Aeroelastic model identification of winglet loads from flight test data

Abstract

Numerical computational methods are getting more and more sophisticated every day, enabling more accurate aircraft load predictions. In the structural design of aircraft higher levels of flexibility can be tolerated to arrive at a substantial weight reduction. The result is that aircraft of the future can be bigger, have better performance and less mass. The performance of an aircraft can be even further enhanced by the use of winglets or other wing tip devices. A more flexible structure in combination with larger dimensions can lead to substantial structural deflections. Due to these larger deflections, the interaction between the aerodynamics and structural mechanics is of increasing importance. Due to their outboard position, the aerodynamic performance of wing tip devices is obviously significantly influenced by the deformation of the flexible wing. Off course, a safe and reliable operational life of the aircraft has to be guaranteed and proven with adequate design calculations. The goal of this thesis was to develop an algorithm to enable the identification of flexibility effects on the outer wing within a manoeuvre loads context based on the Maximum Likelihood Method. The main difference with approaches of existing publications is that the models considered here are based on distributed local data rather than on the net effect on aircraft performance. While this requires the size of the specific models to be much larger, the identified models allow a much more detailed physical interpretation of the observed performance benefits or penalties of winglets or wing tip devices. There are many references that address the topic of aerodynamic performance of wing tip devices and also of winglets in particular. These studies are all based on either wind tunnel measurements or pure aerodynamic (CFD) analysis, thus valid for rigid aircraft. These studies are very important in understanding the complicated flow condition at the wing tip in order to arrive quicker at even more efficient designs. However, flight test measurements have shown that flexibility of the airframe has to be taken into account when predicting the (aerodynamic) loads on the winglet. An algorithm was developed that is able to identify the parameters in a nonlinear coupled aero-elastic manoeuvre loads model. The algorithm is based on the Maximum Likelihood method which is capable of solving even rank-deficient problems. This identification procedure is applied for a loads relevant industrial case using real flight test data. The identification procedure is performed five times using these in-flight measurements with modifications in the aerodynamic modelling on the wings and winglets. One model was developed that describes the nonlinear rigid behavior and could be optimised in an identification for a best fit to the flight test measurements. It was found that especially the local alpha-gradients on the winglet are much larger in this model as predicted by the corresponding results derived from the original aerodynamic database. These identified gradients were compared with the gradients determined from the CFD-simulations and it was shown that they correspond very well. The success of the identification of a specific model strongly depends on the structure of the model and the assumed initial values. The model must be sophisticated enough to capture/describe the phenomena contained in the measurements, however simple/small enough to enable its identification with the available computational resources. The identification algorithm from this thesis was shown to be a very good means to quantify model improvements during model development. Secondly, it can obviously be used to identify the most optimal values for the free model parameters.