Development of an ICAD Generative Model for Blended Wing-Body ...

AIAA 2002-5447

9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization 4-6 September 2002, Atlanta, Georgia

DEVELOPMENT OF AN ICAD GENERATIVE MODEL FOR BLENDED WING BODY AIRCRAFT DESIGN G. La Rocca, L. Krakers, M.J.L. van Tooren Delft University of Technology Systems Integration Aircraft department Kluyverweg 1, 2629 HS Delft, The Netherlands

Introduction Aircraft design is a complex process that requires, since the earliest phase of the project, inputs and guidelines from all disciplines involved. A good design is the result of a strong integration of all the disciplines during the whole product development. This integrated and collaborative approach in design is generally addressed as concurrent engineering. ∗ A tool to make the concurrent engineering approach feasible on a large scale, as required by a civil aircraft-like product development, is not yet available. One hot issue for this design approach is the capability in managing the huge amount of information relative to the many design Copyright 2002 by the Technical University of Delft, The Netherlands. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Conceptual Design Phase

List of Requirements

Diverging process

Design concepts

- Parametric multi-modeling Preliminary Design Phase

Abstract Aim of the EC sponsored project ‘Multidisciplinary Design and Optimization of Blended Wing-Bodies’ is the development and application of a fully integrated Computer Design Engine (CDE). TU Delft contributed to the project with the development of a Blended Wing-Body Multi-Model Generator, which is able to supply geometries and data to the analysis software, either COTS or tailor made, used by the various disciplinary groups in the project team (aerodynamics, structures, stability and control etc.). A full parametric definition of the aircraft has been implemented in the KTI ICAD environment. The ICAD Multi-Model Generator (or Generative Model) holds the ‘knowledge’ of the Blended Wing Body aircraft, such that consistent models can be generated, at different leve ls of fidelity, suitable for the various disciplines involved in the CDE. A large range of aircraft variants can be generated, just editing the values of the aircraft parameters, which are all collected in one single input file. The optimiser can change the parameters value within the optimisation loop, without the need for user interactive sessions. The generative model can be run in batch mode, even from remote sites.

concepts/variants, which potentially comply with the customer requirements. In figure 1, the typical diverging/converging character of the product design process is sketched 1,2. During the synthesis phase (diverging phase), the design team produces a set of possible concepts for the aircraft, or variants of previous assessed designs. The larger the set of available concepts/variants, the higher the chance to design a quality product. However, it should be considered that the properties of all these solutions have to be derived to allow for a proper trade-off. In this trade-off phase (converging) computer modeling and analysis plays a fundamental role. For each design

- Multidisciplinary analysis Converging process

- Trade-offs

Design solution

Figure 1: the typical diverging/converging design process. concept/variant FE models must be generated and analysed, CFD analysis must be set-up, cost, weight and manufacturing feasibility must be assessed, before the best solution can be selected. The multi-disciplinary analysis requires a lot of repetitive work. A significant part of the manpower is just spent pre/post-processing models or interacting with all the software and analysis tools. Problems of interfacing different analysis tool become an issue, especially when data must be exchanged across actual geographical distances: it should be considered that the companies nowadays are often spread all over the

1

American Institute of Aeronautics and Astronautics Copyright © 2002 by the author(s). Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Customer requirements

Disciplines silos Final configuration

item

Mass_(kg)

X_cg

Y_cg

Z_cg

GROUP_FUSELAGE_(left_half) TED_1_(half) 107.4 44973.1 -1250.3 1250.1 …….. ANTI-ICING-SYSTEM 240.0 12755.8 -6368.1 490.0 OPERATIONAL_ITEMS_(half) 157.5 3000.0 0.0 0.0 CABIN_ARRANGEMENTS_(half) 40.0 3000.0 0.0 0.0 FLUIDS_(half) 3.0 3000.0 0.0 0.0 GROUP_WING_(left_half) TED_4_(iw _ins) TED_5_(iw _out) ……. ANTI-ICING-SYSTEM_(ow) GROUP_WINGLET_(left_half) RUDDER ANTI-ICING-SYSTEM_(wl )

309.1 43256.9 -15067.4 292.0 41879.3 -20362.6

2543.5 2256.1

402.0

1671.6

front trim tank

right wing tank

rear trim tank

left wing tank

174.5 80.0

40304.5 -31062.3

49785.9 -39394.7 47962.0 -39490.0

DATA

Figure 3: paradigm of a Computer Design Engine 1.

Weight & balance

Aerodynamics

others

& EVALUATOR

Cost analysis

CONVERGER

ICAD MULTI-MODEL GENERATOR

Performance analysis

Concept generator INITIATOR

Structural analysis

Position of the ICAD multi model generator inside the MOB Computer Design Engine Primary scope of the EC sponsored project MOB - A Computation Design Engine Incorporating MultiDisciplinary Design and Optimisation for Blended Wing Body Configuration, is the development of a tool able to support aircraft design in a real concurrent engineering environment 3. The CDE is at date in the final phase of development. It mainly consists of a complex software tool that integrates in one flexible system all the resources required to perform the automated multidisciplinary design of a Blended Wing Body (BWB) aircraft. This CDE incorporates software tools for aerodynamic, structural, dynamic, and flight mechanics analysis (both COTS and in-house developed tools), which physically resides on different machines, distributed over the several MOB sites in Europe (The Netherlands, UK, Sweden and Germany). Special backbone software (SPINEware by NLR and NEC) has been employed to enable the link of software and computers among the different co-operating partners 4 .

At the heart of the CDE is located the ICAD Multi-Model Generator (MMG) in which the full parametrical description of the aircraft product resides (see figure 2). The MMG gets input (the values for the set of parameters) from the concept/variant generator and (re-)generates different models (aerodynamic, structure, masses distribution) for the various analysis tools incorporated in the CDE. The models consist of geometrical surfaces, numerical values, ASCII files etc. to feed the different analysis boxes (mainly FEM packages, CFD and panel codes). The output generated by the analysis tools is given back to an evaluator/optimisator, which on the basis of some target function, gives a response and issues a new set of input data for the Multi-Model generator (see figure 3). The CDE is definitely not intended to replace the functionality of software tools (i.e. PATRAN, FLUENT, COTS in general or in-house developed tools), but to

Aerodynamic Analysis

globe in various groups or teams of expertise. These challenges call for more powerful and flexible tools to handle the increasing complexity of the product design. Such tools should be able to integrate all the specialized tools from the disciplines into a transparent environment. It should allow designers to investigate the downstream effects of the decisions to be made in the early phase of the product design process. Finally it should be able to considerably reduce all the time consumed by repetitive work so that engineers can focus on the what-if process and more space is left to exploit design skills and creativeness.

4990.1 5531.1 A 2

GROUP_PROPULSION_(left_half) CENTER_ENGINE_(half) 3751.2 43758.0 0.0 4142.9 CENT_ENG_.. 980.7 43758.0 0.0 2185.7 LEFT_ENGINE 7502.3 39750.0 -7501.0 5410.5 LEFT_ENG_STRUC….. 1961.3 39750.0 -7501.0 3453.2

V = A

2 3

b(A +A 1

+A 2

A ) 1

2

1

b

GROUP_LANDING_GEARS_(left_half) NOSE_LANDING_RETRACTED_(half) 594.0 3500.0 0.0 -1298.6 INNER_LANDING_RETRACTED 3415.7 33984.0 -3991.0 -87.1 OUTER_LANDING_RETRACTED 3415.7 33984.0 -7501.0 381.5

integrate them in a wider organised system. A MMG should be able to 'call and communicate' directly with those engineering programs, working as a kind of preprocessor for the analysis applications.

Flutter analysis Structural analysis

ICAD multi-model generator

Structure optimisation Details generation for Multi level optimisation

Figure 2: role of the ICAD Multi-Model Generator inside the MOB Computer Design Engine.

The ICAD Multi-Model Generator and the ‘wingtrunks’ build-up approach In order to be able to comply with the functionalities required by the CDE, starting requirements for the MultiModel Generator are: 1. Definition of a full parametric description of the aircraft. 2. No limits and constraints to the creativity of the designer, which must be able to give a shape to his concept of aircraft.

2

American Institute of Aeronautics and Astronautics

3. Transparency of the aircraft model generation process, which implies avoidance of hard coded constraints or hidden decisions deep nested inside the modeler routines. These would make critical both the output interpretation and the proper generation of the input. 4. Software structure modularity, fundamental in view of upgrading and expanding the system with new functionalities whenever required for more evolved instantiations of the CDE. The ICAD environment has been selected for the development of the MMG to serve the CDE 5. The Multi-model generator basically consists of a set of routines programmed in IDL (ICAD Design Language), which is a super set of LISP. ICAD is an advance tool to support Knowledge Based Engineering (KBE). It is an Object Oriented software that contains both the typical features of expert system languages and the geometry handling possibilities of the most advanced CAD programs. The ICAD product model instantiation has a typical tree structure, the ICAD Product Tree, where the inheritance concept is applied. See figure 4 for the BWB product tree generated by the MMG. A generative model is not simply a CAD model, but represent the engineering intent behind the geometric design. It captures the How and Why, in addition to the What, of the design. It captures the design strategy required to produce a particular product from a specification. It is the set of engineering rules (not

only rules involving geometry) used to design the product. At a high level of abstraction, the whole BWB aircraft can be seen as an organised assembly of parts such as a fuselage (center body), wings, winglets, etc. At one deeper level of detail, all the components of this assembly can be seen as a multiple instantiation of the same socalled wing trunk element (See figure 5). The wing trunk represents the primitive element to build up the complex assembly called Blended Wing Body Aircraft. Intermediate airfoil Root airfoil

Lofted surface Wing trunk

Tip airfoil Connection element

Figure 5: Wing trunk and connection element

Figure 4: The BWB product tree generated by the ICAD MMG.

The definition of a wing trunk requires the assessment of a set of parameters that unambiguously and univocally defines external and internal (structure) shape, orientation and positioning in space 6. One single IDL routine has been developed to generate all the wing trunks. The so called ICAD Contextual Instantiation Principle has been applied: the same wingtrunk routine is invoked multiple times by the MMG, which supplies every time a different set of parameters as input. According to the specific input file, the same generic wing trunk becomes a fuselage, a winglet, a wing etc. The user can select via the input file the reference axis to be used for the wing trunk construction, such as the leading edge line, or the quarter chord line etc. Once this reference axis has been selected all the geometrical parameters (span, sweep, twist and dihedral angle) are consistently and univocally defined. More degrees of freedom have been used for the BWB center-body wing trunk definition, so that a rather free forming is possible respect to the planform layout. A set of airfoils must be selected to allow the MMG to actually generate the lifting surfaces. The user must only specify the location of each airfoil as a percentage of the wing trunk span, while the MMG will automatically

3


generate the double curvature surface, which interpolates the specified wing sections. The airfoils selected to build up the wing trunk are predefined in a library, which the user can always update or integrate with new entries. The airfoils are stored as a set of co-ordinates. A stretching factor, specified for each airfoil, allows the generation of different depth wing sections, eve ntually using the same airfoil definition from the library. If there are changes in dihedral angle between contiguous wing trunks, the model generator will automatically generate proper connection elements in order to always get a continuous and closed surface. The extended parametrical definition of the wing trunk element enables unlimited generation of aircraft variants, just modifying the values of the basic parameters such as span, chord length, angles etc. Furthermore, the use of the wing-trunk (and similar fuselage-trunk like elements) build-up approach reduces even the effort in generating aircraft with very different configuration 1, 2. The wing-trunk structure generator The ICAD contextual instantiation principle has been applied again to make the MMG able to automatically generate the complete structure into any given wing trunk surface 7. In this way there is an efficient (re-)use of the code modules and, at the same time the user can specify a different structure configuration at different aircraft locations. This option makes the tool powerful and very flexible at the same time. The structure is always tailored to the outer aerodynamic surface. A different number of spars and a different set of ribs can be placed in each trunk of the aircraft, while the MMG is automatically taking care of generating the proper connection elements, needed to guaranty integrity and continuity in the structure. Number and position of spars can be assessed just specifying their start/end points as percentage of the root/tip chord of each wing trunk (see figure 6). The

spar cap lines

spar lines

spar webs

specification of one start (or end) point plus an angle respect to the flight direction is also allowed to generate a spar. Each single rib can be positioned and oriented both in flight direction and/or with an angle respect to any spar. As usual the positioning values must be specified as percentage of the trunk length. Some user-friendly features have been incorporated to automatically position ribs at the root and tip of the trunk without caring about the actual span size or eventual dihedral and sweep angles used (see figure. 7). All the rib-lets in the leading and

upper rib cap line rib cut plane n 0

rib- reference- point lower rib cap line rib- orienting- angle Figure 7: ribs generation in the wing trunk elements. trailing edge parts of each wing trunk are also automatically generated by the MMG. The user has the possibility to assign, via the MMG input file, a label to each placed rib, indicating the actual structural functionality (hinge rib, bulkhead rib, etc). Once the MMG detects a hinge rib, a routine automatically generates all the hinge brackets to attach the respective moveable surface. The trailing edge moveables (TED’s and rudders) generation is also incorporated in the MMG. The user can specify number and size of moveable parts at the trailing edge of each wing trunk. The position of the hinge lines of each TED can be assigned. The moveable closure spar is automatically generated, so that a realistic representation of the load introduction path from the moveables on to the torsion box is defined for the wing structural analysis. The moveables defined in the MMG are not really deflectable inside the ICAD model, but they are actually deflected in the aerodynamic model incorporated in the CDE 8. For the structural analysis, the actual aerodynamic load, evaluated for deflected condition, is applied on the un-deflected structural model generated by the ICAD MMG.

Figure 6: Spars generation in the wing trunk elements. 4


The non-structural mass items generation tool Together with the structural items such spars, ribs and skins, also the so-called non-structural mass (NSM) items need to be incorporated in the aircraft model, to be used both for the structural analysis and for the weight and balance discipline. Engines, landing gears, systems, cargo are some of the typical NSM items on board, whose inertia gives a significant contribution to the aircraft loading. In order to include their contribution in the structural analysis, a complete NSM model has been incorporated into the ICAD multi-model generator. The model is rather simplified since the actual items are not modeled in detail, but they are represented as lumped masses, properly located respect to the aircraft structure. The MMG does not actually evaluate the weight of each item, but simply processes some reference weight values fed into the MMG as input. Other tools in the CDE are in charge of supplying the MMG with a good estimation of the weight of each item, even if automatic weight scaling of some items like the trailing edge device (TED) controls or the deicing systems is performed by the MMG whenever the size of the aircraft is changed by the CDE during an optimisation loop or a parametric study. The MMG generates, for each aircraft configuration, a formatted table, where, the weight (eventually scaled) and the 3D-location of the center of gravity, is indicated for each NSM item respect to a datum (see figure 8). As the geometry of the aircraft changes, the positioning of the non–structural items follows automatically and in a consistent way. The NSM ITEM

Mass_(kg)

GROUP_FUSELAGE_(left_half) TED_1_(half) …….. ANTI-ICING-SYSTEM OPERATIONAL_ITEMS_(half) CABIN_ARRANGEMENTS_(half) FLUIDS_(half) GROUP_WING_(left_half) TED_4_(iw_ins) TED_5_(iw_out) ……. ANTI-ICING-SYSTEM_(ow) GROUP_WINGLET_(left_half) RUDDER ANTI-ICING-SYSTEM_(wl)

X_cg

Y_cg

Z_cg

44973.1

-1250.3

1250.1

12755.8 -6368.1 3000.0 0.0 3000.0 0.0 3000.0 0.0

490.0 0.0 0.0 0.0

309.1 43256.9 -15067.4 292.0 41879.3 -20362.6

2543.5 2256.1

402.0

40304.5 -31062.3

1671.6

174.5 80.0

49785.9 -39394.7 47962.0 -39490.0

4990.1 5531.1

43758.0 43758.0 39750.0 39750.0

4142.9 2185.7 5410.5 3453.2

107.4 240.0 157.5 40.0 3.0

GROUP_PROPULSION_(left_half) CENTER_ENGINE_(half) 3751.2 CENT_ENG_.. 980.7 LEFT_ENGINE 7502.3 LEFT_ENG_STRUC….. 1961.3

0.0 0.0 -7501.0 -7501.0

GROUP_LANDING_GEARS_(left_half) NOSE_LANDING_RETRACTED_(half) 594.0 3500.0 0.0 -1298.6 INNER_LANDING_RETRACTED 3415.7 33984.0 -3991.0 -87.1 OUTER_LANDING_RETRACTED 3415.7 33984.0 -7501.0 381.5

Figure 8: Non Structural Mass items table

nacelle/aircraft-body clearance values are also parameters to be set in the input file. The modular nature of this ICAD application enables a possible incorporation of a dedicated module for the NSM estimation, in order to generate the reference weights file directly within the MMG. Since the masses are represented by the MMG as a grid of lumped masses, an application has been developed to automatically assess, for each structural element (or part of it), the proper connectivity with the NSM items. The NSM items included in the model can actually be connected to a single or eventually more elements of the BWB structure (e.g. a de-icing system is attached to all the leading edge rib-lets of the given wing trunk; landing gears are attached both to a number of spars and ribs elements). An example of this structure/non-structural mass item connectivity by means of the RBE is sketched in figure 9 9. The MMG is able to interpret the current TED’s control system CG’s

TED 5 TED 6 TED 7 TED 8

RBE

Rear spar segments Ribs

Closure spar segments

Figure 9: Example of NSN items/structure connectivity. structure configuration and NSM items distribution, and based on their relative position in the space, is able to recognize which mass is attached on which structure element. This information is then formalized assigning to each single structural element an attribute (or label, see The KBE approach: the Fem tables, later in this document), which reports the name/s of the connected NSM item/s. Two trim-tanks (parametrically described) can be located in the aircraft center-body. The user can modify capacity, shape and positioning of both the tanks, just setting in the input file for the MMG the amount of fuel required and the position of the tanks relatively to the aircraft datum. Weight and positioning of the tanks center of gravity are also supplied by MMG as output. The MMG Input/Output data flow system All the parameters used to define the BWB aircraft product are combined in engineering rules inside the various modules programmed in IDL, while the numerical values that actually give the flavour to the parametric model are univocally assigned in what we called so far the MMG INPUT file. This is nothing else than a ASCII file, editable as such, where all the parameters are userfriendly grouped in different sections, according to their logical function. There is an input section for the aircraft 5


surface parameters (span, chord length, angle etc.), one for the airfoils assignments, one for the structural items definition, one for the eventual trim tanks, engines and landing gears positioning/sizing etc. The INPUT file univocally determines the definition of the final output. From each INPUT file, one specific aircraft configuration is derived. This input data system allows for both interactive and batch job use. The user himself can edit directly the INPUT file, but the parameters change can also be done by the CDE integrated optimiser, if the MMG is operative in a closed loop process. The initiator module, previously shown in figure 3, can also initiate the design process. Figure 10 shows the actual I/O data flow for the MMG in the CDE. USER or DATABASE

INPUT DATA SET

Airfoil library

Batch.lisp

Positioning rules for NSM items

Non-structuralmass.lisp

INPUT DATA SET EDITING

ICAD local database

Input.lisp

Stretch function for CFD-points

Input.lisp

ICAD MULTI-MODEL GENERATOR

Batch.lisp Airfoil.lisp

IGES FILES

ICAD output reports

FEM-TABLES TANK VOLUMES

…...

NONSTRUCTURAL MASSES

CFD-POINTS HF/LF

PLANFORM POINT

STRUCTURE AEROELASTICITY AERODYNAMIC

Figure 10: the ICAD MMG input/output data flow. The MMG is built up to output a huge amount of information (not only geometrical information) to feed the discipline analysis tools. Aerodynamic models at high and low fidelity level, structural models, planform models for initial flutter analysis, mass distribution models etc., can be explicitly requested by the user (also for batch run), based on the specific CDE architecture in which the MMG is integrated. The ICAD environment supports the socalled Demand Driven Instantiation: ICAD evaluates only what is needed to the user immediate requirements and does not generate the entire model (i.e. if the user demands the generation of the outer surface geometry, no calculation resources are spent by the MMG to evaluate the internal structure). In this way the amount of computation is kept to the strictly required. Different standard output formats have been used to code the MMG; these include IGES files, VRLM file and a variety of table and formatted text (ASCII files). The input/output data format is a main issue when working within a strongly integrated analysis environment such as a CDE, since each analysis tool

(COTS or in-house developed) often requires a specific (not always standard) input data format. Link with aerodynamic analysis tools The MMG is able to export the complete aircraft surface, eventually split in upper and lower panels, directly embedded in IGES format. A special application has been developed to transfer the ICAD generated surfaces also through a ‘cloud of points’ like format (see figure 11). This represents a valuable alternative to link the MMG with those codes (often inhouse developed), which are not able to handle the IGES format 8. All the BWB surfaces generated by the MMG are automatically processed by a specific module (still inside the MMG) and translated into a set of surface points (the cloud of points). The MMG then reports in a formatted text file the Cartesian co-ordinates of each single point, respect to a reference datum. This specific ICAD interface/translator is suitable both for high and low fidelity aerodynamic models, since the MMG user can customize (via the INPUT file) the amount and distribution of grid-points (the density of the cloud) 8, 10. The user can specify, independently for each single wing trunk and connection element, the amount of wing sections at which the MMG must extract the points. I.e. the BWB center section (fuselage) will require a higher number of sections/profiles due to its large span, while for the small connection elements a couple of sections will be enough to provide a good surface discretization. The number of points to extract from the upper and lower part of each section/profile is again a user parameter to set in the MMG input file (i.e. a profile discretization suitable for HF aerodynamic analysis will require hundreds of points, while few points will be sufficient for LF applications). The stretching function (a sinusoidal function) applied to extract the points from each section is also parametric, so

Figure 11: generation of the ‘cloud of points’ for the CFD analysis tool.

6


that the user can eventually vary the points density (distribution along the various sections/profiles) near the leading edge and/or trailing edge. It should be noted that for certain kind of analysis, the cloud of points might directly be interpreted as mesh seeds for a grid. Link with structural analysis tools A main issue here is the definition of a proper strategy to build up a direct and smooth link between the MMG and the FE analysis environment. On the other hand, no attempt has been made trying to code any FEM package directly inside the MMG. The basic idea is to give the discipline expert the opportunity to play with his ‘favourite toys’ (PATRAN/NASTRAN, ABAQUS and/or other in– house available codes), but taking advantage of the power of the flexible geometry generator developed in ICAD.

ICAD environment

SURFACES FRAGMENTATION

PATRAN environment

structural items configuration of the aircraft model and automatically fragments skins and web in readily meshable surfaces patches. Every time the MMG is invoked by the user (or the CDE optimiser) for a new analysis loop, and the airfoils shape and/or the number of structural elements in each wing trunks or even the number of wing-trunks is changed, the splitting-routine still is able to deliver a consistent set of cut surfaces. When a surface patch is generated in ICAD, a special identification label is applied. A special MMG module performs the scanning of the product tree, collects all the surface patches and distributes them in a predefined set of IGES files, ready to be imported in the FE environment. This ‘labeling/scanning’ strategy (see also The KBE approach: the FEM tables, later in this document) has been developed as main procedure to extract in a systematic way knowledge/information from the product tree and make it available, customized for the CDE downstream processes. Time efficiency and size of the output data flow are major issues, which have been considered, as well as topology consistency and geometry accuracy 2, 7. The design variable definition In order to optimise the thickness distribution of all the structural elements, a special ICAD application has been developed to assign each surface element a proper design variable. When the structural optimisation is performed, all the surface patches with the same design variable will be sized with the same thickness. A major user requirement is the possibility to tailor both the amount and the local distribution of the design variables. A higher number of design variables will provide a better structure optimisation, but will be more expensive in terms of time and needed calculation power. Based on the specified input data, a proper code number is automatically generated by the ICAD MMG, which identifies the membership of each surface element to a

Figure 12: The ICAD model surfaces are split in small patches before being imported in PATRAN. Efforts have been made to completely automate the repetitive pre-processing job, in order to directly pass from the geometry definition in the MMG to the analysis and post-processing phase. A special procedure has been programmed in the MMG to automatically split all the structural surfaces (spars, ribs, skins) along their intersections. Spars are split along the intersection with all ribs; skin panels are split in patches along the intersections both with the ribs and the spars (See figure 12). This is a typical, time consuming, operation that the FE package operator must always perform prior to meshing. In this case, the MMG interprets the

BWB aircraft zones

Span-wise areas

Chord-wise areas

Structural elements

Figure 13: grouping principle for the design variable area assignment.

7


given design variable area. The generated design variable code is then added to the list of attributes/labels of the given element (see The KBE approach: the FEM tables later in this document). The user has the possibility, via the input file, to split each BWB trunk in a free number of span wise extending zones. He can also choose to set all the surface segment of one rib into one design area. It is possible also to specify one design variable area for a complete upper or lower skin panel. On the other hand, in case of significantly wide skin panels, it is possible to generate an extra surface segmentation, by means of so-called “virtual spars” (see figure 13). The design variable code is finally a number derived from the combination of the 4 different sub-code numbers, which respectively identify the specific BWB aircraft zone (i.e. the center body section, the outer wing, winglet etc.), the specific structural element (i.e. a spar, rib, upper etc.), the chord-wise and the span wise extending areas that have been defined on each wing trunk. The KBE approach: the FEM-tables The IGES standard, widely employed for CAD/CAE activities, is suitable to transfer only geometries, whilst all the non-geometrical information (the actual ICAD model added value) required for the MMG/FE integration, needs a different output strategy. A specific ICAD application has been incorporated inside the MMG (the FEM-tables system) in order to transfer all the product knowledge hold by the BWB product tree, which is lost during the IGES translation for the FE environment. A set of look–up tables is automatically output by the MMG in parallel with the IGES files. For each surface patch that is exported into an IGES files, all the related (useful) information generated by the MMG is transcribed in a properly formatted text file (the FEM-table). This information includes (see figure 14) a first thickness estimation, material and Surface_ID_number 2000023 Isoparametric? T Membership INNER-WING-INSIDE Type QUAD-SEGMENT Design_variable_group 2010203 Material AL_ZI_PLATE Thickness_(mm) 6.0 Attach_non_struc_mass DE-ICE_SYSTEM Disturbed_by_door_cutout NIL Number_of_nodes 4 Node_ID X Y Z 92 49542.0 -39936.5 8381.3 93 49454.4 -39895.0 8173.1 94 49871.0 -39859.8 8061.9 95 49962.0 -39926.1 8383.2

Figure 14: example of one FEM-Table.

design variable area specification, non-structural masses connectivity, plus other generic element features such as ‘mesh-ability’, number of edges, coordinates of the corner nodes, etc. All this data have been made available by the MMG via the attributes/labels that we mentioned before in this document. One FEM-table is generated for each IGES file. A proper routine (MAPSURF) has been developed to map the contents of the FEM-tables, directly into the PATRAN database that is created to import the IGES files. The co-ordinates of the nodes of each surface patch represent the bi-univocal link between each surfaces IGES coded and its representation in the PATRAN environment (see figure 15) 8, 9. Geometry link (IGES FILES) Geometry info in the IGES file y

Geometry info in PATRAN s5

s k

s4

b

a w

s3

s2

s1 x

a1

z2

z3

z4

z1

Surface Mapping link (NODES COORDINATES) ICAD element ID : x node: m (co-ordinates x, y, z)

PATRAN element ID : z2 node: w (co-ordinates x, y, z)

node: m + 1 ( “ node: m + 2 ( “ node: m + 3 ( “ element knowledge

node: w + 1 ( “ node: w + 2 ( “ node: w + 3 ( “ element knowledge

x, y, z) x, y, z) x, y, z)

x, y, z) x, y, z) x, y, z)

Knowledge link (FEM-TABLES)

Figure 15: the ICAD/PATRAN connection via the geometry and knowledge links. A flexible PATRAN session file has been programmed in PCL (PATRAN Command Language), to read automatically the set of IGES files, locate and position the non structural mass items, generate mesh, apply constraints and properties and run the FEM analysis and/or optimisation 9. The special use of PATRAN/NASTRAN via the programmed session file allows the complete integration of the MMG with the FE environment, so that even models with different topology can smoothly flow from one environment to the next. Input model for the low fidelity flutter analyses A special application has been developed to extract the simplified planform model from the complete BWB model generated by ICAD. The BWB planform is automatically split in a set of fixed and moveable (flapped) panels, based on the actual position of the moveable surfaces defined on each aircraft section. This application is fully parametrical so that amount and positioning of each panel is always consistent with the actual BWB configuration specified by the ICAD input file. The coordinates of the corner points of each panel are 8


Fuselage trapezoids Un-flapped panel n.1 of 3 X Y Z 0.000 0.000 0.000 44085 0.000 1231.4 44080 –3011.8 1567.9 6032.9 –3011.8 231.75

flapped panel

flapped panel n.1 X Y 44085 0.000 47981 0.000 47833 –3011.8 44080 –3011.8

of 3 Z 1231.4 1340.2 1668.2 1567.9

Un-flapped panel

Figure 15: 2D-model for aeroelastic computation based on the ICAD planform point model. evaluated relatively to a datum and output in a properly structured ASCII file (see figure 15). The planform panel model is used as typical input for aeroelastic analysis tools 11. Model generation for the multi level optimisation approach The optimisation of a complete and detailed aircraft structure easily becomes an un-manageable problem, due to the large number of possible design variables and related constraints. The same difficulties are encountered in the actual modeling phase of the aircraft (keep in mind the routine developed to split the surfaces in patches along the intersections of the structural items). In order to make details modeling and optimisation a feasible task, a multi-level methodology is applied. The model (and the optimisation problem as well) is decomposed into a set of sub models (problems), still taking into account the essential structural coupling effects. These optimisation processes normally regard the analysis of some typical items as doors or windows cutout. A door cutout is parametrically defined in the

ICAD MMG. The presence of the cutout is not taken into account in a first instance, during the application of the surfaces segmentation procedure (see Link with structural analysis tools above in this document), in order not to increase the level of geometrical complexity in the model. A routine automatically scans all the surface segments in the BWB and isolate those influenced by the presence of the door-cut. These surfaces are finally extracted from the global model, processed and sent as input for the sublevel optimisation loop [ref. The cutting and trimming process on these surface elements will not affect the global model; on the other hand the ‘disturbed’ elements will always reflect all the geometrical changes occurring in the global model (see figure 16). This approach can also be suitable to generate windows cut out, inspection panels or enlighten holes in the ribs web. Functional analysis and tool exploitation The development of a multi model generator, able to incorporate the parametrical definition of parts and assemblies, appears to be a major goal for aircraft industry and, more in general, for companies that develop complex systems. The modular structure of the ICAD code allows, in principle, unlimited possibility of growth for the generative model. Many other applications can be developed and bolt-in the MMG to incorporate other disciplines (production and manufacturing, acoustic and costs) or to model and analyse phenomena at different scale level, e.g. bonded or mechanical fastened joints, piezo-elements for active noise reduction 1, 12, 13, 14. While MDO activities are nowadays generally expensive in terms of run time and labour due to the pre/post processing activities, the integration of a powerful MMG in CDE systems can reduce not only the running time, but also limit significantly the need for user interaction. Acknowledgements The study reported in this paper is carried out in EC research project MOB - A Computational Design Engine Incorporating Multi-Disciplinary Design and Optimisation for Blended Wing Body Configuration (Contract Number G4RD-CT1999-0172). The authors wish to thank the EU for partly funding the project in the GROWTH program and all partners who contributed to the functional modules which together for the CDE.

1.

Figure 16: detail of a FE-model for the door cutout optimisation.

2.

9

References M.J.L. van Tooren, L. Krakers, G. La Rocca, A. Beukers; Design and Technology in Aerospace, Flying High; Onderzoek Integraal Ontwerpen Architectuur & Techniek, Verliefdheid of Hartstocht; April 2002. G. La Rocca; L. Krakers; M.J.L. van Tooren; Development of an ICAD generative model for


3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

aircraft design, analysis and optimisation; International ICAD User Group conference, Boston; June 2002. A. Morris; MOB A European Distributed MultiDisciplinary Design and Optimisation Project; AIAA 2002 conference, Atlanta, GA, USA, September 2002, AIAA-2002-5444. W.J. Vankan; A SPINEware Based Computational Design Engine for Integrated Multi- Disciplinary Aircraft Design; AIAA 2002 conference, Atlanta, GA, USA, September 2002, AIAA-2002-5445. Knowledge Technologies International; The KBO environment documentation; Release 2.0; 2001. G. La Rocca, L. Krakers, M.J.L. van Tooren; Description of the ICAD BWB-surface generator code; report MOB/6/TUD/D6iiia; Augustus 2001. G. La Rocca, L. Krakers, M.J.L. van Tooren; Description of the ICAD BWB-structure generator code; report MOB/6/TUD/D6iiib; Augustus 2001. M. Laban; P. Arendsen et al; A computational design engine for multi-disciplinary optimisation with application to a blended wing body configuration; AIAA 2002 conference, Atlanta, GA, USA, September 2002, AIAA-2002-5446. D. Pearson; MOB structural model generation by PATRAN session file; report MOB/6/BAE/Report/3; September 2001. N. Qin; Aerodynamic Studies of Blended Wing Body Aircraft; AIAA 2002 conference, Atlanta, GA, USA, September 2002, AIAA-2002-5448. M. Stettner; R. Voss; Aeroelastic, Flight Mechanic, and Handling Qualities of the MOB BWB Configuration; AIAA 2002 conference, Atlanta, GA, USA, September 2002, AIAA2002-5449. H.E.N. Bersee, M.J.L. van Tooren, A. Beukers; Manufacturing of a thermoplastic composite structural aircraft component; The 5th International ESAFORM Conference on Material Forming, proceedings; Poland; April 2002. Gleich, D.M., van Tooren, M.J.L., Krakers, L.A., Beukers A.; A numerical method to analyse and design adhesively bonded joints in structures; International IASS symposium on lightweight structures in civil engineering, Warsaw; June 2002. Krakers, L.A., van Tooren, M.J.L., Beukers A.; A design engine to evaluate sound damping of flat panels in the low frequency range; ECCM10 proceedings, Brugge (Belgium); June 2002.

10
