Cluster-based aircraft fuel estimation model for effective and efficient fuel budgeting on new routes
File(s)
Published version
Author(s)
Yanto, Jefry
Liem, Rhea P
Type
Journal Article
Abstract
Fuel burn accounts for up to 25%
of an aircraft’s total operating cost and has become one of the most important decision factors in the airline industry. Hence, prudent fuel estimation is essential for airlines to ensure smooth operation in the upcoming financial year. Challenges arise when airlines need to estimate the total fuel consumption of new sectors where data are not available. This necessitates the derivation of a robust parametric model that can represent the characteristics of the new route even in the absence of relevant data. To address this issue, we propose a two-step approach to derive a model that can accurately estimate the aircraft fuel needed. The developed approach involves both unsupervised learning and a regression model. For the unsupervised learning step, hierarchical density-based spatial clustering of applications with noise (HDBSCAN) is used to cluster the principal component analysis (PCA)-reduced data. This step can automatically separate flight sectors based on their underlying characteristics, as revealed by their principal components, upon filtering the noise in the data. Afterward, multivariate linear regression (MLR) is used to derive the equations for each cluster. The PCA-based clustered model is shown to be superior to using a global model for a single aircraft type. This approach yields fuel estimation with less than 5%
root mean square error for existing routes within each cluster. More importantly, the proposed method can accurately estimate the total fuel of a new route with less than 2%
aggregate error, thereby addressing one of the current limitations in the airline fuel estimation study.
of an aircraft’s total operating cost and has become one of the most important decision factors in the airline industry. Hence, prudent fuel estimation is essential for airlines to ensure smooth operation in the upcoming financial year. Challenges arise when airlines need to estimate the total fuel consumption of new sectors where data are not available. This necessitates the derivation of a robust parametric model that can represent the characteristics of the new route even in the absence of relevant data. To address this issue, we propose a two-step approach to derive a model that can accurately estimate the aircraft fuel needed. The developed approach involves both unsupervised learning and a regression model. For the unsupervised learning step, hierarchical density-based spatial clustering of applications with noise (HDBSCAN) is used to cluster the principal component analysis (PCA)-reduced data. This step can automatically separate flight sectors based on their underlying characteristics, as revealed by their principal components, upon filtering the noise in the data. Afterward, multivariate linear regression (MLR) is used to derive the equations for each cluster. The PCA-based clustered model is shown to be superior to using a global model for a single aircraft type. This approach yields fuel estimation with less than 5%
root mean square error for existing routes within each cluster. More importantly, the proposed method can accurately estimate the total fuel of a new route with less than 2%
aggregate error, thereby addressing one of the current limitations in the airline fuel estimation study.
Date Issued
2022-10
Date Acceptance
2022-10-17
Citation
Aerospace, 2022, 9 (10)
ISSN
2226-4310
Publisher
MDPI AG
Journal / Book Title
Aerospace
Volume
9
Issue
10
Copyright Statement
© 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
License URL
Identifier
http://dx.doi.org/10.3390/aerospace9100624
Publication Status
Published
Article Number
624
Date Publish Online
2022-10-20