K. Hasanzadeh , S. Bourgault-Cote , E. Laurendeau , C. Brette , I. Paraschivoiu
Département de Génie Mécanique, École Polytechnique, C.P. 6079, Centre-Ville,
Montréal (QC), H3C 3A7, Canada
Département de Advanced Aero, Bombardier Inc., 2351 Alfred-Nobel
Montréal(QC), H4S 2A9, Canada
An Abstract to 62nd CASI Aeronautics Conference and AGM
May 19-21, 2015, Montreal, QC, CANADA
A newly developed CFD based two-dimensional ice accretion, NSCODE-ICE is presented. The method is used to predict iced airfoil shapes and performance degradation with a multi-step approach. A Multi-Block Navier-Stokes CFD code, NSCODE2D, a Multi-Block elliptic grid generation, NSGRID2D, and a Multi-Block Eulerian Droplet solver, EUCODE2D, have been coupled with the CANICE2D-NS icing framework. The new coupling allows fully automated multi-layer icing simulation while also permitting flow analysis and performance prediction of iced airfoils. Grid and flow sensitivity, Lagrangian and Eulerian droplet computation, effects of uniform surface roughness in quasi-steady ice accretion simulation, the number of time-steps, and the modification of the laminar flow heat transfer around the stagnation point are analyzed through different NATO validation test cases. The results demonstrate the benefits and robustness of the new framework in predicting ice shapes and aerodynamic performance parameters.
The Navier-Stokes flow solver NSCODE2D, a finite volume two dimensional multi-block Euler/Navier-Stokes flow solver with Full Approximate Storage Multi-Grid developed by Laurendeau et al. Implemented turbulence models include the widely used Spalart-Allmaras and k-ω-SST, γ-Rθ equations. The code has the wall treatment roughness model Boeing which is implemented and validated with the Spalart-Allmaras turbulence method. NSCODE2D is applied for steady/unsteady flow analysis and has the implementation of Chimera method. The code has been validated and verified through a variety of steady and unsteady case studies.
NSGRID2D is a two dimensional multi-block multi-grid elliptic/parabolic grid generation code developed at Ecole Polytechnique Montreal. The code is capable of grid generation and smoothing for complex domains such as 2D experimental glaze ice shapes. It includes a variety choice of control functions ensuring spacing, curvature and orthogonality requirements, including a novel blended approach. A novel automated curvature control algorithm is also implemented. The mesh generation solver includes several solution algorithms (Point-Jacobi, Gauss-Seidel, ADI, Line SOR) within the context of a full multigrid operator. NSGRID2D has been validated through different complex grid generation cases.
The developed code CANICE2D-NS includes four basic modules: grid generation, external flow simulation, droplet trajectory and local catch efficiently calculation (Lagrangian or Eulerian), surface thermodynamic balance and ice accretion, Figure 1. The Eulerian droplet solver, EUCODE2D, is developed at Ecole Polytechnique Montreal.
Fig. 1 CANICE2D-NS code structure.
Focus is on the development of a fully automated multi-step coupling procedure, capable of analyzing long ice accretion accumulation times in a quasi-steady formulation. The complexity of grid generation within the multi-step to account for large deformations with concave/convex topologies is addressed. The choice for automation of the meshing process via mesh regeneration is justified. Effects of icing roughness modeling, important in rough rime ice accretion conditions are examined. The effects of Lagrangian and Eulerian droplet computation on the robustness of ice accretion simulation are tested. CANICE2D-NS multi-time steps process has been validated using 8 individual NATO icing cases that are types of Rime, Glaze, and Mixed ice forms. Two examples are described here. NATO cases C09 and C17 input icing conditions are shown in Table 1.
Table 1 NATO cases test conditions.
It has been observed, the low skin friction and close wall velocity in the area of laminar flow (around stagnation point) computed by NS solver, results in a very low ice accumulation compared to the experimental data. This issue is modified by adding an elliptic smoother to smooth the skin friction and subsequently improved the computed heat transfer coefficient and ice shape, shown in Figure 2.
Fig. 2 Comparison: Skin friction (Cf), top left; Heat transfer coefficient (hc), top right; Ice shape comparison, down.
A number of results of CANICE2D-NS for the case C09 are shown in Figures 3 to 8.
Fig. 3 Beta comparison, Lagrangian/Eulerian (TS=1)
Fig. 4 Convergence (ks=0.0001, TS=5): Grid (left), Flow (middle), Eulerian droplet (right).
Fig. 5 Ice shape comparison (TS=5): ks=0.0001 (left), ks=0.0005 (middle), ks=0.001 (right).
Fig. 6 Ice shape comparison (ks=0.0001).
Fig. 7 Generated grid, (ks=0.001), (TS=5).
Fig. 8 Computed flow, (ks=0.001), (TS=5).
A number of results of CANICE2D-NS for the case C17 are shown in Figures 9 to 14.
Fig. 9 Beta comparison, Lagrangian/Eulerian (TS=1).
Fig. 10 Convergence (ks=0.0001, TS=5): Grid (left), Flow (middle), Eulerian droplet (right).
Fig. 11 Ice shape comparison (TS=5): ks=0.0001 (left), ks=0.0005 (middle), ks=0.001 (right).
Fig. 12 Ice shape comparison (ks=0.0005).
Fig. 13 Generated grid, (ks=0.001), (TS=5).
Fig. 14 Computed flow, (ks=0.001), (TS=5).
There have been in total 144 individual icing run for all 8 NATO cases. Results prove the robustness of the elliptic grid generation using the developed blended approach. Also it shows the robustness of Eulerian droplet approach compared to Lagrangian, as the number of Eulerian crashed cases was (2) compared to number of Lagrangian crashed cases (6). The skin frication elliptic smoothing (around the stagnation point) shows improvement in heat transfer coefficient computation and ice accumulation mass around the stagnation point, especially for the Rime ice problem. The average computation time of each icing layer is equal to 12.5 minutes. The main input parameters in NSGRID2D, NSCODE2D, EUCODE2D, and CANICE2D-NS has been adjusted the way to increase the robustness of the multi-time steps icing computation.
The work has been performed through a Collaborative R&D Grant No. 341083–06 with Bombardier Aerospace and the Natural Sciences and Engineering Research Council of Canada (NSERC). An appreciation to: Mrs. Vafa, Mr. Fortin, Dr. Pueyo, and Dr. D. Germain, for their help and support.
Topics: Aerodynamics of airfoils, wings, wing/fuselage interactions, nacelles, etc., inclu , Topics: Computational Fluid Dynamics as applied to any of the above, including surface mod
