Fix level2 detailed takeoff example

[kanekosh]

Summary

This PR improves and fixes the level-2 FLOPS-based detailed nominal takeoff example.

• I added a mass_link constraint to the solved 2DoF problem configurator, like the height-energy configurator already does. Without this, the initial mass is fixed because it sets fix_initial=True for the mass state in the FLOPS ground-roll phase. Now it overrides the mass state to fix_initial=False and add an equality constraint to impose initial mass state = mission:summary:gross_mass.
• I made minor changes to num_segments and optimizer settings so the level-2 detailed takeoff example converges.

Questions

In the height-energy problem configurator, the mass link constraint is only added when the simple takeoff component is not added. I'm not sure if we should do the same for the solved-2DoF configurator (I guess the question comes down to: do we ever use the simple takeoff comp along with solved-2DoF EoM for main mission, or solved-2DoF is exclusively for FLOPS detailed takeoff?)

Also the height-energy configurator adds the range and mass residual constraints but the solved-2DoF configurator does not. This works fine for my case because I only use solved-2DoF mission for takeoff, but if the solved-2DoF phases are intended to cover the entire mission, range and mass residual constraints would be required. Currently a user would have to manually add these in their runscript (which in my opinion perfectly fine)

Note

It might also help other cases if we set a default limited_memory_max_history in Aviary's set_driver method. This is analogous to SNOPT's Hessian frequency. IPOPT's default for this parameter is 6, but in my experience carrying more history for Hessian approximation improves convergence.

Related Issues

• Resolves #

Backwards incompatibilities

None

New Dependencies

None