The Air Traffic Controller Workforce Imperative: Staffing Models and Their Implementation to Ensure Safe and Efficient Airspace Operations (2025)
Chapter: 3 Facility Staffing and Safety
3
Facility Staffing and Safety
Safety has always been the priority for the Federal Aviation Administration’s (FAA’s) air traffic control (ATC). It was initially established in the United States in 1935 by a consortium of airlines to keep aircraft in flight safely separated from one another.1 At the request of Congress, the committee for the 2014 National Academies of Sciences, Engineering, and Medicine (NASEM) report examined ATC facility staffing models and the Air Traffic Organization’s (ATO’s) execution of a staffing plan to determine whether each of ATOs facilities was adequately staffed. The committee’s stated position was that safety should be the primary consideration in setting staffing levels at facilities.
The committee’s Statement of Task for this report includes a requirement to “assess FAA’s progress in furthering the recommendations in the 2014 National Academies of Sciences, Engineering and Medicine (NASEM) The Federal Aviation Administration’s Approach for Determining Future Air Traffic Controller Staffing Needs report.” This chapter begins with an overview of aviation safety related to ATC. It then reviews the responses of FAA to the safety recommendations in the NASEM 2014 report and another report by fatigue experts with similar recommendations released in 2024. The chapter then examines data on overtime by controllers provided to the committee by FAA that might be informative about the adequacy of facility staffing levels. The final section provides the committee’s findings, conclusions, and recommendations.
___________________
1 See https://www.faa.gov/about/history/photo_album/air_traffic_control.
SAFETY
The NASEM 2014 study examined ATC facility staffing models and the ATO’s execution of a staffing plan to ensure that each of its 313 facilities was adequately staffed for safety. FAA’s Office of Finance and Management (AFN) staffing demand models are based on coverage of all positions at each facility that should be open for a 90th percentile day of traffic for each facility. These open positions are postulated on being adequately staffed to safely separate traffic. If staffing at facilities is sufficient for all positions to be covered on peak travel days, that should provide for safety. However, as noted in Chapter 2, 31% of ATC facilities were more than 10% below their FAA’s Office of Finance and Management (AFN) staffing standards in Fiscal Year (FY) 2024; 23% of facilities were 15% below them (see Table 2-4). Controllers at understaffed facilities can be required to work more overtime or longer shifts which could result in fatigue as explored later in this chapter.
Accidents
Regarding accidents, the 2014 NASEM report committee reviewed National Transportation Safety Board (NTSB) fatal accident reports from 1990 to 2012 in which NTSB investigators implicated ATC as a contributing factor. A total of 66 ATC-related accidents occurred over that 13-year period (NRC 2014, Table 2-2). Relative to all fatal crashes, the 0.8% of them in which ATC played a role was viewed by the committee as an indicator of the relative success of ATC staffing in ensuring safety.2 Fortunately, there have been several years since 2012 without any domestic commercial passenger airline hull losses resulting in fatalities, as noted in Chapter 2. In the few fatal accidents that have occurred between 2012 and 2024 involving commercial passenger, cargo, tourist, and charter flights, ATC was not directly implicated.3 For the January 29, 2025, midair collision between a PSA Airlines Bombardier CRJ700 Airplane and a Sikorsky UH-60 Military Helicopter near Reagan National Airport, NTSB had an investigation
___________________
2 Most of the accidents, 57 or 86%, occurred in general aviation. Of the nine involving commercial aviation, three occurred in air carrier operations and six occurred among air taxi operations.
3 Before 2025, there were seven fatal accidents involving commercial passenger airlines since 2012, five of which occurred on the ground that included suicides, individuals illegally on the airfield, crew members ingested by engines, and an overrun of a landing that killed one passenger. An additional accident resulted from an engine failure that killed one individual while in flight and another was a suicide in a stolen aircraft. There have been three cargo aircraft fatal accidents involving pilot errors (crash landings and a crash into a mountain). There have been two fatal accidents involving tourist flights and a charter flight accident caused by overloading of the aircraft.
under way as this report was being completed.4 Speculation about possible roles of ATC personnel as contributing causes in the accident were widely reported in the media at the time, but details will not be known until the NTSB has completed and released its report.
General Aviation accidents occur much more frequently than among passenger carrier operations, as described in Chapter 2. The 2014 NASEM report committee observed that some GA accidents before 2012 could possibly have been related to lack of timely information from overly busy or fatigued controllers to pilots, but lack of records regarding the circumstances and controllers involved inhibited drawing evidence-based conclusions.
Close Calls
Although fatal aviation accidents in which ATC played a significant role are rare (NRC 2014), close calls happen with some regularity. Close calls are an important leading indicator of accident risk. Such events are investigated by FAA, and the most concerning of them are investigated by NTSB to determine contributing causes. In February 2023, FAA issued a broad-based Safety Call to Action due to an “uptick in serious close calls.”5 In March 2023, FAA held a Safety Summit with more than 200 industry leaders to address recent near miss incidents, runway incursions, and other safety indicators.6
Runway Incursions
According to FAA, a runway incursion is “any occurrence at an airport involving the incorrect presence of an aircraft, vehicle or person on a runway. Incursions are caused by operational incidents, pilot deviations, and vehicle/pedestrian deviations; and they vary greatly in type and severity.”7 In mid-2024, the NTSB found “the lack of critical safety technology and incorrect assumptions by an air traffic controller led to a near-collision between a Southwest Airlines B-737 and FedEx B-767 on an Austin, Texas runway,” in February 2023.8 A runway incursion at JFK airport in the preceding month also nearly led to a fatal accident, primarily due to the pilot ignoring controller instructions to hold before crossing a runway, which the controller did not notice.9 In the NTSB report of the Austin incident, NTSB
___________________
4 See https://www.ntsb.gov/investigations/Pages/DCA25MA108.aspx.
5 See https://www.faa.gov/aviation-safety-call-to-action.
6 Ibid.
7 See https://www.faa.gov/airports/runway_safety/resources/runway_incursions.
8 See https://www.ntsb.gov/news/press-releases/Pages/NR20240606.aspx.
9 See https://www.ntsb.gov/investigations/Pages/DCA23LA125.aspx.
Chair Jennifer Homendy stated that “one missed warning, one incorrect response, even one missed opportunity to strengthen safety can lead to tragedy and destroy public confidence in our aviation system, which is precisely why we must learn from near misses such as these.”10 FAA classifies runway incursions in four categories (see Box 3-1), of which the first two (Categories A and B) could possibly have resulted in a fatal crash. Although not indicative of changes in risk not captured by such indicators, Category A and B incursions represent less than 1% of incursions and the rate (per 1 million operations) in FY 2022–2023 is similar to prepandemic rates (see Figure 3-1).
Loss of Separation
According to FAA, “Standard separation is a specified separation minima between airborne aircraft in controlled airspace. Breaches of such minima are based on airborne loss event data. Losses of standard separation are only reported by Air Route Traffic Control Center (ARTCC)” (FAA, 2024a, 29). Separation loss event data at the time they occurred that could be related to staffing levels are not published by FAA, but annual summaries of such events pre- and post-COVID-19 indicate that through 2023, the frequency of such events remain below prepandemic levels (see Table 3-1).
Without more complete information, the committee is unable to draw inferences from the above trends regarding changes in accident risk due to
BOX 3-1
Definitions of Runway Incursions
Category A is a serious incident in which a collision was narrowly avoided.
Category B is an incident in which separation decreases and there is a significant potential for collision, which may result in a time critical corrective/evasive response to avoid a collision.
Category C is an incident characterized by ample time and/or distance to avoid a collision.
Category D is an incident that meets the definition of runway incursion such as incorrect presence of a single vehicle/person/aircraft on the protected area of a surface designated for the landing and takeoff of aircraft but with no immediate safety consequences.
___________________
10 See https://www.avweb.com/aviation-news/ntsb-cites-lack-of-safety-technology-and-controller-error-for-near-collision-in-austin.
NOTES: Rate shown is for Category A and B incursions per 1 million operations. Data are current as of December 31, 2024, and shown through November 30, 2024. Caveat: It often takes weeks for official classification of any event. Most recent 2 months of airport operations are preliminary counts because OPSNET numbers are not finalized until 20 days after the last day of the previous month.
SOURCE: Air Traffic Organization Safety and Technical Training. https://www.faa.gov/closecalls.
TABLE 3-1 Losses of Standard Separation Pre- and Post-COVID-19 Pandemic
| FY 2017–2019 Average | FY 2022 | FY 2023 | FY 2023 Compared to FY 2017–2019 | |
|---|---|---|---|---|
| Incidents | 1,221 | 961 | 1,026 | −16% |
| Rate/Million Operations | 10.33 | 8.2 | 8.59 | −17% |
SOURCE: FAA. 2024. “Air Traffic by the Numbers.” 29. Rates calculated by the committee using annual operations data provided by FAA.
actions by air traffic controllers. However, FAA has confidential information available to it beyond those mentioned above that provide more insights into not only what happened in the above and other safety incidents but also potentially why they happened. As examples:
- ATO has an automated system, Aviation Risk Identification and Assessment, that uses data extracted from radar and other surveillance data to identify air traffic operations that represent potential safety risks, even if operations are technically deemed compliant.
- ATO personnel involved in or witnessing any unsafe air traffic or technical operations event are required to ensure that Mandatory Occurrence Reports (MORs) are completed by relevant personnel and are submitted to ATO’s Comprehensive Electronic Data Analysis and Reporting (CEDAR), which is an online comprehensive data reporting, collection, and analysis tool required by FAA Order 7210.632A.11
- Employees can also complete safety reports to ATO’s Voluntary Safety Reporting Program (VSRP). As with other aviation VSRPs, employees reporting such incidents are not subject to recrimination for reporting except for certain egregious violations. This is done in the spirit of encouraging an open reporting culture that focuses on addressing causes of lapses in safety defenses to prevent accidents rather than blaming those involved in them.
- Also available to ATO are pilot or flight crew reports of Near MidAir Collisions (NMACs) when aircraft pass within 500 feet of each other.
FAA has an ongoing quality assurance program that analyzes and categorizes these and other data in CEDAR by severity. Incidents that could
___________________
11 See https://www.faa.gov/documentLibrary/media/Order/JO_7210.632A.pdf.
have resulted in an accident can be further investigated through interviews with those involved. Also available to FAA, as well as the general public, are NTSB ATC close call investigations of some of the more serious incidents. The final reports review all relevant factors to make judgments about contributing causes. These independent NTSB assessments are the most valuable for understanding whether the multiple defenses against accidents failed in some way and how these defenses can be strengthened, but they are only done for selected incidents, which are not necessarily representative. Of particular concern to the committee is whether staffing levels can affect safety results due to fatigue when controllers at short-staffed facilities must work too many hours or because understaffing results in a single controller being responsible for a combined position when having assistance for that controller would have been a prudent safety measure. As discussed next, one of the NASEM 2014 report recommendations addressed this very issue.
FAA RESPONSES TO NASEM 2014 REPORT RECOMMENDATIONS
FAA’s responses to the 2014 report committee’s recommendations are reviewed in the context of another important report issued in 2024 on controller fatigue and safety. In December 2023, in response to New York Times articles about exhaustion among air traffic controllers due to staffing shortages and a series of close calls,12 the FAA Administrator convened three respected experts “to evaluate the latest science on human sleep needs and fatigue considerations as applied to FAA’s current air traffic controller workforce, work requirements, and scheduling practices.”13 Their report, “Assessing Fatigue Risk in FAA Air Traffic Operations: Report by Scientific Expert Panel on Air Traffic Controller Safety, Work Hours, and Health” (Rosekind et al. 2024) was released in April 2024. Rosekind et al. (2024) identified 58 opportunities for improving ATC operations to reduce fatigue and improve safety, several of which parallel those made in the NASEM 2014 report. In reviewing the responses of FAA to the 2014 report safety recommendations, this committee draws on the scientific expert panel’s observations and recommendations along with presentations made by FAA officials regarding how they have responded to the recommendations made in both reports.
Both reports addressed staffing at facilities, and both found controller fatigue to also result from the shift scheduling practices that ATO uses to operate in a 24/7/365 environment. As noted by Rosekind et al. (2024),
___________________
12 See https://www.nytimes.com/2024/04/19/business/faa-air-traffic-controllers-close-calls.html and https://www.nytimes.com/2023/12/02/business/air-traffic-controllers-safety.html.
13 See https://www.faa.gov/newsroom/media/fatigue_standard_directive_4192024_signed.pdf.
Daily, 45,000 flights carry 2.9 million passengers to their destinations, guided by more than 13,000 air traffic controllers operating in 313 facilities. Around-the-clock, every day, these air traffic controllers confront known challenges to humans working in 24/7 operational settings, amplified even further in safety-sensitive environments where errors can mean lives lost and people injured. Fatigue related to sleep loss and circadian disruption is created when human operators work schedules around-the-clock and is known to have significant adverse effects across safety, performance, health, and mood.
The commitment of the dedicated professionals in the ATC workforce to safety is reflected in these controllers being “part of a remarkably safe system that allows the current high-performance and effectiveness of the NAS [National Airspace System]” (Rosekind et al. 2024). In examining whether the reduced staffing level of the ATC workforce since 2012 could be compromising safety, FAA’s shift scheduling patterns at facilities, which have included widespread use of shifts that induce fatigue, must also be accounted for.
The committee for the NASEM 2014 report made 22 recommendations in the body of the report, which it summarized in four overarching recommendations made in its Executive Summary. Although most of the recommendations in the 2014 report are related to safety to some degree, three of the overarching ones address safety directly regarding
- the collection of data to support research on the role of staffing levels and shift schedules on controller fatigue, performance, and safety;
- implementation of a robust Fatigue Risk Management System (FRMS) to supplement the rules and policies governing existing fatigue management; and
- development or acquisition of shift scheduling software for facility manager use that adheres to fatigue rules and best practices in Fatigue Risk Management (FRM) while making efficient use of existing staff.
The committee for the 2014 report made the second and third recommendations out of concern about the shift scheduling practices in widespread use at the time (rotating counterclockwise shifts exemplified by the 2-2-1 schedule) that were known to induce fatigue and degrade controller performance, as discussed below.
Among the four main opportunities for improvement recommended by Rosekind et al. (2024), three overlap with the concerns expressed, and recommendations made, by the 2014 NASEM report committee:
- Integrate prescriptive fatigue-oriented rules and policies with FRMS into a single system for ATO guidance and use in FRM;
- Develop a plan to eliminate the 2-2-1 counterclockwise shift schedule; and
- Update current prescriptive policies to address identified fatigue factors, especially to avoid known schedule practices that induce fatigue, including minimum rest periods between shifts, especially before a midnight shift.
The committee for the NASEM 2014 report made recommendations to conduct research on the effects of fatigue related to inadequate staffing, FRM, and improved shift scheduling tools to help facility managers schedule staff in full compliance with fatigue rules and best practices in FRM.
Research Related to Fatigue and Staffing
Fatigue resulting from overwork or inadequate time off for sleep and recovery is known to be a risk factor in aviation. The committee for the NASEM 2014 report noted that
fatigue is a risk factor for errors and accidents, and it is frequently encountered in operations that need to be sustained 24/7, like ATC. Fatigue can be broadly defined as “a physiological state of reduced mental or physical performance capability.” Numerous factors are known to induce or contribute to fatigue (ICAO 2011):14
- Acute sleep loss (i.e., being awake for a prolonged period of time),
- Chronic sleep restriction (i.e., not getting enough sleep per 24 hours on a regular basis),
- Circadian rhythm (e.g., working at night when the body is programmed to sleep or sleeping during the day when the body is programmed to be awake),
- Low-quality sleep due to sleep fragmentation that is induced internally (e.g., by sleep disorders) or externally (e.g., by noise),
- Sleep inertia (i.e., a period of reduced performance capability immediately after waking up),
- Time on task (i.e., prolonged work periods without breaks),
- High workload, and
- Trait like (i.e., likely genetic) interindividual differences in susceptibility to the above factors.
___________________
14 Rosekind et al. (2024, 33 and 66) provide many citations from the sleep science literature documenting the performance decrements associated with disrupted and inadequate sleep and chronic fatigue.
The safety of the National Airspace System (NAS) depends on continuously high levels of controller performance. At the same time, the tasks of controllers can be complex and demanding, and they require a high level of attention. The latter is affected profoundly by sleep loss, circadian misalignment, and other fatigue-inducing factors (Lim and Dinges 2008, 2010).
Although the importance of fatigue’s impact on performance and safety was well understood, the status of data collection and research in 2014 was not adequate for the 2014 NASEM report committee to draw firm conclusions about quantitative relationships among staffing levels, fatigue, and safety to inform specific adjustments that should be made to FAA’s staffing models. In its Recommendation 2-5, the 2014 NASEM report committee stated:
To ensure that staffing decisions are properly informed and to help communicate the basis of these decisions throughout the organization, FAA should promote data collection and analysis to improve identification of key relationships between controller staffing and safety. Specifically, (a) the data should consistently include relevant staffing and fatigue concerns so that they can be considered in the analysis, considered in accident and incident investigations, and incorporated into the appropriate questions in the forms underlying safety reports; (b) the data should be compiled and stored in a way that allows integration of data sets (e.g., integration of staffing records with incident reports); (c) the appropriate data analyses should be conducted to examine where and how controllers have contributed to accidents and incidents to help inform the models of controller activity underlying staffing standards—including whether controllers have adequate time both to separate traffic and to issue safety alerts, as required by FAA Order 7110.65—and to identify relationships between staffing and safety; and (d) especially in view of the limitations on FAA resources, FAA should ensure that enough researchers have access to the data to provide the range of perspectives and expertise necessary for identifying how staffing concerns relate to safety.
FAA officials reported to the committee in FY 2024 that ATO has developed and implemented a program required by FAA Order 7210.632A of systematically collecting information associated with incidents and accidents in which controllers may have contributed, and it analyzes controller self-reports of fatigue on an ongoing basis. As one example of the results of such analysis, watch staffing15 at Towers was increased to reduce the risk of close calls. As valuable as such practices are in mitigating risk, this data collection and analysis has remained within FAA and has not led to
___________________
15 See https://www.faa.gov/air_traffic/publications/atpubs/foa_html/chap2_section_6.html.
the conduct of external research that would apply scientific methods to the study and understanding of the interrelated roles of facility staffing, fatigue, and safety. This lack of scientific research has inhibited the committee for this report in reaching judgments about the relationship between staffing levels and safety.
Of particular concern to the 2014 NASEM report committee was widespread use of counterclockwise rotating weekly shift schedules, as exemplified by the 2-2-1 “rattler” (see Figure 3-2). As an example of this kind of schedule at the time of the 2014 report, over the course of 1 week a controller would begin the first 2 days of a weekly shift at 1:30 PM, then report to work at 7 AM on the next 2 days, then come back to start the midnight shift at 11 PM on the fourth day (with only 8 hours off between stopping a day shift and beginning a midnight shift).16 The controller would complete work at 7 AM on the fifth day. This schedule has the advantage of sharing responsibility for midnight shift coverage across the workforce rather than having workers devoted to either day or night shifts. It also compresses the workweek to less than 5 days and gives controllers 80 to 90 hours off depending on the start time of their shift on the first day of the second week. These 80+ hours off have been popular with many, but not all, controllers (Rosekind et al. 2024). Surveys of controllers by FAA and the National Aeronautics and Space Administration (NASA) in the late 1990s and 2012 found this type of shift schedule in use roughly 60% of the time (as cited in Rosekind et al. 2024). This widespread use of such schedules was consistent with analysis of shift schedules in January 2024 by Rosekind et al. (2024). A significant disadvantage of such schedules is that controllers do not have adequate time or opportunity for restorative sleep between the end of a morning shift and beginning of a midnight shift. Not only is there inadequate time between shifts for commuting back and forth, participation in regular family activities, and ability to engage in restorative sleep, it is difficult for many individuals to go to sleep during the day at a time counter to their bodies’ circadian rhythms. Rosekind et al. (2024) cite FAA’s own surveys showing that controllers have complained about inadequate sleep before a midnight shift and achieving as little as 3 hours of sleep between shifts, which is simply not adequate to prepare controllers for the night shift where their sleep debt will be exacerbated by the circadian system promoting sleep rather than alertness.
The 2-2-1 schedule is referred to as the “rattler” because of how disruptive it is to sleep schedules and performance. A schedule that rotates counterclockwise to normal circadian rhythms is particularly fatigue inducing, yet many ATC facilities have relied on such schedules for decades. The
___________________
16 The minimum rest period in 2024 had been extended to 9 hours, and this recovery period was increased to 12 hours before a midnight shift following release of Rosekind et al. (2024).
NOTE: This figure shows a minimum of 8 hours between the end of a day shift and the beginning of a midnight shift, but at the time the analysis was done, FAA had already increased that to 9 hours.
SOURCE: Rosekind et al. 2024, 39.
2014 NASEM report committee expressed concern that it was used at all in light of the insights gained through sleep science in recent years. Perhaps its widespread use was justified by FAA’s and other past research showing that other alternative shift schedules in use also created fatigue that led to significant performance decrements (as cited in Rosekind et al. 2024, 42). However, Rosekind et al. (2024, 42) note that the comparison across types of shift schedules in these studies did not include comparison with other alternative schedules (not in use at the time) that would have been less fatigue inducing: “There are a number of alternative schedule designs that could be more effective in addressing fatigue during 24/7 operations that should be considered when exploring alternatives to the counterclockwise rotating 2-2-1 schedule.”
Immediately upon release of the scientific expert panel’s report in 2024, the FAA Administrator required expansion of minimum rest periods between shifts, which initiated elimination of the 2-2-1 schedule, and required the integration of fatigue-oriented prescriptive rules and policies and its FRMS into a single source for ATO FRM.17 Following agreements reached between FAA and the National Air Traffic Controllers Association (NATCA), beginning in January 2025 the new rest periods (10 hours off duty between all shifts and 12 hours before and after a midnight shift) were implemented at all operational facilities. Also, in accordance with the agreement reached, collaborative workgroups were established to work with the scientific panel and make recommendations on strategies to eliminate the 2-2-1 counterclockwise rotating schedule in 2026 and another group to make recommendations on the establishment of minimum durations for recuperative breaks on all shifts. Whereas FAA has taken decisive action about eliminating the 2-2-1 shift schedule, which should reduce fatigue, the effects of facility staffing levels on fatigue and safety remain unproven empirically. Even so, such risks could be managed through an effective FRMS.
Fatigue Risk Management (FRM)
The committee for the NASEM 2014 report noted that
full implementation of a Fatigue Risk Management System (FRMS) [see Box 3-2] could go beyond duty time limitations and mitigate fatigue in a more comprehensive manner. ICAO [International Civil Aviation Organization] defines an FRMS as “a data-driven means of continuously monitoring and managing fatigue-related safety risks, based upon scientific principles and knowledge as well as operational experience that aims to
___________________
17 See https://www.faa.gov/newsroom/media/fatigue_standard_directive_4192024_signed.pdf.
BOX 3-2
Fatigue Risk Management Systems
Fatigue Risk Management Systems (FRMS) are data-driven approaches to managing fatigue in fields such as transportation that operate 24/7 and tax the abilities of operators to obtain sufficient restorative sleep between shifts. ICAO (2018) defines FRMS as “A data-driven means of continuously monitoring and managing fatigue-related safety risks, based on scientific principles and knowledge as well as operational experience that aims to ensure relevant personnel are performing at adequate levels of alertness.”
FAA allows an approved FRMS to serve as an alternate means of compliance with regulations that limit hours of duty (FAA 2013) and has its own FRMS for the Air Traffic Organization (ATO) (JO-1030.10A).
Based on the principles of Safety Management Systems (SMS), effective FRMS depend on continuous monitoring and risk assessment. Two of the basic challenges for FRMS is this regard are that (a) fatigue is difficult to define and measure and (b) there is considerable individual variability in how fatigue affects performance. Even so, the applications for monitoring such as bio-mathematical models, self-report measures, and performance monitoring technologies have been advancing.
Short duration testing of fatigue using alertness monitoring tests are subject to both Type 1 and Type 2 errors. Techniques for monitoring also include bio-mathematical models to alert managers of possible issues with flight crew schedules and automated visual screening of operator eye movements and wearable devices to detect and mitigate fatigue. Reports from wearable devices by controllers might be a way to alert managers of controllers whose performance is falling below an acceptable level. While promising, measurement of fatigue using wearable devices requires further development (Martins et al. 2021). Few studies have evaluated the effectiveness of FRMS themselves, but components of FRMS are associated with improved safety metrics (Sprajcer et al. 2022).
ensure relevant personnel are performing at adequate levels of alertness.” An FRMS shares characteristics with safety management systems (SMSs), including effective safety reporting, senior management commitment, a process of continuous monitoring, a process for investigation of safety occurrences that aims to identify safety deficiencies rather than apportion blame, the sharing of information and best practices, integrated training for operational personnel, effective implementation of standard operating procedures, and a commitment to continuous improvement.
The committee’s Recommendation 2-2 stated that
the safety of the NAS has highest priority. FAA should ensure full implementation of an FRMS, with sufficient resources for proactive analysis of potential fatigue concerns and mitigations and with follow-up evaluation
of all changes in policy, training, and scheduling. Before claiming success for any recent (and future) initiatives addressing controller fatigue, FAA must ensure that these initiatives are assessed properly and are having their intended impact once they are implemented at local facilities.
At the time of this 2014 recommendation, FAA was already at the forefront of application and oversight of FRM for commercial air carrier operations. In 2010 the agency published guidance to operators and subsequently specified this guidance in regulation. Even before the 2014 NASEM report committee’s recommendation about an FRMS for ATC operations, in 2011 FAA had implemented an FRMS for ATO through Job Order 1030.10A. However, in reviewing the 14 years of ATO’s FRMS, Rosekind et al. (2024) noted that it lacked evidence of ongoing evaluation of operations that is inherent in any FRMS or Safety Management System (SMS). Rosekind et al. state that “it is unclear whether there have been any evaluations conducted to determine the overall effectiveness of the current mechanisms or individual elements.” The committee for the current report questions how an effective FRMS could have allowed continuance of schedules such as the 2-2-1. Moreover, in the next section evidence is provided of frequent exceedances of fatigue rules in shift schedules that should have been caught and ended by an effective FRMS. Rosekind et al. (2024) note that ATO has only 1.5 Full-Time Equivalents (FTEs) devoted to its FRMS, whose activities are spread across 30 initiatives; thus, budget and staffing constraints may be influencing ATO’s ability to carry out an effective FRMS.
Scheduling Tool That Is Fatigue Policy Compliant and Efficient
A shift scheduling tool that complies with fatigue mitigation rules and policies and schedules available staff efficiently is important for two reasons. For safety, it should incorporate both prescriptive fatigue management rules and policies as well as best practices in FRM. Second, it should also optimize use of existing staff in designing schedules. Drawing on findings about shift scheduling tools that appear in three of the chapters in the 2014 NASEM report, the 2014 NASEM report committee’s overarching recommendation stated that
FAA should, as a matter of priority, continue its efforts to develop an improved scheduling tool capable of creating efficient controller work schedules that incorporate fatigue mitigation strategies. The agency should collaborate closely with the National Air Traffic Controllers Association in implementing this improved scheduling capability, notably in adopting schedules that reflect science-based strategies for managing the risks associated with controller fatigue.
As recently as 2023, the U.S. Department of Transportation’s Office of Inspector General (OIG 2023) reported that FAA had not yet implemented a “standardized scheduling tool to optimize controller scheduling practices at air traffic control facilities,” repeating a finding it had made in an earlier report devoted to that subject (OIG 2018).
As FAA explained to the committee for the current report, it acquired a Commercial Off-The-Shelf (COTS) shift schedule software package in 2010. The Operational Planning and Scheduling (OPAS) tool is an online scheduling system based on this software package; it uses mathematical algorithms and on-position demand to optimize facility start times and shifts. It is intended to give facilities the ability to produce efficient, data-driven schedules.
FAA officials reported to the committee that OPAS was used initially as a management and oversight tool by ATO and AFN and was then used at two Centers in FY 2013–2014. The 2016 Collective Bargaining Agreement (CBA) expressed a commitment to implement OPAS. It was successfully implemented at one Center, after which NATCA requested a second test implementation at another Center, which was also successful, including endorsement by the local NATCA representative. As had been recommended by the committee for the NASEM 2014 report, ATO collaborated with NATCA in an analysis of OPAS capabilities and its intended use in all facilities. FAA made numerous enhancements to OPAS in response to 264 ATO-NATCA working group and user requests.18 However, the working group concluded that many more would be required. Among other problems with OPAS, NATCA representatives explained to the committee that the software had many bugs, lacked full online capability, was incompatible with the 2016 CBA, and required 2 weeks of training for facility managers to learn how to use it.
ATO dropped all plans to implement OPAS in 2017, but OPAS has continued to be used at the national level to generate optimal facility shift schedules and consistency with policies (an example of which is provided below). At the time the committee was briefed in mid-2024, the maximum contract period for OPAS had been reached. FAA had initiated a new open-ended bidding process for OPAS or a better replacement system if one emerged. Meanwhile, most facilities use a “WMT Web Schedules” system for setting schedules. FAA analysis of WMT capabilities indicates that “WMT makes no effort to prevent schedulers from creating schedule violations of all but a very narrow set of existing fatigue rules.”19
___________________
18 FAA email of March 17, 2025, in response to committee request for information.
19 Email correspondence between FAA and the committee.
Recent evidence indicates that fatigue rule violations occur regularly. One of the top four priorities for ATO identified by Rosekind et al. (2024, 2) is to
identify and determine specific circumstances around a subset of representative scheduling policy and agreement exceedances then implement mechanisms to monitor and eliminate such exceedances. This effort should be focused on developing and implementing these mechanisms and not involve punitive actions for past circumstances.
Rosekind et al. (2024) carried out a preliminary analysis of FAA ATC shifts during the first 10 weeks of 2024. They found numerous, if mostly relatively minor, exceedances of fatigue rules as well as a small number of egregious ones. Subsequently, FAA carried out its own analysis of FY 2024 schedules using OPAS and found
in excess of 4,000 fatigue rule violations present in published schedules for 2024 alone. Further, there are violations across all of the individual fatigue rules such as minimum time off between shift, more than 6 consecutive workdays, and required 30 hours off in every rolling 7-day period.20
At the time of this writing in early 2025, it is unclear whether these fatigue rule violations are due to lack of robust shift-scheduling software or inadequate staffing at many facilities, or some combination of the two.
The analyses of exceedances identified by Rosekind et al. (2024) and by FAA itself indicate the importance of evaluating the actual schedules implemented as well as those that are planned by facility managers. Rosekind et al. (2024) note that actual schedules worked reflect the realities of staffing, such as accommodating unexpected surges in traffic, extra efforts required to deal with severe weather, and unanticipated demand for sick leave or flextime. Planned shift schedules are sometimes checked by ATO headquarters for compliance with FAA rules and policies, but actual schedules worked are better indicators of efficient use of existing staff and facility staffing adequacy.
The committee for this report concludes that ATO made a good-faith effort to comply with the 2014 NASEM report committee’s recommendation, and the system acquired was expected to be a substantial advance over (then) current practice. However, the inability to implement OPAS at local facilities results in the lack of adequate shift scheduling tools that enforce fatigue rules and policies, best practices in FRM, and efficient shift
___________________
20 Ibid.
scheduling of existing staff. Issues with efficient shift scheduling are discussed in more detail below.
OVERTIME AND OTHER MEASURES OF WORKFORCE STRAIN
As described in Appendix B, at the committee’s request FAA provided an extensive data file on the ATC workforce since FY 2010 through FY 2024. In addition to information about changes in the size of the workforce since 2010, the committee requested information about regular and overtime hours worked, which the committee expected would yield insights about strains on the workforce, including fatigue. FAA officials indicated to the committee that they viewed an annual facility overtime percentage in excess of 2% as a potential indicator of inadequate staffing. Ongoing review of such evidence would indicate the consequences of reduced staffing and could help validate whether FAA’s facility staffing models were estimating appropriate staffing levels to ensure safety. Review of broad trends in the workforce related to staffing levels and their divergence from what is expected are followed by (a) discussion of further counterintuitive trends in overtime use at facilities that are both within and outside of the AFN staffing standard range and (b) possible other approaches that might help partially validate the relationship between AFN model fill rates and safety.
Broad Trends in Traffic Operations, Staffing, and Regular and Overtime Hours Worked
Before arriving at measures of overtime at facilities, it is necessary to begin with an appreciation of general trends in staffing levels, traffic operations, regular hours worked, and overtime hours worked. Figure 3-3 displays aggregate trends across all facilities of total FTEs and total operations. Between FY 2010 and 2024, total FTEs declined 13% while total operation increased by 4%. (It should be noted that some individual airports had increases in total operations in the double digits.)
The committee expected that regular time worked would have to increase in order to compensate for reduced staffing and increased operations. However, regular hours worked per FTE have declined over the period as shown in Figure 3-4. This figure shows regular hours worked in the three general categories for aggregating reported time. Time on Position (TOP) is time recorded when individuals are managing traffic; this is the primary operational duty of controllers. The decline in TOP shown in Figure 3-4 represents a drop from 50% TOP in FY 2010 to 47% TOP in FY 2024. (Average FTE TOP in FY 2024 was 790 hours, a decline of 106 hours/FTE compared to FY 2010.) Available time (something of a misnomer) covers recovery periods following TOP and meals; thus, it represents time when
SOURCE: Generated by the committee from data provided by FAA.
SOURCE: Generated by the committee from data provided by FAA.
controllers are not available for TOP or Other Duties. Other Duties include training, staff meetings, special assignments, and many other categories of work when not on position or in a recovery period or break. (The committee takes note of the disagreements within FAA regarding how well ATO’s timekeeping system accurately accounts for the categories of work that controllers carry out on Other Duties. The disagreement is not about the amount of time recorded, but whether some time spent on Other Duties is reported as Available Time or is simply not done. This disagreement is discussed further in Chapter 4 but is not central to this chapter.)
The first of several curious trends is that recorded regular hours worked have declined about 4% (an average of 79 hours per year per FTE over the 15-year period). Even more curious is that average TOP per FTE has declined 7% per 8-hour shift. These trends are counterintuitive because traffic operations increased 4%, FTEs are down 13%, yet the workforce has been able to manage the increased traffic with fewer staff working fewer regular hours and less time actually managing traffic. FAA officials hypothesized that improvements in traffic flow (the “regularization” of traffic flow) might have made it possible for controllers to become more productive. The committee has no way to test this theory, but, if true, it indicates that the task load models within FAA’s facility staffing modeling process described in the next chapter need to be recalibrated.
Part of the explanation for these trends is that overtime has increased substantially over the period as shown in Figure 3-5. Overtime per FTE has increased by a total of 126 hours per FTE annually (308%), even as other regular hours worked declined. FAA officials did not offer an explanation for these trends. The increased use of overtime may be related to the hiring constraints that FAA experienced beginning in FY 2013 (as described in Chapter 2), since overtime begins increasing after FY 2013. However, these increased overtime hours are not limited to facilities that are short-staffed by AFN’s staffing standards, as described later.
The declines in regular TOP hours worked per FTE shown in Figure 3-6 are perplexing given that operations increased and staffing decreased. The committee does not have an explanation for these counterintuitive trends. They may be related to how well facility managers utilize their staffs in scheduling shifts, which might be due to lack of an efficient and enforceable shift scheduling tool as described above. FAA provided the committee two case studies of facilities whose excessive use of overtime was partially explained by inefficient shift scheduling (FAA 2022, 2024). See the discussion below in the section title “Implications of Shift Scheduling Practices” for the committee’s effort to estimate the prevalence of such practices.
SOURCE: Generated by the committee from data provided by FAA.
SOURCE: Generated by the committee from data provided by FAA.
Effort to Partially Validate AFN Models
To return to the focus of this section, the committee’s interest is whether the substantial increase in overtime since FY 2013 is explained by the growing number of facilities and growing number of FTEs working at facilities that are more than 10 or 15% below the AFN staffing standard (see Tables 2-4 and 2-5). If this is the case, then it would be possible to partially validate the AFN facility staffing models by comparing the facilities that are within their staffing standards, where overtime should be low because they are
adequately staffed, with facilities that are below the staffing standard range, where facilities would need more overtime to compensate for reduced staffing levels. Facilities that are heavily dependent on overtime could be expected to have staff that are more fatigued and more prone to making errors that would result in increased incidents or close calls. Facilities that are staffed 10 to 15% more than their staffing standards should have even less overtime than those that are staffed +/−10% of the staffing standard range. Staff working at these facilities should be the least fatigued and have the fewest close calls related to fatigue.
As indicated in Table 3-2, use of overtime has increased substantially across all Facility Levels regardless of how well staffed they are by AFN’s standards.21 As shown in the bottom row of Table 3-2, facility use of overtime has increased from 2 to 9% over the FY 2010–2024 period. As shown in the top lines of the three groups of Facility Levels, facilities staffed more than 15% above the staffing standard grow from 1% overtime in FY 2010 to 6 to 7% in FY 2024. At the other extreme, facilities staffed 15% below the staffing standard in all three groups rose from 4% overtime in FY 2010 to between 9 and 12% by FY 2024. In each group, facilities within +/−10% of the staffing standard show similar patterns but with anomalies across fill rates such as Level 7–9 facilities within 100 to 105% of the staffing standard having as high a percentage of facility overtime (11%) as those that are 15% below the staffing standard.
The committee ran a series of correlation checks between facility staffing levels shown in Table 3-2 relative to their AFN staffing standard and these facilities’ reliance on overtime (see Table 3-3).22 The facilities are grouped by operations complexity Level in recognition of the lack of central tendency across facility AFN fill Levels as described in Chapter 2 and the generally increased reliance on overtime from facilities in complexity Levels 4–6 up to complexity Levels 10–12 as shown in Table 3-2. The correlations have the expected sign: as facility fill rates increase, facility reliance on overtime goes down.
Although the correlation gets stronger as facility complexity Level increases, especially in Level 10–12 facilities, the overall correlation is −0.403, which is not indicative of a strong relationship between fill rates and overtime across Levels 4–9. Plots of overtime using AFN or Collaborative Resource Working Group (CRWG) fill rates provided in Appendix B indicate very little correlation. Apparently, something else is also causing
___________________
21 Facility percentage overtime is calculated by dividing total overtime hours by the sum of regular hours worked and overtime hours worked in this and Tables 3-2 and 3-3.
22 The four Combined Control Facilities are included in the total correlation coefficients reported but their coefficients are not shown because of the small number of facilities in this group.
TABLE 3-2 Overtime (OT) Across Facilities by Complexity Level, FY 2010, 2017, and 2024
| FY Values 2010 | 2017 | 2024 | |||||
|---|---|---|---|---|---|---|---|
| Facilities | AFN Standard | Facilities (N) | Facility % OT | Facilities (N) | Facility % OT | Facilities (N) | Facility % OT |
| Level 4–6 | >115% | 38 | 0 | 41 | 0 | 36 | 6% |
| 110–115% | 3 | 0 | 5 | 0 | 7 | 8% | |
| 105–110% | 3 | 1% | 7 | 0 | 9 | 6% | |
| 100–105% | 7 | 0 | 3 | 2% | 6 | 6% | |
| 95–100% | 4 | 0 | 7 | 0 | 9 | 4% | |
| 90–95% | 8 | 0 | 2 | 3% | 11 | 6% | |
| 85–90% | 7 | 0 | 6 | 0 | 7 | 5% | |
| <85% | 20 | 0 | 33 | 0 | 19 | 9% | |
| Total | 90 | 2% | 104 | 4% | 104 | 6% | |
| Level 7–9 | >115% | 35 | 0 | 38 | 0 | 32 | 7% |
| 110–115% | 10 | 0 | 9 | 0 | 11 | 9% | |
| 105–110% | 11 | 1% | 14 | 4% | 16 | 9% | |
| 100–105% | 17 | 2% | 15 | 5% | 12 | 11% | |
| 95–100% | 16 | 1% | 12 | 5% | 21 | 7% | |
| 90–95% | 15 | 3% | 10 | 4% | 7 | 7% | |
| 85–90% | 14 | 2% | 10 | 3% | 13 | 8% | |
| <85% | 36 | 4% | 34 | 5% | 31 | 11% | |
| Total | 154 | 2% | 142 | 5% | 143 | 9% | |
| Levels 10–11 | >115% | 19 | 1% | 20 | 6% | 8 | 7% |
| 110–115% | 7 | 2% | 5 | 4% | 3 | 9% | |
| 105–110% | 7 | 2% | 8 | 5% | 12 | 7% | |
| 100–105% | 8 | 4% | 9 | 5% | 6 | 10% | |
| 95–100% | 6 | 5% | 12 | 6% | 6 | 9% | |
| 90–95% | 8 | 6% | 4 | 7% | 6 | 9% | |
| 85–90% | 5 | 4% | 3 | 7% | 7 | 9% | |
| <85% | 8 | 4% | 5 | 11% | 18 | 12% | |
| Total | 68 | 2% | 66 | 6% | 66 | 10% | |
| Grand Total | 312 | 2% | 312 | 5% | 313 | 9% | |
| Facility Level | Centers (21) | TRACONs (25) | Tower and Approach Control (121) | Tower (142) | Correlation Coefficient |
|---|---|---|---|---|---|
| 4–6 | — | — | –0.474 | –0.244 | –0.306 |
| 7–9 | — | –0.227 | –0.331 | –0.298 | –0.298 |
| 10–12 | –0.593 | –0.560 | –0.833 | –0.440 | –0.453 |
| All Levels | –0.593 | –0.449 | –0.284 | –0.365 | –0.358 |
growing use of overtime at facilities since declines in facility-level staffing at Levels 9 and below do not correlate well with the increased reliance on overtime. Thus, the committee’s expectation that growing overtime use would be primarily explained by facilities that are less than 10 or 15% of the staffing standard fill rates proved to be incomplete. Facility reliance on overtime alone cannot be used to partially validate the AFN staffing standards model. Other indicators are needed to explain the widespread use of overtime and where and how much it is contributing to workforce strain in ways that could affect safety.
Prevalence of 6-Day Workweeks
Another approach to considering strain on facility workforces that could increase fatigue would be the prevalence of 10-hour days and 6-day workweeks. Reports suggest that controllers at 40% of facilities worked 6 days in a row at least once per month in FY 2022 and that “several facilities required 6-day workweeks regularly.”23 FY 2022 was a year when the nation was recovering from the COVID-19 pandemic and traffic operations had rebounded sharply while many facilities were still recovering from the pandemic and overall staffing continued to decline (see Figure 3-3).
The data FAA provided the committee on the controller workforce and described in Appendix B are aggregated in 2-week pay periods rather than by day or shift, thus the committee was unable to estimate hours worked on a daily or weekly basis. However, it could estimate the prevalence of 6-day workweeks across entire pay periods to estimate how many facilities are requiring them. To do so, the committee calculated the number of 2-week time sheets in the first three quarters of FY 2024 that recorded Greater Than or Equal to (GTE) 96 hours (more than 12 8-hour days over 2 weeks). Overall, an average of 6% of controller time sheets over the 9-month period
___________________
23 See https://www.natca.org/2024/04/19/natca-calls-on-faa-to-collaborate-on-air-traffic-controller-fatigue.
recorded GTE 96 hours worked, but the distribution was highly uneven (see Table 3-4). Nine facilities (3%) had more than 20% of time sheets reporting Greater Than (GT) 96 hours worked and 38 (12%) had more than 10, up to 20%, of time sheets recording GTE 96 hours worked. The majority (266 or 75%) had less than 10% of time sheets with GTE 96 hours worked. Having more than 20% of time sheets indicating 6-day workweeks at 3% of facilities is a relatively small percentage of the total but is nonetheless an important indicator of potential fatigue at these facilities that could lead to increased errors by controllers.
The nine facilities with 20% or more of time sheets recording GTE 96 hours represents a wide range of facilities by type and Level. They include a very large center in New York; two Terminal Radar Approach Controls (TRACONs) (one near New York City and one in central Florida), three Towers at airports in Oakland, Orlando, and New Orleans with significant commercial traffic; and three Towers at small airports.24 Six of the nine facilities are staffed 10% or more below the staffing standard range; five of the six are below 15%. However, three are within or above their staffing standards, one of which is 27% above. The committee did not have the resources to investigate these anomalous individual cases.
The number and share of controllers within these nine facilities who are estimated to be working 6-day workweeks in the first three quarters of FY 2024 is shown in Table 3-5. Of the 420 controllers across these 9 facilities who reported GTE 96 hours worked per pay period, 77% come from the New York Center; thus, its results are dominating the total. As can be seen, 35.5% of controllers in these facilities recorded GTE 96 hours worked in five or fewer pay periods. At the other extreme, 26% of controllers in these nine facilities recorded working GTE 96 hours worked for half of the total pay periods during the first three quarters of FY 2024. Such an extent of
| Time Sheets GTE 96 Hours Worked | Facilities (N) | (%) |
|---|---|---|
| GT 20% | 9 | 3 |
| 10.1–20% | 38 | 12 |
| 1–10% | 219 | 70 |
| LT 1% | 47 | 15 |
| Total | 313 | 100 |
___________________
24 New York TRACON (N90), Fort Lauderdale Executive Tower (FXE), New Orleans Tower (MSY), New York Center (ZNY), Oakland Tower (OAK), Central Florida TRACON (F11), Westchester Tower (HPN), Orlando Tower (MCO), Midland Tower (MAF).
| Number of Controllers Reporting GT 96 Hours per Pay Period | ||
|---|---|---|
| Pay Periods | Controllers (N) | (%) |
| 0–5 | 149 | 35.5 |
| 6–10 | 162 | 38.6 |
| 11–15 | 94 | 22.4 |
| GT 15 | 15 | 3.6 |
| Total | 420 | 100.0 |
NOTE: Totals may not add to 100% due to rounding.
sustained overtime without sufficient time for recovery is a concerning indicator of fatigue at these facilities.
Implications of Shift Scheduling Practices
Shift scheduling practices may be an important influence on the anomalous patterns of overtime use shown above. As an example of the effects of shift scheduling on overtime use, FAA provided the committee a detailed case study of the Columbus Tower (CMH), which experienced increased overtime in FY 2024 compared with FY 2019, even though it was adequately staffed according to AFN’s staffing standard and traffic in FY 2024 was below that of FY 2019 (the last prepandemic year) (FAA 2024b). The actual shift scheduling at CMH was less efficient than the optimal shift schedule guidance provided to CMH by FAA (see Table 3-6). As can be seen, in some days (Mondays, Saturdays) CMH scheduled 8–9 more controllers than the optimal schedule. CMH’s less efficient scheduling combined with other facility scheduling practices resulted in it exceeding its overtime budget by $330,000 with 6 weeks remaining in 2024.25
The CBA between FAA and NATCA (Article 32, Section 2) requires that shift guidelines be set as a result of negotiation at the local level between the facility management and local union officials. Shift guidelines are defined as the “number of employees to meet forecasted workload requirements for core and ancillary shifts.” In addition, the shift guidelines are incorporated into the local Basic Watch Staffing agreement, which means they are more than just advisory.26 There is no apparent requirement from
___________________
25 FAA calculated this for the “leave year,” which applies to the number of 2-week pay periods, which can vary from one calendar year to another.
26 Watch staffing is set based on the number of positions that are deemed necessary to staff positions that must be open to meet projected traffic on any given shift.
TABLE 3-6 Optimal Versus Actual Shift Schedules at Columbus Tower
| Schedule | Sun | Mon | Tues | Wed | Thu | Fri | Sat |
|---|---|---|---|---|---|---|---|
| Optimal | 10/9/3 (22) |
6/8/4 (18) |
8/8/4 (20) |
8/8/4 (20) |
10/7/4 (21) |
9/7/4 (20) |
8/6/4 (18) |
| Actual | 11/10/5 (26) |
11/11/5 (27) |
11/11/5 (27) |
11/11/5 (27) |
11/11/5 (27) |
11/11/5 (27) |
11/10/5 (26) |
NOTE: In the cells of the table the numbers separated by slashes indicate the three daily shifts starting in the morning, afternoon, and before midnight.
SOURCE: FAA 2024, 10.
FAA that shift guidelines strive for the most efficient use of existing staff while also complying with rules and policies to minimize fatigue.
There were other drivers of CMH’s overtime, such as allocating more controllers on peak days than required by the shift guidelines, which FAA (2024) characterized as: “CMH employees have complained of ‘mandatory’ 6-day workweeks with ‘nothing to do’ on the days they are assigned overtime.” Decisions local managers made also allowed leave in excess of what the local CBA agreement required. Between FY 2019 and 2024 CMH experienced an 8% decline in operations and a 3% increase in Certified Professional Controllers (CPCs) and CPCs-In Training (CPC-ITs), which makes the 23% increase in overtime perplexing.27 Moreover, the increase in overtime was accompanied by a 6% decrease in average TOP from 4 hours and 37 minutes to 4 hours and 20 minutes, which is not surprising given the adequacy of its staffing. Although used for facility staffing modeling and not an operational requirement,28 CMH’s average annual TOP was already below ATO’s stated policy guidance that average annual facility TOP be no more than 5 hours per 8-hour shift (see Chapter 4).
Since the committee did not have data on the shift patterns at individual facilities, it relied on the data it did have to determine the number of other facilities that had a pattern similar to that of CMH; that is, declining operations, increased staff, and increased overtime between FY 2019 and 2024. For this analysis, the change in operations simply compared total operations at each facility in FY 2019 to that of FY 2024. For staffing, the committee estimated how the fill rate (AFN staffing standard/FTEs) at each facility changed between the 2 years. The change in overtime was based on the change in overtime/FTE at each facility between the 2 years. The committee also estimated the change in average TOP at each facility between
___________________
27 Overtime use per FTE of 12.5% was already high at CMH FY 2019. It increased to 15.4% in FY 2024.
28 This sentence was revised after release of the report to clarify that ATO’s policy guidance applies to modeling facility staffing standards.
the 2 years. The aggregate change represents the average across individual facilities. Including CMH, there were 71 facilities (23% of all facilities) that met these criteria. The results for these 71 facilities were an average 14% decline in operations, a 16% increase in AFN fill rate, an 82% increase in overtime/FTE and a 3% decline in average TOP. See Table 3-7.
The analysis was extended to estimate the number of facilities where operations and AFN fill rate both declined to determine the impact on overtime. Comparing FY 2019 and FY 2024, the committee found 59 facilities (19%) that met the criteria of decreasing operations, decreasing staffing, and increased overtime. The results for these facilities are an average 9% decrease in operations, a 12% decrease in AFN fill rate, a 109% increase in overtime, and a 2% decrease in average TOP. For this group, one would expect overtime to be higher than the first group. However, the magnitudes are similar between the two groups, indicating that something other than changes in staffing is driving overtime usage in these two groups.
To complete the analysis, the committee checked for the number of facilities with increased operations and either increases or decreases in staffing while also experiencing increased overtime. As shown in Table 3-7, the 51 facilities where both operations and staffing increased experienced a 106% increase in overtime (OT)/FTE. Note that this change in overtime use is comparable to facilities where operations decreased and staffing decreased. At facilities where operations increased but staffing decreased, overtime per FTE increased the most of any group (+138%), which would be expected given that operations were up and staffing was down. The facilities defined
| Changes Operations and Staffing | Number Facilities | Change in Operations | Change in AFN Fill Rate | Change in OT/FTE | Change in Facility TOP |
|---|---|---|---|---|---|
| Operations Down/Staffing Up | 71 | −14% | +16% | +82% | −3% |
| Operations Down/Staffing Down | 59 | −9% | −12% | +109% | −2% |
| Operations Up/Staffing Up | 51 | +15% | +11% | +106% | 0% |
| Operations Up/Staffing Down | 82 | +15% | −13% | +138% | −2% |
by the criteria applied in Table 3-7 totaled 263, thereby leaving out 50 facilities where overtime did not increase as a result of changes in operations or staffing. These facilities were excluded because the committee’s interest was in the facilities where OT/FTE increased and whether that could be explained by declining staffing levels.
The committee did not have data on actual shift scheduling practices; thus, it does not have a direct way to empirically test whether the increase in overtime since FY 2010 is primarily due to inefficient shift scheduling at facilities. The data the committee does have show that the CMH pattern of decreasing operations, increasing staffing, and increasing overtime is shared by 71 other facilities. When staffing at facilities declines with operations, the increase in overtime for these 59 facilities is greater than for the first group of 71 facilities, but not so much greater to rule out that inefficient scheduling is a plausible reason for increased overtime in both groups. When operations increase and staffing decreases (82 facilities), the largest increase in OT/FTE results. An increase in OT/FTE in these 82 facilities would be expected under these conditions, but not for the 71 facilities when operations decreased and staffing increased.
The pattern of decreased TOP for three groups along with no change for another are difficult to explain given their varied changes in operations and staffing. As noted, FAA staff had speculated that controller productivity might be increasing in the post FY 2010 period because of the regularization of traffic due to the improved flow control measures introduced by ATO. If true, then it is possible that many facilities are not scheduling their existing workforce efficiently, but they are nonetheless able to carry out their primary function (TOP) regardless of whether staffing increased or decreased relative to changes in operations. In any event, the above analysis of perplexing trends reinforces the lack of correlation between facility overtime and staffing levels shown in Table 3-3.
Alternate Approach to Validation
Given the inability to rely on overtime as a measure of workforce strain for all facilities, there may be other ways to partially validate fill rates with safety by examining their possible correlations with safety indicators discussed above, such as losses of separation and runway incursions and how they relate to shift schedules and position coverage at the time. Regarding the latter, when traffic gets heavy, managers with adequate staff will either open new positions to help controllers manage demand or add controllers to assist the principal controller. If staffing is inadequate to respond in these manners, it poses a safety risk. Rosekind et al. (2024) suggested that ATO evaluate available approaches for real-time monitoring of controllers staffing combined positions (Rosekind et al. 2024, 6). As also discussed above, some of the information that would be needed to test the staffing-safety
relationship is confidential or not otherwise available to the committee. However, because FAA has these data, it could examine whether such a partial validation might be possible. It may be that the close calls in the air or on the tarmac that could have resulted in an accident (Level A and B for runway incursions, and the least distance between aircraft in losses of separation in the air) are too few to provide meaningful results and may not provide consistent trends due to small numbers and changes in definitions. FAA could try using other indicators, such as those described above (MORs and NMAC reports). FAA could include Level C incursions with MORs, NMACs, and losses of separation if employee voluntary confidential reports are raising concerns about fatigue and errors in facilities that are staffed below the staffing standard range. It could also check whether the numerous fatigue rule violations in shift scheduling discussed above correlate with facility fill rates. Available data may not be sufficient for establishing statistical correlations for the purpose of model validation, but close investigation of operational errors and tracing them back to possible causes is an essential element of the safety assurance components of ATO’s fatigue risk and safety management systems.29
One other possible approach to model validation might be based on the amount of facility shift-level TOP and the frequency of individual controllers exceeding 2 hours on position without a break. Significant differences in measures of TOP between facilities adequately staffed and those that are more than 10 or 15% below their AFN staffing standards may help validate the models. However, establishing correlations may prove difficult for reasons such as inefficient shift scheduling, but would call for a deeper investigation. Even if inconclusive because of multiple possible causes, ongoing monitoring of shift-level TOP would serve as a useful safety assurance metric. The committee’s understanding is that ATO has individuals at headquarters monitoring this metric on an ongoing basis for facilities that it is concerned about. Such monitoring could be implemented on a more widespread basis and integrated into ATO’s safety assurance processes.
FINDINGS AND RECOMMENDATIONS
Finding 3-1: In response to Recommendation 2-5 of the NASEM 2014 report that FAA collect necessary data and conduct research to document the relationship between facility staffing levels and safety, FAA implemented a process of data gathering and analysis of close calls and other incidents that may have been caused or influenced by controller staffing, including fatigue from overwork. However, this study
___________________
29 See https://www.faa.gov/documentLibrary/media/Order/1030_7A.pdf and https://www.faa.gov/air_traffic/publications/media/ATO-SMS-Manual.pdf.
committee is unable to evaluate the safety impact of declining staffing on fatigue and safety because the incident data collected have not been used to support rigorous independent, peer-reviewed research to understand relationships between staffing and safety issues such as controller fatigue. Having considered the anecdotal evidence provided by FAA from its own data collection and analysis, the committee is not able to isolate effects from understaffing at many facilities from other fatigue-inducing factors that can affect safety, including the practice of scheduling shifts (such as 2-2-1) that are known to be associated with higher levels of worker fatigue.
Finding 3-2: The committee had intended to use facility reliance on overtime as a means of gauging strains on the workforce at facilities staffed below their staffing standard ranges. The increasing use of overtime across facilities, regardless of staffing levels, rendered such an approach untenable. Contrary to expectations, the analysis of the overtime and staffing data since FY 2010 did not reveal strong relationships among facility staffing levels, overtime, and regular hours worked per FTE. Similarly, other trends in regular hours worked and TOP are perplexing in light of increased traffic and reduced staffing. Shift scheduling practices other than compressed workweeks (2-2-1 schedule) may be affecting these relationships in ways that the committee was unable to document with actual shift scheduling data with the time and resources available to this project. Such practices deserve scrutiny by FAA.
Recommendation 3-1: The research recommended in 2014 by NASEM to relate staffing levels at individual facilities to safety should be conducted by FAA researchers. In doing so, FAA’s fatigue-related data and its possible relationships with staffing levels, shift scheduling practices, and on-position assignments exceeding 2 hours without a break should be evaluated by FAA and perplexing trends in regular hours worked and TOP explained. FAA should also make relevant fatigue and deidentified safety indicator data linked to staffing levels, shift schedules, and position assignments available for independent research on the relationships among facility staffing levels, shift scheduling, and safety.
Recommendation 3-2: FAA should examine whether the confidential safety indicator data it has could be used to test possible relationships between facility staffing levels and safety incidents in ways that could help validate its facility staffing models.
Finding 3-3: FAA’s adoption of extended rest periods between controller shifts advances the goal of reducing fatigue in the ATC workforce. The ultimate impact on facility staffing standards from adoption of the policy changes is unclear but important to understand.
Recommendation 3-3: FAA should conduct a thorough analysis of the impact on facility staffing standards of the effects of extended rest periods and termination of compressed and backward rotating shift schedules, recalibrate its staffing models accordingly, and report the results in a future annual Controller Workforce Plan.
Finding 3-4: A robust FRMS with strong ongoing monitoring and evaluation components are essential elements of minimizing fatigue and ensuring safety in 24/7 operations.
Recommendation 3-4: FAA should follow up on the NASEM 2014 report recommendation on FRM and those of Rosekind et al. (2024) and implement an effective FRMS with a robust ongoing evaluation component and keep abreast of developments in validated alertness testing and monitoring that might be added to it in the future. Congress should resource FAA for this purpose and monitor for timely follow-through.
Finding 3-5: FAA made good-faith efforts to acquire and implement shift scheduling software that is intended to reduce the incidence of controller fatigue, as recommended in the NASEM 2014 report. The recommended implementation would incorporate prescriptive fatigue rules and policies and best practices in FRM while seeking to optimize the use of existing staff. However, a joint FAA-NATCA commitment in the 2016 CBA to implement OPAS foundered when NATCA withdrew and ATO ultimately concurred. It is unclear to the committee whether it was impossible or simply challenging to implement OPAS. It was used successfully at two Centers. The inability to date to implement a software package that accomplishes the goals means that facility managers do not have available to them a robust shift scheduling tool compliant with best practices in FRM.
Recommendation 3-5: FAA should identify and evaluate reasons why available shift scheduling software to reduce fatigue and utilize staff efficiently has not proven adaptable to local scheduling practices at FAA facilities. The evaluation should include determining whether difficulties in implementing available scheduling software that apparently led to its abandonment is justified by the cost of being unable to implement shift schedules at all facilities that are fully consistent with fatigue rules
and best practices in FRM, while also efficiently scheduling existing staff.
Recommendation 3-6: FAA should procure or develop and implement shift scheduling software designed to reduce the incidence of controller fatigue while also making efficient use of available controller resources, as recommended in the NASEM 2014 report. This should include a robust requirements and Request for Proposals process that includes a build (custom) versus buy (COTS) decision. This process should also include implementation and organizational change management support for greatest likelihood of successful implementation and adoption. Congress should resource FAA for this purpose and monitor timely follow-through.
REFERENCES
FAA (Federal Aviation Administration). 2013. FAA. 2013. Fatigue Risk Management Systems for Aviation Safety, Advisory Circular 120-103A. https://skybrary.aero/sites/default/files/bookshelf/4032.pdf.
FAA. 2022. “Comprehensive Review of Controller Overtime and Other Operational Considerations at Jacksonville Air Route Traffic Control Center (ZJX ARTCC).” Unpublished white paper.
FAA. 2024a. “Air Traffic by the Numbers.” https://www.faa.gov/air_traffic/by_the_numbers/media/Air_Traffic_by_the_Numbers_2024.pdf.
FAA. 2024b. “CMH Schedule Analysis.” Unpublished white paper.
ICAO. 2018. Annex 6 to the Convention on International Civil Aviation, Operation of Aircraft, Part 1, Chapter 4. https://www.icao.int/safety/CAPSCA/PublishingImages/Pages/ICAO-SARPs-(Annexes-and-PANS)/Annex%206.pdf.
Martins, N., et al. 2021. Fatigue Monitoring Through Wearables: A State-of-the-Art Review. Frontiers in Physiology. https://pmc.ncbi.nlm.nih.gov/articles/PMC8715033/pdf/fphys-12-790292.pdf.
NRC (National Research Council). 2014. The Federal Aviation Administration’s Approach for Determining Future Air Traffic Controller Staffing Needs, TRB Special Report 314. The National Academies Press. https://doi.org/10.17226/18824.
OIG (U.S. Department of Transportation Office of Inspector General). 2018. “FAA Remains Several Years Away from Having a Standardized Scheduling Tool.”
OIG. 2023. “FAA Faces Controller Staffing Challenges as Air Traffic Operations Return to Pre-Pandemic Levels at Critical Facilities.”
Rosekind, M., L. Flynn-Evans, and C. Czeisler. 2024. “Assessing Fatigue Risk in FAA Air Traffic Operations: Report by Scientific Expert Panel on Air Traffic Controller Safety, Work Hours, and Health.”
Sprajcer, M., et al. 2022. “How Effective Are Fatigue Risk Management Systems (FRMS)? A Review.” Accident Analysis and Prevention 165.
This page intentionally left blank.
