Sample Decision Support
ATM Pr ojects
Integration of National Airspace System Models
Modeling NAS Infrastructure Investments
J. Kobza, A.A. Trani, H.D. Sherali, H. Baik and C. Quan
Sponsor
FAA Office of Program Analysis and Investment
(Dr. Randy Stevens)
NEXTOR - National Center of Excellence for Aviation Operations Research
1 of 29
Integration of National
Airspace System
Models
Project Objectives
•
To investigate alternatives to integrate existing National Airspace
System models as a suite of decision making tools
•
To connect new algorithms with existing NAS airspace and airport
simulation models (e.g., SIMMOD, RAMS, etc.)
Status of the Project:
•
An object-oriented programming mechanism to connect to the
architecture of both models with external processes is under
development
•
The approach will be demonstrated next using a generic ground
network simulation capability to extend RAMS
NEXTOR - National Center of Excellence for Aviation Operations Research
2 of 29
Integration of Airspace/Airfield Models to
External Processes
•
To demonstrate the integration process RAMS and SIMMOD are
being used as testcase
+ RAMS - The Reorganized Airspace Mathematical Simulator
+ Airspace operational model developed by Eurocontrol to
simulate advanced airspace concepts
+ SIMMOD - the FAA airfield and airspace model
•
Curiously, none of these models was designed with a plug-in
architecture in mind (SIMMOD designed in the 1970s and RAMS
in the 1980s)
•
Both are NARIM tools
•
RAMS lacks detailed ground simulation capabilities (perceived as
weakness by many)
NEXTOR - National Center of Excellence for Aviation Operations Research
3 of 29
Integration
Appr oach
•
Develop an integration procedure to link dynamic events into
RAMS and SIMMOD
•
Ground simulation component is the common element for the
integration procedure (RAMS has little ground logic)
RAMS
SIMMOD
Runway Object
Runway Ops. Functions
(NodeA and B events)
VPI Object-Oriented
Ground Model
Mimics Ground ATM
Area of Interest
NEXTOR - National Center of Excellence for Aviation Operations Research
4 of 29
Object-Oriented Ground ATM Model
Link_Class
Node_Class
Point_Class
Controller_Class
SP_Class
Flight_Class
AirpNwk_Class
Aircraft_Class
Message_obj
AcftFoll_Class
NEXTOR - National Center of Excellence for Aviation Operations Research
5 of 29
Model F eatur es
•
Continuous (or discrete-event) micro-simulation model
•
The total delay due to network congestion can be analyzed
•
Implements an “aircraft-following” model
+ The dynamic behavior of moving aircraft can be captured
•
Communication delays and frequency congestion is incorporated
•
Continuous updating the shortest path (Quasi-Dynamic)
+ More realistic results in reasonable computation time.
•
Ground control model attempts to produce a continuous dynamic
equilibrium
•
Three types of data structures (Array, Linked-list, Mixed)
depending on the data type
NEXTOR - National Center of Excellence for Aviation Operations Research
6 of 29
Departur e State Transition Diagram
yes
Parked
(at gate)
Call for
push-back
clearance
Pushingback
Waiting to
Taxi
Call for
Taxi
Clearance
Got
Cleance?
Communication
Taxing
Waiting to
Taxi
No
Communication
Get
Cleance?
No
Waiting in
Line
(R/W dep.
Queue)
Gate Hold
Area
Holding
(Holding
Area)
Waiting on
Runway
Call for
Take-off
clearance
LiftOff
Rolling
(until initial climb rate
and speed)
Get take-off
Clearance
Communication
no
clearance
(with const. accel.
until flying speed)
Communication (pilot standpoint)
Sending
(Request)
Message
Waiting
for
Response
Recieving
(Control)
Message
Sending
Confirm.
NEXTOR - National Center of Excellence for Aviation Operations Research
7 of 29
Contr oller State
Transition Diagram
Proc.
Oper.
Const's
Sending
Control
Message
Standby
Waiting
for
Confirm
Receiving
Request
Message
Standby
Sending
Control
Message
Waiting
for
Confirm
Process
Conflict
Situation
Resolving
Conflict
Checking
it again
Sending
Control
Message
Waiting
for
Confirm
NEXTOR - National Center of Excellence for Aviation Operations Research
8 of 29
Air craft-F ollo wing Logic (second-order
model)
Distance control logic
&x&nt ++1∆t = k[( xnt − xnt +1 ) − D], where k = design parameter, D = safety distance
If &x&nt ++∆1 t > &x&max , then &x&nt ++∆1 t = &x&max
Speed control logic
&x&nt ++∆1 t = k ( x&nt − x&nt +1 ) if &x&nt ++∆1 t > &x&max , then &x&nt ++∆1 t = &x&max
Distance-speed control logic
&x&nt ++∆1 t
= α [ x&nt
−
x&nt +1 ]
( x&nt ++∆1 t ) m
where, α = C t
[ xn − xnt +1 ]l
NEXTOR - National Center of Excellence for Aviation Operations Research
9 of 29
Collision Detection at Intersections
Second or later flights on this link
(do follow the leading flight
by car-following logic)
First flight on this link
(Need to check
the potential collision)
Conflicting flights coming
toward thecommon intersection
15.6(sec)
Current position
of flight
20.8(sec)
Start point
of Intersection
Legend :
the current operation direction
30.7(sec)
F2
Expected arrival time to the start
point of the common intersction
(ETi)
NEXTOR - National Center of Excellence for Aviation Operations Research
10 of 29
Intersection Resolution
where,
AG: Acceptable Gap
ET: Expected arrival time to the intersection
P: Priority of a certain flight
Algorithms
Every delta seconds?
Yes
Is this the first aircraft
in the Link?
No
Yes
For all first flights on the backward stars of this flight's next node
Yes
Pthis == High
No
Exist conflict flight
having Pconf (= High)?
Exist conflict flight
having Pconf (= High)?
No
Yes
Yes
|ETthis - EThigh| < AG
|ETthis - ETconf| < AG
No
No
Yes
(then slow down)
Yes
(then FIFO)
ETthisnew = EThigh + AG
ETthis >= ETconf
No
Yes
No
Exist conflict flight
having Pconf (= LOW)?
Yes
No
Yes
|ETthis - ETconf| < AG
Yes
NEXTOR - National Center of Excellence for Aviation Operations Research
11 of 29
Sample Air port Netw orks
C1-32
B1-22
M1
L1-10
M2-6
E1-15
K1-19
F111
5000ft
DCA
G116
H1-18
M712
M1320
M21
5000
ft
ORD
NEXTOR - National Center of Excellence for Aviation Operations Research
12 of 29
Sample Data Files
File 1: Node Data
1. NodeId
2. NodeType (Gate/Taxiway/Runway)
3. TT
4. Restriction
File 2: Link Data
1. FromNode
2. ToNode
3. LinkType
4. LinkId
5. Restriction
6. Direction
File 3: Aircraft Model
1. Model_Id
2. Wingspan
3. Length_f
4. MaxAccel
5. Etc.
File 4: Flight Plan
1. Flt No.
2. Acft_Type
3. Flt_Type
4. S_Time: Start time
5. O_Node: Origin Node
6. D_Node: Destination node
7. etc
File 5: MinPath Matrix
from\to 0
1
2
0
0
1
1
1 -9999
0
2
2 -9999 -9999
0
3 -9999 -9999 -9999
4 -9999 -9999 -9999
5 -9999 -9999 -9999
3
4
3
3
-9999
4
-9999 -9999
0
4
-9999
0
-9999 -9999
NEXTOR - National Center of Excellence for Aviation Operations Research
5
3
4
5
4
5
0
13 of 29
Rele vance of the Integration
•
RAMS is the subject of three integration processes to enhance its
capabilities as a decision support tool (our viewpoint)
+ RAMS-OPGEN
+ RAMS-Weather
+ RAMS-Ground_Sim
•
While moderate gains in airspace capacity might be possible in
NAS under various future Concept of Operations the airport
capacity limits of the system remain a critical ATM issue
NEXTOR - National Center of Excellence for Aviation Operations Research
14 of 29
Modeling N AS Infrastructur
e In vestments
Research Objectives:
•
Develop a modeling framework to estimate NAS investments, their
effects in NAS operational metrics (delay, capacity, etc.) and
macroeconomic impacts
Approach:
•
A macroscopic modeling framework to assess the development of
NAS has been postulated using Systems Dynamics
•
Backbone of the modeling framework has been developed
•
Implementation of the framework using State-Flow1 is in progress
(proof-of-concept)
1.State-Flow is a simulation environment developed by the Mathworks in Natick, MA.
NEXTOR - National Center of Excellence for Aviation Operations Research
15 of 29
Investment Models Cycle (NARIM)
• A conglomerate of models (belonging to three specific domains per Odoni et al., 1996):
+ Operational models (RAMS, SIMMOD, ACM, etc.)
+ Architectural models (NASSIM - National Airspace Systems
Simulation)
+ Investment analysis modeling
Airline Schedule
NAS Resource Simulator
Optimization
LOS = f (Safety, Delay,etc.)
Concept of
Operations
Demand Function
Socio-Economic Model
Critical F eedback
Cost/Benefit
Analysis
NEXTOR - National Center of Excellence for Aviation Operations Research
16 of 29
Complementary Approach to NARIM
• We are not trying to create a new NARIM
• The framework being investigated could be used (if successful)
as an integration approach that binds NARIM models together
• The difficulty with the current approach in NARIM is that each
case scenario needs to be studied independently and models run
individually
• The Systems Dynamics approach proposed uses a response curve
approach coupled with a dynamic model to ‘encapsulate’ agent
behavior researched elsewhere (i.e., ATM Group # 1)
• The approach tries to answer questions without running each
model every time
NEXTOR - National Center of Excellence for Aviation Operations Research
17 of 29
Systems Dynamics
Variables (STELLA
II TM Repr esentation)
• Level variables - represent accumulation of resources (also called
state variables of the system)
• Rate variables - represent the rates of change of level variables
• Auxiliary variables - other variables used to estimate rate
variables
Multiplier
Air Transport Demand
Feedback
Loop
ATD Rate of Change
Population
Level of Service
ATS Capacity
GNP
NEXTOR - National Center of Excellence for Aviation Operations Research
18 of 29
Issues in Modeling NAS Agents
• Level of aggregation
+ NAS agents can be modeled individually or using some level
of aggregation
+ The level of aggregation in modeling is very critical
• Who should be considered?
+ FAA (infrastructure facilities such as centers, sectors, CNS
assets; staff,etc.)
+ Airlines (fleet assets, gaming strategy)
+ Users (the basis for air transport demand)
• Modeling agent dynamics
+ Part of NEXTOR’s challenge is to research how agents
respond to short and long term system changes
+ Agent dynamics are being studied concurrently
NEXTOR - National Center of Excellence for Aviation Operations Research
19 of 29
Modeling Framework Structure
Economic Model
Population, Labor, GNP
yields Demand
Investment Strategy
What?, Where?
When?, How much?
Air Transportation System Model
ATC
Airports
Airlines
Air Transportation System Metrics
Safety Capacity
Delay
Load Factor
NEXTOR - National Center of Excellence for Aviation Operations Research
20 of 29
Example of Airline Asset Modeling
• The following example illustrates in a very simplified way how
an airline might add or retire fleet assets
Desired Fleet Age
AT Demand
Retiring Acft Normal
Aircraft Fleet
Airline Market Share
From Economic
Model
Fleet Age
New Aircraft
Load Factor
Total Capacity
Retired Aircraft
Seat Capacity
In practice the aircraft fleet level variable could be replaced
by n levels representing various aircraft age groups
NEXTOR - National Center of Excellence for Aviation Operations Research
21 of 29
Example of FAA Resources Modeling
Staffing Level
Frac to Staff
New staff
From Economic
Model
Traffic Demand
Retirements
FAA Unit Level Capacity
FAA UL Budget
Frac to Comm
Communication Assets
Comm Proc
Level of Service
Comm Tech multiplier
Comm Ret
Surveillance Assets
Frac to Sur
Sur Tech multiplier
Sur Proc
Sur Ret
NEXTOR - National Center of Excellence for Aviation Operations Research
22 of 29
US Socio-Economic Development Model
(SEM)
• The model includes the following relevant variables from the
following sectors:
+ Industrial
+ Infrastructure (includes transportation)
+ Social
+ Population
• Predicts socio-economic activity over time
• Predicts demand functions (regional, city pair, national, etc.)
• Ties in with air transport model (described in the following
pages)
• An adaptation of ACIM can be made to link traditional economic
parameters (GNP) and the technical model
NEXTOR - National Center of Excellence for Aviation Operations Research
23 of 29
US Air Transportation System Development
Model (ATSDM)
• A technology model to predict air transport specific variables
consists of the following sectors:
+ Air Traffic Control
+ Airlines
+ Airport
• Tracks airport infrastructure, air fleet, and ATC assets over time
• Contains the dynamics of agents evolve as system state variables
change
• Uses the results of SEM (economic model)
NEXTOR - National Center of Excellence for Aviation Operations Research
24 of 29
Why is Feedback Modeling Important?
• Traditional NAS metric predictions are based on a point
performance model (base and horizon years)
• Agents (FAA, airlines, users, and the economy itself) exhibit
various degrees of adaptive behavior that are impossible to
capture with such a model
• Feedback occurs naturally as these agents adapt over time and
thus a dynamic model offers an interesting alternative (as long as
we know how this adaptation behavior occurs)
• A dynamic model is capable of responding to external inputs as
the model evolves (gaming is possible with such a model)
• Life cycle analysis is a secondary benefit of a dynamic model
without gross assumptions in evolution
NEXTOR - National Center of Excellence for Aviation Operations Research
25 of 29
Model with Explicit Capacity Formulation
Population
ROC Population
Jobs
Taxes
Captivity Factor
Region Attractiveness
AT Demand
Revenue
Third Order
Feedback System
ROC ATD
AT Sys Attractiveness
Frac of Rev to Investment
Cost
Investment to AT
System Delays
Demand Capacity Ratio
AT Infrastructure Const
Capacity of AT System
ROC Capacity
Unit Cost of Capacity
NEXTOR - National Center of Excellence for Aviation Operations Research
26 of 29
Scenario Analysis
• Scenarios of interest to air transportation service providers, users
and FAA can be constructed and analyzed
• The proposed modeling framework can be used to understand
important investment strategies for NAS and how these affect
NAS metrics
+ Level of ATC automation
+ Level of airline investments and return
+ Free flight impacts on NAS metrics
+ Levels of investment needed
• Perhaps most important is the question when to invest to achieve
any benefit on investment (the classical ROI airline and FAA
question)
NEXTOR - National Center of Excellence for Aviation Operations Research
27 of 29
Expected Results of the Model
• Notional patterns for various NAS capital investment policies
(1.2) Nominal
Nominal
(0.80) Nominal
(0.80) Nominal
Nominal
(1.2) Nominal
Year
NEXTOR - National Center of Excellence for Aviation Operations Research
28 of 29
Closing Remarks
• Systems Dynamics is just one method to understand ties between
variables of social, economic and technological order in complex
dynamic systems like NAS
• A properly developed and calibrated model of NAS behavior
could serve as a living laboratory where FAA can examine
investment policies before implementation
• The approach presented here is a first order modeling effort that
complements the our understanding of the NARIM umbrella of
models
• While it can be argued that establishing causality between
technology and economic variables is difficult it is important to
realize that many concurrent (or previous studies and models)
can be useful to sort out these details
NEXTOR - National Center of Excellence for Aviation Operations Research
29 of 29