Numerical simulation of blade / casing rub interaction

Size: px

Start display at page:

Download "Numerical simulation of blade / casing rub interaction"

Bernadette Johnson
5 years ago
Views:

1 Numerical simulation of blade / casing rub interaction Olivier Beaupain, Techspace-Aero PP Jeunechamps, JP Ponthot University of Liège, Belgium

2 Partners involved University of Liège, LTAS-MN²L Aerospace & Mechanical Engineering department, Belgium Techspace-Aero (WP leader), Belgium Snecma, Villaroche, France Sulzer METCO, Zurich, Switzerland 2

3 Technical approach : Overview Original numerical wear method Validation of wear method on Sulzer rub test Finite element simulation of a blisk test 3

4 Original numerical wear method : framework Framework to numerically describe wear process: No new global Degrees Of Freedom into the system Focus on the blade/casing geometrical interaction (not on the local abradable behavior) Define a wear profile on the contact tool (constant during any time step) Use of constant wear profile to compute contact forces and achieve iterative equilibrium Update wear at the end of the step (post processing operation) Increase of CPU cost remains under control 4

5 Original numerical wear method : Definition of wear surface Abradable contact surface feature: - Must consist of 1 or several contact surfaces - Contact surface limited by a 4-lines wire - Contact surface can be a plane, a ruled surface, a Coons, a revolution surface 4 lines wire Meshed abradable contact surface 5

6 Original numerical wear method: definition of wear surface Meshing of contact side: Transfinite mesh Contact side for wear Wear cells = Quads (bilinear Coons) Wear points: ξ, η in side isoparametric space Initial and final wear stored at each Blade top wear point Correction of penetration of blade in eta abradable ( Gap) by interpolation inside each wear cell Geometrical wear law used: newwear = α(gap - oldwear) α = wear parameter to be identified gap = penetration of blade in abradable, function of contact pressure. Wear point Meshed contact side in isoparametric space Wear cell ksi 6

7 Original numerical wear method : Detection of weared points Detecting all active contact nodes at times t and t+ t ) X,Y,Z space Searching for every active blade contact Edge (edge of blade FEM mesh for which nodes are in contact at time t and t+ t). Defining a wear quad including all wear points to be weared between t and t+ t Blade edge at t eta Blade edge at t+ t For each wear point: Compute if wear point is inside the wear quad Compute local coordinate (ksi,eta) in wear quad space Evaluate gap by bilinear interpolation Update wear ksi Wear isoparametric space 7

8 Original numerical wear method : Simple pin test application 8

05mm/s Incursion depth: 1mm Numerical model: 2268 finite elements Elastic-plastic model for blade

9 Validation on Sulzer rub test : Description of test Experimental data: Rotation Speed : 3800RPM corresponding to tip speed of 200m/s Incursion rate : 0.05mm/s Incursion depth: 1mm Numerical model: 2268 finite elements Elastic-plastic model for blade behavior Wear model parameters: Penalty factor P Friction coefficient µ Wear parameter α CPU cost: between 1 and 2 days for 100 revolutions 9

blue lines correspond to top tours) Experimental Finite element Blade vibration modes OK 1st mode 2rd

10 Validation on Sulzer rub test : correlation test Contact periodicity: Experimental: essentially contact at every revolution (sometimes after 2 revolution) Finite element: contact at every revolution (bright blue lines correspond to top tours) Experimental Finite element Blade vibration modes OK 1st mode 2rd mode 3th mode 4th mode Experimental 200Hz 900Hz 1150Hz 2300Hz Finite element 250Hz 850Hz 1200Hz 1900Hz 10

11 Validation on Sulzer rub test : onset of divergence for some wear parameters values, change in frequency modes 2 revolutions before divergence Beginning of divergence Divergence New frequencies appear 2 revolutions after divergence Important bending of blade 11

Simulation of blisk test 56 blades 13000 volumic elements Overlength blade: 25 elements along height 22 elements along width 2 elements along thickness Other blades: 8 elements along height 11

12 Simulation of blisk test 56 blades volumic elements Overlength blade: 25 elements along height 22 elements along width 2 elements along thickness Other blades: 8 elements along height 11 elements along width 2 elements along thickness Blisk imposed rotation at the some nodes Initial unbalancing by applying a concentrated mass of 29.4g.cm Incompatible mesh between blades and blisk sticking contact considered (tied interface) Casing geometry: Revolution surface Overlength blade Imposed displacement P.P. Jeunechamps, J.P. Ponthot, LTAS- MN²L, ULg Initial unbalance 01/04/

13 Simulation of blisk test : Strain gauges position y y (mm) z (mm) Orientation ( ) SG SG SG SG Leading edge Trailing edge Trailing edge Leading edge SG 2 SG SG 2 2 SG 4 SG SG 1 1 SG 1 SG 3 P.P. Jeunechamps, J.P. Ponthot, LTAS- MN²L, ULg 01/04/

14 Simulation of blisk test : Evolution of wear footprint P.P. Jeunechamps, J.P. Ponthot, LTAS- MN²L, ULg 01/04/

15 Simulation of blisk test : Evolution of contact forces between overlength blade and casing irregular contact forces (bright blue lines corresponding to top tours) Full signal 250 contact revolutions 500 contact revolutions 750 contact revolutions 1000 contact revolutions P.P. Jeunechamps, J.P. Ponthot, LTAS- MN²L, ULg 01/04/

16 Conclusions and future work New numerical model able to accurately describe blade / casing interaction and able to describe the onset of divergence Wear model: only one simple wear model is implemented (geometry based). More sophisticated material models for abradable can be implemented but these models require relevant physical experiments to identify wear constitutive model parameters Blisk test: No onset of periodic wear footprint due to the perfect circular casing geometry insert elliptic casing or initial wear of abradable material To be correlated with future experimental test CPU cost to be improved, even if wear computation cost remains low with respect to blisk computation P.P. Jeunechamps, J.P. Ponthot, LTAS- MN²L, ULg 01/04/

Example 24 Spring-back

Example 24 Spring-back Summary The spring-back simulation of sheet metal bent into a hat-shape is studied. The problem is one of the famous tests from the Numisheet 93. As spring-back is generally a quasi-static