Abstract

The use of virtual prototypes and the simulation-driven design have become vital factors in lowering costs and reducing the time-to-market while developing increasingly complex products like aeroengines. To simulate the structural behaviour of a design, the relevant physics of the real-world problem must be abstracted and idealized to form a mathematical description of the physics of the system. While there exist standardized procedures to verify the numerical representation of the computational model, the validation of the different modelling assumptions remains a major challenge. Since experimental data is limited during the virtual design process, the validation of the modelling assumptions often cannot be supported by a physical test. The NAFEMS Engineering Simulation Quality Management Standard (ESQMS) recognizes two methods for validation not supported by physical test: the comparison of simulation results with those from alternative mathematical models and the review by subject matter experts. In this paper the method of comparing models of different fidelity is adapted and applied to a turbine casing assembly of an aeroengine, consisting of a high-pressure turbine casing (HPTC), a low-pressure turbine casing (LPTC), an exhaust outlet guide vane (EOGV) and a rear bearing support structure (RBSS). For subsequent detailed analysis of the various engine components, the accuracy of the calculated interface forces needs to be ensured. To showcase the influence of different modelling assumptions on the simulation prediction, the models are correlated to a high-fidelity representation of the casing structure. Since the interface forces differ depending on the loading conditions, the difference in the stiffness and mass properties of the modelled structures are first evaluated by correlating the results of the corresponding free-free modal analyses. Static analyses and correlation of the models subject to a bearing load is performed to assess and compare the impact on the load share and load distribution caused by the modelling changes. Different methods are demonstrated which help to quantify the effect of the different assumptions.