mdolab
Orthotropic material
Description of feature
I'm wondering if composites laminates would be straightforward to implement in OpenAeroStruct ? Would be neat for aeroelastic tailoring
Potential solution
Using a non-diagonal stiffness matrix would be relatively easy I reckon.
Yes, if you want to add it into OAS and make a PR, it's page 16 from this thesis.
https://www.diva-portal.org/smash/get/diva2:609349/fulltext01.pdf
I'm down ! Can you give me pointers about where that matrix is defined ? I have a couple of test cases at hand to verify the outputs.
Cheers !
@benoitpaillard I recommend using GitHub dev and doing a global search for the stiffness matrix definition in the source code. I don't know the code base well enough to point out the exact file.
Hi all, I've tracked down the stiffness matrix definition to these lines
Can anyone confirm that appending this would be enough for the definition of an orthotropic material ?
Hi @benoitpaillard that looks right. However, if you also want derivatives working for the orthotopic material so you can do optimization, you must also modify the compute_partials() routine just below it.
How do you think you will add the orthotopic stiffness matrix? As an if-check or something else?
At first I would like to try and implement something quick and dirty, most likely hard coded, and check the results against our results from paper
I noticed you also work on hydrofoils, are you interested in doing some testing as well? Do you have any good reference case in mind?
The experimental and computational results you linked would be a great test case. I would suggest starting with a simpler case though since a NACA hydrofoil is not exactly a beam given the cross-section. We'd also want to rule out that the flow forces are not the cause of any discrepancy so a simpler external load would be better than going straight for a VLM + composite beam test case. It seems like you added viscous corrections to your AVL model that would potentially disagree with the OAS VLM.
Since the FEM is a beam model, perhaps some of the composite beam validations from the VABS paper with a simple tip load would be good. Then, there is this distributed pressure load case from Fig 4. from Liao et al 2019
If any other test cases come to mind, I'll let you know, but this is probably a decent start. I don't really have the bandwidth to support any testing with OAS myself, but happy to provide general pointers on implementing this.
I'd like to pick up where this was left off, as it even with that work appears to be the easiest way to get rather explicit aeroelasticity modeling working to:
...roughly estimate how much structure/structural weight has to be incurred to meet airworthiness requirements for some innovative ultralight setups that deliberately exploit wing flex, for example via Dawn One-style outboard horizontal stabilizer tails acting stabilizing against flutter and providing roll authority through wing warping (moving the control reversal speed below the stall speed). Adding aeroelastic tailoring to e.g. reverse the sign of the aeroelastic feedback coupling that would normally result in divergence, to clamp the peak bending loads a vertical gust could impose, might be crucial to hit compliance with the rather substantial gust and maneuvering loads lighter classes tend to require.
Given that there is already support for B-spline control of tubular spar diameter and thickness, I'd imagine reasonable parametrization of fiber direction distribution around the cross-section of the spar, as e.g. wing flex loads prefer just the top and bottom to be thickened with unidirectional spanwise fiber, beyond what +-45degree wrapping torsional loads want. I think it would be reasonable to squish all those details down into a fully coupled stress/strain matrix for the beam and just record where the strain limit of failure modes (I guess just buckling and ultimate?) for the resulting linear 1D beam fall, to enable the optimizer to shave off structure until failure happens under prescribed loading, be it static spar structure or exceeding other limitations (like traditionally, demanding no control reversal at operating speeds). Then take this, I assume mere B-spline, model (in each variable parameter of the 1D beam model) and feed the normal aerostructural engine from it.
At least that covers steady flight; rolling out a sudden gust from an initial steady state for some reasonable duration unfortunately doesn't quite seem covered.
