Dynamically controlled variable-fidelity modelling for aircraft structural design optimisation. (June 2014)

• Record Type: Journal Article
• Authors: Allen, Jonathan G; Coates, Graham; Trevelyan, Jon
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 228:Number 8(2014:Aug.)
• Page Start: 1434
• Page End: 1449
• Publication Date: 2014-06
• Subjects: Aircraft conceptual design -- structural optimisation -- variable-fidelity modelling -- parameter control; Aeronautics -- Periodicals; Astronautics -- Periodicals; Airplanes -- Design and construction -- Periodicals; Aerospace industries -- Periodicals; 629.1
• DOI: 10.1177/0954410013493074
• Languages: English
• ISSNs: 0954-4100

Abstract

Structural mass optimisation of an aircraft design is important in increasing the likelihood that a high quality airframe is designed of minimal weight whilst providing necessary resistance to load. Analysis of such structures is often performed at a single level of model fidelity, the selection of which can lead to either excessive computational time or reduced accuracy of results. Alternatively, variable-fidelity modelling may be employed to reduce such computational expense whilst maintaining accuracy, traditionally performed using predetermined levels of fidelity for specific periods of the optimisation process. This paper investigates dynamically controlled variable-fidelity modelling during aircraft conceptual design optimisation wherein fidelity is controlled as a dynamic parameter of the optimisation process. Consequently, model fidelity is adapted during optimisation to promote early discovery of promising design characteristics prior to detailed analysis of the best designs available. Models are constructed through the grouping of similar structural members within elements, thus reducing the number of degrees of freedom and subsequent computational effort required for analysis of each design. A case study is performed to verify the results of analysis and obtain benchmark results for optimisation with static model fidelity prior to the investigation of various set-ups of dynamically controlled variable-fidelity modelling. The results of this study indicate improved design quality using dynamically controlled variable-fidelity modelling compared to using static model fidelity whilst reducing the necessary computation time.