The aim of axiomatic aerodynamic modelling is introduced, namely to understand the validity of aerodynamic modelling in the context of aircraft dynamics with particular emphasis on understanding the relationship between full-scale flight behaviour as compared with that predicted from data based on wind tunnel tests.

An axiomatic aerodynamic model is formulated for the attached flow past a wing and tailplane, based on a vortex lattice representation, with the wing wake fully relaxed at each instant of time. Conventional derivatives are estimated based on this model.

Predicted responses, using the above derivatives, are compared with ‘exact’ responses, using the axiomatic model for various elevator inputs and for both non-swept and swept aircraft configurations.

Surprisingly, even in these simpler cases, differences appear between prediction and ‘exact’ responses.

The axiomatic aerodynamic model in Part 2 has been extended to an aircraft moving in a general coupled longitudinal-lateral motion but with the flow attached throughout the motion. Results are presented for the responses of a swept wing aircraft due to separate rudder and aileron inputs. For some aileron responses there are inexplicable significant differences between actual response and predicted response, based on derivatives, in roll rate and pitch rate. No such differences appear with rudder inputs.

Measurements made at a Mach number of 0.18 and a chord-based Reynolds number of 4·2 x 106 on a constant-chord model having a NACA 4412 aerofoil section are described and compared with the results of flow field calculations.

Both the experimental arrangement and the difficulties initially experienced in achieving an adequate approximation to two-dimensional flow above the wing are briefly outlined.

The measurements include static pressure distributions on the wing surface and on the wind-tunnel walls above and below the mid-span section of the wing. The main emphasis in the experiment was, however, on defining the development of the upper surface boundary layer through separation (at about 20% chord ahead of the trailing-edge) and on into the wake, making extensive use of laser anemometry to measure mean velocities. In addition, Reynolds stresses were measured in certain parts of the flow field by hot-wire anemometry.

The flow field calculations are of the semi-inverse kind in which an inverse momentum-integral treatment of the shear flow, used to avoid difficulties at separation, is coupled to a direct solution of the inviscid flow problem. The main features of the method are outlined.