As essential parts of the engine, aero-engine casings have a considerable influence on the criticalspeed of the rotor and the dynamics of the whole engine. The accuracy of the dynamical predictionsobtained from the FE model of whole engine is largely dependent on the FE model quality of thecasings. However, the finite element analysis results, i.e. natural frequencies and mode shapes, are notalways consistent with measured results due to modeling errors and uncertainties of the structuralparameters. In order to improve initial prediction capability of the FE model, a model validationtechnique is investigated in this paper and applied to a typical assembled aero-engine casing.Firstly, the concept and the approaches of model validation process are systematically introducedand special attentions are paid to test planning, correlation, error location and model updatingmethods. In order to obtain the optimum experiment data, test planning analysis using the initialmodel is employed to design an optimum modal test in terms of specifying the best suspension,excitation and response locations. Using the Equivalent Element Modal Strain Energy (EEMSE) andEquivalent Element Modal Kinetic Energy (EEMKE), a novel method is proposed to localize errors inthe finite element model. Finite element model updating, using a first-order optimisation method, isapplied to correct the errors in the model.Then, the proposed model validation technique is illustrated with a simulated aero-engine casingand a real assembled aero-engine casing. Because of the complexity of assembled casing, a two-stepstrategy is proposed in the updating process. In the first step, model updating is carried out on the twosingle casings separately in order to validate the casing component models. In the second step, theassembled casing is updated with emphasis on updating joint parameters. The joints are modeledusing a layer of continuous solid elements that have material properties to be adjusted. The finalupdated FE model for the casing is able to predict natural frequencies and mode shapes close to themeasured ones. Therefore, it is capable to apply the model to the whole engine dynamic analysis.Finally, the influence of bolt joints on the dynamics of aero-engine casing is studied byexperimental means. The experimental results reveal that the dependence of the natural frequency onthe joint stiffness is found to be non-linear and appears to approach an asymptotic value from belowas the stiffness becomes large. The experimental results show that the randomness of assemblyconditions induced by bolted joints also has a significant impact to the nature frequency of jointedcasing. |