Title: Incredible Challenges of the Air Traffic Control System Modeling, Control and Optimization in the National Airspace System
1Incredible Challenges of the Air Traffic Control
SystemModeling, Control and Optimization in
the National Airspace System
- Dr. Banavar Sridhar
- NASA Ames Research Center
- Moffett Field, CA 94035
- Banavar.Sridhar_at_nasa.gov
UCSC Seminar May 27, 2004
2Outline
- What is the National Airspace System (NAS)?
- Scope
- Influence on the economy
- Transformation
- Comparison with other networks
- Technology Research in Traffic Flow Management
(TFM) - Strategic flow models
- FACET simulation and modeling capability
- Transformation of the NAS
- Questions?
3Visualization of Air Traffic Data
4(No Transcript)
5Hierarchy in TFM
- Centralized command and control structure
- Command Center, Herndon, VA
- 20 Centers
- 830 high and low-altitude sectors
6Time-Scales in Air Traffic Management
Ref Boeing/Aslaug Haraldsdottir
7Inter-Center Traffic Flow
ZBW
ZLC
ZSE
ZOB
ZMP
ZNY
ZAU
ZDV
ZKC
ZID
ZDC
ZOA
ZME
ZAB
ZFW
ZLA
ZTL
ZJX
ZHU
ZMA
8TFM problem
- Capacity
- Theoretical maximum flow rate supported by the
separation standard - Throughput
- Rate of flow realized in operation
- Efficiency
- How close is throughput to capacity?
- Objective
- Maximize flow rate to meet traffic demand
9Types of Control (TFM actions)
- Ground Delay Program
- Controlling aircraft departure time to manage
aircraft arrival rates - Metering (Miles-in-Trail)
- Controlling flow of aircraft into a center by
imposing flow restrictions on aircraft one or
more centers away - Reroutes
- Congested En-route area
- Weather
- Special Use Airspace
- Playbook
- Effort to provide a common understanding of
re-routing strategy under previously defined
situations
10Transforming the NAS
11September 11, 2001 Chronology of events
- 845 a.m. A large plane crashes into World Trade
Center north tower. - 903 a.m. A second plane crashes into World Trade
Center south tower. - 917 a.m. FAA shuts down all New York City area
airports. - 940 a.m. FAA grounds civilian flights
- 1024 a.m. FAA reports that all inbound
transatlantic aircraft flying into the United
States are being diverted to Canada. - 1230 p.m. The FAA says 50 flights are in U.S.
airspace, but none are reporting any problems.
12(No Transcript)
13Commercial Transport Enplanements
800.0
700.0
Large Air Carrier Passenger Enplanements
(Millions)
600.0
500.0
400.0
Forecast
Actual
300.0
200.0
100.0
0.0
05
06
07
08
09
10
11
12
13
14
95
96
97
98
99
00
01
02
03
04
Calendar Year
Source 1990-2002 U.S. Air Carriers, Form 41,
U.S. DOT 2003-2014 FAA Forecasts
14System Reaching Saturation
Target Year
Source LMI, Alternatives for Improving
Transportation Throughput and Performance, March
2002
15What is at stake in air transportation?
- Lost growth and output from air transportation
due to demand outstripping capacity - Unserved demand of 180 billion Revenue Passenger
Miles (RPMs) resulting lost annual economic
output of 23 billion by 2015 (23B does not
include additional impact of lost user
productivity) - Major policy / operational alternatives within
the current air transportation architecture
recaptures only a small fraction of unserved
demand and economic output - Large, continuing security costs to protect the
system from acts of terrorism - Difficult to measure efficacy
Rising costs, rising frustrations, lost
opportunities
Source LMI, Alternatives for Improving
Transportation Throughput and Performance, March
2002
16What makes NAS different?
- Safety is paramount
- Human-in-the-loop decision making at all levels
- System capacity limits established by human
performance - Changes need to be done while the system is in
operation - Difficulty in modeling user reaction to events
- Availability/absence/uncertainty of information
- Need to get consensus among various parties FAA,
unions, airlines, aircraft manufacturers, etc. - Status of automation/decision support tools
17Strategic Flow Models
18Outline
- Strategic Flow Models
- Linear Time-variant Dynamic System representation
- Flow Matrix
- Forcing Function
- Example
- Bounds on the Model
- Concluding remarks
19Traffic Flow Models
Detailed Aggregate
Deterministic CTAS, FACET, CRCT Flow models
Stochastic Sector Congestion Queuing models
- Detailed models
- Useful for developing algorithms affecting
individual aircraft - Controller/Traffic Manager decision support tools
- Aggregate models
- Useful for understanding the general behavior of
the system - Effectively address system uncertainties and long
term behavior
20Traffic Flow
Different Centers
Atlanta Center on different days
21Linear Time-Varying Dynamic Traffic Flow Model
22A Matrix (May 6, 2003 6 hour average, 5-11P.M,
PST)
23Variation of A Matrix
Daily Variation May 6,7,8 2003 5-11 P.M
Variation during May 6 11P.M- 5A.M, 5-11 A.M,
11A.M-5 P.M
24Variation of A Matrix During May 6, 2003 5-11 P.M
Hourly Variation
Two-Hour Variation
25Modeling A(k) Constant for different time
intervals
26Normalized mean and standard deviation of Error
27Modeling of the forcing function (BuCw)
Departure Counts (May 6, 2003 Every 10 Minutes)
28Effect of using A from previous days
Atlanta Center (ZTL) Traffic Counts for May 8,
2003 predicted using May 7 and May 6 flow
matrices
29Departure Counts (May 6-8, 2003 Every 10 Minutes)
May 6
May 7
May 8
30Modeling departures using mean value
31Error Bounds for Model
32Modeling departure errors as gaussian
33Concluding Remarks
- Described linear time varying models to represent
traffic flow for developing strategic TFM
decisions. - Linear dynamic traffic flow system model with a
slowly varying transition matrix and Gaussian
departure representation adequately represents
traffic behavior at the Center-level. - Error bounds around nominal traffic counts in the
Centers was described. - Numerical examples presented using actual
traffic data from four different days to
demonstrate the model characteristics. - Advantages
- Unlike trajectory-based models, these models are
less susceptible to uncertainties in the system, - The model order is reduced by several orders of
magnitude from 5000 aircraft trajectories to 23
states at any given time - Tools and techniques of modern system theory can
be applied to this model because of its form. - Capabilities of this class of models for
strategic traffic flow management will be
explored in the future.
34Future ATM Concepts Evaluation Tool (FACET)
35Future ATM Concepts Evaluation Tool (FACET)
- Environment for exploring advanced ATM concepts
- Balance between fidelity and flexibility
- Model airspace operations at U.S. national level
(10,000 aircraft) - Modular architecture for flexibility
- Software written in C and Java programming
languages - Easily adaptable to different computer platforms
- Runs on Sun, PC and Macintosh computers
- 3 Operational Modes Playback, Simulation, Hybrid
- Used for visualization, off-line analysis and
real-time planning applications
36FACET Architecture
Winds
Applications Air and Space Traffic
Integration Airborne Self-Separation Data
Visualization Direct-To Analysis Dynamic
Density System-Level Optimization Traffic Flow
Management
Flight plans Positions
Route Parser Trajectory Predictor
Climb Cruise Descent
Centers Sectors Airways Airports
37FACET Displays
Traffic
Winds
3-D
Convective Weather
38ATL Arrivals (Purple) and Departures (Green)
39FACET Display
16
17
40Severe Weather Playbook Reroutes(Eastbound
Traffic over Watertown)
41Alternative effects of TFM actions
42Integrated traffic counts in ZMP Sector 16
A Nominal Counts, B Playbook Reroute, C
Playbook MIT, D Playbook MITLocal Reroute.
43EWR and LGA Delay Contours
44FACET for AOC Applications
- March 2001 request by Aircraft Dispatchers
Federation (ADF) team to increase NASA research - FACET modified to work with Aircraft Situation
Display to Industry (ASDI) data - Developed a version of FACET for AOC use
- Enable efficient operations planning by AOC
- Risk analysis
- Departure planning and congestion assessment
- Integration with weather
- Commercial Technology Office to license the
software to Flight Explorer (FACET release in FE
6.0, October 2004)
45Comments from Airline Dispatchers
- I usually (almost always) plan for the worst
case scenario. The ability to tailor fuel uplift
to individual flights with a very high degree of
confidence in the probability of en route delay
is worth tens of millions of dollars to the
airlines. It costs me about 400,000 a year to
carry one additional minute of fuel on each
flight. If I am carrying an average of 35
minutes, and I really only need a system-wide
average of 15 minutes, that would be worth 8
million per year to my airline alone. - I would find the predictive data very helpful in
planning routing and fuel load. - The concept of alerting a dispatcher regarding
ATC sector overload and inbound ATC reroutes is
an excellent idea. - To the dispatcher at the desk, I think it would
give him a huge advantage to see, understand,
plan, fuel and brief the crews on possible ATC
initiatives based on volume issues. - FACET would be great because when the Command
Center says, or the ATC community says These are
your three options, we could say You know, you
might want to consider a fourth option here that
we could game or model on FACET. - Weve been asking for a common situation display
for a long time. This may be the basis for it.
46Transformation of the NAS
47Commission on the Future of the United States
Aerospace Industry
- Recommendations
- 2 The Commission recommends transformation of
the U.S. air transportation system as a national
priority. - Rapidly deploy a new highly automated ATM system
- 3 The Commission recommends that the U.S.
create a space imperative. - 9 The Commission recommends that the federal
government significantly increase its investment
in basic aerospace research, which enhances U.S.
national security, enables breakthrough
capabilities, and fosters an efficient, secure
and safe aerospace transportation system.
48Partners in Development ofNational Plan for the
Future NAS
- JPO develops and maintains National
Transformational Plan which includes - Associated policies, technology, processes
- Overall operational concepts
- Supporting research
- Implementation strategies
- Policy and implementation commitments
Strategic Plan Perf. Goals
Strategic Plan Perf. Goals
Strategic Plan Perf. Goals
RD Plan
Pgm. Plan
49Capability
Time
50Issues in the transformation of NAS
- Automation
- Need
- Impact
- Human Factors
- Policy
- Regulations
- Certification
- Equity
- Allocation of scarce resources
- Sharing of information
- Cost of equipment
- Integration with existing systems
- Software verification and validation