Class 8 Truck External Aerodynamics Choice of Numerical Methods Portland, March, 19 th 2013 Dinesh Madugundi, Anna Garrison PVE Vehicle Analysis Security Classification Line 1
Product Validation Engineering Vehicle Analysis Agenda � Motivation � Review of existing literature � Understanding Truck aerodynamics � Choice of numerical methods for Truck aerodynamics � Numerical methods comparison study � References Source: PVE Vehicle Analysis Daimler Trucks North America 2
Product Validation Engineering Vehicle Analysis Motivation Class�8 Truck with Standard 53‘ Trailer � Accurately predicting complex flow phenomenon in Truck aerodynamics using numerical methods can be challenging. Tractor trailer gap Trailer back face Wake interaction between the drive tires and trailer bogies Source: PVE Vehicle Analysis Daimler Trucks North America 3
Product Validation Engineering Vehicle Analysis Motivation Why CFD to Evaluate Class 8 Truck Aerodynamics? � Standard Class 8 trucks are ~2.8m wide and ~22m long including the trailer. � Very few full scale wind tunnels can accommodate full tractor trailer configuration, while simulating real road wind conditions. � Advantages in using CFD — Full tractor trailer configuration — Simulating real road conditions — Predicting performance of an aero component before modifying or installing — Wide range of design validations — Detailed flow visualization Source: PVE Vehicle Analysis Daimler Trucks North America 4
Product Validation Engineering Vehicle Analysis CFD Modelling Standard Numerical Methods � Reynolds Averaged Navier Stokes (RANS) � Unsteady RANS � Large Eddy Simulation (LES) � Detached Eddy Simulation (DES) � Detached Eddy Simulation (DES) Source: PVE Vehicle Analysis Daimler Trucks North America 5
Product Validation Engineering Vehicle Analysis Experimental Setup (Reference) NASA/TM – 2006�213489 � Bruce L. Storms AerospaceComputing Inc. “A Summary of the Experimental Results for a Generic Tractor-Trailer in the Ames Research Center 7- by 10-Foot and 12-Foot Wind Tunnels”. Generic Conventional model (GCM) of class;8 tractor;trailer 1/8 th scale results available for � validation study. � Simplified model of standard class;8 tractor;trailer, no grille opening, no underhood components. � � Experiments were performed at Re* = 1.1e6 – Experiments were performed at Re* = 1.1e6 – 6.2e6. � For this study, results from Re = 6.0e6 are compared, no T;T gap aero treatment, no trailer aero treatment. *Re was calculated based on Truck width. PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America 6
Product Validation Engineering Vehicle Analysis CFD Setup Geometry and CFD Mesh Overview � GCM 1:1 scale, closed grille, flat underbody. � No T;T gap aero treatment, no trailer aero devices � Computational domain with far field domain, moving ground, and spinning tires. � Mesh settings — — Base size = 40mm Base size = 40mm — Trim mesh, surface size 5mm – 40mm — Wake refinement — Low Re prism mesh — Total number of volume cells ~15M PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America 7
Product Validation Engineering Vehicle Analysis CFD Results of GCM Transient vs Steady � Flow conditions, constant density, Re = 6.0e6, 0yaw and 6yaw. No side extenders. � Solvers RANS, URANS and DES with Spalart;Allmaras turbulence model, are compared. � The three solvers predicted Cd that matched relatively close to the measurements. � RANS and URANS results matched well with each other. � DES results are closer to the measurements at yaw, compared to RANS. More accurate wake predictions? 80mph @0yaw: DES 80mph @0yaw: URANS Time Avg Ptotal Plots Time Avg Ptotal Plots 80mph @6yaw: DES 80mph @6yaw: URANS Plane View PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America 8
Product Validation Engineering Vehicle Analysis CFD Results Understanding Truck Aerodynamics 1/2 l ∑ Cumulative Cd Cd = x x x − l i i 1 − i 1 = � Production Cascadia sleeper, 45” T;T gap, and 53’ standard trailer. Trailer drag � CFD Methods: Current DTNA best practices. ~ 50% � About 50% of total drag is from tractor. Tractor drag � Yaw effects are predominant on trailer bogies and ~ 50% trailer back face. Tr � Sections of drag effects — A;surface � Stagnation pressure on the grille Cumulative Cd[;] plot over the length of the Truck, normalized by total vehicle Cd 0yaw . � Flow around the bumper and hood � Stagnation pressure on wind shield � Flow over the roof cap � Effectiveness of roof deflector � Effectiveness of side extenders and chassis fairings PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America 9
Product Validation Engineering Vehicle Analysis CFD Results Understanding Truck Aerodynamics 2/2 l ∑ Cumulative Cd Cd = x x x − l i i 1 − i 1 = � Sections of drag effects, cntd… — Underhood pressure: flow below the bumper determines underhood pressure. T�T gap Pressure — Underbody flow: chassis components and drive tires are exposed to high speed flow. — T;T gap: Pressure in T;T gap influences effectiveness of side;extenders and roof deflector. side;extenders and roof deflector. — Trailer bottom and trailer back face. Cumulative Cd[;] plot over the length of the Truck, normalized by total vehicle Cd 0yaw . Plan view: Z;section along tire center: 55mph, 0yaw Time Avg Velocity Magnitudes Plan view: Z;section along tire center: 55mph, 6yaw 55mph, 0yaw 55mph, 6yaw PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America 10
Product Validation Engineering Vehicle Analysis CFD Modeling Choice of Numerical Methods for Truck Aero � Choice of numerical methods is critical to capture complex flow phenomenon of Truck aerodynamics. — Surface bounded flow (Current industry standards, RANS, k;e, or SA). — Under the cab wake interaction (accurate prediction of vortex shedding). — Tractor trailer wake interaction. � RANS methodology with Low;Re mesh can achieve accurate boundary flow. � LES methodology to capture vortex shedding. — Highly mesh dependant in BL. — Can be computationally expensive. � For Truck aero applications, hybrid model DES (Detached Eddy Simulation) can deliver best aspects of RANS and LES methodologies. — Less sensitive to boundary layer mesh with RANS methodology. — Low Re mesh to accurately predict flow separation. — LES methodology to predict wakes; sensitive to mesh wake refinements. — Computationally less expensive than LES Source: PVE Vehicle Analysis Daimler Trucks North America 11
Product Validation Engineering Vehicle Analysis CFD Results RANS vs DES 1/6 � The CFD models are created using STAR�CCM+ v6.06.017 . The following numerical methods are compared for this study — RANS � Turbulence model – Spalart Allmaras � Time dependency – Steady � Segregated Flow � Segregated Flow � Wall Treatment – All y+ — DES (As per DTNA’s best practices) � Turbulence model– Spalart Allmaras Detached Eddy � Time dependency – Implicit Unsteady � Segregated Flow � Wall Treatment – All y+ PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America12
Product Validation Engineering Vehicle Analysis CFD Results RANS vs DES 2/6 l ∑ Cumulative Cd Cd = x x x − l i i 1 − i 1 = � The difference in total vehicle drag between RANS vs ∆Cd ~ 15% � 20% DES is about 15% ; 20% depending on yaw condition. � Delta Cd on tractor is about 5% ; 8%; resultant difference is on trailer bogies and back face. � Flow Comparison, ∆Cd ~ 5% � 8% — Flow separation over the hood. — More diffusion under the bumper. � Effects underhood pressure. � Higher drag on chassis and drive tires. Cumulative Cd[;] plot over the length of the Truck, normalized by total vehicle Cd 0yaw . — Difference in T;T gap pressure (influences roof deflector and side extenders’ performance). DES Time Avg Velocity Magnitudes RANS PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America13
Product Validation Engineering Vehicle Analysis CFD Results RANS vs DES 3/6 l ∑ Cumulative Cd Cd = x x x − l i i 1 − i 1 = � RANS predicted drag on the trailer is ~12% ; 15% ∆Cd ~12% � 15% lower. — Difference in drag is higher at 0yaw; can be accounted to wake interaction. — At 6yaw, the wake interaction is reduced due to free stream effects; shift in wake direction. — — Similar differences on trailer back face at 0yaw and Similar differences on trailer back face at 0yaw and 6yaw. � Transient phenomenon with controlled wake under the trailer? For example, trailer skirts. Cumulative Cd[;] plot over the length of the Truck, normalized by total vehicle Cd 0yaw . DES DES Time Avg Velocity Magnitudes RANS RANS PVE Vehicle Analysis 09.04.2013 Daimler Trucks North America14
Recommend
More recommend
