Multi-Fidelity Whole Conceptual Aircraft Analysis
This following is based largely on the abstract from my DIPART 2018 Presentation delivered a the CFMS, Bristol 27th Nov 2018
The availability of ever-increasing computing power at reasonable cost has resulted in many industrial, research and academic organisations bolting together ever more complex design methodologies to model complete air vehicles. Is this the most effective approach? Maybe, maybe not.
The extensive capabilities and benefits of high-fidelity methods are in many cases are, without question, extremely impressive in the right hands and used in the right way, particularly for defining the physical aircraft components. However, any perceived connection between increasing accuracy and increasing fidelity must be avoided. It can even be argued that recent airframe and engine projects do not always hit their targets despite using the highest level of analysis with highest level knowledge on component design and modelling definition.
My Opinion
My opinion is that the most appropriate approach for modelling future air vehicles in the conceptual and preliminary design phase is a combination of higher- and lower-fidelity analysis methods with appropriate movement between, but with a bias towards the lower fidelity methods for whole aircraft analysis.
At this stage, any novel technologies may be initially modelled in low fidelity ('level 1 & 2') methods based on engineering judgement from the most qualified expert or research available with some uncertainty analysis used to gauge design robustness.
If required and enough time/resource is available, higher-fidelity methods ('level 3') can be employed in novel areas in a ‘Design of Experiment’ (DoE) mode to provide the whole aircraft impact of any novel technologies, possibly using the same expert who provided the initially engineering judgement. The DoE results should be in the form of Pareto fronts/response surfaces representing optimal design solutions as a function of top-level aircraft attributes to form new surrogate models that replace the 'judgement' in the lower-fidelity design tool with adjustment factors to make further technology adjustments.
The lower fidelity multi-disciplinary design tool is used to rapidly explore the design space to define potential market ‘sweet-spots’, design constraints as well as the sub-system targets (and their boundary conditions) to achieve the whole aircraft level requirements, i.e. they provide a clear line of sight to the customer requirements.
At this stage, if enough benefit has been forecast at the lower fidelity level, a preliminary design phase may be launched in which the higher fidelity methods are critical to determine exactly how these each sub-system (i.e. aerodynamics, structure, propulsion, various systems, etc.) targets are to be physically achieved. Individual or multiple high-fidelity methods in combination are used as required - always use the minimum.
The knowledge gained for this point design is used to further calibrate the lower fidelity methods and to confirm or modify the design sweet spot.
GIGO
Higher-fidelity analysis methods, in isolation or in combination, must be used carefully. The results typically rely on many inputs such as detail geometry, material properties, mechanical design, detail architectural design choices, technology availability, boundary conditions, surface finishes, the meshing strategy, component properties, the underlying analytical code. As such, they provide highly detailed results for the input sets provided, but the relevance of the results is a function of the quality of ALL the inputs (GIGO – ‘Garbage In, Garbage Out’ or ‘Great Inputs, Great Outputs’).
GIGO equally applies to lower-fidelity methods, but the inputs are much fewer and it is generally substantially easier to define or estimate a reasonable (i.e. ‘Great’) value for all of them. This can achieve a ‘Great’ output in much less time (both physical and CPU). These methods do not model stress or aerodynamic flows, but they forecast the outcomes (weight, drag, SFC & performance, etc) of what a high-skilled multi-disciplinary design team would be expected to achieve when they use high fidelity tools over an extended period to realise a successful certificated aircraft design.
The lower-fidelity methods also allow the impact of market-based (non-technical) requirements as well as their sensitivity to design changes and uncertainty to be rapidly evaluated.
