SPEED-UP GEARBOX SIMULATIONS BY INTEGRATING SCORG Dr. Christine Klier, Sahand Saheb-Jahromi, Ludwig Berger*
CFD SCHUCK ENGINEERING Engineering Services in computational fluid Dynamics (CFD) 25 employees at 3 locations in Germany more than 25 years of experience established in Heidenheim in 1990 offices in Munich (1999) Gaimersheim (2014) Pressure [Pa] Office Heidenheim Office Munich Office Gaimersheim Bahnhofplatz 3 Ingolstädter Str.22 Am Augraben 21 D-89518 Heidenheim D-80807 München D-85080 Gaimersheim Tel. +49 (0) 7321 34 93-3 Tel. +49 (0) 89 35 82 80-6 Tel.: +49 (0) 89 35 82 80-82 Fax +49 (0) 7321 34 93-59 Fax +49 (0) 89 35 82 80-89 info@cfd-schuck.de www.cfd-schuck.de
OUTLINE 1. Motivation 2. State of the art Gear Simulation 3. Modelling Setup and Mesh 4. Results 5. Conclusions and Outlook
1. MOTIVATION Gear lubrication poses a significant concern in various industries. Prototype testing does not always provide the information detail required. CFD model prediction is an effective tool for analysing oil injection systems and oil flow around rotating components in a gearbox: Improve the efficiency of transmissions Minimization of load-independent spin power losses Assessment of wall effects on gear housing Thermal management
1. MOTIVATION Current CFD-methodologies available for gear system investigations: Particle based methods Grid based methods => STAR-CCM+ Choice of methodology depends on real task Lubrication => short response times => particle methods High accuracy results => include all physics => STAR-CCM+ CAD data from KISSsoft (www.kisssoft.ag) => limited due to computational time Feasibility study using SCORG Accuracy Robustness Effort (man-h, hardware, software)
2. STATE OF THE ART (STAR-CCM+) Generic spur-gear system Oil sump different filling levels & heat transfer Oil sump & gear contact Oil sump & injection system & oil film thickness
3. MODELLING SETUP AND MESH 3.1 GEOMETRY AND BOUNDARY CONDITIONS Material data: Phase 1: Oil (C12H26, ISO VG 220, 100 C) Phase 2: Air (ideal gas) Solver Settings (STAR-CCM+ Vers.11.04 ) k-w-sst-turbulence model Gear housing: d = 280 mm, l = 100 mm Gears: d=130 mm, l=58 mm, n T =20 Operation Conditions: Gear 1,2 = 2000 rpm Total number of rotations 2 Physical Model VOF No zero gap interface
3. MODELLING SETUP AND MESH 3.2 MESH Overset Mesh: 4.4 Mio cells half model SCORG: 3.2 Mio cells full model
4. RESULTS 4.1 ROTATION IN AIR Minor differences but same level
4.2 RESULTS: OIL FLOW (T= ½ R) Overset Mesh SCORG
4.2 VELOCITY DISTRIBUTION (2 ROTATIONS) Overset Mesh SCORG Similar velocity distribution Higher peaks of velocity with Overset Mesh
4.2 PRESSURE DISTRIBUTION (2 ROTATIONS) Overset Mesh SCORG Similar compression/expansion zone at gear contact Oil shearing at tooth tips => Pressure Peaks
4.2 VOF (2 ROTATIONS) Overset Mesh SCORG Similiar distribution of oil Different oil paths below the upper gear
4.2 TORQUE (GEAR 1) Similar level of torque after 2 rotations
4.2 VOF (GEAR 1) Similar level of 90%-VOF on gear surface after 2 rotations
4.2 COMPUTATIONAL EFFORT Overset (Half model) SCORG (Full model) Number of cells (-) 4.4 Mio 3.3 Mio Background grid 3.1 Mio 1.5 Mio Rotor grid 1.3 Mio 1.8 Mio Number of inner iterations 30 15 Simulation Δt 2.08E-05 6.00E-05 Time per Iteration & 1 mio. cells & 1 CPU (s) 0.41 3.86 Time per rotation (h) N cores 26 h 80 Cores 1) 28 h 6 Cores 2) 19 h 12 Cores 2) 1) supercomputing facility 2) personal computer
4.2 COMPUTATIONAL EFFICIENCY (SCORG) Non linear speed up with influence of parallel architectures Linear curve progression within MPI range => Efficient usage of HPC possible
4.3. EVALUATION Overset Mesh SCORG (accuracy) 0 0 physical models 0 - robustness 0 0 effort 0 + Man-h 0 0 CPU 0 ++ costs 0 - Based on a single investigation => preliminary character Accuracy-proof requires comparison with test data
5. CONCLUSIONS AND OUTLOOK Investigation of using SCORG for simulation of a spur gear systems Feasibility study to proof accuracy, robustness and estimate effort for application Comparison with results from Overset Mesh methodology Good agreement concerning torque and oil movement High potential for reducing simulation time (5-10 times faster) Further works to achieve Higher accuracy Comparison with test data Test of additional model capabilities Fluid film model / Heat transfer More complex geometries
Thank you for attention!!