Multi-Stage Formation Flight Planning

Multi-Stage Formation Flight Planning

Doole, M.M.

Visser, H.G. (mentor)

Aerospace Engineering

Air Transport and Operations

2016-08-24

This thesis investigates the concept of fuel-reduction via formation flight of long-haul commercial aircraft on a network-wide scale, particularly addressing the issue of combinatorial complexity in global formation routing and allocation. A multi-stage formation flight planning (MS-FFLIP) tool is proposed as a computationally tractable approach to formation allocation. Assuming all trailing aircraft receive a 10% reduction in the fuel-burn by flying in formation, the optimal routes including the joining and splitting points are computed in four specific stages along the flight depending on the desired formation size. Within a stage, the interconnected allocation problem takes a set of flights, their possible formation combinations and associated costs and optimally assigns them in a fuel-cost minimising formation fleet. The formation flight routing and the associated fuel burn are enumerated rapidly. This fast computation allows the large-scale allocation problem to be solved via a Mixed Integer Linear Program and optimised through an external optimiser solver. Within the context of this thesis, a stage is defined as a collection of events where all flights merge into formation at optimal joining points, driven by an event-based simulation. The MS-FFLIP tool sequentially creates stages until the largest allowable formation size is created. Additionally, the model is capable of handling large sets of flights and in creating large formations of up to sixteen-aircraft allocated to a formation within a real-time operational time limit. Similar models exist in the current literature, however they are highly prone to combinatorial complexity and as a result, their computation times diverge dramatically with increase in aircraft number and formation size. For this reason, the models were not able to capture the majority of the potential fuel-savings. The MS-FFLIP model was compared against the conventional single-stage centralised formation flight planning approach for which 50 eastbound transatlantic flights were evaluated for optimal routes and assigned to formation fleet of up to four-aircraft per formation. The experiment for the centralised formation flight planning approach showed a 5.72% reduction in fuel-burn against its corresponding solo flights, while requiring 58 CPU minutes of computation time. On the contrary, the MS-FFLIP model performed the same experiment within a time frame of 14 seconds while demonstrating potential fuel savings of 5.66%. Moreover, a case study for 267 transatlantic flights was simulated by the MS-FFLIP tool. The results revealed a 6.88% in potential formation fuel savings against solo flights, with up to sixteen-aircraft allocated to a formation, while performing the computation within 9 CPU minutes. The developed model also showed significant fuel saving potentialcompared to local optimisation-based cooperative planning frameworks.

Formation Flight
Centralised Formation Flight Planning
Flight Operations Optimisation
Airline Fuel Saving Strategy
Combinatorial Complexity
Combinatorial Explosions
NP-hard

http://resolver.tudelft.nl/uuid:44cb036b-51b8-4b1e-ba1e-a7d59936caa1

Student theses

master thesis

(c) 2016 Doole, M.M.