Ada 450964

Safety Analysis of TCAS on Global Hawk using Airspace
Encounter Models
by
2Lt. Thomas B. Billingsley
B.S., Aeronautical Engineering (2004)
United States Air Force Academy
Submitted to the Department of Aeronautics and Astronautics

in Partial Fulfillment of the Requirements for the Degree of
Master of Science in Aeronautics and Astronautics
at the
Massachusetts Institute of Technology
June 2006
Signature of Author.......................................................................................................................
Department of Aeronautics and Astronautics
May 26, 2006
Certified by ...................................................................................................................................
Dr. James K. Kuchar
Assistant Group Leader, Group 42, MIT Lincoln Laboratory
Thesis Supervisor
Certified by ...................................................................................................................................
Dr. Jonathan P. How
Associate Professor of Aeronautics and Astronautics
Thesis Supervisor
Accepted by ..................................................................................................................................
Dr. Jaime Peraire
Professor of Aeronautics and Astronautics
Chair, Committee on Graduate Students
Form Approved
Report Documentation Page OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.
1. REPORT DATE 2. REPORT TYPE 3. DATES COVERED

01 JUN 2006 N/A -
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Safety Analysis of TCAS on Global Hawk using Airspace Encounter 5b. GRANT NUMBER
Models
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
AFIT/CIA, Bldg 125 2950 P Street WPAFB, OH 45433 11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT

Approved for public release, distribution unlimited
13. SUPPLEMENTARY NOTES
The original document contains color images.
14. ABSTRACT
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE
UU 66
unclassified unclassified unclassified
Standard Form 298 (Rev. 8-98)

Prescribed by ANSI Std Z39-18
Safety Analysis of TCAS on Global Hawk using Airspace
Encounter Models
by
Thomas B. Billingsley
Submitted to the
Department of Aeronautics and Astronautics
on May 26, 2006,
in Partial Fulfillment of the requirements for the degree of

Master of Science in Aeronautics and Astronautics
Abstract
This work is sponsored by the Air Force under Air Force Contract #FA8721-05-C-0002.
The views expressed in this article are those of the author and do not reflect the official policy or
position of the United States Air Force, Department of Defense, or the U.S. Government.
The U.S. Air Force’s RQ-4 Global Hawk unmanned aerial vehicle (UAV) is a high
altitude, long endurance aircraft used for surveillance and reconnaissance. Because of the
potential for close proximity to manned aircraft in civil airspace, collision avoidance is a major
concern, and the Air Force is seeking to equip Global Hawk with the Traffic Alert and Collision
Avoidance System (TCAS) to reduce the probability of mid-air collision. Currently, Global
Hawk is equipped with a Mode S transponder and uses chase aircraft, ground observers and/or
ground radar contact to comply with the collision avoidance requirement.
In order to evaluate TCAS effectiveness, a fast-time simulation tool has been developed
at MIT Lincoln Laboratory that computes the mean probability of a near mid-air collision for a
large number of close encounters between two aircraft. Airspace encounter models enable sets
of encounters to be simulated that are statistically representative of the aircraft encounters that
actually occur in the airspace. The TCAS logic is implemented in the simulation tool and the
aircraft responses during the encounters, with and without TCAS, are simulated in parallel. By
observing measured vertical miss distance at the closest point of approach between the two
aircraft, it is possible to quantify the reduction in collision risk provided by TCAS, termed the
risk ratio.
Global Hawk’s flight characteristics differ from a conventional aircraft. Its mission
profile through civil airspace includes slow, steep climbs and descents, and shallower turns than
a conventional aircraft. Its vertical acceleration and climb rate limits can hinder its response to a
TCAS resolution advisory (RA). Communication latency also may occur. For this thesis,
encounter models were developed that reflect Global Hawk’s flight characteristics. The new
encounter models were then implemented in the simulation tool, and millions of encounters
between Global Hawk and a conventional aircraft were simulated. These encounters were
compared against encounters between two conventional aircraft to observe how Global Hawk’s
flight characteristics changed the effectiveness of TCAS.
2
Assuming a standard pilot response to TCAS RAs, TCAS provided a significant safety
improvement to Global Hawk over a Mode S transponder alone, yielding risk ratios in the range
of 0.003 to 0.079. Global Hawk’s flight characteristics generally caused a decrease in TCAS
effectiveness from the original encounter models. Encounters were also simulated where Global
Hawk’s response to TCAS RAs was delayed by increasing amounts to simulate communication
latency. A delay of approximately 15 seconds was tolerable before TCAS provided less safety
than a Mode S transponder alone.
Thesis Supervisor: Dr. James K. Kuchar

Title: Assistant Group Leader, Group 42
MIT Lincoln Laboratory
Thesis Supervisor: Dr. Jonathan P. How

Title: Associate Professor of Aeronautics and Astronautics
3
Table of Contents
Abstract ......................................................................................................................................2
Table of Contents........................................................................................................................4
List of Figures.............................................................................................................................6
List of Tables ..............................................................................................................................8
List of Acronyms ........................................................................................................................9
Acknowledgments.....................................................................................................................10
1. Introduction and Background ............................................................................................11
1.1. Thesis Objectives and Approach ................................................................................11
1.2. Global Hawk Collision Avoidance Requirement ........................................................11
1.3. Traffic Alert and Collision Avoidance System ...........................................................16
1.3.1. TCAS Description..............................................................................................16
1.3.2. TCAS on UAVs .................................................................................................18
2. Collision Avoidance Safety Analysis.................................................................................22
2.1. Safety Analysis Approach..........................................................................................22
2.1.1. Dynamic Simulation ..........................................................................................23
2.1.2. Simulation Overview .........................................................................................26
2.1.3. Encounter Generation.........................................................................................29
2.2. Global Hawk Encounter Model Modification.............................................................36
2.2.1. Airspeed.............................................................................................................38
2.2.2. Vertical Maneuver Probability ...........................................................................40
2.2.3. Climb Rate.........................................................................................................44
2.2.4. Descent Rate......................................................................................................45
2.2.5. Bank Angle ........................................................................................................48
2.2.6. Ceiling ...............................................................................................................48
2.2.7. Vertical Acceleration .........................................................................................49
2.3. Summary ...................................................................................................................51
3. Simulation Results.............................................................................................................53
3.1. Test Conditions..........................................................................................................53
3.2. Results.......................................................................................................................54
4
3.2.1. ICAO Encounter Model .....................................................................................55
3.2.2. European Encounter Model................................................................................58
4. Conclusions and Recommendations...................................................................................62
4.1. Global Hawk Flight Profile ........................................................................................62
4.2. TCAS RA Response Latency.....................................................................................63
4.3. Global Hawk Vertical Acceleration and Climb Rate Limits .......................................63
4.4. Summary and Recommended Research......................................................................64
References ................................................................................................................................66
5
List of Figures
Figure 1-1: RQ-4 Global Hawk .................................................................................................12
Figure 1-2: Global Hawk Communications and Data Link Networks.........................................13
Figure 1-3: Global Hawk Pilot Interface (Training Simulator)...................................................14
Figure 1-4: Example Arrival and Departure Routes from Beale AFB (Nominal Climb Profile
Altitudes Shown)...............................................................................................................15
Figure 1-5: TCAS Coordination Levels .....................................................................................17
Figure 1-6: TCAS Display [6] ...................................................................................................18
Figure 1-7: Prototype Global Hawk TCAS Display [4]..............................................................20
Figure 2-1: Risk Ratio ...............................................................................................................24
Figure 2-2: Example Relative Altitude Error Distribution (Aircraft at 18,000 ft) .......................25
Figure 2-3: Simulation Schematic .............................................................................................27
Figure 2-4: Example Scenario Output........................................................................................29
Figure 2-5: Encounter Model Parameter Dependency................................................................30
Figure 2-6: Encounter Model Altitude Layers (Probabilities in Brackets) ..................................31
Figure 2-7: European Encounter Model Vertical Maneuver Classes ..........................................32
Figure 2-8: Aircraft Encounter Parameters ................................................................................35
Figure 2-9: Global Hawk Encounter Model Parameter Dependency ..........................................37
Figure 2-10: Global Hawk Airspeed Envelope [12] ...................................................................38
Figure 2-11: Global Hawk Airspeed Profile Based on European Encounter Model....................39
Figure 2-12: Initial Airspeed Histograms, Layer 3.....................................................................40
Figure 2-13: Vertical Maneuver Probabilities, European Model, Layer Three............................43
Figure 2-14: Vertical Maneuver Probabilities, Global Hawk Modified European Model, Layer
Three.................................................................................................................................43
Figure 2-15: Global Hawk Climb Rate [3].................................................................................44
Figure 2-16: Global Hawk Climb Rate vs. Altitude, Gross Weight = 26,750 lbs ........................45
Figure 2-17: Global Hawk Descent Rate [3] ..............................................................................46
Figure 2-18: Global Hawk Initial Vertical Rate Histogram, All Altitude Layers ........................47
Figure 2-19: Conventional Aircraft Initial Vertical Rate Histogram, All Altitude Layers ...........47
Figure 2-20: Global Hawk Bank Angle Schedule [3].................................................................48
6
Figure 2-21: Global Hawk Vertical Acceleration Profile, European Encounter Model...............49
Figure 2-22: Vertical Maneuver Example..................................................................................51
Figure 2-23: Summary of Global Hawk Effects on Encounter Generation and Simulation.........52
Figure 3-1: Risk Ratio Comparison, ICAO Model vs. Global Hawk-Modified ICAO Model .....56
Figure 3-2: Risk Ratio vs. RA Latency, Global Hawk Modified ICAO Model...........................57
Figure 3-3: Risk Ratio Comparison, European Model vs. Global Hawk-Modified European
Model................................................................................................................................59
Figure 3-4: Risk Ratio vs. RA Latency, Global Hawk Modified European Model .....................60
Figure 3-5: Running Risk Ratio, Altitude Layer 3, Conventional TCAS Response ....................61
7
List of Tables
Table 2-1: Vertical Maneuvers ..................................................................................................33
Table 2-2: Example Vertical Maneuver Parameters, Altitude Layer 1: P[A1,A6,0] .....................33
Table 2-3: Example Global Hawk Vertical Maneuver Probabilities ...........................................41
Table 3-1: Simulation Parameter Combinations.........................................................................53
8
List of Acronyms
ACAS Airborne Collision Avoidance System
ACASA Airborne Collision Alerting System Analysis
ACC Air Combat Command
AFB Air Force Base
ATC Air Traffic Control
CPA Closest Point of Approach
COA Certificate of Waiver or Authorization
FAA Federal Aviation Administration
HMD Horizontal Miss Distance
ICAO International Civil Aviation Organization
IFR Instrument Flight Rules
KCAS Knots Calibrated Airspeed
KTAS Knots True Airspeed
NAS National Airspace System
NMAC Near Mid-Air Collision
RA Resolution Advisory
ROA Remotely Operated Aircraft
SCRSP Surveillance and Conflict Resolution Systems Panel
TCAS Traffic Alert and Collision Avoidance System
UAV Unmanned Aerial Vehicle
VMD Vertical Miss Distance
VSI Vertical Speed Indicator
9
Acknowledgments
This work is sponsored by the Air Force under Air Force Contract #FA8721-05-C-0002.
The views expressed in this article are those of the author and do not reflect the official policy or
position of the United States Air Force, Department of Defense, or the U.S. Government.
I would like to thank MIT Lincoln Laboratory Division 4, under the leadership of Dr.
Bob Shin, for supporting my research during my two years here. I would also like to extend my
appreciation to Group 42, Air Surveillance Systems, under the direction of Jim Flavin, for the
research assistantship opportunity. My experience at Lincoln Laboratory has been thoroughly
enjoyable. I heard it said that there was an “expert for everything” here at Lincoln, and I found
that to be true as I researched and wrote this thesis.
I want to extend my heartfelt appreciation to Dr. Jim Kuchar, Assistant Group Leader of
Group 42, for his patience, insight and guidance as my thesis supervisor throughout my time here
at the Lab. The learning curve was steep, but you made sure I was onboard and up to speed
before diving into the project. The feedback you gave during my research and thesis writing
process was timely, reliable and very helpful in producing quality work. Thank you for the time
you invested to ensure that I receive excellent advice and support.
I also want to thank Val Heinz of Group 42 for the well-informed advice he gave me as I
selected a topic for research and completed this thesis. I appreciate your taking the time to help
me even as you were switching groups.
Numerous members of the Group 42 staff assisted me while I researched and wrote this
thesis. In particular, I would like to thank Brian O’Donnell, Kathy Sinclair, Maggie Herring,
Ann Drumm, and Barbara Chludzinksi for their help.
My thanks go to Roger Elstun and Tee Mans at Air Combat Command, Langley Air
Force Base, VA, for their sponsorship of the work that went into this thesis.
My wife’s constant and loving support these past two years during my academics and
research has meant so much. I appreciate your confidence in me, and I want to thank you for
selflessly caring for me and for our daughter. I love you!
I am grateful to God for the opportunity He has given me to complete a portion of my
education here at MIT. It has been an enriching experience which I will always remember
fondly.
10
1. Introduction and Background
1.1. Thesis Objectives and Approach

The U.S. Air Force Air Combat Command is seeking to operate the RQ-4 Global Hawk
unmanned aerial vehicle (UAV) in the National Airspace System (NAS) as part of its mission
profile. The Federal Aviation Administration (FAA) requires UAVs to fly with an “equivalent
level of safety” as that of manned aircraft [1]. Global Hawk operations are currently based on a
Certificate of Waiver or Authorization (COA) issued by the FAA that requires a five-day
advance notice to the FAA, along with the use of chase aircraft, ground observers and/or ground
radar contact to ensure that Global Hawk avoids other aircraft. To improve operational
flexibility, the Air Force is interested in using assets onboard Global Hawk to provide a collision
avoidance capability. As part of this process, Air Combat Command (ACC) has been funding a
study to quantify the level of safety potentially provided by equipping Global Hawk with the
Traffic Alert and Collision Avoidance System (TCAS).
While numerous levels of collision avoidance protection exist, this thesis focuses only on
the effect of Global Hawk’s flight characteristics and communication latency on the level of
safety provided by TCAS in collision avoidance. All other effects on Global Hawk collision
avoidance are assumed to be similar to manned aircraft. These effects may include procedural
steps and aircraft separation provided by the air traffic management system. This thesis also
does not discuss the operational flight test and evaluation that would be necessary to obtain
certification to equip Global Hawk with TCAS.
1.2. Global Hawk Collision Avoidance Requirement

The development of unmanned aerial vehicles (UAVs), especially in the United States,
has increased greatly in recent years. During the four decades of the Cold War, all but 4 of 16
major U.S. military UAV efforts were cancelled, but in the decade since, only 4 of 12 have been
cancelled, and three of those have been brought back to life [2]. Several factors have caused this
streak of successful UAV programs. First, highly accurate navigation is now possible due to
Global Positioning System (GPS) satellites and ring laser gyros, and quicker microprocessors
have expanded mission capabilities for military UAVs. In addition, decreased hostility of over-
theater airspace has enabled UAVs to be utilized where they would not have been before. UAVs
11
have been deployed in each of the United States’ last five major regional conflicts (Persian Gulf,
Bosnia, Kosovo, Afghanistan and Iraq) in increasing numbers, types and roles. During the
conflict in Iraq, the missions performed by UAVs included surveillance, reconnaissance, strike,
target designation, diversionary decoy, and base security. Because of their success in these
conflicts as well as a need for homeland defense applications, funding for UAVs has increased
over the past decade and will likely continue to increase in the future.
One of the most extensively used UAVs currently in operation by the U.S. Air Force is
the RQ-4 Global Hawk. The RQ-4 is a self-deploying, long-dwell, high-altitude unmanned
military reconnaissance aircraft produced by the Northrop Grumman Corporation and assigned
to ACC. During its limited use, Global Hawk has already passed several milestones. The RQ-
4A model was deployed to both Operation Enduring Freedom and Operation Iraqi Freedom,
during which it flew 16 day-long missions over Iraq and collected 25 percent of all the airborne
reconnaissance imagery taken during the conflict [2]. Development has already begun on the
RQ-4B, an upgraded model with an increased payload capacity and double the power previously
generated onboard. With a wingspan of 116 feet and a maximum takeoff weight of 26,750 lb,
Global Hawk is the largest UAV in operation by the military. The aircraft is equipped with
electro-optical (EO), infrared (IR), and synthetic aperture radar (SAR) sensors for use in
intelligence, surveillance and reconnaissance (ISR) [3]. The long-term plans are for Global
Hawk to take over the ISR missions currently handled by the Air Force’s U-2 fleet based at
Beale AFB. Figure 1-1 shows a simple diagram of Global Hawk.
Figure 1-1: RQ-4 Global Hawk
12
Figure 1-2 shows a schematic of the major data links and communication networks upon
which Global Hawk command and control are based. It can be seen from Figure 1-2 that Global
Hawk’s command and control network has numerous components. Global Hawk is operated
remotely from a common ground station (CGS) which consists of a launch and recovery element
(LRE), a mission control element (MCE), and supporting equipment [3].
UHF SATCOM
INMARSAT
Ku-band SATCOM
UHF
CDL CDL
Air
Traffic
Control
Mission Control Element Launch and Recovery Element
Figure 1-2: Global Hawk Communications and Data Link Networks
Command and control for Global Hawk is accomplished using both Line of Site (LOS)
radio communication and beyond line of site (BLOS) Satellite Communication (SATCOM) data
links. The communication system includes common data link (CDL), ultrahigh frequency (UHF)
radio, International Maritime Satellite (INMARSAT), Ku-band and UHF SATCOM. Takeoff
and landing are accomplished through LOS link between the aircraft and the LRE, and control is
handed over to the MCE for climb, cruise, mission operations, and descent through LOS and
BLOS links.
13
Figure 1-3 shows Global Hawk’s pilot control interface from a training simulator. Global
Hawk is operated using a computer mouse and standard keyboard. Therefore, the pilot is unable
to directly input flight controls such as pitch and roll rate, as he would with a control stick or
yoke. However, it is possible to override the preprogrammed flight plan by modifying the
mission waypoints or commanding airspeed, heading, altitude or vertical velocity.
Figure 1-3: Global Hawk Pilot Interface (Training Simulator)
Global Hawk’s missions may require flight operations out to 1200 miles to a mission
area, up to 24 hours on-station, and then return flight to the operating base [3]. Domestically,
Global Hawk is currently fielded at Edwards Air Force Base near Lancaster, California, as well
as the U.S. Air Force’s 12th Reconnaissance Squadron at Beale Air Force Base near Yuba City,
California. Global Hawk requires flight operation in the NAS to achieve its operational
14
objectives, which may include transition through civil or uncontrolled airspace near its operating
location. Figure 1-4 shows a map of the area surrounding Beale AFB (shown in the upper-left
corner of the figure), along with example departure and arrival routings for Global Hawk. Also
shown in Figure 1-4 are several points at which the aircraft would reach certain altitudes during
its climb-out phase, assuming a nominal climb rate. The example routes pass through the Class
C airspace surrounding Beale (the solid magenta circles in the upper-left portion of the figure), as
well as in the vicinity of several other Class C, D, and uncontrolled airfields.
40,000 ft
10,000 ft
20,000 ft 30,000 ft
Figure 1-4: Example Arrival and Departure Routes from Beale AFB (Nominal Climb Profile Altitudes
Shown)
Because of the requirement to fly through this populated airspace in the NAS, a collision
avoidance concern exists for the multiple Global Hawk aircraft slated for and currently in
operation.
According to FAA Order 7610.4, the U.S. military’s remotely operated aircraft (ROA)
flying in the NAS are required to be equipped with a collision avoidance system that provides an
“equivalent level of safety, comparable to see-and-avoid requirements for manned aircraft” [1].
15
Various methods for unmanned collision avoidance have been considered, and include radar
observation, forward or side looking electro-optical/infrared (EO/IR) cameras, electronic
detection systems, visual observation from one or more ground sites, monitoring by a patrol or
chase aircraft, or a combination of one or more of these methods.
Initial airworthiness for Global Hawk to fly in civil airspace was established in March
1999 through flight tests at Edwards AFB, California, and it has already flown over 5000 hours
in civil airspace worldwide [4]. In August 2003, Global Hawk was the first UAV to receive a
national COA from the FAA to fly routinely in U.S. national airspace [5]. Although an
Instrument Flight Rules (IFR) flight plan and an emergency plan are required to be briefed to the
FAA before takeoff, the COA represents significant progress in obtaining routine access to civil
airspace because it shrinks the processing and approval timelines from 60 to as few as five days.
In accordance with the COA, Global Hawk’s operators currently coordinate all flights into the
NAS at least five working days in advance with the local FAA Enroute Center [5]. Chase
aircraft and ground radar are used to assist in collision avoidance, and communication is required
between the Global Hawk pilot and air traffic control (ATC).
1.3. Traffic Alert and Collision Avoidance System
1.3.1. TCAS Description

A series of mid-air collisions over the course of thirty years (1956-1986) led the FAA to
initiate the development and implementation of the Traffic Alert and Collision Avoidance
System, or TCAS [6]1. TCAS was developed through extensive analysis and flight evaluation,
and it has served a successful role in reducing the risk of midair collisions between aircraft.
TCAS operates independently of ground-based ATC and is effective for a wide range of
aircraft types. TCAS relies on the same radar beacon transponders that are used to communicate
with ground-based ATC radar, and therefore cannot protect against non-transponder equipped
aircraft.
TCAS operates by interrogating the transponders of aircraft in its vicinity once per
second, and based on the replies received, tracks the slant range, altitude, and bearing of
surrounding air traffic. Based on several replies over time, TCAS then calculates the time to
1
Internationally, TCAS is known as the Airborne Collision Avoidance System, or ACAS.
16
reach the closest point of approach (CPA) with the intruding aircraft by dividing the slant range
by the closure rate. If a tracked aircraft becomes a threat, TCAS issues an alert to the pilot of the
equipped aircraft. Two types of alerts may be issued by TCAS: traffic advisories (TAs), which
alert the pilot to an intruder aircraft and assist in the visual search for traffic, and resolution
advisories (RAs), which recommend vertical maneuvers that will increase vertical separation
between the two aircraft while minimizing perturbations to the existing flight path. If both
aircraft are equipped with TCAS, coordinated avoidance maneuvers are performed to ensure
compatibility. Figure 1-5 shows the levels of coordination provided by TCAS.
TCAS-Equipped
Detection and
Maneuver Coordination
TCAS-Equipped
Detection
Transponder-Equipped
Figure 1-5: TCAS Coordination Levels
The TCAS pilot interface includes two cockpit displays – the traffic display and the RA
display. These two displays may either be separate or incorporated into one single unit. Figure
1-6 shows an example traffic display and RA display incorporated into one unit. The
information shown by the traffic display includes the position, vertical speed and altitude of
other aircraft relative to its own aircraft, with increasing threat levels depicted by different
symbol colors and shapes. The RA display provides the pilot with the appropriate resolution
maneuver by displaying either the vertical rate or pitch angle to fly or avoid. The RA display
shown in Figure 1-6 is implemented in a vertical speed indicator (VSI). The RA shown in Figure
1-6 is a “climb” RA at a vertical rate of 1500 to 2000 ft/min.
17
Figure 1-6: TCAS Display [6]
As the safety benefits of installing TCAS became evident, several countries participating
in the International Civil Aviation Organization (ICAO) began to mandate the carriage of TCAS
on aircraft exceeding a certain weight or passenger capacity. The United States was the first
member of ICAO to require TCAS in 1993, followed by all European countries, Argentina,
Australia, Chile, Egypt, India, and Japan. Following a midair collision between a Saudi Boeing
747 and a Kazakhstan Ilyushin 76 in 1996, ICAO proposed worldwide mandatory ACAS II (RA-
capable) carriage on all aircraft above certain weight or passenger seat limits, beginning in 2003.
Starting on 1 January 2005, a worldwide ICAO requirement for ACAS II equipage was
mandated for all aircraft with more than 19 passenger seats or with a maximum takeoff weight of
more than 5,700 kg.
1.3.2. TCAS on UAVs

Because it has already been developed and is currently operational, TCAS presents a
potential near-term safety benefit for Global Hawk. The U.S. Air Force ACC established a
requirement to equip RQ-4 aircraft with TCAS in order to increase the safety of operations to
comply with FAA Order 7610.4. Because Global Hawk’s operator is able to control vertical
velocity, it would be possible to provide TCAS traffic and RA displays to the operator and allow
18
him to conduct traffic avoidance maneuvers to comply with TCAS RAs. The question, then, is
whether such operation would be safe.
Several concerns exist in equipping UAVs with TCAS. UAVs were not originally
considered when the various TCAS and ACAS mandates were drafted by ICAO. Although
Global Hawk is above the weight requirements for ACAS II equipage, TCAS on UAVs was not
intended to be part of the ICAO mandate. The safety studies that were conducted to certify
TCAS all assumed that the aircraft would have a pilot onboard, and therefore TCAS would
provide a benefit in terms of increasing situational awareness and assisting in the visual search
for air traffic. Since UAVs do not have a pilot onboard to look outside the aircraft, situational
awareness does not apply in the same way as manned aircraft. Due to the bearing error and
update rate of TCAS, the FAA and ICAO have stated that the TCAS display alone is not
sufficient to provide the operator with enough situational awareness to avoid the threat. The
inability to perform visual acquisition means that the traffic display information cannot be
corroborated by the UAV operator, and therefore the certification authorities are concerned about
its use.
In addition, because TCAS operates by interrogating transponders on equipped aircraft,
non-cooperative traffic, or aircraft without transponders, are not tracked by TCAS. Aircraft are
only required to be equipped with altitude-reporting transponders in Class A, B, and C airspace,
and Class E airspace above 10,000 feet. In the low-altitude Class E and uncontrolled airspace
through which UAVs may fly, TCAS would be unable to detect unequipped intruders, and there
is currently no capability for a UAV pilot to visually acquire these types of threats.
From prior simulation analysis, it is also known that slow pilot responses to resolution
advisories can increase the risk of collision to greater than what it would be without TCAS.
Because of the BLOS nature of Global Hawk’s control strategy at long ranges, a certain amount
of communication latency may exist between the aircraft and the pilot on the ground. This may
cause delayed responses to RAs if the pilot were to command the aircraft to execute avoidance
maneuvers, and safety could possibly be degraded.
Rather than equipping UAVs with a TCAS II system capable of generating RAs, an
additional option under consideration would be to equip UAVs with TCAS in TA-only mode,
which would alert the pilot on the ground of an intruder aircraft but not provide a collision
avoidance maneuver recommendation. Figure 1-7 shows a prototype TCAS traffic display for
19
Global Hawk, based on this concept. Equipping UAVs with TCAS in TA-only mode is under
debate since pilots are not authorized to respond to TAs alone, and there is no capability to
visually acquire the threat.
Figure 1-7: Prototype Global Hawk TCAS Display [4]
Another option would be to allow TCAS to take control of the UAV and automatically
execute the avoidance maneuver. This level of automation would effectively take the UAV pilot
out of the collision avoidance process. Numerous studies have shown that the level of safety
provided by TCAS increases with the accuracy and consistency of RA responses. Safety would
be improved with an immediate and correct TCAS response each time an RA is issued, if there
are no system failures and all the information is perfectly accurate and complete. However,
TCAS has not yet been integrated with an autopilot system and is not certified for automatic
maneuver response.
Recently, more consideration has taken place as to the steps needed to evaluate the
performance of TCAS on UAVs, serving as a topic of discussion at recent ICAO Surveillance
and Conflict Resolution Systems Panel (SCRSP) meetings. The original development of TCAS
required extensive tests, analyses, computer simulations and operational evaluations, at a cost of
approximately $400 million in FY2001 dollars [7]. Such development has not occurred for
UAVs, and it is clear that a rigorous effort is needed to determine the necessary requirements and
procedures for operation of TCAS on UAVs. Until this is accomplished, the safest way to
operate UAVs in the NAS, as recommended by ICAO, is to equip them with a 25-foot
20
altitude-reporting Mode S transponder. The Global Hawk program office has responded to the
requirement for additional research and testing by funding ICAO-recommended TCAS safety
studies for Global Hawk. In addition, flight tests are planned for late 2006 which will evaluate
the performance of TCAS onboard Global Hawk.
21
2. Collision Avoidance Safety Analysis
2.1. Safety Analysis Approach

Historically, the process of conducting a collision avoidance safety analysis for a manned
aircraft has included three complementary techniques. The first technique is a static analysis that
uses a “fault tree” which identifies and analyzes failures in the entire collision avoidance
operating environment. The second method is to use dynamic simulation to evaluate TCAS
performance computationally. The third main technique is analysis of TCAS performance on an
aircraft accomplished through flight testing. The combination of these three techniques allows
safety to be quantified and a collision avoidance system to be thoroughly evaluated. Safety
analysis can be conducted in a similar manner for UAVs.
An event tree, or fault tree, provides both a quantitative and qualitative means to identify
and analyze failure modes in the end-to-end collision avoidance process. All possible means by
which a collision could occur are identified and organized into a logical structure. These failure
modes are not limited to TCAS alone. They may include air traffic control, radio, navigation or
communication system failures, pilot errors, or aircraft flight control system failures. After the
fault tree is generated, each process leading to failure can be studied and the root causes and
interactions identified. System failures can have a significant impact on risk, and so this is an
important area of study.
The dynamic simulation technique involves robust computational simulation of aircraft
encounters and subsequent statistical analysis of the results. The development of TCAS for
piloted aircraft included millions of computer simulations to quantify the protection provided by
TCAS in a wide range of traffic encounters. Safety studies were performed using updated TCAS
algorithms, to estimate how TCAS could further reduce the risk of mid-air collisions. This
method assumes that TCAS works properly. For the purposes of simulation, this means TCAS
receives accurate information, such as range, bearing and altimetry information from the intruder
aircraft, and there are no power supply failures or general faults. The simulation requires an air
traffic model, as well as models for pilot and aircraft response and the TCAS logic. The purpose
of dynamic simulation is to predict TCAS performance over the entire expected life of the
22
system. In other words, a wide range of possible scenarios in which a TCAS-equipped aircraft
may encounter another aircraft are considered.
The third technique includes operational evaluations that may be conducted which test
TCAS performance in flight. Information acquired from the flight tests may be fed back into the
fault tree or dynamic simulation. The TCAS logic has occasionally been updated based on
results from these evaluations.
For this thesis, a dynamic simulation was developed and executed to evaluate TCAS
performance on Global Hawk. In prior TCAS studies, ground-based radar data were collected
and examined to find encounters between two aircraft where TCAS would alert the pilots of a
potential mid-air collision [8]. Structured airspace encounter models, or traffic models, were
then developed which reflected the statistical characteristics of the airspace in which these radar
data were collected. Millions of encounters between two aircraft were then generated based on
this statistical model and simulated using the TCAS logic implemented in computational tools.
For the Global Hawk study, existing airspace encounter models were modified to reflect Global
Hawk’s flight characteristics. Encounter models are discussed in further detail in Section 2.1.3.
2.1.1. Dynamic Simulation

A Monte Carlo simulation was used to evaluate the effectiveness of TCAS in lowering
the risk of a near mid-air collision (NMAC). NMACs are defined as encounters where the
vertical miss distance (VMD) is 100 feet or less and the horizontal miss distance (HMD) is 500
feet or less. In general, during the simulation a climbing or descending maneuver may be
performed by one or both aircraft in response to TCAS advisories to change the VMD of the
aircraft at the closest point of approach (CPA).
The effect of TCAS on collision risk is expressed through the “risk ratio,” which is
defined as the probability of an NMAC with TCAS divided by the probability of an NMAC
without TCAS. Although TCAS cannot entirely eliminate the collision risk, it can reduce the
probability of an NMAC to less than what it would have been without TCAS. A risk ratio of less
than one indicates that TCAS is effective in improving safety by lowering the risk of a collision.
Figure 2-1 graphically shows how risk ratio is computed. The left graph in Figure 2-1
schematically indicates the probability of an NMAC without TCAS, p, based on a large number
of Monte Carlo simulations. The right graph indicates the probability of an NMAC with TCAS
23
when run over the same set of encounters. In this example, 90% of the pre-existing collision risk
is resolved by TCAS, but two components of risk remain: unresolved risk and induced risk.
Unresolved risk represents those encounters where TCAS fails to remove a pre-existing risk, and
induced risk is the potential for TCAS to cause a collision that did not exist in its absence.
Without TCAS With TCAS

Resolved
NMAC
P(NMAC) = p 0.9p
No NMAC No NMAC
NMAC Unresolved
NMAC
0.1p
Induced
NMAC
0.067p
Figure 2-1: Risk Ratio
Equation 2-1 shows how the unresolved and induced components of risk ratio are
calculated. The total risk ratio is the union of these two components. In the example from
0.167 p
Figure 2-1, the risk ratio is , or 16.7%. The unresolved risk ratio component is 10%, and
p
the induced component is 6.7%.
P[ NMAC with TCAS]

Risk Ratio = = Unresolved RR + Induced RR
P[ NMAC without TCAS] Equation 2-1
P[ NMAC w/ TCAS, NMAC w/out TCAS] P[ NMAC w/ TCAS, NMAC w/out TCAS]
= +
P[ NMAC w/out TCAS] P[ NMAC w/out TCAS]
Risk ratio is an important metric because it represents a quantitative measure of safety

that can be obtained through fast-time simulation of aircraft encounters.
The Monte Carlo simulation uses each aircraft’s measured altitude at CPA to compute
VMD. However, the measured VMD may not be equal to the actual VMD due to altimeter error.
If the distribution of combined altimeter error is known, it is possible to compute the probability
of an actual VMD less than 100 ft, which would result in an NMAC during the encounter.
24
Assuming a Laplacian probability distribution function as specified by the ICAO standards, the
altimeter error, e, for one aircraft is distributed according to Equation 2-2:
1  − | e |
p(e) = exp  Equation 2-2 [9]
2λ  λ 
where λ is a statistical parameter based on altitude. The probability of an NMAC for a single
encounter can then be obtained by integrating the combined altimeter error distribution (using
the convolution of two of the distributions given in Equation 2-2) of the two aircraft over the
range of errors that would result in a true vertical separation of less than 100 feet. Figure 2-2
shows an example combined relative altitude error distribution for a case where aircraft at 18,000
ft have a measured separation of 200 ft.
-3
x 10
9
7
Probability Density
2
Measured VMD
1
0
-300 -200 -100 0 100 200 300
Altimeter Error (ft)
Figure 2-2: Example Relative Altitude Error Distribution (Aircraft at 18,000 ft)
If the measured VMD for a particular encounter without TCAS were 200 ft, the NMAC
probability for that encounter would be determined by integrating the distribution from 100 to
300 ft, since any altimeter error in this range would result in a true vertical separation of less than
100 ft. The result is an NMAC probability of approximately 0.24. If one or both aircraft
maneuvered in response to TCAS and the measured VMD for this encounter increased to 300 ft,
the probability of an NMAC would be found by integrating altimeter error between 200 ft and
400 ft, which gives 0.054. Risk ratio for that particular encounter would then be 0.054/0.24 =
25
0.22. Each encounter run k yields a P0(k) (without TCAS) and P1(k) (with TCAS), where P0 and
P1 are the probabilities of an NMAC for the run. Then,
N
1
∑ N P (i )
i =1
1
RR = N
Equation 2-3
1
∑
i =1 N
P0 (i )
In this way, the risk ratio can be computed for a set of thousands of encounters using the
mean probabilities of NMAC with and without TCAS.
2.1.2. Simulation Overview

To obtain the necessary results for calculating risk ratio, a simulation tool is required that
focuses on dynamic interactions between several components of the collision avoidance process.
The simulation tool used for this thesis focuses on interactions between TCAS, pilot response,
and aircraft response during a close encounter between two aircraft that occurs in a 50- to 60-
second window near the CPA.
Figure 2-3 shows a schematic of the simulation process. Each scenario involves two
aircraft already in a close encounter. No other considerations of events leading up to the
encounter are included in the simulation process, such as procedures, pilot error or loss of
separation by ATC.
The simulation process begins by generating an encounter that specifies each aircraft’s
initial conditions and planned trajectory. Once the initial conditions for an encounter are
obtained, they are supplied to the fast-time simulation which runs two possible trajectories for
each aircraft in parallel: one without TCAS (planned or scripted motion), and the other in which
the aircraft responds to TCAS advisories. It is assumed that there are no effects that modify the
planned trajectory other than TCAS. Pilot response and aircraft dynamics represent the aircraft
motion in response to TCAS, such as a vertical acceleration in response to an RA, and may
include some amount of delay in pilot reaction. The simulated motion of each aircraft is updated
each time step and in turn affects how the TCAS logic behaves in the next time step. After the
simulation is completed, VMD with and without TCAS are recorded and converted into
P[NMAC] for each run. Finally, risk ratio is computed by dividing the mean P[NMAC with
26
TCAS] by the mean P[NMAC w/out TCAS] over a set of several thousand encounters, as
described in Section 2.1.
Encounter Global Hawk

Flight Profile & Constraints
TCAS
Pilot Response Aircraft Dynamics
Initial Logic
Output
Conditions
TCAS
Pilot Response Aircraft Dynamics
Logic
RA Response Global Hawk

Latency Flight Constraints
Figure 2-3: Simulation Schematic
In order the evaluate TCAS effectiveness on Global Hawk, one aircraft in each simulated
pair of aircraft in the encounter is replaced with Global Hawk. This in turn causes changes as
indicated by the dashed boxes in Figure 2-3.
First, Global Hawk’s mission profile and aircraft design cause several of its flight
characteristics to be different than a conventional aircraft, and affect its planned trajectory during
an encounter with another aircraft. Several parameters governing the encounter geometry for
one of the two aircraft (i.e. vertical maneuver, airspeed, and bank angle), were modified to reflect
Global Hawk’s flight profile and constraints.
Second, specific dynamic interactions between Global Hawk and another aircraft are
affected by Global Hawk’s TCAS RA response. Global Hawk’s climb rate and vertical
acceleration limits affected the “Aircraft Dynamics” block, and varying amounts of RA response
delay were simulated as part of the “Pilot Response” block. RA delay is the amount of time that
passes between when an RA is issued and when the aircraft begins responding to it. Because the
actual communication latency for Global Hawk is unknown, a parametric study was conducted to
determine how much delay is acceptable before TCAS fails to improve safety.
27
The software used for simulation in this thesis was developed in the Matlab computing
environment and has three components: a graphical user interface (GUI) used to define the initial
conditions of the simulation, an encounter generator that uses the conditions specified in the GUI
to generate millions of encounter scenarios, and a dynamic simulation developed in Simulink
that uses Honeywell TCAS software to simulate aircraft responses during each scenario [10]. A
six degree of freedom point mass dynamic model is used to simulate aircraft motion. Models are
included for range, bearing, and altitude error as well as pilot response.
Post-processing of the results includes the calculation of P[NMAC] with and without
TCAS for each run, the average value over a set of runs, and the calculation of risk ratio for each
set of runs, as described previously.
Shown in Figure 2-4 is an example output of a single simulation run. In the example, the
two aircraft trajectories are shown in the top three graphs, one in black and the other in blue. A
top-down plan view is shown in the first graph, and two side views are shown in the second and
third graphs, one with the two aircraft approaching each other and the other with the two aircraft
trajectories vs. time. CPA occurs at t = 40 seconds. The dashed black line represents the
aircraft’s planned trajectory, while the solid line represents its response to a TCAS RA. In this
example, only the black aircraft changes its planned trajectory in response to TCAS. In the
bottom three graphs, range, vertical acceleration and vertical speed are shown for each aircraft
throughout the scenario. The required vertical rates from TCAS RAs are shown by the gray and
light blue shaded areas in the last graph. In this example encounter, the two aircraft approach
each other at a fairly small closure angle and shallow vertical rates. At t = 13.5 sec, the blue
aircraft receives a “Do Not Climb” RA from TCAS, and at t = 15 sec, the black aircraft receives
a “Climb” RA. Both aircraft respond to the RAs five seconds after they are issued, and a “Do
Not Descend” RA is issued to the black aircraft at t = 25 sec. The aircraft responses cause the
VMD for the example encounter to increase from 560 feet to 788 feet.
28
2000 6000 6000
0 5000 5000
Altitude (ft)
Altitude (ft)
-2000 4000 4000
North (ft)
-4000 3000 3000
-6000 2000 2000
-8000 1000 1000

-6000 -4000 -2000 0 2000 0 20 40 60 80 0 10 20 30 40 50
East (ft) Time (s) Time (s)
4
x 10
7000 1
0.2
6000 Vertical Acceleration (g)
Vertical Speed (ft/min)

0.5
5000 0.1
Range (ft)
4000
0 0
3000
2000 -0.1
-0.5
1000
-0.2
0 -1
0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s) Time (s)
Figure 2-4: Example Scenario Output
2.1.3. Encounter Generation

To reflect the types of two-aircraft encounters that actually happen during flight within a
given airspace, a viable simulation requires an airspace encounter model built from thousands of
hours of radar data. The model is based on such parameters as altitude, airspeeds, or maneuvers
performed by one or both aircraft in close encounters. Observed frequencies of events from the
radar data are then tabulated in probability tables. By selecting new parameter values based on
the tabulated statistics from the radar data, millions of encounters can then be generated and
simulated that reflect actual situations.
Each encounter focuses on two aircraft (treated as point masses) in a 50- to 60-second
window around the CPA. Depending on the specific encounter parameters, vertical (climb or
descend) or horizontal (turning) maneuvers may be planned by one or both aircraft, in the
absence of TCAS, during the encounter.
Two previously developed encounter models were used for this thesis. One was defined
by the International Civil Aviation Organization (ICAO) and the other was developed during the
2000 European Airborne Collision Alerting System Analysis (ACASA) study [9, 11]. These
29
encounter models and the process discussed in this section have been widely used and accepted
by the FAA and ICAO. The following discussion of each parameter will focus on the European
model. The ICAO model uses a somewhat similar process to generate encounters [9].
Figure 2-5 shows the parameter dependency for the European model. The first parameter
to be selected is the altitude layer in which to generate encounters. The ICAO encounter model
is separated into six altitude layers and the European model contains five. All other parameters
for an encounter are based on the altitude layer.
Aircraft Performance Class
Vertical Maneuver Types ( joint, both aircraft )
level or rate-to-level / other

crossing / not crossing
Vertical Rates

Vertical Acceleration
Decision
Vertical Acceleration Duration

and Completion Time
Altitude Layer
Horizontal Miss Distance
Vertical Miss Distance
Non-crossing Separation Tendency

Approach Angle
Horizontal Maneuver Decision (turn / no turn)
Turn Duration and Completion Time
Bank Angle
Ground Speed at Start and End of Encounter
Figure 2-5: Encounter Model Parameter Dependency
The altitude layer limits are shown in Figure 2-6, along with the probability that a
generated encounter lies within each layer. Thousands of encounters are generated within each
altitude layer and divided according to their respective weights in order to reflect the
probabilities of aircraft encounters throughout the entire airspace.
30
6 [0.01]
41500 41000
40000
5 [0.004414]
35000
5 [0.07]
30000 29500
Altitude (feet)
25000 4 [0.121628]
21500
20000 20000
3 [0.153016] 4 [0.22]
15000
13500
10000 2 [0.328396] 10000

3 [0.32]
5000 5000 5000
1 [0.392545] 2 [0.25]
2300
1000 1 [0.13]
0
European Model ICAO Model
Figure 2-6: Encounter Model Altitude Layers (Probabilities in Brackets)
Based on the altitude layer, two aircraft are selected from a performance class table
which determines aircraft performance characteristics such as vertical rate limits, maximum and
minimum airspeeds, and ceiling.
A planned vertical maneuver type is then selected for each aircraft. Vertical maneuvers
are based on initial and final vertical rates, and are categorized into nine classes. Figure 2-7
shows the nine possible vertical maneuvers for the European traffic model based on beginning
and ending vertical profile (climb, level, or descend).
31
End Level Descent Climb
Begin
Level
Descent
Climb
Figure 2-7: European Encounter Model Vertical Maneuver Classes
As an example, a level-climb profile would represent an aircraft initially at a level flight

condition and starting a climbing maneuver. To provide some variety in the vertical profile of
the encounter mixture, a level segment can include any vertical rate between ±400 ft/min.
Selection of a vertical maneuver also includes the consideration of whether the two aircraft will
have flight paths whose altitudes cross each other during the encounter, termed “cruxality.”
Since each encounter involves two aircraft and there are nine maneuvers possible for
each one, a total of 162 combinations of maneuvers can occur for each encounter (including
crossing and non-crossing encounters). All 162 combinations are incorporated into probability
tables based on flight data, from which one is selected randomly for each encounter.
32
Table 2-1: Vertical Maneuvers
Number Vertical Maneuver Type

A1 Descent-Descent
A2 Descent-Level
A3 Descent-Climb
A4 Level-Descent
A5 Level-Level
A6 Level-Climb
A7 Climb-Descent
A8 Climb-Level
A9 Climb-Climb
1 9 9
∑ ∑ ∑ P[ A , A , x] = 1
x =0 j =1 i =1
i j Equation 2-4
As defined in Table 2-1, the nine vertical maneuver types are labeled A1 through A9.
Equation 2-4 shows the total probability of all possible two-aircraft vertical maneuver
combinations. The sum of the probabilities of all combinations, including nine vertical
maneuver types for each aircraft (Ai, Aj) and the cruxality of the encounter (x=0,1), is equal to
one. The ICAO model defines vertical maneuvers slightly differently. Rather than specifying
climb, level or descend, it simply defines level and transition, which could be a climb or descend.
The same general method, however, is used to select a vertical maneuver combination for the
encounter.
An example excerpt from the European model vertical maneuver probability table is
shown in Table 2-2.
Table 2-2: Example Vertical Maneuver Parameters, Altitude Layer 1: P[A1,A6,0]
Cruxality Starting Vertical Ending Vertical Starting Vertical Ending Vertical Probability
Profile (AC1) Profile (AC1) Profile (AC2) Profile (AC2)
Non-Crossing Descent Descent Level Climb 0.010432
33
Table 2-2 indicates about a 1% chance of selecting event [A1, A6, 0] in altitude layer one,
a non-crossing encounter where one aircraft performs a descent-descent maneuver and the other
a level-climb maneuver.
Once a vertical maneuver is selected for each of the two aircraft, starting and ending
vertical rates are selected from an appropriate probability table. The timing of any vertical
acceleration is selected from another table to begin no sooner than five seconds into the 50-
second window during which the encounter occurs.
The CPA in the European encounter model is constrained to occur 40 seconds into the
50-second simulation. VMD at CPA is uniformly selected from a window based on the vertical
maneuver type, and the HMD is selected from a uniform distribution between 0 and 500 feet.
Since HMD is always less than 500 feet, every encounter has the potential to be an NMAC with
TCAS.
The tendency of the flight paths to diverge or converge in altitude is randomly selected,
as is the lateral approach angle of the two aircraft.
Parameters governing the horizontal maneuver of the aircraft are selected next, including
whether each aircraft performs a turn, its duration, and the bank angle during the turn.
Finally, beginning and ending ground speeds are selected for each aircraft.
Reverse kinematics are then used to back up the scenario from the VMD and HMD
situation at CPA into the required initial conditions for each aircraft. Thirteen parameters are
ultimately required for each aircraft to begin simulating each encounter scenario. These include
initial true airspeed, vertical speed, heading, three-dimensional position, and several parameters
that define the planned vertical- and horizontal-plane motion of the aircraft.
Figure 2-8 shows a top-down view and a plot of altitude vs. time for an example two-
aircraft encounter. Each aircraft begins at its respective initial position [E, N, h] with initial true
airspeed [v], heading angle [ψ] and vertical speed [hdot]. One aircraft performs a turn to the right
in combination with a vertical acceleration, and the other continuously descends along a straight
horizontal path. The aircraft whose trajectory is colored in blue performs an instantaneous bank
to the angle [φ] at time [t3] to begin the turn, and comes out of the bank at time [t4], again
instantaneously. Its vertical maneuver includes a vertical acceleration [hdd] at time [t1] to
increase its climb rate. This vertical acceleration ends at time [t2].
34
Top-Down View Altitude vs. Time
North Altitude
[t4]
[a]
[φ]
East [t1] [t2] Time
[t3]
[hdd]
[ψ , N, E] [v, hdot, h]
v Initial True Airspeed ft/s

hdot Initial Vertical Speed ft/s
ψ Initial Heading Angle rad
N Initial North Position ft
E Initial East Position ft
h Initial Altitude ft
t1 Start Time of Vertical Maneuver sec
t2 End Time of Vertical Maneuver sec
hdd Vertical Acceleration during Vertical Maneuver ft/s2
t3 Start Time of Turn sec
t4 End Time of Turn sec
φ Bank Angle during Turn rad
a Horizontal Acceleration ft/s2
Figure 2-8: Aircraft Encounter Parameters
35
2.2. Global Hawk Encounter Model Modification
As a high-altitude long-endurance aircraft, Global Hawk flies differently than a
conventional aircraft. However, operational radar data are not currently available to create a
specific encounter model that reflects this difference in performance. Therefore, existing
encounter models were modified based on data from Global Hawk’s flight manual. In order to
accurately model the performance of Global Hawk during encounter simulations, several
parameters from the ICAO and European air traffic models needed to be modified to reflect its
flight characteristics.
Figure 2-9 shows the parameter dependencies and highlights the changes required to
develop a Global Hawk-specific encounter model. The same altitude layers were used for the
Global Hawk modified encounter models as for the original models. Because the proportion of
Global Hawk encounters that might occur at each altitude layer is not known, the altitude layer
weights were not applied to the Global Hawk encounter models. Instead, encounters were
generated separately within each altitude layer and the results for each layer were compared
individually against the results from the original encounter models. Of the thirteen parameters
shown in Figure 2-9, six were modified to reflect Global Hawk’s flight characteristics.
36
Aircraft Performance Class
Vertical Maneuver Types ( joint, both aircraft )
level or rate-to-level / other

Vertical Rates

Decision
Vertical Acceleration Duration

and Completion Time
Altitude Layer
No changes required
Horizontal Miss Distance
Global Hawk specific
Vertical Miss Distance
model required
Non-crossing Separation Tendency

Approach Angle
Horizontal Maneuver Decision (turn / no turn)
Turn Duration and Completion Time
Bank Angle
Ground Speed at Start and End of Encounter
Figure 2-9: Global Hawk Encounter Model Parameter Dependency
Global Hawk’s flight profile involves relatively steep climbs up to its cruising altitude, as
well as steep descents to landing, slow airspeeds, and shallow turns. The parameters for one
aircraft in each encounter were modified to reflect these performance constraints. In addition,
the vertical maneuver mix (combinations of climb, level, and descend geometries) for Global
Hawk was modified based on its higher probability of being in a climb or descent when an
encounter occurs than might be the case for a conventional aircraft. These changes allowed
encounters to be simulated that involved one aircraft with conventional flight parameters
distributed according to the original encounter model, and one aircraft exhibiting Global Hawk’s
flight characteristics. The cumulative effect of these changes could then be observed by
comparing the simulation results from the original encounter model (encounters between two
conventional aircraft) against encounters between a conventional aircraft and Global Hawk.
Both the ICAO and European encounter models were modified to obtain two Global
Hawk encounter models. The specific parameters that were changed are discussed in the
following sections.
37
2.2.1. Airspeed
Global Hawk’s airspeed limits were obtained from its flight manual [3, 12]. Figure 2-10
shows the Global Hawk airspeed envelope vs. altitude and weight, along with its limits. The red
lines in Figure 2-10 show the airspeed limits selected for the Global Hawk encounter model
modifications. The airspeed limits were based on maximum gross weight while climbing and
minimum weight while descending, in order to capture the full range of airspeeds both during the
climb to initial cruising altitude and during the descent to landing. Because the units in Figure
2-10 are given in KCAS, these speeds were converted to KTAS using the standard atmosphere.
70 27
Above 60kft,
65 consult Airspeed Envelope vs Altitude and Weight 26
Maximum
Altitude Alert Autonomous Speed Schedule
60 25
Bulletin
55 0.65 Mach Limit
24
59kft
50 23
45 Min Override Speed vs Altitude 22
Vne -- Structural
For Climb and Alt Hold Li it
40 21
Weight - Klbs
Altitude - Kft
(decend mode reverts to

autonomous schedule)
35 20
Max Spd Range from 11,500 to 14,200 lbs
Min Spd For WT < 11,500
30 Min Spd vs WT when 19

alt>10kft
25 Max Spd for WT > 14,200 lbs 18
20 Min Spd vs WT when 17
WT>14,200
Min Spd for
alt<10kft
15 16
Max Spd For WT < 11,500 lbs
10 15
Below 10kft Min Spd
5 93kts 14
is based on WT only
0 13
-5 80 90 100 110 120 130 140 150 160 170 18012
Airspeed - kts CAS The min speed at any altitude is the greater of either
the 'Min Override Spd' OR the 'Min Speed vs WT'
Figure 2-10: Global Hawk Airspeed Envelope [12]
Figure 2-11 shows Global Hawk’s airspeed limits and its nominal climbing airspeed
profile over all five altitude layers for the European encounter model. The airspeed limits shown
in Figure 2-11 were added to the maximum and minimum speed tables in the original encounter
models. Global Hawk airspeeds were selected according to a uniform distribution between the
minimum and maximum limits.
38
40000 Speed range within
altitude layer
35000
5
30000
25000 4
Altitude (ft)
Nominal profile
20000
3
15000
10000
2
Altitude layers
5000
1
0
100 150 200 250 300 350
True Airspeed (kt)
Figure 2-11: Global Hawk Airspeed Profile Based on European Encounter Model
Figure 2-12(a) shows a histogram of Global Hawk initial airspeeds in altitude layer three,
while Figure 2-12(b) shows a histogram of initial airspeeds for a conventional aircraft. It is clear
that Global Hawk’s airspeed limitations have a large effect on the airspeed distribution of the
encounters. Rather than being broadly spaced across a wide range like the conventional aircraft,
the airspeeds are condensed within Global Hawk’s relatively small range of allowable speeds.
39
0.06
Relative Frequency 0.05
0.04
0.03
0.02
0.01
0
0 100 200 300 400 500 600
Velocity (KTAS)
(a) Global Hawk
Relative Frequency
0.1
0.05
0
0 100 200 300 400 500 600
Velocity (KTAS)
(b) Conventional Aircraft
Figure 2-12: Initial Airspeed Histograms, Layer 3
2.2.2. Vertical Maneuver Probability

The Global Hawk aircraft was designed to climb steeply and quickly up to its cruising
altitude. When Global Hawk encounters another aircraft, it is most likely that Global Hawk will
be performing a steep climb or a descent to landing, rather than being in level cruise. The
probability of Global Hawk flying a climbing or descending vertical maneuver during an
encounter was therefore initially selected to be 80%. Although a specific probability is not yet
available based on actual radar data, this number was selected as a reasonable estimate to
develop and test the Global Hawk modified ICAO and European traffic models.
The method used to derive probabilities for each vertical maneuver type involved the
definition of conditional probabilities. As discussed in Section 2.1.3, 162 possible combinations
of vertical maneuvers exist for each two-aircraft encounter. Table 2-3 shows the selected
probabilities of all vertical maneuver types for Global Hawk. The 0.80 probability of
continuously climbing or descending was split equally into climb-climb and descent-descent
40
maneuvers. A 0.14 probability was assigned to level flight, and the remaining 0.06 probability
was distributed equally to the other six vertical maneuvers.
Table 2-3: Example Global Hawk Vertical Maneuver Probabilities
Number Maneuver Type Probability

A1 Descent-Descent 0.40
A2 Descent-Level 0.01
A3 Descent-Climb 0.01
A4 Level-Descent 0.01
A5 Level-Level 0.14
A6 Level-Climb 0.01
A7 Climb-Descent 0.01
A8 Climb-Level 0.01
A9 Climb-Climb 0.40
First, the probability of each vertical maneuver type for a conventional aircraft was
obtained by summing all possible combinations where one aircraft performed the maneuver. For
example, the probability of a conventional aircraft following a descent-descent maneuver is:
1  9 9 
P[ A1 ] = ∑  ∑ P[ A1 , A j , x] + ∑ P[ Ai , A1 , x] − P[ A1 , A1 , x ]  Equation 2-5
x =0  j =1 i =1 
where Aj represents all nine possible maneuvers for the other aircraft, for both crossing
and non-crossing encounters. In altitude layer one of the European model, for example, the
probability of a descent-descent maneuver by one aircraft, P[ A1 ] , is computed to be 0.23845.
Next, the conditional probabilities for each vertical maneuver were calculated. For
example, the probability of a non-crossing encounter with a level-climb maneuver for one
aircraft, given a descent-descent maneuver for the other aircraft, would be:
P[ A1 , A6 ,0] + P[ A6 , A1 ,0]
P[ A6 | A1 ,0] = Equation 2-6
P[ A1 ]
41
In layer one of the European model, the probability of a non-crossing encounter with a
level-climb maneuver (A6) for a conventional aircraft given a descent-descent maneuver (A1) for
the other aircraft is computed to be 0.04375.
Finally, using the Global Hawk probabilities defined above, new probability tables were
created by multiplying the Global Hawk vertical maneuver probabilities by the conditional
probabilities. For example, the probability of an encounter where a Global Hawk descent-
descent meets a conventional aircraft level-climb is:
P[ A1GH , A6 ,0] = P[ A6 | A1 ,0] ⋅ P[ A1GH ] Equation 2-7
The calculated probability for this example non-crossing encounter is 0.0175, which is
almost double the value for this type of encounter from the original European encounter model.
The effect of Global Hawk’s vertical maneuver probabilities on encounter generation was
a higher likelihood of a Global Hawk climb-climb or descent-descent maneuver during an
encounter with another aircraft. Figure 2-13 shows the probabilities for each type of vertical
maneuver for the European encounter model at altitude layer three, and Figure 2-14 shows these
same probabilities for the Global Hawk modified model. In the European encounter model, it
can be seen that climbing and descending maneuvers are rarely selected for both aircraft. In the
Global Hawk model, the Global Hawk aircraft is more likely to be climbing or descending. The
highest likelihood is for a Global Hawk descent-descent maneuver (continuous descent) and the
other aircraft remaining level throughout the encounter.
42
D-D
D-L 0.25
Aircraft One Vertical Maneuver D-C
0.2
L-D
L-L 0.15
L-C
0.1
C-D
C-L 0.05
C-C
D-D D-L D-C L-D L-L L-C C-D C-L C-C
Aircraft Two Vertical Maneuver
Figure 2-13: Vertical Maneuver Probabilities, European Model, Layer Three
D-D 0.18
D-L 0.16
Global Hawk Vertical Maneuver
D-C 0.14
L-D 0.12
0.1
L-L
0.08
L-C
0.06
C-D
0.04
C-L
0.02
C-C
0
D-D D-L D-C L-D L-L L-C C-D C-L C-C
Conventional Aircraft Vertical Maneuver
Figure 2-14: Vertical Maneuver Probabilities, Global Hawk Modified European Model, Layer Three
43
2.2.3. Climb Rate
Global Hawk’s climbing maneuvers are generally flown at a steeper flight path angle (~7
deg) than that of a conventional manned aircraft (~4 deg). Based on the assumptions of
maximum gross weight and no temperature deviation, several data points were collected from
Global Hawk’s climb rate chart shown in Figure 2-15.
Figure 2-15: Global Hawk Climb Rate [3]
The data points were plotted against altitude, and a linear regression was performed
which yielded an equation for maximum rate of climb as a function of altitude, shown in Figure
2-16.
44
4000
from Rate Of Climb chart
Linear Regression
3500
3000
Increase
2500 Climb
Climb Rate (ft/min)
RA
2000
Climb
1500 RA
1000
500
0
0 1 2 3 4 5 6
Altitude (ft) x 10
4
Figure 2-16: Global Hawk Climb Rate vs. Altitude, Gross Weight = 26,750 lbs
The maximum climb rate was used as a constraint both in the encounter model
geometries and when simulating Global Hawk TCAS response. For Global Hawk encounter
generation, the limits on climb rate were capped based on the data in Figure 2-16. The standard
rate of climb for a TCAS resolution advisory is 1500 ft/min for an initial “climb” RA and 2500
ft/min for a subsequent “increase climb” RA. Figure 2-16 indicates that at altitudes greater than
20,000 feet, Global Hawk cannot exceed 2500 ft/min, and above 34,000 feet, it cannot exceed a
1500 ft/min climb rate. During the Global Hawk encounter simulation, if TCAS commanded a
climb rate that was greater than what Global Hawk could perform, the actual rate of climb was
constrained to the performance limit shown in Figure 2-16, based on altitude.
2.2.4. Descent Rate

Using the descent rate chart shown in Figure 2-17, the maximum Global Hawk descent
rate for each altitude layer was calculated and added as a constraint to the probability tables for
the ICAO and European encounter models.
45
Figure 2-17: Global Hawk Descent Rate [3]
Assuming an aircraft weight of 13,000 lb and rapid descent conditions, the time to travel
from the high to the low limits of each altitude layer was calculated. The total height of each
layer was divided by the time to descend through that layer to give the average descent rate. The
maximum Global Hawk descent rate was approximately the same for each altitude layer, 4000
ft/min. This value was implemented in the Global Hawk modified encounter models.
Figure 2-18 shows Global Hawk’s initial vertical rate as randomly generated over
200,000 runs. Global Hawk’s initial vertical rate does not exceed 4000 ft/min, and a large
number of vertical rates are in the 1000 – 4000 ft/min range.
46
0.25
0.2
count/# samples
0.15
0.1
0.05
0
-5000 -4000 -3000 -2000 -1000 -400 -200 -100 0 100 200 400 1000 2000 3000 4000 5000
Vertical Rate (ft/min)
Figure 2-18: Global Hawk Initial Vertical Rate Histogram, All Altitude Layers
In contrast, Figure 2-19, generated in a similar way but for a conventional aircraft,
indicates a much higher probability of an initial vertical rate of less than 100 ft/min and slightly
higher vertical rate limits.
0.25
0.2
count/# samples
0.15
0.1
0.05
0
-5000 -4000 -3000 -2000 -1000 -400 -200 -100 0 100 200 400 1000 2000 3000 4000 5000
Vertical Rate (ft/min)
Figure 2-19: Conventional Aircraft Initial Vertical Rate Histogram, All Altitude Layers
47
The difference between the histograms in Figure 2-18 and Figure 2-19 arises from a
combination of the Global Hawk climb and descent rate limits as well as the vertical maneuver
probabilities discussed in Section 2.2.2.
2.2.5. Bank Angle

Global Hawk’s bank angle schedule was used to determine limits on bank angle during
maneuvering. Figure 2-20 shows Global Hawk’s bank angle limits throughout its altitude
envelope.
Figure 2-20: Global Hawk Bank Angle Schedule [3]
In the modified European model, the bank angle limit is 20 deg for altitude layers one
through four, and 15 deg for altitude layer five. If a turning maneuver is performed during the
maneuver, the bank angle randomly selected by the encounter model will not exceed these limits.
This only affects the encounters where a turning maneuver is performed, and where the selected
bank angle exceeds Global Hawk’s limit.
2.2.6. Ceiling
Global Hawk’s operating ceiling is 65,000 feet. This is higher than either of the
encounter models’ highest altitude. This allows encounters to be simulated at any altitude in the
model, rather than capping the altitude at a lower limit as it is for other aircraft classes.
48
2.2.7. Vertical Acceleration
Global Hawk’s pitch rate limit is two degrees per second. When TCAS issues a
resolution advisory, it expects the aircraft to perform a vertical maneuver with a normal
acceleration of 0.25g for an initial RA or 0.35g for an increase or reversal. Because of Global
Hawk’s 2 deg/sec pitch rate limit, at low speeds in certain encounters its vertical acceleration
cannot meet this requirement.
Equation 2-8 shows how vertical acceleration, a, is calculated from airspeed, v, and pitch
rate, γ& (assuming a small bank angle).
a = vγ& Equation 2-8
Figure 2-21 shows the pitch rate limit and its effect on vertical acceleration applied to
each altitude layer.
0.18 g
0.25 g 0.35 g
40000
35000
30000
25000 Nominal profile

Altitude (ft)
20000
15000
Can exceed 0.35g

10000
Can reach 0.25g but not 0.35g
5000 Cannot reach 0.25g
0
100 150 200 250 300 350
True Air Speed (kt)
Figure 2-21: Global Hawk Vertical Acceleration Profile, European Encounter Model
49
Global Hawk’s vertical acceleration at very slow airspeeds cannot reach the 0.25g
commanded by TCAS. At other airspeeds, it can meet the 0.25g but not the 0.35g for an increase
or reversal. All altitude layers except layer five are affected to some extent by this constraint.
The vertical acceleration limit causes a less aggressive avoidance maneuver to be performed at
lower airspeeds.
As discussed in Section 2.2.3, the target vertical rate for an initial TCAS “climb” RA is
1500 ft/min. Figure 2-22 shows simplified altitude trajectories for three aircraft performing
vertical pitch-up maneuvers in response to a TCAS “climb” RA: two aircraft perform this
maneuver at 0.25g (conventional) with delays of 5 and 10 seconds, and one performs the
maneuver at 0.18g (Global Hawk worst case) with a delay of 5 seconds. The dashed lines in the
figure represent the time at which each aircraft reaches a climb rate of 1500 ft/min. The solid red
line in the figure represents a conventional aircraft trajectory response to a TCAS “climb” RA at
1500 ft/min. The aircraft pitches up at 0.25g and reaches 1500 ft/min at t = 8.1 sec. The total
altitude gained after 30 seconds is 586 ft. The blue line demonstrates how Global Hawk’s
vertical acceleration limit affects its ability to meet the TCAS commanded climb rate of 1500
ft/min. Referring to Figure 2-21, 0.18g is the worst case vertical acceleration for Global Hawk,
when it flies at its slowest airspeed. Because it can only meet 0.18g, it takes slightly longer to
achieve the 1500 ft/min climb rate (t = 9.3 sec) and its total altitude gain after 30 seconds is 571
ft, slightly less than the conventional aircraft. The solid black line represents a conventional
aircraft maneuvering at 0.25g but with an added delay of 5 seconds. The additional delay
represents a slower pilot response to TCAS. This aircraft achieves a 1500 ft/min climb rate at t =
13.1 sec with a total altitude gain of only 461 ft.
50
600
500 h&& =0.25g

Delay = 5 sec
400
h&& =0.18g
Delay = 5 sec
h&& =0.25g
Altitude (feet)
300 Delay = 10 sec
200
100
-100
0 5 10 15 20 25 30
Time (sec)
Figure 2-22: Vertical Maneuver Example
It can be seen from Figure 2-22 that a conventional aircraft performing the vertical
maneuver at 0.25g but with a slight delay is comparable to performance with a lower
acceleration, like the aircraft representing Global Hawk. The effect of performing the maneuver
with 0.18g vertical acceleration is equivalent to performing it with an additional latency of 0.6
sec. This suggests that Global Hawk’s vertical acceleration limit has a similar effect on the risk
ratio as does a slower TCAS pilot response.
2.3. Summary
Figure 2-23 highlights the changes that were needed for the simulation to reflect Global
Hawk’s characteristics. Global Hawk’s flight profile, as well as some of its performance
constraints, affected the way in which the geometries of the two-aircraft encounters were set up.
Several of its performance constraints also affected the way Global Hawk responded to TCAS
RAs during the encounter simulation. The combination of these changes allowed Global Hawk-
specific encounter models to be developed, where one aircraft was conventional and the other
51
was specified as Global Hawk. The subsequent effect on risk ratio could then be observed after
simulating thousands of these encounters.
Global Hawk Flight Profile Global Hawk Performance Constraints
P[climb] Airspeed
P[descent] Descent Rate
P[level] Bank Angle
Ceiling
Climb Rate
RA Response Latency
Encounter Global Hawk

Geometry TCAS Response
Figure 2-23: Summary of Global Hawk Effects on Encounter Generation and Simulation
52
3. Simulation Results
Having described the method for airspace encounter model use and the generation of
Global Hawk-specific encounter models, this section presents the results of the simulation of
millions of two-aircraft encounters using the Global Hawk models.
A total of 1,860,000 runs were simulated for each test condition (described in Section
3.1) using the Global Hawk modified ICAO model, 310,000 runs in each of six altitude layers.
For the European model, 1,000,000 runs were simulated for each test condition, 200,000 runs in
each of the five altitude layers.
3.1. Test Conditions

All simulations involved two aircraft. The first was always a conventional aircraft with
flight characteristics based on the original ICAO or European encounter model and equipped
with and responding to TCAS. The second aircraft was controlled to be either a conventional
aircraft or Global Hawk. By changing the aircraft’s flight characteristics, whether it had TCAS
onboard, or how it responded to TCAS resolution advisories, multiple effects on risk ratio could
be observed. Several combinations of encounter model, aircraft response and equipage were
used; these are shown in Table 3-1 and described below.
Table 3-1: Simulation Parameter Combinations
TCAS-Equipped Conventional Aircraft vs.

Encounter Model TCAS Response Equipage
ICAO Conventional Mode S Transponder
Global Hawk-Modified ICAO Global Hawk TCAS
European Global Hawk + Delay
Global Hawk-Modified European
Encounter Model:
The original ICAO and European encounter models were modified as discussed in
Section 2.2 to develop models for Global Hawk. Two Global Hawk models were created, one
based on the ICAO model and the other based on the European model. Simulations were run for
the ICAO and European model encounters and for the two Global Hawk model encounters, and
risk ratios were then compared. This allowed the effect of Global Hawk’s vertical flight profile
and performance constraints on risk ratio to be observed.
53
TCAS Response:
A conventional TCAS response indicates that during the simulation, the aircraft
responded to TCAS with a vertical acceleration and climb rate that met the requirements
discussed in Section 2.2.3 and 2.2.7 (0.25g and 1500 ft/min for an initial RA, and 0.35g and 2500
ft/min for a strengthened RA). The RA latency for a conventional response is 5 seconds for an
initial RA and 2.5 seconds for a subsequent RA. A Global Hawk TCAS response indicates that
limits were imposed on climb rate and vertical acceleration that could prevent the aircraft from
meeting the vertical acceleration or climb rate commanded by TCAS. The RA response latency
for this condition was still 5 seconds (initial) and 2.5 seconds (subsequent). The third type of
TCAS response, Global Hawk + Delay, indicates an additional latency in the response to
resolution advisories. This represents a total latency caused by round-trip communication delays
in addition to human operator response delays. The delays for both initial and subsequent RAs
were increased by 5, 10, and 15 seconds so that the final set of encounters had Global Hawk
responding with a 20 second initial RA latency and 17.5 second subsequent latency. In addition,
a set of encounters was simulated with no delay in response to either an initial or subsequent RA,
in order to represent a best-case, automated response to TCAS on Global Hawk.
Equipage:
The equipage on Global Hawk was set to be either a Mode S transponder alone or in
combination with TCAS. During cases where Global Hawk was only equipped with a Mode S
transponder, the only avoidance maneuver commanded or performed was by the other aircraft,
equipped with TCAS. Risk ratio would therefore generally be higher for these cases than for
those where both aircraft are TCAS-equipped.
3.2. Results
The following figures show two types of effects. First, the effect on risk ratio of
equipping both aircraft with TCAS (instead of Mode S vs. TCAS) is highlighted by comparing
the heights of a red and a blue bar at a given altitude layer. Second, the effect of Global Hawk’s
encounter model and TCAS response on risk ratio is shown by comparing the first pair of bars to
the second pair at a given altitude layer. The results are presented in this way for both the ICAO
and European encounter models. The description of what each bar represents is annotated for
only one altitude layer in the figures, but applies to all other layers as well.
54
3.2.1. ICAO Encounter Model
Figure 3-1 shows the results for the ICAO encounter model and the Global Hawk
modified ICAO model with one aircraft carrying a Mode S transponder alone (red bars) and then
TCAS (blue bars). All of the simulated encounters were run versus a conventional TCAS-
equipped intruder aircraft. The first two bars (light red and light blue) represent the
improvement TCAS provides to a conventional aircraft. The second two bars (dark red and dark
blue) represent this same effect using the Global Hawk modified encounter model.
TCAS decreases the risk ratio by more than 75% from the Mode S value at all altitude
layers in the original ICAO model. TCAS does not perform as well against an unequipped
intruder performing a climb or descent as it does when the intruder is level. Therefore, because
of Global Hawk’s higher likelihood of climbing or descending, risk ratio is generally larger for
the Global Hawk Mode S case (dark red) than the ICAO (conventional) Mode S case (light red).
This is the case in the first four altitude layers. However, at layers five and six, the risk ratio
when Global Hawk is unequipped with TCAS decreases from that of a conventional aircraft.
The most likely reason for this is that at high altitude layers, although Global Hawk has a higher
likelihood of climbing, it also climbs at slower vertical rates than a conventional aircraft. Since
Global Hawk’s climb rate is not as high at these altitudes, TCAS would be more effective against
a Global Hawk aircraft unequipped with TCAS than against an unequipped conventional aircraft.
55
0.40
.358
0.35
ICAO GH-Modified
Model ICAO Model
0.30
.259
.246
0.25 .233
Risk Ratio
.217
0.20 .183
.154
0.15 .136 .132
.126
GH Mode S vs. TCAS

.113
Mode S vs. TCAS
0.10
TCAS vs. TCAS
TCAS vs. TCAS .066

.056
0.05 .042
.012 .014 .016

.008 .009 .010 .007 .007
.004 .003
0.00
1 2 3 4 5 6
Altitude Layer
Figure 3-1: Risk Ratio Comparison, ICAO Model vs. Global Hawk-Modified ICAO Model
Equipping Global Hawk with TCAS in addition to a Mode S transponder significantly

reduces risk ratio at all altitude layers in the modified ICAO encounter model.
The differences in risk ratio between altitude layers are primarily due to variations in the
encounter geometries that are dependent upon altitude. For instance, aircraft vertical profiles,
horizontal accelerations and closure angles are different at high altitudes than at lower altitudes
due to the statistical distribution of the collected radar data used to generate the encounter
models. These combined changes can cause risk ratio to increase or decrease with altitude layer
in both encounter models.
Figure 3-2 shows the results for the Global Hawk RA latency study. Risk ratios are
shown for cases where Global Hawk initial RA response times were increased from zero seconds
(autonomous) to twenty seconds. All of the risk ratios are normalized to the Mode S-only risk
ratio, where Global Hawk is only equipped with a transponder.
56
3
2.5
Normalized Risk Ratio
Layer 2
1.5
Layer 3
Layer 1
Mode S Only
1
Layer 6
Layer 4
0.5
Layer 5
0
0 5 10 15 20
Initial RA Response Time (sec)
Figure 3-2: Risk Ratio vs. RA Latency, Global Hawk Modified ICAO Model
A response time of zero seconds indicates an autonomous response to TCAS RAs with no
delay, and the remaining four points along a given altitude layer line represent 5, 10, 15 and 20
seconds delay between an initial RA and the start of a vertical acceleration in response. The
five-second delay case is equivalent to a conventional aircraft pilot response as described in
Section 3.1. One important result in Figure 3-2 is the latency allowed before TCAS begins to
perform worse than a Mode S transponder alone, which would be indicated by normalized risk
ratios greater than one. At altitude layer one the maximum allowable latency appears to be less
than 15 seconds. At RA latencies greater than 15 seconds, Global Hawk’s initial vertical
avoidance maneuver is performed so late that TCAS may have already issued a reversal to the
other aircraft, and the two aircraft may actually converge rather than diverge, decreasing the
vertical miss distance and hence increasing the probability of an NMAC.
Global Hawk’s initial TCAS RA response time was a significant factor in the simulation.
After 15-20 seconds delay in RA response, the safety provided by TCAS is much lower than it
57
would be with a standard five-second pilot response, an autonomous Global Hawk response, or
with Global Hawk not even carrying TCAS.
3.2.2. European Encounter Model

Figure 3-3 shows the results for the European encounter model and the Global Hawk
modified European model. The first two bars represent risk ratios from the original European
model with a conventional aircraft response and Mode S transponder equipage (light red), then
with TCAS equipage (light blue). The second two bars (dark red and dark blue) are the risk
ratios from the Global Hawk modified European model with a Global Hawk aircraft response,
also equipped with a Mode S transponder and then with TCAS.
It can be seen that TCAS reduces the risk ratio significantly at all altitude layers for both
the European and Global Hawk modified European encounter models. The Mode S risk ratios
are higher for the Global Hawk model than for the European model. Like the ICAO model, this
is due to the effect of Global Hawk’s flight profile (more climbing and descending maneuvers)
and performance constraints. Risk ratio is also reduced in the TCAS-equipped case compared to
the Mode S-only case.
One significant result is that the risk ratio is greater than 1.0 at altitude layer five for the
Global Hawk Mode S-only case (the dark red bar). The mean probability of a mid-air collision
with TCAS on the other aircraft increases from what it would be without TCAS. Essentially, at
layer five the results indicate that TCAS apparently reduces safety when the intruder aircraft is
TCAS-equipped and Global Hawk is only Mode S-equipped. This effect is due to the
combination of a peculiarity in the European encounter model and a known vulnerability in the
TCAS logic involving slow rate-of-closure encounters [13]. Actual exposure to this risk at high
altitudes is believed to be overestimated by the European model and is currently being
investigated by other researchers.
58
1.80
1.642
1.60
1.40
1.20
Risk Ratio
1.00
European GH-Modified
Model European Model .836
0.80 .731
.624
Mode S vs. TCAS

0.60
TCAS vs. TCAS
TCAS vs. TCAS

GH Mode S vs. TCAS
0.40 .369
.209
0.20
.103 .112 .093
.058 .063 .079
.052
.022 .021 .014 .014 .011 .011 .028
0.00
1 2 3 4 5
Altitude Layer
Figure 3-3: Risk Ratio Comparison, European Model vs. Global Hawk-Modified European Model
Figure 3-4 shows the risk ratios for Global Hawk initial RA response times normalized to
the Mode S-only risk ratio. The same general observations can be made about RA latency for
the Global Hawk modified European model as for the modified ICAO model. A latency of
approximately 15 seconds is acceptable before TCAS begins to perform worse than a Mode S
transponder alone. By the time Global Hawk reacts to the TCAS RA, TCAS has already
determined that the recommended RA is not succeeding, and is likely to have reversed the RA
(i.e. “climb” changes to “descend”), with the result that VMD decreases rather than increases.
59
3
Layer 1
2.5
Normalized Risk Ratio
1.5
Layer 2
Mode S Only
1 Layer 3
0.5
Layer 4
Layer 5
0
0 5 10 15 20
Initial RA Response Time (sec)
Figure 3-4: Risk Ratio vs. RA Latency, Global Hawk Modified European Model
In order to check the statistical validity of the results presented in this section, a metric
called “running risk ratio” can be calculated over a large set of runs. Plotting the running risk
ratio allows the risk ratios to be seen converging to statistically stable values as the number of
runs increases. To calculate the running risk ratio, the average P[NMAC w/ TCAS] and P[NMAC
w/out TCAS] over the last n simulation runs are computed from n = 1 to 200,000.
Figure 3-5 shows the running risk ratio for an example set of 200,000 encounters that
were simulated at altitude layer three for the European and Global Hawk modified European
encounter models.
60
0.4
Global Hawk Encounter Model
European Encounter Model
0.35
0.3
0.25
Running Risk Ratio
0.2
0.15
0.1
0.05
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Encounter Number x 10
5
Figure 3-5: Running Risk Ratio, Altitude Layer 3, Conventional TCAS Response
After approximately 40,000 runs, the results begin to stabilize with a consistently higher
risk ratio for the Global Hawk modified European model than the conventional European model.
The mean difference in running risk ratio when n = 200000 is approximately 2σ, where σ is the
standard deviation of the risk ratio over the same n. This appears to dominate the effect of noise
in the results, so that the higher risk ratio for the Global Hawk model is truly representative of
changes in its flight profile, such as the steeper climb and descent rates discussed above. The
results indicate that 200,000 is a large enough number of encounters to simulate at each altitude
layer in order to obtain statistically valid results.
61
4. Conclusions and Recommendations
Unmanned aerial vehicles (UAVs) are becoming increasingly important in military and
homeland security applications both overseas and domestically. Because of their ability to
perform the same mission as manned aircraft without the risk of pilot loss, the number of UAVs
operated in the United States will continue to increase in the future. This causes a concern for
collision avoidance between UAVs and manned aircraft in civil airspace. The U.S. Air Force’s
Air Combat Command is seeking to equip the RQ-4 Global Hawk UAV with the Traffic Alert
and Collision Avoidance System (TCAS) to address this concern. Since TCAS was not designed
for use on UAVs, the level of safety provided by the system must be such that equipping Global
Hawk with TCAS would be advantageous.
Two Global Hawk-specific encounter models were constructed to help evaluate TCAS
effectiveness. One was based on a European Airborne Collision Alerting System Analysis
(ACASA) traffic model and the other was based on a model defined by the International Civil
Aviation Organization (ICAO). The Global Hawk encounter models reflected two types of
modifications: one from Global Hawk’s flight profile that included limits on several flight
parameters such as airspeed and climb rate, and the other that affected the mixture of encounters,
i.e. how often a climbing or descending vertical maneuver was selected for Global Hawk.
Global Hawk’s climb rate and vertical acceleration limits were also imposed in the dynamic
simulation to observe how its response to TCAS would be different than a conventional aircraft.
4.1. Global Hawk Flight Profile

Global Hawk’s vertical profile can raise or lower risk ratio, depending on which
encounter model was used and the altitude layer in which the encounters were simulated. At
high altitudes in the modified European encounter model, the risk ratio increased from that of
two conventional aircraft, both for the TCAS-equipped and the Mode S-equipped Global Hawk
cases. For the case where Global Hawk was equipped with only a Mode S transponder, the risk
ratio rose to approximately 1.6 even with TCAS on the intruder aircraft. The risk ratio of 1.6
indicates a 60% safety degradation with TCAS, as opposed to an expected increase in safety.
This is due to an issue in the European encounter model which involves a particular vulnerability
62
in TCAS during slow rate-of-closure encounters, rather than a prediction of unsafe operations
[13].
In the modified ICAO model, the changes to risk ratio due to Global Hawk’s flight
profile were not nearly as drastic. These results indicate that although the flight profile of Global
Hawk does have some effect on risk ratio, the effect is not widespread across both encounter
models. Additional testing could include a study of how Global Hawk’s proportion of
continuous climbing or descending vertical maneuvers affects risk ratio. 80% was selected as a
reasonable estimate for this number, but radar data or the aircraft’s flight data recorder may
indicate a number that is higher or lower. New sets of encounters could be generated and
simulated that reflect an increase or decrease in this number, as well as its effect on risk ratio.
4.2. TCAS RA Response Latency

TCAS simulations using both the modified European and ICAO encounter models
indicate that there is room for approximately 15 seconds of communication latency before TCAS
performs worse than not having TCAS on Global Hawk. Especially at lower altitude layers, risk
ratio begins to rise rapidly after an RA response time of more than 5-10 seconds. An
autonomous RA response for Global Hawk appears to be a promising solution. With an
immediate RA response time, the simulations gave a maximum risk ratio of just 18.5% of the
Mode S-only value for the modified ICAO model and 9.5% for the modified European model.
4.3. Global Hawk Vertical Acceleration and Climb Rate Limits

Global Hawk’s vertical acceleration and climb rate limits during the simulation appear to
have a slight effect on risk ratio. The effect was limited primarily to altitude layer one of the
modified ICAO encounter model. As discussed in Section 2.2.7, this effect is due to the slower
airspeeds flown by Global Hawk at lower altitudes that at times prevented it from reaching the
vertical acceleration commanded by TCAS. At the higher altitude layers, the climb rate limit
had only a slight effect on risk ratio. The increases were much smaller for the modified
European model, and in general none of the increases were causes for concern. TCAS does have
the capability to inhibit the climb or increase climb RA rate limits due to high altitude or aircraft
configuration [6]. This capability was not studied for Global Hawk, and because of the very
small change the climb rate limit has on risk ratio, this may not be an issue warranting further
63
investigation. The effect of Global Hawk’s vertical acceleration limit on its ability to respond
properly to TCAS RAs needs to be discussed with the TCAS community.
4.4. Summary and Recommended Research

The results of the dynamic simulation conducted for Global Hawk indicate that a
significant improvement in safety is provided by equipping the aircraft with TCAS in addition to
the Mode S transponder it currently has. Risk ratios from the TCAS-equipped cases lie in the
range of 0.003 – 0.079, compared to 0.004 – 0.058 for conventional aircraft. Several important
matters remain to be studied and tested.
One issue that needs to be resolved is how the actual communication latency will affect
Global Hawk’s response to a TCAS RA in LOS and BLOS ranges of its command and control
equipment. A pilot response delay adds to the total communication latency: the current TCAS
procedure for manned aircraft is to disengage the autopilot and take manual control in response
to an RA. Because Global Hawk does not have standard flight controls, as discussed in Section
1.2, the process of responding to RAs would be different than a manned aircraft and may result
in an increased total round-trip delay time. The maximum value for the total delay time could
provide insight as to whether TCAS would improve or degrade safety, if pilot response to TCAS
is selected as a collision avoidance solution. If the delay is greater than 15 seconds, an
alternative solution appears to be integrating TCAS into Global Hawk’s flight management
system and programming it to respond autonomously to RAs.
Another important issue for UAVs in general and for Global Hawk in particular, is the
lack of visual acquisition capability for collision avoidance and how to resolve this. Several
systems are currently being studied for use in combination with TCAS, either to increase pilot
situational awareness on the ground or to assist in autonomous collision avoidance.
TCAS failure modes must be analyzed carefully as well. A fault tree can be constructed
to pinpoint particular sources of failure and identify how each source contributes to the overall
probability of system failure and its impact on safety.
The tools used for this thesis focused on the analysis of close encounters between two
aircraft. There may also be situations where Global Hawk encounters multiple manned aircraft
while flying in the National Airspace System (NAS). A separate analysis still needs to be
64
performed that examines the likelihood and criticality of multiple-aircraft encounters that may
occur.
Civil airspace has changed in several important ways in recent years. Aircraft types,
speeds, accelerations, approach patterns, and density have all affected how aircraft encounter
each other in civil airspace. For example, a program called Reduced Vertical Separation
Minimum (RVSM) was implemented by the FAA in domestic airspace. RVSM reduces the
minimum vertical separation requirement from 2000 ft to 1000 ft at higher altitudes, thereby
enhancing airspace capacity by enabling aircraft to operate at additional flight levels. New radar
data needs to be collected to develop an encounter model that is more reflective of the types of
two-aircraft encounters that occur in the NAS. Once an updated encounter model is developed,
additional analysis of TCAS on Global Hawk can be performed.
65
References
[1] “Order 7610.4K: Special Military Operations,” U.S. Department of Transportation,
Federal Aviation Administration, 19 Feb 2004.
[2] “UAVs: A Global Perspective,” UVS International, 2004.
[3] “Global Hawk UAV Technical Data Center,” http://globalhawk.jatdi.mil/. Feb 2006.
[4] “Traffic Alert & Collision Avoidance System Concept of Operations for the RQ-4 Global
Hawk Aircraft,” Global Hawk Program Office, ASC/RG, 21 Mar 2005.
[5] Certificate of Waiver or Authorization, Global Hawk Remotely Operated Aircraft (ROA)
Operating Area, Federal Aviation Administration, 13 Aug 2003.
[6] “Introduction to TCAS Version 7,” U.S. Department of Transportation, Federal Aviation
Administration, Nov 2000.
[7] A. Drumm, L. Nivert, J. Anderson, “Information Paper: Use of TCAS/ACAS on Global

Hawk,” 2001.
[8] H. Hutchinson and K. Carpenter, “Safety of TCAS II Version 7 in the UK,” National Air
Traffic Services Ltd, London, Jun 1999.
[9] “Annex 10 to the Convention on International Civil Aviation – Aeronautical

Communications,” International Civil Aviation Organization, 11 May 1998.
[10] Kuchar, J., “Collision Avoidance System Safety Assessment Tool (CASSATT) v1.0,”
MIT Lincoln Laboratory, 26 Apr 2005.
[11] T. Miquel and K. Rigotti, “European Encounter Model,” ACASA/WP1.1.5, 4 Dec 2001.
[12] “Global Hawk Task 1, Reduced List of Data Needs for Global Hawk Safety Study,
Version 2.0,” 12 Apr 2004.
[13] Kuchar, J., “Evaluation of Proposed Changes to the ACAS Modified Tau Calculation”,
SCRSP WG A, May 2006.
66

Ada 450964

Uploaded by

Copyright:

Available Formats

Ada 450964

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ada 450964

Uploaded by

Copyright:

Available Formats

Safety Analysis of TCAS on Global Hawk using Airspace

2Lt. Thomas B. Billingsley

B.S., Aeronautical Engineering (2004)

United States Air Force Academy

Submitted to the Department of Aeronautics and Astronautics

Massachusetts Institute of Technology

1. REPORT DATE 2. REPORT TYPE 3. DATES COVERED

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

12. DISTRIBUTION/AVAILABILITY STATEMENT

15. SUBJECT TERMS

Standard Form 298 (Rev. 8-98)

in Partial Fulfillment of the requirements for the degree of

Thesis Supervisor: Dr. James K. Kuchar

Thesis Supervisor: Dr. Jonathan P. How

1.1. Thesis Objectives and Approach

1.2. Global Hawk Collision Avoidance Requirement

Figure 1-1: RQ-4 Global Hawk

Mission Control Element Launch and Recovery Element

Figure 1-2: Global Hawk Communications and Data Link Networks

Figure 1-3: Global Hawk Pilot Interface (Training Simulator)

1.3. Traffic Alert and Collision Avoidance System

1.3.1. TCAS Description

Figure 1-5: TCAS Coordination Levels

1.3.2. TCAS on UAVs

Figure 1-7: Prototype Global Hawk TCAS Display [4]

2.1. Safety Analysis Approach

2.1.1. Dynamic Simulation

Without TCAS With TCAS

Figure 2-1: Risk Ratio

P[ NMAC with TCAS]

Risk ratio is an important metric because it represents a quantitative measure of safety

Altimeter Error (ft)

2.1.2. Simulation Overview

Encounter Global Hawk

RA Response Global Hawk

Figure 2-3: Simulation Schematic

-4000 3000 3000

-6000 2000 2000

-8000 1000 1000

Vertical Speed (ft/min)

Figure 2-4: Example Scenario Output

2.1.3. Encounter Generation

Aircraft Performance Class

Vertical Maneuver Types ( joint, both aircraft )

level or rate-to-level / other

crossing / not crossing

Vertical Acceleration Duration

Horizontal Miss Distance

Vertical Miss Distance

Non-crossing Separation Tendency

Horizontal Maneuver Decision (turn / no turn)

Turn Duration and Completion Time

Ground Speed at Start and End of Encounter

Figure 2-5: Encounter Model Parameter Dependency

10000 2 [0.328396] 10000

Figure 2-6: Encounter Model Altitude Layers (Probabilities in Brackets)