See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228828069

Reliability-Based Design and Optimization of Self-Twisting Composite Marine

Rotors

Article · July 2009

DOI: 10.1115/OMAE2009-80067

CITATION READS
1 163

3 authors, including:

Michael Motley Yin Lu Young

University of Washington Seattle University of Michigan
13 PUBLICATIONS   439 CITATIONS    164 PUBLICATIONS   4,281 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Debris-Fluid-Structure Interaction: Minimalist Numerical Approach and Validation View project

From Disney Movies and Video-Games to Rigorous Engineering Applications View project

All content following this page was uploaded by Yin Lu Young on 15 January 2015.

The user has requested enhancement of the downloaded file.

 Proceedings of the ASME 28th International Conference on Ocean, Offshore and Arctic
Engineering
OMAE2009
May 31-June 5, 2009, Honolulu, Hawaii

OMAE2009-80067

RELIABILITY-BASED DESIGN AND OPTIMIZATION OF SELF-TWISTING

COMPOSITE MARINE ROTORS

Michael R. Motley∗ Yin L. Young Jack W. Baker

Dept. of Civil and Env. Eng. Dept. of Civil and Env. Eng. Dept. of Civil and Env. Eng.
Princeton University Princeton University Stanford University
Princeton, New Jersey 08544 Princeton, New Jersey, 08544 Stanford, California, 94305
Email: [email protected] Email: [email protected] Email: [email protected]

ABSTRACT structures that interact with their environment [2]. State-of-the-

The objective of this work is to develop a reliability-based art methods in reliability-based design and optimization are pre-
design and optimization methodology to improve the efficiency sented in [2] with regard to aeroelastic structures, where the au-
of self-adapting composite marine rotors. The goal is to quan- thors presented an FSI model for the analysis of a 3-D wing struc-
tify the influence of material and operational uncertainties on the ture and the first-order reliability method is used to evaluate per-
performance of self-adapting marine rotors, and to present a de- formance sensitivities. Probabilistic design methods have been
sign and optimization scheme to maximize the performance and used to optimize the design of composites in [3; 4]. Response
reliability of these structures. surface techniques were used to analyze parametric sensitivities
for a thin-walled composite cylinder in [5]. These methods, how-
ever, all focused on aerospace structures. Similar probabilistic-
INTRODUCTION based design methodology is also needed for the optimization of
The focus of the current work is on passive, self-adapting adaptive marine structures.
marine structures that utilize fluid-structure interactions (FSI)
to improve structural performance and function. More specif-
ically, the current work presents a reliability-based design and Self-Adapting Composite Marine Propellers
optimization methodology to improve the energy efficiency of Marine propellers are traditionally made of nickel-
self-adapting composite marine rotors. A previously validated 3- aluminum-bronze (NAB) due to its excellent stiffness, yield
D fluid-structure interaction model is used to determine the per- strength, and anti-biofouling characteristics. They are designed
formance functions to generate response surfaces. The response to be rigid, and the blade geometry is optimized to yield the max-
surfaces were used to perform a Monte Carlo analysis in an effort imum efficiency at the design flow condition. However, when the
to evaluate the global sensitivity to material and load uncertain- ship speed or the shaft rotational frequency moves away from
ties, and to optimize the design parameters. the design values, the blade geometry becomes sub-optimal rel-
The objective of reliability-based design and optimization ative to the changed inflow, leading to a decrease in energy ef-
is to ensure a level of reliability with respect to structural and ficiency. This problem can be avoided or minimized by using
operational uncertainties through minimization of unacceptable blades made of carbon fiber reinforced plastics (CFRP). In addi-
performance. A recent literature review in this area can be found tion to the well-known higher specific stiffness and higher spe-
in [1]. Of the available literature, little work focuses on flexible cific strength of CFRP, the intrinsic deformation coupling behav-
ior of anisotropic composites can be utilized to improve the pro-
peller performance by passive tailoring of the load-induced de-
∗ Address all correspondence to this author.
1 c 2009 by ASME
Copyright °
formations according to the changing inflow. As demonstrated to be greater than a minimum target efficiency for all flow condi-
by recent experimental [6; 7] and numerical [8–16] studies, a tions, εη :
properly designed self-adapting composite marine propeller can
achieve higher energy efficiency and improved hydrodynamic
performance compared to its rigid counterpart when operating
at off-design conditions or behind a spatially varying wake. gob j (S, R) = η (S, R) − εη (2)

prob prob
Objectives subject to two probabilistic limit state functions g1 and g2 .
Significant advancements have been made recently related
to deterministic design and optimization of self-adapting com- prob
gj = p f j − p (g f j (S, R) < 0) ≥ 0; j = 1, 2 (3)
posite rotors. However, material properties, geometry, bound-
ary constraints, operational conditions, and environmental con-
ditions are all subject to natural or man-made random variations. where the constraint functions g f j are defined
Hence, the objectives of this work are to (1) quantify the influ-
ence of material and load uncertainties on the performance of
self-adapting composite marine propellers, and (2) optimize the PST (S, R)
g f 1 (S, R) = 1 − (4)
design to achieve the desired level of reliability in the structural Prigid (J)
performance.

PROBLEM DEFINITION ∆max ∆ (S, R)

g f 2 (S, R) = − (5)
Reliability-based design and optimization are common prac- D D
tice for many rigid and/or non-adaptive structural engineering
systems. The objective is to ensure the reliability requirements subject to an acceptable probability of failure, p f = [p f 1 p f 2 ]T .
with respect to uncertainties in structural parameters and oper- We denote η (S, R) as the efficiency of the self-twisting propeller
ating conditions by minimizing the probability of unacceptable and PST (S, R) and Prigid (R) as the power demand of the self-
performance. Performance and probability of failure calculations twisting and rigid propellers, respectively. Note here that the
are all made while explicitly considering uncertainties in struc- rigid propeller is a function of the loading condition represented
tural properties and loading conditions. by the advance coefficient J only because the objective is to opti-
To perform a reliability-based evaluation of the structure, mize the design variables for the self-twisting propeller such that
two performance measures are evaluated. First, the probability it yields equal or better performance compared to the already op-
of unsatisfactory performance is found by defining a limit state timized rigid propeller. Hence, the rigid propeller is only used as
function, g (S, R), where S is a vector of design variables, either a reference to evaluate the performance of the adaptive propeller.
deterministic or random, R is a vector of random variables rep-
In the application considered here, the vector of random
resenting uncertain structural properties and loading conditions.
variables is defined R = [J, E1 , E2 , G12 , ν12 , ν21 ]T , where J =
The function g (S, R) can either be implicit (e.g., the outcome of
V /nD is the advance coefficient defining the flow condition rep-
a numerical FSI code), or explicit (e.g., an approximate equation
resenting the operational parameter with propeller advance speed
obtained using the response surface methodology described be-
V , rotational frequency n, and diameter D; for the sake of sim-
low). The function g (S, R) is also chosen such that g (S, R) = 0
plicity, the blades are assumed to be made of a single layer of
defines a boundary between satisfactory and unsatisfactory per-
orthotropic lamina with material properties E1 , E2 , G12 , ν12 , and
formance (with g (s, r) < 0 indicating that the structure has un-
ν21 oriented at angle θ counterclockwise relative to the spanwise
acceptable performance). The performance state associated with
direction. Again, for simplicity, the only design variable con-
the boundary g (S, R) = 0 is denoted a “limit state.” Given this
sidered is the fiber orientation angle, S = θ. By running the
formulation, the optimization problem can be written as
propeller fluid-structure interaction analysis model [14; 15] for
extreme values of the material parameters (E1 , E2 , G12 , ν12 , ν21 ),
it was found that these parameters do not have a notable effect
£ ¡ ¢¤
min p gob j (S, R) ≤ 0 (1) on the system response [17]. For the purposes of this paper it is
S
assumed that variations from the design values in these parame-
ters have a negligible effect on propeller performance. As such,
where gob j (S, R) is the objective function, based on the effi- the random variable vector R can be assumed to only contain the
ciency (η) of the adaptive composite propeller, which is required advance coefficient, J.

2 c 2009 by ASME
Copyright °
 Equation (4) is used to ensure that the required power for
the self-twisting propeller is less than that of the rigid propeller
for acceptable performance, which will guarantee that the self-
twisting propeller provides higher energy efficiency on average.
Further, we define ∆(S,R)
D as blade tip deflection normalized by
the diameter, D, of the propeller, a parameter which is limited by
the maximum allowable normalized blade tip deflection, ∆max D .
The blade tip deflection needs to be restrained to limit the possi-
bility of blade strength and stiffness failures. Composite blades
made of CFRP can have many possible material failure modes,
as well as hydroelastic instability failure modes, most of which
can be correlated to the tip deflections. As such, the more eas-
ily measured tip deflection provides a standard of safety regard-
ing multiple possible structural stability and integrity character-
istics. Hence, Eqn. (5) is used to represent the safety limit, while deformed
Eqn. (4) is used to represent the serviceability limit. This limits undeformed
the optimal design range and the objective function (Eqn. (2)) is
used to find the fiber orientation angle that optimizes propeller Figure 1. DEFORMED AND UNDEFORMED GEOMETRY OF PRO-
performance as defined by the overall efficiency of the propeller. PELLER 5474 AT THE DESIGN FLOW CONDITION.

RESPONSE SURFACE METHODOLOGY

A fully-coupled boundary element method-finite element
PROBLEM SETUP method (BEM-FEM) model [14; 15] has been developed for the
The propeller herein is modeled using a single layer for sim- design and analysis of adaptive composite marine rotors. Al-
plicity, but the actual model will have many layers and will be though the coupled BEM-FEM analysis method is much faster
stacked in a sequence such that the load-deformation character- than coupling a Reynolds-averaged Navier-Stokes (RANS) with
istics will be the same as the effective single layer [18]. The an FEM method, it can still be computationally expensive to use,
material selected is Hexcel IM7-8552 carbon epoxy composite with time requirements ranging from 5 minutes to 2 hours for
[19]. The mean-load geometry is based on that of propeller 5474 a single simulation, depending on if the analysis is steady, un-
(Fig. 1), one of the composite propellers manufactured by AIR steady, with or without cavitation. For a Monte Carlo analysis
Fertigung-Technologie GmbH and designed and tested in coop- large enough to successfully achieve a reliable optimization, use
eration with the Naval Surface Warfare Center, Carderock Divi- of this model becomes impractical. Since the behavior of the per-
sion (NSWCCD). The model-scale propeller has a diameter of formance (power, deflection, and efficiency) are expected to be
D = 0.6096 m. The design rotational frequency is n = 780 rpm. smooth functions of J and θ, the response surface methodology
The design advance coefficient is J = V /nD = 0.66. More details is a more practical analysis alternative. Specifically ordered data
of propeller 5474 can be found in [7; 12]. For the possible range points obtained from the BEM-FEM model were used to pre-
of J values in forward operation, the self-twisting propeller is de- dict the behavior of the self-twisting composite propeller which
signed to be overpitched in its unloaded configuration. The self- are then used to generate response surfaces via two-dimensional
twisting propeller de-pitches due to twisting motion induced by regression analysis. The resulting coefficients of determination,
bending deformation caused by the fluid loading, which changes representing the goodness-of-fit of the surfaces, for the power de-
with J. The design requirement is that (1) at J = Jdesign = 0.66, mand, blade tip deflection, and efficiency, respectively, are 0.997,
the deformed geometry of the self-twisting propeller matches the 0.997, and 0.988, where
optimized rigid propeller geometry to achieve equivalent perfor-
mance between the two propellers, and (2) at J 6= Jdesign , the self-
twisting propeller should outperform its rigid counterpart. The Σ (gBEM−FEM − ḡBEM−FEM )2
result is a propeller that is, on average, more energy efficient than R2 = 1 − (6)
its rigid counterpart, requiring overall less power to perform and Σ (gBEM−FEM − g (S, R))2
less variation in power, which reduces strain and load fluctua-
tions, and hence extends the fatigue life of the engine. For details where gBEM−FEM is the data obtained from the BEM-FEM
about the design procedure or fluid-structure interaction analysis model, ḡBEM−FEM is the mean of all data obtained from the
methodology, readers should refer to [10–12; 14–16; 18; 20]. BEM-FEM model, and g (S, R) is the data from the regression

3 c 2009 by ASME
Copyright °
analysis. Values closer to 1.0 represent higher accuracy. The
three response surfaces are shown in Figs. 2,3, and 4. Note that x 10
4

J is dimensionless and θ is in degrees. 6

The power requirement is more sensitive to J than to θ. 5
Lower values of J correspond to higher angles of attack and J = 0.66
higher loads and thereby higher power demands. At higher loads, 4

P (W)
the change in pitch caused by the fluid-structure interaction is 3
also greater, and hence the power demand is more sensitive to θ
2
at lower J values. At high J values, the power demand is lower
and is less sensitive to θ due to small changes in pitch caused by 1
90
the hydrodynamic load induced bending-twisting deformation. 75 θ = 32o
The maximum deflection is a strong function of both J and 60
45
θ. This is because, as the fiber orientation angle becomes larger, 30 0.9
0.8
the blades are less stiff along their primary (longitudinal) axis 15 0.6 0.7
θ (deg) 0 0.4 0.5
(which, at θ = 45o becomes oriented more as the secondary axis). design J
As a result, the blade tip deflections have nonlinear growth with
fiber orientation angle. The increasing of the tip deflection with Figure 2. POWER DEMAND, P, RESPONSE SURFACE FOR THE
decreasing J is also expected due to increasing longitudinal load. SELF-TWISTING PROPELLER.
The efficiency is highest at the design values (J = Jdesign =
0.66, θ = θdesign = 32o ), which means that the design objectives
are satisfied. Note that the efficiency of the adaptive compos-
ite propeller has a strong dependence on J, which is inversely
proportional to the angle of attack, but a weaker dependence on
0.1
θ. It is of note, however, that there exists a quadratic element J = 0.66
to the behavior of the surface based on the fiber orientation an- 0.08
gle. This curvature switches directions at J = 0.66 and the local 0.06 θ = 32
o

maximum and minimum point is located at θ = θdesign , a char-

∆/D

acteristic which the response surface takes into account. This 0.04

change in curvature is because for θ > 32◦ and θ < 32◦ , the 0.02
change in tip pitch angle, ∆φ, will be less than ∆φ |θ=θdesign =32◦ .
0
For J < Jdesign = 0.66, if ∆φ < ∆φ |θdesign , the loaded pitch distri- 90
75
bution will be further away from the theoretical ideal value and 60 0.8
0.9
45
hence η < η |θdesign ; for J > Jdesign = 0.66, if ∆φ < ∆φ |θdesign , the 30 0.6
0.7
loaded pitch distribution will be closer to the theoretical value 15 0.5
θdesign (deg) 0 0.4
J
and hence η > η |θdesign .
The rigid propeller power requirement does not require re-
sponse surface methodology as it is only a function of J; how- Figure 3. NORMALIZED BLADE TIP DEFLECTION, ∆/D, RESPONSE
ever, fitting a curve to define the behavior of the rigid propeller SURFACE FOR THE SELF-TWISTING PROPELLER.
is also faster than using the BEM model to compute the behav-
ior at each J value (FEM analysis is not needed since the blades
are designed to be rigid). Using polynomial fitting techniques, a Carlo analysis. The response surface rather than the fully cou-
second-order curve was fit (R2 = 0.999) to the data for the rigid pled BEM-FEM propeller FSI solver is used to evaluate the struc-
propeller: ture performance.
The first step beyond the response surface methodology in-
volves determining how to define the random distribution of the
variables. It is typical for a manufacturer to provide a fiber orien-
Prigid (J) = Prigid (R) = −38572J 2 − 5091J + 50416 (7) tation tolerance around 2−3o in the construction of the laminates
for propeller or turbine blades, with a confidence level of 95%.
With this as a reference point, it is reasonable to assume that the
DESIGN SAMPLE fiber orientation angle has a Gaussian distribution with a mean
A design example is presented based on the reliability and value of θdesign . A tolerance of 3o with 95% confidence can be
response surface methodology above using a standard Monte approximated by a standard deviation of 1.5o (for a normal dis-

4 c 2009 by ASME
Copyright °
 however, there is little variation in the failure probability across
the entire range of θ. The limit states play a very important
role beyond the objective function. The constraint functions each
o
0.8 θ = 32 have definitive boundaries for acceptable performance. The ser-
viceability constraint is that the power requirement of the self-
0.7
twisting propeller is lower than that of the rigid propeller on av-
0.6
erage, which can only be satisfied if 31o ≤ θ ≤ 81o . Second,
η

the safety constraint limits the fiber orientation angle to θ < 34o .
0.5 Hence, what seemed initially to be a wide range of viable options
for the design variable based on the objective function is limited
0.4 J = 0.66
to a small range of 31o ≤ θ ≤ 34o . In this case, the probability
90
75 of failure of the objective function ranges between 5.1 − 5.4%,
60 0.9
45 0.8 which represents approximately 94% confidence that the self-
30 0.7
15 0.5
0.6 twisting propeller will exhibit safe and improved performance
θdesign (deg) 0 0.4
J over the rigid propeller for a realistic range of operating condi-
tions.
Figure 4. PROPELLER EFFICIENCY, η, RESPONSE SURFACE FOR
THE SELF-TWISTING PROPELLER. 1
Optimal Design Range
p(gobj < 0)
0.75

tribution, 95% of values are within 2 standard deviations of the 0.5

mean). 0.25

Further, it can be assumed that the propeller will operate 00 30

θdesign 60 90
near the design advance coefficient (Jdesign = 0.66) under most 1
operating conditions and that this would be an appropriate mean
p(gf1 < 0)

0.75 Serviceable Operating Range

value. A standard deviation of 0.10 and a normal distribution 0.5
will provide a realistic range of operating conditions. 0.25
A second step is to define an acceptable maximum tip de- 00 30 60 90
flection and minimum target efficiency. An inherent problem of θdesign
1
self-twisting propellers is that they can be subject to hydroelastic
Safe Operating Range
p(gf2 < 0)

0.75
instabilities and resonance issues. As described above, by limit-
ing the tip deflection these issues can be avoided or minimized. 0.5

Extending this value too high can lead to static divergence (dur- 0.25

ing deceleration or backing), increased stresses, and higher sus- 00 30 60 90

θdesign
ceptibility to resonance. The value ∆max /D = 0.05 is selected for
this design example, which prevents the stresses from reaching Figure 5. PROBABILITY OF FAILURE OF THE OBJECTIVE AND LIMIT
their peak value and sets a minimum allowable natural frequency STATE FUNCTIONS.
for the system within a safe range. The minimum target effi-
ciency is set at εη = 0.60. Finally, the values of p f must be set.
By setting the serviceability limit state to be that the self-twisting
propeller outperform its rigid counterpart at least 50% of the time
(i.e. p f 1 = 0.50), it is ensured that the self-twisting propeller CONCLUSIONS
yields better averaged performance over all possible flow condi- The objective of this research is to develop a reliability-
tions. Further, a balance must be reached between limiting the based design and optimization methodology to improve the en-
deflection and allowing the blades to bend and twist enough to ergy efficiency of self-adaptive composite marine rotors while
provide hydrodynamic efficiency improvements. In this exam- considering material and load uncertainties. Using Response
ple, p f 2 < 0.001 is used as the constraint to ensure that deflec- Surface and Monte Carlo analysis, it was shown that the optimal
tions do not grow to the extent to cause instabilities and excessive fiber orientation angle for the adaptive propeller is 31o ≤ θ ≤ 34o ,
stresses. which will yield a 94% probability of acceptable performance
The results of the objective function and limit states are based on three criteria: a serviceability limit state based on pro-
shown in Fig. 5. According to the top figure, the optimal fiber peller power requirement, a safety limit state based tip deflec-
orientation angle in terms of the objective function is about 59o ; tion, and an objective function that ensures a maximum energy

5 c 2009 by ASME
Copyright °
efficiency. The serviceability limit state and constraint functions REFERENCES
are designed such that, on average, the adaptive propeller outper- [1] Frangopol, D., and Maute, K., 2005. Engineering Design
forms its rigid counterpart. The safety limit state and constraint Reliability Handbook.
functions are designed to limit the tip deflection to a specified [2] Allen, M., and Maute, K., 2005. “Reliability-based shape
value to prevent excessive deflections, stresses, and to reduce the optimization of structures undergoing fluid-structure inter-
susceptibility to hydroelastic instability failures. Finally, the ob- action phenomena”. Computational Methods in Applied
jective function was used to determine the optimal fiber orienta- Mechanics and Engineering, 194, pp. 3472–3495.
tion angle that will maximize the energy efficiency of the self- [3] Chamis, C., 2004. “Design of smart composite structures
twisting propeller. in the presence of uncertainties”. Journal of the Chinese
The adaptive propeller was previously designed using deter- Institute of Engineers, 27, pp. 771–781.
ministic techniques to satisfy the following design condition: the [4] Lekou, D., and Philippidis, T., 2008. “Mechanical property
adaptive propeller must yield equal or better performance than variability in frp laminates and its effect on failure predic-
its counterpart over all flow conditions. The effects of material tion”. Composites Part B: Engineering, 39, pp. 1247–1256.
and load uncertainties were not considered. [5] Rais-Rohani, M., and Singh, M., 2004. “Comparison of
global and local response surface techniques in reliability-
In this work, we developed and applied a probabilistic based
based optimization of composite structures”. Structural and
method to (1) analyze the performance of the adaptive propeller
Multidisciplinary Optimization, 26, pp. 333–345.
subject to random variations in load and fiber orientation angle,
[6] Gowing, S., Coffin, P., and Dai, C., 1998. “Hydrofoil cav-
and (2) find the optimal fiber orientation angle that will maximize
itation improvements with elastically coupled composite
the energy efficiency while simultaneously minimize the engine
materials”. In Proceeding of 25th American Towing Tank
power demand and limit the maximum blade tip deformation.
Conference.
The optimal fiber orientation angle (31o ≤ θ ≤ 34o ) was found
[7] Chen, B., Neely, S., Michael, T., Gowing, S., Szwerc, R.,
to be similar to the deterministic design (θ = 32o ). However, at
Buchler, D., and Schult, R., 2006. “Design, fabrication and
θ = 32o , the probability that the overall energy efficiency of the
testing of pitch-adapting (flexible) composite propellers”.
adaptive propeller will either equal or exceed its rigid counterpart
In The SNAME Propellers/Shafting Symposium ’06.
is 94%, which suggests that there are small regions within the
[8] Lee, Y., and Lin, C., 2004. “Optimized design of com-
design space where the rigid propeller is a better choice than the
posite propeller”. Mechanics of Advanced Materials and
self-twisting propeller.
Structures, 11, pp. 17–30.
The results show that a probabilistic approach is more appro- [9] Lin, C., and Lee, Y., 2004. “Stacking sequence optimiza-
priate than a deterministic approach for the design and optimiza- tion of laminated composite structures using genetic algo-
tion of adaptive composite structures that rely on fluid-structure rithm with local improvement”. Composite Structures, 63,
interaction. This is because such structures are inherently more pp. 339–345.
sensitive to random variations in material properties, geomet- [10] Young, Y. L., Michael, T. J., Seaver, M., and Trickey, S. T.,
ric configurations, and loading conditions. Additional work is 2006. “Numerical and experimental investigations of com-
needed to assess the effect of material, geometry, and load uncer- posite marine propellers”. In 26th Symposium on Naval
tainties on initiation and evolution of failure modes. This is more Hydrodynamics.
complex due to the need to consider the many layers of laminates [11] Motley, M., Liu, Z., and Young, Y., 2008. “Utilizing fluid-
and the many possible modes of failure, as well as uncertainties structure interactions to improve energy efficiency of com-
in the failure modeling of CFRP. posite marine propellers in spatially varying wake”. Com-
It should be emphasized here that although the methodolo- posite Structures, (under review).
gies presented here focused on adaptive composite marine pro- [12] Young, Y., and Liu, Z., 2007. “Hydroelastic tailoring of
pellers, the framework is also generally applicable to other flexi- composite naval propulsors”. In 26th International Confer-
ble structures that undergo fluid-structure interactions, including ence on Offshore Mechanics and Arctic Engineering.
wind or tidal turbines as presented in [21]. [13] Lin, H., and Lin, J., 1997. “Effect of stacking sequence on
the hydroelastic behavior of composite propeller blades”.
In Eleventh International Conference on Composite Mate-
rials, Australian Composite Structures Society.
ACKNOWLEDGMENTS [14] Young, Y., 2008. “Fluid-structure interaction analysis of
The authors are grateful to the Office of Naval Research flexible composite marine propellers”. Journal of Fluids
(ONR) and Dr. Ki-Han Kim (program manager) for their fi- and Structures, 24(6), pp. 799–818.
nancial support through grant numbers N00014-07-1-0491 and [15] Young, Y., 2007. “Time-dependent hydroelastic analysis of
N00014-08-1-0475. cavitating propellers”. Journal of Fluids and Structures, 23,

6 c 2009 by ASME
Copyright °
 View publication stats

pp. 269–295.
[16] Liu, Z., and Young, Y., 2008. “Utilization of bending-
twisting coupling effects for performance enhancement of
composite marine propellers”. Journal of Fluids and Struc-
tures, (under review).
[17] Young, Y., Baker, J., and Motley, M., 2009. “Reliability-
based design and optimization of adaptive marine struc-
tures”. Journal of Applied Mechanics, (submitted).
[18] Young, Y. L., Liu, Z., and Motley, M., 2008. “Influence of
material anisotropy on the hydroelastic behaviors of com-
posite marine propellers”. In 27th Symposium on Naval
Hydrodynamics.
[19] Camanho, P., and Lambert, M., 2006. “A design method-
ology for mechanically fastened joints in laminated com-
posite materials”. Composites Science and Technology, 66,
pp. 3004–3020.
[20] Plucinski, M., Young, Y., and Liu, Z., 2007. “Optimiza-
tion of a self-twisting composite marine propeller using a
genetic algorithms”. In Proceedings of 16th International
Conference on Composite Materials.
[21] Young, Y., Motley, M., and Yeung, R., 2009. “Hydroelastic
response of wind or tidal turbines”. In 28th International
Conference on Ocean, Offshore and Arctic Engineering.

7 c 2009 by ASME
Copyright °