points (circular points) and
boundary points (diamond points) between the feasible and
infeasible spaces are shown in Figures 13, 14. With the
feasible range achieved in property space, the feasible space of
the volume fraction of TiB2 is identified by setting HD-EMI4
and HD-EMI5 for mapping models (f4 and f5) as unity. Then,
we search the entire feasible ranges in the structure space (the
feasible spaces of grain size (d), TiB
2
particle size (d
p
) and
volume fraction of TiB
2
(xTiB
2
)) with given property
requirements. The required HD-EMIs (HD-EMI4 and HD-
EMI5) for mapping models (f4 and f5) are set as greater than
or equal to unity, which ensures satisfaction of the requirement
that all quantified uncertainty is within a target level. Among
the obtained feasible space of d, d
p
and xTiB
2
we select the
value of temperature of processing (temp) that has the largest
feasible space for the rest of the properties (d, d
p
,, and xTiB
2
)
because, as explained previously, it is desired to maintain the
feasible region as large as possible until the end of the design
process to achieve robustness under MSU.

The achieved feasible space of the volume fraction of TiB2 is
within the ranges [0.0322, 0.450] and [0.0656, 0.0995] [Figure
15]. This indicates that the achieved space of the volume
fraction of TiB
2
, [0.0322, 0.450]; [0.0656, 0.0995] and
thickness of shell, 10 (mm), with processing temperature of
Copyright 2009 by ASME 10
634 deg C guarantee that the submersible performance
satisfies the given requirements while maintaining all
quantifiable uncertainty within bounds (5% performance
variability), both for each of the mapping models (f1, f3, f4, f5,
f7, f8 and f9) and for its propagation through the model chain
into the final performance space.


Figure 13 Feasible Design Space for MODULE 5

Figure 14 Feasible Design Space for MODULE 4

Figure 15 Feasible Design Space for MODULEs 1 and 3
We see from Figure 13 that higher yield stress
y
values and
lower density values are favorable for the design and are
associated with higher HD-EMI values. It can also be
concluded from Figure 14 that lower grain size (d) and lower
TiB2 particle size (d
p
) yield higher HD-EMI values and hence
are favorable for the design. From Figure 17 it can also be
concluded that higher volume fraction of TiB
2
yields more
feasible design structures of the composite.

On increasing the HD-EMI values to [1.4, 1.8, 1.8] for (f7, f8,
f9) and [1.4, 1.2] for (f4, f5) and [1.8, 1.8] for (f1, f2); we see in
Figures 16 and 17 that the feasible spaces at each level of
design reduces and feasible space of volume fraction of TiB2
falls to within the range [8.84, 9.75].

As illustrated, considering MSU in the mapping models, a
designer may then select the best or preferred solution(s) for
materials and product using IDEM. Ranged sets of design
specifications are identified with given product/system
performance requirements considering quantifiable
uncertainty. Based on obtained feasible solution sets, designers
may have more freedom for choosing their decisions,
emphasizing product performance, achieving robustness
against MSU, or compromising between them.

Figure 16 Feasible Design Space for MODULE 4


Figure 17 Feasible Design Space for MODULEs 1 and 3

At this point, the reader may wonder what additional
information is afforded by this approach in designing or
selecting a material beyond the rather obvious effects of
reducing particle size and spacing through increase of particle
volume fraction. It might seem obvious that these steps would
be necessary to develop a high specific strength in situ Al
MMC in such an application. However, there are several
important points to make. First, the specific ranges of
microstructure attributes are directly coupled in the present
methodology with the overall systems design (material plus
submersible). Hence, changes in performance requirements
are directly reflected in the ranges of microstructure attributes
that emerge from application of IDEM. Second, this approach
can be readily extended to include performance requirements
that impose multifunctional, multiphysics requirements on the
material design aspect. For example, if high thermal
conductivity is also required as part of the present design (as
might be the case for heat transfer from the submersible
interior), it may very well drive a decrease in volume fraction
of non-metallic particles, which conflicts with strength
requirements. Such competing modes of requirements in
materials design are common and serve as a compelling basis
for the present systems-based robust design approach.
Moreover, if one is interested in selecting different process
routes (e.g., in-situ versus ex-situ Al MMCs or other matrix
materials), the assessment of feasibility is quite difficult
without considering the full contributions of the process-
structure-property-performance relations. In other words, it is
not just a classical materials selection problem [18].
Copyright 2009 by ASME 11
5. CLOSURE
In this paper we introduce the microstructure-mediated design
construct. We present a methodology to pursue concurrent
robust design of a robotic submersible and Al-based metal
matrix composite that embodies the microstructure-mediated
robust design construct. Recently developed tools have been
used such as the Inductive Design Exploration Method, which
facilitates top-down searching for design solutions including
process path and microstructure based on bottom-up
simulations.

The work presented in this paper constitutes one of the most
complete applications of IDEM. The primary challenge
involves managing uncertainty in over seven empirical and
theoretical models over four levels of design. Starting from a
hull thickness parameter in MODULE 5, the feasible design
spaces for various mechanical properties along with higher
yield stress and lower density are identified using IDEM.
Similarly, the feasible space of material properties and various
material and processing parameters are identified using IDEM,
leading to preference for lower grain size and higher volume
fraction of TiB
2
in the MMC; this is based on use of
MODULEs 4, 3 and 1. Upon analyzing the results we
conclude that the microstructure mediated design construct
holds promise in designing both the product and the material
from which the product is made.
ACKNOWLEDGMENTS
This work stems from collaborations within the Georgia Tech-
IIT Kharagpur (GT - IIT KGP) Alliance, and involves
collaboration of students from IIT KGP who interned at
Georgia Tech in the summer of 2008 and were jointly
supervised for the remainder of the year by faculty at the two
institutions.

MODULE Student IIT Faculty GT Faculty
1 A. Patra T.K. Kundu T.H.B Sanders
2 S. Lenka S. Ghosh H. Garmestani
3 S. Bagchi M. Chakraborty
R. Mitra
K. I. Jacob
4 M.K. Singh S. Ghosh
D.K. Pratihar
D. L. McDowell
5 A.K.
Srivastava
D.K. Pratihar
C.S. Kumar
J. H. Panchal
F. Mistree
6 A. Sinha C.S. Kumar J.K. Allen
J. H.Panchal
Coordinators M. Chakraborty F. Mistree

The support of Drs. Mitra, Pratihar, Sanders, Jacob is
gratefully acknowledged. The financial support from the IIT
Foundation for the six interns to spend eight weeks at Georgia
Tech was instrumental in fostering this collaborative research
effort and is much appreciated.
REFERENCES
[1] Anestiev L., Froyen L. and van Vugt L., 2000, "On the
Kinetics of Diffusion of Controlled Precipitation under
Microgravity", Journal of Applied Physics, vol. 88, p.
2130.
[2] Avrami M., 1939, "Kinetics of Phase Change. I: General
Theory ", Journal of Chemical Physics, vol. 7, p. 1103.
[3] Choi, H-J., 2005, " A Robust Design Method for Model
and Propagated Uncertainty," Ph.D. Dissertation, George
W. Woodruff School of Mechanical Engineering,
Georgia Institute of Technology, Atlanta, GA.
[4] Choi, H-J. and Allen J. K., "A Metamodeling Approach
for Uncertainty Analysis of Non-deterministic Systems:
an Example of Linear Cellular Alloy Heat Exchanger,"
ASME Journal of Mechanical Design, in press.
[5] Choi, H-J, Allen, J.K., Rosen, D.W., McDowell, D.L.,
and Mistree, F., 2005, An Inductive Design Exploration
Method for the Integrated Design of Multi-scale
Materials and Products, ASME Design Automation
Conference, (Hae Chang Gea, Ed.) Long Beach, CA.
ASME DETC2005-85335.
[6] Choi, H-J., McDowell D. L., Allen J. K. and Mistree F.,
2008, "An Inductive Design Exploration Method for
Hierarchical Systems Design under Uncertainty,
Engineering Optimization, vol. 40, no 4, pp. 287-307.
[7] Choi H-J., McDowell D. L., Rosen D., Allen J. K. and
Mistree F., 2008, An Inductive Design Exploration
Method for Robust Multiscale Materials Design, ASME
Journal of Mechanical Design, vol. 130, no. 3, pp.
031402-1/13.
[8] Das A., Ji S. and Fan Z., 2002, "Morphological
development of solidification structures under forced
fluid flow: a Monte-Carlo simulation", Acta Materialia,
vol. 50, pp. 4571-4585.
[9] Fan Z., 2002, "Semisolid Metal Processing: 30th
Anniversary Review", International Material Reviews,
vol.47 no. 2, pp. 49-85.
[10] Herbert M. A., Maiti R., Mitra R. and Chakraborty M.,
2006, " Microstructural evolution and Wear properties of
In-situ Al-4.5Cu-5TiB2 Composite Processed in Mushy
State", Solid State Phenomena, vol. 116-117, pp. 217-
220.
[11] Herbert M. A., Sarkar C., Mitra R. and Chakraborty M.,
2007, "Microstructural Evolution, Hardness and
Alligatoring in the Mushy State Rolled Cast Al-4.5Cu
Alloy and In-situ Al4.5Cu-5TiB2 Composite",
Metallurgical and Materials Transactions A, vol. 38 no.
9, pp. 2110-2126.
[12] Johnson W. A. and Mehl R. F., 1939, "Reaction Kinetics
in Processes of Nucleation and Growth", Transactions of
the American Institute of Mining, Metallurgical and
Petroleum Engineers, vol. 135, pp. 416-458.
[13] Joubert P.N., Some Aspects of Submarine Design, Part
1. Hydrodynamics, under contract to Maritime Platforms
Division Platforms Sciences Laboratory, October 2004.
[14] Kumar, R.P, Kumar, C.S., Dasgupta, A. and Sen, D.,
2009, Discrete Time-Delay Control of an Autonomous
Underwater Vehicle: Theory and Experimental Results
as an Invited Paper in Special Issue on Autonomous
Copyright 2009 by ASME 12
Underwater Vehicles, Ocean Engineering, Volume 36, pp
74-81.
[15] Kumar, R.P., 2007, "Dynamics, Control and Optimal
Motion Planning of Autonomous Underwater Vehicles"
Ph.D. Dissertation, IIT Kharagpur.
[16] Lloyd D. J., International Materials Review, 34 (1989) 1.
pp.1-29.
[17] Mandal A., Maiti R., Chakraborty M. and Murty B. S.,
2004, " Effect of TiB2 particles on aging response of Al-
Cu alloy", Material Science Engineering A, vol. 386, pp.
296-300.
[18] McDowell, D.L. and Olson, G.B., 2008, "Concurrent
Design of Hierarchical Materials and Structures",
Scientific Modeling and Simulation (CMNS), vol. 15 no.
1, pp. 207-240.
[19] Mirkovic D., Grbner J., Schmid-Fetzer R., Fabrichnaya
O. and Lukas H. L., 2004, "Experimental study and
thermodynamic re-assessment of the AlB system",
Journal of Alloys Compounds, vol. 384, p. 168.
[20] Montgomery D.C., 2004, Design and Analysis of
Experiments, 6th Edition, Wiley.
[21] Murty B. S., Kori S. A. and Chakraborty M., 2002, "
Grain refinement of aluminium and its alloys by
heterogeneous nucleation and alloying [J]", International
Material Reviews, vol. 47, pp. 2-29.
[22] Olson G. B., 2000, "Designing a New Material World",
Science, vol. 288. no. 5468, pp. 993-998.
[23] Panchal, J. H., Choi H. J., Allen J. K., McDowell D. L.
and Mistree F., 2007, A Systems Based Approach for
Integrated Design of Materials, Products, and Design
Process Chains, Journal of Computer-Aided Materials
Design, vol. 14, n supplement 1., pp. 265-293.
[24] Rohatgi P. K., Asthana R. and Das S., 1986,
"Solidification, Structures, and Properties of Cast Metal-
Ceramic Particle Composites", International Material
Reviews, vol. 31, pp. 115-139.
[25] Sharp A. G., Woods Hole Oceanographic Institute
Technical Memorandum 3-81 Failure Curves of
Cylindrical/Spherical Pressure Vessels and Flat End
Caps, August 1981,
[26] Sundman B. and gren J., 1981, "A Regular Solution
Model for Phases with Several Components and
Sublattices, Suitable for Computer Applications",
Journal of Physics and Chemistry of Solids, vol. 42, p.
297-301.
[27] Witusiewicz V. T., Bondar A. A., Hecht U., Rex S. and
Ya. Velikanova T., 2008, "The AlBNbTi system: I.
Re-assessment of the constituent binary systems BNb
and BTi on the basis of new experimental data ",
Journal of Alloys and Compounds, vol. 448, pp. 185-194.
[28] Witusiewicz V. T., Bondar A. A., Hecht U., Rex S. and
Ya. Velikanova T., 2008, "The AlBNbTi system: III.
Thermodynamic re-evaluation of the constituent binary
system AlTi ", Journal of Alloys and Compounds, vol.
465, pp. 64-77.
[29] Young W. C., 1989, "Roark's Formulas for Stress and
Strain", 6th Edition, McGraw-Hill.
[30] Zhange Z. and Chen D. L., 2006, "Consideration of
Orowan Strengthening Effect in Particulate-reinforced
Metal Matrix Nanocomposites: A Model for Predicting
their Yield Strength", Scripta Materialia, vol. 54, pp.
1321-1326.
[31] Yang, B., Y.Q. Wang and B.L. Zhou, 1998, The
mechanism of formation of TiB2 particulates prepared
by in-situ reaction in molten aluminum, Metall. Mater.
Trans., vol. 29B, p. 635.