aeroelasticity study

[jcmbl]

Hi guys,

Does anyone have any experience in doing some aeroelasticity studies? I mean coupling CFD (e.g., Fluent) and FEA and issues such as pressure distribution from CFD into FEA mesh or type of appropriate FEA job. I've read a paper wherein a FEA bump is subject to aeroelastic forces. Each FEA static increment (NLGEOM on) was followed by a trained aeroelastic CFD data, since they are interdependent, and so on till the equilibrium was reached. There was no dynamic explicit involved.
Maybe there is someone who can share this sort of experience with others.

Kind regards,

J

 

[pja]

I do plenty of aeroelasticity work..mostly coupling our in-house CFD codes to ANSYS. However I don't know what exactly you want me to explain. There are plenty of papers out there that deal with aeroelasticity. If you have a specific topic in mind let me know and I can give you more specific references.

 

[jcmbl]

pja, thanks for the feedback.

From what I know, an iterative procedure until the convergence (i.e., oscillatory) criteria is met. I'm not sure exactly how to approach my problem: an airfoil with a simple actuated flap. I deflect the flap with a force F, run CFD, extract pressure p=1/2*rho*V^2*Cp, apply this p onto the structure and run FEM model with F+p fron unstrained state or just p from equilibrium state with F?
And so on with the virtue of convergence.

I've also carried out some optimizations on the airfoil, with no aeroelastic study. Jut wondering, is there any point in doing non-aeroelastic study and then continue with these results and add some aeroelasticity so that your final model to be feasible?

Mapping issue: if one has Cp on the airfoil, how can he map it onto the camber of the airfoil, considering there's nothing between the camber and skin? As I deal with camber in FEM.

I'd appreciate any hint.

J

 

[pja]

You should apply the full load you get from the CFD to structure at each timestep since the structural model is (normally) set up in a Lagrangian frame of reference. You then compute the deflection, move your fluid mesh with the new airfoil coordinates, compute the new pressure,etc. Since you are "loosely" coupling the fluid and structure you will need to subiterate this process within each timestep.

With regards to mapping the Cp to the structural model you will need to do some sort of interpolation. Since your model is 2D you will need to set up the interpolation using the x and y coordinate in your fluid and structural models.

With regards to the optimization..I'm not sure since I don't deal with this. BUt I would think it would depend on how flexible this airfoil is. The more flexible it is the more likely it will be that you will need to carry out the optimzation purely in an aeroelastic sense. If the model is not very flexible then you may be able to get away with running a purely CFD model, converge your optimization then add in the aeroelasticity for the final optimized model.

 

[jcmbl]

pja,

I'm still not clear. You are talking about timestep, so I guess your analysis is an explicit and time-dependent. That leads me to the complex part of aeroelasticity, i.e., dynamics (flutter, buffeting etc). I'll better start with static aeroelasticity, where your time step has no significance, but simply represent the size of the step. I'm talking in terms of Abaqus settings.
As I told you last time, I begin the procedure with FEM. I apply mechanical force F, get deflections, run CFD, get pressure p. In my understanding, in next iteration I run FEM with initial F plus p, which leads to another deflection state and then run CFD on it and get another pressure distribution p2. In the next iteration, I run FEM with F + p + p2 and then CFD, and so on. Please correct me if I'm wrong.

Indeed, the structure itself can be too stiff to be affected by aeroelastic forces, and only in this case I should alter the final FEM-on-based results with some CFD forces.

Thanks for the feedback

 

[pja]

Just apply the mechanical load F plus the current pressure distribution...the presure loads should not be additive. Think of it this way. If you were actually solving the fully coupled problem you would solve the system of equations once and be done with it. However since you are loosely coupling the two physics (fluid and structure) you need to iterate to make up for the inherent lag you are introducing. However at each iteration you only need to apply the current F and p loadings.