//img.uscri.be/pth/0b20dfb52639b35d9555aa413c173d58ed58bc59
Cet ouvrage fait partie de la bibliothèque YouScribe
Obtenez un accès à la bibliothèque pour le lire en ligne
En savoir plus

European Congress on Computational Methods in Applied Sciences and Engineering ECCOMAS

De
15 pages
European Congress on Computational Methods in Applied Sciences and Engineering ECCOMAS 2004 P. Neittaanmäki, T. Rossi, S. Korotov, E. Oñate, J. Périaux, and D. Knörzer (eds.) Jyväskylä, 24—28 July 2004 1 MULTI-DISCIPLINARY OPTIMISATION OF A SUPERSONIC TRANSPORT AIRCRAFT WING PLANFORM G. Carrier* *ONERA - Applied Aerodynamics Department 29, avenue de la division Leclerc, BP 72, 92322 Châtillon Cedex, France E-mail: , web page: Key words: Supersonic, High-Speed Civil Transport (HSCT), Multi-disciplinary Analysis and Optimisation (MDAO), Optimisation. Abstract. A Multi-Disciplinary Analysis and Optimisation (MDAO) system for the evaluation and optimisation of the performance of a High-Speed Civil Transport aircraft has been developed at ONERA within the context of the CISAP project. This paper first describes the MDO system implemented at ONERA. This MDO system is constructed by coupling a Multi-Disciplinary Analysis (MDA) process, developed in the present research project, with different optimisation algorithms including a gradient-based optimiser and a Genetic Algorithm (GA). The MDA process embeds the different disciplines modules and schedules and monitors their execution. It is capable of evaluating the global aircraft performance for a specified mission.

  • cisap project

  • multi-disciplinary

  • has been

  • optimisation

  • supersonic cruise

  • design variable

  • mdo system

  • aircraft

  • disciplines analysis


Voir plus Voir moins

European Congress on Computational Methods in Applied Sciences and Engineering
ECCOMAS 2004
P. Neittaanmäki, T. Rossi, S. Korotov, E. Oñate, J. Périaux, and D. Knörzer (eds.)
Jyväskylä, 24—28 July 2004
MULTI-DISCIPLINARY OPTIMISATION OF A SUPERSONIC
TRANSPORT AIRCRAFT WING PLANFORM
*
G. Carrier
*
ONERA - Applied Aerodynamics Department
29, avenue de la division Leclerc, BP 72, 92322 Châtillon Cedex, France
E mail: gerald.carrier@onera.fr, web page: www.onera.fr
Key words: Supersonic, High-Speed Civil Transport (HSCT), Multi-disciplinary Analysis
and Optimisation (MDAO), Optimisation.
Abstract. A Multi-Disciplinary Analysis and Optimisation (MDAO) system for the evaluation
and optimisation of the performance of a High-Speed Civil Transport aircraft has been
developed at ONERA within the context of the CISAP project.
This paper first describes the MDO system implemented at ONERA. This MDO system is
constructed by coupling a Multi-Disciplinary Analysis (MDA) process, developed in the
present research project, with different optimisation algorithms including a gradient-based
optimiser and a Genetic Algorithm (GA). The MDA process embeds the different disciplines
modules and schedules and monitors their execution. It is capable of evaluating the global
aircraft performance for a specified mission. Among the different disciplines considered
during the MDA, the aerodynamics and structure disciplines are given special care and
analysed with high fidelity methods, respectively Computational Fluid Dynamics (CFD) and
Finite Element Method (FEM). The other disciplines such as engines performance and flight
mechanics are evaluated with simpler methods.
This MDO system has been applied to optimise two aircraft variants: a Mach 2.0 and a
Mach 1.3 aircraft architectures. The overall objective has been to maximise the aircraft range
while multiple design constraints were considered. The results of these optimisations are
presented in the second part of this paper.
1G. Carrier.
1. INTRODUCTION
The story of civil air transportation has been marked in 2003 by the last scheduled
commercial flight of Concorde. Despite this event, which marks a break in supersonic
commercial transport, a continued interest in High Speed Civil Transport (HSCT) has been
existing in Europe. If the recent European industrial efforts have been mostly concentrated on
large civil transport aircraft, for which a market is clearly identified, HSCT with its reduced
transportation time offers a complementary answer to the continuously developing demand of
commercial air transport. From the experience of a more than 30 years long period of regular
Concorde service, the improvements required over Concorde in term of fuel efficiency and
environmental acceptability to render a second generation HSCT viable are significant,
putting challenging demands on design methodologies to be used in the development of such
an aircraft.
An important knowledge has been consolidated in Europe during the last decades in the
different individual key-disciplines required to design a Mach 2.0 HSCT aircraft. For
instance, in the aerodynamics discipline, wing design and optimisation have been
demonstrated[1] and low-speed performance improvements achieved[2] for a Mach 2.0
aircraft. However, very few researches have been conducted in Europe for Mach number
lower than 2.0, while an interest toward aircraft designs cruising at intermediate Mach number
between 1.0 and 2.0, have recently been expressed by industry. Indeed, choosing a supersonic
cruise speed lower than Mach 2.0 may enable to use existing conventional technologies or, at
least, the less sever operating conditions would facilitate the development of the key
technologies required for HSCT and specific to such an aircraft, such as engine performance
and propulsion integration issues.
In this context, CISAP, a collaborative project between Airbus (AI) and the Association of
European Research Establishments in Aeronautics (EREA), was set-up to investigate the
potential of HSCT aircraft concepts designed for cruise Mach numbers below 2.0 and to
examine the implications for the optimal wing planform of lowering the cruise speed. Such
objectives of new aircraft concepts evaluation require the different disciplines affecting the
overall aircraft performance to be considered through a multi-disciplinary approach. The
Multi-Disciplinary Analysis and Optimisation (MDAO) technology[4] provides an
appropriate method for evaluating and designing aircraft architectures presenting strong
interaction between the different disciplines, such as HSCT[3] and BWB[5] aircraft.
The present paper describes the multi-disciplinary optimisation work performed at
ONERA within the CISAP project. After an introduction of the context of the CISAP project,
the development of the MDAO system for HSCT is described in the second section. The
application of this MDAO system has been made to optimise, first a Mach 2.0 and second a
Mach 1.3 aircraft, and the optimisation results obtained are described in the last section.
2G. Carrier.
2. CONTEXT: THE CISAP PROJECT
CISAP, acronym for “Cruise speed Impact on Supersonic Aircraft wing Planform”, was a
collaborative project between Airbus and EREA. This project was conducted by the Research
Establishments DLR (co-ordinator), NLR, ONERA and QinetiQ, over an 18 months period
running from July 2002 to December 2003.
This project intended to investigate the multi disciplinary effects of changing the cruise
Mach number of an High Speed Civil Transport (HSCT) aircraft, and especially the effect on
the wing planform[6]. Three different cruise Mach numbers were commonly chosen with AI:
M=2.0, M=1.6 and M=1.3.
The research conducted in CISAP was organised in three work packages:
• WP1 consisted in defining the conceptual design of both M=1.6[7] and M=1.3 aircraft
architectures, intended to serve as starting point for the MDO work of WP2. The
M=2.0 aircraft concept was provided by AI;
• WP2 was the main work package and included the development, validation and
application of MDO methodologies by the different partners[8], each partner having
developed its own MDO system on the basis of a common core set of tools;
• WP3 included complementary investigations regarding especially the low speed
performances, alternative structural layout and off design structural behaviour, for one
of the optimised aircraft configuration of WP2.
The next sections of this paper gives an overview of the work performed at ONERA within
WP2 of CISAP.
3. MULTI-DISCIPLINARY OPTIMISATION METHODOLOGY AND SYSTEM
3.1. Optimisation problem formulation
The design/optimisation problem to be solved has been defined as a maximisation problem
of a system(aircraft) level objective defined as the achievable range of an aircraft having:
• a fixed payload of 250 passengers;
• a fixed Maximum Take Off Weight (MTOW) of 340,000 kg.
This problem formulation has been preferred to the alternative of minimising the weight of
an aircraft having the same fixed payload and a minimum achievable range.
3.2. Design variables, constraints and constants
The design space investigated in the present work concerned aircraft design variants
differing by the wing shape and position relatively to the fuselage and also by the altitude at
3G. Carrier.
which the supersonic cruise leg is started. To explore this design space, the MDA procedure is
parameterised through a total of 16 parameters (design variables):
• 7 variables are used to define the wing planform, defined as a double-trapezoidal shape
(Figure 1);
• 6 variables are used to control the thickness and twist in 3 wing sections ( root, crank and
tip sections). Linear interpolations are used in-between;
• 2 variables define the wing position relatively to the fuselage;
• 1 variable is used to control the “start of cruise” altitude (this altitude was actually an
implicit result of the aircraft mission analysis, the actual explicit design variable used
being the angle of attack of the aircraft in supersonic cruise).
A set of system level constraints have been commonly defined and used in this CISAP
project[6]. Different types of constraints can be distinguished:
• geometry constraints affecting wing shape and position to insure that realistic and
realisable geometry of the investigated aircraft designs (minimum thickness at different
wing locations, maximum spanwise wing bending, etc);
• a low speed aerodynamic constraint (stability);
• a trim ability and a maximum angle of attack constraints for the supersonic cruise
condition.
Furthermore, some aircraft/mission specifications were kept constant during the MDO
design optimisations. These parameters can be viewed as design constants or constraints
inside the MDA process. The most significant constants are:
• payload and MTOW (identical for M=1.3 and M=2.0 optimisations, see previous section
for values);
• geometry of the fuselage;
• weights and centre of gravity of the aircraft elements, except the wing weight (but some
equipment weight differ between M=1.3 and M=2.0 aircraft variants);
• maximum allowable stresses in the wing material considered during the wing structure
sizing process(200 MPa).
4G. Carrier.
3.3. MDO system description
Multi-Disciplinary Feasible (MDF) approach
System Optimiser:
Maximise (Range)
Tip s.t. design constraints
chord
outer
Crank
chord
MDA
inner Root
chord
Sub-system 1
Analysis
Wing X
position Sub-system 2
Analysis
Sub-system n
Analysis
Figure 1: Parameters defining the wing planform and
Figure 2: MDO system coupling the system-longitudinal position (8 among the 15 geometry parameters).
level optimiser with the MDA procedure.
The MDO system developed and applied by ONERA in CISAP is based on the
“traditional” MDO technique referenced in literature as the Multi Disciplinary Feasible[9]
formulation. This MDF approach, which ensures that all disciplines are evaluated and all
inter disciplinary dependencies to be satisfied, is recognised as the most robust, although
computationally expensive. In the MDF approach, the MDO process is constructed by
coupling a system level optimisation routine with a MDA process in charge of performing the
complete system analysis, including the different disciplines analysis and possible inner-
loops. As illustrated in Figure 2, the system level optimiser then explores the design space and
optimises the system level objective (range in this work) by successively candidating new
aircraft designs to be evaluated by the MDA process. The MDA process, in turn, returns to the
optimiser values for the objective function and for the amplitude of violation of the design
constraints (if constraints are violated).
5
jjh
crank
span
Design Variables
Range +
ConstraintsG. Carrier.
MDA process description
Cruise altitude
Angle Altitude
of
Attack
CFD
CL( ) CD inviscid ( ) CD total
computation Total drag
elsA
CFD Pressure: Flight
mesh CP ( )
MechanicsEngine
Mesh Model
Aircraft Range,Cruise Maneuver generationgeometry loads extrapolation Altitude,AircraftICEM-CFD
Geometry Geometry database
ConstraintsGenerator
Wing structure
FuelWingCSM Weights &
optimisation Balancemesh weight
NASTRAN
Figure 3: Structure (simplified) of the Multi-Disciplinary Analysis (MDA) system developed for High-Speed
Civil Transport (SCT)
The most important part is therefore the MDA process itself, which is described in Figure
3. In this MDA procedure, the different disciplines required to evaluate the aircraft range are
evaluated through dedicated mono-disciplinary analysis modules. These disciplines are:
• aerodynamics,
• structures,
• engine performance,
• flight mechanics.
The two most important disciplines in term of their impact of the operating range of a
supersonic aircraft were anticipated to be aerodynamics (in cruise condition) and structures.
High-fidelity methods were therefore used to perform the analyses corresponding to these two
disciplines. The cruise aerodynamics was evaluated using Computational Fluid Dynamics
(CFD) while the wing structures was evaluated and sized using Finite Element Method
(FEM).
All other disciplines were evaluated by means of simpler analysis methods, such as semi-
empirical methods (aerodynamic performances in low-speed and transonic flight conditions),
which provide explicit analytical formula, or by means of “relatively inexpensive” iterative
procedures (mission integration, engine performance).
The MDA process implemented at ONERA has been constructed on the basis of a set of
modules including simple in house programs, modules from the common “multi model
generator” developed by NLR[10], the ONERA CFD solver elsA[12] , the commercial
® ®
structured mesh generator ICEM Cfd and the commercial FEM code MSC-NASTRAN .
6
?aaaaG. Carrier.
The object-oriented script language Python[11] has been used to develop the wrapper code
that embeds the different mono-disciplinary analysis modules into a completely automated
MDA process. This Python code acts as a scheduler and controls the execution of the different
discipline modules, insuring the necessary data exchanges between these modules.
Figure 5: CFD mesh of the Mach 2.0
Figure 4: FEM model of the Mach 2.0 datum
datumaircraft; External boudary.
aircraft (right) wing.
60000
50000
40000
3030000000
Datum M=1.3 , altitude
20000 Optimised M=1.3 , altitude
10000
00
00 11000000 20020000 30003000 40004000 55000000 66000000
RaRannggee
Figure 6: Mission profile considered for the range calculation; Comparison between the datum and
optiumised configurations illustrating the range improvements achieved through MDO.
As presented in Figure 3, the MDA process is passed a set of design variables defining the
wing geometry and the cruise condition angle-of-attack for which the analysis has to be
performed. As a first step, the “Multi Model Generator” (MMG)[10] creates both the FE
model and the pointwise description of the aircraft wing and fuselage geometry. Different
parametric FE models, one for each cruise Mach number, have been implemented with
“rubberised” layouts to fit the external wing shape (see Figure 4). The main wing structural
elements are modelled: ribs, spars, upper and lower skins and stringers are present, as well as
the holes due to the landing gears bay. The two engines (for a half wing) are modelled as
point masses.
The second step corresponds to the CFD mesh generation and calculation, using the
®commercial ICEM Cfd mesh generation software run in batch mode. From the separate
7
altitudeG. Carrier.
geometry descriptions of the wing and the fuselage, generated by the MMG, a CAD database
of the aircraft (wing-body configuration without engine is considered for the supersonic
®
aerodynamic modelling) is generated with ICEM-Cfd Med, including the wing fuselage
intersection. Then the topological mesh elements are created and fitted to the actual aircraft
®
geometry. Finally ICEM-Cfd HEXA is used to generate a multi-blocks structured mesh
around the configuration (see Figure 5). This mesh is then used to calculate the solution of the
3D Euler compressible equations, with the ONERA elsA[12] solver, for the angle of attack
specified as one of the design variables. The results of this flow calculations are the
aerodynamic non-dimensional lift, pressure drag and pressure coefficients. The lift coefficient
CL is used to determine the start-of-cruise altitude, as the unique altitude at which the
instantaneous Lift=Weight equation is satisfied. The start-of-cruise altitude is used to evaluate
engine performances and friction drag (flat plate assumptions).
In the third step the wing is evaluated by sizing the different structural elements to
withstand a load case corresponding to a 2.5 g maneuver (performed at cruise Mach number
®
and altitude, in zero-fuel condition), using the commercial FE code MSC-NASTRAN and its
built-in optimisation module SOL200. The aerodynamic loading for this load case is deduced
from the pressure distribution of the CFD calculation using a simple linear extrapolation
procedure. In this sizing process, the actual load case depends on the aircraft empty weight
which in turn depends on the wing weight. An iterative procedure has been used to solve this
coupled aerodynamic-structure-weight system, requiring a FEM optimisation at each iteration.
The final results of the this iterative wing sizing procedure are then used to determine the total
fuel mass embarked in the aircraft. Indeed, with the fixed MTOW considered, any kilogram
saved on wing structure is directly converted into an additional kilogram of fuel (it is check
that this is compatible with the maximum fuel volume offered by the wing geometry).
Finally, the last step consists in the range calculation. Two methods are included in the
present MDO process. The simplest one consists in a direct application of the Breguet-Leduc
range formula, with correction for the climbing and transonic phases fuel consumption. The
second method consists in a “mission module” that integrates the flight mechanics equations
for the investigated aircraft design performance figures (supersonic cruise aerodynamic
performance, embarked fuel, etc…) and returns the operational aircraft range. The module
assumes a mission profile composed of fixed low speed, climb and transonic cruise (at
M=0.95) legs of 500 nm, and of the final supersonic cruise leg as it is illustrated in Figure 6.
The length of ascending (“constant lift coefficient”) supersonic cruise leg is variable, depends
on the aircraft design performance figures and is calculated so that the aircraft arrives at the
“scheduled destination” with a fuel quantity corresponding exactly to the what is needed to
perform a 30 minutes hold + 250 nm diversion flight and land at alternate destination with the
mandatory fuel reserve (right part of the Figure 6 shows the diversion flight).
Optimisation methods
Several optimisations algorithms have been introduced in the MDO system and used for
this study. Two local optimisers based on gradient information (finite differences), the
©feasible directions algorithm COPES/CONMIN[13][14] and the SQP algorithm CFSQP [15]
8G. Carrier.
have been applied. In addition, the Genetic Algorithm GADO[16] has also been used, as an
alternative (stochastic) algorithm to the gradient optimisers.
3.4. Computer related issues
A natural approach for building the present automated MDO system would have been to
implement and execute the complete system with all its software pieces on the same machine.
A distributed MDO architecture has been preferred because:
• it was felt to provide a more promising basis for possible future MDO system
developments;
• the use of several commercial and licensed software packages (for mesh generation and
structural optimisations) that installed on dedicated machines makes it much easier to
simply execute these codes on their native machines;
• this provided a way to speed-up the complete procedure by running the most expensive
part on powerful machine (the CFD code was run on a vector NEC SX6 supercomputer).
Finally, the complete multi-disciplinary analysis of one aircraft configuration required
approximately 15 minutes (wall-clock time), the most time consuming parts being the high-
fidelity analyses and especially the cruise aerodynamics evaluation (CFD mesh generation +
flow calculation).
4. MULTI-DISCIPLINARY OPTIMISATIONS RESULTS
The CISAP project investigated three cruise Mach numbers: M=2.0, M=1.6 and M=1.3.
Mach 2.0 and Mach 1.3 were studied at ONERA (in collaboration with QinetiQ[8]) by
applying the MDO system described in the previous section to optimise the combination
aircraft / mission profile (cruise altitude) for each of these cruise Mach number.
The Mach 2.0 case was studied by all partners, and was used as a test case to check the
validity, compare and calibrate the MDO systems implemented by the different partners. For
these Mach 2.0 optimisations, a datum configuration was supplied by AI and was used as the
starting point for the optimisation runs. This datum M=2.0 aircraft was designed to start its
supersonic cruise at FL510 and has a range (Breguet-Leduc) of 5060 nm, according to the
ONERA MDA system.
For the Mach 1.3 case, a datum configuration had been generated by ONERA within the
CISAP WP1 work, using an integrated conceptual design tool, and was intended to serve as
an initial design for the optimisations. This aircraft configuration, designed to start its
supersonic cruise at FL400, has a range (Breguet-Leduc) of 5040 nm, according to the
ONERA MDA system.
9G. Carrier.
4.1. Mach 2.0 aircraft optimisations
The ONERA MDO system was first applied to optimise the Mach 2.0 aircraft
configuration. The optimisation was performed with the “gradient based” optimisers in two
steps. The first step was initiated from the datum Mach 2.0 configuration (see Figure 8) It
proved the validity and operability of the MDO system and yielded a 6% improvement of the
aircraft range within approximately 500 iterations (MDA evaluations).
However, the results of this first step proved to be only a local minima of the design space.
Therefore a second optimisation step was initiated, starting from a promising design
configuration identified by QinetiQ. This additional optimisation step permitted to reach a
11% improvement of the range (compared to the datum configuration). The final optimised
configuration resulting from these optimisation at Mach 2.0 is presented in Figure 8. The
aerodynamic wing pressure distribution of the initial (datum) and optimised configurations at
Mach 2.0 are compared to in Figure 9. The MDO optimisations resulted in an unloading of
the outer wing, which proved to allow wing weight saving (Figure 10).
5400 0 10
-1
5300
-2 5
5200
-3
5100 -4 0
-5
5000
-6 -5
Breguet Range
4900 Cruise Trim constr
Wing Inner sweep -7
Wing Outer sweep
4800 -8 -10
100 200 300 400 500
iter
Figure 8: Comparison of the Mach 2.0
Figure 7:History of convergence of the first stage of optimised and datum configurations.
the Mach 2.0 optimisation.
Finally, for this Mach 2.0 multi disciplinary optimisations, the 11% range increase is the
results of the combination of an improvement in the cruise aerodynamic efficiency (3.3%) and
a wing structural weight reduction (about 9%). As mentioned earlier, the cruise altitude was
actually optimised simultaneously with the aircraft geometry during the MDO process. The
optimum “start of cruise” altitude associated with this Mach 2.0 optimised configuration was
FL 525, slightly higher than for the datum M=2 aircraft.
10
Range
Constraint #3
LE Sweep variation (%)