New trajectory builder and 1D performance model roadmap

[lrobion]

This is a summary of a discussion with @speth, @WyattGiroux, @ShreyaSharma2023 and I, please correct me if I misrepresent anything!

The goal of these changes is to make AEIC trajectories more flexible and representative of real flights, and allowing for the use of more detailed performance models (e.g. PIANO or TASOPT).

Flexible cruise altitude

Currently, AEIC sets the constant cruise altitude for an aircraft as ceiling - 7000ft. This is:

1. Not representative of the actual flight altitude distribution flown by an aircraft type
2. Does not allow for step climbs nor more flexible cruise altitude changes (e.g. contrail avoidance).

The new AdjustableLegacyBuilder enables overriding the the constant cruise altitude with another value passed by the user. This somewhat solves 1. except for the fact that the overriden altitude can cruise below the cross-over altitude (where the aircraft flies at its target cruise Mach) meaning we need to fly at a target CAS this sub-optimal cruise altitude but we have no performance data to support this with the LegacyPerformanceModel (I am not 100% sure, please correct me on this).

Point 2. should be addressed by Wyatt's new trajectory builder that implements a succession of flexible flight segments.

Non-legacy performance models

Current performance models passed to AEIC are 0-D (term coined by Wyatt, specifying FL and mass fully determines performance). TASOPT and PIANO have extra dimensions e.g. you would need to specify a Mach number as well (1D performance model). This performance models can be grouped by capability: essentially how many variables do I need to specify to estimate performance at a given point?.

My understanding is that Wyatt's new trajectory builder is built around that idea: it looks at how much flexibility does the performance model give the trajectory builder to specify the next flight segment.

Implementation

I assume all of this should be post V1, so this could be done on a branch of main while we finish V1?
This is just a suggested plan and I'd like to discuss this with everyone:

1. Extend make_performance_model.py to support creating AEIC .toml files from PIANO outputs
2. Create a 1DPerformanceModel (temp. name) performance model which under simplifying assumptions can be collapsed to a LegacyPerformanceModel that uses another data source (instead of the legacy BADA)
3. Modify the AdjustableLegacyBuilder (or make a new one given this is no longer legacy) to find a reasonable cruise speed when cruising below the crossover altitude
4. Add Wyatt's new trajectory builder
5. Add more realistic cruise segments (step climbs etc...) using the new builder and performance model capabilities

For point 2 and 3, part of that discussion would be what kind of data we want to get from PIANO/TASOPT into their corresponding 1D performance model in addition of what's needed by legacy.

[ian-ross]

Working on a branch before we do a V1 release would be fine. We're pretty close to V1 now: there are a few small things that @aditeyashukla needs to look at, then I'll cut a release and rerun simulations and emissions for 2019 (and then all years from 2013 to 2025). There shouldn't be any changes in the meantime that would interfere with working on new performance models or trajectory builders.

[lrobion]

Thanks! That makes sense, I'll create a new branch and make a PR on it soon.

[lrobion]

Ok it turns out I do not have write access MIT-LAE/AEIC to create a new branch. I tried doing so by creating a new local branch from up-to-date main named postv1-new-traj-perf-models and running:

git checkout postv1-new-traj-perf-models
git push -u upstream postv1-new-traj-perf-models

Would it be possible to create a branch for this project?

[ian-ross]

I don't have permissions to do that either, I think. @askprash Could give me and Louis write permissions in this repo? Not admins, but just enough to create branches, etc.