Optimisation of composite wind-tunnel wing models for frequency, flutter and divergence

A comparison has been made between the composite beam designs produced by minimum mass optimisation using two different sets of constraints. The first approach constrained the design to have a given separation between fundamental bending and fundamental torsional natural frequencies; the second constrained the design to have a given flutter and divergence speed. The beams are modelled as a series of elements, stepped in thickness at discrete nodes, with the Dynamic Stiffness Method being used for calculation of their natural frequencies. The aeroelastic constraints are obtained from the Fortran program CALFUN. The results show that for similar flutter and divergence speeds, the optima produced using aeroelastic constraints have a slightly lower mass (up to 4% lower) and a less ‘hard’ flutter onset. However, the time taken to produce these optima is significantly longer (in excess of 2 orders of magnitude). A preliminary study discusses the merits of a combined optimisation method where frequency constrained optimisation is used to provide a near-optimum starting point for flutter and divergence constrained optimisation. In addition, a wind-tunnel model of one of the optima has been manufactured and subject to both modal analysis and wind-tunnel tests to validate the flutter speed calculations. This shows that when using strip theory, CALFUN predicts a conservative value of flutter speed for this design. Further investigation has shown CALFUN's lifting surface theory to be more accurate for low aspect ratio models.