Empirical Assessment of the End-Around Taxiway's ...docs.trb.org/prp/13-5272.pdfMassidda A.,...
Transcript of Empirical Assessment of the End-Around Taxiway's ...docs.trb.org/prp/13-5272.pdfMassidda A.,...
Massidda A., Mattingly S. 1
TRANSPORTATION RESEARCH BOARD 1
92nd Annual Meeting – January 13-17, 2013 2
3
TITLE: Empirical Assessment of the End-Around Taxiway’s Operational Benefits at 4
Dallas/Fort Worth International Airport Using ASDE-X Data 5
6
PAPER NUMBER: 13-5272 7
8
AUTHORS: *Antonio Massidda, Faculty Research Associate 9
Department of Civil Engineering 10
University of Texas at Arlington 11
P.O. Box 19308, Arlington, TX 76019-0308 12
Phone: 214-208-0808 13
E-mail: [email protected] 14
15
Stephen P. Mattingly, Associate Professor 16
Department of Civil Engineering 17
University of Texas at Arlington 18
P.O. Box 19308, Arlington, TX 76019-0308 19
Phone: 817-272-2859 20
E-mail: [email protected] 21
* - corresponding author 22
23
KEY WORDS: End-Around Taxiway, Perimeter Taxiway, Airport, Runway Crossing, Runway 24
Safety, Throughput, Delay, Capacity, Taxi Time, Airfield Planning, Dallas/Fort Worth, DFW, 25
ASDE-X, ADS-B, Database 26
27
WORD COUNT: 5,050 words + 4 tables + 5 figures = 7,300 28
29
ABSTRACT 30
At the present time, only a few airfields in the world have an End-Around Taxiway (EAT). 31
Since December 2008, an EAT serves Dallas/Fort Worth International Airport’s (DFW) runways 32
17L, 17C, and 17R, with the purpose of reducing the number of runway crossings and therefore 33
improving safety and capacity. 34
This paper describes the results of the research project (funded by the FAA and 35
performed by the authors) to assess the safety impacts of DFW’s EAT in terms of reduction in 36
number of runway crossings. In addition, this paper empirically defines the enhancement in 37
departure and arrival throughput achieved after the construction of the EAT. These assessments 38
are based on data from DFW’s Surface Detection Equipment – Model X (ASDE-X) database. 39
This study has found that the EAT has improved runway safety, increased capacity and 40
reduced departure delay at DFW, although for several reasons its usage is essentially limited to 41
runway 17L arrivals. The EAT has eliminated on average 51% crossings on runway 17C daily 42
and over 83% percent of runway 17L arrivals use the EAT or cross runway 17C using low-risk 43
taxiways. In fact, the EAT has nearly eliminated all mid-runway 17C crossings due to 17L 44
arrivals. 45
The EAT has exceeded the expected enhancements of departure and arrival capacity. 46
Compared to pre-EAT operations, the ASDE-X data reveals that both departure and arrival 47
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 2
demand have increased at DFW. However, EAT operation has allowed the daily mean arrival 48
and departure maximum throughput rates to increase by 40% and 25%, respectively, while the 49
mean daily maximum departure delay has decreased by 38%. 50
51
1 INTRODUCTION 52
The End-Around Taxiway (EAT) is an innovative runway safety infrastructure concept that the 53
Federal Aviation Administration (FAA) has introduced in recent years at some of the busiest 54
airports in the United States. The primary EAT goal is to reduce the number of runway 55
crossings, thereby reducing the opportunities for runway incursions; as a result, on these 56
runways, which experience fewer crossings, additional expected operational benefits from an 57
EAT include (1) an increase in departure rate and therefore a reduction in departure queue delay. 58
At the present time, Hartsfield-Jackson Atlanta International Airport (ATL) and 59
Dallas/Fort Worth International Airport (DFW) have one EAT each; San Francisco International 60
Airport (SFO), Houston George Bush Intercontinental Airport (IAH), and Los Angeles 61
International Airport (LAX) have considered implementing EATs. In Europe, only Amsterdam 62
Schiphol (EHAM) and Frankfurt Main (EDDF) have an EAT (2). 63
The original plan for DFW’s EAT system included a total of four EATs, each located in 64
one of the airport quadrants (1). FAA performed an early study of this EAT system in 1998 (1); 65
five years later, in cooperation with DFW’s management and air traffic control (ATC), the FAA 66
conducted a human-in-the-loop simulation at the NASA Ames Research Center (ARC), which 67
included a control tower simulator and a full flight simulator (1). In this simulation, actual DFW 68
air traffic controllers and pilots interacted in a simulation of DFW operations without and with 69
the proposed EAT system (1). The results of this simulation were used as an operational 70
baseline for the subsequent EAT planning process. 71
The FAA Airport Obstruction Standards Committee (AOSC), in 2005, per DFW request 72
evaluated the compatibility of an EAT located on the Southeast quadrant with departure 73
operations. The U.S. Standard for Terminal Instrument Approach Procedures (TERPS), which 74
required protection of the 40:1 Obstacle Clearance Surface (OCS) from penetrations by the tails 75
of taxiing aircraft (3), provided the technical reference for this assessment. Given the slope of 76
the OCS and a distance of about 0.5 miles from the runways’ 17C and 17R thresholds to the 77
EAT (Figure 1), the AOSC observed that aircraft with tails up to 65 feet (Group V) would not 78
penetrate the departure surface (3). 79
80
81 82
FIGURE 1 Sketch of EAT compatibility with OCS. (17) 83
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 3
As a result, the AOSC approved unrestricted operations on the Southeast EAT, but 84
limited to South flow airport configuration (3). 85
In 2007, one of the authors estimated that the probability to penetrate the OCS for an 86
aircraft taxiing on the EAT was nearly zero, based on transponder height data analysis (4). 87
DFW and FAA started the construction of the Southeast EAT in 2006, and it opened in 88
December 2008; as shown in Figure 2, this EAT is essentially formed by extensions of the north-89
south taxiways “P”, “M”, “JS”, and by taxiway “ES” that links these extensions along their south 90
ends. This EAT serves the three parallel runways on the East side of the airport 35L/17R, 91
35C/17C, and 35R/17L. 92
93
94 95
FIGURE 2 DFW EAT, runways 35L/17R, 35C/17C, and 35R/17L. (5) 96
97
The current operating procedure for this EAT, issued by FAA DFW ATC in April 98
2011(6), states that during VFR conditions the EAT should be preferred for the arrivals leaving 99
runway 17L via taxiway “ER” (unless ATC approves an exception), and for runway 17C arrivals 100
only when the wait time to cross runway 17R is expected to be more than five minutes. 101
In recent years, the FAA installed Airport Surface Detection Equipment – Model X 102
(ASDE-X) at many of the nation’s busiest airports, including DFW (7). The primary purpose of 103
ASDE-X is to prevent runway incursions by enhancing the situational awareness of tower 104
controllers through the visualization of real time localization and identification of aircraft and 105
vehicles on an electronic map of the airport, which also includes airborne aircraft flying the final 106
approach segment. The ASDE-X data can be stored in a database for further processing and 107
analysis, and it can be enriched by merging data acquired by different sources, such as automatic 108
dependent surveillance broadcast (ADS-B), and radar regarding airborne operations beyond 109
ASDE-X’s radius of reception. As a result, this combined data provides valuable information 110
pertaining to aircraft operations; for example, the database integrates several characteristics of 111
individual flights, such as call sign, tail number, gates, specific surface movements (e.g. 112
runways, taxiways, and taxi times), and navigation fixes used. This study uses the DFW ASDE-113
X database created and maintained by NASA/FAA North Texas Research Station (NTX). The 114
specific ASDE-X data used are presented in the data analysis section. 115
The importance of research using ASDE-X data continues to increase; however, at this 116
time, there is limited runway and surface safety research based on this data. Typical research 117
using ASDE-X data address impacts associated with surface movement characteristics, new 118
approach and departure procedures, and surface management systems at a single airport. 119
Engelland (2) uses the DFW ASDE-X database to quantify EAT usage volume and to compare 120
the taxi times for arrivals using the EAT with those crossing the runways. Simaiakis et al. (8) 121
develop a pushback rate control strategy to reduce delay and maximize throughput for departures 122
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 4
based on an analysis of the Boston Logan International Airport ASDE-X database. Balakrishnan 123
et al. (9) develop a taxi route plan that optimizes surface operations using Integer Programming. 124
ASDE-X data is proving to be critically desirable for assessing the impacts of airport 125
infrastructure and operational policy changes. 126
Given the increasing congestion of the National Airspace System and at major airports, 127
the problems of airfield capacity and throughput have received significant attention from several 128
researchers and the FAA. Several approaches for estimating or calculating airfield capacity and 129
throughput have been proposed. In one such study, Hansen et al. (10), proposes a method for 130
empirically analyzing the runway throughput. This study uses this methodology for analyzing 131
the impacts of the EAT on arrival and departure throughput (more details are provided in section 132
3). While all previous studies on EAT impacts have been based on simulation results, this 133
research seeks to evaluate the actual EAT impacts on surface operations using DFW’s ASDE-X 134
data. 135
Section two describes the approach used for conducting a before and after evaluation of 136
surface operations at DFW. In section three, the ASDE-X data is used to analyze the EAT’s 137
impacts on runway crossings, departure and arrival throughput, and departure delay. The final 138
section summarizes the overall impacts of the EAT on safety and operations, and proposes 139
additional research involving the EAT and ASDE-X data. 140
141
2 Research Approach 142
This research used the ASDE-X database to compare observed runway crossings before/after the 143
EAT construction. In addition, the study assesses the impacts on demand, delay, and throughput 144
for runway 17R departures, and on arrival throughput for runways 17C and 17L. The study uses 145
the Surface Operations Data Analysis and Adaptation (SODAA) tool, developed by Mosaic 146
ATM, Inc. to access the database and export the information to custom spreadsheets developed 147
by the research team. The authors develop SODAA queries to retrieve taxi routes for runway 148
17L arrivals during South flow operating conditions. 149
Data sets are randomly selected from both the before and after periods based on required 150
minimum sample size estimates (see 11 for more information). The research team defines the 151
Before EAT time period from April 2008 to November 2008, which corresponds to the entire 152
pre-EAT operations portion of the ASDE-X database. To control for seasonal effects, the After 153
EAT period begins in April 2011 at the same time as the current EAT usage policy (1) and 154
extends for the same length of time as the before period, until November 2011. The researchers 155
have verified (using online historical aviation meteorological information, see 11) that during 156
every randomly selected day only regular South flow operations were conducted and that the 157
weather conditions were compatible with visual flight rules (VFR) (1). 158
A preliminary analysis of Before EAT data shows that the volume of runway 17C arrivals 159
that use the EAT appear negligible. This confirms the findings of the aforementioned 160
Engelland’s study (2). Therefore, the research team compares the distribution of runway 17C 161
crossings observed before and after the EAT. However, the effects of runway 17R crossings are 162
considered during the departure delay and throughput analyses. 163
164
3 Data Analysis 165
This section uses observed operations retrieved from the DFW ASDE-X database to assess the 166
impact of the EAT. The changes in the runway crossing distribution relate to both the 167
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 5
operational and safety benefits associated with the EAT while the throughput and delay analysis 168
characterize both the capacity and operational improvements. 169
170
3.1 Runway Crossing and EAT Operation Distributions 171
During south flow configuration at DFW, the locations of taxiways “Y”, “Z”, and “EJ” increase 172
the potential severity of a runway incident due to runway incursion; in fact, aircraft landing on 173
runway 17C reach those points at considerable speed, leaving less time for avoiding a collision, 174
and increasing the potential severity. Taxiway “EL” poses similar concerns for both arrival 175
operations on runway 17C and departure operations on runway 17R. Taxiways B and ER are 176
located south of the designated land and hold short point for runway 17C; therefore, those 177
taxiways are often used while conducting land and hold short operations (LAHSO). Figure 3, a 178
detail from the airport diagram, shows those runways, taxiways, and the EAT. 179
180 181
FIGURE 3 Runways 17R, 17C, intersecting taxiways, and the EAT. (5) 182
183
3.1.1 Methodology 184
The taxiway intersections with runway 17C are aggregated into three groups based on the 185
similarity of their safety impacts: North, Middle, and South. For the After EAT case only, the 186
EAT, represented by taxiway “ES”, is added as a fourth group. The taxiway groups are shown in 187
Table 1. 188
Table 1 Taxiway Groups 189
190
Before
EAT
After
EAT Taxiways
Taxiway
Groups
North
Y
Z
EJ
EK
K8
Middle EL
EM
South
B
A
EQ
ER
N/A EAT ES
191
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 6
The Before EAT runway crossing distribution is obtained by combining runway 192
crossings from twelve randomly selected data sets (days) with a total number of 772 operations 193
(only runway crossings), and the After EAT frequency distribution is obtained by combining 194
runway crossings from fifteen randomly selected data sets (days), with a total number of 807 195
operations (runway crossings and EAT operations). The study confirms the suitability of these 196
sample sizes by using the methodology proposed by Nisen et al. (13); for the minimum sample 197
size determination, α and β are set at 0.05 and a meaningful change is set at ten percent. With 198
these values, the minimum required sample size is 477 operations, which both the Before EAT 199
and After EAT cases surpass. The researchers conduct validation tests (see 10 for more 200
information) to provide additional verification that the randomly selected samples adequately 201
represent their respective sampling frames. 202
The Before EAT distribution is compared with the After EAT distribution using the 203
Pearson Chi Square test (also known as “goodness of fit”). This test evaluates if a change in 204
runway crossing distribution has occurred. The tested null and alternative hypotheses for this 205
goodness of fit test are: 206
207
• H0 = the After EAT distribution is not significantly different than the Before EAT 208
distribution (FA North = FB North, FA Middle = FB Middle, FA South = FB South) 209
210
• H1 = the After EAT distribution is significantly different than the Before EAT distribution 211
(FA North ≠ FB North or FA Middle ≠ FB Middle, or FA South ≠ FB South) 212
213
Where: 214
FA North/Middle/South, = After case frequency for the category 215
216
FB North/Middle/South, = Before case frequency for the category 217
218
The Chi Square values are calculated with the following: 219
220
��������� � ����� ���
�
��� (1)
221
Where: 222
χ2 = Pearson's cumulative test statistic 223
224
Afteri = After value of category “i” 225
226
Beforei= Before value of category “i” 227
228
j = Number of categories 229
230
3.1.2 Results 231
Table 2 compares the frequencies of taxiway group usage between the Before EAT and the After 232
EAT case. 233
234
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 7
Table 2: Comparative Assessment Runway 17C Crossing Before and After the EAT 235
236
Taxiway
Groups
Frequency
Before EAT
Frequency
After EAT Change (%)
North 0.18 0.15 -17%
Middle 0.44 0.02 -95%
South 0.38 0.32 -16%
EAT N/A 0.51 N/A
237
When comparing only the North, Middle and South distributions, the Chi Square test 238
statistic is equal to 446, which is an extremely high value; this statistic corresponds to a p-value 239
that is essentially zero. 240
241
The safety benefits from DFW’s EAT are very clear. This demonstrates that a significant 242
difference exists between the Before and the After case runway crossing frequency distributions, 243
and that the EAT has a positive safety impact on runway crossings. After the EAT, the number 244
of runway crossings reduced on all taxiway groups; on average, the EAT removes fifty-six 245
crossings per day. Most importantly, the “Middle” group’s crossings, which had the highest 246
severity risk in case of a runway incursion, have been significantly reduced and virtually 247
eliminated. In a hypothetical safety assessment process, part of the FAA Safety Management 248
System, the EAT functioned as a risk mitigation strategy to reduce the risk associated with those 249
crossings, based on the FAA risk matrix (14). 250
Within the South group, taxiway “B” accounted for about three quarters of the group’s 251
crossings during the Before EAT case, but after the EAT construction, its frequency decreased to 252
only four percent. This reduces the risk of collisions when land and hold short operations 253
(LAHSO) are conducted on runway 17C. In fact, while “B” is located just 235 ft beyond the 254
LAHSO stop bar, taxiway “ER” crosses that runway’s South threshold, is located more than 255
2,700 ft beyond. 256
After the EAT, 83% of runway 17L arrivals either used the EAT or crossed runway 17C 257
at the low risk “South” intersection. Overall, the risk associated with surface operations appears 258
to have decreased significantly because of the EAT. 259
260
3.2 Departure and Arrival Throughput 261
In addition to providing safety benefits, the FAA and DFW expected the EAT to improve 262
operations and increase capacity. At the time of the EAT’s initial conception, DFW experienced 263
severely constrained operations due to high demand. Throughput represents a commonly used 264
measure of effectiveness for assessing capacity constrained operations. While throughput lacks a 265
standard calculation methodology, its ability to measure the rate of operations makes it extremely 266
useful for assessing the impacts of infrastructure and policy changes. This study uses the 267
methodology discussed in the next section for determining departure and arrival throughput. 268
Figure 4 presents an example of the observed pattern of EAT usage and departure and arrival 269
operations for one day. One of the expected, but clear patterns is that the departure rate and 270
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 8
arrival rate do not peak at the same time, since arrivals generate runway crossings for 17R that 271
periodically interrupt departure operations. 272
273
274 275
FIGURE 4 Departure, arrival, and EAT operations. 276
277
The EAT usage pattern shows that the ATC uses it to allow uninterrupted operations on the 278
departure and arrival runways, during high demand. The impacts of this usage strategy on the 279
departure and arrival throughput are assessed in section 3.2.2. 280
281
3.2.1 Methodology 282
The reduction in runway crossings and the change in their distribution achieved by the EAT’s 283
construction shows that the EAT likely has a positive impact on both arrival and departure 284
operations by reducing their interruption due to crossing aircraft. As mentioned in the 285
introduction, the researchers use the methodology proposed by Hansen et al. to calculate hourly 286
departure and arrival throughput (10) for runways 17R and 17C, respectively. This study 287
replaces the data from the Airline Service Quality Program (ASQP), used by that study, with 288
DFW’s ASDE-X data. Since the ASQP data refers to only the ten largest airlines in the U.S. at 289
the time, which accounted for sixty percent of the flights at DFW (10), the throughput rates 290
calculated by the older study slightly underestimated the total rates (10). Since the ASDE-X 291
database includes all flights at that airport compared to older databases, its data may be used to 292
develop more accurate assessments of airport operational performance. While ASDE-X data 293
gaps may affect accuracy, however, the authors have developed strategies to close the few 294
observed gaps (report). 295
Although the aforementioned study estimated DFW’s arrival throughput for all runways 296
in an aggregate form, the same methodology may be used to determine arrival and departure 297
throughput for an individual runway. To calculate a maximum rate, Hansen’s study found the 298
shortest time in which N flights arrived at their gates. That number had to be “large enough to 299
yield a stable average, but small enough to avoid lulls in demand” (10); as a result, these 300
researchers set N equal to thirty flights. The “hourly arrival rate based on the shortest time 301
interval for thirty arrivals at the airport is defined as the Daily Maximum Throughput Rate 302
0
5
10
15
20
25
30
35
40
45
50Runway 17C and
17L Arrivals
17R Departures
EAT Operations
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 9
(DMTR).” This study replicates this methodology using the DFW ASDE-X data, which 303
provides the departure and arrival times, runway entry, exit, and time in queue. For more 304
detailed information on how this data is processed and interpreted by the SODAA tool, the 305
reader can refer to the user manual. Using the same method for calculating the DMTR, this 306
study calculated the Daily Maximum Departure Demand (DMDD) and the Daily Maximum 307
Average Departure Delay (DMAD). 308
A sample size of five for the before and after cases is adequate for assessing the impact of 309
the EAT on throughput. Five days are selected using Simple Random Sampling from the before 310
and after cases; once again, the researchers verify that VFR conditions exist for the selected 311
days. When VFR conditions do not exist the day is discarded and another day is randomly 312
selected to replace it. Each randomly selected day represents a single data point in the analysis 313
because the maximum daily throughput values are used. The mean daily maximums (DMTR, 314
DMDD and DMAD) for the before case are compared with the same values after the EAT’s 315
construction. A t-test is used to evaluate if EAT has a significant impact on the throughput and 316
delay. 317
318
3.2.2 Results 319
Departure Throughput 320
For the departure throughput, this study focuses on runway 17R, the primary departure runway 321
on the East side during South flow, to determine if the EAT increases departure capacity as 322
measured by departure DMTR. The reduction in runway crossings that result from use of the 323
EAT should have a positive impact on capacity; however, this effect is somewhat muted because 324
the EAT is not used by the vast majority of the arriving aircraft (93%). The study also 325
determines Runway 17R’s DMDD and DMAD. The departure demand differs from the 326
departure throughput because the first is the rate at which aircraft reach the departure queue, 327
which is determined by the software’s logic based on aircraft position, while the second is the 328
rate at which the runway serves the departing aircraft. Table 3 shows the results of the departure 329
throughput analysis for runway 17R and provides the significance level observed for the change 330
in mean values between the before and after cases. 331
332
Table 3 Comparison of Departure Performance for Runway 17R Before and After the 333
EAT 334
335
Runway 17R
Daily Maximum
Throughput Rate
DMTR
(departures/hr)
Runway 17R
Daily Maximum
Departure Demand
DMDD
(aircraft/hr)
Runway 17R
Daily Maximum
Average Departure
Delay
DMAD
(minutes)
Mean Std.
Deviation Mean
Std.
Deviation Mean
Std.
Deviation
Before EAT 39.14 3.51 37.88 2.53 10.95 6.35
After EAT 48.83 3.74 47.34 3.40 6.80 1.00
Significance Level 0.002 0.001 0.222
336
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 10
When comparing the mean values for the DMTR and DMDD observed before and after 337
the EAT for runway 17R departures, the extremely low p-values demonstrate that a significant 338
increase has occurred after the implementation of the EAT; the reduction in crossings appears to 339
provide a direct operational benefit and permits runway 17R to accommodate higher demand. 340
The simulation conducted during EAT development (1) estimated an average increase of the 341
total departure rate by 18 aircraft per hour when considering all departure runways. Given that 342
DFW ATC, essentially, uses two runways for departures during South flow, the results of that 343
simulation can be divided by two to consider only one departure runway, by assuming that 344
departures may be equally distributed between East and West side runways. This study observes 345
an average increase in DMTR of 9.69 aircraft per hour, which appears to validate the simulation 346
study results. The high variability in the DMAD makes its assessment more difficult. While a 347
thirty-eight percent reduction is observed, a much larger sample size is required to verify this 348
result statistically. Departure delays should be investigated in more depth in future research. 349
350
Arrival Throughput 351
This analysis measures the increase in average Daily Maximum Arrival Throughput Rate 352
(DMATR) for runway 17C as a result of the fewer crossings observed after the EAT; in addition, 353
the combined arrival throughput for runways 17C and 17L is calculated (see Table 4). 354
355
Table 4 Comparison of the Arrival Throughput Rate Before and After the EAT for 356
Runway 17C Alone and Combined with 17L 357
358
Daily Maximum Arrival Throughput Rate
(DMATR)
(departures/hr)
Runway 17C Runways
17C and 17L
Mean Std.
Deviation Mean
Std.
Deviation
Before EAT 31.87 0.88 39.96 3.15
After EAT 34.05 0.97 56.52 5.49
Significance Level 0.004 0.001
359
Although the increase of the average DMATR observed after the EAT for runway 17C is 360
statistically significant, its magnitude is limited. In average, only about two more arrivals per 361
hour are observed. However, when considering runways 17C and 17L combined, their average 362
DMATR increased dramatically after the EAT; these runways are now serving over sixteen more 363
arrivals. To exclude that these results are due to a lull in departure demand during the DMATR 364
time interval, the researchers compared the observed demand and throughput rates for runway 365
17R and found that they had increased after the EAT. Before the EAT, the mean departure 366
demand during runway 17C DMATR was 26 aircraft per hour, while after the EAT the mean 367
demand increased to 30.2; when considering runways 17C and 17L combined, the mean 368
departure demand increased from 25 to 26.6. 369
370
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 11
4 CONCLUSIONS AND FUTURE RESEARCH 371
This research compares ASDE-X data for operations at DFW before and after the 372
implementation of the EAT, which represents the current ATC operating procedures. The 373
findings clearly indicate that the EAT at DFW achieves its goals of improving runway safety and 374
airfield capacity. Firstly, it improves the safety of East side surface operations by reducing the 375
number of crossings on runway 17C, used for arrivals, and by relocating other crossings away 376
from more dangerous locations. Secondly, the East side runways’ capacity increases because the 377
arrival and departure operations can be better sequenced and coordinated with the conflicting 378
surface operations. 379
This study finds that the EAT is used by between 47% and 55% of runway 17L arrivals; 380
however, only between 6% and 8% of all the East side runways 17C and 17L arrivals use the 381
EAT. These EAT usage levels correspond to an average daily reduction of 56 runway crossings. 382
In addition, the EAT has been a catalyst for implementing operational procedures that mitigate 383
runway crossing risk due to their locations, considering the FAA risk assessment matrix (14). 384
The EAT operating procedure adopted by DFW ATC allows using the low risk taxiway “ER” to 385
cross runways 17C and 17R on in lieu of using the EAT. Therefore, virtually all of the average 386
48 runway crossings per day observed before the EAT runway crossings at the midfield taxiway 387
EL, which posed the most safety concerns among all crossing due to its location and the potential 388
conflict with departure and arrival operations at high speed, have been eliminated. Also, when 389
conducting LAHSO on runway 17C, this observed increase in use of taxiway “ER” has mitigated 390
the risk by increasing the minimum potential distance between a landing aircraft and a crossing 391
aircraft from about 235 feet to about 2,700 feet. 392
This study demonstrates the improvement in runway capacity due to EAT operations by 393
quantifying their daily maximum throughput. The observed mean DMTR after the EAT, is 394
significantly larger than the mean DMTR observed before the EAT. The departure capacity 395
finding is consistent with the simulation performed in 2003 by FAA and DFW in cooperation 396
with NASA (1); however, the simulation failed to identify the improvement in arrival capacity 397
identified in this study. The ASDE-X data reveals its flexibility for use in empirically validating 398
simulated or modeled results. Although this study uses an earlier study’s methodology for 399
determining the daily maximum throughput, the results of these two studies are not comparable 400
because in the meantime the Area Navigation (RNAV) departure and arrival procedures, which 401
have improved DFW’s terminal airspace efficiency (15), have been implemented. 402
Although the observed mean DMAD is not statistically significant, the improvement in 403
departure delay appears to be tangible. In fact, despite the increase in observed departure 404
demand after the EAT, the mean DMAD has reduced from eleven minutes to about seven 405
minutes. For all these reasons, this retrospective empirical study demonstrates the improvement 406
in operational efficiency that result from the reduction in runway crossings. 407
These safety and capacity benefits by themselves show promise for justifying the EAT’s 408
construction, and if the number of runway 17C arrivals using the EAT would increase, the 409
observed effectiveness in improving surface operation safety and in increasing arrival and 410
departure capacity would likely increase further. 411
Finally, study provides opportunities for future research based on ASDE-X data; these 412
projects may include comparisons between actual airfield performance with the estimated results 413
of the models typically used by the FAA (16), and other models under development. This study 414
can be extended to evaluate the entire airfield performance, and the factors impacting the 415
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 12
throughput, such as the fleet mix (8) can be examined more closely to determine their importance 416
and magnitude. 417
In an effort to seek future research on EAT performance, the authors compared the taxi 418
times for aircraft landing on runway 17L and using different routes to reach terminal C, as shown 419
in Figure 5. 420
421
422 423
Figure 5 Examples of taxi routes. 424
425
Taxi time for the aircraft routed via taxiway “EL” was 8 minutes and 33 seconds, while 426
the aircraft using the EAT took 9 minutes and 50 seconds, eight seconds less than the taxi time 427
for the aircraft using taxiway “ER”. Although these are only three examples of taxi times and do 428
not prove any assumption, they show that this EAT route can have similar taxi times to runway 429
crossing taxi routes. Most importantly, these examples confirm that EAT operations deserve 430
further investigation to assess its impacts on the arrival delay, which is a critical performance 431
measure for all air carriers. 432
Based on a detailed analysis of taxi times, the operational costs for air carriers and the 433
environmental impacts associated with EAT use can be estimated to complete the assessment of 434
this innovative infrastructure. The results of all of these studies, which integrate the safety and 435
operational benefits together, may also be used by the FAA and the airports to evaluate or to 436
validate the cost-effectiveness of the EATs. 437
438
439
ACKNOWLEDGEMENTS 440
441
FAA Southwest Region Office, Fort Worth, TX 442
443
FAA/NASA North Texas Research Station, Fort Worth, TX 444
445
Mosaic ATM, Inc., Leesburg, VA 446
447
Taxiway EL
Taxiway ER
EAT
TRB 2013 Annual Meeting Paper revised from original submittal.
Massidda A., Mattingly S. 13
REFERENCES 448
1) VFR 2030: Vision of The Future. Realized. Dallas/Fort Worth International Airport, 2009 449
update. 450
2) Engelland S., Analysis of Perimeter Taxiway Operations. Presented at the 10th AIAA 451
Aviation Technology, Integration, and Operations (ATIO) Conference. 452
3) FAA Airport Obstructions Standards Committee (AOSC) Decision Document #06 Summary 453
Approved: June 8, 2005 Dallas / Fort Worth (DFW) End-Around Taxiway System 454
4) Satyamurti, S. D., Mattingly, S.P., Flight Track Data Analysis to Estimate Aircraft Height 455
Above a Perimeter Taxiway System. Presented at 87th
Annual Meeting of the Transportation 456
Research Board, Washington, D.C., 2008. 457
5) Federal Aviation Administration, Airport Diagrams online database, accessed on May 1, 458
2012. www.faa.gov/airports/runway_safety/diagrams/. 459
6) N DFW ATCT 7110.518 – Southeast End-Around Taxiway Operations and RY 17R/17C 460
Departures 461
7) Airport Surface Detection Equipment, Model X (ASDE-X), Federal Aviation Administration 462
www.faa.gov/air_traffic/technology/asde-x/. Accessed May 12, 2012. 463
8) Simaiakis, Ioannis; Balakrishnan, Hasmsa; Khadilkar, Harshad; Reynolds, Tom G.; 464
Hansman, R. John; Reilly, B.; Urlass, S. Demonstration of Reduced Airport Congestion 465
Through Pushback Rate Control. http://18.7.29.232/handle/1721.1/60882. Accessed on May 466
22, 2012 467
9) Balakrishnan H., Jung Y., A Framework for Coordinated Surface Operations Planning at 468
Dallas-Fort Worth International Airport. American Institute of Aeronautics and Astronautics 469
Conference. Hilton Head, NC, August 2007. 470
10) Hansen, M., Tsao, J., Huang, S., Wei, W., Empirical Analysis of Airport Capacity 471
Enhancement Impacts: A Case Study of DFW Airport. Presented at 78th Annual Meeting of 472
the Transportation Research Board, Washington, D.C., 1999. 473
11) Mattingly S., Massidda A., Preliminary Study on the Safety Impact of the End-Around 474
Taxiway at the Dallas/Fort Worth International Airport using FAA/NASA Electronic Data. 475
Final Report, May 2012. 476
12) Weatherunderground.com weather history, http://www.wunderground.com/history/. 477
Accessed from December 2011 to July 2012. 478
13) Nisen J., A simple method of computing the sample size for Chi-square test for the 479
equality of multinomial distributions. Computational Statistics and Data Analysis, Elsevier, 480
2008. 481
14) Federal Aviation Administration, Order 5200.11. Effective 08/30/2010. 482
15) Mayer, R., Evaluation of RNAV Departure Operations at Dallas Fort-Worth International 483
Airport. Presented at the 25th Digital Avionics Systems Conference, 2006 IEEE/AIAA. 484
Portland, OR, 15-19 Oct. 2006. 485
16) Federal Aviation Administration, Advisory Circular (AC) 150/5060-5 (Change 2), 486
Airport Capacity and Delay. 487
17) Davis, William “Perimeter Taxiways and Improved Surface Safety,” Office of Runway 488
Safety, FAA, Washington D.C., 2002 489
490
491
TRB 2013 Annual Meeting Paper revised from original submittal.