Impact act of STAR-CCM CM+ + v7.0 in the Aut utom omotiv tive e Ind ndustr ustry Frederic derick k J. Ross, ss, CD-ada adapc pco Direct ector or, , Ground nd Transpor nsporta tatio tion
Vehic icle le Simulation mulation Compone ponent nts Vehicle Aerodynamics • Design Studies • Aeroacoustics • Water/Dirt Management Vehicle Thermal Management • Front End Cooling • Component Temperature Prediction Cabin Simulations • HVAC • Deice/Defog • Passenger Thermal Comfort Manufacturing Simulations • Paint Dipping • E-coat 2
Vehic icle le Simulation mulation Compone ponent nts Vehicle Aerodynamics • Design Studies • Aeroacoustics • Water/Dirt Management Vehicle Thermal Management • Front End Cooling • Component Temperature Prediction 3
Vehic icle le Simulation mulation Compone ponent nts Vehicle Aerodynamics • Design Studies • Aeroacoustics • Water/Dirt Management Vehi ehicle cle Aerod odynamics ynamics News ws Unsteady Aerodynamics • When to use? Reducing Turn-around Time • Taking Advantage of Coupled Solver for reducing • Automation Water/Dirt Management • New wall film model in STAR-CCM+ being used for water management 4
Steady eady vs vs Uns Unstead eady y Simulation mulation Should ould we us use steady ady or un unsteady eady simulat ation? ion? – Statement: Vehicle Aerodynamics is Unsteady True Statement • Use turbulence model to capture the unsteady nature • Model the Unsteady structures using LES type model – Statement: Need to have the right physics to get the correct solution • LES Models physics more accurately – Flow structure capture depends upon: » Grid Size/Time Step Size – Wall interaction: 2 Layer » 20-25 layers – Drag value is based upon a time average » Time length really depends on vehicle Example: Class 8 trucks: 10-30 seconds
Sample ple Stud udy S-Cl Clas ass Sedan dan – Model • A-pillar noise generation: target size 1mm • Side Window, Tire Wakes: Resolve to 2mm • Boundary layer on exterior: 2-Layer on wall. 20 layers Model Size: 500,000 million cells – Computer Resource • High Performance Computing Center Stuttgart (HLRS) • Cray XE6 Supercomputer – 3552 compute nodes 113664 cores Status: Currently running on cluster
RANS S Compr promi omise se (DES ES) Direct ct Ed Eddy y Simu mulat ation ion (DES ES) ) allows ws large e struc uctu tures res with th LES ES while le small ll, , un under der-res resolve e regions ions with th RAN ANS Positive Features • Capture large wake structures on typically able to be replicated with pure RANS solution • Reduces the need for fine grid to capture small structures – RANS can be used in regions of coarse grid • Can run with larger time step Negative Features • Still require long transients to get time averaged solution
RANS S Solut ution ion Run unnin ning g pur ure steady ady reduc duces s run un time Positive Features • Reduced Run Time • Accurately models flow approach to the vehicle • Accurately captures separation points – Still recommended 2-Layer grid for boundary layer development and prediction of separation. • Do not need to run long for averaging of results. • Not as sensitive to grid density Drawbacks • Not time accurate Wake structures not correctly produce • – Not ideal for multi-vehicle drag prediction – Looking at curvature correction to improve wake structures, but there is a limit to how far these will take us.
Exam ample ple 2: GCM Truck ck Benchmar hmark Det etail il Ex Experime rimenta ntal l data was taken en for a g generic neric truc uck. Purp urpose se was to help p test t and valida date e CFD Ran case with full Yaw Sweep: • 0,1,2,3,4,5,7,9,7,5,4,3,2,1,0 Between 2-3 degrees, solution is not stable
GCM Truck ck Benchmar hmark 0.8 Det etail il Ex Experime rimenta ntal l data was taken en 0.7 for a g generic neric truc uck. Purp urpose se was to Exp help p test t and valida date e CFD 0.6 RANS matches fairly well to Yaw 0.5 STAR- study 0.4 CCM+ Results are not symmetric, and only • 0.3 half yaw sweep completed -15 -10 -5 0 5 10 15 Averaged RANS fits reasonably on Yaw sweep • In location of high instability with RANS, running unsteady solver may provide higher time accurate result • Alternative: can look at case using DES DES Flow Animation
Uns Unstead eady y Simulations mulations Where e DES S is being g used: – Aeroacoustic Studies – Examining Rotating Wake Interactions • Fans, Blowers, Wheels – Vehicle Handling DES Sim imulat lation ion
Star ar-CCM CM+ + v7.02: Overse set t Grids ds More e comple lex passing sing requires uires more re comple lex grid d motio tion. n. – Studies have used rotating regions to aid in simulation of overtaking. – Mesh morphing has been used to change ride height. – Overlapping grids can simplify grid motion in the future. DES Sim imulat lation ion
Where here Steady eady RAN ANS S solutions utions are being ng us used? ed? Vehicle Design Tool Formula 1 Design High accuracy – Daily comparison to wind tunnel tests available • Fast Turn-around – Need to minimize CPU usage • DOE Studies Optimizing multiple variables – Running 100’s of studies to look at drag – reduction Being used today for Trains, Trucks, – Passenger Cars and Performance Vehicles Recent Workshop: “STAR -CCM+ CCM+ has pro roven en to be as accurat urate e – as our ot other in-hous ouse tools and easy y to setup etup while e pro roviding g a much ch faster er overall rall turn- around time” “ CD CD-adapco apco idea of DOE licens nse e scheme me – very appre recia ciated ed for design n optimi miza zation on studies ”
Vehic icle le Simulation mulation Compone ponent nts Vehicle Aerodynamics • Design Studies • Aeroacoustics • Water/Dirt Management Vehi ehicle cle Aerod odynamics ynamics News ws Unsteady Aerodynamics • When to use? Reducing Turn-around Time • Taking Advantage of Coupled Solver for reducing • Automation Water/Dirt Management New wall film model in STAR-CCM+ being used for water management • 14
Accelerating erating Converge ergence: nce: Couple pled d Solv lver er Coup uple le solver er has really ly been een pus ushed ed Elapsed Time (Hours) hard d over er the e past year ar by our ur F1 64 Cores teams ms – Latest testing shows significant 5.29 Segregated: 0 Yaw gains for all vehicles Couple: 0 Yaw 2.22 STAR AR-CC CCM+ M+ has some me feat atures s to help make runnin ing Segregated: 6 Yaw 13.36 com omple lex cases ses easier sier with h cou ouple le solv lver 7.67 Couple: 6 Yaw Expert initialization: Grid sequencing • CFL value should be similar to target CFL (120- 200) Elapsed Time (Hours) Expert Driver: Intelligent system which makes it easier to run with high CFL value 64 Cores AMG Acceleration: Bi Conjugate Gradient Stabilized 11.3 Segregated: Grid 1 • Improves rate of convergence. 9.5 Segregated: Grid 2 Couple: Grid 2 2.7 In 2012, we are investigating improvements to extend coupled solver to the vehicle thermal management
Vehic icle le Simulation mulation Compone ponent nts Vehicle Aerodynamics • Design Studies • Aeroacoustics • Water/Dirt Management Vehi ehicle cle Aerod odynamics ynamics News ws Water/Dirt Management • Rain management and soiling affect the safety, handling and aesthetics of passenger vehicles. 16
WATER ER DEPOSITION POSITION Modeling deling Rain in Water ter/Mist Mist: • Lagrang grangian ian droplet oplet model, del, active ive or froz ozen en gas as fie ield ld • Low w comp mput utat ation ional al cos osts ts • Genera nerally lly Freeze eeze flo low fie ield ld • Define ine mass/ ss/pa particle rticle siz ize • Eule lerian rian Multi lti-Phas Phase • Can n seed ed flo low fie ield ld with ith rang nge e of part rticle icle siz izes es Courtesy tesy M. Islam, am, Audi AG
Water er Managem ement nt: : Wall Film lm Model Capturin uring g ful ull affects cts of water er managem agemen ent: t: – Aerodynamics field for vehicle, – Lagrangian-Eulerian two-phase and coupled particle tracking of discrete particles in an airflow continuum, – liquid-film formation due to impingement of droplets on the surface, resulting liquid-film transportation, – and droplet stripping (re-entrainment) back into the continuum airflow due to edge effects of film instability.
Sample: ple: Side e Window Soiling iling
Plott tting ing Results sults Add dd two scala lar r di displ player ers: For the e dr dropl plets ts sel elec ect t as “Part” the “Water” Lagr grangi gian Phase e and d “Velocity Magnitude” as the e scalar. For the e Fl Flui uid d Fi Film sel elec ect t as “Part” the desired shells and “Fluid Film Thickness” as scalar.
Simulating mulating Wiper er Motion on Wall Fi Film Influenc ence e by win inds dshie ield d wip iper er – Part of the design of the A-Pillar is influenced by flow pushed by the windshield wiper blades Note: v7.04, expanding wall film for multi-component evaporation and condensation. This is a needed feature for SRC (Selective Reduction Catalyst) simulations for users to migrate applications from Star-Cd
Recommend
More recommend
