Knowledge based engineering techniques to support aircraft design and optimization

Abstract

Since the 1960s, the demand for air transportation has doubled every 15 years, resilient to every oil crises and international events. However, the current capability of the air transport management system, the demand of increasingly growing levels of quality, comfort, safety and security, and, above all, an environmental sensitivity as high as never before, seem to constrain any further growth. The Advisory Council for Aeronautical Research in Europe (ACARE), similarly to NASA in the United States, has indicated a set of challenging objectives and devised a roadmap to help the aerospace industry stepping into a new age of sustainable growth. However, major technological advances will not be possible without significant improvements to the current design methodology. In this regard, the present research work aims at the development of new design methods and tools that are able to sustain the evolutionary improvement of current aircraft designs, as well as to support the investigation of novel aircraft configurations. To be successful, these design methods and tools must be able to facilitate the aircraft development process as it is currently carried across large and distributed supply chains. Besides, they must account for the increasing scarcity of intellectual resources and the consequent need to increase engineers’ productivity and freeing time for innovation. The Multidisciplinary Design Optimization (MDO) approach appears to be the most promising design methodology in the field of aircraft design, both to improve the performance of traditional aircraft configurations and to support the development of novel concepts. However, a number of technical and non-technical barriers have prevented full exploitation of the MDO approach and, so far, limited its industrial application to detail design cases. To this purpose, the concept of Design and Engineering Engine (DEE) has been developed at the Faculty of Aerospace Engineering in Delft, which is a modular, loosely integrated design system able to support distributed multidisciplinary analysis optimization by automating as far as possible the repetitive and non creative activities that hamper the design and analysis process. One of the DEE technology enabler is the Multi Model Generator (MMG), which actually represents the main outcome of this research work. The Multi-Model Generator (MMG) is a Knowledge Based Engineering (KBE) application developed with the twofold intent of 1) providing designers with a parametric modeling environment to define generative models of conventional and novel aircraft configurations and 2) feeding various analysis tools with dedicated aircraft model abstractions, as required for the verification of the generated design. To meet these objectives, two types of functional blocks have been developed, which constitute the main ingredients of the MMG: the High Level Primitives (HLPs) and the Capability Modules (CMs). Four High Level Primitives have been defined, namely Wing-part, Fuselage-part, Engine and Connection-element. These can be figured out as a suite of advanced LEGO blocks that designers can manipulate to assemble the geometry (external surfaces and structural layout) of the aircraft concept they have in mind. Each HLP has been programmed as a class using the object-oriented programming language of the employed KBE system. This has allowed capturing the design rules that give the HLPs the capability to automatically adapt their own shape and topology, or to trigger events as a reaction of input changes. By means of the editable MMG input file, designers can assign different values to the attributes of each HLP class and call for multiple HLPs instantiations. In this way, both conventional and novel aircraft configurations can be automatically generated and then stretched/morphed into an infinite amount of variants. During the conceptual design phase, designers “see” the aircraft as an assembly of basic solutions to fulfill functionalities, such as generate lift and accommodate payload, rather than an assembly of points, curves, surfaces and solid features. The capabilities to support the designer’ functional thinking and capture knowledge in terms of design rules, have yielded the MMG primitives the “high level” connotation, in contrast with the “low level” primitives of conventional CAD. Once the model of the given aircraft is available, the preparation for the verification phase starts, which requires the set up of the various discipline abstractions (or views) that must be fed to the analysis tools. In the traditional design process, the preparation of these disciplinary models is acknowledged to be lengthy and repetitive, particularly when high fidelity analysis tools are involved. Up to 80% of the overall design process can be wasted just for these preprocessing activities. However, it has been observed that 1) independently from the aircraft configuration at hand, the same analysis tools and preprocessing methods are generally used by specialists; 2) large part of the preprocessing activities is rule-based and require a large deal of geometry manipulation, which actually represent the strengths of KBE technology. To support this phase of the design process, a set of Capability Modules (CM) has been developed to capture the “model preprocessing knowledge” of discipline experts and reuse it to automate the generation of models for a broad range of low and high fidelity analysis tools, both proprietary and commercial off the shelf. The implemented approach has enabled the use of high fidelity analysis tools, such as FEM and CFD, already in the early stages of the design process, which not only increases the level of confidence in the designed product, but provides essential means for the study of innovative aircraft configurations, where semi-empirical and statistics based methods fail and first principle analysis is the only way to go. Due to its ability to be accessed in remote, via web connections, and operated in batch, the MMG also demonstrated to be a valuable asset to support MDO processes across distributed design frameworks. The capability of the MMG has been demonstrated by means of several example applications and two relevant study cases addressed in this work. The first case concerns with the European project MOB, on distributed multidisciplinary design optimization of blended wing body aircraft configurations. The second deals with a MDO system developed in collaboration with Airbus to redesign the vertical tail of an existing passenger aircraft. A side objective of this work was to improve the dissemination of KBE technology, which is still a relatively young discipline that has not yet found the deserved level of attention and understanding, both in the world of industry and academia. To this scope, an extensive and original investigation on the Artificial Intelligence roots of KBE is provided and its object oriented paradigm thoroughly discussed. A best practice section to the development of KBE applications is included as well.