Advances in computer-aided engineering
Bibliotheek TU Delft
11111111111
C 2117163engIneerIng
CAD/CAM-research at Delft University of Technology
Report of the VF-project CAD/CAM 1989-1994Delft University of Technology Delft, June 1994
Delft University Press Stevinweg 1 2628 CN Delft The Netherlands Telephone +31 15 783254 Fax +31 15 781661
CIP-DATA KONINKLIJKE BIBUOTHEEK, DEN HAAG Advances
Advances in computer-aided engineering: CAD/CAM-research at Delft University of Technology (Report of the VF-project CAD/CAM 19891994). -Delft : -Delft University Press. -111.
ISBN 90-407-1017-1 NUGI 841
Subject headings: design; product modelling
Copyright c 1994 by Faculties of Industrial Design Engineering, Aerospace Engineering, Mechanical Engineering and Marine Technology, Technical Mathematics . and Informatics at Delft University of Technology
All rights reserved.
No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without permission from the publisher: Delft University Press, Stevinweg 1, 2628 CN Delft, The Netherlands.
Contents
Introduction
Faculty of Aerospace Engineering
The Aircraft Design and Analysis System (ADAS): an overview 9 C. Bil
Aircraft aerodynamic design 17
I. Middel
Interactive programs for aircraft structural design and optimization 30 A. Rothwell
Analytical techniques for the optimum conceptual design of subsonic and
supersonic transport aircraft 40
E. Torenbeek
Faculty ofMechanical Engineering and Marine Technology
A design program based on the Monte Carlo method with applications 61 C.M. Kalker-Kalkman
Evaluation of discrete component systems 71
A.L. Schwab, K. van der Werft
A software environment for integrated design and manufacturing of
mechanisms: CIMOME 81
W. Zhang, HA Crone, K. van der Werft
Concept Exploration Model for multi-purpose container carriers 89 H. Boonstra, C. Georgescu
Shipmotion calculations in the ship design process 102 l.M.l. lournée, A. Versluis
SUBCEM, a concept exploration model for underwater vehicles 110 C.G.J.M. van der Nat
PROPEL: Propulsion Installation Selection 118
Faculty of Industrial Design
Supporting multidisciplinary product development 129 A. P. Bremer, T. Schätti
Fast shape designer: a surface modeIer based upon hand sketched curves 137 P A. van Elsas, C.G.C. van Dijk
Cost information tools for designers 145
TJA. Haan, L.S. Wierda
A conceptual sketching device for the early phase of design 153 R. Kol/i, R. Stuyver, l. Hennessey
Research spin-off: design projects in practice using integrated CAD/CAM 163 A.F. Lennings
Robust NC path generation for rapid shape prototyping with a sculpturing
robot system 171
I.W.H. Tangelder, lJ. Broek, PJ. de lager, A.F. Lennings A. Kooijman, A. de Smit, l.S.M. Vergeest
CAD data exchange and model sharing 179
l.S.M. Vergeest, T. Wiegers
From databases to data management within product development 187 R.W. Vroom
Faculty ofTechnical Mathematics and Informaties
Constructive Solid Geometry 197
W.F. Bronsvoort, F.W. lansen
Feature modelling 211
W.F. Bronsvoort, M. Dohmen, W. van Holland, KJ. de Kraker
Ray tracing and radiosity algorithms for photorealistic image synthesis 225 AJ.F. Kok, F.W. lansen
Geometrie icons for flow visualization 233
W.C. de Leeuw
Applying artificial intelligence for intelligent design 244 R.A. Vingerhoeds, B.D. Netten, H. Koppelaar
Preface
This book contains a collection of articles describing on-going CAD/CAM-research at several engineering faculties (Industrial Design Engineering, Aerospace Engi-neering, Mechanical Engineering and Marine Technology, Technical Mathematics and Infonnatics) at Delft U niversity of Technology. This research fonns part of the VF-project TUD-LR-02/85-35 CAD/CAM, which was initiated in 1985. This book is the final report of this project for the period 1989-1994.
The research within this project covers mainly two themes:
Conceptual Design of Complex Products. The emphasis here is on optimization and evaluation of a design of complex products, such as airplanes, ships and other mechanical produets and systems, in a stage where the product is not yet fully detailed. This requires a design support system th at can apply knowledge and ana -lysis techniques to as yet abstract and incomplete design representations. It also requires fast multi-variate optimization techniques to interactively explore large parameter spaces. These topics are the main subject of research at the faculties of Aerospace Engineering and Mechanical Engineering and Marine Technology. Product Modelling and Product Data Management. The emphasis is here on tools for the input, visualization and exchange of geometrie and functional product repre-sentations, such as systems for interactive geometrie modeling, feature modeling, fast prototyping, rendering, data visualization, engineering databases and product data exchange. These topics are mainly covered by research at the faculties of Industrial Design Engineering and Technical Mathematics and Infonnatics.
The book is organized as follows. It starts with an introduction describing the ten projects at the faculties that constitute the VF-project. Then four sections of articles follow describing the research within each faculty. The last section contains a bibliography of the publications of the project.
Keeping up-to-date with new design-supporting technology is of utmost impor-tanee for the university and its engineering curricula. The TU Delft has a strong tradition in design engineering. Actively participating in research on CAD/CAM is the best way to foster this tradition.
F.W. Jansen Project coordinator
Introduction
The YF-project CAD/CAM was one of the largest projects in its kind within the TU Delft. It brought together 10 separate research projects of 4 faculties, in total about 30 fte of research. Coordination was provided by the board of project leaders that met monthly to discuss the research policy, coordination on hardware and soft-ware, and the organization of special events. During the five year period (1989-1994), 7 seminars were held covering the different research topics within the pro-ject. Speakers were researchers from within the project and from outside. In 1992, a CAD/CAM-symposium was organized that attracted 750 participants.
In the following, a short description is given of the different projects and the research teams involved.
AE-a. Conceptual design of airplanes
The design process for airplanes is a complex process where a large number of design parameters, constraints, goal functions and problem structures play an important role. The ADAS system (Aircraft Design and Analysis System) has been developed to support the designer in exploring his design concepts and optimizing the chosen concept. Other research topics within this project are aerodynamic con-figuration research (for unconventional airplane designs), engine design, and research on the relation between airplane design and airplane operation.
Staff (2.6 fte):
prof.ir. E. Torenbeek (project leader) dr.ir. C. Bil
ir. J. Middel dr.ir. H.G. Visser
K.E. Shahroudi MSc. (AIO) Ir. A.H.W. Bos (AIO)
AE-b. Computer-aided engineering and production of airplanes
Structural design of airplanes requires the accurate description of airplane geo-metry. This can only be done with sophisticated surface modeling techniques. Within this project the use of commercial CAD-systems is explored and interfaces are being developed for interfacing with finite element systems for combined aero-dynamic and structural optimization, as weil as for interfacing to engineering data-bases.
Staff (4.1 fte): prof.dr. J. Arbocz
prof. dr. A. Rothwell (project leader) ir. T. van Baten
ir. F. van Dalen ir. P.D. Kempen
ir. M.E. Heerschap (AIO) ir. Zhang Guo
Qi
(AIO)MEMf-a. CAD of discrete component systems
Design of kinematic systems requires the input and analysis of the form and func-tion of an assembly of discrete components. Within this project techniques and systems are being developed for geometric modeling, analyzing and optimizing systems like transmissions and mechanisms.
Staff (3.9 fte):
prof.drjr. K. van der Werff (project leader) prof.ir. H.A. Crone
ir. A van Dijk
ir. C.M. Kalker-Kalkman dr.ir. A.I. Klein Breteler ir. AL. Schwab
Zhang Wenjun, M.S. Eng. (AIO) MEMT-b. Computer-aided ship design
The research in this project aims at developing techniques for exploration of ship concepts in an early stage of the design process. Other topics are the estimation of ship properties and interactive techniques for constructional design.
Staff (3.6 fte):
prof.dr.ing. C. Gallin prof.ir. J. Klein Woud ir. H. Boonstra (project leader) dr.ir. K. Saurwalt dr.ir.E. Deetman ir. J.M.J. Journée ing. A Versluis ir. J.H. Vink ir. J. de Wilde
Dipl.ing. C. Georgescu (AIO)
/DE-a. Computer-aided design and model production
Within this project, surface modeling techniques are being developed for fitting sur-faces through a mesh of points, for interactive design of sursur-faces with use of sketched design curves (fast shape design), for the automatic generation of foam models with an NC milling machine, and for communication between CAD/CAM-systems (i.e. STEP).
prof.ir. B.B. Schierbeek (project leader until 1.4.92) prof.ir. P. de Ruwe (project leader from 1.4.92) dr. lS.M. Vergeest
ir. J.l Broek ir. A.F. Lennings ir. C.G.C. van Dijk (AIO)
drs.
c.F.
Louwe Kooijmans (OIO) drs. A.E. Vries-baaijens (AIO) /DE-b. CAD for conceptual designThis project aims at exploring the possibilities of applying CAD-techniques in the conceptual phase of the design process. In particular the following topics are inves-tigated: developing design rules for modeling, assessing cost information, creating and accessing data collections, and the use of expert systems. Also organizational aspects of CAD are investigated.
Staff (5.1 fte):
prof jr. B.B. Schierbeek (project leader until 1.4.92) prof. ir. P. de Ruwe (project leader from 1.4.92) ir. A.P. Bremer
ir. T.J.A. Haan ir. J.F. Prins ir. R.W. Vroom
ir.S.H.P. van Zanten (AIO) dr. lS.M. Vergeest ir. ing. C.l Lupker
IDE-e. Ideate /Designer's Toolkit
CAD systems generally offer a very non-intuitive user interface. This hampers the interaction between designer and system, in particular in the early phase of the design process. Within this project interactive user interface and 3D sketch tech-niques are developed to provide a more intuitive user interface. The techtech-niques are implemented in a design framework, the Designer's TooIkit.
Staff (3.2 fte):
prof. lM. Hennessey (project leader) R.Kolli
dr.ir. M.J.O.M. van Emmerik ir. O.J. Pasman (AIO) ir. M. Oribnau (AIO)
TM/-a. Geometrie modeling and product modeling
Within this project CSO modeling techniques have been developed for input, visualization of solid models and for finite element mesh generation form
eso
modeis. Emphasis is now on feature modeling which offers a more complete repre-sentation by combining a geometrical description with a functional description of a product.
Staff (3.5 fte):
prof. dr. D.J. McConalogue (project leader) prof.dr.ir. F.W. Jansen
dr. W.F. Bronsvoort ir. M. Sepers rr.E.Boender(AIO) rr. R van Kleij (AIO) ir. M.H.J.P. Dohmen (AIO) ir. K.J. de Kraker (OIO) ir. W. van Holland (AIO)
TM/-b. Visualization
Visualization is important for designers, both to support the interactive design of geometric models as weIl as for the presentation of the design and its properties to other participants in the design process. Within this project, techniques are being developed both for 3D interaction, realistic rendering of 3D models and for visuali-zing multi-dimensional data (e.g. the results of a finite element analysis).
Staff (3.0 fte):
prof.dr.ir. F.W. Jansen (project leader) ir. F.H. Post
drs. P.R van Nieuwenhuizen ir. AJ.S. Hin (AIO)
ir. AJ.F. Kok (AIO) rr. T. van Walsum (AIO) ir. W.c. de Leeuw (OIO)
TM/-c. Application of knowledge-based techniques
Earlier work inthis project was on scheduling for manufacturing and on error-con-trol in flexible manufacturing systems. Emphasis is now on AI-techiques that can enhance and complement numerical techniques for design and optimization, in par-ticular case-based and constraint-based reasoning techniques for conceptual design. Staff (3.7 fte):
prof.dr. H. Koppelaar (project leader) prof. dr. D.J. McConillogue
dr.ir. H. de Swaan Arons dr.ir. RA. Vingerhoeds
ir. E.P. Jansen dr.ir. E.J.H. Kerkhoffs drs. J. Stigter (AIO)
ing. P.R.van der Weerd (AIO)
ir. B. Goedhart ir. B.D. Netten
The Aircraft Design and Analysis System (ADAS):
An overview
C.Bil
1. Introduction.
Aircraft design follows a top-down approach: in the pre-design stage the aircraft configuration as a whole is defined and optimised to best meet the specified requirements. This process is very complex because changes in the design will effect all the major characteristics of the design, such aerodynamics, weight and balance, stability and con trol, field performance, economics, noise, etc. (interdisci-plinary). Decisions have to be made on which design parameters are important, which criteria define an optimum design and can performance requirements be changed in favour of a better overall design ("trade-off studies"). The designer is faced with all these issues at a point where only very little information is known and where only fust-level methods are available to analyse the design with a rea-sonable degree of accuracy. In 'down-stream' stages, design is focused on detail and the process becomes more mono-disciplinary. At this stage the design configu-ration will be frozen and the engineering cost will increase exponentially. Major changes in the configuration become very costly and are undesirable. This means that it is essential for an aircraft to be economical and competitive that the design concept is sound and well-balanced.
The development of the Aircraft Design and Analysis System (ADAS) started in 1982 with the objective to establish a computer-based design methodology specifi-cally for conceptual and preliminary aircraft design1,2,3,4;5,6. Initial development was carried-out on the so-called Interfaculty CAD-Installation (ICI).
2. The aircraft design process.
The first step in the development of ADAS was to analyse the basic activities that take place in a typical aircraft design process. Because design is essentially trans-ferring an idea or a concept into something substantial, ADAS must allow the designer to remain in the loop. The design process usually starts from a basic description of the mission and performance requirements: the design specification. There are many different design concepts that would meet those requirernents and at this stage it is the objective to find the optimum configuration. The designer will
look at different layouts and will also determine the best wing size and shape for the mission ("sizing"). This process is iterative, as shown in Figure 1, and starts
no TENTATIVE DESIGN DESIGN DEFINITION DESIGN __ ANALYSIS ANALYSIS - METHODS DESIGN EVALUATION
+
BEST DESIGN DESIGN REQUIREMENTS/ FIGURE OF MERITFigure 1: Basic steps in an aircraft design cycle.
from a tentative design based on experience and perhaps a few basic calculations ("design definition"). Using suitable methods this design is then analysed in more detail to asses its performance ("design analysis"). The results are evaluated and compared with the predefined requirements ("design evaluation"). If the design is not satisfactory some design changes have to be made and the process repeats. This procedure is commonplace in traditional design but would also fit very wen in a
computer-based design system. Moreover, the high-speed computing power intro-duces opportunities for a more systematic design approach, such as explicit and implicit design optimisation. In explicit optimisation the system generates a large number of designs by systematically changing selected design parameters. Design trends and characteristics can be visualised through computer graphics. The princi-pal advantage of this approach is that it gives sensitivity information useful for trade-off studies. In implicit optimisation or multivariate optimisation the design process is basically driven by an optimisation algorithm ("optimiser') which searches for the optimum design according to a specified objective function and constraints. This approach allows many design parameters to be optimised but only gives results on one design. These procedures are to a large extend complementary and a design system should incorporate both features as an option.
3.
TheADAS system architecture.
The ADAS system architecture is schematically illustrated in Figure 1. The system
t
@jJ
Dl
r
T
rutoPLOT,.-~~aJ
files .-~
ft
ADAS Executive Program Figure 2: ADAS system architecture.comprises several components which will be discussed in more detail in the fol-lowing sections.
3.1. Geometry interface.
Most of the information required in analysis is related to the geometry, i.c. dimen-sions, areas, volumes, etc. Within ADAS the user can define a design by a3-view configuration drawing created with a standard CAD-package (Figure 3). Initially
ADAS was coupled to MEDUSA, a drafting and modelling system available on the ICI, but later ADAS-versions were converted to DXF. The ADAS/AutoCAD
Figure
3:
Example ADAS/AutoCAD 3-view configuration drawing. interface is based on a protocol that requires drawing entities to be labelled with specific layer names. From this layer name definition the interface module can rec-ognise basic drawing entities as particular aircraft components and the relevant geometric properties are available in subsequent analysis computations.3.2. ADAS programming interface.
In principle the ADAS-system regards analysis methods as input information to be supplied by the user in the form of a Fortran-coded design prograrnme. Although this requires some programming skills, it increases the flexibility and applicability of the system considerably. However to avoid duplication of effort, standard analy-sis methods can be selected from a program library ("method base") and used as building-blocks. Ideally the program library will contain modules which not only cover a wide gamut of technical disciplines but also have different levels of accu-racy, so the designer can select a method that is best suited for the information available, accuracy required, CPU-demand, etc. Most of the current ADAS devel-opment effort is concentrated on extending and upgrading the program library.
3.3.
The ADAS executive program.
Once a design program is completed. tbe next step is to compile it and link ino tbe ADAS executive program. Tbe executive program controls the processing of a user-supplied design program according to a specified control mode (Figure 4).
no
no
no
yes
no - -....
---I.
IP- design point analysis mode lP- optimisation mode _ _ ~.~ parametric survey modeGD
The design program (DSPROG) is ealled at different 'strategie' locations in the program flow whieh allows the user to arrange eomputations for eaeh step in the analysis process. By including specifie subprograms whieh read neeessary infor-mation and automatieally set the exeeutive program to run in the selected eontrol mode. If neither of these modes are selected the design program will simply be executed for only one design ("design point analysis").
4. Design evaluation.
The final step in a design eyele in the design evaluation phase. ADAS features sev-eral options to retrieve and display analysis results. A separate program, ealled AutoPLOT, allows the user to ereate several types of engineering graphs from gen-eral parametrie data. An example of a typical representation of a parametric data.
12 11 10 9
.s
-Q2
8I
til 7-<
6 5 4 3200 3400 3600 3800 4000 4200 4400 Wing Loading (N/m2)Figure 5: A typical representation of parametric data in the farm of a contour plot with constraints.
The graphs can be imported into AutoCAD for display wherein they can be edited and generally prepared for publication. Most wordprocessors can import graphics files directly in to the text.
ADAS can also transfer geometry information back into the original configuration drawing. This would very useful in the case the geometry has changed for example by implicit optimisation. Optionally a 3-dimensional model can be created auto-matically by association of the 3 views. Such a model can be imported into Auto-CAD and where it can be viewed from various angles or rendered with different display techniques, as shown in Figure 5.
Figure
5:
Sluuled image of a 3-dimensional model of an unmannedflight vehicle generated with ADAS7.5. Conclusions.
The Aircraft Design and Analysis System (ADAS) has come a long way. Since its conception now almost 10 years ago, the system is been effectively used in student project work for various types of applications. ADAS has attracted international attention both from the academic world (RMIT, Australia) and from industry (IPTN, Indonesia). Short courses have been organised and presented at various institutions (RMIT, University of Glasgow, TU Berlin, Institut Teknologi Band-ung).
Tbe general architecture of the system has not changed significantly over the years and current development effort is mainly directed towards enhancing and upgrad-ing the program library. This is particularly evident in a research project in cooper-ation with the Ncooper-ational Aerospace Laboratory (NLR) and the structural design group at the Faculty of Aerospace Engineering8. This project aims to adapt and implement CfD- and FEM-based methods with the objective to involve structural design earlier in the design process. This project has currently entered its comple-tion phase in which the system will be applied to the Fokker-SO aircraft as a test case. Fokker Aircraft is offering assistance by providing valuable data on the air-craft structure and on test results.
6. References.
1. Bil, C.: Applications ofComputer-Aided Engineering to Subsonic Aircraft
De-sign in a University Environment, ICAS-paper 86-3.1.1, ISth Congress of the International Council of the Aeronautical Sciences, September 7-12, London 1986.
2. Bil, C.: Development and Application of a Computer-Based Systemfor
Con-ceptual Aircraft Design, ISBN 90-6275-484-8, Delft University Press, Delft
1988.
3. Bil, C.: ADAS: A Design System for Aircraft Configuration Development, AIAA-paper 89-2131, AIANAHS/ASEE Aircraft Design, Systems and Oper-ations Conference, July 31-August 2, Seattle 1989.
4. Bil, C., Middel, 1.: Some new developments on the Aircraft Design and Analysis
System (ADAS), ICAS-paper 90-2.6.4, 17th Congress of the International
Council of the Aeronautical Sciences, September 9-14, Stockholm 1990. S. Bil, C.: Aerodynamic Analysis Integration in a Computer-Based Aircraft
De-sign System, paper 3A-3, International Aerospace Congress, May 12-16,
Mel-boume 1991.
6. Bil, C.: Aircraft Design and Analysis System (ADAS), Users Manual, Report LR-HO, Faculty of Aerospace Engineering, Delft University of Technology, November 1992.
7. Thompson, L.A., Abanteriba, S., Bil,
c.:
A Multi-purpose Autonomous FlightVehicle System, Sth Australian Aeronautical Conference, September 13-15,
Melboume, 1993.
8. Bil, C., Dalen, F. van, Rothwell, A., Arendsen, P., Wiggenraad, 1.F.M.:
Struc-tural Optimization in Preliminary Aircraft Design: - A Finite Element Ap-proach -, 18th Congress of the International Council of the Aeronautical Sciences, September 20-2S, Beijing, 1992 (also NLR report TP 92459L).
Aircraft Aerodynamic Design
J. Middel Foreword,
Computer-Aided Engineering (CAB) is still not generally accepted in tbe concep-tual and preliminary design phases. Tbe intuitive and heuristic nature of tbese ini-tial phases of tbe design process conflicts witb tbe formalized structure of CAB. However, 80 percent of tbe overall design costs are related to decisions taken in these design phases, and CAB is cost effective in the downstream phases and improves design quality. Tberefore, the computerized Aircraft Design and Analy-sis System (ADAS) for the conceptual and preliminary design phases has been developed, witb tbe aim of reducing overall design costs and improving design quality.
Tbe initial development of ADAS was focused on its configuration, conforming witb tbe established conceptual and preliminary design approach. Tbe aircraft analysis methods built into ADAS were still based on traditional semi-empirical methods. A major drawback of these methods is that they are based on existing technology and apply to conventional aircraft configurations only. Advanced and unconventional aircraft configurations cannot be handled.
However, ADAS allows to implement more advanced calculation metbods tbat are already established in downstream detailed design phases, effectively leading to a more integrated design approach. This paper discusses ADAS, focusing on tbe implementation of an aerodynamic tooibox based on fundamental aerodynamics ratber tban semi-empirical methods.
Summary
ADAS (Aircraft Design and Analysis System) is a computer aided design tooI, providing a user-friendly environment for aircraft multi-disciplinary conceptual and preliminary design.
This paper discusses an aerodynamic tooibox that is based on fundamental aero-dynamics rather than semi-empirical methods, and which is implemented into ADAS. Tbe tools are based on a panel method, a finite element like method for cal-culating the potential flow around (lifting) geometries.
This tooibox improves the quality of the conceptual and preliminary design phases: an aircraft design can be analyzed more tboroughly at an early stage. This may yield a more mature design, improved design quality and reduced overall costs. Moreover, unconventional design configurations can be investigated too.
This aerodynamic tooibox is developed to support two kinds of applications: The standard ADAS application where the designer supplies a design program with caUs to tooibox (library) stored subroutines, and, several resident stand-alone tools for standard applications, e.g. analysis of the surface pressure distribution.
Two applications show the abilities of the aerodynamic tooibox: The sizing and positioning of the canard, main wing and tail of a "three-surface" aircraft, and the modification of airfoil sections close to the wing-fuselage junction, compensating for wing-fuselage interference effects.
General ADAS concept.
Tbe Department of Aerospace Engineering of the Delft University of Technology avails of the Aircraft Design and Analysis System (ADAS) for research purposes into conceptual and preliminary design. This system is distributed over several computers: a CONVEX supercomputer for number crunching, and SUN and Sili-con Graphics workstations for pre- and post-processing and smaller design tasks, all running under the UNIX operating system.
This system consists of several modules (fig. 1):
~
Q
....
Cl
t
t ,
t ,
-rJITl--blUJ-
,
I - - - ,I
I-Cl
Q-I~I
I I
I~I IINGRES! - . . . . - L---AVS-J
fig. I. ADAS structure.
• A 2-D CAD drawing module (MEDUSA 14.2) for definition of a baseline geometry and data visualization. The baseline is defined in a standard layout, i.e. top and side views and optional fuselage cross-sections, including reference labels (fig. 2). These labels refer to library stored standard items, e.g. airfoils and engines. This module also displays numerical results, selected and retrieved from the database, in various graphical engineering formats.
• A relational database system (INGRES) for the preparation, storage, selection, combination and retrieval of numerical data. INGRES is used to store design data (e.g. design speeds and altitudes), engine and airfoil data. and optimization
_.cc
A2
fig. 2. ADAS aircraft definition sheet lor aerodynamic cakulations.
and parametric variation data along with design monitoring and survey data.
• The ADAP program controls the design process. At start-up this program reads the sheet and database, links the reference labels to the associated data, effec-tively building an intemal model of an airplane, and forwards this model and control parameters to the design program. This design program is provided by the user (designer) and contains the design strategy, design knowledge and design decisions. This major feature of ADAS exempts the designer from using a hard-wired, rigid design method. Within ADAP, several design modes are available: single analysis, optimization, parametric variation or some combina
-tion. The design mode is selected depending on the contents of the optimization and parametric variation database tables. The calculated data may be transferred to the database and/or MEDUSA drawing.
• Severallibraries, containing pre-programmed design and analysis methods. These methods can he incorporated into the design program. These library stored, pre-compiled design and analysis methods allow to save considerable development time.
Functional requirements of the aerodynamic tooibox
The ADAS aerodynamic library is designed to allow the designer to conceive
tai-lor made tools to investigate the aerodynamics of both conventional and innovative aircraft configurations as weIl as to explore new design procedures, without exces-sive needs for computer requirements and software development time. This wide range of applications implies that the library should be equipped with both
analy-sis, i.e. calculation of the properties of a specified geometry, and inverse design functions, i.e. determination of a geometry that satisfies specified requirements. The ADAS parametric and optimization modes should he supported, for true multi-disciplinary design where the effects of global design variables such as aspect ratio, area and taper ratio of the lifting surfaces are investigated. Sensitivity of the design objectives and constraints to design parameters is determined by ana-lyzing the baseline and several, parameter driven, derivative aircraft configura-tions.
Moreover, detailed aerodynamic analysis and design should be supported, offer-ing local geometry refinement e.g. design of fairings.
Within the ADAS optimization/parametric variation loops, all tools should be able to operate autonomously and to be mixed with other, non-aerodynamic meth-ods. Usually user- interactive control is preferred if tools are operated in stand alonemode.
The library is set up as a "tooibox": The user may select from parallel, coexisting routines, fulfilling the same tasks but with different detail level or methods. Sets of routines may he functionally grouped, even stand alone tools are included or easily built.
Selecting basic analysis method
The scope of application exclude the common, weIl established, semi-empirical methods, which are based on fundamental relations dependent on a few global dimensions, matched with the real numbers through statistical "scaling" factors, derived from data on existing aircraft. Hence one is forced to opt for more funda-mental (numerical) analysis methods. The selection of the basic method should consider the required computational resources, which should be limited, especially regarding optimization and parametric variation. This effectively exc1udes compu-tational fluid dynamics that are based on volume or finite difference methods ..
A remaining option is the "panel method" as the basic analysis tooI i.c. NLRAERO. This program is developed by the "Nationaal Lucht- en Ruimtevaart-laboratorium". A panel method is capable of calculating the accurate velocity and pressure distributions over complex aircraft geometries in flow conditions where the potential flow model is justified: rotation free, isentropic, linear compressible, viscous free and ideal gas properties. Effectively, this limits the application to flows without strong shockwaves and where boundary layers are confined to the geometry surface. Furthermore, NLRAERO assumes that the lifting surfaces are thin and mildly cambered. However, geometry effects like fuselage displacement and interference effects, essential to the evaluation of many designs, are modeled.
The simplification of the governing aerodynamics yield high computational efficiency and low cost. The flow is determined throughout the physica1 domain requiring a computational grid at aircraft surface and wake (fig. 3).
fig. 3. Paneled aircraft configuration. Implementation
Panel methods are effective tools in the downstream design phases, requiring an accurate detailed geometry. However, at the conceptual design level, global dimen-sions, like aspect ratio and area, have to be established first before an accurate geometry can he defined. Furthermore, the conceptual designer is merely inter-ested in global properties like overall lift and drag, aircraft neutral point, and their optimum values, which are sometimes only a function of a few global dimensions, rather than pressure distributions.
To improve the effectiveness and efficiency the panel code some features have to he added or re-configured:
• An automated panelling function (grid generator) is provided for, yielding a finite element like model (grid) from the internal ADAS representation. This grid generator runs in (semi-) automatic mode. The grid is automatically refined at geometry areas with intersections and high curvature. In the stand-alone ver-sion or the ADAP analysis mode the user may influence the grid spacing by specifying the numher of panels in two directions, e.g. span and chord. While running in optimization or parametric variation mode, the grid is automatically adapted to the derivative configurations, approximately preserving local panel dimensions.
par-ticularly essential when running in the parametric or optimization design modes. These functions allow e.g. to change the aspect ratio, sweep, area, posi-tion of a lifting surface, and deftecposi-tion of control surfaces.
• The panel code itself is split into functional segments to allow generation and combination of multiple solutions, preventing recurrent tasks.
• A Trefftz-plane (momentum based) calculation is included to determine the induced drag. This theory allows to calculate the induced drag using the circula-tion distribucircula-tion in the aircraft wake in a suitably chosen (Trefftz) plane aft of the aircraft configuration. The method is an alternative for the surface integrated pressure distribution, but needs only the spanwise circulation and geometry to beknown.
• The viscous effects account typically for half of the overall drag. Two different viscous effects modules are present: A full three dimensional momentum inte-gral boundary layer program for detailed calculations, and a simple but fast vis-cous correction function. Both visvis-cous effects modules are based on the local velocity distribution, but the latter has a fast algorithm for lifting surfaces, with the viscous drag calculations based on a family of pressure (velocity) distribu-tions rather than actual ones. Required for each spanwise location are: locallift coefficient, free stream Mach number, (aerodynamic) local sweep, thickness Renolds number and chord.
• This assumed shape of the pressure distribution may also be used to determine the airfoil pitching moment characteristics. This airfoil family concept allows fast and easy assessment of the effects of airfoil variations, without the need for airfoil development.
• An optimization algorithm to determine the lift distribution yielding minimum
induced (plusviscous) drag is available. Both trimmed and untrimmed ftight conditions can be addressed. This allows to access the trim-related drag penal-ties and lift distributions.
This method is based on a Trefftz-plane method and the airfoil family con-cept, and shares the grid with the panel code. The lift and the induced and vis-cous drag are expressed in the unknown spanwise circulation distribution. The fuselage pitching moment is externaIly supplied, usually from a panel method analysis.
• Customized interfaces for color-coded visualization of pressure and velocity distributions have been developed (AVS) have been developed to improve design interpretation, in addition to the standard ADAS data visualization meth-ods.
Mter the first global dimensions are fixed, an important goal in the aerodynamic design is to find an aircraft (detailed) shape which fulfils such requirements as high maximum lift coefficient, high Mach number for which shockwaves first appear,
and low zero lift drag, appropriate boundary layer stability and stall progression etc. These requirements however depend on the appropriate swface pressure distri-bution. To avoid the design of the aircraft shape by trail and error, especially trou-blesome for complex configurations (e.g. wing-body- tail-nacelle combinations), an inverse method which computes the swface geometry for the specified pressure distribution has been developed.
The method adopted here is a geometry mode concept, based on the panel code. The user specifies a target pressure distribution and series of geometry modes (shape changes) and their location. For each geometry mode with a yet fixed amplitude, its effect on the local pressures at the target distribution is evaluated. Finally an optimization procedure determines the amplitudes of each geometry mode, attempt to find the best fit of the resulting and the target pressure distribu-tion.
This inverse design problem is nonlinear and badly posed by nature: lt is a
pri-ori unknown if a feasible geometry yielding the target pressures does exist. Several
restarts, even redefinition of target pressures and/or geometry modes and areas may be necessary to obtain satisfactorily results.
Though ADAS users provide their own design programs, several recurrent tasks, employed by various users, exist. So several standard tools are provided, all oper-ating on the aircraft grid. Among these tools are:
• Pressure distribution analysis.
• Boundary layer analysis, in cooperation with the panel code itself.
• Determination of the optimallift distribution, for minimum drag conditions,
both trimmed and untrimmed ftights.
• A warping function to twist the lifting swfaces, imposing a prescribed spanwise lift distribution.
• Calculation of the configuration maximum lift (coefflcient).
• A trim function, where a prescribed lift and pitching moment are satisfied by angle of attack and control swface deftections.
"Three surface" aircraft design
The aviation industry is continuously developing more efficient aircraft. This
includes research into novel aircraft configurations, one of them being the "three-swface" aircraft. This configuration is distinguished from the conventional aircraft in having an additional horizontallifting swface (canard) in front of the main wing in addition to the main wing and horizontal tail swface, see e.g. fig. 3. The three
lifting swface layout offers several advantages over conventional designs:
• The extra swface of the "three-swface" aircraft allows to minimize drag throughout the entire ftight envelope, by selecting the optimal combination of
angle of attack, tail and canard (control) swface deftections. The lift distribu-tions over the two lifting surfaces of conventional and canard configuradistribu-tions in trimmed flight are fully determined for a given center of gravity and weight: The tailplane (elevator) deftection and aircraft angle of attack enforce the lift-weight and pitching moment equilibriums, however at the cost of trim drag. • Stability requirements usually restrict the aft position of the airplane center of
gravity. For conventional aircraft this of ten yields a download on the tail, increasing overall (induced) drag. For a canard aircraft, having a canard and main wing but no tail swface, this condition results in a highly loaded canard and due to its relative small span, high induced drag levels. A redistribution of the lift over canard and tail of the "three swface" configuration may result in a moderately loaded canard and reduced tail download.
• Additional benefits of the "three swface" configuration include: unobstructed pressure cabin and reduced weight. The main wing may be positioned aft of the cabin pressure bulkhead, leading to a much simplified wing-fuselage structural joint. Careful positioning of the main gear may allow simpier and more effec-tive load paths.
• Some disadvantages arise: The optimization of the lift distribution between
sur-faces requires a more complex control system, and the total wetted area may increase, adding viscous drag. The canard attachment may cause structural and geometrical problems: the pilots view might be restricted.
Evaluation of the "three-surface" aircraft concept
The present investigation concerns the selection of the wing, canard and horizontal tail major parameters, i.e. area, incidence and twist, of a "three swface" aircraft configuration. The major design challenge is to exploit the canard-main wing-tail-plane-fuselage aerodynamic interference effects: The canard downwash/upwash and fuselage displacement effects are leading design driving phenomena. The "three-surface" aircraft aerodynamic concept is investigated in two ways: • Evaluation of a family of aircraft at design cruise conditions, design weights
and design centers of gravity. Each member has a different combination of
canard, tail and main wing area and main wing longitudinal position.
• The drag polars of are evaluated for a few aircraft. Drag polars are determined for trimmed conditions with varying weight and center of gravity. Three
differ-ent methods for trimming are investigated.
A total of 100 configurations are derived, based on a single "three-swface"
base-line design. This family spans from near-canard to near-conventional aircraft
con-figurations, the fuselage is kept constant. The areas of the main wing, canard and horizontal tail are varied such that the minimum (stall) speed, pitch-up capability
and cruise conditions are (approximately) preserved. This reduces the independent design parameters to the canard area and the main wing longitudinal position.
Tbe baseline and an extreme aircraft derivative are depicted in fig. 4 and fig. 5.
fig. 4. Baseline paneled aircraft.
fig. 5. "Extreme" configuration.
Among the data determined for each configuration are:
• Aircraft weight and center of gravity, based on semi empirical weight formulae. • The neutral point and statie stability margin.
• The optimallift distribution yielding minimum tota! drag for trimmed cruise condition.
• Planform lift coefficients.
The results for the design conditions are shown in a contour plot (fig. 6). Along the horizonta! axis the canard area is depicted, along the vertieal the main wing posi-tion. Lines for constant drag and statie stability levels are ineluded. It shows that absolute minimum drag is given for a canard configuration. However, this configu-ration does not meet the (inherently) stability requirement. For typical statie
stabil-2 2.5 3 3.5 4 4.5 5
5 (corwrd) (11121 roos CII2)
- - - StablLity __ gin
- - - StabHlty __ gin FLootlng canard
- - - Cl Ctol LJ
- - - ClCconord)' CLClIOln "Ing)
C .... ter of grovlty ~Ing on colculot-.l cOIIf'O"*'t ".'ghta
o
Sos.l,,,.
design-, ,
!
Afig. 6. "Three surface" aircraft drag and stability characterisncs.
ity margins, tbe lowest drag is achieved for tbe conventional aircraft. The de-stabilizing effect of tbe canard surface is also apparent from tbe statie margin dif-ferences from tbe ftoating and fixed incidence canard arrangements. Hence tbe "three surface" configuration might be interesting provided statie stability margins requirements are relieved.
Moreover, aircraft hardly ever fty at a single design condition. In practice, tbe payload and initial fuel weight will vary from ftight to ftight. The fuel weight even varies during each ftight. Therefore, aerodynamic performance at off-design condi-tions are investigated too. Por some aircraft derivatives, tbe lift and drag character-istics are determined as functions of varying weight and center of gravity. Por each aircraft, tbe lift distribution is optimized at tbe design center of gravity and design weight. Por each center of gravity and weight, tbe aircraft is trimmed and analyzed. Three different trimming metbods are considered. Por all aircraft tbe angle of attack is optimized and trim is achieved through:
• canard and tail, minimizing overall drag. • canard.
Fig. 7 shows the resultant overall drag as a function of lift and center of gravity. To
-.
... lil baseline trlMe<! b!I.
~ - - canard e tolL
...
canard enlll toll enlll ---'I no tr 1 _ cend i t I en be!iond thl. point <-Stoble I t.nstoble -> I:
/~ ' " I ~ .... ' ."","""" "---I -0.75 -0.5 -0.25 0 0.25 0.5 0.75 center oF 9"0yltll shiFt Clllwlth respect to the design center oF 9"0ylty fig. 7. Lift and center of gravity dependent drag.
prevent unrealistic results, the local (sectional) lift coefficient is constraint. Appar-ently, trirnming with both canard and tail at typical cruising conditions typically save up to 4 percent in drag compared to the altemative methods. Trimming using the canard only at typical cruise lift coefficients, gives a lower drag penalty than the tail trimmed case. Apparently, changes in the negative tail-Ioad cannot outper-form slightly positive canard lift variations. At higher lift coefficients the tail trimmed variant gains over the canard trimmed one. The canard is already highly loaded, while the tailload may be positive. The center of gravity range is increas-ingly restricted with increasing weight: The maximum (negative or positive) sec-tionallift coefficient is reached somewhere.
Corrections for wing-fuselage inlerference effects
This application shows the detailed aerodynamic design of a fuselage-wing junc-tion of a small single engined aircraft with a high wing arrangement (fig. 8). Basic (laminar flow) airfoils, wing planform, twist and fuselage design have been fixed. At the fuselage-wing intersection however, the local geometry should be tuned to cope with the fuselage turbulent boundary layer which requires a significant change in the target pressure distribution to prevent early flow separation.
Further-fig. 8. High winged aircroft. with location for redesign.
more, tbe fuselage interference effects cause a loss of lift, which should he regained if possible.
First, a small aircraft witb a high wing arrangement is prepared in an ADAS input sheet, and a panelled (FEM) model generated. Next, tbe chordwise pressure distribution on tbe wing lower side adjacent to tbe fuselage is analyzed (fig. 8 and
-cP
\ • 2 weight factors along chord.
\.0 0.8 0.6 O.Q 0.2 -0.2 -O.q -0.6 target pressures. resulting pressures. 0.\ 0.2 0.3 O.Q 0.5 0.6 0.7
fig. 9). Subsequently, the designer sets a target pressure distribution at this location,
and defines the allowable local geometry shape functions. In this case the lower
airfoil side is modified, using 6 different shape functions. The inverse design tooI determines the amplitude of each shape function. The resulting pressure
distribu-tion is compared to the target one and the geometry is checked for geometrica!
con-straints. This process is repeated 3 times until satisfactorily results have been
obtained (fig. 9). Fig. 10 shows the initia! and revised airfoil cross section .
...
;:: 0.11.1
-0.1
Interactive programs for aircraft structural design and
optimization
A. Rotbwell
Emphasis in this research is given 10 tbe development of programs to serve as a direct aid to tbe designer in different stages of the design process, in which a computer-generated structural model provides geometrie (and other) data input, and in which tbe current status of a design can be displayed in graphical or numerical form at any stage of tbe design. Optimization procedures (single- or multi-Ievel) are employed to search for a feasible and efficient solution to tbe design problem, while allowing tbe designer the freedom to steer tbe design by suitable choice of constraints imposed on it. The work carried out can be described under three headings:
development of a set of interactive, special-purpose programs for specific detail design problems, including programs for a wing and fuselage cross-section.
use of graphics pre- and post-processors coupled to a finite element analysis for rapid model generation, sensitivity analysis and optimization of complex structures. The current application is to tbe design of tbe cut-out for a fuselage door.
development of structural sizing and optimization routines for incorporation into tbe aircraft preliminary design program ADAS. Aim of this work is a multi-disciplinary optimization witb aerodynamic, structural, aeroelastic and otber constraints.
Each of tbese is described in more detail in tbe following three sections.
1. Detail design programs
For some commonly occurring problems in aircraft structural design, the layout of tbe structure can be described in a general enough way to apply to a wide range of designs, and at tbe same time tbe usual engineering formulae are adequate for much of tbe structural analysis. An automated design procedure can tben be botb quick and highly effective. Examples of such problems are tbe
design of a wing and fuselage cross-section. The extemal shape varies according to the chosen wing profile, or fuselage type. The structural fonn is typically a thin shell reinforced by stringers in the longitudinal direction and transverse ribs or frames, for which the design freedom is primarily the skin thickness and the design of the reinforcement. The conventional design procedure is an iterative process, in which the margins of safety are progressively brought to the same level in different failure modes under the specified loading cases. However, if an automated design procedure can be coupled 10 an optimization routine this assists in finding a feásible solution as weIl as directing the solution towards a minimum weight design. Programs developed for this purpose can relieve the designer of much tedious, repetitive work, enable a wider range of designs 'to be explored, and are valuable in parametric studies at an early stage of a design. An essential requirement is that the designer must remain in control of his design at every stage, and be able to adapt it as necessary. An interactive program enables him to monitor the design, allows him to steer the design by modifying constraints and imposing additional dimensional restrictions, and provides a user-friendly input and output. Development of suitable programs also requires a precisely defined design methodology and can, therefore, stimulate development of the design process itself.
Design programs have been developed for a wing [1](2) and a fuselage (3)
cross-section. The generic form of structure for a wing cross-section (i.e. the torsion box of the wing) is shown in figure 1. The extemal shape of the cross-section may be input in a variety of ways, and displayed on the screen. To reduce the
number of design variables, standardised shapes of stringer are used (corresponding to tbose found in practice) and restrictions are placed on tbe number of different skin thiclrnesses. The loading is defined by various combinations of shear force, bending moment and twisting moment on tbe
cross~section. For tbe fuselage tbere is also tbe loading due to cabin
pressurisation which, due to tbe presence of frames and stringers, causes a complex stress distribution in the fuselage skin. Constraints include stress limitations for static strength and fatigue, buclding and post-buckling behaviour, and stiffness requirements. Optimization takes place by a two-level procedure, in which individual stringer/skin panels are optimized at tbe lower level and assembied into tbe complete structure at tbe upper level. Control of the design is achieved primarily by selecting design variables to be either active or inactive, and by choice of upper and lower bounds on dimensions. The
programs are fully interactive, input of data being in response to prompts which
appear on the screen. The user can choose between tbe options offered by various menus displayed by the program. Figure 2 shows tbe menu structure of the program WingDesign.
Main Menu
t - -.... Bd1t Menu
§
l\etum to Main Menu Bd1t pometry Bd1t dimensions Bd1t IIIIIterialaBd1t design requi~ta 1---.... OptimizaUon Menu
t=:
l\etum to Main MenuSelect design variables and bounda Optimize
t - -.... l\esults and Save Menu
§
l\etum to Main Menu Show status of the design Print results"rite LeIS __ ero Sa". to 111e ' - - -... Job Menu
I---.l\etum to Main Menu
~prOCeed .ith a ne. design session
The further development of tbese detail design programs will he into an integrated, more versatile package AeSOpS (Aerospace Structural Optimization System) in which specific analysis routines used in tbe design can also be accessed. The programs are at present restricted to metal structures, for which suitable crack growtb and damage toleranee routines will be incorporated. Metbods are also heing investigated to extend tbe use of tbe programs to composite structures, including laminate optimization. Appropriate routines for tbe design of ribs and frames are being developed. These improvements have to he accompanied by a more powerful multi-Ievel optimization procedure. The proposed organization of AeSOpS is shown in figure 3. The wing and fuselage design programs are in regular use in design work at Delft and elsewhere, and have already demonstrated tbeir usefulness. With the improvements referred to above, a highly versatile package should result.
2. Fuselage cut-out
damage toleranee
analysis
Figure 3. Organisation of AeSOpS. laminate
design
While for tbe type of design problem in the previous section a standardised geometry was satisfactory, for more complex structures such as tbe wing-fuselage connection or the reinforcement around a cut-out for a wing-fuselage door this does not offer sufficient design freedom. Furthermore such problems do not lend tbemselves to analysis by tbe conventional engineering formulae; a finite element analysis is called for. Although tbis has become tbe standard method for tbe analysis of structures of complex shape, its use in design is somewhat
more problematic. It is, of course, possible to perfonn an iterative procedure in which the structure is re-analysed a number of times, the structure being increased or reduced in thickness (or other dimensions) at each step according to the stress in that part. However, this procedure does not necessarily lead to an optimum, producing in general an adequate structure (i.e. one with the required minimum strength) but not always one of minimum weight. A fundamental difficulty in the above procedure is that, in a statically-indetenninate structure, the stress in a particular part is affected not only by the part itself but also by changes in allother parts of the structure. This makes an intuitive re-sizing to achieve the required stress levels throughout the structure virtually impossible, or at least highly time consuming. The alternative is to make use of a so-called sensitivity analysis, in which the change in stress in all parts of the structure resulting from some change in thickness (or other dimension) in any component of the structure is calculated. Use of computer graphics becomes essential to handle the large quantity of data generated and to present this data to the designer in a convenient fonn to enable him to improve his design. Sensitivity data provides a basis for "what-if' studies to predict the effect of design changes. The same sensitivity data can also be used in a fonnal optimization procedure for a minimum weight design satisfying specified stress constraints. This can serve as the starting point for a more practical design, with the benefit of a sensitivity analysis to identify critical areas of the structure.
The program CUFUS [4] has been developed for the design of a cut-out in a
fuselage, implementing the sensitivity analysis referred to above. The program uses MSC/NASTRAN for the finite element analysis, together with PATRAN for pre- and post-processing. Maximum use is made of the graphics capability of the latter software package. For this reason CUFUS is wriuen almost entirely in the PCL-Ianguage offered in PATRAN. All software produced is in addition to available commercial software, i.e. no source code modifications of the commercial software are called for. CUFUS offers valuable facilities for model development and modification. This is particularly useful in a design program for a fuselage door, when specific cu stomer requirements can lead to frequent design changes. The program enables the user to define the shape of the cut-out and the geometry of the fuselage containing it, including the position of stringers and frames, and to define doubler-plates and edge members reinforcing the cut-out. The finite element mesh is generated automatically, and built-in mesh refinement procedures enable .the user to obtain more detail in critical areas such as the corners of the cut-out. The program also enables standard fonns of cross-section for stringers, frames, etc. to be selected, as
illustrated in figure 4. The background of this figure shows the finite element mesh (one quarter of the structure only) with the cut-out located at lower-right. A typical sensitivity plot (figure 5) shows contours of the change in principal stress in the structure due to a given increase in thickness of a doubler-plate at the corner of the cut-out (the actual screen display in colour is easily readable!). Qnly the part of the structure near the rounded corner of the cut-out is shown in the figure. A highly complex behaviour is observed, with reduction in stress at the doubler-plate itself but significant increases (positive contours) elsewhere in the structure. This confrrms the earlier statement that an intuitive design process would be ineffective, if done without the benefit of a sensitivity analysis.
The program described above is specifically for cut-outs in a fuselage structure. However, many of its felj.tures are valuable for a wider range of design problems, and these are being incorporated into a "tooI box" of routines for use in the development of programs for other structural design problems. The tooi box will include mesh adaption routines, routines for generation and display of sensitivity data, and routines for easy property definition.
3. Multi-disciplinary design
In the traditional aircraft design process the external form of the aircraft is fixed at an early stage, primarilyon the basis of aerodynamic and performance requirernents. This has to be based on empirical estimates for the weight of the structure, the actual design of which takes place at a later stage in the whole design process. However, there are considerabie advantages to be gained frorn a more integrated approach. By carrying out some preliminary structural design simultaneously with other aspects, a better compromise can be reached between the conflicting requirements of aerodynamic and structural design. Aspecific example of this is aeroelastic tailoring, in which the flexibility of the structure is used to improve the flight characteristics of the aircraft. In general, parameters such as wing thicknesslchord ratio and sweepback angle have major, but conflicting, influence on both structural and aerodynamic design. If the full advantage is to be gained from the use of composites and other new materials in aircraft design, an integrated approach to design becomes unavoidable.
In a program(4)[Sl developed in collaboration with the National Aerospace Laboratory (NLR), a single finite element type of model is used for both aerodynamic and structural calculations, with a reduced version of the same
~=
... _
....
__
...
_
.
Figure 4. Menufor standard sections used in CUFUS.
, ft . . . . "'DI
-,..
-..
_.
nu_
.....
--j~ .. , .... ..,..,.-."...
..
_:
• • _ . . , . . . . 1Figure 5. Sensitivity plot obtained by CUFUS.
...
...
""'" ---~. -... nmodel for aeroelastic analysis. The model is produced by an automatic mesh generation routine from a configuration drawing of the aircraft in AutoCAD. Modifications of the model are directly reproducible in the configuration drawing, and vice-versa. The user has the freedom to adapt the model, place additional structural elements and make other changes as necessary. Changes in configuration, such as change in sweepback angle, are automatically reflected in the mesh. Figure 6 shows a typical mesh generated by the program. Information
Figure 6. Au/oma/ic mesh genera/ion in ADAS.
supplied by the user with regard to materiais, forms of construction and other data is stored in a data base, together with the results of the various types of analysis. Loading on the structure is deduced from a pressure distribution on the aircraft obtained from a panel-method program, and from the mass distribution which as weil as payload, fuel and systems also has to include the mass of the structure itself. The design of the structure is based on a two-Ievel optimization. At the upper level a structural analysis is performed with the finÎte element program B2000, in conjunction with the optimizer B20PT. The design of the structure at this level has to satisfy strength, stiffness and aeroelastic requirements. Strength requirements are based on allowable stresses deduced at the lower level, which largely makes use of the conventional design formulae. The analysis at this level includes buckling, fatigue and damage toleranee. Optimization of individual panels, shear webs and other structural elements determines a maximum allowable stress consistent with constraints placed on the design.
Tbe weight of tbe structure obtained by the above opnmlzation process is retumed to ADAS (Aircraft· Design and Analysis System) which contains appropriate routines for the overall design of the aircraft, including perfonnance calculations. Tbe organisation of the structural design procedure witbin ADAS is illustrated in figure 7. In ADAS a third (system) level ofoptimization can be
L
_.r~~~~~~;L4 Leyel 1__
-E·tructura' de.lgn ADAS: alrcraft g.am.try and m •• h gen.ratlan d.tall _ _ _ _ _ . , . _ _ ->L",e!.:!ye.,..I ... 2..,.. _ _ _ d ... ign aera.lastlc analysls '--r---'--:c==an::.:r.r=alnt d.flnitlon: 1 . / - -... - - -.. allowobl • • tr . . . ond o.!!,!1 ~fI~"!:l.. _ compo.lt. lay-upFigure 7. Multi-Ievel structure of ADAS.
carried out, with any objective chosen by the user. Frequently this is maximum take-off weight but could also be, for example, range or direct operating cost. Extensive use of computer graphics is made at this stage, so tbat tbe user can explore fully tbe characteristics of the design problem. As alternative to a fonnal optimization, tbe user may perfonn parameter studies to improve tbe design. Tbe same graphics routines can be used to display the results of such parameter studies, as weIl as tbose of the previously described structural design and optimization. Studies are currently being made - based on actual aircraft data - to test tbe effectiveness of the system, and to investigate to what extent tbis integrated, multi-disciplinary approach can achieve some improvement in aircraft design.
Acknowledgement
Tbe work described in section 2 was carried out with the financial assistance of Fokker Aircraft bv, and that in section 3 under contract to the National Aerospace Laboratory NLR. The author greatly appreciates the continuing support and technical collaboration of both organisations.
References
1. Zoon~es, R.P.G. and Rothwell, A. WingDesign: program for the structural
design of a wing cross-section. Report LR-627, Delft University of Technology, Faculty of Aerospace Engineering (April 1990).
2. Schilder, H.F. User's guide for the computer program WingDesign 4.00. Memorandum M-679, Delft University of Technology, Faculty of Aerospace Engineering (March 1994).
3. Pluim, M. Computer program for the analysis and optimization of a fuselage cross-section (in Dutch). Ir-thesis, Delft University of Technology, Faculty of Aerospace Engineering (January 1993).
4. Heerschap, M.E. An interactive computer aided design system for cut-outs in pressurized aircraft fuselages (to be published).
S. Bil, C., van Dalen, F., Rothwell, A., Arendsen, P. and Wiggenraad, lF.M. Structural optimization in preliminary aircraft design: a finite-element approach. Proc. 181h Congress Inl. Council of lhe AeronaUlical Sciences, pp.1505-1515 (Beijing, China, September 1992).
6: van Dalen, F., Bil, C., Rothwell, A. and Arendsen, P. Finite-element based preliminary design procedures for wing structures. To be presented at the 19th Congress Int. Council of the Aeronautical Sciences, Anaheim, USA, September 1994.
