Search

During early design stages, limited information is available to conduct decisions on the goodness of aircraft configurations. In this work, the physical information is supplemented by the uncertainty underlying the applied analysis modules, propagated through the complete design workflow. Thereby, it is possible to not only state that one concept is preferred above another, but furthermore the level of certainty of such a statement can be quantified.

The present study developed a multi-fidelity workflow to derive physics-based conceptual design methods from models of higher-fidelity usually employed during preliminary aircraft design. The multi-fidelity workflow consists of a design of experiments, a multi-fidelity loop, and symbolic regression as surrogate modeling technique. Results are presented for conventional and unconventional aircraft configurations.

Aircraft conceptual design often focuses on unconventional configurations like for example forward swept wings. Assessing the characteristics of these configurations usually requires the use of physic based analysis modules. This is due to the fact that for unconventional configurations no sucient database for historic based analysis modules is available. Nevertheless, physic based models require a lot of input data and their computational cost can be high. Generating input values in a trade study manually is work-intensive and error-prone. Conceptual design modules can be used to generate sucient input data for physic based models and their results can be re-integrated into the conceptual design phase. In this study a direct link between a conceptual design module and an aerodynamic design module is presented. Geometric information is generated by the conceptual design module and the physic based results, in form of the Oswald factor, are then fed back. Apart from the direct link, an equation for determination of the Oswald factor is derived via a Symbolic Regression Approach.

The coupling of physical effects and the rising complexity of modern aircraft necessitate an intense collaboration of disciplinary specialists in preliminary airplane design. Additionally, a growing number of suppliers and outsourced design activities aggravate the design processes even further. Novel multidisciplinary design environments intend to enable specialists to better integrate analysis codes and constitute the consistent technical basis for their cooperation. In this manner different technical aspects can be evaluated quickly, since the data update and analysis capabilities are available throughout the design team. The exchange of information is therefore of crucial importance in multidisciplinary design. Due to the fact that in data exchange the number of interfaces is the critical factor for the flexibility of a design environment, a central information model is a key feature. The central information model reflects among other things the common namespace of the design team and can be seen as the meta-model for all of the deduced analysis models. The architecture of such a novel design environment is strongly linked to principles of model-based architectures and the chosen underlying software engineering techniques influence strongly the efficiency of the resulting design processes. An information model for a design environment as described above consists of two aspects. On the one hand the elements, attributes and their structure need to be defined in a schema definition. On the other hand the explicit content has to be stored in a data set conform to the schema definition. Whereas the data set is mainly used for the exchange of information, the schema definition is utilized for documentation, model validation and model generation. Several possibilities of attributes for a quality information model are named in the literature and are outlined in this research. These attributes or requirements include important aspects like holism, accessibility, transparency and ordering mechanisms. Additionally, several abstraction methods need to be taken into account that are mostly based on principles of object oriented modeling. Information models commonly used in preliminary airplane design include STEP, XML and most recently the Unified Modeling Language (UML). The Standard for the Exchange of Product Data (STEP) is described in ISO 10303. STEP is widely spread in the industry. The meta-models are built up using the object flavored modeling language EXPRESS. Content models are stored either via plain text or XML formats. The Common Parametric Aircraft Configuration Schema (CPACS) is a DLR intern standard. It is used in several multidisciplinary projects that handle the scope of preliminary airplane design and further aspects like climate impact and transport systems. The standard is based on XML and benefits from a growing number of tools that are adapted to it. As UML is concerned, there is not yet a standard for the modeling of preliminary aircraft data, this work provides therefore an outlook on some of the research in this area. Traditionally, the UML is a powerful software modeling language and the only of the introduced language concepts that features a truly integrated model. The different modeling languages and approaches are reviewed in respect to the quality characteristics already mentioned. Their development history is displayed and existing approaches for the application of these techniques to preliminary airplane design in the academic as well as industrial field are shown. An evaluation of the named modeling languages is made with respect to the existing multidisciplinary design environment at the DLR. Future developments for the different modeling languages are outlined and potential crossovers highlighted. It is shown that CPACS can be translated to an UML model. Additionally, a CPACS to STEP translator is introduced outlining the software mechanisms applied and showing a validation example using the geometry data of the VFW-614 ATTAS. This research concludes in an evaluation of several techniques for information modeling taken from computer science to a multidisciplinary engineering design environment for preliminary airplane design. The central information model contained therein forms the basis for distributed modeling and simulation. Its proper parameterization allows therefore a more firm and consolidated assessment of known aircraft designs as well as of some of the potential aircraft technologies of tomorrow.

A multidisciplinary approach to preliminary airplane design is seen as one of the major improvements for future design tasks. Several approaches can be found in the literature. The key to a multidisciplinary approach is a central model that contains all data and is accessible for all domains. The goal of this work is to analyze three different models in respect to their benefits for data integration in preliminary airplane design. The models are Cpacs, developed at the Dlr, Step, published by the Iso, and the Uml, released by the Omg. For this purpose a lexical overview of the important terms is given. Additionally several methods for the classification of information are introduced and requirements for information models are set up. The main part of the work is contributed to the analysis of the different models. A conclusion is drawn that suggests a solution for a future information model using a combination of the Uml / SysML and Xml. Subsequently a prototype for a converter tool, that processes Cpacs data to Step is developed. The converted geometry data of the Attas Vfw 614 is used for validation and shows the quality of the tool.