Reassessing the ICAO’s Standard Taxi/Ground Idle Time: A Statistical Analysis of Taxi Times at 71 U.S. Hub Airports

Abstract

Taxi time plays a critical role in airport capacity, aircraft fuel consumption, and emissions. It is defined as the time from touchdown to the gate and from the gate to liftoff. The International Civil Aviation Organization (ICAO) established a standard taxi/ground idle time-in-mode (TIM) of 26 min in the landing and take-off (LTO) cycle for modeling turbine engine aircraft emissions. However, actual taxi times vary significantly across airports. While a simplified standard streamlines emissions modeling, the 26 min assumption may not accurately reflect real-world conditions. While using airport-specific taxi times may not always be practical, hub classifications of U.S. commercial airports may affect taxi time and serve as a compromise between airport-specific taxi times and a simplified standard. Therefore, this study statistically analyzed Federal Aviation Administration (FAA) data from 71 U.S. commercial hub airports to compare reported taxi times with the ICAO’s standard and assess the influence of airport hub classifications. The exploratory findings indicate that the 26 min ICAO taxi/idle TIM does not represent reported taxi times at 70 of the 71 sampled airports. Moreover, total taxi time varied by hub classification: small-hub airports had a mean taxi time of 19.82 min (median: 18 min), medium-hub airports had a mean taxi time of 19.72 min (median: 18.25 min), and large hubs had a mean taxi time of 26.98 min (median: 25.08 min). When hub classifications were ignored, the overall mean taxi time was 23.78 min (median: 22 min), indicating a statistically significant difference between the ICAO’s standard 26 min assumption and the observed taxi times at most airports.

1. Introduction

Airport taxi time contributes to airport capacity and aircraft fuel consumption and emissions. The Federal Aviation Administration (FAA) [1] defines taxi and ground movement operations as follows: “issue the route for the aircraft/vehicle to follow on the movement area in concise and easy to understand terms”. When aircraft taxi, engine emissions are emitted to the airports and surrounding areas where people live and work. Aircraft emissions have significant effects on the atmosphere [2], human health [3], and mortality [4]. Longer taxi times may contribute to more engine emissions, fuel burn, and airport congestion.

The International Civil Aviation Organization (ICAO) proposed the landing and take-off (LTO) cycle to estimate engine emissions for the engine certification process in the ICAO Environmental Report [5]. The LTO cycle consists of four modes—approach, taxi/ground idle, take-off, and climb—with a specific time-in-mode (TIM) for each mode of the cycle. These TIMs and specified thrust settings have been extensively used as references in emissions estimations and engine certification. In the ICAO Environmental Report [5], the TIM for the taxi/ground idle mode is 26 min.

Table 1. ICAO: Four modes of LTO cycle and their respective times-in-mode (TIMs) [6].

Modes	Time (in Minutes)
Approach	4.0
Taxi/Ground Idle	26.0
Take-off	0.7
Climb	2.2

shows the four modes and the time in each mode. The ICAO [6] (pg. I-1-2) defined taxi/ground idle as “the operating phases involving taxi and idle between the initial starting of the propulsion engine(s) and the initial of the take-off roll and between the time of runway turn-off and final shutdown of all propulsion engine(s)”.

This paper aims to compare the actual reported taxi times at airports (and hub classifications) with the ICAO-specified 26 min. The FAA Airline System Performance Metrics (ASPM) dataset [7] and the FAA National Plan of Integrated Airport Systems (NPIAS) dataset [8] are selected in this study for the retrieval of airport operations data and U.S. commercial airport hub classifications. According to the ASPM dataset [9] (para. 40–42), “taxi in time is the time difference between the Wheels On time and Gate In time in minutes and taxi out time is the time difference between the Wheels Off time and Gate Out time in minutes”. In this study, average quarter-hour taxi-in times and taxi-out times are collected from the ASPM dataset for analysis. The FAA publishes NPIAS reports [8] for administrative and funding purposes. In the NPIAS report, airports are categorized as non-hub, small-hub, medium-hub, and large-hub airports [8]. This paper compares the sum of the average quarter-hour taxi-in time and average quarter-hour taxi-out time to the TIM of the taxi/ground idle mode (26 min) from the ICAO LTO cycle, including the potential effects of hub classifications and individual airports. The researchers use both parametric and non-parametric tests to conduct data analysis.

The LTO cycle is used as a model for engine emissions estimation and certification. This paper aims to investigate if the TIM of the taxi/ground idle mode from the LTO cycle reflects real-world reported airport taxi times. Exploring the relationships between the LTO cycle and reported taxi times may contribute to reducing taxi times and engine emissions and increasing airport capacity. This paper may also provide insights to researchers, aircraft manufacturers, and aviation regulators who may use the ICAO LTO cycle for emissions estimations.

2. Background

Taxi time affects airport capacity, fuel burn, and aircraft engine emissions. Different approaches have been used to explore potential ways to predict and reduce aircraft taxi times. Previous research considered the number of flights [10,11,12,13,14,15], weather conditions [12,13,15], runway configurations [11,12,15,16], taxi distance [14,15,16,17,18], taxi speed [14,19], number of turns [19], aircraft weight [18,19], number of hot spots [20,21], airport hub classifications [20,21], and presence of specialized airport equipment [22] to explore the relationships of these factors with taxi times using statistical methods [11,13,14,15,17,20,21,22], machine learning [10,18,19], and simulation models [12,16,23,24].

Aircraft engines run during taxi operations. The longer the taxi times are, the more emissions are produced. The ICAO [25] recommends two approaches to calculating aircraft engine emissions for the LTO cycle. One of the approaches includes a methodology to calculate greenhouse gases or emissions that have significant impacts on human health, such as nitrogen oxides, carbon monoxide, high hydrocarbon, and non-volatile particulate matter [25]. The equation for the calculation is given as Equation (1) [25] (p. 3-A1-13):

$${Ei}_j=\sum \left({TIM}_{jk}\times 60\right)\times \left({FF}_{jk}\right)\times \left({Ei}_{ijk}\right)\times \left({Ne}_j\right),$$

(1)

where “Ei_j stands for total emissions of pollutant i produced by aircraft j for one LTO cycle; FF_jk stands for fuel flow in each mode k for each engine on aircraft type j; Ei_ijk stands for emission index for each pollutant i, in mode k, for each engine of aircraft j; Ne_j stands for number of engines used on aircraft j; and TIM_jk stands for time-in-mode for different modes of the LTO cycle (approach, taxi/idle, climb-out and take-off)” [25] (p. 3-A1-13). This methodology has been widely used in previous research to estimate aircraft engine emissions [26,27,28,29,30,31,32,33,34,35,36,37]. When estimating taxi emissions, some research used the 26 min of TIMs from the LTO cycle [26,27,28,29,30,31,32,33]. Other research used real operation data such as flight trajectory data [34], flight data recorder (FDR) data [35], self-observed data from airlines [36], and a third-party database [37] to estimate emissions instead of using ICAO TIMs.

The engine certification process was developed in the late 1960s, and the TIMs of the LTO cycle were established based on aircraft performance during that time. The TIMs have not changed since then. Researchers argued that the TIMs from the LTO cycle do not represent the real times of current operations at some airports. Using Automated Radar Terminal System (ARTS) data from six major airports in the U.S., Rice [38] suggested revising the TIM for the climb mode from 2.2 min to 1.1 min to improve estimation accuracy. Unique [39] collected operation data from nine aircraft/engine combinations at Zurich Airport to compare the TIM with the ICAO LTO cycle. The average idle mode at Zurich airport is 43% less (14.8 min) than the 26 min from the LTO cycle [39]. Patterson et al. [35] used a total of 2824 records from the FDR to compare the TIMs, fuel flow rates, and LTO fuel consumption in the take-off and approach modes with the values from the ICAO LTO cycle. The results suggest that the TIMs of the take-off and approach modes from the LTO cycle are significantly different from the actual time of take-off and approach for most of the aircraft/engine combinations [35]. The researchers of another paper [36] (p. 3502) reported that the 26 min TIM from the LTO cycle is “far longer than the actual value observed” at Brisbane International Airport. However, previous research focused on comparisons of TIMs with real operation data at a single U.S. airport [38], European airports [39], or Australian airports [36], or used mixed data from airports globally [35]. There is no identified research specifically comparing the TIM of idle mode with real operation data at U.S. hub airports.

The research in this study builds upon the previous taxi time exploratory studies [20,21,22]. In Gupta et al. [20], the researchers studied the effect of airport hub classification and number of airport hotspots on taxi-in times and taxi-out times; in Wang et al. [21], the researchers compared taxi-in times with taxi-out times across U.S. hub airports and number of hotspots, and in Wang et al. [22], the researchers studied the effect of Airport Surface Detection Equipment, Model X (ASDE-X) on taxi-in and taxi-out times. The purpose of this study is to use average quarter-hour taxi time data from the ASPM dataset and compare the total taxi times at 71 U.S. hub airports with the TIM of taxi/ground idle mode (26 min) to examine if the 26 minutes represent the reported taxi times at small-, medium-, and large-hub U.S. commercial airports.

3. Materials and Methods

This study compares the sum of average quarter-hour taxi-in times and average quarter-hour taxi-out times with the idle mode time of the ICAO LTO cycle. The paper explores two research questions.

Research Question 1:

Does the TIM of the taxi/idle mode of the ICAO LTO cycle (26 min) represent the real taxi times at each of the 71 sampled U.S. hub airports (treating each airport individually)?

Research Question 2(a):

Does the TIM of the taxi/idle mode of the ICAO LTO cycle (26 min) represent the real taxi times at U.S. hub airports (treating the 71 airports as one group)?

Research Question 2(b):

Does the TIM of the taxi/idle mode of the ICAO LTO cycle (26 min) represent the real taxi times at small-, medium-, and large-hub airports (segregating the 71 airports by NPIAS hub)?

The FAA collects and publishes data of the average quarter-hour taxi times of 77 U.S. airports in the ASPM dataset [7]. The FAA NPIAS report [8] classifies the U.S. hub airports into 4 categories: non-hub, small-hub, medium-hub, and large-hub airports. Among the 77 airports with available taxi time data in the ASPM, 71 of them are classified as hub airports by the FAA NPIAS report [8]. This paper uses the average quarter-hour taxi time data from these 71 airports to conduct data analysis. Among the 71 sampled airports in this study, 11 are small-hub airports, 30 are medium-hub airports, and 30 are large-hub airports.

The researchers collected the average quarter-hour taxi-in time and average quarter-hour taxi-out time of the 71 airports from 1 January 2023 to 31 December 2023, from 6:00 a.m. to 10:00 p.m. In the collected data, the researchers identified average quarter-hour taxi times as zeros. A zero in the average quarter-hour taxi time data means that there were no departure or arrival operations in that specific quarter-hour period. In this paper, all quarter-hours that contain zeros were removed from the collected sample, as taxi times could not be zero. Therefore, this study explores the research questions based on the assumption that there was at least one departure operation and at least one arrival operation in each quarter-hour.

The researchers identified a large number of outliers in the distributions of the taxi times, as well as violation of the homogeneity and normality assumptions for t-tests. However, the large sample size in this study may result in improved statistical power with t-tests. The researchers used both t-tests and 1-sample Wilcoxon non-parametric tests to compare the mean and median taxi times in this paper.

Table 2. Research questions and hypotheses.

Research Questions		Hypotheses
		Parametric	Non-Parametric
RQ1		Ho: µ(taxi-in+taxi-out) = 26 min Ha: µ(taxi-in+taxi-out) ≠ 26 min for each of the 71 airports	Ho: η(taxi-in+taxi-out) = 26 min Ha: η(taxi-in+taxi-out) ≠ 26 min for each of the 71 airports
RQ2	(a)	Ho: µ(taxi-in+taxi-out) = 26 min Ha: µ(taxi-in+taxi-out) ≠ 26 min for a combined sample of 71 airports	Ho: η(taxi-in+taxi-out) = 26 min Ha: η(taxi-in+taxi-out) ≠ 26 min for a combined sample of 71 airports
	(b)	Ho: µ(taxi-in+taxi-out) = 26 min Ha: µ(taxi-in+taxi-out) ≠ 26 min for airports grouped by S/M/L hubs	Ho: µ(taxi-in+taxi-out) = 26 min Ha: µ(taxi-in+taxi-out) ≠ 26 min for airports grouped by S/M/L hubs

shows the hypotheses of this study.

To answer the research questions, the sum of the average taxi-in time and the average taxi-out time in each quarter is used as a new variable, average quarter-hour total taxi time [T_total], for comparison with the TIM of the idle mode of the ICAO LTO cycle (26 min). Equation (2) shows the calculation for T_total. The researchers compare the means of the average quarter-hour total taxi time with the TIM of the taxi/idle mode of ICAO LTO (26 min) using the t-statistic, as calculated in (3) [40].

$$T_{total}=T_{in}+T_{out}$$

(2)

$$t={\displaystyle \frac{\overline{X_{T_{in}+T_{out}}}-{\mu }_o}{S_{T_{in}+T_{out}}/\sqrt{n}}}={\displaystyle \frac{\overline{X_{T_{total}}}-{\mu }_o}{S_{T_{total}}/\sqrt{n}}}$$

(3)

According to Minitab® Version 21.4.2 [41], the 1-sample Wilcoxon non-parametric test uses “the number of pairwise averages (also called Walsh averages) that are larger than the hypothesized median, plus one half the number of pairwise averages that are equal to the hypothesized median” as Wilcoxon statistics (W). Equation (4) is used for the calculation of Z values for the 1-sample Wilcoxon non-parametric test in this paper [41,42].

$$Z={\displaystyle \frac{W-\frac{n(n+1)}{4}}{\sqrt{\frac{n(n+1)(2n+1)}{24}}}}$$

(4)

To answer Research Question 1, the samples of T_total are segregated into 71 groups by airport. The researchers compare T_total with the TIM of the idle model individually for each of the 71 airports. To answer Research Question 2a, a combined sample of the T_total of 71 airports is used to compare the mean taxi time with the TIM of the idle mode of the LTO cycle. The airports are not segregated by hub classification in RQ2a. To answer Research Question 2b, the samples of T_total are segregated into three groups based on the NPIAS hub classifications of the airports (S/M/L). The researchers individually compare the T_total of each hub classification (S/M/L) with the TIM of the idle mode of the LTO cycle.

Table 3. A sample of the consolidated data.

Data Source	ASPM Dataset 2							NPIAS 2023–2027 3	Average Quarter-Hour Total Taxi Time 4 (minutes) TTotal
Facility 1	Date	Hour of the Day (6:00 a.m.–10:00 p.m.)	Quarter of the Hour	Departures for Metric Computation	Average Taxi-Out Time (minutes) TOut	Arrivals for Metric Computation	Average Taxi-In Time (minutes) TIn	NPIAS Hub Classification
ABQ	1/1/2023	6	2	1	14	1	5	M	19.00
ABQ	1/1/2023	10	3	3	7.67	2	5.5	M	13.17
ABQ	1/1/2023	10	4	2	13.5	1	3	M	16.50
CLT	10/12/2023	10	3	3	18.67	10	8.9	L	27.57
CLT	10/12/2023	10	4	2	20	6	9.33	L	29.33
CLT	10/12/2023	11	1	31	23.42	3	5	L	28.42
LGB	1/4/2023	7	2	1	14	1	3	S	17.00
LGB	1/4/2023	7	4	2	15.5	1	3	S	18.50
LGB	1/4/2023	8	4	3	12	1	3	S	15.00

¹ Abbreviations in the Facility column correspond to the 3-letter airport code or identifier assigned by the International Air Transport Association (IATA) for each of the 71 airports in the study. For instance, ABQ refers to the Albuquerque International Sunport Airport; CLT refers to the Charlotte Douglas International Airport; and LBG refers to the Long Beach Municipal Airport Daugherty Field. ² FAA Aviation System Performance Metrics Dataset [7]. ³ FAA 2023–2027 National Plan of Integrated Airport Systems (NPIAS) report [8]. ⁴ The data in this table has similar headings as previous research studies [20,21,22].

shows a sample of the consolidated data for this study.

4. Results

This section presents the results of data analysis. Minitab® Version 21.4.2 was used to conduct all statistical analyses using an alpha level of 0.05.

4.1. Research Question 1

To answer RQ1, the collected samples were segregated into the 71 airports, and the mean and median total taxi time was compared to 26 min, individually for each sampled airport. Using the one-sample t-test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population mean of the average quarter-hour total taxi time at each of the 71 airports is equal to 26 min, except for Detroit Metropolitan Wayne County Airport (DTW). Using the one-sample Wilcoxon Signed Rank Test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population median of the average quarter-hour total taxi time at each of the 71 airports is equal to 26 min.

Table 4. Consolidated results for Research Question 1.

Descriptive Statistics (in Minutes)								Inferential Statistics (in Minutes)
${\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}={\mathit{T}}_{\mathit{i}\mathit{n}}+{\mathit{T}}_{\mathit{o}\mathit{u}\mathit{t}}$ ${\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}=\mathbf{Average}\mathbf{Quarter}-\mathbf{Hour}\mathbf{Total}\mathbf{Taxi}\mathbf{Time}$ ${\mathit{T}}_{\mathit{i}\mathit{n}}=\mathbf{Average}\mathbf{Quarter}-\mathbf{Hour}\mathbf{Taxi}-\mathbf{In}\mathbf{Time}$ ${\mathit{T}}_{\mathit{o}\mathit{u}\mathit{t}}$ = Average Quarter-Hour Taxi-Out Time								$1-\mathbf{Sample}\mathit{t}-\mathbf{Test}$ ${\mathbf{H}}_{\mathbf{o}}:{\mathit{\mu }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}=26\mathbf{min}$ ${\mathbf{H}}_{\mathbf{a}}:{\mathit{\mu }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}$ ≠ 26 min		$1-\mathbf{Sample}\mathbf{Wilcoxon}\mathbf{Signed}\mathbf{Rank}\mathbf{Test}$ ${\mathbf{H}}_{\mathbf{o}}:{\mathit{\eta }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}=26\mathbf{min}$ ${\mathbf{H}}_{\mathbf{a}}:{\mathit{\eta }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}$ ≠ 26 min
Facility 1	Hub	N 2	Min 3	Max 3	Mean 3	Median 3	Std. Dev. 3	95% Confidence Interval for Mean	p-Value t-Test	95% Confidence Achieved for Median	p-Value Wilcoxon
BHM	S	5132	6	130	18.54	17	6.76	(18.36, 18.73)	<0.001	(17, 17)	<0.001
BUF	S	5842	8	168	20.02	17.5	9.5	(19.78, 20.27)	<0.001	(17, 17.5)	<0.001
DAY	S	1870	8	117	21.47	19	8.06	(21.1, 21.83)	<0.001	(19, 19.5)	<0.001
HPN	S	8717	6	184	20.98	19	8.13	(20.81, 21.15)	<0.001	(19, 19)	<0.001
ISP	S	894	9	116.5	16.85	16	6.76	(16.4, 17.29)	<0.001	(15, 16)	<0.001
LGB	S	5137	4	98	16.34	15	5.53	(16.19, 16.49)	<0.001	(15, 15)	<0.001
MHT	S	1773	6	90	19.33	17.5	7.45	(18.99, 19.68)	<0.001	(17, 18)	<0.001
PSP	S	4865	10	86	21.6	20	6.89	(21.41, 21.8)	<0.001	(19.83, 20)	<0.001
PVD	S	4391	8	120	19.19	17	8.04	(18.95, 19.42)	<0.001	(17, 17)	<0.001
SDF	S	10,069	7	154	21.12	19.75	7.14	(20.98, 21.26)	<0.001	(19.64, 20)	<0.001
TUS	S	4727	9	498	18.7	17.5	9.47	(18.43, 18.97)	<0.001	(17, 17.5)	<0.001
ABQ	M	7403	8	163	18.99	17.67	6.2	(18.84, 19.13)	<0.001	(17.5, 18)	<0.001
ANC	M	17,206	5	128	21.29	19.5	7.09	(21.19, 21.4)	<0.001	(19.47, 19.67)	<0.001
BDL	M	7316	8	156.5	20.48	18.67	8.09	(20.3, 20.67)	<0.001	(18.5, 19)	<0.001
BUR	M	10,176	7.5	103.33	19	17.5	6.63	(18.87, 19.13)	<0.001	(17.5, 17.75)	<0.001
CLE	M	10,752	10	117	22.52	20.5	7.99	(22.37, 22.67)	<0.001	(20.33, 20.5)	<0.001
CVG	M	14,650	11	180	21.62	20	6.75	(21.51, 21.73)	<0.001	(20, 20.2)	<0.001
DAL	M	15,119	7.63	157	17.4	16.11	5.95	(17.31, 17.5)	<0.001	(16, 16.17)	<0.001
HNL	M	16,733	10.75	96	24.69	24	5.2	(24.61, 24.77)	<0.001	(23.95, 24)	<0.001
HOU	M	13,048	8.25	222	17.36	16.5	5.21	(17.27, 17.45)	<0.001	(16.33, 16.5)	<0.001
IND	M	12,150	9.75	146	21.5	19.92	7.04	(21.37, 21.62)	<0.001	(19.75, 20)	<0.001
JAX	M	9067	8	179.5	19.5	18	6.42	(19.37, 19.63)	<0.001	(18, 18.17)	<0.001
MCI	M	11,342	5	158	18.65	17.2	6.52	(18.53, 18.77)	<0.001	(17, 17.33)	<0.001
MEM	M	9728	5.25	270.6	22.98	21.5	9.7	(22.78, 23.17)	<0.001	(21.42, 21.62)	<0.001
MKE	M	8402	7	133	19.72	17.5	8.66	(19.54, 19.91)	<0.001	(17.33, 17.5)	<0.001
MSY	M	12,461	5.25	128.33	17.78	16.8	5.2	(17.69, 17.87)	<0.001	(16.67, 17)	<0.001
OAK	M	13,367	10	105	18.91	18	4.87	(18.83, 18.99)	<0.001	(17.83, 18)	<0.001
OGG	M	10,027	6	145	19.08	16.83	8.91	(18.9, 19.25)	<0.001	(16.5, 17)	<0.001
OMA	M	7165	7.67	136	18.33	16.5	7.68	(18.15, 18.51)	<0.001	(16.13, 16.5)	<0.001
ONT	M	9453	6	103	18.74	17.75	5.62	(18.63, 18.86)	<0.001	(17.6, 18)	<0.001
PBI	M	10,759	9.33	150.5	20.82	19.2	6.87	(20.69, 20.95)	<0.001	(19, 19.33)	<0.001
PDX	M	15,307	6.5	156.5	18.52	17.66	5.11	(18.44, 18.6)	<0.001	(17.5, 17.67)	<0.001
PIT	M	11,536	8.33	100.43	19.92	18	7.18	(19.79, 20.05)	<0.001	(18, 18)	<0.001
RDU	M	13,521	7	150	22.26	21	7.14	(22.14, 22.38)	<0.001	(20.8, 21)	<0.001
RSW	M	9463	10	129	19.47	17.7	7.47	(19.32, 19.62)	<0.001	(17.67, 17.85)	<0.001
SAT	M	10,820	7	161	17.6	16.57	5.6	(17.5, 17.71)	<0.001	(16.5, 16.67)	<0.001
SJC	M	12,838	6	90.63	17.04	16.25	3.93	(16.97, 17.11)	<0.001	(16.2, 16.33)	<0.001
SJU	M	12,957	5	121	19.29	18.95	5.09	(19.21, 19.38)	<0.001	(18.75, 19)	<0.001
SMF	M	12,765	8	100.8	17.5	16.67	4.54	(17.43, 17.58)	<0.001	(16.67, 16.75)	<0.001
SNA	M	11,962	7.4	99.67	20.28	18.8	6.63	(20.16, 20.39)	<0.001	(18.67, 19)	<0.001
STL	M	13,880	7.33	141	18.39	17.5	5.19	(18.3, 18.48)	<0.001	(17.47, 17.5)	<0.001
ATL	L	18,747	13	108.78	25.1	24.17	5.48	(25.02, 25.18)	<0.001	(24.08, 24.23)	<0.001
AUS	L	15,304	6	161	22.45	21.18	6.38	(22.35, 22.55)	<0.001	(21.1, 21.27)	<0.001
BNA	L	15,883	10.75	144	22.6	21.61	5.8	(22.51, 22.69)	<0.001	(21.5, 21.69)	<0.001
BOS	L	18,036	11.83	127.57	28.49	26.4	8.94	(28.35, 28.62)	<0.001	(26.29, 26.5)	<0.001
BWI	L	16,015	9.67	157.5	20.07	18.79	5.94	(19.98, 20.16)	<0.001	(18.71, 18.86)	<0.001
CLT	L	15,694	8	115	31.44	30.66	7.44	(31.33, 31.56)	<0.001	(30.54, 30.76)	<0.001
DCA	L	14,715	9.5	116.14	27.24	25.77	8.27	(27.1, 27.37)	<0.001	(25.62, 25.9)	<0.001
DEN	L	18,093	7.2	132.14	26.81	24.31	9.05	(26.67, 26.94)	<0.001	(24.21, 24.4)	<0.001
DFW	L	17,735	13	184.33	30.7	29.28	8.95	(30.57, 30.83)	<0.001	(29.18, 29.38)	<0.001
DTW	L	15,814	11	121.4	25.88	23.9	8.1	(25.76, 26.01)	0.071	(23.83, 24)	<0.001
EWR	L	18,704	8	238	34.98	32.5	12.59	(34.8, 35.16)	<0.001	(32.36, 32.67)	<0.001
FLL	L	17,468	10.4	163.29	25.19	23.17	8.52	(25.06, 25.31)	<0.001	(23.09, 23.25)	<0.001
IAD	L	14,087	11.5	138	24.27	23	6.91	(24.16, 24.38)	<0.001	(22.97, 23)	<0.001
IAH	L	16,363	12	151.58	26.78	25.18	7.73	(26.66, 26.9)	<0.001	(25.06, 25.31)	<0.005
JFK	L	20,038	17.78	277	37.59	34.6	12.66	(37.41, 37.76)	<0.001	(34.47, 34.75)	<0.001
LAS	L	18,280	12.67	96.71	25.72	24.25	6.64	(25.63, 25.82)	<0.001	(24.16, 24.33)	<0.001
LAX	L	20,481	11	80.38	27.81	26.99	5.37	(27.74, 27.89)	<0.001	(26.89, 27.03)	<0.001
LGA	L	15,019	13	249.7	31.32	28.83	10.95	(31.15, 31.5)	<0.001	(28.69, 28.97)	<0.001
MCO	L	17,776	12	152	29.8	27.86	8.78	(29.67, 29.93)	<0.001	(27.77, 27.95)	<0.001
MDW	L	14,775	8	180.33	19.84	18	7.24	(19.72, 19.95)	<0.001	(18, 18.14)	<0.001
MIA	L	20,054	10	132.9	28.64	26.54	8.77	(28.52, 28.76)	<0.001	(26.47, 26.67)	<0.001
MSP	L	16,614	10	176.5	24.81	21.8	10.03	(24.66, 24.96)	<0.001	(21.7, 21.88)	<0.001
ORD	L	19,044	10.5	149	33.68	32.11	9.85	(33.54, 33.82)	<0.001	(32, 32.26)	<0.001
PHL	L	15,905	4.5	155.86	25.17	23.43	8.84	(25.03, 25.31)	<0.001	(23.33, 23.5)	<0.001
PHX	L	17,700	8	74	21.53	20.7	4.7	(21.47, 21.6)	<0.001	(20.63, 20.77)	<0.001
SAN	L	15,048	10.5	92.94	22.98	21.5	7.26	(22.87, 23.1)	<0.001	(21.4, 21.6)	<0.001
SEA	L	18,771	12	203.67	29.48	28.42	7.54	(29.38, 29.59)	<0.001	(28.33, 28.5)	<0.001
SFO	L	18,171	10	129.03	26.86	25.75	6.64	(26.77, 26.96)	<0.001	(25.67, 25.89)	<0.001
SLC	L	15,568	9.33	128	25.22	23	8.44	(25.08, 25.35)	<0.001	(22.9, 23.05)	<0.001
TPA	L	15,989	8.5	105	19.91	18.7	5.9	(19.82, 20)	<0.001	(18.66, 18.75)	<0.001

¹ Abbreviations in the Facility column correspond to the 3-letter IATA airport code or identifier for each of the 71 airports in the study. ² N refers to the number of taxi time observations for each airport. ³ Min, Max, Mean, Median, and Std. Dev. refer to the minimum, maximum, mean, median, and standard deviation values of the average quarter-hour total taxi time T_total, in minutes.

shows the consolidated results for RQ1. Using the collected samples, the researchers concluded that the TIM of the idle mode (26 min) from the LTO cycle does not represent the real reported average quarter-hour total taxi time at each of the 71 sampled airports, except for DTW.

4.2. Research Question 2a

To answer RQ 2a, a combined sample (ignoring hub classifications) is used to compare T_total with the TIM of the idle mode from the LTO cycle. The data consolidation resulted in a sample size of 916,681. The T_total mean was found to be 23.78 min with a 95% confidence interval (CI) of (23.76, 23.78) minutes. The T_total median was found to be 22 min.

Using the one-sample t-test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population mean of the average quarter-hour total taxi time equals 26 min. Using the one-sample Wilcoxon Signed Rank Test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population median of the average quarter-hour total taxi time is equal to 26 min.

Table 5. Consolidated results for Research Questions 2a and 2b.

	Descriptive Statistics (in Minutes)							Inferential Statistics (in Minutes)
${\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}={\mathit{T}}_{\mathit{i}\mathit{n}}+{\mathit{T}}_{\mathit{o}\mathit{u}\mathit{t}}$ ${\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}=\mathbf{Average}\mathbf{Quarter}-\mathbf{Hour}\mathbf{Total}\mathbf{Taxi}\mathbf{Time}$ ${\mathit{T}}_{\mathit{i}\mathit{n}}=\mathbf{Average}\mathbf{Quarter}-\mathbf{Hour}\mathbf{Taxi}-\mathbf{In}\mathbf{Time}$ ${\mathit{T}}_{\mathit{o}\mathit{u}\mathit{t}}$ = Average Quarter-Hour Taxi-Out Time								$1-\mathbf{Sample}\mathit{t}-\mathbf{Test}$ ${\mathbf{H}}_{\mathbf{o}}:{\mathit{\mu }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}=26\mathbf{min}$ ${\mathbf{H}}_{\mathbf{a}}:{\mathit{\mu }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}$ ≠ 26 min		$1-\mathbf{Sample}\mathbf{Wilcoxon}\mathbf{Signed}\mathbf{Rank}\mathbf{Test}$ ${\mathbf{H}}_{\mathbf{o}}:{\mathit{\eta }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}=26\mathbf{min}$ ${\mathbf{H}}_{\mathbf{a}}:{\mathit{\eta }}_{{\mathit{T}}_{\mathit{t}\mathit{o}\mathit{t}\mathit{a}\mathit{l}}}$ ≠ 26 min
	Hub	N 1	Min 2	Max 2	Mean 2	Median 2	Std. Dev. 2	95% Confidence Interval for Mean	p-Value t-Test	95% Confidence Achieved for Median	p-Value Wilcoxon
RQ 2a	-	916,681	4.00	498.00	23.78	22.00	9.10	(23.76, 23.80)	<0.001	(22.00, 22.00)	<0.001
RQ 2b	S	53,417	4.00	498.00	19.82	18.00	7.91	(19.75, 19.89)	<0.001	(18.00, 18.00)	<0.001
	M	351,373	5.00	270.60	19.72	18.25	6.76	(19.69, 19.74)	<0.001	(18.25, 18.29)	<0.001
	L	511,891	4.50	277.00	26.98	25.08	9.34	(26.95, 27.00)	<0.001	(25.05, 25.10)	<0.001

¹ N refers to the number of taxi time observations for respective samples. ² Min, Max, Mean, Median, and Std. Dev. refer to the minimum, maximum, mean, median, and standard deviation values of the average quarter-hour total taxi time T_total, in minutes.

shows the consolidated results for RQ 2a. Using the collected samples, the researchers concluded that the TIM of the idle mode (26 min) from the LTO cycle does not represent real reported average quarter-hour total taxi times.

4.3. Research Question 2b

To answer RQ 2b, the collected samples were segregated into three groups by airport hub classification (S/M/L). The mean and median total taxi times of each group were compared to 26 min individually. The 95% confidence intervals for the mean of T_total at small-, medium-, and large-hub airports are (19.89, 19.75), (19.69, 19.74), and (26.95, 27.00).

Using the one-sample t-test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population means of the average quarter-hour total taxi time at small-, medium-, and large-hub airports are equal to 26 min. Using the one-sample Wilcoxon Signed Rank Test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population medians of the average quarter-hour total taxi time at small-, medium-, and large-hub airports are equal to 26 min. Table 5 shows the consolidated results for RQ 2b. Using the collected samples, the researchers concluded that the TIM of the idle mode (26 min) from the LTO cycle does not represent real reported average quarter-hour total taxi times at small-, medium-, and large-hub airports.

5. Discussion and Conclusions

This paper collected average quarter-hour taxi time data from the FAA ASPM dataset and compared them to the TIM of the taxi/idle mode of the ICAO LTO cycle (26 min). The researchers compared average quarter-hour total taxi times with 26 min (1) across each of the 71 airports, individually, (2) using a combined sample of data from 71 airports, and (3) by segregating the 71 airports using NPIAS hub classifications. This paper is affected by the following limitations and delimitations:

• The FAA ASPM dataset publishes the airport operations data of 77 U.S. commercial airports. Among these 77 airports, 71 airports are categorized as hub airports (small, medium, large). In this study, the average quarter-hour taxi times for the 71 U.S. hub airports were collected and analyzed;
• The quarter-hour taxi time observations that contained a zero for either taxi-in time or taxi-out time were removed from the collected data sample. A zero taxi-in or taxi-out time indicates no landing or take-off operations, respectively. Therefore, in this study, for a quarter-hour taxi time observation to be considered for data analysis, at least one departure and one arrival operation must have occurred in that quarter-hour;
• According to the FAA ASPM dataset [9], taxi-in time is the “time difference between the Wheels On time and Gate In time, in minutes” (para. 40); taxi-out time is the “time differences between the Wheel Off time and Gate Out time, in minutes” (para. 42). In ICAO [6] SARPs Annex 16, taxi/ground idle for a standard LTO cycle is defined as “the operating phases involving taxi and idle between the initial starting of the propulsion engine(s) and the initial of the take-off roll and between the time of runway turn-off and final shutdown of all propulsion engine(s)” (pg. I-1-2). The potential discrepancies in the definitions of taxi times and idle mode may affect the analysis and results of the study. The FAA ASPM definitions do not consider the time when an aircraft stays at the gate with the engine(s) on, which is included in the ICAO definitions, whereas the ICAO definitions do not include the time spent on the runway, which is included in the ASPM data. While the differences in these two definitions are significant, the ASPM dataset does not provide a means to bridge the gap between them; taxi time and/or idle time data from other reliable and credible source(s) may be needed to eliminate these discrepancies;
• The FAA ASPM dataset provides the average taxi-in time and average taxi-out time for each quarter-hour, and not the specific taxi-in time or taxi-out time for each operation. The samples collected may have a higher standard deviation than the taxi time data of individual flights;
• The collected samples for each of the research questions violated the normality and constant variance assumptions for t-tests. Despite these violations, the large sample sizes in each RQ dataset may ensure the robustness and reliability of the data analysis results due to the Central Limit Theorem (CLT);
• The TIMs of the ICAO standard LTO cycle are designed for the estimation of emissions for turbine engine models. The quarter-hour taxi times collected from the FAA ASPM dataset may include taxi times for aircraft equipped with other types of engines. Type-of-engine data are not available from the ASPM dataset. The lack of engine-type data may reduce the effectiveness of the statistical analysis;
• Other factors that may cause variations in taxi time such as aircraft taxi speed, airport layout, and weather conditions are not considered in this study,

This study may provide insights to airport managers and regulators regarding congestion mitigation and capacity promotion. Even though the ICAO proposed TIMs in the standard LTO cycle for emissions estimation purposes, the 26 min idle mode has been widely used in research that focuses on airport congestion, taxi optimization, and airport planning. This study may be useful to researchers focusing on taxi time estimations and predictions, airport efficiency, and airport planning in terms of choosing the appropriate data source of ground idle time.

In this study, the average quarter-hour taxi time data of 71 U.S. commercial airports were collected, consolidated, and compared to the TIM (26 min) of the taxi/idle mode of the ICAO standard LTO cycle. In RQ 1, the total taxi times at each of the 71 airports in this study were compared to 26 min using parametric (one-sample t-test) and non-parametric (one-sample Wilcoxon Signed Rank Test) statistical tests. The researchers rejected the null hypothesis that the estimated population means of the average quarter-hour total taxi time are equal to 26 min for 70 of the 71 sampled airports. The researchers failed to reject the null hypothesis for Detroit Metropolitan Wayne County Airport (DTW). The researchers rejected the null hypothesis (p < 0.001) that the estimated population medians of the average quarter-hour total taxi time are equal to 26 min for each of the 71 sampled airports. It is important to note that the average quarter-hour total taxi time was greater than 26 min for 15 of the 71 sampled airports, with each of these 15 airports being large hubs.

In RQ 2a, the researchers used a combined sample of all 71 airports in the study to compare the mean and median total taxi times to 26 min. For the combined sample, the mean total taxi time was found to be 23.78 min, and the median total taxi time was found to be 22 min. Using the one-sample t-test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population mean of the average quarter-hour total taxi time equals 26 min. Using the one-sample Wilcoxon Signed Rank Test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population median of the average quarter-hour total taxi time is equal to 26 min.

In RQ 2b, the T_total data were segregated into three groups (S/M/L) based on the NPIAS hub classifications of airports. The researchers compared the total taxi times of each of the three groups to 26 min individually. The following mean and median total taxi times, respectively, were found for each hub classification: small (19.82; 18), medium (19.72; 18.25), and large (26.98; 25.08). Using the one-sample t-test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population means of the average quarter-hour total taxi time at small-, medium-, and large-hub airports are equal to 26 min. Using the one-sample Wilcoxon Signed Rank Test, the researchers rejected the null hypothesis (p < 0.001) that the estimated population medians of the average quarter-hour total taxi time at small-, medium-, and large-hub airports are equal to 26 min. While the 95% CI for mean total taxi times was less than 26 min for small (19.75, 19.89)- and medium (19.69, 19.74)-hub airports, it was more than 26 min for large (26.95, 27.00)-hub airports. These differences may be attributed to factors such as the airport size, runway length and pattern, taxi distance, presence of end-around taxiways, traffic volume and flow patterns, aircraft types, and operational management.

The results of this study indicate that the ICAO standard taxi/idle time of 26 min may not accurately represent reported taxi times at U.S. commercial airports (as shown in RQ1) and that actual taxi times vary significantly across airports. In addition, the results of this study indicate that while the mean and median total taxi times at large-hub airports may be closer to 26 min, these numbers are under 20 min for small- and medium-hub airports, being significantly different from the ICAO’s TIM of 26 min for the taxi/idle mode. While using airport-specific taxi times for modeling would enhance accuracy, it may not be a practical solution due to variability and implementation challenges. Using a universal standard simplifies these challenges, and if a single standardized value is necessary, the findings suggest that a taxi/idle time closer to 22 min (as shown in RQ2a) may provide a more real-world representative estimate. Alternatively, incorporating hub classifications and their respective average taxi times (as shown in RQ2b) may offer a compromise between airport-specific taxi times and a single standardized number. This practical middle-ground approach may acknowledge the operational differences among small-, medium-, and large-hub airports, providing a more refined estimation framework without requiring individual values for each airport. The overarching findings of this study highlight the opportunity for future assessments to consider adjustments to ICAO’s assumptions, perhaps representing a call for the reassessment and revision of the standard taxi/ground idle TIM, ensuring that accurate taxi/idle times that reflect real-world variations are used in future studies.