Beruflich Dokumente
Kultur Dokumente
Peyman Davoudabadi
Confidence by Design Workshop Chicago, IL June 14, 2012
1 2011 ANSYS, Inc. June 18, 2012
Agenda
Introduction to Turbulent Flows
Characteristics of a turbulent flow Challenges in simulation of turbulent flows
Agenda
Introduction to Turbulent Flows
Characteristics of a turbulent flow Challenges in simulation of turbulent flows
Large eddies carry small eddies The behavior of large eddies is different
in each flow Sensitive to upstream history The behavior of small eddies is more universal in nature
Reynolds Number
The Reynolds number is defined as
where U and L are representative velocity and length scales for a given flow. L = x, d, dh, etc.
Internal Flows
Natural Convection
where
5 2011 ANSYS, Inc. June 18, 2012
(Prandtl number)
Energy Cascade
Injection of energy from mean flow
Dissipation of energy
Flux of Energy
Dissipating eddies
3/ 4 l ReT
U i U i p Uk xk xi xk t
U i x j
DNS
= (n 3 / e) 1/4;
8 2011 ANSYS, Inc. June 18, 2012
t = (n /e ) 1/2;
v = (ne ) 1/4
Implication of Scales
Consider a mesh fine enough to resolve smallest eddies and large enough to capture mean flow features
Ncells ~ ( 3Ret )
where Ret = ut H / 2n
9/4
l (n 3 / e )1 / 4
Ncells = 4x107 !
0.003H Ret ut
June 18, 2012
Reynolds Averaging
Transport equations for mean flow quantities are solved All scales of turbulence are modeled Transient solution ( t ) is set by global unsteadiness
Filtering (LES)
Transport equations for resolvable scales Resolves larger eddies; models smaller ones Inherently unsteady ( t ) dictated by smallest resolved eddies
Both methods introduce additional terms that must be modeled for closure
10
Energy Cascade
Injection of energy from mean flow
Dissipation of energy
Flux of Energy
Dissipating eddies
3/ 4 l ReT
Resolved
Modeled
SRS
(Scale Resolving Simulations)
RANS
(Reynolds Averaged NavierStokes Simulations)
Numerically solving the full unsteady Navier-Stokes equations No modeling is required A useful research tool only
Includes Large Eddy Simulation (LES) The motion of the largest eddies is directly resolved in the calculation, in at least a portion of the domain, but eddies smaller than the mesh are modeled Inherently unsteady method
Solve Reynolds-averaged Navier-Stokes equations (time-average) Steady state solutions are possible All turbulence is modeled. Larger eddies are not resolved RANS turbulence models are the only modeling approach for steady state simulation of turbulent flows This is the most widely used approach for industrial flows
Feasible only for simple geometries and low turbulent Reynolds numbers
DNS is not suitable for practical industrial CFD
12
U Time-average of velocity
U Instantaneous velocity
RANS turbulence models provide closure for the Reynolds stress tensor, which represents the effect of turbulent fluctuations on the mean flow. This allows us to perform steady state simulations of turbulent flow.
13 2011 ANSYS, Inc. June 18, 2012
Below: RANS
140 cells, steady calculation, run time is order of minutes
Important
For wall-bounded flows, in a more typical 3D industrial geometry, RANS would still be 2 orders of magnitude fewer cells and run times of hours versus days.
15 2011 ANSYS, Inc. June 18, 2012
RANS
Agenda
Introduction to Turbulent Flows
Characteristics of a turbulent flow Challenges in simulation of turbulent flows
16
17
Example: Embedded Large Eddy Simulation of Flow Around the Ahmed Body (challenging/popular test case)
ANSYS-FLUENT R13.0 has been utilized RANS Model: Unsteady SST k SAS Model: Zonal LES-RANS technique: ELES Detailed LDA Measurements from Becker and Lienhart
U velocity profiles
18 2011 ANSYS, Inc. June 18, 2012 Ref: D. Caridi, et al., 8 th MIRA International Vehicle Aerodynamics Conference, 2010.
RANS
ELES
20
Example 1: DES/DDES of Separated Flow around a Realistic Car Model Exposed to Crosswind
Model Drag (SCx) Exp. 0.70 DDES 0.71 DES 0.75 LES 0.69
DES
DDES
Courteys PSA Peugeot Citron
22
Agenda
Introduction to Turbulent Flows
Characteristics of a turbulent flow Challenges in simulation of turbulent flows
24
Overview
The Role of Steady State (RANS) Turbulence Modeling Overview of Reynolds-Averaged Navier Stokes (RANS) Modeling Capabilities in ANSYS CFD Overview of models Wall treatment Model extensions and other interesting new features in R14.0 and R14.5
25
Providing state-of-the-art RANS modeling capabilities remains an important focus of ANSYS development
Base design
End design
Example: Optimization study achieves 1/3 reduction in pressure drop in U-bend over 30 different design iterations
26 2011 ANSYS, Inc. June 18, 2012
4% 11% 8% 11%
S-A SKE RKE RNGKE KE-Other K-W Std. SST NLEVM RSM RSM-BSL Other
Informal survey of single phase RANS model usage based on papers published in the Journal of Fluids Engineering during 2009 2011
The CFD user community requires a broad range of models to choose from in order to meet its needs
Over 2/3 of all simulations reported using some variation of 1 or 2 equation model (SA, k-e family, k-w family) In some applications, one model may be more dominant than others (example: aerodynamics & SST, cyclones & RSM), but for a broad range of applications, a variety of models is needed to match the appropriate model to the appropriate application
27 2011 ANSYS, Inc. June 18, 2012
ANSYS Models
It is not enough just to provide many choices More importantly, for the models that are available, emphasis is placed on
Correct implementation Models should be well understood and tested Accurate and validated for some class(es) of applications Robust performance on all mesh topologies Interoperability with other physical models, e.g. multiphase, dynamic mesh, combustion, etc. Wall treatment
Example: Solids suspension in a tall, unbaffled tank. Reynolds stress model together with Eulerian granular multiphase model
Courtesy of the University of Bologna
29 2011 ANSYS, Inc. June 18, 2012
u y
u Ut
30
33
Sensitivity of the skin friction coefficient to mesh density in an incompressible flat boundary layer modeled with Spalart-Allmaras
34
Boundary layer velocity profile modeled with Standard k-e for three different mesh densities using Enhanced Wall Treatment
35
5 m/s
Case 1
Air
Water
5 m/s
Air
Case 2 (Single phase case with only air flowing over moving wall) Wall Velocity = 1 m/s
36
EARSM
SST
38 2011 ANSYS, Inc.
Left: In-plane component of velocity vectors for Periodic flow in a square duct. EARSM (above) predicts secondary flow patterns with velocity ~2.4 percent of bulk velocity. SST (below) predicts no secondary flow Above and Right: Flow in a rectangular, asymmetric diffuser. EARSM correctly predicts pressure coefficient on bottom surface
June 18, 2012
Contours of velocity showing the impact of a porous jump on velocity in bordering cells
39
SRS/LES is a non-trivial subject: Optimal model selection is problem dependent Users need to understand models, application, grid and time step
requirements Optimal numerical settings important
ANSYS Hybrid Models are a substantial advancement in providing SRS capability to industry on today's computers
41 2011 ANSYS, Inc. June 18, 2012
Backup Slides
42
Assume l is a characteristic size of a larger eddy. Dimensional analysis is sufficient to estimate the order of large eddy supply
rate for k as: k / t turnover . The order of t turnover can be estimated as l / k 1/2 i.e., t turnover is a time scale associated with the larger eddies). Since e ~ k / t turnover, e ~ k 3/2 / l or l ~ k3/2 / e . Comparing l with :
l l (k 3 / 2 / l )1/ 4 3/ 4 3 Re T (n / e )1/ 4 n 3/ 4 l
Prediction Methods
45
The actual flow-field will not be known until you have computed the solution (and indeed it is sometimes unavoidable to have to go back and remesh your model on account of the computed Y+ values).
To reduce the risk of needing to remesh, you may want to try and predict the cell size by performing a hand calculation at the start. For example:
Air at 20 m/s
y
The question is what height (y) should the first row of grid cells be. We will use SWF, and are aiming for Y+ 50
(Recall from earlier slide, flow over a surface is turbulent when ReL > 5x105)
46
The target y+ value and fluid properties are known, so we need Ut, which is defined as: t
Ut
y y Ut
Ut
tw 0.82 m/s
The wall shear stress ,tw ,can be found We know we are aiming for y+ of 50, from the skin friction coefficient, Cf: hence: 2 t w 1 C f U y 2 9x10 -4 m y Ut A literature search suggests a formula for the skin friction on a plate1 thus: our first cell height y should be
0.2 C f 0.058 Rel
approximately 1 mm.
1 An
equivalent formula for internal flows, with Reynolds number based on the pipe diameter is Cf = 0.079 Red-0.25
2011 ANSYS, Inc. June 18, 2012
47