Inquiry: Dynamic Throttle Allocation for Heterogeneous Engines in 2DOF

[OscarEspinosaBarcenas]

Hi Aviary Development Team,

I am working on a mission analysis using two heterogeneous engines in a push-pull arrangement. Up to this point, I have been using the HEIGHT_ENERGY equations of motion, and everything has been working great. However, I am looking to transition my mission phases to use TWO_DEGREES_OF_FREEDOM.

Specifically, I need guidance on how to properly set up dynamic throttle allocation for this heterogeneous push-pull configuration within a 2DOF phase. I am already aware of the "perks" regarding SOLVED_2DOF (that it integrates over distance, struggles with optimization, and is slated for removal), so I am exclusively focusing on the standard TWO_DEGREES_OF_FREEDOM implementation.

Additionally, I would like to ask if Aviary can manipulate the angle of attack (alpha) and Thrust (T) directly node-by-node as control variables within the optimizer, instead of prescribing the Mach number and altitude using polynomials?

Thank you for your time and any help you can provide.

Best regards,
Oscar

[jkirk5]

Hi Oscar, sorry for the late reply, this discussion post seemed to have slipped through the cracks.

The standard 2dof equation of motions currently don't support multiple engines due to a lot of hardcoding, which I have been trying to untangle the last few weeks (see #1304). 2DOF is still relatively inflexible for throttle values (a byproduct of the GASP python translation we heavily relied on in early development), so right now I only have dynamic throttle for cruise, while the other phases still require user-set static values in phase_info for other mission phases.

For changing the controls, I believe this would require a minor refactor of the ODE and phase_builder to set up the desired controls and connect them properly. In this case you are basically removing the thrust=weight and lift=drag baked in assumptions. We have balancecomps in many ODEs that vary AoA or throttle to make those happen, in your case instead of target thrust being connected from aerodynamic's drag calculation, you'd want target thrust directly connected to your controls. And you could entirely remove the AoA balance since you are directly setting that variable, and actual lift would be a fallout of aerodynamics (and not guaranteed to be sufficient to lift the vehicle)