The Aerodynamic Effects of Runback Ice

The objectives of this PhD were to investigate the aerodynamic performance effects due to runback ice accretion with particular interest on the EASA 45 minute hold case in icing conditions. The sponsors, Airbus and Cranfield University collaboratively identified this aim as a result of the successful creation and capture of full-scale runback ice using the Cranfield Icing Tunnel. The hold phase represents typical icing conditions but with increasing demands on airports and subsequent knock-on effects to increased holding times has led to airliners paying more attention to this phase of flight. Typical icing conditions occur during the hold phrase of flight and place increasing demands on airports. A further challenge is that the EASA 45 minute hold case fails to take into account the large supercooled liquid droplets (SLD) when certifying the airplane. Published literature open to the public domain on flowfield interaction with realistic runback ice shapes, force coefficient losses and heat transfer interaction is limited. This is despite the fact that these parameters play a significant role in the examination of ice accretion. Inspection of the runback ice casting highlighted regions of two-dimensional features which were used for aerodynamic analysis. These high-fidelity two-dimensional runback ice shapes were utilised throughout this project. A single hex-core hybrid mesh from an ANSYS ICEMCFD script was designed and served the dual purpose of assisting the process of optimisation and validation. The numerical validation procedure analysed three separate studies with differing airfoils. The three studies examined were B737-700, B737- 200ADV and NACA 23012. The first two studies were in clean cruise configuration and the third simulated forward-facing quarter-round ridge ice. The experimental validation process investigated the drag associated for the three run back ice shapes. Tests were conducted in the atmospheric boundary layer wind tunnel. A multi-objective Tabu Search optimiser was coupled with ANSYS FLUENT solver for investigations on ice location, shape optimisation using freeform deformation, vertical tail plane shape optimisation and leading edge heattransfer effects. The primary finding in relation to ice location optimisation was an observed sensitivity to chord location. Further exploration of this data identified two district characteristics for this sensitivity. These characteristics are the variation of run back ice height relative to the size of boundary layer thickness. More specifically, if the boundary layer thickness is smaller than the runback ice height, trends identified from this project are adhered to. However, if the boundary layer thickness is larger than the run back ice height, no visible trends are observed. The ice location study found optimum chord locations closer to the leading edge of the airfoil and sensitivity to small chord movements. The B737 shape study was conducted as a preliminary optimisation test to highlight best practices for shape optimisation using free-form deformation. The results showed more stringent requirements for geometrical constraint handling when dealing with shape optimisation. Small changes to the NACA vertical plane improved both clean and iced performance. This was attributed to the leading edge design producing a lower velocity around the airfoil. This was a highly constrained optimisation problem where improved results required an exhaustive and competent search mechanism. The MOTS code was best placed to conduct this search and was therefore used throughout this study. Anti-Icing optimisation incorporated heat transfer providing a more complete runback ice optimisation study. This more complete design comprised a multipoint optimisation code with three objectives and one variable; to minimise leading edge temperature output, maximise lift and minimise drag with variable heat input. The findings corroborate the results seen in the ice location optimisation study. The ice location study would benefit from an increased range of variability to observe the development of the trends found. Solver prediction capabilities for ice accretion studies would benefit from experimental data on roughness parameters associated with icing. The two-dimensional airfoil simulations would benefit from further CFD runs using three-dimensional swept wings. The Pareto-optimal designs are ideal candidates to compare the flowfield inadequacies alluded to by literature. As an interim study before threedimensional simulations are conducted, an extension to the cruise configuration airfoil study would be to deploy high-lift profiles. It would provide insight into ice accretion scaling methods by running a second NACA tail plane optimisation run using runback ice shapes scaled based on boundary layer thickness. Findings would provide invaluable information on the interaction of the two scaling methods with the boundary layer and how changes to flowfield characteristics effect force coefficients. Finally, further aerodynamic investigations with wind tunnel testing of representative full-scale runback ice shapes on a sectioned full-scale swept wing would be beneficial. This would complete the years of commitment and innovation by Cranfield University, sponsors and various students on the issue of ice accretion, particularly runback ice.