EDS/ETD Deployment Program: Modeling and Simulation Approach

Transcription

1 EDS/ETD Deployment Program: Modeling and Simulation Approach Presentation at the TRB Annual Sunday Simulation Workshop Omni Shoreham Hotel, Washington, D.C., January 12, 2003 By: Evert Meyer, Ph.D. Mark Lunsford Principal, Leigh Fisher Associates Principal, Leigh Fisher Associates 1

2 Role of Modeling and Simulation Develop and test new generic screening concepts: Hybrid screening solutions, Pre-ticketing versus post-ticketing screening, Simultaneous ticketing/screening (e.g., ETD-on-a-Stick), and In-line/bag room solutions. Estimate EDS and ETD equipment requirements for each airport. Assess system performance at each airport: Queue lengths; and Average, 95th percentile, and max waiting times. Assist with selection of preferred solution for each airport. Continuous improvement (beyond December 31, 2002). 2

3 Use of Simulation for Terminal Planning in a New Environment Existing terminal design standards for security and baggage processing space allocation have become obsolete (e.g., FAA Terminal Planning and Design Guidelines and International Air Transport Association Airports Development Reference Manual). Need for analytical performance evaluation (e.g., queuing theory) or simulation on case-by-case basis. Security/baggage screening is a stochastic process (randomness in processing rate) resulting in overall system performance that is non-linear and may be sensitive to small changes. Processing time Demand/capacity ratio 3

4 Example of Nonlinear Queuing Effects and Variable Passenger Processing Times Potential risks of using manufacturers processing times and rates: Idealized laboratory rates versus real-world achievable rates often differ by a factor of 2; nd Could result in underestimating waiting times and queue lengths by a factor of 4. Potential risks of using average processing times as constants: Example: System A has a constant service time of 2 min; System B has a variable service time with an average of 2 min and a standard deviation of 1.5 min. Then, a standard rule of thumb from queuing theory predicts that, for the same demand pattern, the waiting times and queues at System B could be twice as long as those at System A. Potential combined risks: Using manufacturers rates and ignoring the variance in processing times rates could result in underestimating waiting times and queue lengths by a factor of 8 (say 5 to 10). 4

5 Alarm Rates and EDS Throughput A 5-point reduction in alarm rate results in about a 25% increase in EDS throughput. 300 EDS Throughput Rate (bags per hour) Secondary screeners 3 Secondary screeners 2 Secondary screeners 1 Secondary screener EDS Alarm Rate 5

6 Boeing Team Modeling and Simulation Tools EDS-SIM (LFA) Bags per Hour ZONE LEVEL DESIGN DAY BAGGAGE FLOW Zone : 8B JFK - NEW YORK(KENNEDY) NY Capaci ty 600 Req'd 565 bph bph EDS 300 Capaci ty PaxSim (Preston) 0 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 Time of Day EDS throuhput Design Confidence Interval ETD throughput Expected Mean Flow Rate Airlines served : 46%, 100%, 100% Avg % of Bags Avg % of Bags Total Design Screened with Screened with Day Bags EDS ETD 4, % 8.9% ZONE 8B EDS CAPACITY Flow Model (LFA) Avg. EDS Avg. ETD Hours of Utilization Utilization Active Use in During Active During Active Design Day Hours Hours % 8.5% RECORD AutoMod (commercially available proprietary model) 6

7 Modeling and Simulation Strategy Airport Types Assessment TSA Review Survey Design GROUPS A/B * Most challenging terminals Flow Model PaxSim AutoMod (as necessary) * Others Flow Model EDS-Sim PaxSim or AutoMod (as necessary) GROUP C Flow Model EDS-Sim (as necessary) GROUP D Flow Model Flow Model = High-level queuing model (concept stage; preliminary requirements) EDS-Sim = Discrete-event simulation model (design stage; final requirements) PaxSIm = Discrete-event simulation model with animation output (design stage; final requirements) AutoMod = Discrete-event simulation model with 3D animation output (design stage; final requirements) 7

8 Flow Model Bags per Hour ZONE LEVEL DESIGN DAY BAGGAGE FLOW Airlines served : 46%, 100%, 100% Zone : 8B JFK - NEW YORK(KENNEDY) NY 0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 Time of Day EDS throuhput Design Confidence Interval Total Design Day Bags Avg % of Bags Screened with EDS Avg % of Bags Screened with ETD 4, % 8.9% Hours of Active Use in Design Day Avg. EDS Utilization During Active Hours Avg. ETD Utilization During Active Hours % 8.5% Capaci ty Req'd 565 bph 360 bph EDS Capaci ty ETD throughput Expected Mean Flow Rate ZONE 8B EDS CAPACITY RECORD Based on fluid approximations with statistical surging. Used for quickly estimating EDS and ETD equipment requirements for a variety of protocols. Integrated with database containing: Official Airline Guide (OAG) schedules, Industry average trends/data, Airline/airport-specific data, and Data collected in field. Calibrated to the 10-min performance criteria and redundancy. Rapid set-up and application typically less than 8 h. 8

9 EDS-SIM Used for rapid concept evaluation and requirements for most airports where a standard solution is proposed: Modular design to implement solutions from standard set of templates; Captures unique metering/dependency effects associated with pre-ticketing or post-ticketing solutions; and Evaluates requirements based on 95-percentile, 10-min design objective. Relatively rapid set up and processing time: Group A: 3 to 5 days; Group B: 2 to 3 days; and Group C: 0.5 to 1 days. Integrated with database containing: OAG schedules, Industry average trends/data, Airline/airport-specific data, and Data collected in field. Technical details: Discrete event, simulation model; and Pre-processor, a simulation engine, and a post processor modules. 9

10 PaxSim Fully animated simulation program: Passengers and their luggage modeled visually as individual objects; State-of-the-art, object-oriented modeling environment; and Superior graphics capabilities. Appropriate for difficult, controversial layouts: Assist with the refinement of requirements and concept layouts as well as their visualization; Quantify and visualize the impact of congestion on passenger flows and dwell times; and Obtain buy-in for concepts with unique flow/queuing characteristics. Relatively rapid set up and processing time: Group A: 8 to 10 days; Group B: 5 to 10 days; and Group C: 4 to 8 days. Technical details: Object Oriented Model Development Environment, C/C++ Native Code, Linux Platform, Rule-based decision structure, and Free-flow grid paradigm (not node-link structure as used older technology products). 10

11 Data Management Data sources Schedule (OAG) Airline Data Transaction statistics Check-in Splits Load Factors % O&D Field data Detailed time and queue statistics for 60 airports including all CAT I & II Site survey data for all other airports (429+) Database Repository Relational Database Processing scripts (SQL/Delphi) Schedules August 2000 March 2003 Charter Activity Processed data for: Bags/party distribution Party size distribution Check-in splits Load Factors O&D splits Process time distributions Probability Fitted Distributions Passenger process time (seconds) GAMMA NORMAL Observed 11

12 Major Modeling and Simulation Tasks Every airport has received some type of modeling: Flow Models: 429 airports EDS-Sim models: 127 terminals, 90 airports PaxSim models: 60 terminals, 30 airports Development of automation processes and tools necessary for airport system analysis: System-wide model to evaluate policy/machine allocation strategies; Flow model development and support for use by planners/architects to assess machine requirements; and Analysis tools in support of continuous improvement program. 12

13 MCO Post-Ticketing Drop-n-Go Concept 13

14 LAX Terminal 1 (SWA) Simultaneous Screening/Ticketing 14

15 MCI Simultaneous Ticketing/Screening 15

16 ORD T1 Animation (Original Concept) 16

17 ORD T1 Animation (Revised Concept) 17

18 ORD T1 Animation (Close-Up) 18

19 LAX TBIT 19