Design of a decision support system for the fuel ordered in an airline
Loading...
Date
2019
Authors
Director
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Javeriana de Cali
Abstract
Description
item.page.descriptioneng
In an airline operation, the fuel cost is the main driver of the total expenses. To deal with the volatility of the jet fuel price and the high-cost percentage, fuel management projects with the aim of delivering higher fuel efficiency need to be made. After an exploratory analysis, it was found that there was a great amount of fuel that could be considered as waste. It was also detected that there is a strict correlation between this quantity and the extra fuel tanked at briefing. According to this, the purpose of the study was to design a decision support system to be used at briefing, in order to give a better estimation of the extra fuel, based on the remaining fuel prediction. To develop the system three components were considered: database, model and user interface.
After selecting the respective database, linear regression and neural networks were developed for the second category. The first one performed slightly better in terms of RMSE and R2, nonetheless, statistical assumptions such as normality of the residuals did not comply. Therefore, the artificial neural network was selected as part of the DSS model. The user interface was designed to be included in the fuel order template of the company following a user-centered design. Once the DSS was structured, the savings were estimated by using historical data. Finally, the system was validated using a system dynamics approach in order to study fuel consumption behavior.