SimSAC

Simulating Aircraft Stability and Control

Characteristics for use in Conceptual Design

Specific Targeted Research Project

Start: 01 November 2006
Duration: 36 Months

CEASIOM Tier I and Tier I+ modules

for simulation of aircraft
stability and control characteristics

Prepared by:

Paul Molitor

KTH

Document control data

Deliverable No:

D 3.2-5

Due date:

Juli 31, 2009

Version:

Task manager:

KTH

Date Delivered:

July 30, 2009

Project manager:

Prof. Arthur Rizzi

Project co-funded by the European Commision within the Sixth Framework Programme (2002-2006)
Dissemination level
PU Public
x
PP Restricted to other programme participants (including the Commission Services)
RE Restricted to a group specified by the consortium (including the Commission Services)
CO Confidential, only for members of the consortium (including the Commission Services)

Deliverable
Version
Page

KTH

D 3.2-5
1
2 of 79

Change Log
ISSUE
1

PAGE
all

CHAPTER
all

DESCRIPTION
First Issue

DATE
31.07.2009

VIS
Paul Molitor

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
3 of 79

Contents
1 Introduction

2 Mathematical modeling of aircraft aerodynamics to investigate flight dynamics problems

2.1 Coordinate system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.2 Control sign convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.3 Aerodynamic tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.4 Simulation and Dynamic Stability Analyzer . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.5 Mathematical model of aerodynamics of the aircraft

. . . . . . . . . . . . . . . . . . . . . 11

2.6 Aeroelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Presentation of Tier I and Tier I+ tools and geometries

13

3.1 Adaptive-Fidelity geometry modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.2 Adaptive-Fidelity CFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Testcases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Tier I: The semi-empirical handbook method - Digital Datcom code

18

5 Tier I: Linear flow physics - The Tornado vortex lattice code

19

5.1 Classical Vortex Lattice Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5.1.1 Historical Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.1.2 Mathematical Background of the Vortex Lattice Method . . . . . . . . . . . . . . . 20
5.2 The Tornado software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2.1 Main features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2.2 Geometry representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2.3 Programming environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2.4 Extended Implementations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.3 Other VLM codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.3.1 AVL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.3.2 VORSTAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.3.3 SHAMAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.4 Calculation of unsteady derivatives using VLM . . . . . . . . . . . . . . . . . . . . . . . . 25
5.4.1 Frequency domain software

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

5.4.2 Time Domain method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

5.4.3 Comparison of the two methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.5 Fuselage model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5.1 Cruciform fuselage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5.2 Slender body theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.5.3 Comparison of the two methods and conclusion . . . . . . . . . . . . . . . . . . . 36

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
4 of 79

5.6 Leading edge vortex correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

5.6.1 Leading edge vortices and their influence on low aspect ratio wing aerodynamics . 37
5.6.2 Comparison with experimental data for Delta wings . . . . . . . . . . . . . . . . . . 38
5.6.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.7 Viscous corrections

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

5.7.1 Torfoil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.7.2 Component Buildup Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.7.3 Comparison of the two viscosity correction models . . . . . . . . . . . . . . . . . . 44
5.7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.8 Stall model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.8.1 Empirical Stall model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.8.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.9 Final version of the vortex lattice method Tornado in CEASIOM . . . . . . . . . . . . . . . 51
6 Tier I+: Non-linear flow physics - The Euler solver Edge

52

7 Benchmarking

54

7.1 f12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7.2 X-31 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
7.3 Ranger 2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8 CEASIOM in design

68

8.1 TCR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
8.2 GAV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
9 Recommended best practices

72

References

75

Appendix :

79

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
5 of 79

Summary

In the current report the mathematical model of the unsteady aerodynamics on the aircraft in arbitrary
motion is briefly discussed.
The tool used to generate the aero-data is CEASIOM, the Computerized Environment for Aircraft Synthesis and Integrated Optimization Method. CEASIOM uses adaptive-fidelity CFD to create the aerodynamic dataset of the aircraft and then uses the dataset to analyze flying qualities. The adaptable-fidelty
modules are referred to as Tier I, Tier I + and Tier II.
1. Tier I: Handbook method DATCOM and steady and unsteady TORNADO vortex-lattice code
(VLM) for low-speed aerodynamics
2. Tier I+: Inviscid EDGE Euler code for high-speed aerodynamics and aero-elasticity
3. Tier II: RANS (Reynolds Averaged Navier-Stokes) flow simulator for high-fidelity analysis of extreme flight conditions
In this report, the main attention is given to Tier I and Tier I+ modules.
A review of the software based on the vortex lattice method (VLM) to calculate the stability, control,
rotary and unsteady aerodynamic derivatives is presented. The possibility to use the VLM software
in steady and unsteady problems is discussed. The fuselage is represented using the slender body
approximation, and methods to account for the viscosity and leading edge vortices are analyzed.
The handbook method DATCOM and the inviscid EDGE Euler code are also presented and briefly
discussed.
Finally the Tier I and Tier I+ modules are applied to several testcases. A conclusion is drawn out of
these calculations under which conditions which module should be used.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
6 of 79

List of acronyms
Acronym
AMB
AR
CAD
CEASIOM
DOF
FCS
FCSDT
GAV
LQR
MAC
RANS
RHS
S&C
SAS
SDSA
SimSAC
TCR
VLM

Aerodynamic Model Builder

Aspect Ratio
Computer Aided Design
Computerized Environment for Aircraft Synthesis and Integrated Optimization
Degree of Freedom
Flight Control System
Flight Control System Designer Toolkit
General Aviation Vehicle
Linear Quadratic Regulator
Mean Aerodynamic Chord
Reynolds-Averaged NavierStokes
Right Hand Side
Stability and Control
Stability Augmentation System
Simulation and Dynamic Stability Analyzer
Simulating Aircraft Stability and Control Characteristics for use in Conceptual Design
Transonic Cruiser
Vortex Lattice Method
Table 1: List of acronyms

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
7 of 79

1 Introduction

Present trends in aircraft design towards augmented stability and expanded flight envelopes call for an
accurate description of the non-linear flight-dynamic behavior of the aircraft in order to properly design
the Flight Control System (FCS). Currently aircraft static and dynamic stability derivatives as well as
control ones are determined in conceptual design by semi-empirical methods (DATCOM/ESDU) based
on prior testing, later corrected in preliminary design by experimental measurements. Relying on current
semi-empirical relations in conceptual design restricts progress because such relations make extensive
use only of statistical methods based on previous experience. This increases the risk associated with
new concepts since their prototypes are absent.
Advances in computing power over the last decades together with recent progress in software for computer aided design (CAD) now permit the introduction of linear aerodynamic methods (Tier I level of
analysis) to replace the semi-empirical approach in the current conceptual design. The increase of the
knowledge about stability and control (S&C) at early stage of aircraft development will result in the diminishing of mistakes in the FCS design and architecture. It is well known that later correction of these
mistakes proves very costly since of unscheduled activities related to additional wind tunnel and flight
tests. Apart from potential for significant cost overruns, such problems would also lead to a delay in
aircraft certification and delivery. In addition, modifications usually result in some penalty to the aircraft
design and may in fact pose some operational limitations.
In the beginning of current report the conventional mathematical model of unsteady aerodynamics
used to simulate aircraft 6DOF flight dynamics is briefly presented. In this model aerodynamic forces
and moments are assumed to depend linearly on small changes in flight parameters experienced from
its equilibrium flight. These linear additions to steady trim flight aerodynamics are expressed using
the static stability, rotary and unsteady damping derivatives as well as aerodynamic control derivatives.
Nonlinear, high-order, frequency-dependent or time-dependent terms are assumed to be negligible.
This linear model is used everywhere during flight dynamics simulation. Accuracy of this model is
totally satisfactorily for attached flow conditions.
The computational models considered here range from handbook methods (USAF Digital DATCOM[12]),
through linear singularity methods (Vortex Lattice Method[13], Panel Methods such as dwfSolve[15]) to
full non-linear Euler and RANS compressible flow CFD packages[22]. The lower fidelity Tier I models
are acceptable for low angles of attack and low speed, whereas the Tier I+ Euler model extends the
predictable region of the envelope by capturing compressibility effects, and the Tier II RANS models
include also viscous effects.
In conclusion of the report the implemention of Tier I and Tier I+ calculations of aircraft stability and
control characteristics in simulation system called Computerized Environment for Aircraft Synthesis
and Integrated Optimization (CEASIOM) is presented.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
8 of 79

2 Mathematical modeling of aircraft aerodynamics to investigate flight

dynamics problems
2.1

Coordinate system

Flight dynamics, aerodynamics and other home-made codes have their own reference frames. Additionnaly, countries have their own system notations. The cordinate system of the vortex lattice method
Tornado[13], of the geometry definiton XML file[1], AMB[18] and the stability and control tool SDSA[2]
were compared with the international normed coordinates system ISO 1151-1[4]. The largest difference
is between the aerodynamic coordinate system, which is used by the xml file and the aerodynamic
codes, and the flight mechanics coordinate system, which is used by AMB and SDSA.
In the XML file, the origin is the nose of the aircraft. The X-axis goes from the nose to the tail, the Y-axis
goes to the right wing, and the Z-axis points up.
Tornado uses a Cartesian coordinate system with the X-axis along the aircraft body, increasing aft. The
Y-axis is aligned positive out through the starboard wing when no dihedral is present. The Z-axis is
right-hand perpendicular to the X- and Y-axis, thus pointed up. The origin can be fixed by the user, and
is normally fixed at the aerodynamic center. The coordinate system is the same then in the XML file,
just with a different origin.
SDSA uses a body axis system. The origin is placed in a point usually defined by 1/4 MAC (Mean
Aerodynamic Chord). The axis OX is parallel to the MAC and pointed forward to the nose of the aircraft,
axis OZ is pointed down, and axis OY is pointed to the right wing. This is coherent with the international
normed coordinates system ISO 1151-1[4].
In AMB, the origin is the reference point. The reference point can be specified by the user, it can be
the center of gravity, the 1/4 MAC or a user defined point. AMB uses the same coordinate system then
SDSA.

Figure 2.1: Coordinate systems

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

2.2

D 3.2-5
1
9 of 79

Control sign convention

The SimSAC convention will be as defined by Cook[5] which is: a positive control action by the pilot
(from the pilots point of view) gives rise to a positive aeroplane response, whereas a positive control
surface displacement gives rise to a negative aeroplane response. Thus:
in roll: positive right push force on the stick positive starboard stick displacement right aileron
up and left aileron down (negative) starboard wing down roll response (positive)
in pitch: positive pull force on the stick positive aft stick displacement elevator trailing edge
up (negative) nose up pitch response (positive)
in yaw: positive push force on the right rudder pedal positive rudder bar displacement rudder
trailing edge displaced to the right (negative) nose to the right yaw response (positive)
The positive control surfaces deflection are shown on the 2.2.

Figure 2.2: Positive angles

2.3

Aerodynamic tables

A prerequisite for realistic prediction of the S&C behavior and sizing of the FCS is the availability of
complete and accurate aerodata (i.e. the S&C database). The aerodata is represented by an multidimensional array of dimensionless coefficients of aerodynamic forces and moments, stored as a function
of the state vector and control surface deflections. The aerodynamic table in AMB has the following
format[18].
is the angle of attack, M is the Mach number and the side slip angle. q, p and r are the three
rotations in pitch, roll and yaw. The three control surfaces that can be deflected are the elevator (e ), the
rudder (r ) and the aileron (a ). The table is linearised and is build up from 7 three-dimensional tables
with , M and a third parameter (, q, p, r, e , r or a ).

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

x
x
x
x
x
x
x

M
x
x
x
x
x
x
x

x
-

q
x
-

p
x
-

r
x
-

e
x
-

r
x
-

a
x

CL
x
x
x
x
x
x
x

CD
x
x
x
x
x
x
x

Cm
x
x
x
x
x
x
x

CY
x
x
x
x
x
x
x

Croll
x
x
x
x
x
x
x

D 3.2-5
1
10 of 79

Cn
x
x
x
x
x
x
x

Table 2.1: Format of the aerodynamic table

2.4

Simulation and Dynamic Stability Analyzer

Once the aerodynamic coefficients have been obtained along with the mass and inertia properties, for
the flexible aircraft using the eta values the S&C aerodata database is in hand. The S&C analysis can
begin with SDSA described here, or another tool like J2 Universal Tool-Kit[6] or FCSDT[7]. The SDSA
(Simulation and Dynamic Stability Analyzer) provides the following functionalities[2]:
1. Stability analysis:
a. Eigenvalue analysis of linearized model
b. Time history identification (nonlinear model)
2. Six Degree of Freedom flight simulation:
a. Test flights, including trim response
b. Turbulence
3. Flight Control System:
a. Human pilot model
b. Stability Augmentation System (SAS)
c. FCS based on Linear Quadratic Regulator (LQR) theory
4. Performance prediction
5. Miscellaneous (data review, results review, cross plots, etc.)
Figure 2.3 illustrates the structure and functionality of this module.
SDSA uses the classic six DoF mathematical nonlinear model[8], of the aircraft motion for all functions.
For the eigenvalue analysis, the model is linearized numerically by computing the Jacobian matrix of
the state derivative around the equilibrium (trim) point. Eigenvalues and eigenvectors analysis allow
automatic recognition of the typical modes of motion and their parameters. The flight simulation module
can be used to perform test flights and record flight parameters in real-time. The recorded data can be
used for identification of the typical modes of motions and their parameters (period, damping coefficient,
phase shift). The stability analysis results are presented as figures of merits based on JAR/FAR, ICAO,
and MIL regulations. The SDSA embedded flight control system allows a pilot in the loop, and SAS and
FCS based on a LQR approach. The LQR-FCS module allows computing and saving control matrices
for simulations over the whole envelope. In this way, SDSA includes the FCS for stability characteristics
and in flight simulation for the closed loop case. The performance option is designed to compute basic
performance parameters: flight envelope (Vmin and Vmax versus altitude of flight), selected manoeuvres
(e.g. regular turn), range and endurance characteristics. For all mentioned functionalities the starting
point is the computation of the trimmed state with sufficient initial conditions.
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
11 of 79

Figure 2.3: SDSA structure and functionality

2.5

Mathematical model of aerodynamics of the aircraft

The mathematical model of aerodynamics considered for the stability analysis is quasi-steady aerodynamics. Aerodynamic forces, on the instance of force along Ox, can be expectanded into Taylor
series:
Xa = Xa



2 X u2
u2 2!
X
2 X v2
+ v v + v2 2!

2
2
+ X
w + wX2 w2!
w
2
2
+ X
p + pX2 p2!
p

2
2
+ X
q + qX2 q2!
q

2
2
+ X
r + rX2 r2!
r

2
2
+ X
u + uX2 u2!


2
2
+ X
+ wX2 w2!
w w

 u

u+

+
+

+
+
+
+
+
+

3X
u3
3X
v3
3X
w 3
3X
p3
3X
q 3
3X
r 3
3X
u 3
3X
w 3

u3
4 X u4
3! + u4 4! + ...
4 X v4
v3
3! + v4 4! + ... 
4 X w4
w3
3! + w 4 4! + ...

p3
4 X p4
+
+
...
3!
p4 4!

q3
4 X q4
+
+
...
3!
q 4 4!

r3
4 X r4
+
+
...
3!
r 4 4!

u 3
4 X u 4
+
+
...
4
3!
u 4!

w 3
4 X w 4
+
+
...
4
3!
w 4!

...
It was assumed that only primary aeroddynamic derivatives are important with the exception of derivatives of vertical component of acceleration (w).
Then, the generalized aerodynamic forces have the
following form (along Ox as an example):
Xa = Xae +

X
X
X
X
X
X
X
u+
v+
w+
p+
q+
r+
w
u
v
w
p
q
r
w

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

(2.1)

Deliverable
Version
Page

KTH

2.6

D 3.2-5
1
12 of 79

Aeroelasticity

Every aircraft is deformed under influence of the forces from the surrounding air. Especially the efficiency of control surfaces deflection can be considerably decrease by aeroelastic forces, way down to
a level that is not acceptable.
In CEASIOM, NeoCASS can be used to calculate the aeroelasticity. NeoCASS uses a linear vortex
lattice method and a doublet lattice method to predict the elastic efficiency factors. The elastic efficiency
factors are defined as:
Cxx(elastic)
(2.2)
xx =
Cxx(rigid)
These factors will be given to the Stability & Control tools to predict the handling qualities of elastic
aircrafts.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
13 of 79

3 Presentation of Tier I and Tier I+ tools and geometries

The task is to build a tabular model for the aerodynamic forces and moments on the airframe by simulation. The geometry should be represented in a way to be parameterized by a small number, say O(100),
parameters with intuitive interpretation. The computational models considered here range from handbook methods (USAF Digital DATCOM [12]), through linear singularity methods (Vortex Lattice Method
[13], Panel Methods such as dwfSolve [15] ) to full non-linear Euler and RANS compressible flow CFD
packages ([22]), see 3.1. The tools for managing the geometry modeling will be described below with
comments on the workflow, in particular on the degree of automation achievable while preserving the
engineers accountability for the quality of the data compiled. A challenge is to approach automatic
volume mesh generation for Tier I+, with geometries including control surface deflections.

3.1

Adaptive-Fidelity geometry modeling

The relevant part of the CEASIOM package which deals with geometry issues is shown in 3.1. The
geo.xml file defines the geometry with sufficient details for the Tier I computations. The lifting surfaces
are assembled from quadrilateral planforms, twist, dihedral, etc., and airfoil definitions. Body, booms,
cockpits, etc., are described by only a few key parameters.

Figure 3.1: Geo-SUMO module of CEASIOM: from Geo to CFD grids.

The geo.xml file is edited by the ACBuilder GUI which gives visual feed-back of not only external
geometry as needed for aerodynamics but also data necessary for weights and balance estimates.
Panel methods and Euler simulations require much higher fidelity geometry. The aircraft must be represented by a closed surface, smooth enough to support a surface grid with proper refinements at critical
places like leading and trailing wing edges, wing tips, etc. But also the surfaces on the fuselage, canopy,
fairings, etc., must be well-rounded not to create spurious pressure peaks or troughs. The SUMO package [14, 15] builds the aircraft surface from a set of closed spline surfaces and provides a proper GUI
for designing the shapes from cross sections and control points. SUMO calculates the intersections and
can perform local smoothing and closure of features such as open wing tips, as necessary, to make a
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
14 of 79

single closed surface. It can proceed to generate a triangular surface mesh with density controlled by
radii of curvature, etc., from a small set of user parameters.
The geo.xml - SUMO interface provides most of the data necessary, but user interaction is required
when the xml geometry is inadequate. Typically, components such as vertical and horizontal tails and
the rear fuselage may not intersect properly; SUMO will then attempt repair with default parameter
settings and issue error messages; the response called for is to change the geometry using ACBuilder.
Control surface deflections can be done by actual geometry deformation before mesh generation, or
by manipulation of surface normals. The surface deformation currently fills the gaps that are created;
details of multi-element high-lift systems are not supported. The step from surface mesh to volume
mesh is taken by the TetGen[16] package, which needs only a few user parameters to fill the volume
between exterior of aircraft and the far-field sphere by a tetrahedral mesh. The quality of the surface
mesh is crucial. Inadequate surface meshes are often caused by surface irregularities, and call for
geometry repair by the engineer.
The Tier II geometry models require high-quality surfaces with all relevant details. Such high-quality
geometry models can be created by SUMO and sent as IGES[19] files to fully-fledged mesh generator
systems such as Ansys ICEM CFD[17]. For existing aircraft, data, including a CAD model, may be
available for validation experiments and modification exercises. The approximation of a given CAD
geometry by the geo.xml format is not a well defined task. It is currently done manually by the
engineer, by extracting cross sections etc. as native SUMO input, or, with even more radical shape
approximation, by adapting the O(100) parameters of geo.xml to best fit the CAD surface data.
An example of adaptive-fidelity geometry modeling can be seen in 7.8.

Figure 3.2: Ranger 2000 geometry renderings and computational grids. Left to right: xml-geometry
XML, SUMO geometry SUMO, and CAD geometry CAD[23]

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

3.2

D 3.2-5
1
15 of 79

Adaptive-Fidelity CFD

Traditionally, wind-tunnel measurements are used to fill look-up tables of forces and moments over the
flight envelope but wind-tunnel models become available only late in the design cycle. To date, most
engineering tools for aircraft design rely on handbook methods or linear fluid mechanics assumptions.
The obvious option is to use Computational Fluid Dynamics (CFD) early in the design cycle. It has
the predictive capability to generate data but the computational cost is problematic, particularly if done
by brute force: a calculation for every entry in the table. The total entries can number in the tens of
thousands. Fortunately methods are available that can reduce the computational cost.
There are essentially three issues, see Fig 3.3.

Figure 3.3: Architecture of the Aerodynamic Dataset Generator AMB-CFD

First, a spectrum of computational tools are available, from RANS to potential flow models and semiempirical methods. Each of the tools has a range of validity which can be exploited to keep the computational cost down. For the preliminary design of the aircraft and its FCS and as long as the flight
attitude remains well within the limits of the flight envelope in the range of low-speed aerodynamics,
Tier I computational methods can provide the aerodata.
For a refined design of the FCS or for flight attitudes close to the border of the flight envelope, the
linear or inviscid methods used in the Tier I tools fail to predict the proper aerodynamic behavior and
Tier II RANS methods will be used. In addition, data fusion [18] can be used for data from different
sources, with low fidelity / low cost data indicating trends and a small number of high fidelity / high cost
simulations correcting the values.
Secondly, mesh-free interpolation methods can significantly reduce the number of data points which actually need to be computed to fill the table. Some studies [20, 18, 21] of using kriging for the generation
of aerodynamic data have been published.
Thirdly, the identification of parameter regions where the aerodynamics is nonlinear, and hence where
Tier II fidelity is needed, is a sampling problem. Therefore CEASIOMs aerodynamic module develops
along with these three elements.
Here is an overview of the CEASIOMs aerodynamic modules:
1. Tier I:
Digital DATCOM, a semi-empirical method which can rapidly produce the aerodynamic derivatives based on geometry details and flight conditions. DATCOM was primarily developed to
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
16 of 79

estimate aerodynamic derivatives of conventional configurations.

The TORNADO vortex-lattice code is used to simulates low-speed flow at small angles of
attack. TORNADO allows a user to define most types of contemporary aircraft designs with
multiple wings, both cranked and twisted with multiple control surfaces located at the trailing
edge. Each wing is permitted to have unique definitions of both camber and chord. The
TORNADO solver computes forces, moments, and the associated aerodynamic coefficients.
The aerodynamic derivatives can be calculated with respect to: angle of attack, angle of
sideslip, roll-pitch-yaw rotations, and control surface deflections. Unsteady derivatives can
be computed, viscosity is taken into account by an empirical extension and the fuselage is
also taken into account in a simplified way.
2. Tier I+: Inviscid EDGE Euler code for high-speed aerodynamics and aero-elasticity. A Matlab
interface was written in order to allow Edge calculations to be prepared and run. This call runs
the preprocessing routines, launches the calculation and processes the solution for the forces and
moments.
3. Tier II: RANS (Reynolds Averaged Navier-Stokes) flow simulator for high-fidelity analysis of extreme flight conditions. The Tier II CFD tools are currently loosely coupled to CEASIOM because
users are mainly interested in coupling their own RANS CFD tools. However, standard interfaces
and file formats are defined in CEASIOM to which different RANS solvers have been coupled with
MATLAB and Python scripts to perform sequences of runs and collect results.

3.3

Testcases

Two kind of testcases were used: complete configurations and simplified models. The simplified models
were mostly used to validate the different extensions of Tornado. The complete configurations were
used to demonstrate the accuracy of the different CEASIOM modules. AMB needs to have a complete
configuration to run properly all the modules, that is why Tornado was used as stand alone for most of
the simplified models.
The simplified models used to demonstrate the fuselage model, the viscosity correction and the stall
model were a list of different wing and wing-body configurations. Experimental data exist in the form of
NACA reports on windtunnel tests.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
17 of 79

AR

c/4 [deg]

NACA airfoil

Re106

1 (a, b)

9.02

0.4

64210 (a) /65210 (b)

4.4

[57]

37.5

0.49

23012

1.1

[57]

10

5.2

30

0015

0.91

[57]

12

14.04

0.33

23012

0.87

[57]
[57]

13
14

6
6

6
0

0.2
1

23012
0012

0.9
1.0

[57]

21

5.2

-30

0015

0.91

[59]

45

0.6

2-006

2.01

[60]

39.45

0.4

64A006

9.7

Ref.

Model

[56]

Skethches

Table 3.1: Test cases

To test the Polhamus suction analogy, a list of delta wings was used as testcase.
Case
chord (c)
span (b)
aspect ratio (AR)
leading edge sweep angle (SW)
quarter chord sweep angle (1/4SW)

1
1
0.25
0.5
82.88
80.54

2
1
0.5
1.0
75.96
71.57

3
1
0.75
1.5
69.44
63.44

4
1
1.0
2.0
63.43
56.31

5
1
1.25
2.5
57.99
50.19

6
1
1.5
3.0
53.13
45

Table 3.2: Delta Wings

The complete configurations tested consist of an airliner, the f12, an trainer, the Ranger 2000 and an
experimental fighter, the X-31. Two design cases of an unconventional configuration, the transonic
cruiser TCR and the Z-wing configuration GAV, using the CEASIOM tools are also included in the
report.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
18 of 79

4 Tier I: The semi-empirical handbook method - Digital Datcom code

The Data Compendium (DATCOM) is a document of more than 1500 pages covering detailed methodologies for determining stability and control characteristics of a variety of aircraft configurations. In 1979,
DATCOM was programmed in Fortran and renamed the USAF stability and control digital DATCOM.
Digital DATCOM is a semi-empirical method which can rapidly produce the aerodynamic derivatives
based on geometry details and flight conditions. DATCOM was primarily developed to estimate aerodynamic derivatives of conventional configurations[24, 25]. For a conventional aircraft, DATCOM gives all
the individual component (body, wing, horizontal and vertical tail), and aircraft forces and moments.
Digital DATCOM has been implemented in AMB. A DATCOM input file is produced by interpreting and
formatting the XML aircraft data. In addition the flight conditions of interest are added to the DATCOM
file.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
19 of 79

5 Tier I: Linear flow physics - The Tornado vortex lattice code

The mathematical models of aircraft motion contains a number of aerodynamic derivatives with respect
to the kinematical parameters of aircraft motion. The most straight forward way to estimate aerodynamic
derivatives is the Vortex Lattice Method (VLM). In the current section of the report the VLM technique
is briefly discussed as well as the possible extensions.

5.1

Classical Vortex Lattice Method

5.1.1 Historical Background

The classical vortex lattice method originated from the Prandtl Lifting line theory. Since the computers
become available as scientific tools in the 1960s the vortex lattice method has been thoroughly investigated and widespread around the world. In 1975 NASA held a workshop dedicated to various problems
solved using VLM. This workshop[29] came up with a number of papers covered various topics of VLM:
configuration design and analysis, high and vortex lift, lattice arrangement, unsteady flows. Later some
nonlinear problems like vortex shedding from the leading edges of high sweep wings and certain unsteady cases were solved [30, 31, 32]. With the progress of computational aerodynamics, new more
accurate panel methods appeared [33, 34, 35]. In general, panel methods are much more laborious
and require more accurate knowledge of the aircraft geometry. Currently these methods are used in
industry in particular to design the shape of aircraft wing and fuselage. The VLM codes remain the
working horses for quick estimation of aerodynamic stability and control derivatives at the early stage
of conceptual design.
In the online textbook [36] it was noted that VLMs are very similar to panel methods, but easier to use
and capable of providing remarkable insight into wing aerodynamics and component interaction. The
steady vortex lattice method for low speed is based on the solution of Laplaces Equation, and is subject
to the same basic theoretical restrictions that apply to panel methods.
For comparison, vortex lattice methods are:
Similar to Panel methods:
singularities are placed on a surface
the non-penetration condition is satisfied at a number of control points
a system of linear algebraic equations is solved to determine the singularity strengths
Different from Panel methods:
Oriented toward lifting effects, and classical formulations ignore thickness
Boundary conditions are applied on a mean surface, not the actual surface (not an exact
solution of Laplaces equation over a body, but embodies some additional approximations,
i.e., together with the first item, on the lifting surface only Cp can be found, not Cpupper and
Cplower )
Singularities are not distributed over the entire surface
Oriented toward combinations of thin lifting surfaces (Panel methods have no limitations on
thickness).
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
20 of 79

5.1.2 Mathematical Background of the Vortex Lattice Method

The assumption is made that the flow consist of ideal compressible gas. The main movement is only
subjected to small disturbances. Further, the flow is assumpted as irrotational (vortex free) except for
the vehicle lifting surfaces and their wake. The effects of vortex wake turning (roll up) are neglected.
For inviscid and irrotational flow, the disturbance velocity potential can be introduced. With these
assumptions the problem can be treated using potential flow techniques.
The equations are solved in dimensionless form, using a characteristic size of the aircraft (chord, span,
length, etc) and using the velocity of the flow at infinity. The angles of attack and sideslip, the angular
rates of aircraft in pitch, yaw and roll, and the angles of control surfaces deflections are assumed to be
small.
As the flow is irrotational, the disturbance velocity can be expressed using a velocity potential
~v = .
The main unsteady equation is as follows
(1 M 2 )

2
2
2 2 2
2
2
+
+

2M

M
= 0,
x2
y 2
z 2
xt
t2

(5.1)

where M is the Mach number of the flow.

For steady flow, the above equation can be reduced to the Prandtl-Glauert equation
(1 M 2 )

2 2 2
+
+ 2 = 0.
x2
y 2
z

(5.2)

This equation can be reduced to the incompressible Laplace equation using the Prandtl-Glauert transformation for the x coordinate. Thus, for steady flow the compressible problem can be reduced to the
incompressible problem using only coordinate transformations.
The pressure distribution along the body for unsteady flow can be calculated using the Bernoulli equation. By linearization, the Bernoulli equation is
Cp =

p p0

= 2(
+
).
q0
t
x

(5.3)

The first boundary condition, the no penetration boundary condition, requires zero normal velocity
across the body elements is expressed as following
~ ~n = 0
~n V

(5.4)

~ is the gas kinematic velocity as viewed from

where ~n is the unit vector normal to the base element, V
the body frame.
The second boundary condition requires that the flow disturbances due to the body motion through the
gas should diminish far from the aircraft
lim = 0

~
r

(5.5)

where ~r = (x, y, z). The pressure distribution in the vortex wake must be continuous. Using (5.3), one
gets

(+ ) + (+ ) = 0
x
t
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

(5.6)

Deliverable
Version
Page

KTH

D 3.2-5
1
21 of 79

where + and are the disturbance potentials above and below the vortex sheet. This boundary
condition corresponds to the condition of circulation conservation along the liquid contour.
The boundary condition (5.6) must be valid not only for vortex wake but also for the trailing edge of the
wing where it corresponds to the Kutta - Zhukovsky condition.
To solve the above stated mathematical problem, one can replace the lifting surfaces of the aircraft by
vortex lattices. The fuselage of aircraft can be represented by two flat elements of the shape corresponding to its planar projections. This reduce the problem to find the intensity of the circulation of
the lattices. Enlarging the number of vortices in the lattice permits to solve the aerodynamic task with
required accuracy.
Equation (5.1) and boundary condition (5.5) are satisfied automatically for arbitrary time and space
dependencies of vortices circulations. The vortex wake intensity is governed by the aircraft vortex
lattice circulation through (5.6). The Kutta - Zhukovsky condition is satisfied when the last collocation
point in each section is placed between the last vortex and the trailing edge [26].
Thus, the circulation of the vortices can be found directly through the boundary condition (5.4), requiring
zero normal velocity on the aircraft surfaces (no slip condition). Then the pressure distribution along the
body can be found using (5.3). For the lifting surfaces, the aerodynamic loads can be calculated from
the pressure difference betwen the upper and lower part of the surface using the local representation
of Zhukovsky theorem.
p = V0

(5.7)

where is the appropriate component of vortex lattice circulation.

5.2

The Tornado software

5.2.1 Main features

The Tornado code is a vortex lattice method programmed to be used in conceptual aircraft design and
in aerodynamic education. The code was developed at the Department of Aeronautics of the Royal
Institute of Technology (KTH), in Stockholm, Sweden [13]. The first version was released in 2001.
The aircraft geometry in Tornado is fully three dimensionally oriented lifting surfaces, with flexible, freestream following wake. Tornado allows a user to define most types of contemporary aircraft designs with
multiple wings both cranked and twisted with multiple control surfaces. Each wing may have taper of
both camber and chord, which may vary spanwise. The Tornado solver solves for forces and moments
from which the aerodynamic coefficients are computed. Aerodynamic derivatives can be calculated
with respect to: angle of attack, angle of sideslip, roll-, pitch- and yaw- rotations and control surface
deflection. If necessary all of these conditions may be applied simultaneously. Uncompressible fluid
conditions are calculated using the Prandtl-Glauert correction, which gives reasonably good results up
to Mach 0.6. Any user may edit the program and design add-on tools as the program is coded in
MATLAB (tm) and the source code is provided under the GNU-General Public License.
Some difference with comparison of classic VLM approach is in the using of vortex-sling concept instead
of horse-shoe vortices. The vortex sling is essentially an seven segment vortex line, which, for each
panel, starts at the infinity behind the aircraft reaches the trailing edge, moves upstream to the hinge
line of the trailing edge control surface, then forward to the quarter chord line of the panel in question,
going across the panel and then back downstream in a similar way.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
22 of 79

5.2.2 Geometry representation

Tornado supports multi-wing designs with swept, tapered, cambered, twisted and cranked wings with
or without dihedral. Any number of wing may be utilized as well as any umber of control surfaces.
Canards, flaps, ailerons, elevators and rudders may be employed. Winglets, fences and engine mounts
may also be incorporated in the design.
The tornado preprocessor allow the user to enter wing data in the standard aeronautical engineering
format: 1) Apex coordinate, 2) Span, 3) Taper, 4) Sweep, 5) Camber, 6) Dihedral, 7) Twist, 8) Symmetry,
9) Root chord, 10) Flaps, 11) Flap symmetry (If flap is present), 12) Flap chord in parts of root chord (If
flap is present), 13) Number of panels in chord (X) direction, 14) Number of panels in span (Y) direction,
15) Number of panels on the flap (If flap is present).
Sometimes it is convenient but sometimes when only the three view picture of designed aircraft is
available it would be more convenient to specify the lifting surfaces by coordinates of leading and trailing
edges directly.

5.2.3 Programming environment

The choice of platform for the Tornado software is the MATLAB environment. MATLAB integrates mathematical computing, visualization and a powerful language to provide a flexible environment for technical
computing. The Matlab programming language is a high level language intended for numerical simulations. Since the VLM is mainly a matrix problem, the Matlab capabilities in working with vectors and
matrices are very useful. Matlab supports online execution and variable manipulation, as well as scripting or creation of procedures. This makes Matlab ideal for code prototyping and development. Fortran
on the other hand does not have the capacity to let the user modify variables during the runtime.
Code readability is also an important factor, as source code sometimes becomes the only available
software documentation. This is an increasingly important factor if the software is to be distributed and
developed by other persons than the original programmer. It is common knowledge that Matlab code is
more readable than Fortran code.
Another trait of Matlab is the availability of data visualization routines. In Fortran program the visualization is also possible but it takes more time and usually involves a third party software.
A big advantage of Matlab is the possibility to create graphical user interface (GUI). In release 12
of Matlab and later the design tools for graphical user interface are powerful and allow fast design
of complex user interface. This is a property of Fortran as well, but certainly belonging to a high
professional programming level.
Of course, Matlab has some drawbacks also. The main is as follows: codes in Matlab are inherently
slower than their compiled Fortran counterparts. But there is the built in capability in Matlab to call
routine written in C or Fortran. So, the reasonable way is to use Matlab language with programming the
most frequently used subroutine (for example, calculation of induced velocities from horse-shoe vortex)
in Fortran.

5.2.4 Extended Implementations

During the last time some modifications of the original Tornado code were available. The modifications
concern fuselage treatement, unsteady or quasi-steady versions of Tornado, stall models and viscosity
treatement[37]. In this chapter, these modifications are discussed and the final version of Tornado is
decided. Here is a list of all the possible extensions analyzed in this chapter:

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Extension
Time domain
Frequency domain
Cruciform fuselage modelisation
Slender body fuselage modelisation
Polhamus suction analogy
Torfoil
Component Buildup Model
Empirical Stall Model

D 3.2-5
1
23 of 79

included in final version of Tornado

No
Yes
Yes
Yes
No
No
Yes
Yes

Table 5.1: List of possible extensions for Tornado

5.3

Other VLM codes

5.3.1 AVL
There is also a number of other vortex lattice implementations available for design engineer free of
charge or commercially. It worth while to mention the AVL code. The Athena Vortex Lattice program
(AVL) has been developed at MIT [38, 39, 40]. This software handles multiple lifting surfaces and
utilizes the Prandtl-Glauert compressibility correction for high subsonic Mach numbers. The singularity
distribution used is the standard horseshoe vortex. The wake is parallel to the geometrical direction of
longitudinal axis Ox. The angle of attack is limited to small values. There is a possibility to simulate the
fuselage using sink/source distribution along the corresponding part of axis Ox. AVL handles any type
of wing camber for the lifting surfaces. The user interface allow the user to investigate trimmed CL at a
specific angle of attack and create the stability derivative matrix.

5.3.2 VORSTAB
VORSTAB[41] is a computer program for calculating stability derivatives with vortex flow effect developed under NASA contract by C. Edward Lan in 1985. It is based on a quasi vortex lattice method which
is an extension of the classic formulation and takes into account of highly non-linear phenomenas.
Its main features are:
Quasi vortex lattice method
Highly non-linear
Lifting surfaces are modeled with horseshoe vortices
Bodies are modeled with vortex multiplets
Wing vortex flow is simulated using discrete free vortex filaments emanating from the edges
Effect of edge-separated vortex flow (leading and side edges) using Polhamus method of suction
analogy
Augmented vortex lift effect
Strake vortex
Vortex breakdown is model using empirical formulas
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
24 of 79

Boundary layer effects are taken into account

Calculates subsonic and supersonic forces, momenta, longitudinal stability derivates and lateraldirectional stability derivates
Control surfaces
In- and out-ground effects
Pre- and post-stall conditions
High angles of attack
Vorstab is a highly advanced program specialized in calculating complicated aerodynamic phenomena
like non-linear effects or post-stall aerodynamics.

5.3.3 SHAMAN
SHAMAN is a preliminary design prediction and analysis software tool applicable to configurations operating in post-stall flow conditions developed by Nielsen Engineering & Research[42]. It can be used
to predict specified vehicle motions or flow conditions, or it can be coupled with a six-degree-of-freedom
equation-of-motion solver to predict flight trajectories and transient performance characteristics. Poststall flight regimes are handled with empirical correlations of wing data.
The method uses vortex lattice panels and a relaxed wake in steady and quasi-steady simulations.
A vortex cloud model is used for the fuselage vorticity. In steady and quasi-steady simulations, the
separation line is predicted, vortices are shedded and tracked, and interference effects are taken into
account. It uses generic empiricism to model the stall of wing and/or canard, and takes account of
vortex bursting. This tool can be used for complete trajectory simulations.

Figure 5.1: Shaman calculation for the X-31

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

5.4

D 3.2-5
1
25 of 79

Calculation of unsteady derivatives using VLM

5.4.1 Frequency domain software

A method to calculate directly the unsteady derivatives was developed by TsAGI using classic VLM in
their UNST software[43]. This method was implement in Tornado by TsAGI[37].
The aircraft surface is approximated using horse-shoe vortices. For unsteady motion, the circulation of
each vortex is time dependant. Due to Kelvin theorem, the vorticity is shed from the transverse vortex
to the wake between two longitudinal vortices of a horse-shoe. If it is assumpted that the dimensionless
circulation of the vortex varies harmonically
(t) = 0 cos kt
then it was shown [27, 28] that when the reduced frequency tends to zero k 0 the velocities induced
by this vortex system can be expressed as follows
~v = v~(1) cos kt kv~(2) sin kt

where constituent v~(1) corresponds to the ordinary quasi steady part which can be calculated using the
Bio-Savart law for incompressible flow
v (1) =

(cos 1 + cos 1 ).
4r

The constituent v~(2) is the out-of-phase unsteady part and its analytical expression for k 0 is available
[27, 28]. In the general case, both constituents v~(1) and v~(2) are functions of the coordinates of the point
in which the induced velocities are calculated, of the coordinates of the horse-shoe vortex apexes and
of the Mach number.
The motion of an aircraft as a rigid body is described by a number of kinematical parameters
(, , p, q, r, )
where and are aircraft angles of attack and sideslip, p, q and r are angular rates in roll, pitch and yaw,
is the deflection of the aerodynamic control surfaces. In linear approximation, all these parameters
are considered to be small. The boundary condition (5.4) can be satisfied on flat base elements and
gives

ny +
nz = ( xq yp)nz + ( zp + xr)ny + f
y
z

(5.8)

where ny and nz are the corresponding constituents of the external normal vector to the aircraft surface.
Since all the flat elements of the vortex lattice are parallel to the Ox axis, nx = 0 for all panels. Function
f is equal to zero on all surfaces of the aircraft except on deflected control surfaces where f = 1.
Linear approximation permits to reduce the general boundary problem (5.8) to a number of particular equations which can be solved independently. For each particular problem, only one kinematical
parameter can be considered as a harmonic variation in time
(t) = 0 cos kt
(t)

= k0 sin kt
where 0 is the amplitude of a kinematical parameter variation and k is the corresponding reduced
frequency.
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
26 of 79

The total disturbance potential can be presented in this case as follows

X
(x, y, z, t) =
[ (x, y, z)(t) + (x, y, z)(t)]

where is the solution of a particular task when all kinematical parameters but one are equal to zero.
In-phase part of boundary condition for this particular task is expressed as follows

ny +
nz = F (x, y, z)
y
z
where F (x, y, z) are known functions of the aircraft geometry
F
F
Fp
Fq
Fr
F

=
=
=
=
=
=

ny
nz
ynz zny
xnz
xny
f

(5.9)

The out-of-phase part of the boundary condition is as follows

ny +
nz = 0
y
z
The task corresponding to the boundary condition F is known as - task, task with the boundary
condition F is - task and so on.
The strength of each horse-shoe vortex can be also considered in a general case as a sum
X

(t) =
[ (t) + (t)]

If the induced velocities are determined in collocation point, the in-phase and out-of-phase boundary
conditions can be satisfied. Two sets of linear equations are obtained. The in-phase part with omitting
the term of order k2 is as follows
{Aij }{i } = {Fj }
and out-of-phase part is
{Aij }{i } = {Bij }{i }
Here {Aij } and {Bij } are square matrices which depends on coordinates of vortices, control points and
Mach number. Each coefficient of these matrices ({Aij } and {Bij }) is equal to the normal component of
the induced velocities (v~(1) and v~(2) ) from the horse-shoe vortex number i at the control point number j.
Vectors {i } and {i } are required solutions of steady and unsteady tasks. Vector {Fj } corresponds
to the values of the boundary conditions (5.9) of each task which has to be solved in the given control
points. First set of linear algebraic equation corresponds to the steady part of the task. The second set
of the linear algebraic equations corresponds to the unsteady part. The last one can be solved after the
solution of first, steady part has been solved.
The solution of both linear equations gives the steady and unsteady parts of circulations of the vortex
lattice. With the use of local Zhukovsky theorem (5.7), the steady and unsteady parts of the pressure
difference on each panel of the vortex lattice can be calculated
Cpi =

2i
,
hi

Cpi =

2i
,
hi

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
27 of 79

where hi is the length in x direction of panel number i.

At last, the aerodynamic derivatives of aircraft can be calculated
P
CZ =
i nzi Cpi i
P
CY =
i nyi Cpi i
P
Cl = (1/b) i (ynzi znyi )Cpi i
P
Cm = (1/
c) i xnzi Cpi i
P
Cn = (1/b) i ynyi Cpi i
where i is the dimensionless area of the panel number i, c and b are the dimensionless mean aerodynamic chord and the span of the wing. Unsteady derivatives with respect to are calculated using the
same expression with substitution Cpi instead of Cpi .

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
28 of 79

5.4.2 Time Domain method

An unsteady Time Domain model, as proposed by Katz and Plotkin[32], was implemented by Adrien
Brard in his licentiate thesis[44]. This model, was updapted[45] and used to calulate and compare the
time dependant coefficients for several cases.
In this model, a time-stepping loop is introduced and a new element appears: the wake rollup computation. The calculation of the influence coefficients and the right hand sid (RHS) vector are very similar
then in the steady version of Tornado, only that the dynamic wake influence has to be considered.
For the calculation of the forces and moments, there are also some unsteady terms which have to be
considered. The main three parts of Tornado which have to be changed are:
1. the solid boundary condition needs to include the unsteady motion and the downwash due to the
wake vortices
2. unsteady terms of the Kutta-Joukowsky theorem have to be taken into account in the calculation
of the forces (using the modified Bernoulli equation to calculate the pressures)
3. an unsteady wake model has to be established and its rollup has to be computed

Figure 5.2: Schematic flow chart of the computational steps for steady Tornado (left) and unsteady
Tornado (right)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
29 of 79

Boundary condition:
To describe the three dimensional time dependent motion of the body, the zero normal flow boundary
condition is applied in a body-fixed coordinate system:
~ ~r) ~n = 0
( V~0 V~rel

(5.10)

~ the rate of rotation of the bodys

where V~0 is the velocity of the stationary, inertial, systems origin,
frame of reference and V~rel an additional relative motion within the body-fixed coordinate system (e.g.
small amplitude oscillations of the wing,..).

Figure 5.3: Inertial and body coordinate systems

Flightpath information update:

At each time step, all the flight path characteristics have to be updated: speed, angle of attack, angle of
sideslip, pitch, roll and yaw rates, vertical and horizontal heaving motion speeds.

Wake model:
The bound vorticity needs to be continued beyond the wing, and therefore a wake model is needed. In
a wake model, the wake strength at the trailing edge is specified as its shape and location.
In the steady version of Tornado, the wake consist of the vortex lines leaving the horse shoe vortex.
The vortex line can not start and end abruptly in a fluid. The wake leaves the trailing edge and extends
linearly behind the trailing edge to infinity. From the Biot-Savart law, the velocity induced on a point by
a vortex panel is inversely proportional to the distance. Thus, the vortices far away from the wing have
negligible effect on the aerodynamic loads.
In the unsteady time-step version of Tornado, the wake needs to be modeled more detailed. First, at
every time step a new range of vortices is shed at the trailing edge. Theses vortices are moved every
timestep with the local speed which allows a more exact wake shape. Also, the wake vortices will
influence the solid boundary condition and thus the aerodynamic properties.
First, the newly shed wake vortices are positioned on the path covered by the trailing edge during the
current time step. As discussed by Katz and Plotkin[32], the newly shedded wake vortices are placed
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
30 of 79

on the path covered by the trailing edge during the current timestep at 0.25 the distance covered by the
trailing edge in order to correct for wake-discretization errors (0.25 * U * t).
To calculate the strength of the wake vortices and the change in wake circulation the Kelvin condition is
used:
d(t) dW
d
=
+
=0
(5.11)
dt
dt
dt
where (t) is the airfoils circulation and W is the wakes circulation. For the first time step,
(t) + W1 = 0

(5.12)

Wi = [(ti ) (ti1 )]

(5.13)

and for the ith time step

When the wake gets further away than 10 Span, the vortices are discarded (it is assumed that the
effect of wake vortex that are after 10*span is negligible, see Sequeira, Willis and Peraire[46]).

System of Equations:
Define N as the total number of panels on the wing, Nc chordwise panels and Ns spanwise panels. At
each timestep, Ns new wake vortices are shed. The equation system to solve is always based on the
zero normal flow condition, which gives the strength of the vortices of the vortex lattice. But in the unsteady case, the normal velocity component is a combination of the self-induced velocity, the kinematic
velocity and the wake-induced velocity.
The self-induced velocity is represented by a combination of the influence coefficients exactly as in
the steady flow case. The kinematic velocity is known from the kinematic equations, which have to
be recomputed at each time step. The velocity induced by the most recent wake vortices, which have
just been shed at the trailing edge, is unknown, but their influence coefficients can be calculated. The
strength of these newly shed wake vortices can be calculated by adding the Kelvin condition as an
equation for each chorwise stripe of panels and the corresponding downstream newly shed wake vortex.

a11 a1N a1W1 a1WNS

RHS1
1
.
..
..
..
..

..
..
..
..
.
.

.
.
.
.
.

a
a

a
RHS
NN
N W1
N WNS
N
N1
N

PN c
=

(i, t t)
B
0
1

0 W 1
i=1

..
..
..
..
..
..
..

.
.
.
.
.
.
.
.
PN
~
WNS
0
B
0

1
i=N N c+1 (i, t t)

~ is a vector of Nc times 1.
where B

Vortex Wake Rollup

Since the wake is force-free, each vortex of the wake must move with the local stream velocity which
is a combination of the free stream flow velocity, the self-influence of the wake and the influence of the
wing on the wake:
~
~
~i = V
~f low + C
~ wingwake wing + C
~ wakewake wake
V
(5.14)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
31 of 79

Calculation of the forces and moments - unsteady form of the Kutta-Joukowsky theorem
The vortex lattice method calculates forces and moments directly from the vortex strength and the freestream velocity. For each panel, the force and the moment are calculated using the unsteady vector
form of the Kutta-Joukowsky theorem[40]
~ s +  c  V
~ s
f~lif t = V
t V~ 

m
~ lif t
where the normal speed is:

1  ~ 2 2
~
= ~r flif t + V  c cm s
2
~ (V
~ .
V~ = V
s)
s

(5.15)

(5.16)
(5.17)

and is the air density, the strength if the horse-shoe vector associated to the panel, c is the panel
chord, s the normal vector of the panel, cm its pitching moment and ~r the coordinate vector relatively
to the reference point.
This equations give the lift and moment for each panel. The total lift and moment of the whole wing are
calculated by summing the contributions of each panel.

Calculation the time dependant coefficients derivates

From the time-stepping method, on gets as result time dependent aerodynamic coefficients. The unsteady derivatives can be obtained using the linear theory approach here described for the lift coefficient:
CL (t) = CL0 + CL (t) + CLq q(t) + CL (t)

(5.18)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
32 of 79

5.4.3 Comparison of the two methods

The f12 is a windtunnel model of an airliner without control surfaces. This airplane was choosen during
the CEASIOM developement as an reference airplane. Experimental data is available from the windtunnel mesurements from DLR, and plenty of numerical data is available from the different members of
the SIMSAC project who compared their numerical codes using this exemple[47]. The case which was

Figure 5.4: The CEASIOM model of the f12 and the DLR windtunnel model of the f12
studied is a pitching motion of amplitude 4.52 and a frequency of 3 Hz. The angle of attack is 0 and the
speed is 70 m/s.

DLR exp. data

DLR Euler
DLR NS
ONERA NS
FOI NS
Tornado - T-step model
Tornado - Freq. model

CL
5.77
5.62
4.98
5.40
5.45
5.24
5.25

Cm
-2.25
-2.29
-2.09
-1.83
-2.11
-2.85
-2.99

CLq

Cmq

7.15
6.65
6.28
12.95
13.59

-19.76
-17.90
-17.86
-35.32
-38.71

CLq + CL
9.30
6.45
6.30
5.88
6.72
12.26
10.94

Cmq + Cm
-24.24
-25.10
-22.71
-22.44
-23.23
-33.45
-28.95

Table 5.2: Aerodynamic coefficients for the f12

The table shows that both Tornado implementations give similar results, and in comparison with the
experimental data the accuracy is acceptable. The difference between the Tornado calculations and
the Euler/NS calculations are due to the geometry and simpler mathematical model. The f12 wings are
very twisted which is difficult to modelis with only three kinks. Also, no fuselage model was used during
the Tornado calculations.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
33 of 79

Figure 5.5: The time dependant lifting coefficient for a pitching motion
The time dependent lift coefficients are in phase between the two models and have only a slightly higher
amplitude for the time stepping method. In the time stepping model lift curve, one sees that the wake
need some time to install which explains the anomaly at the beginning.

5.4.4 Conclusion
The unsteady version of Tornado was developed in the context of stability analysis in SDSA. SDSA uses
the linear perturbation method for the aerodynamics terms in the aircraft flight dynamics equations. For
the lift coefficient, we have then:

CL (t) = CL0 + CL (t) + CLq q(t) + CL (t)

(5.19)

The goal of the unsteady version is thus to calculate with a reasonable accuracy the unsteady derivates.
Thus, a complete time history of the evolution of the coefficients as calculated by the time stepping
model is not needed.
Finally, it was decided to integrated the frequency domain model into CEASIOM. While the time step
method gives much more information, like the evolution of all the coefficients in time and the strength
and position of the dynamic wake, the only desired results are the unsteady derivates which will be
passed to SDSA. Additionally, codes which are based on a time step method require much more computational time, which is an burden in the early design process, where CEASIOM will be used.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

5.5

D 3.2-5
1
34 of 79

Fuselage model

5.5.1 Cruciform fuselage

Figure 5.6: A flat plate modelisation of a fuselage

An easy way to model a fuselage is to replace it by two flat plates along vertical longitudinal section.
This allows to calculate the derivatives. Using this oversimplification of representing the fuselage as
two flat plates, the fuselage is considered as a wing of very low aspect ratio.

5.5.2 Slender body theory

A better representation of the fuselage is to use the slender body theory as used in AVL[39]. The main
assumption of slender body theory is that the diameter of a body is far less then its length, which is
usually the case for a aircraft fuselage. A serie of sources of strength (x) per length are set along the
axis.

Figure 5.7: Source distribution along the x axis

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
35 of 79

The individual source elements are defined:

~ (U
~ ~s)~s
~ (x) = 2U

(5.20)

The induced velocity by one source element at a point can be calculated using the dipole formula:
!
Z
~
~ (x)
U
R
~u =
A(s)
ds
(5.21)
4
R3
where with integrations by parts

~ ~ (x)
R
R3

~ ~ (x))R
~
(R
~ (x)

3
R3
R5

Figure 5.8: Cp on the B747 fuselage and flow lines

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

(5.22)

Deliverable
Version
Page

KTH

D 3.2-5
1
36 of 79

5.5.3 Comparison of the two methods and conclusion

A testcase is this wing-body-tail combination, where some experimental data exist in form of a NACA
report[57]:

3D wing and partition layout

3D wing and partition layout with fuselage

MAC

MAC

ref point
Aircraft body ycoordinate

Aircraft body ycoordinate

ref point
c.g.

2
0
2
4
6

c.g.

2
0
2
4
6

5
10
Aircraft body xcoordinate

15

6
8
10
12
Aircraft body xcoordinate

(a)

14

16

(b)

Figure 5.9: The geometry, left the fuselage is a projected plate, right the fuselage is modelised by a
serie of sources

Lift coefficient

Pitching moment coefficient

1.4
1.2
1

0.05
Standard Tornado+slender body theory
Standard Tornado+projected horizontal plate
Standard Tornado
experiments

0
0.05
0.1
0.15
Cm

CL

0.8
0.6

0.2
0.25

0.4

0.3
0.2
0.35
0
0.2
5

0.4
0

10

(a)

15

20

25

0.45
5

Standard Tornado+slender body theory

Standard Tornado+projected horizontal plate
Standard Tornado
experiments
0

10

15

20

25

(b)

Figure 5.10: Lift and pitching moment coefficient for the different configurations

Both fuselage models give a significant improvement for the pitching moment coefficient.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

5.6

D 3.2-5
1
37 of 79

Leading edge vortex correction

At high angles of attack, several phenomena can appear. First, the boundary layer can separate from
the airfoil which has as a consequence a drop of lift. When the sweep of the wing is very large, separation tends to occur near the leading edge, but unlike in the low sweep situation, the separated region
is not large and does not reduce the lift. Instead, the flow rolls up into a vortex that lies just above the
wing surface.

Figure 5.11: Separation of the flow at the leading edge of the Concorde wing
Rather then reducing the lift of the wing, the leading edge vortices increase the wing lift in a non-linear
manner. The vortex flow reduces the upper surface pressures by inducing higher velocities on the upper
surface.
The vortex lattice methods apply as boundary condition that the flow is tangential to the wing surface.
This is appropriate at the smallest angles of attack, but for increasing angles of attack the local flow
above the wing leading edge becomes a vortex sheet rolled up into a conical structure for highly swept
wings.
Ideally a theoretical methods for predicting aerodynamics at high angles of attack would be based on
Navier-Stokes or Euler equations with appropriate turbulence modelling. But for preliminary design
applications the use of some simplified approaches is quite adequate. Therefore, a practical approach
would be adding semi empirical corrections to the VLM for flow features for high angles of attack. The
empirical correction analysed here is the method of suction analogy[48, 49].

5.6.1 Leading edge vortices and their influence on low aspect ratio wing aerodynamics
The development of lift for highly swept wings at high angles of attack was analyzed by Polhamus[48].
The lift is built up of two parts, one is due to the attached potential flow handled with any vortex lattice
method, and the other is coming from the curved flow over the leading edge vortices which is leaving
the wing attached at the trailing edge. The potential flow gives a lift contribution CL,p , and the vortex lift
adds CL,v . The total lift is:
CL,total = CL,p + CL,v
(5.23)
The potential flow lift, denoted CL,p , is obtained from the calculations with the normal version of Tornado.
The calculation of the second part of the lift is based upon the observation of the similarity of the flow
patterns in potential flow around a round edge and separated flow around a sharp edge, see figure
5.12. This figure shows the flow in a cut normal to the leading edge. In the upper case it acts parallel to
the wing chord plane, in the lower normal to the plane.
Polhamus write the leading edge vortex lift coefficient, CL.v , as
CL,v = KV sin2 cos
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

(5.24)

Deliverable
Version
Page

KTH

D 3.2-5
1
38 of 79

Figure 5.12: Polhamus Leading Edge Suction Analogy[48]

KV = (KP KP2 Ki )/cos

(5.25)

where is the wing leading edge sweep angle, KP the potential flow lift curve slope and the constant
Ki the induced drag factor defined as:
CD
(5.26)
Ki =
CL2
Comparing Tornado and Polhamus data for KP and KV , Hedman finds good agreement[49]. Polhamus
also treats the calculation of drag due to lift[48]. On a flat thin wing in inviscid flow there should be no
tangential force, but only a normal force:
CD,total = CL,total tan

(5.27)

5.6.2 Comparison with experimental data for Delta wings

In a first step, delta wings are considered, because there exist plenty of experimental data and the
vortex lift is significant. The following cases were studied:
Case
chord (c)
span (b)
aspect ratio (AR)
leading edge sweep angle (SW)
quarter chord sweep angle (1/4SW)

1
1
0.25
0.5
82.88
80.54

2
1
0.5
1.0
75.96
71.57

3
1
0.75
1.5
69.44
63.44

4
1
1.0
2.0
63.43
56.31

5
1
1.25
2.5
57.99
50.19

6
1
1.5
3.0
53.13
45

Table 5.3: Test cases

One sees that the leading edge vortex suction analogy gives good results for sharp delta wing with
aspect ratio up to AR = 2.5. For higher aspect ratios, and sweep angles over 60, the attached potential
lift is more close to the experimental values. Also at high angles of attack, over 30, the upper suface
flow may no longer be abler to reattach before or at the trailing edge, and the prediction becomes less
accurate.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
39 of 79

Figure 5.13: The vortex, potential and total lift for delta wings of different aspect ratio (1)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
40 of 79

Figure 5.14: The vortex, potential and total lift for delta wings of different aspect ratio (2)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
41 of 79

5.6.3 Conclusion
The Polhamus suction analogy predicts with a reasonable accuracy the lift coefficients for simple configurations like Delta wings. Even for Delta wings, it overestimates the lifting coefficient at high aspect
ratio. It is an empirical method, where the results of the application are rather good when the calculated
configuration is close to the wings whose experimental data were used in its development. It is difficult
to rely on the results of this approach for unconventional configuration.
The Polhamus suction analogy also does not take account of vortex breakdown, although there exist some empirical corrections for that[51]. Another problem is that the spiral vortex sheet has also an
influence on the pitching moment coefficient. The Polhamus surction analogy works on the coefficient
level and therefore has no idea about pressure distribution on the wing. Thus, an accurate calculation
of the pitching moment coefficient is impossible for cases where there is significant vortex lift.

Figure 5.15: Lift for a Delta wing of aspect ratio 1

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

5.7

D 3.2-5
1
42 of 79

Viscous corrections

The total drag on a lifting surface can be expressed as following:

CD = CD0 + CDl + CDm

(5.28)

There exist three forms of Drag:

Zero-lift drag (CD0 ) arise from skin friction.
Drag due to lift (CDl ): the drag due to lift is at the moment the only drag considered.
Wave drag (CDm ): wave drag is an effect of shocks and only occurs at high Mach numbers, and
thus is not considered at the moment.

5.7.1 Torfoil
XFOIL is an interactive program for the design and analysis of subsonic isolated airfoils developed by
Mark Drela at the MIT in the eighties[54]. It allows viscous and inviscid analysis of an existing 2D airfoil,
featuring forced and free transition, transitional separation bubble(s), limited trailing edge separation, lift
and drag predictions just beyond CLmax and Karman-Tsien compressibility correction.
XFOIL is run for each one of the wing partitions in viscous and in non-viscous flow for the section
loads obtained by Tornado. The angle of attack differences required to match the lift coefficient are
taken as the influence of the viscosity in the flow. The real wing is then twisted with these angles, and
Tornado is run in non-viscous flow mode to simulate approximately the viscous flow around the real
wing. In the implementation in Tornado, the TOR-FOIL concept, wings are run in Tornado for three
dimensional non-viscous flow and in XFOIL for two dimensional viscous/non-viscous flow.
The main steps in TOR-FOIL are
1. Run Tornado for the nominal wing geometry at nominal to define the section lift coefficients in
non-viscous flow, (CL )invisc .
2. Run XFOIL for each section at its (CL )invisc in non-viscous flow to get the invisc
3. Run XFOIL for each section at its (CL )invisc in viscous flow to get the visc
4. The influence of viscosity for the wing at nominal is defined for each section as = invics visc .
5. To compensate for the loss in lift in viscous flow, the section twist is reduced with . Rerun
Tornado to get the aerodynamic coefficients for the simulation of the wing in viscous flow.
Sven Hedman has made several tests with the TOR-FOIL concept[55]. His conclusion was that the lift
and stall estimations are good, but that drag is not so good. In fact, drag seems only be accurate for
straight, slender wings with small amounts of sweep.

5.7.2 Component Buildup Method

In this chapter, an empirical method is presented to estimate the zero-lift drag (see Raymer[52]). The
component buildup method estimates the subsonic zero-lift drag of each component of the aircraft using a calculated flat-plate skin-friction drag coefficient (Cf ) and a component form factor (FF) that

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
43 of 79

estimates the pressure drag due to viscous seperation. Then, the interference effects on the component drag are estimated as a factor Q and the total component drag is determined as the product of
the wetted area, Cf , FF and Q.
The flat-plate skin friction coefficient Cf depends upon the Reynolds number, the Mach number and
skin roughness.
The laminar friction coefficient is expressed as follows:
1.328
Cf,l =
Re

(5.29)

For turbulent flows, the flat-plate skin friction coefficient is (Eckerts equation[53])
Cf,t =

(log10

0.455
+ 0.144M 2 )0.65

Re)2.58 (1

(5.30)

where Re is the Reynolds number, with as a characteristic length a distance related to the component
(fuselage total length, mean aerodynamic chord length for the wing). If the surface is relatively rough,
the friction coefficient will be higher. This is accounted for by using a cut-off Reynolds number:
Recutof f = 38.21(l/k)1.053

(5.31)

where l is the characteristic length and k a surface roughness value (around 0.634 105 for smooth
paint). The lower of the actual Reynolds number and the cut-off Reynolds number should be used for
calculating the skin friction coefficients.
The transition from laminar to turbulent flow is accounted by fixing a hardcore transition point (t = 10 %
as exemple):
(5.32)
Cf = (1 t) Cf,t + t Cf,l
The form factor is a function of the Mach number, the chordwise location of the airfoil maximum thickness point (x/c)m , the sweep of the maximum thickness line m and the thickness to chord ratio
t/c:

 


0.6
t
t 4 
FF = 1 +
+ 100( )
1.34M 0.18 (cos m )0.28
(5.33)
(x/c)m c
c
Finally, the complete zero lift drag is
P
(Cf,component F Fcomponent Qcomponent Swet,component)
CD0 =
Sref

(5.34)

The main difficulty of this method is to calculate the wetted areas. Like all the empirical methods, its
accuracy can be questioned and needs to be tested intensively.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
44 of 79

5.7.3 Comparison of the two viscosity correction models

A long list of models of different wing and wing-body configurations were used to test and compare
the two viscosity corrections[64]. In the following pages, the lift coefficient (including stall model) and
the drag polar are plotted for these models. Tornado B is the basic version of Tornado, Tornado V the
Tornado version with the component buildup method and stall model and Torfoil is the Tornado version
with XFOIL.
AR

c/4 [deg]

NACA airfoil

Re106

1 (a, b)

9.02

0.4

64210 (a) /65210 (b)

4.4

[57]

37.5

0.49

23012

1.1

[57]

10

5.2

30

0015

0.91

[57]

12

14.04

0.33

23012

0.87

[57]
[57]

13
14

6
6

6
0

0.2
1

23012
0012

0.9
1.0

[57]

21

5.2

-30

0015

0.91

[59]

45

0.6

2-006

2.01

[60]

39.45

0.4

64A006

9.7

Ref.

Model

[56]

Skethches

Table 5.4: Test cases

Models
1a
1b
8
10
12
13
14
21

Ref.
[56]
[57]
[57]
[57]
[57]
[57]
[57]

CD,0 (exp.)
0.006
0.008
0.020
0.020
0.020
0.010
0.012
0.020

CD,0 (wetted area)

0.0086775
0.00833
0.0087960
0.0099300
0.0102800
0.010300
0.012100
0.009929

errors(%)
-45%
4%
56%
50%
49%
-3%
-0.8%
50%

CD,0 (TorFoil)
0.005895
0.007870
0.011800
0.009360
0.009368
0.006500
0.011800

errors(%)
2%
61%
41%
53%
6%
45.83%
41.00%

Table 5.5: Zero-lift drag coefficients from wetted area (Raymers theory[52] and TorFoil[55], compared
with experimental data([56],[57]) (minus sign means overpredicted)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Lift coefficient

D 3.2-5
1
45 of 79

Drag coefficient

1.6
1.6
1.4
1.4
1.2
1.2
1

1
0.8
CL

CL

0.8
0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

15

20

25

0.05

0.1

0.15
C

0.2

0.25

0.3

(a)

(b)

Figure 5.16: (a) Model 1a, wing alone

Drag coefficient
1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient
1.6

0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

15

20

0.05

0.1

0.15

0.2

(a)

(b)

Figure 5.17: (b) Model 1b, wing alone

Drag coefficient
1.4

1.2

1.2

0.8

0.8

0.6

0.6

CL

CL

Lift coefficient
1.4

0.4

0.4

0.2

0.2

0.2

TornadoV
Tornado

experiments
TorFoil

0.2
5

10

15

20

25

0.05

0.1

0.15
C

0.2

0.25

0.3

(a)

(b)

Figure 5.18: (c) Model 18

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Drag coefficient

1.6

1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient

0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

D 3.2-5
1
46 of 79

15

20

25

0.05

0.1

0.15
C

0.2

0.25

0.3

0.2

0.25

0.3

(a)

(b)

Figure 5.19: (d) Model 10

Drag coefficient
1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient
1.6

0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

15

20

25

0.05

0.1

0.15
C
D

(a)

(b)

Figure 5.20: (e) Model 12

Drag coefficient
1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient
1.6

0.6

0.6

0.4

0.4

0.2

0.2

0.2

TornadoV
Tornado

experiments
TorFoil

0.2
5

10

15

20

25

0.05

0.1

0.15
C

0.2

0.25

0.3

(a)

(b)

Figure 5.21: (f) Model 13

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Drag coefficient

1.6

1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient

0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

D 3.2-5
1
47 of 79

15

20

25

0.05

0.1

0.15
C

0.2

0.25

0.3

0.2

0.25

0.3

(a)

(b)

Figure 5.22: (g) Model 14

Drag coefficient
1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient
1.6

0.6

0.6

0.4

0.4

0.2

0.2

0.2

0.2
5

10

15

20

25

0.05

0.1

0.15
C
D

(a)

(b)

Figure 5.23: (h) Model 21

Drag coefficient
1.6

1.4

1.4

1.2

1.2

0.8

0.8
CL

CL

Lift coefficient
1.6

0.6

0.6

0.4

0.4

0.2

0.2

0.2

TornadoV
Tornado

experiments
TorFoil

0.2
5

10

15

20

25

30

0.05

0.1

0.15
C

0.2

0.25

0.3

(a)

(b)

Figure 5.24: (i) Model A

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
48 of 79

5.7.4 Conclusion
The zero-lift drag coefficient predictions using Raymers wetted area method show an equivalent (sometimes even better) accuracy compared with the values from TorFoil(Table 5.7.3). The main problem for
Torfoil is that they need a 2D airfoil as input, while in Tornado 3D lifting surfaces are used. For swept
or unconventional wings Torfoil give unreliable results. For most of the cases, Tornado V can give a
relatively good prediction compared with the experimental data. It can be seen that the preditions for
lift and drag coefficient in Tornado V and TorFoil are equally good for models 1 (TorFoil is not implemented in model1b), 12 and 13 (Figure 5.24(a), 5.20(a) and 5.21(a)), which are straight and slightly
swept wings. Note that although Tornado V doesnt predict well for models 8, 21 and the zero-lift drag
coefficients for models 10 and 12, TorFoil predicts also not well or even worse (Figure 5.18(a), 5.19(a),
5.20(a) and 5.23(a)). It is quite clear that the much complicated TorFoil method can not give much better
predictions than using the much simpler Tornado V for viscous corrections. Noting that neither TorFoil
nor Tornado V can not be used correctly for low aspect ratio and high sweep wings, which should take
into account the leading edge vortex corrections.

5.8

Stall model

A stall is a condition in aerodynamics and aviation where the angle of attack increases beyond a certain
point such that the lift begins to decrease. The angle at which this occurs is called the critical angle
of attack. This critical angle is dependent upon the profile of the wing, its planform, its aspect ratio,
and other factors, but is typically in the range of 8 to 20 degrees relative to the incoming wind for
most subsonic airfoils. The critical angle of attack is the angle of attack on the lift coefficient versus
angle-of-attack curve at which the maximum lift coefficient occurs.
Flow separation begins to occur at small angles of attack while attached flow over the wing is still
dominant. As angle of attack increases, the separated regions on the top of the wing increase in size
and hinder the wings ability to create lift. At the critical angle of attack, separated flow is so dominant
that further increases in angle of attack produce less lift and vastly more drag.
Linear method like a vortex lattice method can not simulate this phenomen. In this chapter, an empirical
method is presented and tested.

5.8.1 Empirical Stall model

An empirical, quasi-analytical method to find the maximum lift and the angle of attack for stall has been
introduced by Askin T. Isikveren in his doctoral thesis[65].

1. Two points, = 0 and = 4 are suggested, are computed by a standard vortex lattice method.
L
2. Next, the zero-lift angle of attack (0L ) is found by extrapolating the lift curve slope ( dC
d ) back to
the point at which CL = 0.

3. Step 3 involves an angle of attack increment of = 10 to yield an estimation of the cessation of

the linear portion of the curve (usually around = 8), which is the beginning of the non-linear lift
leading to stall.
4. The maximum lift can now be obtained by adding 4 times the lift curve slope to the lift found at
step 3.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
49 of 79

Figure 5.25: An empirical method for finding the lift characteristics

5. The angle of attack for stall can now be estimated as follows:
stall = 0L +

73 2AR
3

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

(5.35)

Deliverable
Version
Page

KTH

D 3.2-5
1
50 of 79

For testing this empirical stall mode, the same list of models of different wing and wing-body configurations was used then for the viscosity corrections:
AR

c/4 [deg]

NACA airfoil

Re106

1 (a, b)

9.02

0.4

64210 (a) /65210 (b)

4.4

[57]

37.5

0.49

23012

1.1

[57]

10

5.2

30

0015

0.91

[57]

12

14.04

0.33

23012

0.87

[57]
[57]

13
14

6
6

6
0

0.2
1

23012
0012

0.9
1.0

[57]

21

5.2

-30

0015

0.91

[59]

45

0.6

2-006

2.01

[60]

39.45

0.4

64A006

9.7

Ref.

Model

[56]

Skethches

Table 5.6: Test cases

Models
1a
1b
8
10
12
13
14
21
A
B

CLmax (exp.)
1.36
1.20
1.12
1.05
1.16
1.09
0.97
1.03
1.125
0.92

CLmax (pred.)
1.228
1.1997
0.840
0.953
1.098
1.10
1.08
0.9456
0.7911
0.7592

errors(%)
7.0%
0.025%
25.0%
9.2%
5.3%
-0.9%
-11.3%
8.2%
29.7%
17.5%

stall [deg] (exp.)

15.5
14
22
18.5
16.2
16.3
14.3
22.5
28.9
22.7

stall [deg] (pred.)

16.6
16.18
21.1
20.86
19.27
19.1
20.4
20.5
21.63
22.32

errors(%)
-9.7%
-15.6%
4.1%
-12.8%
-19.0%
-17.2%
-42.7%
8.9%
25.2%
1.7%

Table 5.7: Stall condition prediction compared with experimental data (minus sign means overpredicted)

5.8.2 Conclusion
The approximations of maximum lift coefficient for all the models except models 8, 14, A and B are good,
with errors within 10%. But the stall angles of attack are not well predicted for most of the models since
the method is only approximate. We can consider the errors within 10% or even 20% as acceptable.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

5.9

D 3.2-5
1
51 of 79

Final version of the vortex lattice method Tornado in CEASIOM

Table 9.1 compares the main extensions in Tornado with the Shaman and Vorstab codes.

Compressibility
Viscosity
Unsteady
Fuselage
Leading edge vortex
Vortex breakdown
Stall model

Tornado
Prandtl-Glauert correction
Empirical correction
Frequency domain model
Slender body/cruciform
Not included, use Edge
Not included, use Edge
Simple stall model

Shaman

Time domain model

Vortex cloud method
Polhamus suction analogy
Empirical corrections
Empirical corrections

Vorstab
Prandtl-Glauert equation
Boundary layer model
not available
Vortex multiplets
Polhamus suction analogy
Empirical corrections
Empirical corrections

Table 5.8: Tornado features compared with two more complex vortex lattice methods
Both the Shaman code and the Vorstab code were built over years. They are full of complex empirical
extension to take into account difficult flow features like vortex fields coming from the fuselage or vortex
breakdowns.
In CEASIOM, Tornado will not be as complex as Vorstab or Shaman. Some extensions like the leading
edge suction analogy were analyzed, but discarded. It is difficult to implement empirical corrections and
to guarante the accuracy of them for unconventional configuration.
CEASIOM goes a different way: Adaptive-Fidelity CFD. In the CEASIOM suite, the computational models goes from handbook methods (USAF Digital DATCOM [12]), through linear singularity methods
(Vortex Lattice Method [13], Panel Methods such as dwfSolve [15] ) to full non-linear Euler and RANS
compressible flow CFD packages ([22]). The user is advised to move from a Tier I tool to a more
complex Tier 1+ tool for more accurate results.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
52 of 79

6 Tier I+: Non-linear flow physics - The Euler solver Edge

Edge is a parallelized CFD flow solver system for solving 2D/3D viscous/inviscid, compressible flow
problems on unstructured grids with arbitrary elements[66]. Edge can be used for both steady state and
time accurate calculations including manoeuvres and aeroelastic simulation. There is also an inviscid
adjoint flow solver that can be used for gradient based shape optimization. There is also functionality
for mesh deformation, postprocessing, file format conversion and other utility tasks.
The flow solver employs an edge-based formulation which uses a node-centered finitevolume technique
to solve the governing equations. The control volumes are non-overlapping and are formed by a dual
grid, which is computed from the control surfaces for each edge of the primary input mesh. The relationship between the dual and primary input mesh is illustrated in Figure 6.1. In this example, a set of
hexagonal control volumes are constructed from a simple triangular input mesh. In any Edge mesh, all
the mesh elements are connected through matching faces. Edge meshes therefore may not contain
hanging nodes.

Figure 6.1: The input grid (solid) also denoted triangular grid and its dual grid (dashed) forming the
control volumes.
In the flow solver, the governing equations are integrated explicitly towards steady state with RungeKutta
time integration. Convergence is accelerated using agglomeration multigrid and implicit residual smoothing. Time accurate computations can be performed using a semi implicit, dual time stepping scheme
which exploits convergence acceleration technique via a steady state form inner iteration procedure.
Every program in the Edge system can be used via a command-line interface, which is implemented as
set of simple Unix shell scripts. However, there is also a platform-independent graphical user interface
(GUI) written in Java, which allows configuring a complete Edge computation with on-line help.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
53 of 79

A Matlab interface was written in order to allow Edge calculations to be prepared and run. This call
runs the preprocessing routines, launches the calculation and processes the solution for the forces and
moments[67].
The control surfaces deflections are not calculated by mesh deflection, but they are simulated by the
transpiration boundary condition. The transpiration boundary conditions in Edge are implemented by
prescribing a wall normal velocity component depending on how much the wetted surface is deflected
or deformed. This feature is implemented in the framework of the aeroelastic functionalities in Edge
and it can currently be used for unsteady calculations only. To deflect one or several control surfaces by
using transpiration boundary conditions to avoid mesh deformation, the user has to select the surfaces,
deflect the surface and run Edge in an unsteady manner. This is done by using a sequence program in
Edge.
The quasi-steady derivatives can be calculated by simulating an unsteady oscillation. In Edge this is
done by prescribing an unsteady motion and in particular a motion under oscillation with the purpose to
compute the static and dynamic derivative of the motion. This feature is implemented in the framework
of the aeroelastic functionalities in Edge, the procedure is similar to what has been described for deflecting control surfaces. This is done by using a sequence of programs in Edge. The basic principle for
prescribing an oscillating motion (with limited amplitude) is to deflect the entire grid and create two perturbed grids corresponding to the plus/minus maximum deflection. Edge will then interpolate between
these two grids. As a result of this procedure, time dependent data is obtained and the coefficients
(forces and moments). The quasi-steady derivatives can be calculated out of this time dependant data
using Fourier series.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
54 of 79

7 Benchmarking

7.1

f12

The f12 is a windtunnel model of an airliner without control surfaces. This airplane was choosen during
the CEASIOM developement as an reference airplane. Experimental data is available from the windtunnel mesurements from DLR, and plenty of numerical data is available from the different members of
the SIMSAC project who compared their numerical codes using this exemple[47].

Figure 7.1: The CEASIOM model of the f12 and the DLR windtunnel model of the f12
The comparison between the CAD model and the CAD file generated from the XML file with SUMO one
sees that for the general dimensions they agree good. Small differences can be seen at the fuselage,
and especially with the dihedral, twist, thickness and airfoils of the wing. With the simplified XML
geometry it is impossible to modelise exactly a complicated wing like the f12 wing.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
55 of 79

(a)

(b)

(c)

Figure 7.2: CAD model of the f12 (red), and the simplified SUMO model of the f12 (blue)

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
56 of 79

The case which was studied is a simple sweep of angle of attack at a speed of 70 m/s.
Lift coefficient

Drag coefficient

1.4

1.4
Tornado (FWVH)

1.2

1.2

AVL (FWVH)
Datcom (FWVH)
Edge (FWVH)
experiments (FWVH)

1
0.8

1
0.8
0.6
CL

CL

0.6
0.4

0.4

0.2

0.2

0.2

0.2

0.4

0.4

0.6
6

2
4
(deg)

10

0.6

12

TornadoV
Tornado

AVL
Datcom
Edge
experiments
0

0.02

0.04

0.06
C

0.08

0.1

0.12

(a)

(b)
Pitching moment coefficient
0.8
0.6
0.4

Cm

0.2
0
0.2
Tornado (FWVH)
Tornado (WVH)
Datcom (FWVH)
Edge (FWVH)
experiments (FWVH)

0.4
0.6
0.8
6

2
4
(deg)

10

12

(c)

Figure 7.3: Lift, drag and pitching moment coefficient for the f12

The Tornado calculated lift and pitching moment coefficient are slightly offset compared with the experimental data because the geometry is too complicated to be represented by the XML file, especially the
very twisted wings.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

7.2

D 3.2-5
1
57 of 79

X-31

The X-31 is an experimental aircraft designed by MBB (now part of EADS) and Rockwall (now part of
Boing). It was designed to test fighter thrust vectoring technology and to test flights at high angle of
attacks, therefore the program was called Enhanced Fighter Maneuverability. The X-31 is configured
with movable forward canards, wing control surfaces and fixed after strakes. It operate with a digital
fly-by-wire flight control system needed because the X-31 is a natural unstable airplane.

(a)

(b)

Figure 7.4: X-31 aircraft (left), and the CEASIOM model of the X-31 (right)

Several paths were investigated. Unluckily, DATCOM does not allow Canard-Wing configurations. As
the XML fuselage is limited and it is difficult to describe the fuselage of the X-31, a second model of the
X-31 was done where the fuselage was recreated in Sumo[14]. The original CAD geometry is called
X31 va, the modfied X31 vb.

Figure 7.5: Paths

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
58 of 79

The Tornado results and the Edge results of the two version were compared with experimental values[68].
The case studied is a simple alpha sweep at Mach number = 0.06.

(a)

(b)

(c)

Figure 7.6: Lift coefficient, drag coefficient and pitching moment coefficient

The version of Tornado used in this example is without fuselage correction and without viscosity correction. Tornado gives reasonable results only at low angle of attack. As this version of Tornado, and
the Euler code Edge, do not take into account the viscosity, the zero lift drag is zero. Remodeling the
fuselage in Sumo improves the coefficients by an average of 4%.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

7.3

D 3.2-5
1
59 of 79

Ranger 2000

The Ranger 2000 aircraft, Fig 7.7, is a mid-wing, tandem seat military training aircraft with a turbofan
engine. The wing and fuselage are manufactured of composite material and the empennage is a metal
T-tail design. The control surfaces are manually operated elevator and rudder, hydraulically assisted
ailerons, a belly mounted speed-brake, and electrically operated split flaps.

Figure 7.7: Ranger 2000 Military Training Aircraft

Complete aerodynamic data from wind-tunnel tests, handling quality flight tests, and CAD geometry for
the Ranger 2000 have been made available to SimSAC by EADS-M, Dr. S.Hitzel[69].

Figure 7.8: Ranger 2000 geometry renderings and computational grids. Left to right: xml-geometry
XML, SUMO geometry SUMO, and CAD geometry CAD

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
60 of 79

It is interesting to compare the CAD model and the simplified model generated from SUMO, see ??.

(a)

(b)

(c)

(d)

Figure 7.9: CAD model used for the HISSS and NSMB calculations (red), and the simplified SUMO
model which was used for the EDGE calculations (blue)

The aerodynamic coefficients were calculated using the different Tier I and Tier I+ methods, DATCOM, Tornado and Edge. In addition, some Euler and Navier-Stokes calculations were done by
the NSMB solver[71] for comparison. These results were compared to the windtunnel[69] mesurement and HISSS[70] calculations, which have kindly been made available to SimSAC by EADS-M, Dr.
S.Hitzel[69].

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

(a)

(b)

(c)

(d)

(e)

(f)

D 3.2-5
1
61 of 79

Figure 7.10: Aerodynamic coefficients for Mach = 0.5

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

(a)

(b)

(c)

(d)

(e)

(f)

D 3.2-5
1
62 of 79

Figure 7.11: Aerodynamic coefficients for Mach = 0.75

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
63 of 79

Some comments about the coefficients:

DATCOM doesnt estimate accurately the lift curve slope and the pitching motion curve slope,
which explains the inaccuracy in predicting the longitudinal eigenmodes (see Stability Analysis
section).
TORNADO and HISSS results differ for some coefficients, this is due to:
TORNADO is run without fuselage model
HISSS predicts better transonic effects
HISSS uses the real CAD geometry, TORNADO only a simplified XML geometry
TORNADO is run without viscosity correction
These differences explain why HISSS is better in predicting the eigenmodes.
The coefficients calculated with EDGE show a significant improvement in comparison with panel
methods or handbook methods, and thus also eigenmodes (see Stability Analysis section).
With the simplified geometry one gets reasonably good results (EDGE) in comparison with CAD
geometry results (NSMB, HISSS). Hence, geometry difference has only a minor effect.
One can separate the different methods in two groups:
TORNADO and DATCOM, whose results are not so good
HISSS, Euler and NS, whose results are much better
Higher Mach numbers rule out linear methods.
The eigenvalues are calculated for the linearized model, which is linearized around the trim. The
Ranger 2000 is trimmed at an angle of attack between -1 and 3, depending on the altitude and
the speed. Thus, one only needs to consider the coefficients between these values. Hence the
larger differences between the datasets at higher alpha will not enter into the stability analysis in
the next section.
The Engine intake is closed in the NSMB calculations, which explains part of the higher drag for
the NS results.
The largest differences at low angles of attack are
for the lift coefficient: DATCOM is the only method which does not calculate accurate the lift
curve slope.
for the drag coefficient: Both TORNADO and EDGE calculations do not take into account the
viscosity and have a zero-lift drag of 0.
for the pitching moment coefficient: All the methods are slightly offset, but most method
predict good the slope.
for the Mach 0.75 calculations: At Mach 0.75, transonic effects play a large role. TORNADO
is using the Prandtl-Glauert correction to estimate the compressibility effects, which is not
accurate when shocks are encountered.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
64 of 79

An interesting question is how strongly the stability depends on aerodynamic data. Even big differences
in aerodynamic data may fail to cause significant differences in S&C results[9]. The section presents
results of stability analysis for five different aero data sets.
1. The first and reference aero data set, is obtained from wind tunnel[69] investigations. The results
were presented in the previous chapter.
2. The high order panel method HISSS[70] provides the second aero data set.
3. The vortex lattice method TORNADO is the third dataset. TORNADO[13] is embedded in the
CEASIOM environment and was tested in comparison with the high order panel method HISSS.
4. The DATCOM, embeded in the CEASIOM, is the source of the fourth aero data set.
5. A fifth dataset is a combined Euler and DATCOM set. The dynamic derivatives were calculated
using DATCOM, while the rest of the coefficients were calculated by EDGE[22]. The Euler calculations were done on the SUMO geometry.
Full stability and maneuvers analysis give many results. The selected graphs allow to compare and
asses the results. Figures 7.12 and 7.14(a) present Short period frequency vs. calibrated airspeed obtained for five data sets: WT - Wind Tunnel, HISSS - high order panel method, TORNADO - embedded
VLM, DATCOM and EDGE.

(a)

(b)

Figure 7.12: Short Period frequency vs calibrated airspeed - results of SDSA for different aero data
sets.
Figures 7.13 and 7.14(b) present other Short Period characteristic - Control Anticipation Parameter,
which characterizes handling qualities well. The differences between results for all aero data sets
except DATCOM are very small.
Figures 7.15 and 7.16(a) present the phugoid period. In this case results are not so good and values
obtained with DATCOM aero data set differs significantly. The difference between values of lift curve
slope and drag prediction is the cause, because lift is the generator of phugoid vibration and drag is the
damper. Figure 7.16(b) presents Phugoid characteristics against a background of ICAO recommendations for aerodynamic data from HISSS[70] and EDGE[22].
Figures 7.17 and 7.18(a) present the results for the Dutch Roll mode - undamped natural frequency
and damping ratio. Both values have very close values for all data sets. It is interesting, that the results are so close, despite significant differences between values of key stability derivatives for Dutch
Roll - Cl/ and Cn/. Also lateral damping derivatives are different for all data sets. Generally
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

(a)

D 3.2-5
1
65 of 79

(b)

Figure 7.13: Short Period CAP evaluation (undamped natural frequency vs. nz / derivative) - results
of SDSA for different aero data sets.

(a)

(b)

Figure 7.14: Short Period frequency vs. calibrated airspeed (a) and CAP evaluation (b) - SDSA results
for HISSS and EDGE data

(a)

(b)

Figure 7.15: Phugoid period vs calibrated airspeed - results of SDSA for different aero data sets.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

(a)

D 3.2-5
1
66 of 79

(b)

Figure 7.16: Phugoid period vs. calibrated airspeed (a) and ICAO figure of merit to classify the phugoid
characteristics (b) - SDSA results for HISSS and EDGE data
lateral stability depends strongly on these values, however sometimes other values (i.e. other characteristics, mass and inertia) cause, that stability analysis is not sensitive on these derivatives. Figure
7.18(b) presents Dutch Roll characteristics against a background of MIL[73] criteria characteristics for
aerodynamic data from HISSS[70] and EDGE[22].

(a)

(b)

Figure 7.17: Dutch Roll frequency vs calibrated airspeed - results of SDSA for different aero data sets.
The obtained results allow to asses computational methods of aerodynamics from stability analysis point
of view. Generally the best results for all modes of motion gives aerodynamic data from TORNADO[13].
It can seem strange, that the simplest flow model gives the best results, but this feature of VLM is known
(Margason[72]) - method is well conformable with experiment, because neglects effects of thickness
and viscosity concurrently. Viscosity effect compensates the thickness of lift surface, in most cases. All
these things cause, that VLM, despite its simplicity, is very attractive tool to prepare aerodynamic data
for stability analysis.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

(a)

D 3.2-5
1
67 of 79

(b)

Figure 7.18: Dutch Roll frequency vs. calibrated airspeed (a) and MIL figure of merit to classify the
Dutch Roll characteristics (b) - SDSA results for HISSS and EDGE data

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
68 of 79

8 CEASIOM in design

8.1

TCR

The TransCruiser (TCR) is a baseline design proposed by Saab for a civil passenger airplane flying in
transonic regime. It is an excellent case study for CEASIOM, since little literature on such an unconventional configuration can be found, and traditional handbook methods fail to predict the flight characteristics with enough accuracy. The design cruise speed is MD = 0.97 at greater or equal altitude to
37,000 ft. The nominal payload is 200 passengers, and the approximate take-off weight is 220,000 kg.
The baseline configuration has been developed by Saab using in-house methods. It is a conventional
mid-to-low-winged T-tail configuration with two wing mounted engines. Ailerons and rudder are used
together with an all-moving horizontal tail for control. Flaps and slats are used as high-lift devices. The
landing gear is a conventional tri-cycle type where the main gears are mounted in the wing. Some
preliminary sketches are shown in Figure 8.1.
The aerodynamic coefficients in low speed have been calculated using the Tier I method Tornado (VLM).
Using the aerodata obtained, the stability & control module SDSA calculates the trim characteristics.
The CEASIOM trim analysis confirms the guess of the handbook methods (Figures 8.2 & 8.3). Considering a velocity of 160 m/s at sea level (M 0.47), it seems clear that, although the angle of attack
is not too big (around 3.5 degrees), a tailplane deflection of 11 degrees is totally unacceptable. Since
the trim characteristics are so poor there is no use in designing a control system for this configuration.
Plus, the aeroelastic module NeoCASS predicts that the T-tail will flutter in transonic speed.
The poor trim characteristics of the baseline configuration together with the predicted flutter of the T-tail
call for a re-design of the TransCruiser. An alternative all-moving canard configuration was proposed,
and Saab requested to further investigate it by varying the longitudinal position of the canard wing, the
longitudinal position of the main wing and the canard area, while keeping the main wing shape. The
goal is to find a configuration with better trim characteristics than the original one, but without making
the aircraft unstable (i.e. the static margin must be positive).
The table below shows the angle of attack and canard deflection required to trim four different canard
configurations, at a velocity of 160 m/s and sea level.
Two of these configurations (TCR-C2 and TCR-C15) have the canard wing located in the fuselage (in
the foremost position, in order to maximize the arm of the moment produced by the canard around the

Figure 8.1: Spreadsheet sketch (left) and CAD sketch (right) of the TCR.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
69 of 79

Figure 8.2: TCR angle of attack required for trim, for different airspeeds and altitudes.

Figure 8.3: TCR horizontal tail deflection required for trim, for different airspeeds and altitudes.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
70 of 79

Figure 8.4: V-g plot of TCR T-tail configuration.

TCR-C2
TCR-C8
TCR-C15
TCR-C17

xW (m)
0.26
0.27
0.26
0.26

xC (m)
0.13
0.017
0.12
0.0017

SC (m2 )
65
47
72
65

(deg)
2.4
2.2
3.2
2.2

(deg)
6.6
6.6
5.9
4.2

Static margin (%)

4.42
4.64
3.12
-2.65

Table 8.1: Angle of attack, canard deflection and static margin for trim at 160 m/s and sea level, for four
different canard configurations of the TCR.
centre of gravity) and in the other two (TCR-C8 and TCR-C17) the canard is placed in the nose cone.
The latter option has the inconvinience of the canard possibly interfering with several systems housed
in the nose cone, such as the weather radar. Besides, it is not a very aesthetic configuration. However,
the former option means that the doors cannot be located in the fore fuselage. A solution might be to
place them in the mid-fuselage.
Some general conclusions can be drawn from Table 8.1. When increasing the moment arm (i.e. the
distance between the canard aerodynamic centre and the centre of gravity of the aircraft), the canard
needs to produce a smaller lift force to obtain the same moment around the centre of gravity, so a
smaller deflection is required. However, the aerodynamic centre of the whole aircraft is being placed
closer to the centre of gravity. Thus, the aircraft will become less stable. This can be observed comparing the TCR-C2 and the TCR-C17. Featuring a larger distance between the canard and the main wing,
the TCR-C17 requires a smaller canard deflection for trim, but it is also less stable than the TCR-C2 (in
fact, it becomes unstable).
The effect of increasing the canard area is similar. A larger canard area means that, to obtain the
canard lift needed to trim the aircraft, a smaller canard deflection is required. However, a larger canard
area also moves the aerodynamic centre of the aircraft further fore, resulting in less stability. This is
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
71 of 79

appreciated when comparing the TCR-C8 and the TCR-C17. The latter one has a larger canard area,
so the canard deflection for trim is smaller, but the aircraft is more unstable.
To summarise, increasing the canard area or the distance between canard and main wing results in a
smaller canard deflection for trim, but also in a less stable aircraft. Therefore, a compromise between
good trim characteristics and acceptable stability must be reached. For that reason, the TCR-C15 was
chosen as the most appropiate configuration. It considerably reduces the trim deflection with respect to
the original T-tail configuration (around 50%), and the aircraft is slightly stable. The aerodynamic centre
is expected to be shifted furhter aft in the transonic regime, improving the aircraft stability but slightly
worsening the trimability.
All the aerodynamic and stability results will be compared to those obtained from wind tunnel tests
carried out by SimSAC partner TsAGI, on a model built by Politecnico di Milano.

Figure 8.5: TCR wind tunnel model: detail of the attachment to the internal balance through rear and
top fuselage (left) and the final model assembling (right).

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
72 of 79

9 Recommended best practices

CEASIOM consist of a range of different aerodynamic methods. The choice of the aerodynamic method
depends on many parameters like the desired accuracy, calculation time, enveloppe, geometry. Nevertheless, some general recommandations can be given. Datcom should always be used only for a
general and fast guess. As Datcom is a semi-empirical handbook method, one should not expect very
accurate results.
Envelope

Figure 9.1: Recommended methods depending on the enveloppe

The main Tier I tool, the linear vortex lattice method Tornado, should only be used at small angles
of attack and small Mach number (up to 0.5 - 0.7, depending on the geometry). The Tier I+ Euler
code Edge extends the enveloppe to transonic and supersonic speed. For high angles of attack,
Tier II extern RANS codes are indespensable.
Geometry

For design with highly swept wings, a significant part of vortex lift is generated which is not taken
into account by the vortex lattice method Tornado. One is recommand to use for such cases the
Euler code Edge.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
73 of 79

Mesh
Tornado For a mesh analysis, the Warren-12 platform is used where also some reference
values exist[36]. The number of panels in chordwise and spanwise direction were changed
separately and the influence on the lift curve slope was analysed. In both cases, the panels
are distributed equidistant over the wing. One sees that it is no use to have a very fine mesh

Figure 9.2: Warren 12 wing configuration

Figure 9.3: Number of panels and there influence on the result

in chordwise direction, while making the mesh finer in spanwise direction gives more precise
results. This is due that the chordwise local lift changes continuously, while the spanwise
local lift changes in the form of a wave. Doubling the number of panels in spanwise direction
from 10 to 20 gives only a 2% improvement.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
74 of 79

Edge
A simple mesh analysis was done for the X-31 in Edge, at an angle of attack of 5 and a
speed of M = 0.8. The surface mesh was done in Sumo, and the volumic mesh is done
in tetgen. The growth factor (Tet radius/Edge radius) was changed from 1.6 to 1.1. While
Tet radius/Edge radius
1.6
1.5
1.4
1.3
1.2
1.1

Nodes
116780
127392
143199
170422
225201
353252

Boundary triangles
110144
110144
110144
110144
110144
102564

Tetrahedra
564467
632445
733229
905819
1251075
2066133

CL
3.459
3.489
3.531
3.531
3.540
3.558

CD
0.408
0.385
0.367
0.338
0.301
0.248

Cm
-28.301
-28.599
-29.043
-29.010
-29.050
-29.133

Table 9.1: Mesh influence on Edge results

the lift coefficient and the pitching moment coefficient converge, the drag coefficient does not
converge. This is due that the automatic generated meshes from Sumo are not good enough
for transonic speeds with a lot of schocks. A finer mesh or adaptive mesh should be use in
transonic speeds.

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
75 of 79

Bibliography
[1] Andres Puelles Further Development and Application of CEASIOM to the TCR Transonic Cruiser
Design, Master thesis 2009, Department of Aeronautics, Royal Institute of Technology, Stockholm.
[2] Tomasz Goetzendorf-Grabowski, Franciszek A. Dul., Dawid Mieszalski, Lukasz Stefanek. WUT.
SimSAC Delivry Report 5.4.2: Software and incremental contribution to users and technical manual for SCA system of modules of CEASIOM-050 simulation system.
[3] Melin T., Using Internet Interactions in Developing Vortex Lattice Software for Conceptual Design,
Licentiate Thesis Report 2003, Department of Aeronautics, Royal Institute of Technology, Stockholm.
[4] ISO 1151-1: Mcanique du vol - Concepts, grandeurs et symboles. Partie 1: Mouvement de lavion
par rapport lair, Norme International, 1988.
[5] B.H. Cook: Flight Dynamics Principles, Butterworth Heinemann, Cranfield 1997.
[6] http:www.j2aircraft.com
[7] Alireza Maheri, Chris Beaverstock, Maria Sidoryuk. Delivery D5.3-2: Flight Control System Designer Toolkit Technical Report and User Manual
[8] Goetzendorf-Grabowski T. Modelling of the aircraft motion - theory and application, in Optimalsteuerungsprobleme von Hyperschall-Flugzeugsystemen, Workshop des Sonderforschungsbereichs 255, Greifswald-Munchen

1997.
[9] Goetzendorf-Grabowski T. Influence of Stability Derivatives on a Quality of Simulation (supersonic
flow), Journal of Theoretical and Applied Mechanics, No. 4, Vol 32, pp. 773-791, Warsaw 1994.
[10] Etkin B. Dynamics of Atmospheric Flight. Stability and Control. John Wiley & Sons, NY, 1972
[11] Bushgens G.S. Aerodynamics, stability and controllability of super sonic aircraft. Nauka, Moscow,
1998, 811pp. (In Russian)
[12] Hoak, D. E., et al., The USAF Stability and Control DATCOM, Air Force Wright Aeronautical Laboratories, TR-83-3048, Oct. 1960 (Revised 1978).
[13] Melin T., Using Internet Interactions in Developing Vortex Lattice Software for Conceptual Design,
Licentiate Thesis Report 2003, Department of Aeronautics, Royal Institute of Technology, Stockholm.
[14] D. Eller. SimSAC Delivery report 2.3-5. Mesh generation using sumo and tetgen. Royal Institute of
Technology, Stockholm.
[15] http://fdl8.flyg.kth.se/dwf/dwfs.html
[16] Si,H., and Gaertner,K. Meshing Piecewise Linear Complexes by Constrained Delaunay Tetrahedralizations. In Proc. 14th International Meshing Roundtable, pages 147-163, September 2005,
http://tetgen.berlios.de/
[17] http://www.ansys.com/products/icemcfd-mesh-gen.asp

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

KTH

Deliverable
Version
Page

D 3.2-5
1
76 of 79

[18] Ghoreyshi, M., Badcock, K.J. and Woodgate, M. Integration of Multi-Fidelity Methods for Generating an Aerodynamic Model for Flight Simulation, 46th Aerospace Sciences meeting, (Reno, NV),
2008, AIAA-2008-197.
[19] IGES (ANS US PRO/IPO-100-1996) Initial Graphics Exchange Specification IGES 5.3,
http://www.uspro.org/documents/IGES5-3 forDownload.pdf
[20] Tang, C.Y., Gee, K. and Lawrence, S. Generation of Aerodynamic Data using a Design of Experiment and Data Fusion Approach, 43rd AIAA Aerospace Sciences meeting, (Reno, NV), 2005,
AIAA-2005-1137.
[21] Laurenceau, J. and Sagaut, P. Building Efficient Response Surfaces of Aerodynamic Functions
with Kriging and Cokriging, AIAA Journal, Vol. 46, No. 2, pp. 498507, 2008.
[22] Eliasson P. A Navier-Stokes Solver for Unstructured Grids., FOI/FFA report FOI-R-0298-SE, FOI,
Stockholm, Sweden, 2001.
[23] DASA-TN-R-R-002-M-0011 - RANGER 2000 FR06/RP01 aerodynamic dataset release 1.1, 1994
[24] Razgonyaev, Valery, Mason, W. H., 1995. An evaluation of aerodynamic prediction methods applied to the XB-70 for use in high speed aircraft stability and control system design. 33rd Aerospace
Sciences Meeting and Exhibit, AIAA95-0759, Reno, Nevada.
[25] Williams, J. E. and Vukelich, S. R. The USAF Stability and Control Digital DATCOM, McDonnell
Douglas Astona UTICS Company, St Louis Division, St Louis, Missouri, 63166, 1979, AFFDL-TR79-3032.
[26] Belotserkovsky S.M. Thin lifting surface in subsonic gas flow. Nauka, Moscow, 1965. 242pp. (in
Russian).
[27] Belotserkovsky S.M., Skripatch B.K., and Tabachnikov V.G. The Wing in unsteady Gas Flow.
Nauka, Moscow, 1971. 767pp. (in Russian).
[28] Belotserkovsky S.M., Skripatch B.K. Aerodynamic derivativatives of flying vehicle and wing at subsonic velocities. Nauka, Moscow, 1975. ???pp. (in Russian).
[29] Vortex-Lattice Utilization, A workshop held at Langley Research Center, Hampton, Virginia, May
1781, 1976, NASASP405. 424pp.
[30] Belotserkovsky S.M., Nisht M.I. Separated and unseparated ideal fluid flow about thin wings
Nauka, Moscow, 1978. 351pp. (in Russian).
[31] Belotserkovsky S.M., Study of the Unsteady Aerodynamics of Lifting Surfaces Using the Computer,
Annual Review of Fluid Mechanics, Vol. 9, 1977, pp. 469494.
[32] Katz J., and Plotkin A., LowSpeed Aerodynamics: From Wing Theory to Panel Methods,
McGrawHill, New York, 1991.
[33] Magnus, A. E.; and Epton, M. A. PAN AIR A Computer Program for Predicting Subsonic or Supersonic Linear Potential Flows about Arbitrary Configurations Using a Higher Order Panel Method.
Vol. I, Theory Document (Version 1.0). NASA CR-3251, 1980.
[34] Fornasier, L. Linearized Potential Flow Analysis of Complex Aircraft Configurations by HISSS, a
Higher-Order Panel Method. AIAA Paper 85-0281, 1985.
[35] Maskew, B. PROGRAM VSAERO A Computer Program for Calculating the Non-Linear Aerodynamic Characteristics of Arbitrary Configurations, Users Manual. NASA CR-166476, 1982.
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
77 of 79

[36] Mason W.H. Applied Computational Aerodynamics, VirginiaTech, 1997. http://www.aoe.vt.edu/ mason/
[37] Khrabrov A., Kolinko K., Glazkov S., Sidoryuk M., Boldyrev S., Anoshin J. Delivery D 3.2-1: Review
of steady and unsteady panel codes and Tier I simulation of aircraft stability and control characteristics using vortex lattice method. TsAGI. 31 January 2007.
[38] http://web.mit.edu/drela/Public/web/avl/
[39] Mark Drela. Integrated Simulation Model for Preliminary Aerodynamic, Structual, and Control-Law
Design of Aircraft, AIAA 99-1394, MIT Departement of Aeronautics and Astronautics, Cambridge.
[40] Mark Drela. ASWING 5.81 Technical Description of the Unsteady Extension. MIT.
http://web.mit.edu/drela/Public/web/aswing/
[41] Lan, C. Edward. NASA-CR-172501. VORSTAB: A computer program for calculating lateraldirectional stability derivatives with vortex flow effect
[42] Sta ley C. Perkins Jr., Michael R. Mendenhall. Prediction of Post-Stall Aerodynamic Characteristics
of Maneuvering Aircraft. Nielsen Engineering & Research, Inc. AIAA 99-3115.
[43] Begovschitz V., Khrabrov A., Kolinko K., and Machulsky J., UNST: Package of Interacive Calculation of Steady and Unsteady Aerodynamic Characteristics of Aircraft, User Manual, TsAGI,
Zhukovsky, 1993.
[44] Adrien Brard. Method Development for Computer Aided Engineering for Aircraft Conceptual Design. Licentiate Thesis Report 2008, Department of Aeronautics, Royal Institute of Technology,
Stockholm.
[45] Paul Molitor. Extending the Tornado Vortex Lattice Method for Stbility Analysis using the CEASIOM
Softwar. Master Thesis Report 2009, Department of Aeronautics, Royal Institute of Technology,
Stockholm.
[46] S.J. Sequeira, D.J. Willis and J.Peraire. Comparing aerodynamic models for numerical simulation of
dynamic control of aircraft. In 44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada,
9-12 January 2006.
[47] European Benchmark on Numerical Prediction of Stability and Control Derivatives. AIAA 2009
paper nr 4116. Bruno Mialon & all. 27th AIAA Applied Aerodynamics Conference, 22 - 25 June
2009, San Antonio, Texas.
[48] Polhamus E.C. Prediction of Vortex Lift Characteristics by a leading edge Suction Analogy. J.
Aircraft, 1971. Vol.8, No.4, pp.193199.
[49] Sven G. Hedman. Application of a Computer Program for Calculation of Lift for Delta Wings.
[50] Lamar E. J. Extension of leading edge suction analogy to wings with separated flow around the
side edges at subsonic speed. NASA TR R-428
[51] Lan, C.E. VORSTAB: a Computer Program for Calculating Lateral-Directional Stability Derivatives
with Vortex Flow Effect. NASA Contractor Report 172501, 1987.
[52] Raymer D.P. Aircraft Design: A Conceptual Approach. 4th ed, AIAA Education Series, Reston, VA.
2006.
[53] Eckert, E.R.G., Engineering Relations for Friction and Heat Transfer to Surfaces in High Velocity
Flow, Journal of the Aeronautical Sciences, Vol. 22, 1955, pp. 585-587.
This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
78 of 79

[54] Mark Drela. XFOIL. MIT. http://web.mit.edu/drela/Public/web/xfoil/

[55] Sven G. Hedman. Simulating Wings in Viscous Flow Using the TORNADO and the XFOIL Programs
[56] J. C. Sivells J. C., Spooner S. H.. Investigation in the Langley 19-foot Pressure Tunnel of the Two
Wings of NACA 65-210 and 64-210 Airfoil Section with Various Type Flaps. NACA Report 942.
[57] Purser P., Spearman M. L.. Wind-Tunnel Tests at Low Speed of Swept and Yawed Wings Having
Various Plan Forms. NACA Technical Note 2445, 1951.
[58] Evans, William, T.. Leading-Edge Contours for Thin Swept Wings: an Analysis of Low- and HighSpeed Data. NACA RM A57B11, 1957.
[59] Racisz, Stanley F., and Paradiso, Nicholas J.. Wind-Tunnel Investigation at High and Low Subsonic
Mach Numbers of a Thin Sweptback Wing Having an Airfoil Secton Designed for High Maximum
Lift. NACA RM L51L04, 1952.
[60] Graham, David, and Evans, William T.. Investigation of the Effects of an Airfoil Section Modification
on the Aerodynamic Characteristics at Subsonic and Supersonic Speeds of a Thin Swept Wing of
Aspect Ratio 3 in Combination With a Body. NACA RM A55D11, 1955.
[61] Graham, David, and Koenig, David G.. Tests in the Ames 40- by 80- Foot Wind Tunnel of an
Airplane Configuration With an Aspect Ratio 2 Triangular Wing and an All-Movable Horizontal Tail
- Longitudinal Characteristics. NACA RM A51B21, 1951.
[62] Evans, William T.. Data From Large-Scale Low-Speed Tests of Airplane Configurations With a
Thin 45 Degrees Swept Wing Incorporating Several Leading-Edge Contour Modifications. NACA
RM A56B17, 1956.
[63] Koenig, David G., and Aoyagi, Kiyoshi. Large-Scale Wind-Tunnel Tests of an Airplane Model With
a 45 Degrees Sweptback Wing of Aspect Ratio 2.8 With Area Suction Applied to Trailing-Edge
Flaps and With Several Wing Leading-Edge Modifications. NACA RM A56H08, 1956.
[64] Mengmeng Zhang. Modification and Validation of Aerodynamic Characteristics in Viscous Flow
using the Computer Program TORNADO. Master Thesis. KTH 2009.
[65] Askin T. Isikveren. Quasi-Analytical Modelling and Optimisation Techniques for Transport Aircraft
Design. Doctoral Thesis. KTH 2002.
[66] Edge User Guide FOI, March 2007.
[67] A. Da Ronch. AMB Interface for External CFD Solvers. University of Liverpool, July 2009.
[68] RTO/AVT-161: Assessment of Stability and Control Prediction Methods for NATO Air and Sea
Vehicles. https://extsites.dlr.de/as/UCAV/default.aspx.
[69] DASA-TN-R-R-002-M-0011 - RANGER 2000 FR06/RP01 aerodynamic dataset release 1.1, 1994
[70] Fornasier L., HISSS - A Higher Order Subsonic/Supersonic Singularity Method for Calculation
Linearized Potential Flow. AIAA Paper 84-1646, 1984
[71] http://www.cfse.ch/cfse/site/tools-NSMB.php
[72] Bertin J.J., Smith M.L.: Aerodynamics for Engineers, Prentice-Hall International, Inc. London 1989
[73] MIL-F-8785C - Military Specification - FLYING QUALITIES OF PILOTED AIRPLANES

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

D 3.2-5
1
79 of 79

Appendix :
Fuselage
Nose - Vertical Diameter
Nose - Horizontal Diameter
Nose - Down-sweep angle
Nose - Length/Diameter ratio
Nose - Body front angle
Total lenght
Tail - Vertical Diameter
Tail - Horizontal Diameter
Tail - Down-sweep angle
Tail - Length/Diameter ratio
Tail - Body front angle

f12
0.215 m
0.215 m
2.98
0.9974
19.289
2.239 m
0.215 m
0.215 m
7.735
2.96
16.014

Ranger 2000
1.78 m
1.09 m
11
1.5
40
10.39 m
1.78 m
1.09 m
9.25
2.2
45

X-31
2.062 m
1.167 m
4
2.22
7
13.2 m
1.5 m
1.167 m
0
0.1
0

Ranger 2000
15.5 m2
7.06
10.46 m
Eppler 502
0
0.9
0.2
0
0.87
0.26
0
0.52
0
0
0
0
0
0
0

X-31
20.83 m2
2.42
7.1 m
customized
0
0.42
0.57
-5
0.15
0.95
6
0.05
5
-4
57
4
45
13
70

Table 2: Fuselage
Wing
Area
Aspect Ratio
Span
Airfoils
Root - incidence
Kink 1 - taper ratio
Kink 1 - span ratio
Kink 1 - incidence
Kink 2 - taper ratio
Kink 2 - span ratio
Kink 2 - incidence
Tip - taper ratio
Tip - incidence
Inbord - dihedral
Inbord - leading edge sweep
Midbord - dihedral
Midbord - leading edge sweep
Outbord - dihedral
Outbord - leading edge sweep

f12
0.44 m2
9.35
2.04 m
customized
2.94
0.55
0.36
0
0.30
0.84
-1.84
0.22
-5.37
3.3
32
3.8
32
4.04
32

Table 3: wing

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Horizontal Tail
Area
Aspect Ratio
Span
Airfoils
Root - incidence
Kink - taper ratio
Kink - span ratio
Kink - incidence
Tip - taper ratio
Tip - incidence
Inbord - dihedral
Inbord - leading edge sweep
Outbord - dihedral
Outbord - leading edge sweep

f12
0.10 m2
6.36
0.73 m
customized
0.28
0.38
1
2.3
0.38
2.3
6.05
33.4
6.05
33.4

Ranger 2000
2.79 m2
3.4
3.08 m
Naca 0009
-2
1
0.01
-2
0.68
-2
0
11.9
0
11.9

D 3.2-5
1
80 of 79

X-31
-

Table 4: Horizontal Tail

Canard
Area
Aspect Ratio
Span
Airfoils
Root - incidence
Kink - taper ratio
Kink - span ratio
Kink - incidence
Tip - taper ratio
Tip - incidence
Inbord - dihedral
Inbord - leading edge sweep
Outbord - dihedral
Outbord - leading edge sweep

f12
-

Ranger 2000
-

X-31
2.17 m2
3.34
2.69 m
customized
0
0.99
0.01
0
0.25
0
0
45
0
45

Table 5: Canard

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.

Deliverable
Version
Page

KTH

Vertical Tail
Area
Aspect Ratio
Span
Airfoils
Root - incidence
Kink - taper ratio
Kink - span ratio
Kink - incidence
Tip - taper ratio
Tip - incidence
Inbord - dihedral
Inbord - leading edge sweep
Outbord - dihedral
Outbord - leading edge sweep

f12
0.07 m2
1.52
0.33 m
customized
0
0.39
1
0
0.38
0
0
44.5
0
44.5

Ranger 2000
2.96 m2
1.34
1.99 m
Naca 0009
0
0.95
0.1
0
0.49
0
0
38.5
0
38.5

D 3.2-5
1
81 of 79

X-31
3.95 m2
1.13
2.11 m
customized
0
0.33
0
0
0.33
0
0
48
0
48

Table 6: Vertical Tail

This document is the property of the SIMSAC consortium and shall not be distributed or reproduced without the formal approval of the SIMSAC coordinator.