Excess Delay from GDP: Measurement and Causal Analysis

TL;DR

This paper develops a methodology to measure excess delay caused by Ground Delay Programs (GDPs) and analyzes factors influencing this delay using regression models, revealing key operational and program-related factors.

Contribution

It introduces a novel measurement approach for excess delay from GDPs and applies regularized regression to identify influential factors, based on extensive data from U.S. airports.

Findings

Average excess delay per flight is 35.4 minutes.

Ridge regression best explains variation in excess delay.

Factors like taxi-out time and program duration significantly impact excess delay.

Abstract

Ground Delay Programs (GDPs) have been widely used to resolve excessive demand-capacity imbalances at arrival airports by shifting foreseen airborne delay to pre-departure ground delay. While offering clear safety and efficiency benefits, GDPs may also create additional delay because of imperfect execution and uncertainty in predicting arrival airport capacity. This paper presents a methodology for measuring excess delay resulting from individual GDPs and investigates factors that influence excess delay using regularized regression models. We measured excess delay for 1210 GDPs from 33 U.S. airports in 2019. On a per-restricted flight basis, the mean excess delay is 35.4 min with std of 20.6 min. In our regression analysis of the variation in excess delay, ridge regression is found to perform best. The factors affecting excess delay include time variations during gate out and taxi out for flights subject to the GDP, program rate setting and revisions, and GDP time duration.