Jbeam Aerodynamics : Can we use the old system?

We take backwards compatibility very seriously, so we don't change things lightly and without consideration of the possible side effects.

In the aerodynamics case, i've made the change due to the following reasons:

• First is that the new aerodynamics are considerably faster (around 15%).
  
• They are also more versatile, as adding features to them (like the stallAngle) is cheaper performance wise compared to older aerodynamics.
  
• New aerodynamics are a lot more stable with pressured vessels.
• Full volumetric pressure simulation would be impossible with older aerodynamics.

Apart from the advantages, there are also downsides, some of which you have elaborated about in your posts.

I'll try to explain the way that we work concerning these things. Most of the things that we are doing are very much dependent on research that happens at the same time. If we had in a single moment in time, all the knowledge that would be required to simulate a phenomenon then believe me, the last thing we would be doing is to change things around. We would implement the simulation once, in its complete and full form and then leave it as it is. Unfortunately things aren't so perfect, and as we read more of the bibliography and test more ideas our understanding of the phenomena increases. This can also lead to discovering other better or more versatile approaches.

From above you can easily understand that research is a big percentage of our work on simulation. Nevertheless, we try to balance research needs with modding needs, by trying to minimize the number of big changes that we do or by trying to create "bridges" that translate the old configs to the new ones (like we did with the drivetrain -> powertrain transition for example).

I've wrote all of above to try to communicate to you that we do not take designer' problems lightly. We are putting a lot of thinking and consideration on these problems. If you see problems like the aerodynamics one, it isn't because we didn't care or we didn't put the effort. It just is that some things are really hard, and we are still working on them.

Now concerning the problems due to changing the aerodynamic subsystem that you've described. You might think that having both the old and the new one as an option is something easy to do. Well it isn't. Aerodynamics are interlinked with other subsystems. To be able to have them both as an option, it means that i would have to develop, support, research for two different physics cores. One working with old aerodynamics and one working with the new one. The two physics cores would support different features and would behave in different ways depending on which aerodynamics is used. My workload would be more than doubled. Apart from having to work on double the code, i would need to test with both aerodynamics. I would have bugs that would happen with one or the other or both. And it wouldn't affect just me, the people that work on vehicle designs and vehicle simulation would also see increased workload due to similar reasons.

And the worse thing is what would happen when the research on the new-new aerodynamics produced results. Would we need to have 3 aerodynamic subsystems in there?

For the time being, my suggestion is this. Each aerodynamics approach has certain limitations. The problem with old aerodynamics was that the pressure/normals were not reliable, they moved around all the time and their overall physical characteristics could get bad as they moved around. The possible simulation outcomes could become unpredictable too.

The problem with the new aerodynamics is that it works best with a constant node/triangle density. So with configurations like the one you show, a big surface connected to small perpendicular ones, if the density isn't well distributed then the small surfaces end up affecting the big one in a not balanced way.

Finally i want to say that i care very much about the aerodynamics problems that you have elaborated on. Our own designers have been telling me the same arguments as you. I'm constantly thinking/testing ways to improve the overall situation, but as i've said it is a very hard subject. I also want to repeat that any suggestions/ideas that you have (apart from going back in time to old aerodynamics) are more than welcome .

 

You can just separate the end plates of the wing from the main airfoil. You can also just leave the end plates out, the main purpose irl is to constrain the flow and make the main airfoil more effective for a given area. Not really relevant. I don't think they would really act like effective vertical airfoils when there is a wing on the backside anyway.

 

Ok sure, 5 minutes of jbeam work sounds way harder. Alternatively you could probably attach the end plates with couplers or 0 length beams, but I haven't tried it.

 

The couplers (or 0 length beams) idea that Corey wrote about, is very similar to the old system. It'll also allow the wing to break in a more natural way.

It is also a small change to add the couplers (or 0 length beams), so it shouldn't be that big of a problem.