Audit Report: Contract Towers Are More Cost Effective Than Comparable FAA Towers and Have Similar Safety Records.

Contract Towers Are More Cost Effective

Than Comparable FAA Towers and Have

Similar Safety Records

Report No. AV2020028

April 28, 2020

What We Looked At

Established in 1982 at 5 low-activity control towers, the Federal Aviation Administration’s (FAA)

Contract Tower (FCT) Program currently consists of 254 contract towers in 46 states operated by 3

contractors and the Air National Guard. Managing about 28 percent of the Nation’s air traffic control

operations, contract towers constitute an essential part of the National Airspace System (NAS). Our

audit objective was to assess the FCT Program’s cost effectiveness and safety record. We statistically

grouped towers based on characteristics that affect air traffic controller and tower workloads.

Specifically, we gathered and examined hours of operations, numbers of takeoffs and landings, types

of aircraft handled, and runway configurations. Based on these characteristics, we used two statistical

methods to group 351 air traffic control towers, consisting of 248 contract towers and 103 lower level

FAA towers. Our methods produced groups containing a mixture of comparable FAA and contract

towers. We determined the towers within each group were similar to each other and then analyzed

and directly compared their cost and safety data. We reviewed cost and safety data between fiscal

years 2015 and 2018 for the universe of 351 towers.

What We Found

Between fiscal years 2015 and 2018, contract towers were more cost effective per aircraft handled

than comparable FAA towers, and that the safety records of contract and comparable FAA towers

were similar. On average, contract towers used at least 47.6 percent fewer resources—or incurred

lower controller staffing costs—per aircraft handled per year even though comparable FAA towers

handled more total flights. Furthermore, while contract towers had statistically fewer safety events per

aircraft handled, we do not believe the difference between these numbers and those of FAA’s towers

is meaningful because, among other reasons, the numbers of safety related events across the NAS

were very low relative to the total number of flights.

Our Recommendations

We are making no recommendations.

Contract Towers Are More Cost Effective Than Comparable FAA

Towers and Have Similar Safety Records

Requested by the U.S. House of Representatives Committee on Transportation and Infrastructure and its

Subcommittee on Aviation

Federal Aviation Administration | AV2020028 | April 28, 2020

All OIG audit reports are available on our website at www.oig.dot.gov.

For inquiries about this report, please contact our Office of Government and Public Affairs at (202) 366-8751.

AV2020028

Contents

Memorandum 1

Background 3

Results in Brief 6

Contract Towers Are More Cost Effective and Have Safety Records Similar

to Those of Comparable FAA Towers 7

Conclusion 12

Recommendations 13

Agency Comments and OIG Response 13

Actions Required 13

Exhibit A. Scope and Methodology 14

Exhibit B. Entities Visited or Contacted 18

Exhibit C. List of Acronyms 20

Exhibit D. Cost Comparisons 21

Exhibit E. Comparisons of Safety Events 26

Exhibit F. List of 248 Contract Towers 28

Exhibit G. List of 103 FAA-Operated Towers 35

Exhibit H. Detailed Scope and Methodology 38

Exhibit I. Major Contributors to This Report 65

AV2020028 1

Memorandum

Date: April 28, 2020

Subject: Contract Towers Are More Cost Effective Than Comparable FAA Towers and Have

Similar Safety Records | Report No. AV2020028

From: Matthew E. Hampton

Assistant Inspector General for Aviation Audits

To: Federal Aviation Administrator

The Federal Aviation Administration’s (FAA) Contract Tower (FCT) Program

consists of 254 contract towers

in 46 states operated by 3 contractors and the

Air National Guard. Contract towers constitute an essential part of the National

Airspace System (NAS) because they manage about 28 percent of the Nation’s air

traffic control operations. FAA conducts benefit-cost analyses (BCA) to support

decision-making for establishing and discontinuing towers, and determining the

share of towers’ operating costs that airport sponsors will pay.

Recognizing the program’s importance, the former Chairmen of the U.S. House of

Representatives Committee on Transportation and Infrastructure and its

Subcommittee on Aviation requested that we update our previous work on

contract towers.

Our audit objective was to assess the FCT Program’s cost

effectiveness

and safety record.

We conducted this audit in accordance with generally accepted Government

auditing standards. Exhibit A details our overarching scope and methodology.

Exhibit B lists the entities we visited or contacted, and exhibit C presents a list of

acronyms. This report also contains a detailed scope and methodology in exhibit

As of January 1, 2019, FAA added the Albert J. Ellis Airport in Richlands, NC, and the North Texas Regional Airport-

Perrin Field in Dennison, TX, to FCT, bringing the total number of contract towers to 256.

49 U.S.C. § 47124 stipulates when FAA should conduct benefit-cost analyses.

See Contract Towers Continue To Provide Cost-Effective and Safe Air Traffic Services, but Improved Oversight of the

Program Is Needed (OIG Report No. AV-2013-009), November 5, 2012.

We define “cost effectiveness” as the use of fewer resources per aircraft handled than comparable towers.

We will present the results of our work on the status of the revisions to the BCA, our second audit objective, in a

future report.

U.S. DEPARTMENT OF TRANSPORTATION

OFFICE OF INSPECTOR GENERAL

AV2020028 2

We appreciate the courtesies and cooperation of Department of Transportation

representatives during this audit. If you have any questions concerning this

report, please call me at (202) 366-0500, or Nelda Z. Smith, Program Director, at

(202) 366-2140.

cc: The Secretary

DOT Audit Liaison, M-1

FAA Audit Liaison, AAE-100

AV2020028 3

Background

In 1982, FAA began a program that contracted air traffic control services at five

low-activity control towers that had been closed as a result of the Professional Air

Traffic Controllers Organization’s strike the previous year. In 1998, Congress

provided funding for a cost-sharing program

for existing contract towers that

would otherwise have been ineligible for the FCT Program. During 2008’s

economic downturn, FAA suspended admittance of new towers to the program.

In 2012, Congress capped local cost-share contributions at 20 percent of total

costs.

In our 2012 report, we stated that contract towers provided air traffic control

services at lower costs than similar FAA towers. We reported that on average, a

contract tower cost about $1.5 million less to operate than a comparable FAA

tower, primarily due to lower staffing and salary levels. We also reported that

contract towers had lower numbers and rates of safety events than similar FAA

towers, and that airspace users supported the FCT Program.

In November 2017, FAA began to allow new and replacement towers built by

airport sponsors to enter the FCT Program.

Currently, FAA has no process for

converting existing Agency towers to contract towers,

or converting contract

towers back to FAA towers.

To assess the relative cost-effectiveness and safety of contract and FAA towers,

we grouped towers with similar characteristics. To develop the groups, we

examined the numbers of takeoffs and landings, types of aircraft handled, and

runway configurations—characteristics that affect air traffic controller and tower

workloads. Using FAA definitions, we identified five types of aircraft operations

(see table 1)

The cost-share program allows existing contract towers that fall below the established benefit-cost threshold to

remain in the program by paying a portion of the costs to operate their tower.

FAA Order JO7210.78, FAA Contract Tower (FCT) New Start and Replacement Tower Process, November 29, 2017.

The last FAA tower converted to a contract tower in 1999.

AV2020028 4

Table 1. Types of Aircraft Operations

Source: FAA

We further determined whether each tower manages aircraft from a single

runway or from multiple crossing, converging, or parallel runways. Based on these

characteristics, we used two statistical methods to group a universe of 351 air

traffic control towers, consisting of 248

contract towers (see exhibit F for a list)

and 103 comparable FAA towers (see exhibit G for a list).

We used one

statistical method as the primary method to develop our results and the other

method as a robustness check.

Application of these statistical methods,

described in exhibit H, resulted in four groups, each containing a mixture of FAA

and contract towers that shared similar characteristics. See table 2 for the

breakdown of the number of FAA and FCT towers in each group. Because we

determined the towers within each group were similar to each other, we could

analyze and directly compare their cost and safety data. We used data for the

universe of 351 towers collected between fiscal years 2015 and 2018.

From the universe of 254 contract towers, we excluded 6 that the Air National Guard operates.

We considered only FAA towers with radar operating in Class C or D airspace because all contract towers operate in

these airspaces. Class C airspace is generally airspace from the ground to 4,000 feet above an airport control tower.

Class D airspace is generally airspace from the ground to 2,500 feet above an airport control tower.

A robustness check determines whether results change when using alternative assumptions or methods. When the

results of a robustness check correspond with the results from a primary analysis, the results corroborate the main

results.

Aircraft Operations Definition

Air carrier Operations of aircraft with seating capacities of over 60 or maximum payload capacities

of over 18,000 pounds carrying passengers or cargo for hire or compensation.

Air taxi Operations of aircraft with seating capacities of 60 or less, or maximum payload

capacities of 18,000 pounds or less, carrying passengers or cargo for hire or

compensation.

General aviation All civil aircraft operations not classified as air carrier or air taxi.

Military All military takeoffs and landings at both FAA and FCT facilities.

Local Aircraft operations that remain in the local traffic pattern, execute low passes at the

airport, and operate between the airport and a designated area within a 20−mile radius.

AV2020028 5

Table 2. Tower Groups

Group FAA FCT

1 29 17

2 17 44

3 12 71

4 45 116

Totals 103 248

Note: To prevent competitive advantage or disadvantage in current or future FCT contract bidding,

we do not specify which FAA or contract towers are included in each group. In our detailed scope

and methodology (exhibit H), we include summary statistics for each group and describe the process

we used to arrive at our conclusions.

Source: OIG

The groups have the following characteristics:

• Group 1 handles the highest average number of total aircraft per hour

(22.5), the highest average number of general aviation aircraft (8.3), and

the highest average number of air carrier aircraft per hour (2.8). It also has

a mixture of parallel runway configurations with some converging

runways. In this group, for example, FAA operates Brackett Field (POC) in

California, while contractors operate Brown Field Municipal (SDM) in

California.

• Group 2 handles the second highest average number of total aircraft per

hour (14.7) and on average the most military aircraft per hour (1.0). It

contains no single runway configurations. In group 2, for example, FAA

operates Nantucket Memorial (ACK) in Massachusetts, while contractors

operate Front Range (FTG) in Colorado.

• Group 3 handles the third highest average total number of aircraft per

hour (13.8), the second highest average number of general aviation

aircraft, and the lowest average number of air carrier aircraft. All towers in

this group have just one runway. In this group, for example, FAA operates

San Gabriel Valley (EMT) in California and contractors operate Bellingham

International (BLI) in Washington.

• Group 4 handles the lowest average number of total aircraft per hour

(13.0), but handles more air carrier aircraft than Group 3. It has crossing

and parallel runway configurations. In group 4, for example, FAA operates

Flying Cloud (FCM) in Minnesota while contractors operate Dothan

Regional (DHN) in Alabama.

AV2020028 6

The level of air traffic control complexity differs based on many factors. Factors

such as aircraft speed and performance; runway and taxiway layout, length, and

capacity; airspace class; terrain; proximity to other airports; interaction with

foreign countries; and military operations all effect the complexity of air traffic

control. We created our groupings based on numbers of takeoffs and landings,

types of aircraft handled, and runway configurations.

Results in Brief

Contract towers are more cost effective than comparable

FAA towers and have similar safety records.

To compare contract and FAA towers for this review, we grouped 351 air traffic

control towers with similar characteristics, and examined the numbers of takeoffs

and landings, types of aircraft handled, and runway configurations—

characteristics that affect air traffic controller and tower workloads. We found

that between fiscal years 2015 and 2018, contract towers were more cost

effective per aircraft handled than comparable FAA towers, and that the safety

records of contract and comparable FAA towers were similar. On average,

contract towers used at least 47.6 percent fewer resources than comparable FAA

towers.

Contract towers on average used fewer resources—or incurred lower

controller staffing costs—per aircraft handled per year even though comparable

FAA towers handled more total flights. Furthermore, while contract towers had

statistically fewer safety events per aircraft handled, we do not believe the

difference between these numbers and those of FAA’s towers is meaningful

because, among other reasons, the numbers of safety related events across the

NAS were very low relative to the total number of flights.

We are making no recommendations.

The 47.6 percent fewer resources is based on operating costs without overhead. See table 4.

AV2020028 7

Contract Towers Are More Cost Effective and Have

Safety Records Similar to Those of Comparable FAA

Towers

Based on our methodology, contract towers use fewer resources than

comparable FAA towers. Furthermore, the safety records of contract and

comparable FAA towers are similar.

Contract Towers Use on Average Fewer

Resources Per Aircraft Handled

Based on our methodology, contract towers on average use fewer resources

per

aircraft handled than comparable FAA towers. For the purposes of this report, we

define “use of fewer resources” as cost effectiveness. In this context, cost equals

the sum of all costs for fiscal years 2015 through 2018 divided by the total

number of aircraft operations during the same time period.

We found that between fiscal years 2015 and 2018, contract towers were more

cost effective per aircraft handled than FAA towers with similar runway

configurations and mix of air traffic. On average, they used at least 47.6 percent

fewer resources

than comparable FAA towers (see table 4, Group 4). In our

analysis, we examined the following tower operating costs:

• Labor and Benefits Costs. We first considered labor costs, including all

labor and benefits payments. The average labor and benefits cost of

operating a contract tower is between $10.45 and $21.02 lower per

aircraft handled than those of comparable FAA towers (see table 3).

“Resources” refers primarily to controller staffing costs. The methodologies used to determine controller staffing at

FAA and contract towers differ. At FAA towers, staffing is determined by FAA’s Office of Labor Analysis which bases

crew shifts on the numbers of air traffic operations and minutes of communication required. Contract tower staffing

levels are determined by each contractor and then approved by FAA. Contract towers must have staff of at least the

equivalent of four full-time controllers.

For example, if an FAA tower’s cost per aircraft was $25.87 and a contract tower’s cost per aircraft was $13.55, then

the contract tower uses 47.6 percent fewer resources per aircraft handled.

AV2020028 8

Table 3. FAA and Contract Tower Controller and Labor Costs for Fiscal Years

2015 through 2018

Costs Group 1 Group 2 Group 3 Group 4

Average total FAA labor

and benefits

$15,735,462.12 $15,091,367.00

$10,817,413.79

$11,400,187.53

Average FAA labor and

benefits cost per aircraft

handled

$22.34

$29.63 $19.93

$22.89

Average total contract

tower labor and benefits

$2,676,222.39 $2,410,237.62 $2,304,554.78 $2,311,141.44

Average contract tower

labor and benefits cost per

aircraft handled

$7.41

$8.61 $9.48

$10.87

Difference in average FAA

and contract tower labor

and benefits cost per

aircraft handled

$14.93 $21.02 $10.45 $12.02

Average percent fewer

resources used by

comparable contract towers

66.8% 70.9% 52.4% 52.5%

Please see exhibit D for included cost categories. All figures are within-group averages, including totals.

Total contract tower labor cost includes fringe benefits for contract controllers.

Source: OIG analysis.

• Operating Costs without Overhead.

We then looked at all costs except

overhead, including labor costs, utilities, telecommunications, and leases,

among others. Contract towers’ average cost per aircraft handled is

between $10.67 and $23.06 lower than FAA counterpart towers (see table

4).

The term “overhead costs” refers to the cost of indirect support services provided by FAA staff offices—such as

human resources management—and ATO’s Chief Operating Officer.

AV2020028 9

Table 4. FAA and Contract Towers Operating Costs without Overhead during

Fiscal Years 2015 through 2018

Costs Group 1 Group 2 Group 3 Group 4

Average FAA cost per aircraft handled $25.17

$33.85

$21.95

$25.87

Average contract tower cost per aircraft

handled

$8.90

$10.80

$11.28

$13.55

Difference in average FAA and contract

tower operating costs per aircraft handled

(not including overhead)

$16.26 $23.06 $10.67 $12.32

Average percent fewer resources used by

comparable contract towers

64.6% 68.1% 48.6% 47.6%

Note: Please see exhibit D for included cost categories. All figures are within-group averages, including totals. Due to

rounding, some figures may be higher or lower than the sum.

Source: OIG analysis.

• Operating Costs including Overhead. We then looked at operating

costs including overhead and found that contract tower overhead costs

average remained between $12.64 and $27.20 less per aircraft handled

than FAA counterpart towers (see table 5).

Table 5. FAA and Contract Towers Operating Costs Including Overhead during

Fiscal Years 2015 through 2018

Costs Group 1 Group 2 Group 3 Group 4

Average FAA cost per aircraft handled $29.74 $39.91 $25.93

$30.51

Average contract tower cost per aircraft

handled

$10.50 $12.71

$13.29

$15.94

Difference in average FAA and contract

tower operating costs per aircraft handled

(including overhead)

$19.24 $27.20 $12.64 $14.57

Average percent fewer resources used by

comparable contract towers

64.7% 68.2% 48.8% 47.7%

Note: Please see exhibit D for included cost categories. All figures are within-group averages, including totals.

Source: OIG analysis.

AV2020028 10

The Safety Records of FAA and Contract

Towers are Comparable

Based on our methodology, contract towers have experienced statistically fewer

safety events per aircraft handled than FAA towers. However, in our opinion,

these differences are not meaningful because at both contract and FAA towers,

numbers of reported safety events are very low relative to total flights, which is in

the single digits per million aircraft handled. Moreover, many of these events are

self-reported, and both a mandatory and voluntary report may be submitted for a

single safety event.

Within each of our tower groups, safety events were rare in all metrics. For

example, of the 351 towers we analyzed, 242 towers (68.95 percent) did not

report a single air traffic controller involved risk analysis event or surface risk

analysis event

between fiscal years 2015 and 2018 (see table 6). These 351

towers handled on average 84,328 total aircraft per year. Only 19 towers—17 FAA

and 2 contract—handled over 220,000 total aircraft per year. Furthermore, FAA

towers that we analyzed averaged 8.24 controller-involved reported risk analysis

events and reported surface risk analysis events per 1 million aircraft handled,

while contract towers in our analyses averaged not more than 1.08 events per 1

million aircraft handled.

See exhibit A for definitions of “risk analysis event” and “surface risk analysis event.”

ANC in Anchorage, AK, an FAA tower, is responsible for 1.89 of the 8.24 events per million aircraft handled during

the sample period.

AV2020028 11

Table 6: Towers with Risk Analysis and Surface Risk Analysis Events Involving an

Air Traffic Controller During Fiscal Years 2015 through 2018

Aircraft Handled and Events Group 1 Group 2 Group 3 Group 4

FAA average aircraft handled per year 176,086 127,340 135,711 124,513

FCT average aircraft handled per year 90,263 69,998 60,778 53,160

Average number of risk analysis and

surface risk analysis events involving air

traffic controllers per million aircraft

handled at FAA towers

10.97 7.04 7.37 6.47

Average number of risk analysis and

surface risk analysis events involving air

traffic controllers per million aircraft

handled at FCTs

2.28 1.06 0.52 1.18

Total risk analysis and surface risk analysis

events involving air traffic controllers per

million aircraft handled

8.96 3.53 2.40 3.70

Note: FAA measures collision opportunity as the number of potential pairs of aircraft—a function of the

square of the number of aircraft operating in the area (see FAA-APO-90-7, Establishment and Discontinuance

Criteria for Airport Traffic Control Towers, August 1990). Consequently, we would expect increasingly higher

numbers of reports of safety incidents per operation as the number of aircraft handled increases.

Source: OIG analysis.

In our 2012 report, we noted FAA’s transition to a new approach to safety

oversight at all air traffic facilities. Prior to October 2010, FAA’s air traffic facility

oversight included facility safety evaluations of both FAA and contract towers

every 3 years. In January 2012, FAA transitioned to risk-based oversight of its air

traffic facilities. Under the new system, FAA analyzes safety event data reported

by air traffic controllers to identify specific safety problems or trends at air traffic

facilities. FAA then focuses its oversight based on these analyses.

Since implementing the new system, FAA has completed a higher number of

external compliance verifications (ECV) at FAA towers than contract towers. An

ECV is a safety review conducted onsite by FAA’s Safety Quality Control Group

based on risks associated with reported events and other quality control data

along with observed trends. Between fiscal years 2015 and 2018, FAA conducted

ECVs at 96 (93.2 percent) of the 103 FAA towers we reviewed. Several of these 96

towers received multiple ECVs. During the same time period, FAA conducted

ECVs at 76 (30.6 percent) of the 248 contract towers we reviewed. This trend may

be attributable to the fact that FAA towers handle more aircraft and have a higher

number of reported safety events.

The FCT contract requires contractors to have quality assurance programs that

include internal facility evaluations. Each contractor has established its own

AV2020028 12

quality assurance program and conducts evaluations of its towers. FAA does not

perform regular reviews of these programs. Instead, the Agency relies on

contractors’ annual self-evaluations to assess contract tower performance and

identify areas for improvement.

In fiscal year 2008, FAA implemented a voluntary safety reporting program

known as the Air Traffic Safety Action Program (ATSAP) at air traffic facilities. A

voluntary safety reporting program provides a confidential, but not anonymous,

non-punitive mechanism for employees to report aviation safety events and non-

compliance with FAA directives and Federal safety regulations. Reporters use

their discretion in writing the reports, which they submit electronically to a review

committee. However, for an event involving national security or the immediate

safety of a flight, a mandatory report must be completed irrespective of a

voluntary report.

In our 2012 audit report, we recommended that FAA implement a similar

voluntary safety reporting program at contract towers to ensure more

comprehensive reporting of safety events. In response, in fiscal year 2015, FAA

established the SAFER-FCT

program at contract towers. Contract and FAA-

operated towers now have the same safety requirements and methods for

reporting safety events. However, it is important to note that FAA personnel

voluntarily report significantly more safety events than contract tower personnel

do. Between fiscal years 2015 and 2018, controllers at the 103 FAA towers we

reviewed made over 7,000 voluntary reports while contract tower controllers

made 574. We did not find any particular reason for this reporting disparity,

though FAA personnel’s greater exposure and familiarity with voluntary reporting

may result in more reports.

Conclusion

The FCT Program allows FAA to contract out the operation of some lower-activity

control towers at airports which otherwise would not be able to offer air traffic

control services. In addition to providing tower services at lower cost and with

similar levels of safety, the Program is important because it connects smaller

airports and rural communities to the national air transportation system.

Consequently, the FCT Program continues to be a viable alternative for FAA to

provide cost effective and safe air traffic services to aviation users and the flying

public.

The voluntary safety reporting program for contract towers is called SAFER-FCT or the federal contract towers

safety action program.

AV2020028 13

Recommendations

We are not making any recommendations.

Agency Comments and OIG Response

We provided a draft of this report to FAA on March 16, 2020. On April 6, 2020, we

were notified that the Department would not be issuing a written response.

Throughout the review, we discussed our methodology, findings, and conclusions

with FAA representatives. Where appropriate, we incorporated FAA’s comments

and input received during our meetings at both FAA headquarters and field

locations.

Actions Required

No actions required.

Exhibit A. Scope and Methodology 14

Exhibit A. Scope and Methodology

We conducted this performance audit between August 2018 and March 2020, in

accordance with generally accepted Government auditing standards as

prescribed by the Comptroller General of the United States. Those standards

require that we plan and perform the audit to obtain sufficient, appropriate

evidence to provide a reasonable basis for our findings and conclusions based on

our audit objectives. We believe that the evidence obtained provides a

reasonable basis for our findings and conclusions based on our audit objectives.

The Chairmen of the U.S. House of Representatives Committee on Transportation

and Infrastructure and its Subcommittee on Aviation requested that we update

our prior work. Our audit objectives were to assess (1) the FCT Program’s cost

effectiveness and safety record and (2) the status of the benefit-cost analysis

revisions. In this report, we present the results of our work on the program’s cost

effectiveness and safety record. We will present the results of our work on the

status of the benefit cost analysis revisions in a separate report.

We interviewed officials at FAA Headquarters and two of three regional offices

about different aspects of the FCT program. We selected the two regional offices

based on availability. We conducted site visits at 5 of 103 FAA towers that we

selected based on our statistical groupings and proximity. We also visited 12 of

248 contract towers that we statistically selected. To make this selection, we used

a two-stage stratified probability proportional to size sample in which size was

the 5-year tower operations average. We met with the air traffic managers at

these towers and the airport directors or representatives at the contract towers to

discuss the towers’ air traffic operations. We also met with the three FCT Program

contractors.

To gain perspective on the cost effectiveness and safety of contract towers, we

interviewed aviation industry officials at three trade groups, including the

American Association of Airport Executives, the Aircraft Owners and Pilot

Association, and the National Air Transportation Association. We also interviewed

National Air Traffic Controllers Association officials to obtain air traffic controllers’

perspective on the cost effectiveness and safety of contract towers.

To assess the cost effectiveness and safety record of contract towers, we grouped

towers based on characteristics that affect air traffic controller and tower

workloads. To compare similar towers, we gathered and examined hours of

operations, numbers of takeoffs and landings, types of aircraft handled, and

runway configurations provided by FAA’s ATO. We also gathered and validated

operations data using the Operations Network (OPSNET), the official source of

NAS air traffic operations and delay data. To validate the contract towers’ hours

of operations, we used the seven FCT Program contracts and contract

Exhibit A. Scope and Methodology 15

modifications. We validated the hours of operation for FAA towers using data

from FAA’s Office of Labor Analysis. To validate ATC levels for FAA towers, we

used data from FAA's Office of Labor Analysis and ATO Staffing Workbook. We

found the data reliable enough for the purposes of this audit.

We identified five types of aircraft operations—air carrier, air taxi, general

aviation, military, and local. We also determined whether each tower managed

aircraft from a single runway or multiple crossing, converging, or parallel

runways. Based on these characteristics, we used two statistical methods (see

exhibit H) to group a universe of 351 air traffic control towers, consisting of 248

contract towers (see exhibit F for a list) and 103 comparable FAA towers (see

exhibit G for a list). We arrived at 248 contract towers from a universe of 254 by

excluding 6 towers operated by the Air National Guard. We arrived at the 103

comparable FAA towers from a list of 264 by excluding towers (1) whose ATC

levels were 9 or above

because they would be more complex and have more

traffic operations than contract towers; (2) that were co-located with terminal

radar approach control facilities (TRACON) because they would not be

comparable to contract towers in staffing and operational purposes; or (3) that

did not operate in Class C or D airspace because contract towers operate only in

Class C or D airspace.

Our methods resulted in four groups that we used to analyze cost and safety data

between fiscal years 2015 and 2018. We used one method as the primary method

to report our results and the second method as a robustness check. Our

methodology was reviewed by Michael Simeone, of Arizona State University, and

Jacob LaRiviere of Microsoft; they both found it to be sound.

We received extensive cost data from FAA’s Office of Financial Operations for

both FAA operated and contract towers. See exhibit D for the cost categories we

used in the cost effectiveness analysis. Within each statistical group, we

compared cost metrics among FAA and contract towers. We analyzed costs in

three ways—air traffic control labor/benefit costs only; operating costs without

overhead; and operating costs including overhead. We compared each cost

category on a per aircraft basis because some towers handled more aircraft than

others. We validated FAA’s air traffic control labor and benefit liability

expenditures for FAA towers using FAA's payment records downloaded from

Delphi, DOT’s automated financial accounting system. We validated contract

controller labor costs by comparing the monthly service invoices submitted by

the FCT contractors to the costs approved and paid by FAA.

Each FAA facility is classified as a level 4 through 12, based on numerous factors including traffic volume,

complexity and sustainability of traffic.

Exhibit A. Scope and Methodology 16

Using the same groups and statistical methods, we examined seven safety events

(see exhibit E) and reported on the following:

• Numbers of risk analysis events involving an air traffic controller. A risk

analysis event is a loss of standard separation between aircraft that

requires analysis to determine severity and chance of reoccurrence;

• Numbers of surface risk analysis events involving an air traffic controller. A

surface risk analysis event occurs when there is less than 6,000 feet of

separation between two aircraft on a runway or when an aircraft lands or

departs from a taxiway or closed runway;

• Numbers of FAA ATSAP and contractor SAFER-FCT reports. ATSAP and

SAFER-FCT reports are voluntary, non-punitive, confidential written

accounts of air traffic safety events; and

• Numbers of completed ECVs. An ECV is a safety review conducted onsite

by FAA’s Safety Quality Control Group. ECVs are conducted based on risks

associated with reported events and other quality control data along with

observed trends.

We obtained the safety data from FAA’s Office of Safety and Technical Training

that included information from several FAA databases—Comprehensive

Electronic Data Analysis and Reporting (CEDAR),

Compliance Verification Tool

(CVT),

ATSAP, and SAFER-FCT. The risk analysis and surface risk analysis events

came from CEDAR, ECV data came from CVT, and the voluntary safety reports for

FAA and contract towers came from separate systems called ATSAP and SAFER-

FCT. Officials at FAA’s Office of Safety and Technical Training confirmed that the

Agency provided all available safety data to us and that all the information had

been either self-reported through internal databases or evaluated by ATO’s safety

program.

We identified the following limitations in our methodology:

1. Grouping methods, while widely used, are limited to available data and

inputs. Differences between towers may exist that affect towers’

operational costs and safety but are not fully captured by aircraft

operations and runway configurations.

CEDAR is a web-based, comprehensive data reporting, collection, and analysis tool used by both FAA quality

control and quality assurance.

CVT is a national database that contains checklists, reports, facility information, tracking information, response data,

and other statistical information related to FAA’s compliance verification process.

Exhibit A. Scope and Methodology 17

2. The average FAA tower handled more passengers than the average

contract tower within each group. We did not account for the numbers of

passengers handled by each tower.

3. Our cost differential calculations did not account for several factors

relevant primarily to FAA towers. First, most FAA towers use some

resources to train new air traffic controllers. Second, locality pay may be

more of a consideration for FAA towers, which tend to be located in more

heavily populated areas than comparable contract towers. Third, while

FAA approves staffing levels at contract towers, contract towers are

generally staffed at lower levels than FAA towers. FAA also has more

towers that operate overnight shifts, and requires overnight shifts to be

covered by at least two controllers.

4. Numbers of reported safety events were very low relative to overall

aircraft operations for all towers. Based on our methodology, contract

towers experience statistically fewer safety events per aircraft handled

than FAA towers. However, we believe that these differences are not

statistically meaningful due to the rarity of safety events.

Three of 248 contract towers operate 24 hours a day, 7 days a week (1.2 percent), while 26 of 103 FAA-operated

towers (25.2 percent) operate this way.

Exhibit B. Entities Visited or Contacted 18

Exhibit B. Entities Visited or Contacted

FAA Headquarters, Washington, DC

Air Traffic Organization (ATO)

• FAA Contract Tower Program Office

• Office of Safety and Technical Training

Office of Finance and Management

• Office of Financial Services

• Office of Labor Analysis

Policy, International Affairs and Environment

• Office of Aviation Policy and Plans

FAA Field Locations

• ATO, Office of Safety and Technical Training, Service Center, Quality

Control Group, College Park, Georgia and Fort Worth, Texas

FAA Air Traffic Control Towers

• Van Nuys, California (VNY)

• Pensacola International, Florida (PNS)

• Dekalb-Peachtree, Georgia (PDK)

• Lakefront, Louisiana (NEW)

• Manassas Regional/Harry P Davis Field, Virginia (HEF)

FAA Contract Towers

• Whiteman, California (WHP)

• Redding Muni, California (RDD)

Exhibit B. Entities Visited or Contacted 19

• Destin Executive, Florida (DTS)

• Northwest Florida Beaches International, Florida (ECP)

• Lewiston-Nez Perce County, Idaho (LWS)

• Idaho Falls Regional, Idaho (IDA)

• Frederick Muni, Maryland (FDK)

• Easton/Newnam Field, Maryland (ESN)

• Santa Fe Muni, New Mexico (SAF)

• Four Corners Regional, New Mexico (FMN)

• Lawrence J Timmerman, Wisconsin (MWC)

• Kenosha Regional, Wisconsin (ENW)

FCT Program Contractors

• Midwest Air Traffic Control Service, Inc.

• Robinson Aviation (RVA), Inc.

• Serco, Inc.

Other Organizations

• American Association of Airport Executives (AAAE)

• Aircraft Owners and Pilots Association (AOPA)

• National Air Transportation Association (NATA)

• National Air Traffic Controllers Association (NATCA)

Reviewers

• Michael Simeone, Director of Data Science and Analytics for Arizona State

University Libraries

• Jacob LaRiviere, Principal Researcher, Microsoft, Inc.

Exhibit C. List of Acronyms 20

Exhibit C. List of Acronyms

ATC air traffic controller

ATO Air Traffic Organization

ATSAP Air Traffic Safety Action Program

DOT Department of Transportation

ECV external compliance verification

FAA Federal Aviation Administration

FCT FAA contract tower

NAS National Airspace System

OIG Office of Inspector General

Exhibit D. Cost Comparisons 21

Exhibit D. Cost Comparisons

We consulted with FAA’s Office of Financial Operations to determine what

cost categories FAA tracks for both FAA operated and contract towers. We

used FAA financial data to compare and analyze the cost associated with

operating FAA and contract towers. We conducted three cost

comparison—air traffic controller (ATC) labor and benefits only; total costs

without overhead; and total costs including overhead.

In our ATC labor and benefits costs comparisons:

Cost categories included

FAA or FCT cost

FAA ATC labor

FAA

FAA ATC benefit liabilities

FAA

Contract controller labor

FCT

n our total costs without overhead comparisons:

Cost categories included

FAA or FCT Cost

FAA ATC labor

FAA

FAA ATC benefit liabilities

FAA

Contract controller labor

FCT

FAA ATC non-labor

FAA

Technical operations labor

Both

Technical operations benefit liabilities

Both

Technical operations non-labor

Both

FCT Insurance

FCT

Logistics

Both

Telecommunications

Both

Utilities

Both

Leases

Both

Note. Some FAA towers did not incur utilities and leases costs. Furthermore, our analysis

did not include costs incurred by airport sponsors. As a result, some contract towers did

not have technical operations labor, technical operations benefit liabilities, technical

operations non-labor, logistics, telecommunications, utilities, and leases costs.

Exhibit D. Cost Comparisons 22

In our total costs including overhead comparisons:

Cost categories included

FAA or FCT Cost

FAA ATC labor

FAA

FAA ATC benefit liabilities

FAA

Contract controller labor

FCT

FAA ATC non-labor

FAA

Technical operations labor

Both

Technical operations benefit liabilities

Both

Technical operations non-labor

Both

FCT insurance

FCT

Logistics

Both

telecommunications

Both

Utilities

Both

Leases

Both

FAA Headquarters indirect support (overhead)

Both

ATO indirect support (overhead)

Both

Note. Some FAA towers did not incur utilities and leases costs. Furthermore, our analysis

did not include costs incurred by airport sponsors. As a result, some contract towers did

not have technical operations labor, technical operations benefit liabilities, technical

operations non-labor, logistics, telecommunications, utilities, and leases costs. Lastly,

contract towers that are airport sponsored, owned, and maintained had no ATO indirect

support costs.

Descriptions of Cost Categories

1. FAA ATC labor. Includes the compensation and benefits of personnel

directly associated with providing ATC services at terminal locations.

2. FAA ATC benefit liabilities. FAA recognizes the full cost of pensions

and other retirement benefits during an employee’s active years of

service through a combination of costs financed by FAA’s

appropriations and imputed costs. Although it is not funded by FAA’s

annual operations appropriation, the liability represents a cost to the

Federal Government. The portion of unfunded benefit costs for air

traffic controllers is computed for each facility, and is therefore directly

traceable.

Exhibit D. Cost Comparisons 23

3. Contract controller labor. Includes contract labor cost associated with

ATC services at contract tower locations only. Controller labor costs

paid by FAA are collected from invoices submitted by vendors for air

traffic services provided during the year. These invoices may include

wage determination adjustments for the last quarter of the immediate

prior year if known at the time cost is compiled.

4. FAA ATC non-labor. Includes costs such as supplies and travel

expenses that are directly associated with a specific ATC activity at each

terminal location. The cost of ATC non-labor is directly traceable to the

ATC service using the actual cost for each facility.

5. Technical operations’ labor. Includes the compensation and benefits

of personnel that are directly associated with providing maintenance of

electronics and equipment used in direct support of providing ATC

service and located within the tower.

6. Technical operations’ benefit liabilities. FAA recognizes the full cost

of pensions and other retirement benefits during an employee’s active

years of service through a combination of costs financed by FAA’s

appropriations and imputed costs. The portion of unfunded benefit

costs for technicians is allocated pro rata based on labor charges to all

projects, such as equipment and locations, supported by each

maintenance organization since technicians may be responsible for

maintaining multiple types of equipment at multiple facilities.

7. Technical operations’ non-labor. Includes costs such as guard

services, ground services, and other services that are directly associated

with ATC services. The cost of technical operations’ non-labor is directly

traced to the equipment or facility at each tower location using the

actual cost for each facility.

8. FCT insurance. Acknowledging the devastating effects associated with

aircraft accidents, FAA provides aircraft accident liability insurance

coverage at each of the contract tower locations. FAA has established a

$10,000,000.00 per occurrence amount for each FCT location. FAA

procures the brokerage services necessary to obtain the aviation

liability insurance. The total cost of FCT insurance premiums paid

annually and captured in FAA’s accounting system is allocated evenly

to all contract towers. The total premiums paid may vary from year to

year depending on insurance adjustments and timing of payments.

9. Logistics. FAA’s Logistics Center is a large depot responsible for

maintaining stocks and stores of spare parts for issuance to the field,

performing facility refurbishments, and providing on-site repair

services. Logistics cost are obtained from raw supply data in the

Exhibit D. Cost Comparisons 24

Logistics Center Support System (LCSS). LCSS is used to track the

issuance of equipment from the Logistics Center to the field, including

part information, cost data on the part requisitioned, quantities, the

requisitioning cost center, and related customer information. These

costs are only included if directly traceable to equipment at the tower

associated with providing ATC service.

10. Telecommunications. The cost of leased telecommunications lines to

and from tower facilities, including all activities associated with

maintaining, upgrading, or modifying operational and administrative

communications services required for tower facilities; includes both

costs from FAA’s telecommunications infrastructure program obtained

from the program’s Invoice and Financial Management System, and

telecommunications costs related to agreements with the Defense

Information Systems Agency. The cost of telecommunications is based

on actual cost directly traceable to each facility. Airport-sponsor

incurred telecommunication costs are not included.

11. Utilities. Include costs such as electricity, gas, and water. Utilities costs

are directly traceable to each facility in FAA’s accounting system. In

some cases, utilities may be included as part of a lease agreement or

may be provided by the airport sponsor. Airport-sponsor incurred

utilities costs are not included.

12. Leases. Include costs of operating leases. Capital leases are established

as assets and are included as depreciation. An operating lease is

defined as a lease that meets none of the criteria for a capital lease.

Operating lease payments are treated as charges to operating

expenses, and are not capitalized as assets. In most cases, these

payments are made using facilities and equipment funding.

Information on lease agreements is available from FAA’s Real Estate

Management System. Actual lease costs are directly traceable to each

facility. Airport-sponsor incurred lease costs are not included.

13. FAA Headquarters indirect support (overhead). FAA is organized

such that a majority of its general and administrative services are

provided by centralized organizations, referred to as FAA staff offices.

These organizations provide facility and personnel security; accounting

and budget services; human resources management; Government,

public, and industry affairs; legal and policy services; and executive

leadership. The agency wide indirect rate to recover Headquarters

indirect costs is 8.0 percent of all labor and non-labor operations and

maintenance expenses. This rate is reviewed annually and updated as

necessary.

Exhibit D. Cost Comparisons 25

14. ATO indirect support (overhead). ATO is organized with several

layers of management and oversight. It has several staff offices—

including the Chief Operating Officer, Management Services, Safety

and Technical Training, Joint Planning and Development, and Mission

Support Services—that provide support to air traffic and technical

operations service units. Each service unit provides oversight and

support to its service areas. Each service area provides oversight and

support for the services provided by the district offices it covers. Each

district office provides oversight and support to facilities. The rate for

recovering ATO indirect costs is 11.0 percent of all labor operations

and management expenses. The rate is reviewed annually and updated

as necessary.

15. FAA ATC contract training. Includes the cost of on-site training

provided under contract at select locations based on controllers’

specialized needs. A pro rata allocation is performed to assign these

training costs to each terminal facility based on the actual amount of

training hours invoiced by the training contractor at facility. These costs

are only relevant to FAA towers.

16. FAA Academy training. The FAA Academy, located in Oklahoma City,

is a large training facility that provides agency wide training services.

The cost of ATC and technical operations training at the Academy

includes costs for course development and delivery, and associated

employee travel and per diem costs. The Academy maintains

attendance records with detailed travel and course information. The

portion of travel costs assigned to each tower facility is determined

based on students’ travel records. These costs are only relevant to FAA

towers.

17. FAA ATC medical. Air traffic controllers receive regular medical exams

and drug and alcohol tests. An organization within FAA’s Office of

Aviation Medicine—which resides within the Aviation Safety line of

business—funds, conducts, and manages these exams and tests.

Medical costs are allocated pro rata to all terminal and en route

facilities based on personnel compensation and benefits cost. The

current methodology does not allocate cost to FAA contract towers

because these facilities do not have personnel costs and do not benefit

from the medical program. These costs are only relevant to FAA towers.

Note: We excluded three costs from our comparisons—FAA ATC contract

training, FAA Academy training, and FAA ATC medical—because these

cost categories apply only to FAA operated facilities.

Exhibit E. Comparisons of Safety Events 26

Exhibit E. Comparisons of Safety Events

Safety is extremely difficult to measure. In order to conduct appropriate

comparisons, we sought input from FAA and industry stakeholders to help

identify safety events. Based on discussions with FAA officials and industry

stakeholders, we compared the numbers of the following safety events that

occurred between fiscal years 2015 and 2018.

1. Risk analysis events involving an air traffic controller. A risk analysis

event is a loss of standard separation that has a measure of compliance of

less than 66 percent. According to an FAA official, a risk analysis event can

also be defined as an event that undergoes deeper analysis to determine

severity and repeatability factors associated with an individual occurrence.

2. Surface risk analysis events involving an air traffic controller. A

surface risk analysis event occurs when there is less than 6,000 feet

separation on a runway between two aircraft or when an aircraft lands or

departs from a taxiway or closed runway.

3. Mandatory occurrence reports by type at FAA and contract towers. A

mandatory occurrence report is a self-initiated safety report by a staff

person at an FAA or contract tower on an occurrence involving

airborne loss of separation,

airport surface loss of separation,

terrain/obstruction,

airborne air traffic control anomaly (airspace/altitude/route/speed) not

involving a loss of separation,

airport environment,

communication,

emergency or in-flight hazard,

inquiry, or

technical operations.

4. Possibly significant events by type at FAA and contract towers. A

significant event is any event in the NAS that: attracts media attention—

regional, national, or both—or political attention—regional or nation or

both; involves aircraft proximity with less than 33 percent of the standard;

Exhibit E. Comparisons of Safety Events 27

involves a report of a near midair collision with evasive action; or requires

immediate notification to the FAA service area or Headquarters.

5. Covered event reviews at FAA and contract towers. A covered event

review (CER) is used to supplement and document the air traffic services

rendered during an aircraft accident. A CER requires service delivery

points to review each accident in depth, looking beyond areas of

individual performance. A CER also includes a review of all aspects of

service (individual performance, equipment issues, weather, etc.) and

identifies any issues that cannot be ruled out as having contributed to the

accident.

6. FAA Air Traffic Safety Action Program (ATSAP) and contractor SAFER

FCT reports. ATSAP and SAFER-FCT reports are voluntary, non-punitive,

confidential, written accounts of air traffic safety events.

7. Completed external compliance verifications (ECV). An ECV is a safety

review conducted onsite by FAA’s Safety Quality Control Group based on

risks associated with a reported event and other quality control data

along with observed trends.

Exhibit F. List of 248 Contract Towers 28

Exhibit F. List of 248 Contract Towers

Count State Airport Name Tower ID

Bethel

BET

Kenai Muni

ENA

King Salmon

AKN

Kodiak

ADQ

Mobile Downtown

BFM

Dothan Regional

DHN

Tuscaloosa Regional

TCL

Drake Field

FYV

Northwest Arkansas Regional

XNA

Rogers Executive-Carter Field

ROG

Springdale Muni

ASG

Texarkana Regional/Webb Field

TXK

Chandler Muni

CHD

Flagstaff Pulliam

FLG

Glendale Muni

GEU

Phoenix Goodyear

GYR

Laughlin/Bullhead International

IFP

Ryan Field

RYN

Phoenix-Mesa Gateway

IWA

Castle

MER

Chico Muni

CIC

Fullerton Muni

FUL

Jack Northrop Field/Hawthorne Muni

HHR

Sacramento Mather

MHR

Modesto City-Co-Harry Sham Field

MOD

Oxnard

OXR

Palmdale USAF Plant 42

PMD

Ramona

RNM

Redding Muni

RDD

Riverside Muni

RAL

Sacramento Executive

SAC

Salinas Municipal

SNS

San Carlos

SQL

Brown Field Muni

SDM

San Luis County Regional

SBP

Santa Maria Pub/Capt G Allan Hancock Field

SMX

Southern California Logistics

VCV

Exhibit F. List of 248 Contract Towers 29

Count State Airport Name Tower ID

Whiteman

WHP

General William J. Fox Airfield

WJF

Eagle County Regional

EGE

Front Range

FTG

Grand Junction Regional

GJT

Igor I Sikorsky Memorial

BDR

Danbury Muni

DXR

Groton- New London

GON

Hartford-Brainard

HFD

Tweed-New Haven

HVN

Waterbury-Oxford

OXC

Albert Whitted

SPG

Boca Raton

BCT

Brooksville- Tampa Bay Regional

BKV

Cecil

VQQ

Destin Executive

DTS

Flagler Executive

FIN

Gainesville Regional

GNV

North Perry

HWO

Jacksonville Executive at Craig

CRG

Key West International

EYW

Kissimmee Gateway

ISM

Lakeland Linder Regional

LAL

Leesburg International

LEE

Melbourne International

MLB

Naples Muni

APF

New Smyrna Beach Muni

EVB

Ocala International-Jim Taylor Field

OCF

Miami-Opa Locka Executive

OPF

Ormond Beach Muni

OMN

Page Field

FMY

Northwest Florida Beaches International

ECP

Pompano Beach Airpark

PMP

Punta Gorda

PGD

Northeast Florida Regional

SGJ

Witham Field

SUA

Space Coast Regional

TIX

Athens/Ben Epps

AHN

Fulton County Airport-Brown Field

FTY

Gwinnett County-Briscoe Field

LZU

Exhibit F. List of 248 Contract Towers 30

Count State Airport Name Tower ID

Middle Georgia Regional

MCN

Cobb County International-McCollum Field

RYY

Southwest Georgia Regional

ABY

Guam International

GUM

Ellison Onizuka Kona International at Keahole

KOA

Lihue

LIH

Molokai

MKK

Dubuque

DBQ

Friedman Memorial

SUN

Idaho Falls Regional

IDA

Lewiston-Nez Perce County

LWS

Pocatello Regional

PIH

Veterans Airport of Southern Illinois

MWA

Waukegan National

UGN

Central IL Regional Airport at Bloomington/Normal

BMI

Decatur

DEC

Southern Illinois

MDH

St. Louis Regional

ALN

Monroe County

BMG

Columbus Muni

BAK

Gary/Chicago International

GYY

Delaware County Regional

MIE

100

Topeka Regional

FOE

101

Garden City Regional

GCK

102

Hutchinson Regional

HUT

103

Johnson County Executive

OJC

104

Manhattan Regional

MHK

105

New Century Aircenter

IXD

106

Philip Billard Muni

TOP

107

Salina Regional

SLN

108

Barkley Regional

PAH

109

Owensboro/Daviess County Regional

OWB

110

Acadiana Regional

ARA

111

Chennault International

CWF

112

Houma-Terreborne

HUM

113

Shreveport-Downtown

DTN

114

Westfield-Barnes Regional

BAF

115

Beverly Regional

BVY

116

Barnstable Muni-Boardman/Polando Field

HYA

117

Lawrence Muni

LWM

Exhibit F. List of 248 Contract Towers 31

Count State Airport Name Tower ID

118

Martha's Vineyard

MVY

119

New Bedford Regional

EWB

120

Norwood Memorial

OWD

121

Worcester Regional

ORH

122

Easton/Newnam Field

ESN

123

Frederick Muni

FDK

124

Martin State

MTN

125

Salisbury-Ocean City Wicomico Regional

SBY

126

Hagerstown Regional-Richard A Henson Field

HGR

127

W K Kellogg

BTL

128

Coleman A Young Muni

DET

129

Jackson County- Reynolds Field

JXN

130

Sawyer International

SAW

131

Anoka County-Blaine (Janes Field)

ANE

132

St. Cloud Regional

STC

133

Branson

BBG

134

Columbia Regional

COU

135

Jefferson City Memorial

JEF

136

Joplin Regional

JLN

137

Francisco C Ada/Saipan International

GSN

138

Golden Triangle Regional

GTR

139

Greenville Mid-Delta

GLH

140

Hawkins Field

HKS

141

Olive Branch

OLV

142

Stennis International

HSA

143

Tupelo Regional

TUP

144

Bozeman Yellowstone International

BZN

145

Glacier Park International

GPI

146

Missoula International

MSO

147

Concord Regional

JQF

148

Hickory Regional

HKY

149

Kinston Regional Jetport at Stallings Field

ISO

150

Coastal Carolina Regional

EWN

151

Smith Reynolds

INT

152

Minot International

MOT

153

Central Nebraska Regional

GRI

154

Boire Field

ASH

155

Lebanon Muni

LEB

156

Trenton Mercer

TTN

157

Double Eagle II

AEG

Exhibit F. List of 248 Contract Towers 32

Count State Airport Name Tower ID

158

Four Corners Regional

FMN

159

Lea County Regional

HOB

160

Santa Fe Muni

SAF

161

Henderson Executive

HND

162

Francis S. Gabreski

FOK

163

Niagara Falls International

IAG

164

Griffiss International

RME

165

Stewart International

SWF

166

Ithaca Tompkins Regional

ITH

167

Burke Lakefront

BKL

168

Cincinnati Muni Airport Lunken Field

LUK

169

Bolton Field

TZR

170

Cuyahoga County

CGF

171

Ohio State University

OSU

172

Ardmore Muni

ADM

173

Enid Woodring Regional

WDG

174

Lawton-Fort Sill Regional

LAW

175

Stillwater Regional

SWO

176

Univ of Oklahoma Westheimer

OUN

177

Wiley Post

PWA

178

Aurora State

UAO

179

Eastern Oregon Regional at Pendleton

PDT

180

McNary Field

SLE

181

Rogue Valley International-Medford

MFR

182

Roberts Field

RDM

183

Southwest Oregon Regional

OTH

184

Portland-Troutdale

TTD

185

Lancaster

LNS

186

University Park

UNV

187

Williamsport Regional

IPT

188

Arnold Palmer Regional

LBE

189

Capital City

CXY

190

Fernando Luis Ribas Dominicci

SIG

191

Rafael Hernandez

BQN

192

Donaldson Field

GYH

193

Grand Strand

CRE

194

Greenville Downtown

GMU

195

Hilton Head

HXD

196

Rapid City Regional

RAP

197

McKeller-Sipes Regional

MKL

Exhibit F. List of 248 Contract Towers 33

Count State Airport Name Tower ID

198

Millington Regional Jetport

NQA

199

Smyrna

MQY

200

Arlington Municipal

GKY

201

Brownsville/South Padre Island International

BRO

202

Denton Enterprise

DTO

203

Easterwood Field

CLL

204

Fort Worth-Spinks

FWS

205

Georgetown Muni

GTU

206

Grand Prairie Muni

GPM

207

Laredo International

LRD

208

Conroe-North Houston Regional

CXO

209

McAllen Miller International

MFE

210

McKinney National

TKI

211

Mesquite Metro

HQZ

212

New Braunfels Regional

BAZ

213

Dallas Executive

RBD

214

Valley International

HRL

215

San Angelo Regional/Mathis Field

SJT

216

San Marcos Regional

HYI

217

Scholes International at Galveston

GLS

218

Stinson Muni

SSF

219

Sugar Land Regional

SGR

220

Tyler Pounds Regional

TYR

221

Victoria Regional

VCT

222

TSTC Waco

CNW

223

Ogden-Hinckley

OGD

224

Provo Muni

PVU

225

Charlottesville-Albemarle

CHO

226

Lynchburg Regional/Preston Glenn Field

LYH

227

Henry E Rohlsen

STX

228

Bellingham International

BLI

229

Felts Field

SFF

230

Olympia Regional

OLM

231

Renton Muni

RNT

232

Tacoma Narrows

TIW

233

Walla Walla Regional

ALW

234

Yakima Air Terminal/McAllister Field

YKM

235

Appleton International

ATW

236

Central Wisconsin

CWA

237

Chippewa Valley Regional

EAU

Exhibit F. List of 248 Contract Towers 34

Count State Airport Name Tower ID

238

Kenosha Regional

ENW

239

La Crosse Regional

LSE

240

Southern Wisconsin Regional

JVL

241

Lawrence J. Timmerman

MWC

242

Waukesha County

UES

243

Wittman Regional

OSH

244

Greenbrier Valley

LWB

245

Morgantown Muni-Walter L Bill Hart Field

MGW

246

Mid-Ohio Valley Regional

PKB

247

Wheeling Ohio Co

HLG

248

Jackson Hole

JAC

Source: FAA

Exhibit G. List of 103 FAA-Operated Towers 35

Exhibit G. List of 103 FAA-Operated Towers

Count State Airport Name Tower ID

Ted Stevens Anchorage International

ANC

Juneau International

JNU

Merrill Field

MRI

Falcon Field

FFZ

Grand Canyon National Park

GCN

Phoenix Deer Valley

DVT

Ernest A Love Field

PRC

Scottsdale

SDL

Tucson International

TUS

Brackett Field

POC

Bob Hope

BUR

Camarillo

CMA

Chino

CNO

Buchanan Field

CCR

San Gabriel Valley

EMT

Gillespie Field

SEE

Hayward Executive

HWD

John Wayne Airport- Orange County

SNA

Livermore Muni

LVK

Long Beach/Daugherty Field

LGB

Monterey Regional

MRY

Montgomery- Gibbs Executive

MYF

Napa County

APC

Metropolitan Oakland International

OAK

Ontario International

ONT

Palm Springs International

PSP

Palo Alto

PAO

McClellan-Palomar

CRQ

Reid-Hillview of Santa Clara County

RHV

Sacramento International

SMF

Norman Y Mineta San Jose International

SJC

Santa Monica Muni

SMO

Charles M Schulz-Sonoma County

STS

Stockton Metropolitan

SCK

Zamperini Field

TOA

Van Nuys

VNY

Centennial

APA

Exhibit G. List of 103 FAA-Operated Towers 36

Count State Airport Name Tower ID

Pueblo Memorial

PUB

Rocky Mountain Metropolitan

BJC

Bradley International

BDL

New Castle

ILG

Fort Lauderdale Executive

FXE

Fort Lauderdale/Hollywood International

FLL

Orlando Executive

ORL

Orlando Sanford International

SFB

Pensacola International

PNS

Sarasota/Bradenton International

SRQ

Treasure Coast International

FPR

St Pete-Clearwater International

PIE

Miami Executive

TMB

Vero Beach Regional

VRB

Columbus

CSG

DeKalb - Peachtree

PDK

Kahului

OGG

Aurora Muni

ARR

Chicago Executive

PWK

St. Louis Downtown

CPS

DuPage

DPA

Chicago Midway International

MDW

Purdue University

LAF

Bowman Field

LOU

Lakefront

NEW

Laurence G Hanscom Field

BED

Nantucket Memorial

ACK

Joint Base Andrews

ADW

Ann Arbor Muni

ARB

Oakland County International

PTK

Cherry Capital

TVC

Willow Run

YIP

Crystal

MIC

Flying Cloud

FCM

St Paul Downtown Holman Field

STP

Charles B Wheeler Downtown

MKC

Spirit of St Louis

SUS

Eppley Airfield

OMA

Lincoln

LNK

Manchester

MHT

Exhibit G. List of 103 FAA-Operated Towers 37

Count State Airport Name Tower ID

Essex County

CDW

Morristown Muni

MMU

North Las Vegas

VGT

Reno/Tahoe International

RNO

Republic

FRG

Long Island Mac Arthur

ISP

Hudson Valley Regional

POU

Westchester County

HPN

James M Cox Dayton International

DAY

Richard Lloyd Jones Jr

RVS

Portland-Hillsboro

HIO

Portland International

PDX

Allegheny County

AGC

Northeast Philadelphia

PNE

Luis Munoz Marin International

SJU

Addison

ADS

Fort Worth Alliance

AFW

Jack Brooks Regional

BPT

David Wayne Hooks Memorial

DWH

Fort Worth Meacham International

FTW

Manassas Regional/Harry P Davis Field

HEF

Newport News/Williamsburg International

PHF

100

Richmond International

RIC

101

Cyril E King

STT

102

Boeing Field/King County International

BFI

103

Snohomish County (Paine Field)

PAE

Source: FAA

Exhibit H. Detailed Scope and Methodology 38

Exhibit H. Detailed Scope and Methodology

This exhibit presents the technical details of our scope and methodology. We

present our work in six sections. In the first three sections, we summarize selected

related research, describe the ATC towers included in the analyses, and detail our

data sources and preparation. In the methodology section, we discuss our use of

k-means and hierarchical clustering to sort the selected towers into groups of

comparable facilities, and compare the groups produced by the two methods. In

the results section, we present the hierarchical clustering findings and tests of the

significance of the cost and safety results under both clustering methods. Finally,

in the sensitivity analyses section we examine the effects of two specification

changes on the groupings and cost comparisons.

1. Related Research

Cluster analysis has been widely used to sort observations within datasets when

the researcher may not know, or want to impose, classifications. The technique is

especially useful when group classification depends on more than a single

characteristic. In the transportation literature, Kenneth D. Kuhn used cluster

analysis to select “days that are comparable in terms of the conditions faced

during air traffic flow management initiative planning.”

Another study applied

cluster analysis to a set of 42 airlines to explore the potential for groupings

beyond low-cost and full-service air carriers.

K-means clustering was also used

to find that airport hubs

fall into more than FAA’s three-tier classification.

Information supporting the choice of inputs into the analyses in this report came

from a 2003 FAA study in which a number of air and ground control staff at very

busy towers were interviewed on factors that contributed to complexity levels.

The top four factors were high traffic volume, active runway crossings, frequency

congestion, and aircraft differing in performance characteristics. Runway

configuration was sixth. We accounted for these factors in our cluster analyses.

https://daneshyari.com/article/preview/524716.pdf “A methodology for identifying similar days in air traffic flow

management initiative planning.” Transportation Research Part C 69 (2016) 1-15.

Urban, Marcia, et al. "Airline categorisation by applying the business model canvas and clustering

algorithms." Journal of Air Transport Management 71 (2018): 175-192.

FAA defines hub airports based on annual enplanements: large hubs handle 1 percent or more on annual U.S.

enplanements; medium hubs handle at least 0.25 percent, but less than 1 percent; and small hubs handle at least 0.05

percent, but less than 0.25 percent.

https://www.sciencedirect.com/science/article/abs/pii/S0966692313001099 Ryerson, Megan S., and Hyun Kim. "Integrating

airline operational practices into passenger airline hub definition." Journal of Transport Geography 31 (2013): 84-93.

Exhibit H. Detailed Scope and Methodology 39

2. Tower selection

We included nearly all FCTs while considering all plausibly comparable FAA

towers for fiscal years 2015 through 2018. This time period corresponds with the

most recent FCT contract authorization. We excluded from our data set of FCTs 6

towers operated by the National Guard, leaving us with 248.

We filtered the 264 FAA towers to include only towers that FAA classifies as ATC

level 8 or below.

We also excluded FAA towers co-located with terminal radar

approach control (TRACON)

facilities.

Because no contract tower is co-located

with a TRACON facility, co-located FAA towers are not comparable for staffing or

operational purposes. Lastly, since all contract towers operate in Class C or D

airspace, we excluded FAA towers that do not operate in Class C or D airspace,

and ended up with 103 FAA towers.

3. Data

All data in this report covered fiscal years 2015 through 2018. Data on tower

handled operation counts were downloaded from FAA’s OPSNET. OPSNET

defines and provides counts of five types of aircraft operations—air carrier, air

taxi, general aviation, military, and local.

Over the 4 years in the sample, the 103

FAA towers fitting our selection criteria handled 58 million operations from these

categories, while the 248 contract towers handled 60.4 million. Figure H-1 shows

the distribution of total operations handled for each tower type.

FAA provided data on operating hours, tower operating costs, runway

configurations, and airspace. The runway configurations include single runway,

parallel runways, crossing runways, and converging runways. Towers can control

aircraft on any of these configurations, or a combination of the three non-single

configurations—meaning a tower could have both parallel and crossing runways,

but not single and parallel runways.

The OPSNET data was combined with daily operating hours of each tower. Tower

operating hours can change with season, day of the week, or Daylight Savings

Time. We computed hourly averages using the sum of aircraft handled in each

Each FAA facility is classified as a level 4 through 12, based on numerous factors including traffic volume,

complexity and sustainability of traffic. The controllers working at higher level facilities receive greater compensation

due to the increased level of air traffic volume and complexity.

TRACON or Terminal Radar Approach Control facilities house FAA air traffic controllers who use radar displays and

radios to control aircraft approaching and departing airports generally within a 30- to 50-mile radius up to 10,000

feet, as well as aircraft that may be flying over that airspace.

See table 1 for definitions of these air traffic types.

Exhibit H. Detailed Scope and Methodology 40

category over the total hours operated during the 4-year analysis period. Figures

H-2 and H-3 show the distributions of hourly operations handled and average

daily operating hours by tower type. Towers in the dataset averaged 15.6 hours

of operation per day. In addition, 8.3 percent of towers were open 24 hours a day;

these towers comprised 25.2 percent of the FAA towers and 1.2 percent of the

FCTs. See table H-1 for summary statistics on the inputs used to cluster towers.

Cost and safety data came from requests to FAA. We focused our safety analyses

on risk analysis event (RAE) and surface risk analysis event (SRAE) investigation

data provided by FAA. We combined these data into a single measure, counting

all instances in which an air traffic controller was involved.

An FAA official

characterized these events as having undergone a deeper analysis to determine

severity and associated repeatability factors. These deeper analyses mitigate

some selection concerns because each event is chosen by staff at a higher level

than the controller. Furthermore, more detailed information was available on

whether ATC had played a role in these events.

4. Methodology

FAA and contract towers differ in a variety of ways, with the simplest being traffic

density—or aircraft handled per hour. To allow us to also account for additional

factors affecting air traffic control complexity, we used two clustering methods—

k-means and hierarchical—to sort FAA and contract towers into groups that were

comparable considering three types of characteristics. We grouped towers based

on the number of operations of different types of aircraft and type of runway

configuration. We did not allow the clustering methods to consider whether a

tower was an FCT or FAA tower, so group assignments depended only on tower

characteristics.

K-Means Clustering

he k-means clustering algorithm required us to set the number of clusters, k,

and then run an iterative process sorting the observations into k groups. A

randomized condition determined the initial allocation of towers across the

groups, and the mean or average of each group was then calculated. In the next

RAE documentation is required for any occurrences where less than 66 percent of the separation standard is

maintained. Documentation of an SRAE occurs when there is less than 6000 feet of separation on a runway between

two aircraft or any event in which an aircraft lands or departs from a taxiway or closed runway.

The goal of cluster analysis is to ascertain, on the basis of observed characteristics, whether the observations fall

into relatively distinct groups. The difficulty of determining the group to which each observation belongs depends on

whether the groups are well-separated or somewhat overlapping. For example, clustering methods could include only

FAA towers (or FCTs) in a group if they were not comparable to FCTs (FAA towers) in terms of the characteristics

considered.

Exhibit H. Detailed Scope and Methodology 41

step, observations were assigned to a new group if a distance formula

determined that they were closer to another group’s average than they were to

their currently assigned group’s average. Group means were recalculated and

observations reassigned until no observations changed groups.

Our inputs included both continuous and categorical data. Such mixed data

require the use of standardization or special distance metrics in the k-means

algorithm. Similar considerations can apply with continuous data when one input

to the algorithm is significantly larger in magnitude than the other inputs. We

clustered using Gower’s distance measure, which is appropriate for mixed data.

We also standardized our data by demeaning each variable and then dividing by

its standard deviation to prevent extreme values from exerting outsize influence

on the process.

For the k-means approach, we ran the clustering process setting the number of

groups, k, to range from 2 to 25. We then applied the elbow method to the

results to select the number of clusters or groups to use in the analyses. The

elbow method calculates the within-cluster sum of squares (WSS), plots the value

of WSS for each k, and identifies the optimal number of clusters as occurring

where there is an elbow or bend in the plot. The WSS measures compactness—

how similar observations assigned to the same cluster are to each other—and an

elbow in the plot indicates that improvements to compactness have slowed. We

further informed our choice by substituting two other measures for the WSS in

the elbow method—the logarithm of WSS and eta-squared. The latter measures

the proportional reduction of the WSS for each cluster solution, k, relative to the

total sum of squares (TSS), and is defined as

= 1 – (WSS(k)/TSS)

The solution to the k-means algorithm can be sensitive to its starting point. To

account for this, we applied the elbow method to results generated using a large

number of randomly selected starting points.

Specifically, we randomly selected

200 non-negative integers as starting points or seeds for the k-means algorithm.

For each seed, we implemented the k-means method and calculated the WSS, its

logarithm and eta-squared for k = {2, 3, … , 25}. Using this method, we found an

elbow at four clusters. Analogous plots for the logarithm of the WSS and eta-

squared also each had a kink at four clusters. In addition, we considered the

proportional reduction of error (PRE) coefficient, which measures the percentage

reduction in WSS when the number of clusters increases from k-1 to k; its plot

flattened out after k=4.

In doing so, we follow Makles, A. "Stata Tip 110: How to Get the Optimal k-means Cluster Solution." The Stata

Journal 12 (2012): 347-351.

Exhibit H. Detailed Scope and Methodology 42

Having determined four clusters to be optimal, we conducted further analyses to

determine which four cluster groupings to use. We generated 200 grouping

solutions for k=4 (one for each seed), and calculated the average number of

towers in each group. We then selected the solution whose number of towers for

each group was closest to the average. Thirty-four of the 200 solutions had

exactly the same minimum distance to the average group size.

We picked one

of the listed solutions for use in the cost and safety comparisons. Summary

statistics for the groups in the selected solution can be seen in tables H-2, H-3,

and H-4

Hierarchical Clustering

e also clustered towers using agglomerative hierarchical clustering. In

hierarchical clustering, towers are folded into increasingly larger groups based on

the similarity and dissimilarity of individual observations and groups, measured

using specified metrics. We used Ward’s linkage and Gower’s distance measure to

accommodate the mixed nature of our data. The merging is tracked in a graph

called a dendrogram,

and once the process is complete, the analyst can “cut”

the observations into groups at any of the agglomerative steps. The height of

each vertical line in the dendrogram of the clustering process corresponds to a

dissimilarity measure, with a larger vertical distance implying greater dissimilarity.

We cut the process at the level below which increases in dissimilarity dropped.

This produced four groups or clusters. Tables H-5, H-6, and H-7 report summary

statistics for the groups overall and by tower type.

We note that the hierarchical method clusters more strongly on runway

configuration than on traffic density, compared to the k-means method.

However, both methods grouped towers with single runways together. In both

cases, this group consisted of 12 FAA-staffed towers and 71 FCTs that averaged

nearly 14 operations per hour. While relationships between the other groups

produced by the two methods are not as obvious, table H-8 indicates

considerable overlap between them. It shows the distribution of towers across

groupings. Each number on the diagonal indicates the number of towers that

were sorted into the same group under the two methods. For example, table H-8

reports that 28 towers were sorted into Group 1 under both methods. The off-

diagonal numbers track the towers which the two methods sorted into different

We considered both Euclidean and Chebyshev’s distance functions, where each group is a “coordinate.” For

example, if the number of towers in groups 1 through 4 is (25, 50, 75, 100) for a candidate solution and the group

means across the two hundred solutions are (30, 40, 60, 105), the Chebyshev distance is the maximum difference

between a group’s size and its mean (in this case, it is 15 = 75-60).

A dendrogram is a tree diagram that links all observations according to the hierarchical clustering criteria.

Observations linked lower on the dendrogram are more similar to each other than those linked higher up. Larger

vertical distances on the figure represent more dissimilarity from one linkage to the next. Horizontal differences do

not have an interpretation or meaning.

Exhibit H. Detailed Scope and Methodology 43

groups. For example, 17 towers were sorted into Group 1 by the k-means method

that the hierarchical clustering sorted into Group 3.

5. Results

After creating the groups, we compared the costs and safety of FCT and FAA

towers within each group. In this section, we discuss the statistical assessment of

the comparisons, and present the hierarchical clustering results.

Cost Comparisons

sing hierarchical clustering, we again found that FCTs had lower costs per

aircraft handled than comparable FAA towers, as shown in table H-12. In addition,

the range of cost differences within groups was similar using both k-means and

hierarchical clustering methods. We tested the statistical significance of the cost

differences per aircraft using a pooled t-test within each group under both

clustering methods, and consistently found FAA tower costs were statistically

greater than FCTs’ costs at the 5 percent significance level. The test results were

the same whether we assumed equal or unequal variance of costs.

Safety Comparisons

he annual average safety event numbers in the hierarchical groups appear to be

affected by outliers. Table H-13 shows a strikingly large number of annual

average safety events for Group 1 FAA towers, and a high rate of such events for

Group 1 FCTs. In contrast, the incidence of reported safety events over the

analysis period as a whole is lower for FCTs than FAA towers.

We tested the statistical significance of the differences in total reported safety

events over fiscal years 2015 through 2018 using Poisson regression with a

variable representing exposure to the number of aircraft operations handled by

each tower.

The dependent variable was the sum total of SRAE and RAE events

over the analysis period, while the independent variable was an indicator variable

that equaled 1 if a tower was a contract tower. For all groups under both

clustering methods, the FCT coefficient was negative and significant at the 5

percent level.

We also used two alternative methods to assess the comparison. The first was

weighted linear regression, where the dependent variable was the number of

events per 100,000 aircraft handled, and the weights were counts of aircraft

handled. The second was logit regression, where the dependent variable was an

An exposure variable, in this context, consists of a covariate whose logarithm is used in regression with coefficient

fixed at 1. In count regression models, exposure variables ensure that observations are comparable in their unit of

measurement.

Exhibit H. Detailed Scope and Methodology 44

indicator variable that equals 1 if a tower had no safety risk events and the

controls were the FCT dummy variable and the logarithm of aircraft handled. The

weighted regression confirmed the results of the Poisson regression. The logit

regression revealed that in all cases, FCTs were statistically more likely to have

had no safety risk events, even when controlling for the number of aircraft

handled.

6. Sensitivity Analyses

We performed two sensitivity analyses on both the k-means and hierarchical

clustering. In the first, we added a measure of traffic density on the ninetieth

percentile busiest day of operations—a measure FAA uses to ensure that it has

sufficient staff at each tower.

For the second, we reduced the number of runway

configuration categories considered to two—single runway and other. We

followed the same steps detailed in exhibit A to generate groupings under the

new specifications.

The addition of the traffic density measure changed the clusters produced using

the k-means method to a limited extent, and had no impact on hierarchical

method groupings. Table H-9 shows that in the k-means case 83.2 percent of

towers fell into the same groups as in the original analysis.

The reduction in the runway configuration categories changed the groupings so

that 66.4 percent, under the k-means method, and 63.8 percent, under the

hierarchical method, ended up in the same groups as in the original analyses, as

indicated in tables H-10 and H-11. The k-means clustering now concentrates

carrier operations in one group—which includes five FCTs, as shown in tables H-

16 and H-17. Under the hierarchical method, the optimal number of clusters

drops to three. The “other’ runway towers with low carrier traffic, which had been

split between two groups in the original analysis, are now mainly gathered into a

single group. Tables H-18 and H-19 report statistics for the new hierarchical

method clusters. Recalculating comparative labor costs using the new groupings,

we found a similar range of differences in costs between the two tower types as

we found earlier. Tables H-14 and H-15 report these results.

The ninetieth percentile busiest day is defined by sorting all days based on tower operations and choosing the day

with operations busier than 90 percent but less than 10 percent of all the other days in the analysis period. The

number of aircraft handled on that day is used to calculate traffic density instead of an average for all days for this

alternative analysis.

The selected grouping for each specification under the k-means clustering used a different seed.

Exhibit H. Detailed Scope and Methodology 45

Table H-1. Summary Statistics of Inputs into the Clustering Algorithms

1. Aircraft Type Handled

Average

Standard Deviation

Min

Max

Carrier

25551.81

94093.12

914102

General Aviation

137946.2

106812.2

13147

637552

Taxi

38930.11

57921.51

403933

Military

17622.29

35961.62

186

363009

Local Civil

117260.6

133922.5

967342

Aircraft Handled Total

337311.1

254101.6

26833

1542924

2. FAA Aircraft Type Handled

Average

Standard Deviation

Min

Max

Carrier

74023.91

162798.1

914102

General Aviation

218660.4

135228.7

13147

637552

Taxi

66292.53

81237.03

1804

403933

Military

16387.26

44771.62

555

363009

Local Civil

187854.9

184622.4

967342

Aircraft Handled Total

563219

306971.4

86552

1542924

3. FCT Aircraft Type Handled

Average

Standard Deviation

Min

Max

Carrier

5420.254

14633.1

114114

General Aviation

104423.8

69063.51

15519

347718

Taxi

27565.88

39834

311408

Military

18135.22

31683.67

186

222974

Local Civil

87941.25

91636.59

486

570656

Aircraft Handled Total

243486.4

149863.1

26833

1055970

4. Runway Configuration

Total

Proportion Of Towers

FAA

FCT

Single

0.24

Parallel

0.25

Crossing

200

0.57

140

Converging

0.21

Exhibit H. Detailed Scope and Methodology 46

Table H-2. Tower Groupings – K-Means

Group FAA FCT

1 29 17

2 17 44

3 12 71

4 45 116

Total

103

248

Exhibit H. Detailed Scope and Methodology 47

Table H-3. K-Means Groups Summary Statistics

Group

Aircraft Operations per Tower per Hour

Runway Configuration

Group 1

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

22.46

2.82

8.27

2.40

0.96

8.00

0.00

1.00

0.00

0.26

Std Dev

5.66

2.91

1.72

8.23

0.00

0.44

Minimum

0.000

0.69

0.05

0.03

0.00

1.00

0.00

Maximum

26.07

27.27

15.07

8.22

36.78

0.00

1.00

0.00

1.00

Group 2

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

14.66

0.97

5.70

1.68

1.00

5.32

0.00

0.25

0.64

1.00

Std Dev

2.94

3.96

2.37

1.64

6.36

0.00

0.43

0.48

0.00

Min

0.00

1.21

0.00

0.01

0.00

1.00

Max

20.42

18.38

13.40

8.24

29.34

0.00

1.00

Group 3

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

13.77

0.38

6.18

1.66

0.66

4.88

1.00

0.00

Std Dev

0.82

4.67

2.92

0.96

4.83

0.00

Min

0.00

0.71

0.02

0.01

1.00

0.00

Max

4.92

25.29

21.82

3.98

18.94

1.00

0.00

Group 4

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

13.03

0.50

5.65

1.42

0.70

4.75

0.00

0.17

1.00

0.00

Std Dev

1.98

4.09

2.10

1.68

5.25

0.00

0.38

0.00

Min

0.00

0.67

0.00

0.01

0.00

1.00

0.00

Max

21.21

20.84

14.66

15.53

28.16

0.00

1.00

0.00

Exhibit H. Detailed Scope and Methodology 48

Table H-4. K-Means Groups Operations per Tower per Hour and Runway Configurations by Tower Type

Group 1

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.25

5.85

2.51

1.53

5.99

0.00

1.00

0.00

0.24

Std Dev

0.47

4.31

3.62

2.27

7.13

0.00

0.44

Min

0.00

1.45

0.05

0.03

0.26

0.00

1.00

0.00

Max

1.61

14.89

15.07

8.22

25.07

0.00

1.00

0.00

1.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

4.34

9.69

2.34

0.62

9.19

0.00

1.00

0.00

0.28

Std Dev

6.71

5.94

2.47

1.23

8.71

0.00

0.45

Min

0.00

0.69

0.10

0.03

0.00

1.00

0.00

Max

26.07

27.27

10.65

6.19

36.78

0.00

1.00

0.00

1.00

Group 2

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.31

5.33

1.30

1.16

4.83

0.00

0.14

0.55

1.00

Std Dev

0.76

3.60

2.10

1.86

5.14

0.00

0.35

0.50

0.00

Min

0.00

1.21

0.00

0.01

0.21

0.00

1.00

Max

4.60

16.36

13.40

8.24

27.19

0.00

1.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

2.68

6.65

2.64

0.59

6.57

0.00

0.53

0.88

1.00

Std Dev

5.16

4.74

2.80

0.71

8.86

0.00

0.51

0.33

0.00

Min

0.00

1.32

0.59

0.04

0.01

0.00

1.00

Max

20.42

18.38

10.68

2.47

29.34

0.00

1.00

Exhibit H. Detailed Scope and Methodology 49

Group 3

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.37

5.15

1.17

0.74

4.34

1.00

0.00

Std Dev

0.85

3.13

1.16

1.02

4.30

0.00

Min

0.00

0.77

0.02

0.01

1.00

0.00

Max

4.92

14.10

5.53

3.98

16.81

1.00

0.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.44

12.27

4.60

0.22

8.07

1.00

0.00

Std Dev

0.66

7.28

6.62

0.26

6.55

0.00

Min

0.00

0.71

0.10

0.05

0.21

1.00

0.00

Max

1.69

25.29

21.82

1.01

18.94

1.00

0.00

Group 4

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.11

4.55

1.15

0.69

3.55

0.00

0.09

1.00

0.00

Std Dev

0.25

3.08

1.68

1.29

3.45

0.00

0.29

0.00

Min

0.00

0.67

0.00

0.01

0.14

0.00

1.00

0.00

Max

1.36

16.64

10.57

9.55

21.28

0.00

1.00

0.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

1.50

8.50

2.11

0.74

7.84

0.00

0.38

1.00

0.00

Std Dev

3.56

4.96

2.82

2.43

7.45

0.00

0.49

0.00

Min

0.00

1.60

0.10

0.02

0.00

1.00

0.00

Max

21.21

20.84

14.66

15.53

28.16

0.00

1.00

0.00

Exhibit H. Detailed Scope and Methodology 50

Table H-5. Tower Groupings - Hierarchical

Group FAA FCT

1 55 34

2 12 71

3 8 38

4 28 105

Total

103

248

Exhibit H. Detailed Scope and Methodology 51

Table H-6. Hierarchical Groups Summary Statistics

Group

Aircraft Operations Handled by Tower per Hour

Runway Configuration

Group 1

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

22.29

2.26

8.11

2.06

1.08

8.78

0.00

1.00

0.48

0.30

Std Dev

5.14

5.59

2.67

2.32

8.55

0.00

0.50

0.46

Minimum

0.00

0.69

0.00

0.03

0.00

1.00

0.00

Maximum

26.07

27.27

15.07

15.53

36.78

0.00

1.00

Group 2

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

13.77

0.38

6.18

1.66

0.66

4.88

1.00

0.00

Std Dev

0.82

4.67

2.92

0.96

4.83

0.00

Min

0.00

0.71

0.02

0.01

1.00

0.00

Max

4.92

25.29

21.82

3.98

18.94

1.00

0.00

Group 3

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

12.37

0.66

5.07

1.67

1.09

3.88

0.00

0.52

1.00

Std Dev

1.67

3.21

2.67

1.70

4.02

0.00

0.51

0.00

Min

0.00

1.21

0.00

0.01

0.00

1.00

Max

7.53

12.05

13.40

8.24

18.15

0.00

1.00

Group 4

Total

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

11.07

0.29

5.14

1.36

0.54

3.74

0.00

1.00

0.00

Std Dev

0.96

3.45

1.92

0.98

3.50

0.00

Min

0.00

0.67

0.00

0.01

0.14

0.00

1.00

0.00

Max

8.55

17.25

11.14

8.05

21.28

0.00

1.00

0.00

Exhibit H. Detailed Scope and Methodology 52

Table H-7. Hierarchical Groups Operations per Tower per Hour and Runway Configurations by Tower Type

Group 1

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.25

5.39

1.94

1.58

6.00

0.00

1.00

0.50

0.29

Std Dev

0.46

3.96

2.77

2.39

6.59

0.00

0.51

0.46

Max

0.00

1.06

0.00

0.03

0.26

0.00

1.00

0.00

Min

1.68

16.36

15.07

9.55

27.19

0.00

1.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

3.49

9.80

2.14

0.77

10.50

0.00

1.00

0.47

0.31

Std Dev

6.23

5.81

2.63

2.24

9.21

0.00

0.50

0.47

Max

0.00

0.69

0.10

0.03

0.00

1.00

0.00

Min

26.07

27.27

14.66

15.53

36.78

0.00

1.00

Group 2

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.37

5.15

1.17

0.74

4.34

1.00

0.00

Std Dev

0.85

3.13

1.16

1.02

4.30

0.00

Max

0.00

0.77

0.02

0.01

1.00

0.00

Min

4.92

14.10

5.53

3.98

16.81

1.00

0.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.44

12.27

4.60

0.22

8.07

1.00

0.00

Std Dev

0.66

7.28

6.62

0.26

6.55

0.00

Max

0.00

0.71

0.10

0.05

0.21

1.00

0.00

Min

1.69

25.29

21.82

1.01

18.94

1.00

0.00

Exhibit H. Detailed Scope and Methodology 53

Group 3

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.29

5.14

1.19

1.16

4.24

0.00

0.47

1.00

Std Dev

0.78

3.32

2.20

1.85

4.08

0.00

0.51

0.00

Max

0.00

1.21

0.00

0.01

0.21

0.00

1.00

Min

4.60

12.05

13.40

8.24

18.15

0.00

1.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

2.38

4.71

3.97

0.79

2.16

0.00

0.75

1.00

Std Dev

3.25

2.74

3.59

0.60

3.42

0.00

0.46

0.00

Max

0.03

1.92

1.25

0.13

0.01

0.00

1.00

Min

7.53

10.56

10.68

1.96

10.27

0.00

1.00

Group 4

FCT

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.11

4.60

1.17

0.56

3.44

0.00

1.00

0.00

Std Dev

0.24

3.16

1.71

0.94

3.43

0.00

Max

0.00

0.67

0.00

0.01

0.14

0.00

1.00

0.00

Min

1.36

16.64

10.57

8.05

21.28

0.00

1.00

0.00

FAA

Carrier

General Aviation

Taxi

Military

Local Civil

Single

Parallel

Crossing

Converging

Average

0.98

7.15

2.09

0.45

4.87

0.00

1.00

0.00

Std Dev

1.92

3.82

2.45

1.15

3.58

0.00

Max

0.00

1.60

0.17

0.02

0.32

0.00

1.00

0.00

Min

8.55

17.25

11.14

5.19

16.28

0.00

1.00

0.00

Exhibit H. Detailed Scope and Methodology 54

Table H-8. Comparison K-Means and Hierarchical Clusters

Note: Each count on a diagonal is the number of towers sorted into the same group under both

methods. Percent of all towers in the same groups is 87.7.

Group numbering is not aligned across

appendix tables.

Table H-9. Comparison K-Means Clusters With and Without 90th

Percentile Day Traffic Density

Note: Each count on a diagonal is the number of towers sorted into the same group under both

methods. Percent of all towers in the same groups is 83.2. Group numbering is not aligned across

the tables in exhibit H.

See footnote 36 for an explanation of ninetieth percentile day traffic density.

Exhibit H. Detailed Scope and Methodology 55

Table H-10. Comparison K-Means Clusters with Two Runway

Configuration Categories

Note: Each count on a diagonal is the number of towers sorted into the same group under both

methods. Percent of all towers in the same groups is 66.4 Group numbering is not aligned across

the tables in exhibit H.

Table H-11. Comparison Hierarchical Clusters with Two Runway

Configuration Categories

Note: Each count on a diagonal is the number of towers sorted into the same group under both

methods. Percent of all towers in the same groups is 63.8 Group numbering is not aligned across

the tables in exhibit H.

Exhibit H. Detailed Scope and Methodology 56

Table H-12. Cost Comparisons by Hierarchical Groups

Costs Group 1 Group 2 Group 3 Group 4

Average total FAA labor and benefits $14,874,192.24 $10,817,413.79 $14,233,1

42.75

$10,498,013.02

Average FAA labor and benefits cost per aircraft handled $21.66 $19.93

$36.87

$27.62

Average total contract tower labor and benefits $2,648,566.54

$2,304,554.78

$2,336,412.68

$2,293,368.56

Average contract tower labor cost per aircraft handled $7.84 $9.48

$9.09

$11.01

Average percent fewer resources used by comparable

contract towers

63.81% 52.43% 75.35% 60.13%

Table H-13. Summary of Risk Analysis and Surface Risk Analysis Events involving an Air Traffic

Controller for Hierarchical Groups

Aircraft Handled and Events Group 1 Group 2 Group 3 Group 4

FAA average aircraft handled per year 171,670 135,711 96,501 95,019

FCT average aircraft handled per year 84,472 60,778 64,257 52,068

Average number of risk analysis and surface risk analysis

events involving air traffic controllers per million aircraft

handled at FAA towers

8.71 7.37 8.42 7.05

Average number of risk analysis and surface risk analysis

events involving air traffic controllers per million aircraft

handled at FCTs

1.83 0.52 0.92 1.19

Total risk analysis and surface risk analysis events

involving air traffic controllers per million aircraft handled

7.11 2.40 2.72 3.11

Exhibit H. Detailed Scope and Methodology 57

Table H-14. Cost Comparisons for K-Means Groups Using Two Runway Configuration Categories

Costs Group 1 Group 2 Group 3 Group 4

Average total FAA labor and benefits $13,149,853.94

$23,527,031.36

$10,817,413.79

$10,571,473.66

Average FAA labor and benefits cost per aircraft handled

$16.86

$29.90

$19.93

$30.63

Average total contract tower labor

$3,073,880.06

$3,081,336.05

$2,304,554.78

$2,270,812.92

Average contract tower labor cost per aircraft handled

$5.24

$8.95

$9.48

$11.20

Average percent fewer resources used by comparable

contract towers

68.95% 70.08% 52.43%

63.43%

Table H-15. Cost Comparisons for Hierarchical Groups Using Two Runway Configuration