Science topics: Aerospace EngineeringLES

Science topic

# LES - Science topic

Explore the latest questions and answers in LES, and find LES experts.

Questions related to LES

In openFOAM, there are multiple options for choose in LES, such as WALE model, Smagorinsky model and so on. Which one is suitable for dynamic mesh and why ?

I am a bigginner of LES simulation, as far as I know , the only difference between them is WALE model includes the rotation rate in the calculation of

*νsgs, and*Smagorinsky not. What else should I take into account in simulation?- I am doing LES
- Subcritical and supercritical flow
- Either circle cylinder or square cylinder would work
- Preferably an experimental case for validation

Having the required time-series data and in order to analyze food demand system in a country (the aggregate level), How to implement regression procedures in one of following models: LES, DDS, AIDS or LA/AIDS?

I am open for collaboration

Thanks

Hello, I have a project coming up with Delayed detached eddy simulations.

I am trying to understand how LES works right now. can someone suggest a lecture series or video series on how to understand Delayed detached eddy simulation? How does LES evolve into Delayed detached eddy simulation?

Hello,

I am doing a parameter study for different grids doing wall modeled large eddy simulation for a channel flow. Could someone recommend me some papers which discuss the recommendations for such grids? I read papers which give clear recommendations but I dont find a lot why specific grids perform well. Its clear that at some point the grid is just too coarse, so I relate to fine enough grids with different aspect ration for example. I thought maybe it has to do with typical sizes of eddies in the turbulent channel flow but could not find many information about this. I know that it has much to do with the numerics used but are there also some physical reasons eg. elongated eddies in streamwise direction so that x+/y+ > 1 is reasonable? . Also I wonder about some results, eg. a 160*160*160 (x+*y+*z+) grid performing better than a way finer 40*40*40 grid in means of the mean velocity results. The simulations are carried out at Re_tau = 2000 on a finite volume code using WALE and Werner Wengle wall model.

Hi everyone. I am using DES model to simulation turbulent flow with RE 6500. I am only getting one vortex and not getting any vortex shedding. I tried to use URans model, but it shows no vortices and only wake region behind cylinder. Please inform me whats effect of model on vortex shedding.? Will LES provide vortex shedding or there is some other reason of not getting vortex shedding. Thanks

There are different versions of OpenFOAM. I want to know which version is more suitable for LES.

hi.

I have a problem in my simulation work. I use flow-3d for simulating a surface vortex with an air core in a pipe-tank system. I have been successful simulating the vortex with 1cm mesh size, but I haven't been able to see the air core. Due to this problem I had to had to use finer mesh with size of 0.6cm.

When using finer mesh ,my streamlines of the near surface particles change dramatically in comparison with the streamlines I had when using coarser mesh with 1cm mesh cell size.

In coarser mesh my streamlines are strongly sucked and driven into my pipe but when i reduce my mesh cell to 0.6cm, with exactly the same setup my streamline suddenly get shorter and incomplete. It seems that after changing the mesh size the flow of water in to the pipe is not enough strong to create the enough suction for near surface particles of water and due to this there i see no air entrainment afterward.

I will upload the photos of my streamlines in both conditions, I'll be so delightful if anyone can tell what is the cause of this problem I have and how I can solve it?

Dear all,

I am trying to run a Stress-Blended Eddy Simulation with a modified turbulent viscosity for the RANS part, only.

However, when I use DEFINE_TURBULENT_VISCOSITY to modify the RANS turbulent viscosity, it also modifies the LES subgrid model (what I want to avoid)...

I tried a simple if/else statement based on the value of the SBES blending function:

DEFINE_TURBULENT_VISCOSITY(user_mu_t,c,t)

{

if(SBES_BlendingFunction == 1)

{

mu_t = compute_my_RANS_mu_t(...);

}

else

{

mu_t = C_MU_T(c,t);

}

return mu_t;

}

But It does not give the expected results... I also tried to loop over cells and compute it locally but had issues with the implementation...

Any help or advices would be highly appreciated !

Thanks in advance!

I want to use LES to simulate the flow past a sphere at high Re by non-iterative PISO method, which is much faster than iterative method. However, the result of non-iterative PISO is far from the result in literatures, especially in Strouhal number. Can this method stable and accurate enough to apply to LES in turbulent flow? How can I improve the accuracy?

1.I'm using Ansys 2021 R2 fluent and the base of the meshed cuboid becomes finer when multizone method is used, to have a structured mesh. How do I overcome this issue?

2. How can I mesh a structured hexahedral mesh with the sizing I want ?

Note : The mesh is used to analyze the wind loads on high rise buildings using LES.

I have used body of influence to create multiple scales in the fluid domain.

I used to use RANS and I am now reading in LES. I just need a book about how LES is "applied" in equations. Most books, Pope's, Sagaut, ..., approach the problem mathematically. I just need, maybe, a flow chart or something.

I used, as an example, Finite-Volume in solving N-S equations and when it came to Turbulence I was using RANS. Here, I believe it is the same but to the limit of Sub-grid scales and then I need to model those.

I appreciate any help other than Pope, Sagaut ..

Thank you.

Which model is best to capture separation when simulating either a 2D or 3D of an oscillating airfoil?

For 3D simulation which model of hybrid LES/RANS is most suitable?

How do I improve on the model length scale?

Thanks

Both PANS (Girimaji, 2006) and TFLES (Pruett et al, 2003) provide a self-consistent formulation that allows one to recover RANS versus DNS at opposite limits. Are the two approaches fundamentally different (e.g., in the sense that RANS vs LES employ different ansatz and thus yield different types of closures to be modeled: the Reynolds stress vs the subfilter stress)? Or are they conceptually the same approach, independently developed and thus mainly differing in preferred modeling choices/perspectives?

(note: I am unfamiliar with PANS and only marginally familiar with TFLES; apologies for any misinterpretations)

For the RANS/LES hybrid turbulence model, is it possible to theoretically determine the distribution of grid nodes, including the RANS part near the wall and the respective grids of the LES part in the turbulent core area?

I am trying to have a c-grid with airfoil unit cell extension through Z for LES, when I assemble special domain along with airfoil upper lower and c curve the cells are not orthogonal near the boundary.

UPD (29/08/2018): Is the RANS/LES way of turbulence modeling a converging process, i.e. will it hypothetically converge to a hypothetical big but single PDE system? Or will it inevitably disassemble into several non-overlapping PDE systems/approaches each of which is valid only for a certain turbulent regime?

After more than 5 decades of active search, it seems that the hierarchy of moment equations for the Reynolds stress and higher moments derived from the Navier-Stokes equations does not have a universal closure that could be applied to more or less wide range of flows. Numerous closures have been proposed since then. All of them contain heuristic arguments which are usually changed from case to case. So, does it mean that our understanding of turbulence will never be complete? Or do people still believe that a “magic closure” exists which can explain everything?

I am currently working on a project involving shock / boundary layer interactions using experiments and CFD. I am looking for some books / papers where I can get the theoretical understanding of such interactions.

I plan to model the flow using RANS if possible, but I'm not counting LES or DNS out just yet.

Any additional info on this topic will also be highly appreciated.

What the differences between the SST k-omega and LES turbulence models for Pumps ?

My classmate is trying to run a CFD simulation over a farward facing step with conjugate heat transfer over the step and the working fluid is air and at Re= 5000, 7000,1000. But the size of the physical domain is about 1m length with 0.01 m width and a height of about 0.250m . If the meshing is done with a count of 1 milloin cells the simulation is taking months of time to run with a huge divegrence in the timestep. Is it possible to perform a LES simulation over such a big domain. For performing LES what would be the size of the physical domain and the respective mesh size.

Dear All,

Which type of simulations is good to investigate the rotating stall in axial compressor? RANS, URANS, LES, DES, and DNS. Secondly, a single passage or multi passage is recommended for the investigation of rotating stall.

Anyone know what should be the first layer thickness for the large eddy simulation. Also the quality of the mesh should be how much for LES.

Regards,

In turbulent boundary layer, the transition and turbulent regions could be clearly known by seen the Cf plot and averaged velocity profiles via transformation. However, how can I know where the flow enters streamwise fully-turbulent region in supersonic mixing layer by numerical simulation ? As I learn from open literature , combined with my own work, I get three ideas

1. Some empirical equation to ensure a sufficient length of the computational domain. For example, the criterion for xeff/dm1 > 500 proposed by Papamoschou and Roshko (1988).

2. Velocity power spectrum. Calculate the spectrum of a point and get the -5/3 and -7 slope.

3. The Reynolds stress profiles, which should be self-similar. However, the Urms profiles are hard to overlap, so what is the criterion to define 'self-similar'?

Actually, which one or more are rational? Do you have more answers?

I am trying to use WRF_LES to simulate a dust deposition case. My horizontal spatial resolution is 10m, and vertical spatial resolution is stretched from 0.96 m to 24 meter. As the vertical eddy viscosity is important for me and it has a big difference between using 'isotropic (lx=lz=(dx*dy*dz)^1/3)' and 'anisotropic (lx=(dx*dy)^1/2, lz=(dz))'.

So that would be great if somebody can give me some information about when isotropic/anisotropic is used and why?

Any feedback is welcomed, Thanks!

For performing LES initial soundings have been taken from ERA data. But large scale forcings and surface fluxes also needs to be computed, to provide these as input. Kindly suggest some suitable methods .

how to use steady state solution as initial solution for les ?

the problem is that Rans model produce average velocity field while les need instentenous velocity field,so it there a way to solve that problem?

I want to calculate the skin friction for channel flow.I use openfoam with shearStress utility for calculating the shear stress on the wall (vector form).Is it possible from the shear stress vector to calculate the friction factor using paraview?

When the buoyancy effect is significant in supercritical fluids passing in the tube, some eddies can form in any location of tube, especially at the vicinity of supercritical temperature. So, what is the numerical method to solve like the compressible problem? As my experience and literature, there is no way to calculate this type of problem by using RANS models, so I have just started using LES models. Although it was used a very low time step size (about E-5) and tried all of the subgrid models, I couldn't solve this problem. I would like to be grateful if you could share your recommendations. Thanks

I am trying to make sure I understand the limitations on the SST Transition model I'm using in FLUENT.

It is my understanding that RANS simulations have difficulty modeling large eddies (such as those which would form in the wheels of a race car) accurately, and it is my present understanding that the SST transition model would not completely capture that behavior, so even if one would expect such an eddy to form it would not be modeled accurately.

So if I wanted to model it with perfect accuracy, that would require an LES model.

Am I understanding that mostly right? I'm trying to make sure I have a thorough understanding because I might be questioned on it.

I am interested in the numerical characterisation (3D-CFD) of aerodynamics and flow-induced noise (near-field) generated during the operation of a standalone horizontal axis wind turbine. Since I am attempting this activity for the first time, I was wondering if there are any publicly available CAD and performance data of any test wind turbine to validate the numerical methodology.

On a side note, if anyone could recommend me some pivotal pertinent literature, it'll be a great help.

Also, as my knowledge on the state-of-the-art is limited, I was wondering about the practices for modelling impeller rotation. I understand that specifying an angular velocity to the rotating domain should be relatively straightforward but are there any literature/methodologies available to model the motion of the blades based on air-velocity boundary conditions.

Thank you for reading and contributing.

Hello All

Would like to know whether any one has Best practices/guidelines for setting up Embedded LES in Ansys Fluent.

I am confused how to define RANS in the global region and define LES in specific regions in the mesh. ie, two schemes in a single mesh.

I want to use UDF to modify the definition of subgrid length scale in LES model, which involves (1) judging the direction of local vortex axis (2) calculating the vector between grid vertex and grid center. I would like to ask you which macro command in UDF is needed for these two operations, or how to program them.

Thank you very much.

I have done LES and I want to extract information from that data, so I can understand physical aspects of the problem better regarding turbulence structures and how they affect the shock structure.

In LES non-premixed combustion of propane-air in Ansys fluent, there is no signs of combustion. The temp. is 300K everywhere. How to overcome it?

Hello All!

I am doing LES for flow around the 2d circular cylinder in fluent. I want to verify the Navier-stokes from my CFD results. Can somebody guide me on how I can do that? Thanks!

Hi

I am working on thermal stratification of dam reservoir. Could you please help me about choosing the right turbulence model?

K-epsilon

k-omega

LES

I need a periodic BC where mass flow rate is specified to model fully developed flow, and another periodic BC with no pressure drop. However, once I define the mass flow rate for one periodic BC, another periodic BC is defined with the same value. Can anyone help to solve this issue?

We are using a code where the divergence of velocity is neglected in the stress tensor for a low Mach approximation. We inject a pure helium plume in an air cavity. In PIV experiments the flow is more turbulent than in CFD corse DNS calculations. Kolmogorov scale is evaluated to be around 0.1mm. Mesh cell is around 0.3mm. We performed Smagorinsky LES and coarse DNS calculations. L2 norm convergence is estimated better than 7% of error. We wonder whether the Approx 20% of differences between experimental PIV measurements could be attributed to that neglected Div(u) term in the stress tensor, or not sufficiently converged DNS (or exp. problem). Has someone examples of the importance of this div U term in turbulence development ? Best regards?

After reading the proceeding: "Best Practices in Running Collaborative GPU Hackathons: Advancing Scientific Applications with a Sustained Impact" I came across the paper: "Porting the MPI Parallelized LES Model PALM to Multi-GPU Systems – An Experience Report". It is evident that GPU's are Single Instruction, multiple threads and therefore it is not a cache-optimized computer architecture. My experience is that unless you changed the whole kernel the Hybrid approach MPI+GPU affects the performance a lot. However, there is a boom in HPC with GPU's. In some instances, MPI alone performs better than the hybrid since we do not need to move information back and forth to the GPU and CPU, respectively. We have different good practices but there is no a standard or a reference that we can always take to the bank. The picture becomes even worse when high order schemes are used.

1) Which is your common approach for this issues in CFD?

2) Should we keep the structure of the source code as functional programming, imperative programming with a bunch of do .. end do in each subroutine that affects the performance a lot?

3) Should we use data region (where the data remains local in the GPU) where we packed all the computations even though we are hurting the readability of the source code?

4) Should we update the ghost cells in every time step?

Again the focus is only CFD

I have a question on wind flow around the building using CFD. Why the most of the people are using steady RANS (RANS) models instead of Transient (URANS and LES). The flow will be stochastic turbulence in nature, but most of the reputed papers on wind flow around the building, ventilation, pedestrian comfort etc.., are used SRANS (STEADY RANS), models.

Thanks advance

looking forward the answers.

Following this article

I sometimes had discussion with scientists about the "well-suited" using of URANS formulation for flow problems that are statistically steady (that is with no external time-varying driving force).

URANS solves the time-dependend Navier-Stokes equations on an unresolved grid, exactly as happens using LES. Without any explicit filtering or averaging, the only key difference between the two formulations is for the meaning of the turbulence model that adds some effect.

It is my opinion that for statistically steady flow, a real URANS formulation should drive, after a numerical transient, to a steady state solution to be congruent to its meaning of statistical solution (remembering that a statistical averaging is never really performed in the equations). Conversely, we see unsteady solutions that mimics the LES solution.

What really URANS is? Can be considered nothing else that a LES solution?

Hope to get your opinion,

Hello,

I am planning to run a research-led graduate course in turbulence modeling with a special emphasis on LES techniques. Please let me know which topics would you choose for such a course. Also, I'd like to hear your preferred teaching and assessment method. Any ideas are welcome.

The difference between the modeling system GAMS is that I can estimate the whole problem without thinking of the linear dependence problem with econometric system of equations estimations. However, I do not see how I could do the analysis without adding to external information like Frisch into the econometric model. All this I can get with GAMS in one model. For the interested persons, I attach the GAMS code (it is an amended version of a code I found in GAMS webpage).

hey guys

we are currently working on applying LES model on Lattice Boltzmann Method for simulating flow around cylinder and similar geometry.if there is any open source code which done this job its your kindness to address it to me

any relative comment is welcome

regard

I do not find this publication in research gate. Can I introduce the corresponding full text and in case how can I proceed.

P. Freychet, J. Pelletier, G. Rosselin, Nicole Grenier, Christiane Rouel, G.Fouilleul

DOSAGE DE

L’HORMONE THYRÉOSTIMULANTE HYPOPHYSAIRE OVINE A L’AIDE D’UN SYSTÈME

RADIO-IMMUNOLOGIQUE BOVIN. EFFETS COMPARÉS DE LA THYROÏDECTOMIE ET DE

LA CASTRATION SUR LES TAUX DE TSH ET DE LH PLASMATIQUES CHEZ LES OVINS.

Annales de biologie animale, biochimie, biophysique, 1969, 9 (4), pp.483-496.

1) COMSOL program has only RANS model for the turbulent flow but I need LES model.

2) I want to work only on COMSOL program.

Thanks.

Among RANS and LES which one is the best for simulating transient flow in IC Engine.

Hello All

I understand from literature that a completely resolved (Kolmogrov scale) LES is DNS itself. Any eddy resolving larger than kolmogrov scale are defined as LES. The effect of opted out eddies to the resolved eddies are modelled as subgrid stresses (SGS). Different modelling techniques for SGS include (1)Smagorinsky (2) WALES (3)WMLES etc. which are available in Ansys Fluent.

My doubts are :

1. Reference to Sagaut textbook, grid requirements for wall resolved and wall modeled LES is totally different, wall modelled being very coarse. (y+ min is ~30). In Fluent, various SGS models are available.

So Ansys Fluent provides a wall resolved LES with SGS models ?

**OR**how can we distinguish wall resolved & wall modelled in Ansys Fluent ?2. What is wall modelled LES (coarse meshes) and how can we implement it in Ansys Fluent. Is it just by making coarser meshes as per textbooks (y+~30) and applying required SGS models?

3. If we are making a very very fine mesh so as to capture Kolmogrov scale eddies, can fluent be used to solve DNS?

Hi all

I am trying to simulate turbulent pipe flow using LES method.

Literature:

Eggels (1994) used L/D = 5 for his studies.

In 1999, Blackburn used L/D = 2pi().

Hwang et al (2016) used L/D = 22 for their LES studies on pipe flow.

Q1. How do we arrive at required L/D of a pipe or length of a channel flow etc. when LES is used. As LES is computationally expensive, we cannot do a grid sensitivity. Does the two point correlations of required parameters (velocity, pressure etc.) help to obtain the minimum length required.

Hi,

I have the following parameters in my project (please see attachments).

I submitted a paper. Now, one of the Reviewer's comments is as follows:

**Validation is not enough and convincing. The CFD results should compared with static and dynamic torque and efficiency performance with LES simulations available in the literature and or experimental measurements.**

What is the exact meaning of "static and dynamic torque"?

I am grateful that give me useful links and papers in this case.

My used turbulence model in my project is K-W SST (RANS simulation). Also there are not any LES simulation available in this my field. My work is a 3D turbine.

Thanks.

Hi all

I understand that both are different flow problems.

When we try to simulate channel flow, top surface is defined wall for channel flows. For a flat plate of same dimensions, the top face shouldn't be a wall. Instead it is to be assumed that the flow is continous towards the top surface.

Can anyone help me with the boundary condition for a flat plate ?

Note - I am working with commercial code Ansys Fluent.

I am trying to estimate turbulent wall pressure fluctuations using LES for zero pressure gradient flows. From the literatures(1,2), I arrive at a conclusion about the grid resolution in terms of viscous wall units. delta_x_+ = ~55, del_y+ ~ 1, del_z+ ~ 15.

1. How strictly we need to follow these numbers to get reasonably accurate LES results. I have seen some journal papers using LES without following these grid requirements. Are such results valid ?

2. Do we have to keep these resolutions throughout the domain or in the boundary layer region only? I am using ICEM CFD & Ansys Fluent s/w.

Ref 1: Large-Eddy Simulation : Current Capabilities , Recommended Practices , and Future Research. Georgiadis, Nicholas J et al.

Ref 2 : LES for Acoustics, Sagaut P.

Hi,

Please can I have suggestions of articles and publications devoted for comparing LES results from those obtained by a DNS.

I have all the required statistical fields (all terms appearing in the RANS and FANS equations, in addition to the TKE and Reynolds Stress equations). I would like basically to estimate, numerically, a turbulent viscosity (scalar) from the tensors...

Best Regards

Hi,

In a simple form, the Boussinesq hypothesis reads

**R**= -2 nu_t *

**S**,

where

**R**and**S**are tensors (well known).The previous hypothesis works if the two tensors are aligned; basically if the cosine

( |

**R : S**| ) / ( ||**R**|| ||**S**|| ) > 0.86 (approximation).I would like to have some suggestions for cases of non-aligned tensors. Any method to estimate nu_t?

I thank you in advance and I look for your reply.

BR

Elie

i gave my first cell height is 0.0020 accordingly with roughness measurements (0.0019) Ks< first cell height. Can i run this model in LES. ( Model height is 0.7 , width is 0.15m and Domain height is 1.8 m all are accordingly with wind tunnel test section).

Its is possible to run a terrain profiles simulation in ANSYS Fluent, accordingly with roughness parameters limitation (Ks< first cell height) in LES?

I want to combine my LES (3D time resolved field) with experimental data (2D planar distributions) in order to improve my results. The flow field is highly unsteady. Data assimilation can be applied but probably would not be perfect or even worse. Are there any similar studies?

Dear authors,

How do you generate velocity time series for inlet boundary conditions for LES?

Thank you and kind regards,

George

Is there a later report on this evaluation - which tracks the outcomes after completing the program?

It would be good to talk about this if you have time.

And - thanks for uploading the Spingboard evaluation

Best regards Les

Dear colleague,

How to calculate the lateral stress exercise in the walls of the oedometer apparatus during a compression test in the field of non-linear behavior ?

Please let me offer methematical formul, or any experimental technique !

I wish to know whether it is ok to have a comparably large positive Lyapunov exponents for a hyperchaotic system. I am working on a system which gives me LEs in the order like this : 19.4456, 1.45002, 0.0000, -28.4581, when simulated with the Wolf et al's algorithm. Is such a figure feasible for a 4D continuous-time hyperchaotic system? Thanks

Large eddy simulation grid requirement mention about Delta X+ = 100, Delta Z+ = 30 and y+ = 1. Y+ is known but what do you mean by Delta x+ and Delta Z+, x is streamwise and z is spanwise directions,

les variables adéquates pour les utiliser dans TpsDig

Hi all,

So after failing to get satisfying results to my simulation using the RANS model, I decided to switch to LES. As I am expecting some flow separation, after reading some papers, I decided to use the One Eddy Equation LES model. However, I am not sure about what BC should I set for the two fields; k (kinetic energy) and nuSgs (the subgrid scale viscosity), sepecially at the walls where I have the no Slip condition for the velocity.

Also should my mesh be fine enough at the walls?

I want to perform LES simulation for airflow through Human Respiratory tract and I have approximately calculated the kolmogorov time scale of 10 microns getting confuse about the total number of time steps for ANSYS FLUENT. Please someone help me?

Hi,

I'm trying to validate LES with a turbulent channel case.

Assuming the SFS model works correctly (I actually tried several that gave very good agreement one to other), what can cause the results to be far off from DNS data?

Obviously using the same normalization as the DNS case and hopefully a sufficient mesh (was shown to be suffice in several papers).

Just to point out:

The shape of the results is generally OK, but the wall sheer stress / friction velocity are incorrect, therefore resulting in an incorrect mean velocity profile.

What do you think are the minimum benchmark cases that a LES solver must pass before being validated?

As implied in the question, this is strictly subjective so I would like to get as many opinions as possible. My interests lies in three-dimensional turbulent flows.

Hi,

Can I devise a refinement criteria for AMR (Adaptive Mesh Refinement) in such a way it satisfies implicit LES requirements? What are the requirements for ILES exactly?

Or is it possible to think of a refinement criterion according to one of the explicit models of LES?

I'm mainly interested in flows over a bluff body.

When doing a turbulent simulations with models such as LES, how can I compute the ensemble average of the instantaneous velocities. I know that there is an option for calculating "time statistics" in Fluent, but I think that option just calculate the time average between the initial and final time points (1 value for the whole simulation). What I am looking for is unsteady ensemble average. Actually, I am not completely sure how it is supposed to be done, since velocity fluctuations do not have a single wavelength. So the more general question would be: If I have u(t) how can I calculate U_bar (t) (assuming: u(t)=U_bar(t)+u'(t))?

Any help will be appreciated.

to simulating in-cylinder swirl, what kind of LES model is the best?

I have been working with ILES in OpenFOAM for numerical simulation of partial sheet cavitation. I would like a backup of your opinions about the subgrid model and subgrid dissipation.

I want to simulate the flow field with high resolution especially near the wall region in open channel flow with different Re. Can I use LBM, efficiently? Or VFM solution of Navier-Stokes equations using LES as turbulent model is more efficient from accuracy and computational cost (i.e. time) point of view.

I should be noted that my problem is a incompressible multi-phase flow.