RUNWAY EXIT DESIGNS FOR CAPACITY IMPROVEMENT DEMONSTRATIONS. General Research Overview

Transcription

1 RUNWAY EXIT DESIGNS FOR CAPACITY IMPROVEMENT DEMONSTRATIONS. General Research Overview Center for Transportation Research (Airport Analysis Branch) Department of Civil Engineering Virginia Tech A.A. Trani, A.G. Hobeika, B.J Kim, X. Gu and C. Zhong

2 Research Goals Development of an interactive computer model to design variable geometry runway turnoffs under realistic airport scenarios To establish a new turnoff standard to reduce runway occupancy time To test and validate the designs using computer simulations and FAA flight simulators To demonstrate possible runway capacity gains and aircraft delay reductions

3 Phases in Research Cycle Phase I: Algorithmic Development ( ) Development of aircraft landing dynamics and exit selection algorithms Development of high-speed turnoff algorithms Development of turnoff location optimization routines Phase II: Computer Model Development ( ) Development of a PC-based computer program (REDIM) Inclusion of realistic runway turnoff constraints Phase III: Test and Validation ( ) Validation of REDIM solutions and turnoff algorithms Demonstration of capacity gains and delay reductions New high-speed turnoff standard implementation

4 Runway and Aircraft Population Information Analysis Breakdown Structure REDIM Model Aircraft Landing Roll Simulation Model Runway Turnoff Candidate Generation Procedure Turnoff Characterization and Simulation Mathematical Optimization of Turnoff Locations Runway Simulation Model Runway Occupancy Time Departure and Arrival Delays Life Cycle Cost/Benefits

5 Modeling the Landing Segments Aircraft Speed Touchdown Point Exit Decision Point Nominal Landing Roll Point Turnoff Entrance Point Downrange Distance Adjusted Deceleration Second Free Roll Distance Decision Speed Nominal Deceleration Exit Speed Air Distance Braking Distance First Free Roll Distance

6 Landing Simulation Model Input Parameters Aircraft Landing Weight Factors Aircraft Mix and Aircraft Characteristics Airport Operational Data Airport Environmental Data Runway Gradient Characteristics Aircraft Operational Data Weather Percentages and Aircraft Exit Speeds

7 Normative Aircraft Deceleration Algorithm A non-linear kinematic algorithm adjusts the aircraft deceleration pattern according to current aircraft state and location of a feasible exit location 70 Nonlinear Aircraft Deceleration Model Boeing Data 60 Decision Point Speed (m./sec.) Flare Deceleration Phase Nominal Braking Phase Continuously Adjusting Braking Phase Exit Location ROT 1600 m s m s m s m s m s Distance from Threshold (mts.)

8 Improved Aircraft Deceleration Algorithm A simple algorithm adjusts the aircraft deceleration pattern according to current aircraft state and location of feasible exit locations (models both conservative and motivated behaviors) 70 Boeing Aircraft Model Constant Runway Length (2500 m.) Aircraft Velocity (m/s) Continuous Deceleration Profile Coasting Profile Exit Location 1600 m m Distance from Threshold (m.)

9 Runway Length Influence in Landing Roll Profiles The runway length influence in landing roll behavioral patterns affects the touchdown down point and the decision at which the pilot selects a runway exit Boeing Aircraft Model 70 Single Exit Location at 2100 m. 60 Variable Decision Points Velocity (m/s) Variable Touchdown Points Available Runway Length (m.) 2100 m m Distance from Threshold (m.)

10 Identification of Landing Milestone Points (Decision Point during the Landing Roll) Speed (m/s) Deceleration (m/s-s) The new landing kinematic model uses various milestones along the landing roll to differentiate various man-machine behaviors (i.e., touchdown, thrust reverser/brake application, decision point etc.) A Time (sec.) 15 Time (sec.) 20 A Using time and frequency domain analyses one can identify subtle, yet important, changes in the bahavioral patterns displayed by landing aircraft under various runway/turnoff configurations Point A in the velocity/time profile is identified as the milestone decision point on the runway or the instance at which the pilot shoots for a specific runway turnoff while adjusting its deceleration profile

11 Enhanced REDIM Capabilities (Individual Aircraft Modeling Procedure) Aircraft will be individually modeled rather than grouped (follows future airspace modeling procedures - CACI, 1992) Count Histogram Split By: ACFT TYPE Cell: B ROT (seconds) Count Histogram TD LOCATION (ft)

12 Inter-Airport Landing Roll Differences Statistical techniques such as hypothesis testing indicate that aircraft behave differently at various airports not only based upon their individual performance but also based upon the availability of runway exit locations For example REDIM is able to differentiate landing behavior patterns for various runway length conditions Histogram Histogram 60 µ = 1380 ft µ = 2042 ft Count 60 Count TD LOCATION (ft) TD LOCATION (ft) National Runway 36 (6700 f Atlanta Runway 8L (10,000 ft runway

13 REDIM Runway Predictive Capabilities A Comparison of field observations and the mathematical models developed for REDIM suggest that one can predict the runway occupancy time with high confidence. 80 Boeing Data Atlanta Runway 08-L 70 Speed (m/s) Diistance from Threshold (m)

14 REDIM Runway Predictive Capabilities (Continuation) REDIM well suited to predict individual ROT times for each runway exit Obs. Model Atlanta Hartsfield Runway 8L ROT (seconds) Overall Exit Number

15 REDIM Runway Predictive Capabilities (Continuation) REDIM predicts with reasonable accuracy the exit utilization at every runway in the airfield 80.00% % % Obs. Model Atlanta Hartsfield Runway 8L Data Exit Utilization % % % % % 0.00% Exit Number

16 Sample Landing Distribution Output Stochastic variations of the normative landing roll model applied to a large aircraft population yield landing distributions Approach Speed (m/s) Boeing 727/MD-82 Beechcraft F33 Douglas DC-10 Fokker 100 BAe Saab 340/A BAe 31 Beech BE-300 Boeing Landings 8 m/s Exit Speed Nominal Landing Distance (m)

17 Optimization Procedure Considers all possible configurations of exit locations and computes the corresponding average ROT using a set of landing distances generated in simulation module. Based on the average ROT, the optimal configuration of exit locations is determined. The optimal ranges of exit locations are found by repeating the simulation and optimization

18 Mathematical Optimization Model Performance Index M 2 Min i=1 j=1 k A (i,j) a i p j T ijk y ijk Subject to k A(i,j) y ijk = 1, for i = 1, 2,..., M ; j = 1, 2 Assignment Constraint k A(i,j) K x k k=1 x k 1 N for k = 1,2,..., K Exit Usability Constraint Maximum No. of Exits Constraint y ijk x k for i = 1, 2,..., M ; j = 1, 2; k A(i,j) Feasible No. of Exits Constraint x, y binary Logical Constraint

19 Solution Algorithm Stage (i): The situation in which we are to decide the location of the (i+1)-th exit State (s): Set of all possible locations for the i-th exit Decision (d): Set of all possible locations for the (i+1)-th exit Immediate cost function: Average ROT of aircraft which miss the i-th exit but take the (i+1)-th exit Global cost function: Accumulation of immediate cost function up to stage i Algorithm: For each stage, for every state, find the optimal decision and the corresponding global cost. At the last stage, the optimal configuration of exits is determined by back tracing

20 Turnoff Characterization Models Several approaches are possible using continuous simulation Macroscopic models (first order time varying O.D.E.) Microscopic models (multi-degree of freedom model) Regardless of the modeling approach used turnoff constraints should be imposed Turnoff safety margin against skidding Aircraft mechanical turning limitations Passenger comfort Lateral spacing available between a runway and its nearest taxiway

21 Generalized High-Speed Turnoff Geometry Turnoff Entry Point (Entry Speed) Lead-in Turn Runway Lateral Distance to Taxiway Y X Clearance Point Turnoff Tangent Exit Angle Turnoff Final Speed Point Lead-out Turn Taxiway Turnoff/Taxiway Junction

22 Aircraft Turnoff Macroscopic Model V x V V y R = Min f skid (PSI, V) - V2 g R - f s (R) R2 m -.5 ρ V2 S CL g wb lm 1- lm I zz V J n R an Inertia term dominates the first order non-linear differential system

23 Comparison of Turnoff Tracks 80 Lateral Range (m.) Wet Runway Condition 50% Safety Skid Friction 26.8 m./sec. Exit Speed Boeing Boeing Boeing Horonjeff (KC-135) Downrange Distance (m.)

24 Aircraft Speed Schedule vs. Lateral Turnoff Travel Speed (m./s.) Aircraft Constant Free Roll Decel. (-.375 m/s-s) Lateral Distance from Runway Centerline (m.) 200 Aircraft Deceleration on Tangent (m/s-s)

25 Recommended Lateral Taxiway Separation Criteria (I) The plot shows recommended lateral separation criteria between runway and taxiway centerlines for a modified FAA, 30 Deg. angled exit (427 m. transition spiral) Minimum Lateral Distance to Taxiway (m.) Tangent Deceleration = -.75 m/s-s Turnoff Entry Speed 26 m/s Exit Angle 30 Degrees 25 Degrees 20 Degrees Final Speed (m/s)

26 Aircraft Turnoff Microscopic Model Three Moment Equations Σ L = I xx p - I yz ( q 2 -r 2 ) - I xz (r + pq ) - I xy (q -rp ) - ( I yy - I zz )qr Σ M = I yy q - I xz ( r 2 -p 2 ) - I xy (p + qr ) - I yz (r -pq ) - ( I zz - I xx )rp Σ N = I zz r - I xy ( p 2 -q 2 ) - I yz (q + rp ) - I xz (p -qr ) - ( I xx - I yy )pq Three Force Equations Σ F x = m (u + wq - vr ) Σ F y = m (v + ur - wp ) Σ F z = m (w + vp - uq ) v Y q p Left Hand Sides are: 1. Aerodynamic Forces 2. Gravity Forces 3. Tire Forces 4. Thrust Forces 5. Aerodynamic Moments u X Z w r

27 Turnoff Algorithm Validation The time varying first order turning model used in REDIM 2.0 has been verified with a 4-DOF aircraft model (results shown are for a Lockheed Jetstar) Range (ft.) Lateral Lockheed Jetstar Aircraft Model R3030 Turnoff Maximum Turning Effort Design Turnoff Downrange (ft.)

28 Turnoff Algorithm Validation Tire side forces seem to be within acceptable limits for routine operation (further testing at NASA high-speed landing dynamics facilities is being performed) Side Force (lbs) Nose Gear Main Gear Time (sec.)

29 Turnoff Algorithm Validation Tire side force coefficients seem to be within acceptable limits for routine operation s(further testing at NASA high-speed landing dynamics facility is being performed) Side Force Coefficient Boeing Aircraft Model R3030 Turnoff 26 m/s Entry Speed left main gear nose gear right main gear Time (seconds) 20 25

30 Turnoff Algorithm Validation Tire side forces seem to be within acceptable limits for routine wide body aircraft operations Boeing Aircraft Model Side Force (Newtons) R3030 Turnoff 30 m/s Entry Speed Left Main Gear Nose Gear Right Main Gear Time (seconds)

31 Turnoff Tracking Comparison Tracking of the turnoff centerline geometry was accomplished successfully with a simple proportional/derivative control pseudo-pilot 400 Boeing Aircraft Model Distance (ft.) ft. Tracking Error R3030 Turnoff 30 m/s Entry Speed Desired Track Real Track Lateral Downrange Distance (ft.)

32 Runway Simulation Model to Assess Delay Single Run way Sim ulation model to assess capacity delay gains Full Monte Carlo stochastic simulation Inclusion of both arrivals and departures Coded in SIMSCRIPT II.5 simulation language Terminal Arrival Flow (Poisson, Uniform or Neg. Exponential) Final Approach Fix Departure Flow (Poisson, Uniform or Neg. Exponential) Runway

33 Capacity Gains Under Mixed Operations Consideration of single runway events representative of medium hub airports with large banks of flights Scenario Description Runway Turnoffs WAROT (s) A Baseline 5/90 Deg. Turnoffs 54.5 B Wide Throat 4/WT + 1/90 Deg C Std. 30 Deg. 4/30 Deg. + 1/90 Deg D Mod. 30 Deg. 4/Mod. 30 Deg. + 1/90 Deg E HS R3520 4/R /90 Deg Aircraft Mix: TERP A = 5%, TERP B = 35%, TERP C = 55%, TERP D = 5% All turnoff locations optimal (per REDIM 2.0 Model)

34 Capacity Gains Under Mixed Operations (II) Even under different assumption regarding number of exits HS turnoffs payoff substantial reductions in weighted average runway occupancy time (WAROT) Scenario Description Runway Turnoffs WAROT (s) A Baseline 5/90 Deg. Turnoffs 54.5 B Wide Throat 4/WT + 1/90 Deg C Std. 30 Deg. 3/30 Deg. + 1/90 Deg D Mod. 30 Deg. 3/Mod. 30 Deg. + 1/90 Deg E HS R3520 3/R /90 Deg Aircraft Mix: TERP A = 5%, TERP B = 35%, TERP C = 55%, TERP D = 5%

35 Capacity Gains Under Mixed Operations (III) Small to moderate gains in departure saturation capacity are possible under both VFR and IFR conditions (opening effect of arrival streams) Arrivals/Departure Operations (Aircraft per Hour) VMC Weather Conditions Arrival/Departures E 30 D 50/50 Mix C B Departures with A Arrival Priority 25 E Turnoff Scenarios A - FAA 90 Deg. Angle 20 D B - Wide Throat C C - Std. 30 Deg. Angle D - Mod. 30 Deg. Angle 15 B A E - HS R3520 Geometry Aircraft Mix Weighted Average ROT (s.) 60 A-5%, B-35% C-55%, D-5%

36 Reduction in Departure Delays One of the advantages of HS turnoffs is the opening of larger gaps between successive arrivals thus allowing larger number of departures and ultimately reducing departure queues 10 Average Departure Delay (minutes) Scenario REDIM 3520 Baseline Wide Throat 30 Deg. FAA 30 Deg. Mod Departure Operations per Hour 40

37 Sample Economic Results 50 Departure Delay Cost Savings (millions) Operations affected per day operations/day Life Cycle (Years) 20

38 Causal Diagram for Economic Analysis Airport Air Transport Transport Forecast Forecast Aircraft Mix and Characteristics Maximum Takeoff Weight User Costs REDIM (Exit Location Optimization Program) ROT Airframe (Capacity Program) Existing Exits RUNSIM (Runway RUNSIM (Single Runway Oeration Simulation Capacity/Delay Program) Program) Cost/Benefit Analysis Improved Exits Taxiing Taxing Times Discount Rate Operating Cost Reduction Landing Fees Construction Costs Delays Land Operation Time Savings Operations Rate Landing Fee Fee Rate Rates

39 Comparison of REDIM and FAA Standard Geometries Current high-speed geometries seem to be constrained to design speeds of m/s (with pilots taking them 8-10 m/s below design speed) It seems possible to achieve higher entry speeds if lower exit angles and slightly larger easement radius curves are introduced Longitudinal and lateral constraints should be viewed carefully with tradeoff analyses (REDIM allows users to explore infinite turnoff geometries and exit speed combinations) Some airports in the country have implemented high-speed turnoffs with little concern for lateral spacing criteria

40 Conclusions The implementation of high-speed turnoffs offers small to moderate reductions in runway service times (weighted average runway occupancy times) and their standard deviation Capacity gains arise in the form of reduced queuing delays and larger saturation departure capacities Current airline industry practices (e.g., banking) seem to justify the use of HS turnoffs whenever possible Reductions in WAROT should be viewed as a desirable alternative in a future automated ATC environment Economic gains of HS turnoffs seem to be justified over the infrastructure life cycle

41 Possible Extensions in HS Turnoff Research The algorithms developed could be ported to macro-level capacity and delay analysis programs (e.g., SIMMOD) Life cycle cost analysis to justify high-speed turnoff infrastructure (either added to SIMMOD or to be integrated to REDIM 2.0) Detailed runway-taxiway operation analyses and optimization to the gate interface Extension of the optimization procedures implemented in REDIM 2.0 to include multiple runways and multiple constraints Possible use of limited superelevation to reduce entrance radii of curvature Landing gear mechanical studies to validate computer simulation models Automation of the landing roll phase and runway turnoff navigation complementing current FAA efforts in automated ATC ground control (e.g., incorporation of automated rollout information in a proposed ASTA-2 environment)

42 Primary Goals of Flight Simulation Experiments Determine new high-speed turnoff acceptance by pilots Correlate aircraft turning dynamics with those used in REDIM 2.0 Runway length influence on pilot behaviors (to enhance REDIM 2.0) Calibration of current aircraft kinematic model and turnoff assignment model (REDIM 2.1)

43 Derived Benefits of Flight Simulation Experiments Determination of runway occupancy times for B Correlation of Turnoff Times (TOT) for REDIM and FAA standard geometries Pilot behavioral patterns on runway and turnoffs Gain understanding of pilot tracking dynamics on HS turnoffs (equivalent human plant dynamics)

44 Flight Simulation Experiments (FAA Facility at Oklahoma City) Stage I : Individual turnoff experiments Testing of five turnoff geometries to assess pilot s acceptance to two different exit angles (20 vs. 30 degrees) and two design speeds (30 vs. 35 m/s) Stage II: Turnoff location experiments Testing of four turnoff locations to determine the suitability of REDIM 2.0 optimal locations for a Boeing Attention will be given to ROT performance Stage III : Runway length influence on pilot behavior Testing of three runway lengths to ascertain pilot s behavior under different runway environmental conditions Particular attention will be given to touchdown dispersion and aircraft deceleration profiles

45 Stage I: Individual Turnoff Experiments Goal is to determine turnoff acceptance by pilots Short simulation runs (40-50 seconds) Start at a suitable point before turnoff (534 m. from turnoff) allowing pilot adjustments to a predefined entry speed Simulation ends as the aircraft decelerates to a prescribed final speed (8 m/s) Testing of five turnoff geometries Standard 30 degree angled turnoff geometry (pilot run) REDIM 3030 at 100% cornering safety margin REDIM 3020 at 100% cornering safety margin REDIM 3530 at 100% cornering safety margin REDIM 3520 at 100% cornering safety margin

46 Proposed Scenarios (Individual Turnoff Experiments) Four scenarios proposed using REDIM geometries A fifth scenario is the FAA acute angle exit with 427 m. (1400 ft.) spiral curve to be used as pilot run Scenario Angle Dist. to Safety V exit V final (Deg.) Taxiway (m.) Factor (m/s) (m/s) V VI VII VIII FAA N/A

47 Simulated Scenario for Turnoff Experiments (Stage I) Point to Achieve Desired V exit Turnoff Starts Runway Clearance Point Simulation Ends Simulation Starts Taxiway Acceleration Distance to V exit Crew Awareness Distance (Free Roll) Runway mts (Typical)

48 Expected Pilot Tasks During Turnoff Experiments Aircraft acceleration to prescribed V exit (30 or 35 m/s) Idle thrust position to maintain prescribed speed 3-4 second adjustment time before turnoff entry (no speed loss) Initiation of turnoff maneuver with no braking effort on turnoff curves Braking allowed on turnoff tangent segments Typical Aircraft State Variables During Turnoff Ac_Path Sai V_Path Aircraft Configuration Thrust : Idle Flaps : Landing Config. (30 Deg.) L. Mass : Near MALW (70,000 Kgs.) C.G. : Aft c.g. (~35% MAC) Sea Level Standard ISA Conditions Time 1/27/91 12:32:44 AM

49 Stage II: Turnoff Location Experiments Testing of four turnoff locations Short downrange location for REDIM 3020 turnoff Medium downrange location for REDIM 3020 Short downrange location for REDIM 3520 turnoff Medium downrange location for REDIM 3520 Each turnoff geometry to be tested individually to minimize pilot bias Only one turnoff provided per simulation run in CGI visual

50 Turnoff Location Experiments (Stage II) Braking Termination Runway Clearance Point Turnoff Starts Touchdown Point Braking Initiation Exit Distance Second Free Roll Braking Distance Air Distance First Free Roll Distance

51 Scenarios for Turnoff Location Experiments Four scenarios to be investigated Pilot run using FAA acute angle geometry with 427 m. spiral Scenario Speed Location Location Turnoff Number [(m/s)/knots] (m.) Description Geometry I 35 / Short VI II 35 / Medium VI IV 30 / Short VIII V 30 / Medium VIII Baseline 27 / Standard Standard

52 Stage III: Runway Length Influence on Pilot Behavior Goal is to determine the influence of runway length on pilot behavior All testing under wet pavement conditions Testing of three runway lengths Short runway m. (6,000 ft.) Average runway m. (8,000 ft.) Long runway - 3,050 m. (10,000 ft.) Runway visual range (RVR) will be 2400 ft. for all runs Night simulation in use No turnoffs will be provided to reduce pilot preferences

53 Proposed Scenarios for Stage III Experiments Runway Scenario Runway Length Runway Length Number (meters/feet) Descriptor I 1,800/5,900 Short II 2,450/8,036 Medium III 3,050/10,004 Long Aircraft Configuration Thrust : Idle Flaps : Landing Config. (30 Deg.) L. Mass : Near MALW (70,000 Kgs.) C.G. : Aft c.g. (~35% MAC) Sea Level Standard ISA Conditions

54 Runway Length Influence on Pilot Behavior Braking Termination Aircraft Reaches Prescribed V EXIT Touchdown Point Braking Initiation Second Free Roll Braking Distance First Free Roll Distance Air Distance

55 Experimental Design Pilots will be briefed prior to the simulations to explain the purpose and intentions of each experiment A minimum of ten crews to be exposed to proposed scenarios Crew cycles to minimize transfers between experiments and randomize exposure sequences Each crew will execute the experiments at least three times to sample averages A post-briefing questionnaire will be provided to compare pilot reactions to each experiment Focus group session at the end of each day