No abstract

The overall aim of the work reported in this paper is to explore whether a physics-based simulation approach has the potential to reduce the uncertainty & variability associated with both predicting & managing maintenance costs and improving engine design to optimise through-life economic performance. The main novelty in the paper is to demonstrate how an innovative Digital Geometry model can represent typical in-service component degradation and then support appropriate simulation meshes to permit degraded performance to be predicted. Two examples are given: blade erosion from particulates; and a simulated cooled blade burn-through event.

Most numerical engineering simulation is performed on pristine, as-designed representations of the components and systems in question. Although the rate of through-life performance degradation is hugely important when considering the total cost/benefit of a system, engineers have had profound difficulties in modelling the physical changes that components undergo in service due to the quasi-random and organic nature of the mechanisms such as wear, corrosion, icing and fouling. Typically, the creation of ‘worn’ models is based on a posteriori inspection or scanning of a failing or failed component. This paper presents a novel method for modifying geometry in response to scalar field variables directly accessed from the embedded physical models within physics-based simulation. It uses a distance field, managed as a Level-set, to drive time-dependent changes to the geometry surface, borrowing heavily from technology which has seen widespread use in the computer graphics industry to create and modify items in a natural organic way. A computational mesh can then be constructed around and within the modified geometry so that the simulation can be performed on the now ‘in-service’ version of the components. This greatly improves the predictive power of such simulations and provide a priori predictions of component performance in response to, for example, corrosive environments. The method is robust, can manage and create meshes for arbitrarily complex geometry, is insensitive to large scale topology changes such as hole blockage or passage burn-through, is highly suitable for automated simulation workflows and can represent both additive (fouling, icing) and reductive (erosion, corrosion, burn though) shape change.

This paper presents the first part of the development of a mesh manipulation application, MeshTools. The proposed tool eliminates the necessity of revisiting a CAD system when making geometrical changes on the CFD model by directly modifying the node coordinates. In addition to that, a robust built-in mesh quality optimization routine guarantees a standard quality for the output mesh. MeshTools was integrated with a widely used commercial CFD code and several turbomachinery related test cases are reported.