Bioinspired Genetic-Algorithm Optimized Ground-Effect Wing Design:
Flight Performance Benefits and Aircraft Stability Effects
KARL ZAMMIT, HOWARD SMITH, NOEL SIERRA LOBO, IOANNIS K. GIANNOPOULOS
Centre of Excellence for Aeronautics, School of Aerospace, Transport and Manufacturing,
Cranfield University,
Cranfield, MK43 0AL,
UNITED KINGDOM
Abstract: - This paper presents a bioinspired, genetic-algorithm evolutionary process for Ground-Effect vehicle
wing design. The study made use of a rapid aerodynamic model generation and results evaluation
computational fluid dynamics vortex lattice method software, supervised by a genetic algorithm optimization
Python script. The design space for the aircraft wing parametric features drew inspiration from seabirds, under
the assumption of their wings being naturally evolved and partially optimized for proximity flight over water
surfaces. A case study was based on the A-90 Orlyonok Russian Ekranoplan, where alternative bioinspired
wing variations were proposed. The study objective was to investigate the possible increased flight aircraft
performance when using bioinspired wings, as well as verify the static and dynamic aircraft stability
compliance for Ground-Effect flight. The methodology presented herein along with the study results, provided
an incremental step towards advancing Ground-Effect aircraft conceptual designs using computational fluid
dynamics.
Key-Words: - Bioinspiration, Genetic Algorithm, Ground Effect, Ekranoplan, CFD, Vortex Lattice Method
Received: March 1, 2024. Revised: April 19, 2024. Accepted: April 25, 2024. Published: May 2, 2024.
1 Introduction
Wing-in-ground, WIG-craft, are aircraft vehicles
that fly near a surface, mostly above water surfaces.
The vehicles make use of the Ground Effect (GE)
being the increased lift curve slope and reduced
induced drag of the main lifting surfaces, [1].
GE effects are broadly understood as wing-span
and wing-chord effects, [1]. The wing-span
dominant GE is directly related to a reduction in the
induced drag, which is proportional to the wing’s
spanwise length. When a wing is close to the
ground, there is insufficient space for the full
development of wingtip vortices. Consequently, air
pressure leakage from under the wing to the upper
section is reduced. Additionally, the ground’s effect
pushes the vortices outwards, effectively artificially
increasing the wing’s aspect ratio beyond its
geometric value.
The wing-chord dominant GE involves an
increase in static pressure of the oncoming air
beneath the wing, which could be further enhanced
by utilizing wingtip side plates, [2]. The chord-
dominant GE enables the wing to generate more lift
per unit area, resulting in a higher lift coefficient for
the same power input, [1].
The distance between the wing and the ground
influences many of the effects experienced during
flight. Three distinct models have emerged from the
literature, each focused on a specific height zone
above the surface, [3], [4]. The first zone is the
operational region between the surface boundary
and a flight height corresponding to 20% of the
wing-chord length. In this In-Ground-Effect region
(IGE), the flow experiences significant constriction
in the vertical direction, leading to a predominantly
two-dimensional flow with restricted vertical
freedom. The second zone is referred to as the
region between one wing-chord length and ten
wing-span lengths above the ground. Within this
zone, the wing’s span dominates the model. Inviscid
flow models are commonly employed in this region
and demonstrate a marginal increase in the Lift to
Drag ratio (L/D), compared to the Out-of-Ground-
Effect (OGE) flight. A combination of the two
models is necessary to accurately capture the
aerodynamic behavior of wings operating in the
region between 20% - 100% of chord length, [4].
Above ten wing-span lengths, free-flight models
used in conventional aerodynamic theory for aircraft
design are applicable.
Distinct wing designs can be observed for WIG
craft throughout history. Russian Ekranoplans such
as the Korabl Market, the A-90 Orlyonok, and the
Lun-class craft, used a low aspect ratio straight wing
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
189
Volume 19, 2024
with minimal taper and twist. In contrast, the
German RFB X-114 and Chinese XTW had wings
with a significantly low aspect ratio, a very high
taper ratio, slightly sweptback leading and trailing
edges, and an appreciably large wing setting angle.
More recent WIG craft designs include the soon-to-
enter service Viceroy Seaglider by Regent, which
has a noticeably different wing planform with a high
aspect ratio and quasi-organic shape. Taking an
evolutionary perspective on the development and
design choices in the history of WIG-craft, there has
been a gradual transition towards wings resembling
sea birds. Based on this observation, the present
study aimed to develop WIG-craft wings using
nature-inspired seabird wing designs. By leveraging
the principles of bioinspiration, the study explored
the potential benefits of integrating biologically
derived wing planforms whose efficiency is
endorsed by the natural evolution process and the
possibility of enhancing the design using an
optimization algorithmic approach.
2 Bioinspiration
Research has acknowledged that the wing shapes of
seabirds may have evolved to optimize flight over
the sea surface, [5]. Seabirds predominantly employ
soaring rather than flapping flight, while various
soaring methods and corresponding wing planforms
could be identified from natural observations, [6],
[7].
Albatrosses and Shearwaters that belong to the
order Procellariforms, have long and narrow wings,
well-suited for soaring and gliding through the
windy, middle latitudes of the oceans. These birds
rely on horizontal movements of the atmosphere to
acquire the energy needed for flying, [8]. Their
wings are elongated and slender, often of a high
aspect ratio, enabling efficient long-distance flight.
Their wingtips can be slightly rounded or pointed,
and as such, drag is reduced and enhanced
maneuverability over water is achieved. Notably,
their oceanic flight involves frequent and brief pull-
up maneuvers, converting kinetic to potential
energy, [8].
Frigatebirds, renowned skilled aerial predators,
exhibit remarkable maneuverability and the ability
to stay aloft for extended periods, [9]. Their long
and narrow wings have a distinct forked or scissor-
shaped silhouette. Such a wing planform facilitates
dynamic soaring and allows them to exploit marine
thermals and ascending air currents to gain altitude,
[8]. Their high aspect ratio wings contribute to
efficient gliding and soaring, [9].
Pelicans can adapt their wing for fishing purposes,
making their structure unique. Characterized by
large wings with broad spans and relatively low
aspect ratios, they are designed to support frequent
takeoffs and landings on water surfaces, [10]. The
broad wings provide sufficient lift during low-level
flight, and their relatively short length facilitates
agile maneuvers, [8], [10]. Pelicans rely on vertical
movements, such as diving from the air into water
bodies to capture fish.
The research herein chose the A-90 Orlyonok
Russian Ekranoplan WIG-craft for an initial
benchmark case study. The A-90 main lifting
surfaces were subsequently redesigned with three
different bioinspired wings from the seabird families
mentioned above and evaluated the flying
performance which is presented in the following
section.
2.1 Bioinspired Wings
The first part of the study herein, aimed at re-
designing the wing of the A-90 Orlyonok WIG-
craft, effectively substituting it with bioinspired
wing designs stemming from the Albatross, the
Frigatebird, and the Pelican seabird families. For
that purpose, publicly available information such as
representative still images and animated videos of
the above-mentioned bird families flying above
water surfaces were studied. The investigation
resulted in the simplified conceptual parametric
design proposal shown in Figure 1, where a five-
stations, four-wing sections subdivision was
proposed, shown superimposed upon a typical
seabird wing.
Fig. 1: Sectioned parametric conceptual
representation of a seabird wing
The zeroth station shown in Figure 1,
represented the birds wing or WIG-craft fuselage
centerline, while the first section, between the
zeroth and the first station, represented the part of
the wing within the main body of the bird or the part
of the central wing box embedded within the WIG-
craft fuselage. Wing stations one to four were the
exposed parts of the wing to the free stream
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
190
aerodynamic flow. At each station, the parameters
of span, chord, twist, sweep, and anhedral/dihedral
angles were set, to closely resemble the seabird
soaring above the water surface, as these were
captured from the public domain representative
imagery material. The wings were designed to be of
equal plan area to the A-90 wing. Following the
above-stated assumptions, the resulting three
bioinspired wing shapes were designed, along with
the original A-90 Orlyonok WIG-craft wing. The
results are depicted in Figure 2, Figure 3, Figure 4
and Figure 5 and tabulated in Table 1.
Fig. 2: A-90 Orlyonok wing, [11]
Fig. 3: Albatross bioinspired wing
Fig. 4: Frigatebird bioinspired wing
Fig. 5: Pelican bioinspired wing
Table 1. Wing design parameters about Figure 1
wing sections
2.2 Bioinspired Wings Performance
Having identified three different bioinspired wing
planforms potentially suitable for WIG-craft wing
design and having established the A-90 as the datum
case, an assessment of the bioinspired wing
planforms was undertaken to analyze the wing
performance and benchmark it against the datum
case. The objective was to determine whether the
bioinspired wing planforms were viable candidates
to use as starting points for an algorithmic
evolutionary design process. The methodology
followed was similar to the one presented in [12] on
the application of a proposed evolutionary algorithm
for aerodynamic wing optimization. The
computational fluid dynamic software tool of choice
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
191
was Athenna Vortex Lattice (AVL), [13], a rapid
evaluation vortex lattice method computation tool,
capable of capturing the GE flying effect. The
suitability of AVL for usage in evaluating GE flight
has been studied and benchmarked against
experimental results, [14]. The four wing sections
computationally meshed are shown in Figure 6.
Fig. 6: AVL models of the A-90 and the bioinspired
wings
The wing airfoil for the bioinspired wings was
assumed to be the same along the wing span and the
same as the A-90 wing for all the wings under
study.
The objective function that served as metric for
evaluating aircraft performance was the Lift to Drag
ratio (L/D). The L/D ratio plays an important role in
every aspect of aircraft performance such as the
range to be traveled for a certain amount of fuel
stored, or the takeoff speed required to get airborne
that directly affects the size of the engines and the
runway length. To assess the bioinspired wing
designs performance, the resulting 3D designs' L/D
ratio was evaluated using AVL. The computation
took place at various heights from the ground,
signified by the height from the ground to the wing
mean chord non-dimensional parameter (hte/cm),
indicating the distance of the most outboard tip of
the trailing edge from the ground (hte), divided by
the wing mean chord length (cm).
The computation for the A-90 Orlyonok wing,
set the datum at a maximum L/D ratio of 22.2 at
hte/cm=0.4, which approximately corresponded to the
operating height of the A-90 at 4.2m, [4]. The L/D
ratio was also computed at another three different
heights off the ground to diversify the range for
comparison. Table 2 presents the corresponding
results derived from the AVL simulations, at
identical operating parameters and regimes with the
datum case being a cruise speed of 104ms-1, [4] and
operational height at hte/cm=0.4.
Table 2. Flight performance of the bioinspired
wings versus the original A-90 wing
The three bioinspired wings exhibited a superior
L/D ratio compared to the A-90 Orlyonok wing with
the albatross-inspired wing outperforming the rest.
It was interesting to note that while the Pelican-
inspired wing had a higher L/D ratio than the
Frigatebird derived one closer to the ground at
hte/cm=0.1, the latter does better at the higher
distance from the ground wing placements. The
Pelican-derived wing was the closest to the ground
at the wing tips and had an arching shape capable of
trapping the incoming air, generating greater lift at
lower altitudes, the effect of which lessens with
decreasing ground proximity. Conversely, the
Frigatebird-inspired wing is lowest at the wing root
and has a dihedral-set planform.
The L/D ratio benchmark of the A-90 Orlyonok
wing to the three bioinspired versions indicated that
the bioinspired wings could offer more efficient
alternatives for this type of aircraft. The bioinspired
wing design parameters shown in Table 1, assumed
the starting point of the genetic algorithm
optimization study that followed.
3 Genetic Algorithm Optimization for
WIG Wing Design Parameters
Building on the results of the WIG enhanced flight
performance in terms of the increased L/D ratio
exhibited by the bioinspired wing configurations,
further enhancement of the wing designs was
sought, through the application of algorithmic
optimization. To determine the most suitable
algorithm, factors such as the problem’s continuous,
discrete, or combinatorial nature were considered as
well as the dimensionality of the parameter space;
the availability of derivative information; the
presence of constraints; the characteristics of the
objective function such as smoothness,
multimodality, and noise. Following some
experimentation with different algorithms the
Genetic Algorithm (GA) was selected for the
current optimization problem. The algorithm drew
inspiration from the process of natural selection
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
192
operating on a population of potential solutions
represented as individuals encoded in a
chromosome-like structure, [15], [16], [17].
Implementing the GA in the current research work,
involved the representation of the wing parameters
as chromosomes, selecting the appropriate genetic
operators, and evaluating a fitness function based on
wing performance metrics. Multiple iterations of the
GA were conducted to properly explore and exploit
the solution space. Additionally, implementing
mechanisms such as elitism to preserve the best
individuals over generations can help maintain high-
quality solutions throughout the optimization
process. A Python script was developed to conduct
the optimization exercise on wing designs evaluated
using AVL.
3.1 Chromosome Representation and
Population Initialization
The initial step in the GA was to define the genetic
makeup of the chromosomes. The wing design
geometric variables for the sections shown in Figure
1, were encapsulated within a structured array as
genes within each chromosome. Genetic diversity
was initiated through the creation of an initial
population. This ensemble consisted of the genetic
makeup of the abovementioned wing designs; that
of the A-90 Orlyonok, Pelican, Frigatebird, and
Albatross-inspired wings.
3.2 Fitness Function Definition
The performance metrics selected for qualifying the
prevailing wing design parameter sets, were judged
according to the following fitness function, eq. (1):
eq.(1)
In the fitness function, a weighted sum of the
maximum L/D ratio was taken into account, along
with the lift coefficient at 70% of the wing span.
The second weighted term is related to the
requirement on the shape of the lift curve to meet
wing stall characteristics, necessitating at least 10%
reserve of lift on the ailerons, when the flow on the
root part of the wing is starting to separate, [18].
The weightings w1 and w2 were set to 50%. The
fitness function shown in eq. (1), was considered to
quantify the aerodynamic capability of each
chromosome, yielding a fitness score that reflected
its optimization potential.
3.3 Chromosomal Crossover
The initial step in the GA was to define the genetic
makeup of the chromosomes. Crossover operations
facilitate the exchange of genetic material between
selected chromosomes. Employing techniques like
single-point, multi-point, or uniform crossover, [19],
the GA could simulate genetic recombination and
introduce diversity and innovation within the
population. A single-point crossover mechanism
was first explored. Single-point crossover involves
selecting a random point along the chromosome and
swapping the genetic material between the parents
at that point, creating offspring by combining the
genetic material from both parents, as shown in
Figure 7. However, experience later showed that
this method did not introduce enough diversity. The
desired crossover mechanism would produce
offspring whereby the genetic makeup of both
parents would influence every gene. Hence, a
blended crossover function was used to combine the
genes of the two parents to produce averaged values
via a blending parameter that was set to 50%, as
shown in Figure 7.
Fig. 7: Single-point crossover and blended crossover
3.4 Genetic Mutation
Mutation can be introduced to diversify the genetic
makeup of the GA. Small random changes in
chromosomes encourage the exploration of
unexplored areas in the design space, enhancing the
algorithm’s capacity to identify optimal solutions. In
the current study, a simple mutation function was
initially used; a random perturbation between a
range of mutation rates was applied to the entire
chromosome. High mutation rates promote
exploration by allowing a larger chance for random
changes. Very high mutation rates can lead to
excessive randomization and slow convergence.
Low mutation rates focus more on exploitation by
preserving the existing solutions. Very low mutation
rates can lead to premature convergence and limit
the search space, [19]. While somewhat effective
when using a mutation rate in the range of 10%, a
need for some adjustments was identified to account
for premature convergence or the lack of it. In
opting for an adaptive mutation mechanism, a
mutation rate of 25% was dynamically adjusted
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
193
during the evolutionary process based on the
population’s performance or convergence status.
Adaptive mutation can help strike a balance
between exploration and exploitation, as it allowed
for more exploration in the early stages and
gradually reduced the mutation rate as the algorithm
progressed. When convergence was detected after
the first 50% of the generations, the mutation rate
was halved. This adjustment prevented seeking
convergence prematurely and allowed for finer
exploration around the converged region. If
convergence did not occur, the mutation rate was
doubled to encourage further exploration.
3.5 Selection Mechanism and Population
Renewal
A selection mechanism was employed to determine
the prevailing chromosomes to proceed to the next
generation. Various strategies, such as roulette-
wheel selection, tournament selection, or rank-based
selection, [20], were considered to favor individuals
with higher fitness function evaluation scores,
simulating the evolutionary principle of survival of
the fittest. Rank-based selection using the principle
of elitism was selected in the current study: the
fitness values of the parents and offspring were first
compared and the individuals with the highest
fitness values were carried over as parents for the
next generation.
3.6 Termination Criteria
The iterative process of fitness evaluation, selection,
crossover, and mutation went through a
predetermined number of generations. Each iteration
contributed to the refinement of solutions, fostering
incremental progress towards optimal designs. At
the culmination of the GA process, the optimized
geometric variables were extracted from the final
generation’s chromosomes. These variables
represented the refined wing configurations ready
for further analysis and evaluation.
3.7 Initial Implementation
Given an initial population of the A-90 Orlyonok
wing along the three bioinspired wing versions, and
using the abovementioned offspring generation and
selection mechanisms, every generation considered
a total of ten chromosomes; the four best parent
chromosomes carried on from the previous
generation and their six offspring. The GA initially
was set to iterate through 50 generations for non-
dimensional height from the ground hte/cm=0.1, 0.4,
0.8, 1.2. To ensure the robustness and reliability of
the results, the process was repeated four times for
each height, meaning that a total of 4,864 wing
designs were generated, analyzed, and compared.
The process indicated that the algorithm was being
heavily influenced by the outcomes of the first few
generations. The results differed substantially from
each other, often resembling either of the designs in
the initial population. Moreover, the examination of
the algorithm’s performance patterns unveiled a
trend toward convergence by the 30th generation.
Consequently, the subsequent 20 generations were
observed to contribute relatively less substantively
to the optimization process, thus prompting
reconsideration of the algorithm’s implementation.
3.8 Refinement of the Methodology
From the results described above, it was reasoned
that the GA was picking up an unwanted bias from
the first few generations, primarily attributed to the
randomized mutation process. The population
renewal mechanism was then altered to include the
four wing designs from the initial population in
every generation, along with the four chromosomes
carried over from the previous generation. As a
result, each generation comprised a total of 28
offspring. This adjustment sought to mitigate the
algorithm’s predisposition to early-stage bias.
Mindful of the aforementioned convergence
tendency by the 30th generation and the impending
project time constraints, a decision was made to
limit the GA iterations to 30. However, to maintain
the rigor of repeatability and credibility in the
results, the algorithm was still run four times for
each height-to-chord ratio. This measured approach
aimed to strike a balance between achieving
meaningful optimization outcomes and respecting
the project’s practical limitations. A total of 13,504
wing designs were generated, analyzed, and
compared as a result of the refined methodology,
with the results presented in the following section.
4 Wing Design Results and Discussion
The results of four different optimization cycles
numbered (i) to (iv), at the four different non-
dimensional height values from the ground are
shown in Figure 8 and Figure 9. From these images,
the resulting similarity in the designs was evident as
well as each optimization cycle had resulted in
slightly different wing design parameters, with some
features being more consistent and of a smaller
statistical deviation than others.
Notably, all wings exhibited an inboard section
configured with a dihedral angle, while the
remaining two outboard sections featured an
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
194
anhedral angle. An increased height-to-chord ratio
from hte/cm=0.1 to hte/cm=1.2, led to an increase in
the outboard section sweep angle and an overall
wing aspect ratio of 135.8% and 6.9% respectively.
All wing variations at hte/cm=0.1 had a positive twist
angle. Similarly, at hte/cm=0.4, all sections exhibited
a positive twist except for the outboard section,
where a negative twist resulted. At hte/cm=0.8 and
hte/cm=1.2, all sections adopted a negative twist
angle.
Fig. 8: Wing design results for four different
optimization cycles at hte/cm=0.1
Fig. 9: Wing design results for four different non-
dimensional heights hte/cm
Considering the increase in sweep and aspect
ratio at greater heights, suggested an adaptation to
higher flight conditions. Higher sweep reduces drag
at higher speeds, which became more relevant as the
wing moved away from the ground effect. The
observed trends could be attributed to a combination
of ground effect, aerodynamic interactions, and
design goals. Each height-to-chord ratio likely
represents a unique balance of these factors to
optimise the L/D ratio and overall performance. The
observed changes in the wing profile reflect the
algorithm’s effort to maintain efficient lift
generation while adapting to the reduced ground
effect.
Wings at hte/cm=0.1, 0.8, 1.2, had appreciable
amounts of anhedral and dihedral, creating a more
arched appearance. At hte/cm = 0.1, the wing was
closest to the ground experiencing a strong ground
effect, which led to air being trapped under the
wing, contributing to the observed arched shape.
The interaction between the wing and the ground
increased the risks of flow separation and stall due
to the potential for intricate flow patterns, vortex
shedding, and alterations in the effective angle of
attack, necessitating the arched wing contour to
alleviate these effects. As the height-to-chord ratio
increases, the risk of stall may decrease, allowing
for a flatter wing design.
The results for hte/cm=0.4, may be seen as
outliers to the rest. The wings appeared to be almost
flat with minimal anhedral or dihedral. The reduced
anhedral and dihedral might signify a compromise
between roll stability, discussed later, and
minimizing tip vortex effects while being mildly
influenced by the ground effect. The relatively
neutral appearance implied a design focus on
maintaining aerodynamic efficiency over a broad
range of operating conditions.
At hte/cm=0.8, and hte/cm=1.2, the resurgence of
the anhedral and dihedral akin to hte/cm=0.1 arose
from a combination of several aerodynamic factors.
With a reduced ground effect compared to prior
configurations, alterations in lift distribution and
vortex shedding patterns necessitated heightened
roll stability due to reduced ground influence,
reinstating the anhedral and dihedral.
The average parametric standard deviation at
each of the four height-to-chord ratios is given in
Table 3, and the average deviation of each of the 16
wing design parameters in Table 4. The standard
deviations of the span and chord dimensions in
Table 3 appear to be acceptable across the
investigated height ratios. In contrast to the span and
chord dimensions, the deviations in twist and sweep
were larger. The deviation percentages in Table 4
for twist and sweep indicated substantial
discrepancies, particularly for the twist angle. The
discrepancies suggested that the said parameters
might require further convergence for increased
reliability in the design. The anhedral/dihedral angle
standard deviations tabulated in Table 3 exhibited
variations that differed across the height ratios. It is
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
195
worth noting that these variations were relatively
consistent across the height ratios, which indicated a
consistent pattern in design evolution. The
percentages in Table 4 indicated notable deviations
in anhedral/dihedral angles that warrant closer
attention for convergence. While the deviation
values for span and chord seemed reasonably
consistent and within acceptable limits, the
significant percentage deviations for parameters like
twist, sweep, and anhedral/dihedral angles indicated
potential convergence issues. These parameters
appear to be more sensitive to the GA optimization
process and might benefit from an additional
number of generations to ensure more accurate
results.
Table 3. Wing design average parameter standard
deviation
Table 4. Average deviation (%) indicating the non-
converged wing design parameters
Comparing the best design of each of the four
height-to-chord ratios as deemed by the fitness
function given by eq. (1) with the datum case,
yielded the results shown in Table 5.
The resultant L/D ratios obtained, were 16.3%,
14.1%, 39.0%, and 41.9% greater than the datum
case for hte/cm=0.1, 0.4, 0.8, 1.2, respectively.
Hence, it followed that the wing loading criteria
were slightly worse off than that of the base case;
therefore, some adjustment may be needed to the
fitness equation to rebalance the priority of the
algorithm towards the wing loading criteria should it
be deemed necessary.
Table 5. Resulting bioinspired variation having the
best fitness value
5 Aircraft Flight Stability Effects
Having applied the genetic algorithm optimization
and having resulted in a number of bioinspired wing
designs potentially offering better L/D ratios to the
original case study wing, the follow-on conceptual
design step was to verify the aircraft compliance in
terms of static and dynamic stability. AVL software
is capable of generating linear static and dynamic
stability margins and results about any flight
equilibrium position, which for the present study the
trimmed cruise case was assumed.
The case study aircraft had to be re-designed
with versions of the bioinspired wings, with the
respective AVL models depicted if Figure 10. For
the sake of simplicity, the study aimed to inflict the
minimum possible level of intervention upon the
original aircraft design, in essence, to retain the
existing aircraft fuselage and empennage. In the
case of the project would be decided to be further
matured, follow-up on studies would have needed to
resize and relocate various features of the tailplane
and fin structures. The bioinspired wings had to be
located properly to replace the original Orlyonok A-
90 case study wing. The optimized wings resulted in
a certain anhedral angle for most designs, with the
wing tips pointing towards the ground. It was
suggested for the bioinspired wing placement to
take place on the upper part of the fuselage, contrary
to the case study aircraft which was a low wing
placement aircraft design. The supposed wing mass
that was altered due to higher aspect ratios, led to
the proper positioning of the heavier wings at a
longitudinal location that would retain the same
aircraft center of gravity.
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
196
Fig. 10: AVL models for aircraft static and dynamic
stability studies: (a) original A-90; (b), (c) & (d)
modified wing A-90 versions
A conceptual design and a relatively simplistic
approach to verifying the aircraft static stability is
summarized in eq.(2) and (3), where the aero-
derivatives of pitching (Cm) and yawing (Cn)
moments concerning the pitching (αwb) and yawing
(β) angles are shown. The first equation, eq.(2),
depicts the requirements for longitudinal static
stability about the symmetry plane of the aircraft,
while eq. (3) dictates the requirements for
directional stability. Although a six-degree-of-
freedom rigid aircraft of such designs can exhibit a
certain level of coupling between rolling and
yawing motion, only the directional stability aero-
derivative was checked, as an initial design step.
(2)
(3)
Evaluation of the aero-derivatives of eq. (2) and eq.
(3) via AVL, showed that the bioinspired versions
comply with the requirements of natural static
stability, meaning that the modified aircraft can
remain stable to perturbations about the trimmed
cruise flight state, without the usage of electronic
means of stability-enhancing. Dynamic stability was
checked by plotting the root locus of the dynamic
phugoid, short period, and Dutch roll modes. The
location of the roots was captured on the positive
complex plane for various heights from the ground
and depicted in Figure 11, Figure 12 and Figure 13
for the original and unmodified A-90 aircraft;
Figure 14, Figure 15 and 16 depict the root loci for
representative modified A-90 aircraft with
bioinspired wings. From the above-mentioned
figures, it was evidenced that the dynamic modes
were stable, having roots with negative real part and
similar damping levels for their decaying oscillatory
motion.
Fig. 11: Root locus for the phugoid mode of the
original A-90 version
Fig. 12: Root locus for the short-period mode of the
original A-90 version
Fig. 13: Root locus for the Dutch roll mode of the
original A-90 version
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
197
Fig. 14: Representative root locus plot of the
phugoid mode for the bioinspired wing A-90
Fig. 15: Representative root locus plot of the short
period mode for the bioinspired wing A-90
Fig. 16: Representative root locus plot of the Dutch
roll mode for the bioinspired wing A-90
6 Conclusions
The Vortex Lattice CFD method was successfully
applied in the present conceptual aircraft design
optimization study, through the AVL software
platform. Due to the rapid model generation and
results evaluation software attributes, as well as the
easiness in the coupling and internal communication
with Python, the study generated a genetic algorithm
optimization loop that utilized the VLM input-
output for results evaluation and subsequent
modeling rebuilding based on the results, in a time
efficient manner. The study resulted in conceptually
reconstructing and successfully substituting the
original wing of the A-90 Orlyonok WIG-craft, with
bioinspired wing versions of enhanced L/D ratio.
As a second step, the study made use of the
VLM method to draw preliminary conclusions and
justifications for the static and dynamic stability of
the resulting early conceptual design-level
bioinspired wing modified aircraft.
The methodology presented herein along with
the study results, provided an incremental step
towards advancing Ground-Effect aircraft
conceptual designs using VLM computational fluid
dynamics.
References:
[1] Yun L, Bliault A, Doo J, WIG Craft and
Ekranoplan: Ground Effect Craft Technology,
Springer US, 2010, DOI: 10.1007/978-1-
4419-0042-5
[2] Park K, Lee J, Influence of endplate on
aerodynamic characteristics of low-aspect-
ratio wing in ground effect, Journal of
Mechanical Science and Technology, 22(12),
2578–2589, 2008, DOI: 10.1007/s12206-008-
0805-y
[3] Rozhdestvensky K, Pappas P, Karaminas E,
Stubbs A, Pohl T, Hudson M, Stinton D,
Thomasson P, Ekranoplans: The GEMs of
Fast Water Transport. Discussion,
Transactions of the Institute of Marine
Engineers, 109(1), 47–74, 1997.
[4] Halloran M, O’meara S, Wing in Ground
Effect Craft Review. Melbourne Victoria,
1999.
[5] Shirsath R A, Mukherjee R, Experimental and
computational investigations of aerodynamic
characteristics of a finite rectangular wing-
inground effect, Proceedings of the Institution
of Mechanical Engineers, Part G Journal of
Aerospace Engineering, SAGE Publications
Ltd, 2022, DOI:
10.1177/09544100221114700
[6] Prandtl L, Induced Drag of Multiplanes,
NACA-TN-182, 1965.
[7] Boschetti P J, Quijada G M, Cárdenas E M,
Dynamic ground effect on the aerodynamic
coefficients of a wing using a panel method,
AIAA Atmospheric Flight Mechanics
Conference. American Institute of Aeronautics
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
198
and Astronautics Inc., Washington, D.C.,
USA, 2016, DOI: 10.2514/6.2016-3104.
[8] Suh Y B, Ostowari C O, Drag Reduction
Factor Due to Ground Effect, Journal of
Aircraft, 25(11): 1071–1072, 1988.
[9] Laitone E V, Comment on Drag reduction
factor due to ground effect, Journal of
Aircraft, 27(1): 96–96, 1990.
[10] Phillips W F, Hunsaker D F, Lifting-line
predictions for induced drag and lift in ground
effect, 31st AIAA Applied Aerodynamics
Conference. American Institute of Aeronautics
and Astronautics, San Diego California, USA,
2013, DOI: 10.2514/6.2013- 2917.
[11] Мараев P, Flying over the waves (Ekranoplan
‘Eaglet’), 1992.
[12] Cervenka M, Zelinka I, Application of
Evolutionary Algorithm on Aerodynamic
Wing Optimisation, 2nd European Computing
Conference, Malta, p.344-348, 2008.
[13] Drela M., Youngren H., AVL, [Online].
https://web.mit.edu/drela/Public/web/avl/
(Accessed Date: April 26, 2024).
[14] Zammit K, Smith H, Lobo N S, Giannopoulos
I K, Vortex Lattice CFD Application and
Modeling Validation for Ground Effect
Aircraft, WSEAS Transactions on Fluid
Mechanics, Vol. 19, 2024, pp.49-58,
https://doi.org/10.37394/232013.2024.19.5.
[15] Gehrke A, Guyon-Crozier G, Mulleners, K,
Genetic Algorithm Based Optimization of
Wing Rotation in Hover, Fluids, 3(3), 59,
2018, DOI: 10.3390/fluids3030059.
[16] Cayiroglu I, Kilic R, Wing Aerodynamic
Optimization by Using Genetic Algoritm and
Ansys, Acta Physica Polonica A, 132(3–II),
pp.981-985. 2017, DOI:
10.12693/APhysPolA.132.981.
[17] Holst T L, Pulliam T H, Transonic Wing
Shape Optimization Using a Genetic
Algorithm, in Sobieczky, H. (ed.) IUTAM
Symposium Transsonicum IV. Fluid
Mechanics and its Applications. Springer,
245–252, 2003, DOI: 10.1007/978-94-010-
0017-8_38.
[18] Lai K K, Mishra S K, Sharma R, Sharma M,
Ram B, A Modified q-BFGS Algorithm for
Unconstrained Optimization, Mathematics,
11(6), 1420, 2023, DOI:
10.3390/math11061420.
[19] Tutorialspoint Crossover, Genetic Algorithms
Tutorial, [Online].
https://www.tutorialspoint.com/genetic_algori
thms/genetic_algorithms_crossover.htm
(Accessed Date: April 26, 2024).
[20] Tutorialspoint Parent Selection, Genetic
Algorithms Tutorial, [Online].
https://www.tutorialspoint.com/genetic_algori
thms/genetic_algorithms_parent_selection.ht
m (Accessed Date: April 26, 2024).
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
- Karl Zammit: methodology, investigation, formal
analysis, software, validation
- Howard Smith: conceptualization, supervision,
project administration
- Noel Sierra Lobo: methodology, investigation,
formal analysis, software, validation
- Ioannis K. Giannopoulos: validation, visualization,
writing
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflicts of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on FLUID MECHANICS
DOI: 10.37394/232013.2024.19.19
Karl Zammit, Howard Smith,
Noel Sierra Lobo, Ioannis K. Giannopoulos
E-ISSN: 2224-347X
199