Вы находитесь на странице: 1из 18

American Institute of Aeronautics and Astronautics

Public Release: 88ABW-2012-1694




1
Modeling for Conceptual Design: An Aeroelastic Approach
Edward J . Alyanak
1

Wright Patterson Air Force Base AFRL, WPAFB, OH 45433
Conceptual design traditionally is accomplished with the lowest fidelity tools available to the
design engineer augmented with high fidelity aerodynamics. However decisions made
during this relatively short period of time have a very large impact on the life-cycle cost and
performance of an aircraft. Historically a configuration and associated outer mold line for a
design is selected without considering all the relevant technologies, multi-disciplinary
interactions or all the single disciplines that will have an impact on the vehicle. Most
methods are focused primarily on the aerodynamics with very low fidelity empirical
methods applied to the other disciplines. This paper focuses on a method of bringing multi-
physics based analysis forward, into the conceptual design process. The paper will discuss
improving the design process to keep more than a single configuration in consideration until
they have been evaluated with higher fidelity multi-disciplinary tools. Finally the paper will
introduce a new rapid model generation tool being developed within the Air Force Research
Lab. The rapid model generator generates aeroelastic models for MSC Nastran, ASTROS
and ZAERO. It supports drag prediction from AWAVE for supersonic wave drag and other
tools to estimate viscous drag at zero lift conditions. Additionally aerodynamic models for
rigid vortex lattice methods are generated to estimate drag polars for use in performance
calculations.
Nomenclature
AFRL = Air Force Research Lab
ESAV = Efficient Supersonic Air Vehicle
FLOPS = Flight Optimization System
CAD = Computer Aided Design
FEM = Finite Element Model
CFD = Computational Fluid Dynamics
OML = Outer Mold Line
VSP = Vehicle Sketch Pad
I. Introduction

Development of a new aircraft to meet a specific set of mission criterion is a very large endeavor. Historically
the process involves three phases of design; conceptual, preliminary and detailed. Each phase sees an increase in
fidelity level from closed form equations and curve fits through historical databases, to complete computer aided
design (CAD) assemblies, finite element models (FEM) and computational fluid dynamic (CFD) models. Also
included in the later phases is significant component and possible prototype testing.
Recently within the US Air Force (USAF) there has been an increased emphasis on efficiency and cost. Within
the Air Force Research Lab (ARFL) a careful look at how new air vehicles are developed has been undertaken. It
has been found that many very important design decisions are made very early in the design cycle that have a large
impact on future acquisition costs over the life of a vehicle. Many of these early decisions that have a large impact
on a vehicles topological characteristic are made in the conceptual design phase. Conceptual design has largely
been based on low order equations and historical trend data. This has been the case because tens of thousands of
possibilities need to be evaluated to determine the single best configuration. Carrying out this process with anything
but a very fast analysis tool becomes computationally very intensive.

1
Project Engineer, AFRL/RBAC, 2210 8
th
St. Bld. 146 Rm. 225 WPAFB, OH 45433, AIAA Young Professional.
53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference<BR>20th AI
23 - 26 April 2012, Honolulu, Hawaii
AIAA 2012-1425
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


2
Many papers, text books and computer codes have been developed for the conceptual design phase. Raymer has
published many versions of his book (Raymer, 2006) on conceptual design. NASA Langley Research Center
introduced the Flight Optimization System code FLOPS, (McCullers, 2002). Both of these references refer to the
use of statistical data fits and empirical equations to model the complete system considering things such as
aerodynamics, weight, propulsion, mission performance, takeoff/landing requirements, noise, cost, etc
These processes are effective and complete however for new and future vehicles mission requirements have
become very difficult to achieve. Thus vehicles are becoming more multi-disciplinary and nonlinear in their
behavior. Long range requirements demand very efficient engines, possibly imbedded within the airframe.
Additionally to save weight structures have become more flexible and more susceptible to aeroelastic responses such
as control surface reversal or flutter. New conceptual design processes are required that can handle multi-
disciplinary interactions. It is believed that these processes will include more physics and less empirical information
then in the past. This will certainly increase computational cost. However with the advancement of computing
technology these costs are manageable. What are needed are tools that can take advantage of computational power
existing today.
The idea of bringing fidelity (physics based design) forward in the design process is not new. This merger of
conceptual and preliminary design is being investigated by other government agencies and industry airframes. An
early study from AFRL posses the need for CFD analysis in conceptual design (Snyder, 1990). The paper discusses
the importance of obtaining accurate estimates for aerodynamic information for a given configuration. Though the
structure is not mentioned the paper does motivate the need for high-fidelity physics based information to make
early conceptual design decisions. At this point in time CFD is utilized in conceptual design to refine the outer mold
line of the final configuration that will progress to preliminary design. In fact many of the references below are
focused on supporting CFD analysis.
Recently Vehicle Sketch Pad (VSP) has been developed by NASA Langley (Fredericks, et al., 2010). VSP
defines an aircraft with parametric parameters that are intuitive to non CAD experts. From a given plan form
definition geometry information can be extracted that is sufficient to run body fitted CFD analysis. Additionally
information to generate linear aerodynamic models is also available. VSP prides itself on being quick to learn when
compared to standard CAD setups.
VSP also has a structural model available (Chaput, Akay, & Rizo-Patron, 2011). The goal of the structural
model is to provide higher-fidelity mass estimates then the empirical methods currently used. Structural elements
such as ribs are added individually by locating them on the VSP geometry representation. Meshing of the resulting
FEM structural layout is performed by VSP.
At the highest funding level DaVinci is being developed under the CREATE-AV program (Roth & J ohn,
CREATE-AV DaVinci: Computational Based Engineering for Conceptual Design, 2010). The referenced enforces
the authors statements that many decisions that influence cost over the life of the vehicle are made very early in the
design process, and thus higher fidelity physics based analysis is required. At its core DaVinci also believes in
having high fidelity parametric geometry that is suitable for body fitted CFD methods (Roth, Livingston, Dailey, &
Cline, 2011). The parametric geometry requirements include the support of finite element modeling of diverse
structural layouts. However, no method for creating the structural layout is presented.
In the major bodies of work pushing fidelity forward, VSP and DaVinci, a complete geometry representation is
at the core of the development. This representation is intended to support all fidelities of analysis the designer
may desire. This paper will present a different approach to developing models for conceptual design that include
structural models. The goal of the effort is to support many different levels and types of analysis but not everything.
Additionally this paper will discuss how the conceptual design process can be augmented to allow more than a
single configuration to be considered with higher fidelity tools.

II. The Design Process

Previously it was mentioned that conceptual design tools investigate tens of thousands of configurations. This
process, shown in Figure 1, typically results in one or two configurations after the conceptual design phase. In
many cases the outer mold line of the aircraft has been defined based on rigid computational fluid dynamics
simulations. The weight is estimated based on low order methods that are likely proprietary to an individual air
framer. The performance estimate is based on this weight estimation.
These few configurations are brought forward into preliminary design where multi physics interactions are
considered and higher fidelity tools are used to evaluate the design. A finite element model is generated. The
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


3
structural arrangement is generated by an engineer utilizing a CAD package to draw the arrangement. In some
cases the control surfaces are sized and placed by this same engineer at this stage. Maneuver and aeroelastic loads
are evaluated along with flying qualities. The weight model is improved by breaking the component weights down
to a more refined list. The FEM model may or may not impact the weight model. Finally the performance is re-
evaluated. In many cases the design does not close and the configuration is massaged to generate a closed design.
Then the process moves towards detailed design after a successful preliminary design review.
The focus of this paper is in the massaging of the given configuration. Every configuration will have potential
issues that must be solved. In the process shown in Figure 1, the designer has no choice but to solve the issues of
the configuration selected by the conceptual design process. The desire of this paper is to have many potential
configurations in play at this stage so the design team can select which configuration and associated challenges
make sense to undertake.


Figure 1: Traditional Design Practice

The proposed process shown in Figure 2 brings a multi-disciplinary optimization based group to work with the
conceptual design group. This avoids potential difficulties in merging existing conceptual and preliminary design
groups with their own standard practices together. The goal is to augment the ability of the conceptual design team
to reduce the design space to a single configuration by applying physics based analysis tools to many configurations
before determining the ideal configuration to move forward in the process.
Finally the information passed to the preliminary design team can be utilized to aid in the generation of their
models. For example instead of drawing a configuration based on experience a guide of what the structural topology
and what and where the control surfaces fit within the outer mold line will be available based on previous analysis to
augment the experience of the preliminary design engineer. The intent is to minimize the amount of unexpected
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


4
issues that must be solved as the preliminary design team develops refined models to capture the physics associated
with a configuration or the technologies applied.



Figure 2: Proposed Design Process Preserving Current Design Teams

III. Modeling for Design

To make the process in Figure 2 meaningful a method to produce low-medium fidelity physics based models of
10s to 100s of configurations quickly is required. The desire to support technologies such as active aeroelastic wing,
maneuver load control, gust load alleviation, and flutter suppression along with assessing the performance / stability
of a flexible vehicle in conceptual design requires some model of the structure.
Typically a technology such as active aeroelastic wing would be represented as a weight savings in a weight
assembly equation only when the conceptual design process in Figure 1 is applied. For the process in Figure 2 a
wing structure could be designed with and without active aeroelastic wing technology and an actual weight savings
could be associated with a given configuration.
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


5
Currently lead times for an FEM model are in the 2-5 week range. For this method to be successful an FEM
representation that can reproduce the structural response at a level suitable for conceptual design needs to be
available within days. Additionally aerodynamic models that can be utilized for aeroelastic evaluation along with
the FEM must be available in the same amount of time. Furthermore parametric changes to a given configuration,
such as changing wing aspect ratio while keeping area constant, must be available within minutes assuming the
initial configuration has been generated as shown for the simple wing model in Figure 3. The requirements for
modeling for design become rapid generation of parametric models that can capture physics relevant to the vehicle
and its associated technologies. For this paper the focus are aero elasticity based technologies.


Figure 3: Parametric Variations of a Single Configuration
A. AFRL Modeling for Design Capability

The Air Force Research Lab Multidisciplinary Sciences and Technology Center has generated a prototype
capability to perform modeling for design. The capability is being applied in-house currently to perform conceptual
design studies on Efficient Supersonic Air Vehicles (ESAV) and quite unmanned air vehicle concepts.
The tools discussed thus far have been focused on developing a detailed geometry first, then creating the
different models (aerodynamic, structures, etc) models from this geometry. The new capability does not utilize
sophisticated geometry modeling. The constraint for an aerodynamic model is to support linear panel methods with
thickness and chamber inputs such as that found in ZAERO (ZONA, 2009). Thus a body fitted CFD analysis is not
supported for complicated configurations. This simplifies the code and reduces computational expense. The
method is focused on generating the structure and aerodynamic analysis simultaneously along with spline
information required for aeroelastic analysis. Additionally control surfaces can be cut from the aerodynamic model
so that trim analysis can be performed. Finally design models with stress and aeroelastic constraints are developed
where the design variables are element thicknesses or in the case of composites ply thickness and orientation.
Higher fidelity aerodynamic analysis is available through transpiration based Euler codes such as ZEUS (ZEUS
User's Manual Version 3.1, 2009).
Additionally traditional tools to estimate drag are supported. These include AWAVE (Harris, 1963) for wave
drag estimation, a Virginia Tech FRICTION program written by W.H. Mason (Mason, 2006), and Tornado (Melin,
2000) which also has viscous drag estimation along with rigid vortex lattice based induced drag and lift calculations.

1. Intended Use

The proposed method is not intended to replace existing conceptual design tools or compete with high fidelity
parametric geometry based design methods. It is intended to interface between the two. Where high fidelity
parametric geometry is closer to existing preliminary design tools the proposed method is closer to conceptual
design tools. The goal of the method is to provide more multi-physics information in the conceptual phase when the
overall planform/configuration is selected. Furthermore a structural topology and control surface topology can be
explored with the proposed method for a given configuration and used as a basis for preliminary designers to start
the next phase of the design. The use of a FEM early in the conceptual design phase also has the potential to
improve early weight estimates.
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


6
2. Analysis Capabilities

Outputs from the AFRL modeling for design product are in the form of inputs to other analysis tools. These
include NASTRAN (Reymond & Miller, 1996), ASTROS (Neill & Herendeen, 1995), and ZAERO (ZONA, 2009),
AWAVE (Harris, 1963), Tornado (Melin, 2000), Quadpan (Love & Egle, 1999), Virginia Tech FRICTION (Mason,
2006). The following lists the codes and the types of analysis that can be performed with them that is currently
supported.

Status: March 2012
NASTRAN (Metallic and Composite Material)
o Frequency Analysis
o Trim Analysis
o Flutter Analysis
o Stress Constrained Design with Applied Load
o Trim Constrained Design (Multiple Load Cases Supported)
o Flutter Constrained Design
o Trim and Flutter Constrained Design (Multiple Trim Cases Supported)
ASTROS (Metallic and Composite Material)
o Frequency Analysis
o Trim Analysis
o Flutter Analysis
o Stress Constrained Design with Applied Load
o Trim Constrained Design (Multiple Load Cases Supported)
ZAERO (Linked with NASTRAN or ASTROS Frequency Analysis)
o Aerodynamic Model
o Spline from Aero Model to NASTRAN or ASTROS model
QUADPAN
o Aerodynamic Model for AIC generation
Based on Mach and Angle of Attack
AWAVE
o Supersonic Wave Drag Estimation
FRICTION
o Viscous and form drag estimation for zero lift
Tornado (Rigid)
o Viscous and form drag estimation for zero lift
o Induced drag polar calculation for given Mach
Calculated with Angle of Attack sweep

3. Input Format

The input to the AFRL method is file based. An application programming interface (API) may be created to set
input parameters without going through a file for use in some multi-disciplinary optimization (MDO) environments,
but is not currently available. Because the intended operation of the method is inside MDO settings no graphical
user interface (GUI) is planned for this development.
The input is based on building components and grouping them into assemblies. Currently the main assemblies
are wing like or body/fuselage like. Assemblies can be linked together into a collection that represents a complete
aircraft.

The major components are:
> &CrossSection
> &WingSection
> &PointMass
> &Material

D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


7
The major assemblies are:
> &Wing
> &Fuselage

Additional inputs consist of solution inputs such as aero dynamic settings etc
> &Global
> &GlobalTrim
> &Trim
> &Flutter
> &MachAlphaSweep

The &WingSection component contains a significant amount of the information related to the model. This
information includes the structural topology definition. This can be as simple as a number of evenly spaced ribs and
spars or as complex as a biologically inspired map L system (Kobayashi, LeBon, Pedro, Kolonay, & Reich, 2010).
Additionally the material definition and thickness for each member is set. For design models this thickness becomes
the initial design for the sizing optimization process within NASTRAN or ASTROS.
An example of a basic input file that uses default material properties (Aluminum) is given. The inputs
regarding aeroelastic analysis are not shown to save space. The major components and assemblies are shown. The
example is a half span wing with two control surfaces.

&Wi ng
I D = 1,
Mi r r or = ' no' ,
Aer oRef i ne = 2,
Wi ngSect i ons = [ 1] ,
St r Layer s = 1,
/

&Cr ossSect i on
I D = 1,
Loc = [ 0, 0, 0] ,
Chor d = 1. 5,
Shape = ' 0012' ,
Boundar yAct i ve = 1,
Chor dSpaci ng = 8,
/

&Cr ossSect i on
I D = 2,
Loc = [ 0. 5359, 2, 0] ,
Chor d = 1,
Shape = ' 0012' ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 8,
/

&Wi ngSect i on
I D = 1,
SpanSpaci ng = 6,
Lmap = 0,
Cr ossSect i ons = [ 1 2] ,
Cont r ol Sur f Boxes = [ 6 14; 27 35] ,
Hi ngeLocat i on = {' l e' , ' l e' },
Hi ngeType = {' SYM' , ' SYM' },
Ski nT = 0. 001,
Ri bT = 0. 005,
Spar T = 0. 01,
AddDensi t y = 500,
/

The result from this input file is shown in Figure 4.

D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


8

Figure 4: Basic Wing Geometry with Two Control Surfaces

With very small changes to the input file the topology can be changed. For example holes can be cut in a wing
section, Figure 5, by a small modification to the &WingSection input.

&Wi ngSect i on
I D = 1,
SpanSpaci ng = 6,
Lmap = 0,
Cr ossSect i ons = [ 1 2] ,
Cont r ol Sur f Boxes = [ 6 14; 27 35] ,
Hi ngeLocat i on = {' l e' , ' l e' },
Hi ngeType = {' SYM' , ' SYM' },
Ski nT = 0. 001,
Ri bT = 0. 005,
Spar T = 0. 01,
AddDensi t y = 500,
Hol e = [ 9 18] ,
KeepSki n = [ 1 1] ,
/

Here KeepSkin indicates that the upper and lower skins are not removed by the hole cutting process.


Figure 5: Basic Wing with Hole Cut

To include more advanced topology the user can input any mesh made up of spar/rib like components. A
triangle mesh routine will generate a corresponding skin mesh for the structural model. The map L formulation,
(Kobayashi, LeBon, Pedro, Kolonay, & Reich, 2010), has been integrated into the input file structure. In Figure 6 a
Control
Surface

D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


9
random vector was created as input to the map L routine. The only change to the input file was in &WingSection
where Lmap =0 was changed to Lmap =1. Spline information to couple the aerodynamic and structural model is
automatically created by the program. Spline points on the structure are only used if they are hard points, meaning
they are along a rib or spar like supporting structures. The skin mesh generated is shown in Figure 6 since it is no
longer simple quad elements.



Figure 6: Basic Wing with map L Topology

A Wing body configuration can also be generated that contains a beam based structural model of the fuselage. A
half span arrangement can be seen in Figure 7: Wing Body with Beam FuselageFigure 7. The main additional
input information is contained in the &Fuselage input.


Figure 7: Wing Body with Beam Fuselage

&Fusel age
T = Beams ar e hol l ow el l i pt i cal cr oss sect i ons, t hi s i s t he beamski n t hi ckness i nput ,
Cr ossSect i onConnect = Where wing connects to the fuselage,
AddedMass = Added Mass along fuselage,
Xl oc = X Location where elliptical profiles are defined,
Zl oc = Z Locat i on wher e el l i pt i cal pr of i l es ar e def i ned ( Camber ed Fusel age) ,
A = Fuselage Width at Xloc,
B = Fuselage Height at Xloc,
ACON = Wave drag area rule width constraints,
BCON = Wave drag area rule height constraints,
Mach = Wave drag area rule mach number,
Opt i mi ze = Flag to turn on or off area rule based optimization of the fuselage profile
XDES = X location where area ruled fuselage is applied (May be different from Xloc),
/

D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


10
IV. Complex Example Problem with Analysis Results

A light weight fighter example has been generated for this paper. This example problem was generated from a
blank sheet by the author for this paper. Admittedly other examples similar to this had been generated previously.
From blank sheet to figure generation of the results was around 1 hour. The input file used to create the models for
every analysis presented below is contained in the appendix. This very short time to go from idea to data represents
a real advancement in modeling for conceptual design technology. The aero and structural models for this example
problem are shown in Figure 8.


Figure 8: Light Weight Fighter Example

The first analysis demonstration is from a Nastran SOL 103, or modal analysis. The first six modes are zero as
the model is unsupported. The first flexible mode is wing bending and shown in Figure 9. The mass configuration
for the vehicle is empty around 11,000 lbs. Subsystem added mass is 6450lbs distributed along the fuselage in most
cases with 525lbs internal to the wings to model control surface actuators. The model has not been sized for any
load cases for this analysis. This analysis demonstrates that a structural FEM model has been generated.
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


11

Figure 9: First Flexible Mode

The same model is used to perform a lift only trim calculation using Nastran SOL 144. This analysis
demonstrates that an aerodynamic model has been generated and connected via splines to the structural model. The
analysis is run at Mach 0.9 with a dynamic pressure of 6.5 psi. The only free degree of freedom is angle of attach.
The acceleration in the negative z direction is 9.0g. The resulting deflections are shown in Figure 10. As expected
the wing tips bend up. The resulting angle of attach is 11.77 degrees.


Figure 10: 9g Trim for Lift

Finally a sizing optimization is performed on the same model with the same trim case providing the load. The
model is constructed of aluminum and von mises stress is constrained to be less than 70,000psi. The model includes
104 variables, 9 beams in the fuselage, 34 ribs, 39 spar and 22 skin variables. The iteration history in Figure 11
contains the mass trends for this design analysis. The stress constraints are feasible for each iteration of the
optimization process. The constraints represent the stresses in every element at various locations within the element
depending on its type. The final configuration weighs 8380lbs representing a decrease of 2620lbs in structural
weight. Obviously a true design will take into accounts 100s of load cases that are not accounted for in this
example. The methodology developed can handle N-Trim cases and a Flutter constraint as well. This example only
contains one of the purposes of demonstration.

D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


12


Figure 11: 9g Trim Loaded Sizing Optimization history

Finally aerodynamic analysis is performed that include Cd0 prediction and induced drag prediction. These are
calculated using Tornado (Melin, 2000) for this example. Information for both of these calculations is contained in
a table that is in a format suitable for FLOPS (McCullers, 2002) usage. The Cd0 calculations are very similar to
those produced by FRICTION (Mason, 2006) as well. Combining these analysis produces a drag polar for every
Mach and Altitude desired. A drag polar for 0 ft of altitude and the sub sonic Mach numbers of 0.01, 0.1, 0.5 and 0.8
are shown in Figure 12. These polars are calculated based on a clean configuration.


Figure 12: Drag Polar for 0 ft



lbs
iteration
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


13
V. Conclusions and Future Work

An advanced conceptual design process has been proposed that is minimally invasive to the current conceptual,
preliminary, and detailed design process. The purpose of the process is to increase the fidelity of analysis that is
performed on many configurations before the final configuration is selected. The intent is to have knowledge of the
critical issues associated with a configuration early in the process and avoid costly suppresses downstream.
To demonstrate the feasibility of the process a method to quickly generate many different aircraft configurations
and quickly analyze and size them has been developed at AFRL. The method is currently being integrated into an
advanced conceptual design process for ESAV and quite unmanned air vehicles. The process considers aeroelastic
loads and efficient structural topologies for new air vehicle concepts. Additionally tools to predict both flexible and
rigid aerodynamic performance are in place to be used in mission performance assessment.
A demonstration problem of a relatively complex example has been shown. Analysis results for frequency, trim,
structural design for the applied trim load, and aero drag polar generation is presented. The time to generate this
data was less than 1 hour from a blank sheet. This represents a breakthrough in modeling for design capability.
The capability is advancing rapidly within AFRL and being applied to internal design studies for two classes of
vehicle. Future work focused on the efficient supersonic air vehicle is being presented at the 2012 AIAA MDAO
conference at Indianapolis.

References

Chaput, A., Akay, E., & Rizo-Patron, S. (2011). Vehicle Sketch Pad (VSP) Structural Layout Tool. 49th AIAA
Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition. 4 - 7 J anuary 2011,
Orlando, Florida.

Fredericks, W., Antcliff, K., Costa, G., Deshpande, N., Moore, M., San Miguel, E., et al. (2010). Aircraft
Conceptual Design Using Vehicle Sketch Pad. 48th AIAA Aerospace Sciences Meeting Including the New Horizons
Forum and Aerospace Exposition. Orlando: 4 - 7 J anuary 2010, Orlando, Florida.

Harris, J . R. (1963). An Analysis and Correlation of Aircraft Wave Drag. Hampton, Va.: Langley Research Center
NASA.

Kobayashi, M. H., LeBon, A., Pedro, H. T., Kolonay, R. M., & Reich, G. W. (2010). On a Cellular Division Model
for Multi-Disciplinary Optimization. 51st AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics , and
Materials Conference. Orlando, FL: AIAA 2010-2989.

Love, M. H., & Egle, D. D. (1999). Aerodynamic Analysis for the Design Environment (AANDE) Vol. 2: Users'
Manual. Lockheed Martin, Tactical Aircraft Systems. WPAFB, OH: Air Vehicles Directorate AFRL AFMC.

Mason, W. H. (2006). Software for Aerodynamics and Aircraft Design (W.H. Mason, Virginia Tech). Retrieved
2012, from Skin Friction/Form Factor Drag estimation: http://www.dept.aoe.vt.edu/~mason/Mason_f /MRsoft.html#
SkinFriction

McCullers, A. (2002). FLOPS Flight Optimization System; User's Guide. Hampton, VA.

Melin, T. (2000). A Vortex L:attice MATLAB Implementation for Linear Aerodynamic Wing Applications. Royal
Institute of Technology (KTH).

Neill, D., & Herendeen, D. (1995). ASTROS Enhancements: Volume I - Astros User's Manual. Torrance, CA:
Wright Laboratory WL-TR-96-3004.

Raymer, D. (2006). Aircraft Design: A Conceptual Approach, Fourth Edition. AIAA.
Reymond, M., & Miller, M. (1996). MSC/NASTRAN Quick Reference Guide Version 68. Los Angeles, CA: The
MacNeal-Schwendler Corporation.
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


14

Roth, G., & J ohn, L. (2010). CREATE-AV DaVinci: Computational Based Engineering for Conceptual Design. 48th
AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition. AIAA 2010-
1232.

Roth, G., Livingston, J ., Dailey, C., & Cline, A. (2011). Addressing Geometry Needs of Systems Engineering with
DaVinci Software. 49th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace
Exposition. AIAA 2011-1107.

Snyder, J . R. (1990). CFD Needs in Conceptual Design. AIAA/AHS/ASEE Aircraft Design, Systems and Operation s
Conference. Dayton, OH: AIAA-90-3209.

ZEUS User's Manual Version 3.1. (2009). Scottsdale, AZ: Zona Technology, Inc.

ZONA. (2009). ZAERO Users's Manual: Engineers' Toolkit for Aeroelastic Solutions. Scottsdale, AZ: ZONA
Technology, Inc.

Appendix
&Gl obal
Out put Di r = ' sdm_v00' ,
Debug = 0,
wt mass = . 00259,
auni t s = . 00259,
pl ot = 1,
Sor cer Di r = 0,
Symmet r y = 0,
WaveDr ag = 1,
CheckMachAngl e = 1,
Ast r osVer si on = 11,
Uni t Fact or = 0. 0254,
Boundar y = [ ] ,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ MAI N WI NG
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

$ WI NG
&Wi ng
I D = 1,
Mi r r or = ' yes' ,
Mi r r or Vec = [ 1 1 0 0] ,
Aer oRef i ne = 2,
Wi ngSect i ons = [ 1 2] ,
St r Layer s = 1,
Mat I D = 10,
Ref Ar ea = 1,
/

$ HTAI L
&Wi ng
I D = 2,
Mi r r or = ' yes' ,
Mi r r or Vec = [ 1 1 0 0] ,
Aer oRef i ne = 2,
Wi ngSect i ons = [ 11 12] ,
St r Layer s = 1,
Mat I D = 10,
/

$ VTAI LS
&Wi ng
I D = 3,
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


15
Mi r r or = ' no' ,
Mi r r or Vec = [ 1 1 0 0] ,
Aer oRef i ne = 2,
Wi ngSect i ons = [ 21 22] ,
St r Layer s = 1,
Mat I D = 10,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ CROSS SECTI ON DEFI NI TI ONS
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

&Cr ossSect i on
I D = 1,
Loc = [ 300. 0, 0. 0, 0. 0] ,
Chor d = 160. 0,
Shape = [ 0 1; . 03 . 03; - . 03 - . 03] ,
Boundar yAct i ve = 1,
Chor dSpaci ng = 8,
Cut Ri bSect i ons = [ 1 2 6 7] ,
Type = 0,
Fusel ageI nt er f ace = 0,
Cr eat eConnect i onGr i d = 1,
/

&Cr ossSect i on
I D = 2,
Loc = [ 300. 0, 45. 0, 0. 0] ,
Chor d = 160. 0,
Shape = [ 0. 0000 0. 0050 0. 0075 0. 0125 0. 0250 0. 0500 0. 0750 0. 1000 0. 1500 0. 2000 0. 2500 0. 3000
0. 3500 0. 4000 0. 4500 0. 5000 0. 5500 0. 6000 0. 6500 0. 7000 0. 7500 0. 8000 0. 8500 0. 9000 0. 9500
1. 0000; 0. 00000 0. 00405 0. 00492 0. 00635 0. 00904 0. 01303 0. 01616 0. 01880 0. 02306 0. 02636 0. 02890
0. 03078 0. 03207 0. 03277 0. 03279 0. 03218 0. 03102 0. 02935 0. 02722 0. 02463 0. 02158 0. 01802 0. 01375
0. 00931 0. 00467 0. 00000 ; 0. 00000 - 0. 00272 - 0. 00318 - 0. 00379 - 0. 00471 - 0. 00573 - 0. 00636 - 0. 00682
- 0. 00741 - 0. 00773 - 0. 00785 - 0. 00780 - 0. 00759 - 0. 00720 - 0. 00651 - 0. 00558 - 0. 00449 - 0. 00331 -
0. 00212 - 0. 00096 - 0. 00005 - 0. 00005 - 0. 00005 - 0. 00005 - 0. 00005 0. 00000] ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 8,
Type = 1,
Fusel ageI nt er f ace = 1,
Cut Ri bSect i ons = [ 1 7] ,
/

&Cr ossSect i on
I D = 3,
Loc = [ 415. 0, 185. 0, 0. 0] ,
Chor d = 45. 0,
Shape = [ 0. 0000 0. 0050 0. 0075 0. 0125 0. 0250 0. 0500 0. 0750 0. 1000 0. 1500 0. 2000 0. 2500 0. 3000
0. 3500 0. 4000 0. 4500 0. 5000 0. 5500 0. 6000 0. 6500 0. 7000 0. 7500 0. 8000 0. 8500 0. 9000 0. 9500
1. 0000; 0. 00000 0. 00405 0. 00492 0. 00635 0. 00904 0. 01303 0. 01616 0. 01880 0. 02306 0. 02636 0. 02890
0. 03078 0. 03207 0. 03277 0. 03279 0. 03218 0. 03102 0. 02935 0. 02722 0. 02463 0. 02158 0. 01802 0. 01375
0. 00931 0. 00467 0. 00000 ; 0. 00000 - 0. 00272 - 0. 00318 - 0. 00379 - 0. 00471 - 0. 00573 - 0. 00636 - 0. 00682
- 0. 00741 - 0. 00773 - 0. 00785 - 0. 00780 - 0. 00759 - 0. 00720 - 0. 00651 - 0. 00558 - 0. 00449 - 0. 00331 -
0. 00212 - 0. 00096 - 0. 00005 - 0. 00005 - 0. 00005 - 0. 00005 - 0. 00005 0. 00000] ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 8,
Type = 1,
Cl osedWi ngTi p = 1,
Cut Ri bSect i ons = [ 1 7] ,
/

&Cr ossSect i on
I D = 11,
Loc = [ 500. 0, 0. 0, 0. 0] ,
Chor d = 80. 0,
Shape = [ 0 1; . 025 . 025; - . 025 - . 025] ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Cut Ri bSect i ons = [ 1 3] ,
Type = 0,
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


16
Fusel ageI nt er f ace = 0,
Cr eat eConnect i onGr i d = 1,
/

&Cr ossSect i on
I D = 12,
Loc = [ 500. 0, 45. 0, 0. 0] ,
Chor d = 80. 0,
Shape = ' 0005' ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Type = 2,
Fusel ageI nt er f ace = 1,
/

&Cr ossSect i on
I D = 13,
Loc = [ 550, 108, - 5. 0] ,
Chor d = 30,
Shape = ' 0005' ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Type = 2,
Cl osedWi ngTi p = 1,
/

&Cr ossSect i on
I D = 20,
Loc = [ 450. 0, 0. 0, 0. 0] ,
Chor d = 90. 0,
Shape = [ 0 1; . 025 . 025; - . 025 - . 025] ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Cut Ri bSect i ons = [ 1 3] ,
Type = 0,
Fusel ageI nt er f ace = 0,
Cr eat eConnect i onGr i d = 1,
/

&Cr ossSect i on
I D = 21,
Loc = [ 450. 0, 0, 32] ,
Chor d = 90. 0,
Shape = ' 0005' ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Type = 3,
Fusel ageI nt er f ace = 1,
/

&Cr ossSect i on
I D = 22,
Loc = [ 500, 0, 80] ,
Chor d = 40,
Shape = ' 0005' ,
Boundar yAct i ve = 0,
Chor dSpaci ng = 4,
Type = 3,
Cl osedWi ngTi p = 1,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ WI NG SECTI ON DEFI NI TI ONS
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

&Wi ngSect i on
I D = 1,
Car r yThr ough = 1,
SpanSpaci ng = 2,
Cr ossSect i ons = [ 1 2] ,
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


17
KeepSki n = [ 0 0] ,
Hol e = [ 1 2; 6 7] ;
Ski nT = 0. 5,
Ri bT = 1. 00,
Spar T = 1. 00,
/

&Wi ngSect i on
I D = 2,
SpanSpaci ng = 12,
Cr ossSect i ons = [ 2 3] ,
Cont r ol Sur f Boxes = [ 1 71; 6 56] ,
Hi ngeLocat i on = {' t e' , ' l e' },
Hi ngeType = {' ASYM' , ' ASYM' },
Hi ngeAddedMass = [ 0. 0442, 0. 0286] ,
Hol e = [ 1 71; 7 77] ,
KeepSki n = [ 0 0] ;
Ski nT = 0. 10,
Ri bT = 1. 00,
Spar T = 1. 00,
Per pendi cul ar Ri bs = 0,
/

&Wi ngSect i on
I D = 11,
Car r yThr ough = 1,
SpanSpaci ng = 2,
Cr ossSect i ons = [ 11 12] ,
KeepSki n = [ 0 0] ,
Hol e = [ 1 1; 3 3] ;
Ski nT = 0. 2,
Ri bT = 1. 00,
Spar T = 1. 00,
/

&Wi ngSect i on
I D = 12,
SpanSpaci ng = 4,
Cr ossSect i ons = [ 12 13] ,
Ski nT = 0. 2,
Ri bT = 1. 00,
Spar T = 1. 00,
Cont r ol Sur f Boxes = [ 1 9] ,
Hi ngeLocat i on = {' 0. 333' },
Hi ngeType = {' ASYM' },
Hi ngeAddedMass = [ 0. 0135] ,
/

&Wi ngSect i on
I D = 21,
Car r yThr ough = 1,
SpanSpaci ng = 2,
Cr ossSect i ons = [ 20 21] ,
KeepSki n = [ 0 0] ,
Hol e = [ 1 1; 3 3] ;
Ski nT = 0. 2,
Ri bT = 1. 00,
Spar T = 1. 00,
/

&Wi ngSect i on
I D = 22,
SpanSpaci ng = 4,
Cr ossSect i ons = [ 21 22] ,
Ski nT = 0. 2,
Ri bT = 1. 00,
Spar T = 1. 00,
Cont r ol Sur f Boxes = [ 3 9] ,
Hi ngeLocat i on = {' l e' },
Hi ngeType = {' ANTI SYM' },
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5


American Institute of Aeronautics and Astronautics
Public Release: 88ABW-2012-1694


18
Hi ngeAddedMass = [ 0. 029] ,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ MATERI AL DEFI NI TI ONS
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

&Mat er i al
I D = 10,
Type = ' basi c' ,
E = 10e6
nu = 0. 3,
r ho = . 1,
Yei l d = 70000. ,
Thi cknessLB = 0. 02,
Thi cknessUB = 3. 00,
/

&Mat er i al
I D = 20,
Type = ' beam' ,
E = 10e6
nu = 0. 3,
r ho = . 1,
Yei l d = 70000. ,
Thi cknessLB = 0. 1,
Thi cknessUB = 1. 00,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ FUSELAGE DEFI NI TI ON
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

&Fusel age
XDES = [ 0 65 130 195 260 325 390 455 520 585] ,
T = [ . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 ] ,
Xl oc = [ 0 65 130 195 260 325 390 455 520 585] ,
Zl oc = [ 0 4 4 8 0 0 0 0 0 0] ,
A = [ 0 15 25 32. 5 45 45 45 45 45 45 ] ,
B = [ 0 10 20 40 32 32 32 32 25 15 ] ,
ACON = [ 0 1 1 1 1 1 1 1 1 1 ] ,
BCON = [ 0 1 1 1 1 1 1 1 1 1 ] ,
Cr ossSect i onConnect = [ 0 0 0 0 0 0 1 20 11 0 ] ,
AddedMass = [ 0 1000 1000 150 0 0 670 1033 1033 1033] ,
Mach = 1000,
Opt i mi ze = 0,
Cycl es = 0,
Bui l dSt r = 1,
FuseMat = 20,
/

$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
$ ANALYSI S PARAMETERS
$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$

&Gl obal Tr i m
Aest at = {' ANGLEA' , ' URDD3' },
/
&Tr i m
name = ' 9G LI FT' ,
Mach = 0. 9,
Densi t y = 9. 0532e- 08,
Dynami cPr essur e = 6. 5,
Fi xedVar = {' URDD3' , ' CS1001' , ' CS1002' , ' CS1003' , ' CS1004' , ' CS1005' , ' CS1006' , ' CS1007' },
Fi xedVal = [ 9. 0 0 0 0 0 0 0 0] ,
CSLi nk = [ ] ,
Const r ai ned = [ 1 2 4 5 6] ,
Suppor t ed = [ 3 ] ,
Symmet r y = ' ASYMMETRI C' ,
/
D
o
w
n
l
o
a
d
e
d

b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
C
H
I
G
A
N

o
n

M
a
y

8
,

2
0
1
3

|

h
t
t
p
:
/
/
a
r
c
.
a
i
a
a
.
o
r
g

|

D
O
I
:

1
0
.
2
5
1
4
/
6
.
2
0
1
2
-
1
4
2
5

Вам также может понравиться