A Small Aircraft Transportation System (SATS) Demand Model

Transcription

1 NASA/CR A Small Arcraft Transportaton System (SATS) Demand Model Dou Long, Davd Lee, Jesse Johnson, and Peter Kostuk Logstcs Management Insttute, McLean, Vrgna June 2001

2 The NASA STI Program Offce... n Profle Snce ts foundng, NASA has been dedcated to the advancement of aeronautcs and space scence. The NASA Scentfc and Techncal Informaton (STI) Program Offce plays a key part n helpng NASA mantan ths mportant role. The NASA STI Program Offce s operated by Langley Research Center, the lead center for NASA s scentfc and techncal nformaton. The NASA STI Program Offce provdes access to the NASA STI Database, the largest collecton of aeronautcal and space scence STI n the world. The Program Offce s also NASA s nsttutonal mechansm for dssemnatng the results of ts research and development actvtes. These results are publshed by NASA n the NASA STI Report Seres, whch ncludes the followng report types: TECHNICAL PUBLICATION. Reports of completed research or a maor sgnfcant phase of research that present the results of NASA programs and nclude extensve data or theoretcal analyss. Includes complatons of sgnfcant scentfc and techncal data and nformaton deemed to be of contnung reference value. NASA counterpart of peer-revewed formal professonal papers, but havng less strngent lmtatons on manuscrpt length and extent of graphc presentatons. TECHNICAL MEMORANDUM. Scentfc and techncal fndngs that are prelmnary or of specalzed nterest, e.g., quck release reports, workng papers, and bblographes that contan mnmal annotaton. Does not contan extensve analyss. CONTRACTOR REPORT. Scentfc and techncal fndngs by NASA-sponsored contractors and grantees. CONFERENCE PUBLICATION. Collected papers from scentfc and techncal conferences, symposa, semnars, or other meetngs sponsored or co-sponsored by NASA. SPECIAL PUBLICATION. Scentfc, techncal, or hstorcal nformaton from NASA programs, proects, and mssons, often concerned wth subects havng substantal publc nterest. TECHNICAL TRANSLATION. Englshlanguage translatons of foregn scentfc and techncal materal pertnent to NASA s msson. Specalzed servces that complement the STI Program Offce s dverse offerngs nclude creatng custom thesaur, buldng customzed databases, organzng and publshng research results... even provdng vdeos. For more nformaton about the NASA STI Program Offce, see the followng: Access the NASA STI Program Home Page at E-mal your queston va the Internet to help@st.nasa.gov Fax your queston to the NASA STI Help Desk at (301) Phone the NASA STI Help Desk at (301) Wrte to: NASA STI Help Desk NASA Center for AeroSpace Informaton 7121 Standard Drve Hanover, MD

3 NASA/CR A Small Arcraft Transportaton System (SATS) Demand Model Dou Long, Davd Lee, Jesse Johnson, and Peter Kostuk Logstcs Management Insttute, McLean, Vrgna Natonal Aeronautcs and Space Admnstraton Langley Research Center Hampton, Vrgna Prepared for Langley Research Center under Contract NAS June 2001

4 Avalable from: NASA Center for AeroSpace Informaton (CASI) Natonal Techncal Informaton Servce (NTIS) 7121 Standard Drve 5285 Port Royal Road Hanover, MD Sprngfeld, VA (301) (703)

5 LOGISTICS MANAGEMENT INSTITUTE A Small Arcraft Transportaton System (SATS) Demand Model NS004S1 Executve Summary The Small Arcraft Transportaton System (SATS) s envsoned to take advantage of technology advances n arcraft engnes, avoncs, arframes, navgaton equpment, communcatons, and plot tranng to make t the new generaton of general avaton (GA) that wll let people travel from small arports. SATS not only wll help to break the grdlock at large commercal arports by dvertng traffc to nonhub small arports, t also can generate new ar traffc demand as t can reduce the door-to-door tme for travels from or to a place close to a small arport. Wth hgh speed arcraft, numerous arports, an affordable cost, and easy plot tranng, SATS can provde better door-to-door travel tme, enhance moblty, and stmulate busness actvty. Ths report explans our SATS demand modelng at the natonal, arport, and arspace levels. We constructed a seres of models followng the general systems engneerng prncple of top-down and modular approach. Our three prncpal models are the SATS Arport Demand Model (SATS-ADM), SATS Flght Demand Model (SATS-FDM), and LMINET-SATS. SATS-ADM models SATS operatons, by arcraft type, from the forecasts n fleet, confguraton and performance, utlzaton, and traffc mxture. Gven the SATS arport operatons such as the ones generated by SATS-ADM, SATS-FDM constructs the SATS orgn and destnaton (O&D) traffc flow based on the soluton of the gravty model, from whch t then generates SATS flghts usng the Monte Carlo smulaton based on the departure tme-of-day profle. LMINET-SATS, an extenson of LMINET, models SATS demands at arspace and arport by all arcraft operatons n the Unted States. The models presented n ths report can be the powerful tools to polcy decson makers n ar traffc system plannng, especally n SATS. The models wll help proect SATS demands for arports and arspace. The models are bult wth suffcent parameters to gve users flexblty and ease of use to analyze SATS demand under dfferent scenaros. Several case studes are ncluded to llustrate the use of the models, whch also are helpful n desgnng the new ar traffc management system to cope wth SATS traffc.

6 The fgures we present n ths study are not forecasts; they are the results of whatf studes. The models, albet developed wth emprcal data fttng and flexblty to change the parameters, are not forecast models themselves. The models, however, are constructed so they can easly hook to the SATS Economc Demand Model (SATS-EDM), whch wll generate SATS demand forecast from the arcraft performance data and the socoeconomc data about areas surroundng the arports. Wth SATS-EDM n our model sute, we wll have a complete SATS demand forecast model. In the last chapter we nclude our prelmnary thoughts on how to construct SATS-EDM. v

7 Contents Chapter 1 Modelng Framework Chapter 2 SATS Arport Demand Model SELECTED AIRPORTS IN THE STUDY GA AIRPORT DEMOGRAPHIC AND ECONOMIC DATABASE AN ECONOMIC DEMAND MODEL OF AIRPORT GA OPERATION DEFAULT AIRPORT GA OPERATION MODEL GA TRAFFIC MEASURES AND THEIR USE IN SATS DEMAND FORECAST Chapter 3 SATS Flght Demand Model THE GRAVITY MODEL OF ORIGIN AND DESTINATION DEMAND GA FLIGHT PROFILE Dstance Dstrbutons Tme-of-day Profle MONTE CARLO SIMULATION OF GA FLIGHT DEMAND Chapter 4 LMINET-SATS AIRSPACE FOR PISTON-DRIVEN SATS AIRPLANES Arlnes Specal Use Arspace Mountans SATS AND ATM STAFFING DEFINITION AND OPERATION OF LMINET-SATS Consderatons for Addng a SATS Underlayer to LMINET SATS arports LMINET-SATS Operatons RESULTS An ntal exercse Effects of Improved Strateges Conclusons v

8 Chapter 5 Summary and Future Work References Appendx A Parameter Estmaton of the Gravty Model FIGURES Fgure 1-1. Top-down, Modular SATS Demand Modelng Fgure 2-1. Lorenz Curve of GA Operatons Fgure 2-3. Arport Operatons Model Schematc Fgure 3-1. Dstance Dstrbuton of Sngle Engne Arcraft Fgure 3-2. Dstance Dstrbuton of Mult-Engne Arcraft Fgure 3-3. Dstance Dstrbuton of Jet-Engne Arcraft Fgure 3-4. Monthly GA Flghts Fgure 4-1. Fuel Burn and Assocated Arspeed For Canadar CL600 Regonal Jet Fgure 4-2. Fuel Burn and Assocated Arspeed Fgure 4-3. Varaton of Fuel Burn wth TAS at FL 180, Embraer E Fgure 4-4. Dstrbuton of Dstances of 889 IFR Flghts of Lght Arcraft Fgure 4-5. Thrty nm Zones around LMINET Arports Fgure 4-6. Low-Alttude Sectors for Albuquerque Ar Route Traffc Control Center Fgure 4-7. Geographc Enroute Sectors Fgure 4-8. Enroute Sector Structure For Lght-Arcraft SATS Operatons Fgure 4-9. FAA Hgh-Alttude Sectors Fgure Example Traectory Fgure SATS Arports Fgure Hourly SATS Departures Requred to Meet 1 Percent of RPM Demand Fgure Pston ILS Arrvals at Van Nuys, Calforna Fgure Pston ILS Arrvals at Morrstown, New Jersey Fgure Hourly Demand For Busy Sectors, Fgure SATS TRACON Demands, Fgure Hourly Demand For Busy Sectors, Fgure SATS TRACON Demands, Fgure Hourly Pston-SATS Arrvals to SATS Arports Near EWR v

9 Contents Fgure SATS Operatons to Supply RPM Defct Fgure Pston-SATS Arrvals to VNY, Defct Case Fgure Arrvals to Sector Fgure Arrvals To SATS Arports Near EWR, Defct Case TABLES Table 2-1. Arport Actvty Dstrbuton Table 2-2. Top 10 GA Operatons Arports Table 2-3. Arport Demographc and Economc Database Table 2-4. Total Itnerant Landngs by Category and Regon Table 2-5. Number of Arcraft by Category and Regon Table 2-6. Arcraft Utlzaton Rate by Category and Regon Table 2-8. Actual Use GA and AT Hours Flown n Table 2-9. Total TPM by Engne Type and Traffc Dstrbuton n Table Percentage of Total Operatons Flown by Arcraft Type n Table Constraned and Unconstraned RPM (Bllon) Forecasts Table Addtonal SATS Operatons for 1 Percent of Commercal RPM Table Addtonal SATS Operatons to Fll the Gap of Unsatsfed Commercal Traffc Table 3-1. Percentles of Average Daly GA Operatons at TAF Arports Table 3-2. Samples of ETMS Data n the Dstance Dstrbuton Estmaton Table 3-3. Dstance (nm) Statstcs for Combned Data Sets Table 3-4. Model Parameters for Webull Dstrbuton Table 3-5. Estmated Dstance (nm) Statstcs Usng a Webull Dstrbuton Table 3-6. Tme-of-day Probablty Dstrbuton Functon Used n the Smulaton v

10 Chapter 1 Modelng Framework The transportaton system s a vtal part of a dynamc economy. For centures, ctes and economes have developed at seaports and along rverbanks, at the ralroad and nterstate hghway ntersectons, and more recently near arports. Snce the frst flght made by the Wrght brothers about a century ago, the ar transportaton ndustry has matured. It has developed from a means of delverng mal to a means of travel for the rch to today s necessary means of travel for conductng busness and pursung lesure. In the Unted States alone n 2000, there were more than 670 mllon enplanements and more than 670 bllon RPMs. Commercal ar transport servce has become so mportant to our busness actvty and our lves that any dsturbance n ts servce by nclement weather or nadequate ar traffc capacty s met by publc outcry. Because of the mbalance of ncreasng ar traffc demand and the relatvely constant ar traffc capacty, ar traffc congeston wll become worse. Studes have concluded that our Natonal Arspace System (NAS) wll reach grdlock n about a decade, precludng relable commercal ar transport unless demand s curtaled. The more realstc predcton s that arlnes wll rase tcket prces to curb demand from margnal travelers and curtal operatons to reman wthn the ar traffc capacty. These measures wll be at the expense of natonal busness actvty and the general consumer welfare. Because travel demand s postvely related to populaton and per capta ncome, our ar travel demand wll ncrease wth growng populaton and economy [1, 2]. Our propensty of ar travel, n the meantme, wll also be ncreasng. The Baby Boomer generaton wll retre wth the money and tme to travel. The value of human tme s ncreasngly valuable n the fast-paced nformaton age. The economy demands that goods are manufactured and delvered ust-n-tme. Small Arcraft Transportaton System (SATS) provdes alternatves n ar travel more frequent flghts of small arcraft take off and land at small arports. The recent advances n engnes, avoncs, arframe, navgaton, communcaton, and plot tranng have made a new generaton of general avaton (GA) possble. SATS wll take advantage of the vast pool of small arports to break the grdlock of ar traffc congeston. Ar traffc congeston s mostly arport nduced, ether from nsuffcent runways, taxways, gates, or nsuffcent arspace capacty around arports [3]. The Unted States has more than 5,000 arports and about 12,000 landng strps, but only about 200 arports have et operatons. SATS wll help break the grdlock at large commercal arports by dvertng traffc to small arports. It also can pck up latent 1-1

11 ar traffc demand because t can speed up door-to-door trp tme. Wth hgh speed, numerous avalable arports, affordable costs, and easy plot tranng, SATS can provde better door-to-door travel tme, enhance moblty, and stmulate busness actvty. Ths report explans SATS demand modelng. SATS s stll n ts formatve stage, but many current generaton GA arcraft such as the Crrus-20 already have SATS techncal capabltes. Compared wth current GA, SATS may change the speed of travel tme and the reaches of destnatons; alter avoncs requrements for arport and arspace operatons; and reduce operatng costs and certfcaton requrements. Other factors wll nfluence SATS operatons. Pressurzed cabns wll enable SATS to fly at hgher alttudes, requrng dfferent ar traffc control (ATC) servce and dfferent ar traffc management (ATM) schemes. SATS can be treated as GA wth dfferent attrbutes, whch our model wll address. Our model wll be flexble to account for retrofttng of current GA arcraft, modfed attrbutes, or dfferent percentages of arcraft that have SATS attrbutes. Wth ths flexblty, our SATS model s a GA model, and we wll use SATS and GA nterchangeably n our modelng. GA arcraft are classfed 1 as sngle-engne IFR, sngle-engne VFR, mult-engne VFR, mult-engne pston IFR, mult-engne turbo IFR, and et. Engne types determne the speed and reaches of an arcraft; avoncs equpment (VFR/IFR) determne arport and ATC sector demands under dfferent weather condtons. Jets always are assumed to be IFR because flght alttude requres IFR flghts. 1 Helcopter operatons are not ncluded n ths report s demand modelng because (1) arports are not requred for take off and landng and (2) most helcopter operatons are VFR, and thus result n few recorded flght tracks. We can capture only a few IFR helcopter flghts a day n our ETMS data source. 1-2

12 Modelng Framework Fgure 1-1. Top-down, Modular SATS Demand Modelng SATS-ADM SATS-FDM LMINET-SATS Input: socoeconomc data scenaro defnton Output: operatons at each arport by a/c type Component model: SATS a/c utlzaton model Input: operatons at each arport tme-of-day dstrbuton dstance dstrbuton Output: SATS schedule Component model: gravty dstrbuton model Monte Carlo smulaton Input: ar carrer and SATS schedule SATS a/c & arport performance Output: demand & delays at sectors & TRCONs delays at 2,865 arports Component model: LMINET SATS a/c traectory model SATS arport capacty/delay model Fgure 1-1 shows our SATS demand model bult n a top-down, modular fashon followng the general system engneerng prncple. There are three maor component models wth the followng functonaltes: SATS Arport Demand Model (SATS-ADM) Input demographc and economc data, arport and arcraft performance data; Output annual number of operatons by arport and arcraft type; SATS Flght Demand Model (SATS-FDM) Input output of SATS-ADM; Output GA flght schedule for entre GA arport network; LMINET-SATS Input output of SATS-FDM Output ATC-sector demand for the entre NAS. From SATS-ADM to SATS-FDM to LMINET-SATS, the GA demand s more and more detaled. The models are lnked through nput and output, whch are common traffc measures used. Each model exsts n ts own rght, and users can substtute them by usng best-avalable nformaton. 1-3

13 Chapters 2, 3, and 4 explan our three prncpal models, SATS-ADM, SATS- FDM, and LMINET-SATS. Each model has ts own subcomponent models. Chapter 5 summarzes model capabltes and dentfes future work. The appendxes lst techncal detals of our model. 1-4

14 Chapter 2 SATS Arport Demand Model SELECTED AIRPORTS IN THE STUDY There are more than 5,000 arports and approxmately 12,000 landng strps n the Unted States. The most comprehensve databases about these arports are the FAA s Termnal Area Forecast (TAF) and the Natonal Plan of Integrated Arport Systems (NPIAS). For each of the approxmately 3,000 arports n the database, TAF mantans nformaton about enplanements, operatons, and based arcraft. Of the remanng 2,000 arports not ncluded n the databases, about 1,000 are prvately owned but avalable for publc use. Another 1,000 arports are publcly owned but lack suffcent facltes, or do not have suffcent based arcraft, or are wthn 20 mles of a TAF arport [9]. In the NPIAS, arports (see Table 2-1) are classfed as follows: Large hub enplanement s more than 1 percent of the total U.S. enplanement; Medum hub enplanement s more than 0.25 percent but less than 1 percent; Small hub enplanement s more than 0.05 percent but less than 0.25 percent; Nonhub prmary enplanement s more than 10,000 but less than 0.05 percent of the U.S. total; Other commercal enplanement s more than 2,500 but less than 10,000 annually; and Relever GA arports located close to maor metropoltan areas. 2-1

15 Table 2-1. Arport Actvty Dstrbuton Arport type Number of arports Percentage of all enplanements Percentage of actve GA arcraft Arport type Number of arports Percentage of all enplanements Percentage of actve GA arcraft Large hub Medum hub Small hub Nonhub prmary Other commercal Relever Other GA 2, TAF Total 3, Source: NPIAS. TAF arports cover 98 percent of the domestc U.S. populaton wthn 20 mles of arport rad. The arports are dstrbuted roughly one per county n rural areas and often are located near the county seat. Of all TAF arports, 95 percent are consdered to have good or far runway pavement. We selected the entre database of TAF arports for our study because they cover almost the entre domestc U.S. populaton and serve current GA actvty. Includng addtonal arports does not offer suffcent benefts to ustfy the substantal effort requred to collect addtonal arport data. Wth a selecton of a smaller set of total TAF arports, t s stll possble for us to construct a sound model; however, t s not worth the addtonal effort requred to study out-of-network SATS traffc. If we select all TAF arports, we can gnore all out-of-network traffc as practcally nsgnfcant. We have a network of 2,865 arports after excludng arports n Alaska, Hawa, Puerto Rco, and Guam. If we rank those arports accordng to ther tnerant GA operatons and plot them wth the arport dstrbuton, a Lorenz curve results, as shown n Fgure 2-1 [5]. 2-2

16 SATS Arport Demand Model Fgure 2-1. Lorenz Curve of GA Operatons From ths curve, we see that GA operatons are concentrated. In fact, 100 arports account for 25 percent of GA tnerant operaton, 300 arports account for about 50 percent, and 1,000 arports account for about 83 percent. Table 2-2 lsts 10 arports wth the most tnerant GA operatons n Table 2-2. Top 10 GA Operatons Arports Arport Itnerant GA Operatons n 1997 VNY 373,781 SNA 156,216 DAB 232,059 LGB 221,046 APA 197,230 RVS 185,121 FTW 183,301 BFI 182,124 OAK 175,294 FXE 174,

17 If we use 600 nm radus as a crteron that an arport can reach by GA, then an arport n our SATS arport network can reach between 257 to 1,708 arports. The medan number of arports that an arport can reach s 1,038. The range of the frst quartle and the thrd quartle s from 643 to 1,347. Any arport n the SATS network has a good number of other arports wthn ts reach. There s a slght negatve correlaton between the arport reach and the tnerant operaton. The Spearman correlaton s [5]. We thnk ths negatve correlaton s caused by some arports n Calforna, whch, because of ther geographc locaton, have lmted number of arports among ther reach but have large number of GA operatons. Thus, we can say that the GA operaton at one arport generally does not depend on the number of arports among ts reach, but more on the characterstc of ts reach. GA AIRPORT DEMOGRAPHIC AND ECONOMIC DATABASE The GA Arport Demographc and Economc Database s used to extrapolate economc and demographc nformaton from the current baselne year (1998) to the two future target years (2007 and 2022). A set of reference parameters can be calculated from the 1998 data. Proectons are scaled from these reference parameters. For example, one set of parameters s scaled on total household ncome, then proected to total household ncome n the future target years. The database starts wth census track data mapped for 3,320 offcal FAA montored U.S. arports. The FAA assgns each arport a unque three-letter arport dentfer. The database also ncludes geographc nformaton for regon and state, and hub or relever status. The mportant nformaton regardng arport avaton status ncludes enplanements, based arcraft, and operatons. The term, enplanement refers to one passenger boardng an arcraft; dstance flown and purpose are not relevant. There are fve categores of enplanements: ar carrer, ar tax, commuter, U.S., and total. Because enplanement apples to commercal transportaton only, numerous arports report zero enplanements across all categores. Based arcraft refers to the number of arcraft, by type, located at an arport. The arcraft types are sngle engne, mult-engne, et engne, helcopter, other, and total based arcraft. Because commercal arcraft are never based n any arport, the reported based arcraft are GA or ar tax only. Operaton s defned as ether an arcraft takeoff or landng. Operatons can be classfed by purpose. Operatons can be tnerate (place to place) or local. The type of arcraft s mltary, general avaton, ar tax, or ar carrer. Categores for 2-4

18 SATS Arport Demand Model operatons are ar carrer tnerant, ar tax tnerate, general avaton tnerate, mltary tnerant, general avaton local, mltary local, and total operatons. (There are no local ar carrer or local ar tax operatons). Table 2-3. Arport Demographc and Economc Database Descrpton Locaton d Regon Arport Name Cty State Year Hub Sze Relever Ar Carrer Approaches Ar Tax Approaches General Avaton Approaches Mltary Approaches Total Approaches Sngle Engne Based Arcraft Jet Engne Based Arcraft Mult Engne Based Arcraft Helcopter Based Arcraft Other Based Arcraft Total Based Arcraft Ar Carrer Enplanements Ar Tax Enplanements Commuter Enplanements US Flag Enplanements Foregn Flag Enplanements Total Enplanements Prmary Ar Carrer Overs Prmary Ar Tax Overs Prmary General Avaton Overs Prmary Mltary Overs Secondary Ar Carrer Overs Secondary Ar Tax Overs Secondary General Avaton Overs Secondary Mltary Overs Instrument Operatons Total Overs Total Instrument Operatons Ar Carrer Itnerant Operatons Ar Tax Itnerant Operatons General Avaton Itnerant Operatons Varable Name Locd Regon Arport Cty State Year Hub Sze Relver ArC App ArT App GA App Ml App Tot App SEB Ar JEB Ar MEB Ar HelB Ar OthB Ar TotB Ar AC Enpla AT Enpla Co Enpla US Enpla Fo Enpla To Enpla Pr AC Ov Pr AT Ov Pr GA Ov Pr M Ov Se AC Ov Se AT Ov Se GA Ov Se M Ov InOps TO TotI Ops ACIt Ops ATIt Ops GAIt Ops 2-5

19 Table 2-3. Arport Demograhc and Economc Database (Contnued) Descrpton Mltary Itnerant Operatons General Avaton Local Operatons Mltary Local Operatons Total Operatons Populaton In All Rngs Populaton Rng 1 (Red) Populaton Rng 2 (Green) Populaton Rng 3 (Blue) Households In All Rngs Households Rng 1 (Red) Households Rng 2 (Green) Households Rng 3 (Blue) Total Household Income In All Rngs Total Household Income Rng 1 (Red) Total Household Income Rng 2 (Green) Total Household Income Rng 3 (Blue) Average Household Income In All Rngs Average Household Income Rng 1 (Red) Average Household Income Rng 2 (Green) Average Household Income Rng 3 (Blue) Varable Name MIt Ops GAL Ops ML Ops Tot Ops Pop All PopRng1 PopRng2 PopRng3 House All HouseRg1 HouseRg2 HouseRg3 TtHsIn TtHsIn1 TtHsIn2 TtHsIn3 AvHsInTT AvHsInR1 AvHsInR2 AvHsInR3 It s beleved n NPIAS that 20 mles radus s a good measure surroundng an arport because t corresponds about 30 mnutes drvng to the arport. The regon surroundng an arport s dvded nto three rngs n the database: the frst rng s wthn 10 mles of the arport; the second rng s from 10 to 20 mles from the arport; and the thrd rng s 20 to 50 mles from the arport. Some overlap exsts between arports (.e., a household may le nsde the rngs of multple arports). For each defned rng, the database contans populaton, number n a household, and total and average household ncome. The followng sectons n ths chapter dscuss database use and ts mportance n our arport GA operaton demand modelng. AN ECONOMIC DEMAND MODEL OF AIRPORT GA OPERATION In ths secton, we buld a model of GA operatons based on current observatons of socoeconomc varables. Current GA travel s not the SATS travel envsoned for the future, although the current GA arcraft, especally corporate ets and those operated n the tmeshare program, certanly are SATS-capable. Our constructon 2-6

20 SATS Arport Demand Model of a current GA demand model can offer valuable nsght nto the SATS demand model. Because the SATS forecast eventually must be at the arport level, the best data source s our Arport Demographc and Economc Database. Accordng to NPIAS, ths data set covers about 98 percent of the total U.S. populaton, whch lves wthn 20-mle rad of the arports, or about 30 mnutes of drvng tme to an arport. The lnear model s the smplest because the GA traffc s proportonal to the surroundng populaton, although the proportons may be dfferent for the three surroundng rngs. For each arport, the followng GA demand model s always true: 3 GA = α PopRng, [Eq. 2-1] = 1 where GA s the measure of GA traffc to be specfed, and α 1, α 2, α 3 are the parameters that measure the estmated propensty of the populaton to travel GA. If we thnk the propensty for GA travel s lnearly related to the average household ncome,.e., α = δ + β AvHsInR, =1, 2, 3, [Eq. 2-2] then the GA traffc can be rewrtten as 3 GA = ( δ PopRng + β PopRng AvHsInR ). [Eq. 2-3] = 1 Ths s a lnear model f we treat the products of PopRng and AvHsInR, =1, 2, 3, as separate varables. We thnk ths model probably works only for the GA arports when the large, medum, and small hub arports are excluded, because hub arports are developed for commercal servce. To further account for the some of the unqueness GA arports have to attract travel demand, the GA operatons must be related to commercal and ar tax traffc. Our GA demand model can be wrtten as 3 GA = ( δ PopRng = 1 + β PopRng AvHsInR ) + Ac _ Enpla + AtIt _ Ops. [Eq. 2-4] If we use the tnerant GA operatons as the measure of GA traffc n the above lnear regresson equaton, then we can estmate the parameters. Based on 1998 data, R 2 s about 30 percent 1. Ths s not a terrble model, but t s not a terrfc model ether. 1 R 2 s a goodness-of-ft of the model to the data, whch measures as a percentage the varablty of the observed data that can be explaned by the model. 2-7

21 One cause of the lack of goodness-of-ft s the locaton of the arports. When we run the same model by regon, where each of the nne regons has ts own separate parameter estmates, then the R 2 s range from 30 percent through 60 percent. Ths s an mprovement of the model, although the basc dea that GA traffc s determned by populaton and average ncome s unchanged. When we run the model by state, the R 2 s range from 30 percent through 80 percent, wth model mprovement at the hgh end of R 2 but not at the low end. Ths means the arports are not homogenous enough at regonal or state levels for GA traffc at an arport to be explaned by the surroundng populaton and ts average ncome and locaton. The general dea of ths model s good, and model goodness of ft mproves when we further classfy arports geographcally. Wth addtonal nformaton about an arport s equpage, access, weather statstcs, and surroundngs ncludng ncome dstrbuton and busness actvtes, the model can be mproved. Addtonal nformaton about an arport and ts surroundngs does not change our basc assumpton that GA traffc s related to the surroundng populaton whose propensty for GA actvty s related drectly to ts ncome. Addtonal nformaton wll mprove the GA demand model. Wth more nformaton about the arport, we wll need to construct an arport-specfc GA demand model (.e., each arport demand model wll have ts unque parameters. Can we construct arport-specfc GA demand models based on our GA Arport Demographc and Economc Database? The answer seems to be no. Frst, we have only a few hstorcal data ponts for 1980, 1990, and 1998 for each arport, whch makes the estmaton mpossble. Even wth more data ponts for each arport, we cannot use all the data because of the structural change n the GA ndustry snce enactment of the GA Revtalzaton Act n Fgure 2-2 shows hstorcal and forecast traffc n the 3,320 arports n FAA s 1999 TAF. In the fgure, ITN_AC, ITN_AT, ITN_GA, and LOC_GA, are the operatons of ar carrer, ar tax, tnerant GA, and local GA. The data from 1976 through 1998 are based on observed counts; data beyond 1998 are forecasts. Reversal of the GA traffc declne snce enactment of the General Avaton Revtalzaton Act n 1994 clearly shows the structural change on GA traffc, whch reveals the dependence of GA traffc demand on the government polcy. From ths we can further nfer that the SATS demand wll also depend on the government polcy, or more broadly, on the general SATS operaton envronment. The dffculty to predct the government polces makes the SATS demand forecast even more dffcult. 2-8

22 SATS Arport Demand Model Fgure 2-2. Total Observed and Forecast Operatons n TAF Arports ITN_AC ITN_AT ITN_GA LOC_GA Source: Termnal Area Forecast, FAA In concluson, t s good experence to try to construct a current GA demand model. What we have learned though the GA modelng can be appled to the SATS modelng: SATS demand should be drectly related to the surroundng populaton. Propensty for SATS travel s drectly related to the household ncome of the populaton surroundng an arport. The SATS demand model should be arport specfc (.e., wth unque sets of parameters n the same model structure). DEFAULT AIRPORT GA OPERATION MODEL Dfferent arcraft have dfferent performance capabltes for speed, range, and alttude, and dfferent avoncs equpage determnes what ATC servce they requre and where they can go n dfferent condtons. Ths detaled nformaton s requred by LMINET-SATS to buld an accurate pcture of ATC demand from SATS. For the SATS-ADM, we must generate arport tnerant operaton by sngle, mult, or et engne type. For each arport, TAF forecasts the number of based arcraft n the categores of sngle engne, mult-engne, et, helcopter, and other; and tnerant operatons conducted by ar carrer, ar tax, and GA; and local GA operatons. Our man task n ths secton s to have a model by arcraft type for tnerant traffc. Because all GA arcraft need smlar arport servces, we do not further decompose local GA traffc to dfferent categores. 2-9

23 By the nature of ts servce and lack of further nformaton, we assume the arcraft categorcal composton of ar tax s the same as GA. For our 2,865 SATS arport network, the averages of tnerant GA and ar tax operatons per arport n 1998 are 14,644 and 4,308, respectvely. In other words, the ar tax category accounts for about 22 percent of the combned tnerant operatons. The slght dfference between GA and ar tax categorcal composton wll not make a bg dfference n our model because GA contrbutes most traffc. An arport s tnerant operaton comes from two sources: one from locally based arcraft, the other from arcraft from the outsde. Operatons by locally based arcraft are drectly proportonal to the locally based fleet multpled by ts utlzaton rate. The utlzaton rate s defned as the number of tnerant operatons conducted by one arcraft per year. Lackng other nformaton, we assume the operatons conducted by outsde-based arcraft based outsde are proportonal to the traffc mx surroundng the arport. General Avaton and Ar Tax Assocaton (GAATA) database contans the most detaled nformaton about GA and ar tax tnerant operatons at the regonal level. Because we know the locaton and based arcraft by category for each arport n TAF, we can drectly compute arcraft utlzaton by dvdng total regonal tnerant operatons by total arcraft based n the regon. We assume all tnerant operatons are conducted ntra-regonally. Table 2-4 shows the number of tnerant landngs n 1997 by regon based on GAATA. Table 2-4. Total Itnerant Landngs by Category and Regon Regon Sngle Mult Jet Other ACE 520, ,625 52,546 15,729 AEA 746, , ,411 48,897 AGL 1,177, , ,910 77,090 ANE 363,309 81,865 20,065 33,177 ANM 646, ,036 53,656 51,690 ASO 1,066, , ,729 63,216 ASW 913, , ,623 54,671 AWP 1,123, ,260 70,953 67,

24 SATS Arport Demand Model Table 2-5. Number of Arcraft by Category and Regon Regon Sngle Mult Jet Other ACE 7,586 1, AEA 14,512 2, ,310 AGL 22,652 4,150 1,158 1,174 ANE 5, ANM 15,126 1, ASO 21,503 5,749 1,394 1,231 ASW 15,853 3,426 1, AWP 29,453 4, Table 2-6. Arcraft Utlzaton Rate by Category and Regon Regon Sngle Mult Jet Other ACE AEA AGL ANE ANM ASO ASW AWP We assume the arcraft utlzaton rates are stable n the default forecast. In usng the model, the analyst has the flexblty to use any utlzaton rates at any arport. After the analyst selects the utlzaton rates, the number of annual tnerant landngs s attaned by multplyng the fleet by ts utlzaton rate. The regonal landng dstrbuton by arcraft and regon changes based on the dstrbuton of arcraft n the regon. Table 2-7 shows the 1998 landng rates by arcraft and regon. Table 2-7. Dstrbuton (%) of Landngs By Category and Regon n 1998 Regon Sngle Mult Jet Other ACE AEA AGL ANE ANM ASO ASW AWP

25 At each arport, the dfference of the total tnerant operatons gven by the default FAA TAF and the operatons conducted by the locally based arcraft computed by multplyng the fleet by ts utlzaton rate, s the total operatons by vstng arcraft, whch follow the regonal dstrbuton of the arport. Fgure 2-3 s a schematc of the model. Fgure 2-3. Arport Operatons Model Schematc Local Arport Operaton Itnerant By Local A/C By Outsde A/C Local Fleet A/C Utlzaton Itnerant Operatons by Local A/C Sngle Mult Jet Total Operatons by Outsde A/C Operaton Mxture n the Regon Itnerant Operatons by Outsde A/C After all tnerant operatons at an arport are fully decomposed, the other category s dstrbuted proportonally to the categores of sngle-engne, mult-engne, and et. GA TRAFFIC MEASURES AND THEIR USE IN SATS DEMAND FORECAST Whle the default GA operatons model descrbed n the prevous secton wll enable us to fnd the operatons of each arcraft type for each arport based on the default forecast n the FAA s TAF, the model to be developed n ths secton wll let us predct future SATS, or addtonal GA demand. We follow the system engneerng modelng prncple to have a seres of component models to generate measurable, easy-to-change GA traffc measures. Prmary data sources are the TAF, GAATA, and Arport Demographc and Economc Databases outlned n ths chapter. Two case studes are ncluded to llustrate the methodology and the traffc measures n the what-f scenaros that show SATS pckng up some porton of unsatsfed commercal traffc demand. Many statstcs such as operaton and enplanement are avalable to measure commercal ar traffc, but the most mportant one s RPM. RPMencompasses operaton and enplanement, and, more mportantly, t s closer to measure the purpose of ar traffc movng people from one place to another. GA traffc has been 2-12

26 SATS Arport Demand Model measured only by ts operatons, whle enplanement and RPM have never been measured and reported. There are reasons not to report them. Frst, unlke commercal traffc, there s no mechansm to report GA flght passengers orgn and destnaton. Second, GA flght lacks the concept of revenue passenger; at tmes, a GA flght has no passenger, only a plot. SATS wll share the same knd of operaton as the current GA; t also wll share the same dffculty n reportng ts traffc statstcs. SATS, nonetheless, s envsoned to fulfll the transportaton need of movng people from one place to another. It s unavodable then to defne a smlar measure lke RPM n commercal traffc for SATS or GA. In ths case, a transported passenger may not pay for the flght or may be the plot, as long as he or she takes SATS to fulfll the transportaton need of movng from one place to another. Here we wll use the term Transported Passenger Mle (TPM) n leu of RPM. Followng the same defnton for the commercal traffc, TPM for a SATS or GA flght s defned as follows: TPM = transported_passenger flght_ dstance, [Eq. 2-5] whch can be further broken down to TPM = acraft_sze load_factor dstance. [Eq. 2-6] Equaton 2-5 provdes a practcal means of computng TPM. For example, to fnd the TPM for a category, by arcraft type or by regon, one needs to the sum the TPMs of all the flghts fallng n the category. Table 2-7 lsts TPMs and the converson factors of 1 bllon TPM to operaton by dfferent engne types. Table 2-7. Average GA TPM and Equvalent Number of Operatons of 1 Bllon TRM by Engne Type Engne type Seat Load factor Dstance (mle) TRM Operatons per 1 bllon TRM Sngle ,064,281 Mult , ,136 Jet , ,488 In Table 2-7, the average dstance s based on our data analyss of Instrument Flght Rule (IFR) flghts recorded n ETMS, whle the seats and load factors are the ones we assume reasonable based on arcraft type. The table gves a clear vew of TPMs and how they relate to seat, load factor, and dstance, whch the user can modfy. The last column n the table s the equvalent number of operatons, calculated usng Equaton 2-6. Ths s mportant because t translates the SATS traffc measured n TPM to operaton, whch wll be useful for arport plannng and our constructon of SATS FDM. It s more effcent to carry SATS traffc n et arcraft per operaton, as seen n the converson factor column more passengers can be carred a longer dstance per 2-13

27 flght. The ultmate decson whch SATS flght a transported passenger wll take wll depend on many factors: operatng cost, arcraft equpage, trp msson, arport facltes, etc. Because SATS arcraft can be dfferent n engne types, a SATS demand dstrbuton model s needed to decompose overall demand nto dfferent categores. Such a model s constructed usng GAATA data. Table 2-8. Actual Use GA and AT Hours Flown n 1998 GA & AT hours Corporate Busness Personal Other work Ar tax Total Sngle 212,900 2,192,013 8,044, , ,276 10,966,216 Mult 1,015, , ,009 18, ,827 3,211,823 Jet 1,435,231 24,294 18, ,592 1,598,981 Data Source: GAATA Table 2-9 shows dstance as arspeed multpled by flght hour for total TPM for each engne type and the dstrbuton. Table 2-9. Total TPM by Engne Type and Traffc Dstrbuton n 1998 Arcraft type Total hours Speed Seat Load factor Total TPM TPM (%) Sngle 10,966, ,027,151, Mult 3,211, ,039,188, Jet 1,598, ,689,168, In the above Table 2-9, we took the total flght hours from the GAATA database. We based the speeds on our estmaton of ETMS data. The seats and load factors are our assumptons, and the total TPMs and ther dstrbuton ust follow the formula. Under our seat and load factor assumptons, the sngle-engne arcraft carred about half of GA traffc n 1998, whle the mult-engne and et engne arcraft each carred about a quarter of GA traffc. Model users can modfy the nput parameters or change the traffc allocaton. Table 2-10 shows that n operatons, the sngle-engne arcraft s far more domnant. Table Percentage of Total Operatons Flown by Arcraft Type n 1998 Arcraft type % Sngle engne 73.1 Mult-engne 10.9 Jet 2.7 Rotorcraft 3.5 Other 9.8 Data source: GAATA. 2-14

28 SATS Arport Demand Model The remander of ths secton presents a case study of a proect to determne addtonal SATS demand f SATS s used to pck up unsatsfed commercal traffc. Table 2-11 shows the constraned and unconstraned RPM forecasts where the unconstraned RPM forecasts are from TAF whle the constraned forecasts are based on the ntegrated ASAC ACIM and the operatons models.[2] Table Constraned and Unconstraned RPM (Bllon) Forecasts Year 2007 Year 2022 Unconstraned ,841 Constraned ,650 Delta/ RPM gap Percentage gap 2.6% 10.4% The gap n the two forecasts s 2.6 percent n 2007; t grows to 10.4 percent n In real terms, the gaps are about 24 bllon RPM n 2007 and 190 bllon RPM n Two scenaros present nterestng cases. The frst scenaro assumes a small percentage of dverson (we assume 1 percent) from commercal RPM to SATS TPM n the future. In the second scenaro, SATS carres all unsatsfed commercal traffc. The two scenaros gve the lower and upper bounds of SATS traffc. To calculate addtonal SATS demand dverted from unsatsfed commercal traffc, we frst decompose the traffc by engne type accordng to Table 2-9. Then we multply the TPMs for each engne type by the converson factors n Table 2-7. Ths yelds addtonal SATS operatons by engne type. Table Addtonal SATS Operatons for 1 Percent of Commercal RPM Year TPM (bllon) Sngle Mult Jet ,727,419 1,672, , ,424,426 3,334, ,262 Table Addtonal SATS Operatons to Fll the Gap of Unsatsfed Commercal Traffc Year TPM (bllon) Sngle Mult Jet ,113,789 4,286,171 1,173, ,628,185 37,373,976 10,235,762 In the second scenaro, addtonal SATS represent a 22 percent ncrease n GA operatons beyond the baselne operatons n 2007, and a 140 percent ncrease n operatons n

29 Addtonal future GA operatons conducted by SATS depend on the scenaro and nput parameters. Our case studes show that these tables provde useful ways to mpute future SATS operatons. The tables are constructed for easy understandng, and drectly relate to common statstcs on ar travel and arcraft performance. The tables are populated wth current GA traffc nformaton, yet they can be modfed to reflect future SATS operatons. 2-16

30 Chapter 3 SATS Flght Demand Model In ths chapter, we explan how we constructed an orgn-and-destnaton (O&D) flght demand schedule for GA, whch wll feed nto the LMINET-SATS to compute arport and arspace demand. GA schedule does not mean that the GA operatons wll be scheduled; rather, t s an expresson of the GA flght n terms of O&D and tme. The analyss presented n ths secton s based on the gven arport operaton fgures such as the one developed n Chapter 2. Whle t s temptng to construct the future GA demand schedule based on current GA flghts, we lack suffcent nformaton about current GA operatons and schedules for all flghts. The lowest average total daly operatons at all TAF arports s zero; the hghest average number of daly operatons s 1,024. For an arport wth about 1,000 operatons a day, some hours can have about 100 operatons per hour, whch s close to the fgure for a medum-szed commercal arport. However, typcal arports have an average of only 13 operatons a day, or about one an hour. Table 3-1 shows the average total tnerant GA operatons for all TAF arports n Table 3-1. Percentles of Average Daly GA Operatons at TAF Arports Cumulatve dstrbuton (%) Percentle GA traffc flow s thn, but GA flghts have numerous potental destnaton arports to land. Wthn a 600-mle radus of an arport, there can be from 300 to more than 1,500 arports, whch means GA flghts have far more destnaton choces than commercal flghts. It s not a good dea to assgn O&D traffc on the bass of a few observed GA schedules. Another techncal challenge we face s that ETMS s the only data source from whch we can extract the GA schedule. ETMS contans only IFR flghts, but most GA flghts are VFR. For 1998 (the most recent statstcs avalable), there are 5.4 mllon IFR flght hours for fx-wnged arcraft compared to 24.1 mllon flght hours of all flght plans n the same category. There are 1.6 mllon IFR flght hours for sngle-engne pston arcraft compared to 18.3 mllon flght hours of all flght plans n the same category (see GAATA Table 4-7). In other words, usng ETMS wll make the GA traffc flow nformaton even thnner to cover greater possbltes of GA or SATS schedulng. 3-1

31 We face two techncal problems: (1) the GA operatons are low at the arports but broad n the O&D par; and (2) the GA schedule we can extract represents a small porton of total GA. The technque that we use to construct the GA schedule s to combne the tme-of-day departure profle and dstance of travel profle wth the gravty model to get the O&D dstrbuton, and then use Monte Carlo smulaton. When constructng a GA schedule, we wll take the stance that departure tme s ndependent of the destnaton choce, whch makes t possble for us to have two separate models for the tme-of-day departure profle, and the O&D dstrbuton model. We assume most GA arcraft can travel ust a few hours before refuelng. Because the GA schedule s based on these tme profles, and there wll be few O&D flghts between arports, demand lkely wll be a fracton of a flght. The Monte Carlo smulaton technque wll overcome ths defcency by generatng nteger numbers of flghts n the GA schedule based on the probabltes specfed by the tme-of-day departure profle and O&D dstrbuton model. Many rounds of Monte Carlo smulaton must be run and fed nto the LMINET-SATS to calculate delays. THE GRAVITY MODEL OF ORIGIN AND DESTINATION DEMAND In the smplest form, the gravty model s t m α m α c β =, = 1,2,, N, [Eq. 3-1] where t s the traffc from cty to cty, m and m are the masses of cty and, respectvely, and c s the cost or the attractveness of travelng from cty to cty. In studes, researchers have used populaton, per capta ncome, and other crteron as masses and pecunary expense or tme of the travel as cost. α, α, and β are the model parameters to be estmated. The gravty model has been used wdely n O&D demand modelng. It s called gravty model because t mmcs the form of Newton s Gravty Law. The above gravty model can be rewrtten as t = a b T T c, = 1,2,..., N, [Eq. 3-3] where T s the total traffc from ; T s the total traffc to ; and c s the couplng parameter from to, whch s normally negatvely related to the cost of travel from to or postvely related to the attractveness from to. Because the traffc must satsfy the conservaton, or t = T, = 1,2,..., N, [Eq. 3-5] 3-2

32 SATS Flght Demand Model Then the normalzng constants a and b must satsfy the followng: a b T c = 1, = 1,2,... N; [Eq. 3-7] b a T c = 1, = 1,2,..., N. [Eq. 3-9] The second form of the gravty model Equaton 3-2 offers an advantage when termnal traffc T and T are known and the task s to estmate the traffc for every O&D par. In ths model, we propose usng the dstance probablty dstrbuton functon for the couplng parameter c,.e., c = f ( d ),, = 1,2,... N. [Eq. 3-11] On the aggregate level, the dstance probablty dstrbuton functon reflects the propensty people have for travellng on a partcular type of arcraft equpment. In ETMS, the orgn and destnaton of a flght are recorded, from whch we can compute the flght dstance. In runnng the model, we can assume that both IFR and VFR flghts cover the same dstance statstcally. Ths s a reasonable assumpton because the dfference between IFR and VFR s ust the avoncs equpage. Agan, the model s flexble to take any dstance profle for any group of arcraft. After selecton of the arport par and, we wll calculate ther dstance d. Based on the value of d and Equaton 3-6, we can fnd the couplng parameter c. In runnng the model, users can opt for ther own dstance probablty functon for each category. Appendx A contans the parameter estmaton algorthm of c,, {1,2,,N}. GA FLIGHT PROFILE Dstance Dstrbutons We need to construct the probablty dstrbuton functon, based on ETMS, for sngle engne, mult-engne, and et equpment categores. We selected 12 ETMS samples, shown n Table 3-2, to nclude dfferent seasons, days of the week, and tmes of day. 3-3

33 Table 3-2. Samples of ETMS Data n the Dstance Dstrbuton Estmaton Date Tme Day 6/19/ MON 6/10/ SAT 6/10/ SAT 5/23/ TUES 5/23/ TUES 3/29/ WED 10/1/ FRI 9/30/ THURS 9/29/ WED 9/28/ TUES 4/16/ FRI 4/16/ FRI Fgures 3-1 through 3-3 show hstograms of the flght dstance for the combned data. Fgure 3-1. Dstance Dstrbuton of Sngle Engne Arcraft Frequency More Dstance (nm) 3-4

34 SATS Flght Demand Model Fgure 3-2. Dstance Dstrbuton of Mult-Engne Arcraft Frequency More Dstance (nm) Fgure 3-3. Dstance Dstrbuton of Jet-Engne Arcraft Frequency More Dstance (nm) Dstrbuton fttng shows that dstance traveled accordng to arcraft engne type (sngle, mult, and et) s best modeled by Webull dstrbuton, whose probablty densty, and probablty cumulatve functons n general are as follow: and λ 1 λ x δ x = e, x 0; δ, λ > 0, [Eq. 3-13] δ f ( ;δ,λ) δ λ x 3-5

35 F ( x ;δ,λ) = x λ σ 1 e, x 0; δ, λ > 0, [Eq. 3-15] where δ and λ are the Webull scale and shape parameters, respectvely. We fnd no statstcal sgnfcance that the samples are dfferent. The parameters for the combned sample are shown n Table 3-3. Table 3-3. Dstance (nm) Statstcs for Combned Data Sets Sngle engne Mult-engne Jet engne Mean Std. Dev Varance 38,304 58, ,807 Skewness Kurtoss Table 3-4. Model Parameters for Webull Dstrbuton Sngle engne Mult-engne Jet engne Scale Shape Table 3-5. Estmated Dstance (nm) Statstcs Usng a Webull Dstrbuton Sngle engne Mult-engne Jet engne Mean Std. Dev Varance 38,525 55, ,866 Skewness Kurtoss

36 SATS Flght Demand Model Thus, based on the estmated parameters, the pdf s for each arcraft engne type are as follows: f ( x δ, λ ) s 1.15 ; s s = 237 x x 237 e, [Eq. 3-17a] f ( x δ, λ ) m ; m m = x x 289 e, [Eq. 3 9b] f ( x δ, λ ) ; = x , [EQ. 3 9c] where f s, f, and m f are the pdfs for sngle, mult-, and et engne arcraft, respectvely, whch are depcted n Fgure 3-4. Fgure 3-4. Probablty Densty Functons of the Estmated Parameter Sngle Mult Jet Dstance (nm) 3-7

37 The cdfs for each arcraft engne type are as follows: F ( x δ, λ ) s 237 ; = 1- e, [Eq. 3-19a] s s 1.15 x F ( x δ, λ ) m 289 ; = 1 e, [Eq. 3 10b] m m x 1.16 F ( x δ, λ ) 826 ; = 1 e, [Eq. 3 10c] x 1.16 Tme-of-day Profle where F s, F m, and F are the cdf s for sngle-, mult-, and et engne arcraft, respectvely. The shape parameters of the estmated Webull dstrbuton for the three dfferent engne types are so close, ther dfferences are caused manly by the scale parameter. Because the speeds of the dfferent engne types are dfferent, ths may suggest the shape parameter s more unversal, relatng to more fundamental characterstcs of a flght such as the plot s physcal lmt of dstance requred to stop. We need to construct a smlar dstance probablty functon for SATS arcraft for each engne type. We beleve the dstance probablty dstrbuton functon depends on the range of the arcraft, and probably more mportant, on the duraton of a flght, attrbutable to plots physcal and psychologcal lmts. Because the proposed sngle-engne SATS arcraft wll be capable of hgher speed than the current sngle-engne GA arcraft, the dstance probablty dstrbuton functon wll be a stretched verson of the current sngle-engne GA f we want to keep the duraton of flght unchanged. We beleve there are no sgnfcant range or speed dfferences between the current GA and the proposed SATS n ether mult-engne or et categores. The followng fgure shows the total number of departures recorded n ETMS by the local tme for a few days n Aprl

38 SATS Flght Demand Model Fgure 3-4. Total Number of GA Departures by Local Tme n the Unted States FLIGHT APR96:00 06APR96:00 07APR96:00 08APR96:00 09APR96:00 10APR96:00 Data source: ETMS There clearly s an hourly departure pattern across the days. In modelng, we assumed that the VFR flghts share the same tme-of-day departure profle wth the IFR flghts as recorded n ETMS. We must convert the total number of departures to the probabltes of the daly total departure based on Fgure 3-4. We wll use more ETMS data as t becomes avalable. In the smulaton n ths report, all arports share the same tme-of-day departure probablty functon, whch s estmated by usng the GA flght counts recorded by ETMS durng Aprl 5 10, Users can choose ther own profle for any arport n the system. Table 3-6. Tme-of-day Probablty Dstrbuton Functon Used n the Smulaton Tme Tme Tme Tme Probablty Probablty Probablty Probablty

39 MONTE CARLO SIMULATION OF GA FLIGHT DEMAND Multple samples wll be needed to counter the randomness. After the samples are created, they wll feed nto the LMINET-SATS ndvdually. Average delays wll be computed for all the sample runs. One mportant ssue n Monte Carlo smulaton s to decde how many flghts to generate for one day s schedule. Instead of usng the rgd method of generatng a fxed number of flghts for each arport dstrbuted accordng to ts destnaton and tme of day, we wll generate the entre pool of flghts for all arports selected n the network. The advantage of ths approach s that the schedule generated s more random, and arports wth very few operatons may not be covered by the schedule for a random day. The total number of flghts of one arcraft category n the entre network, N Daly s gven as N Daly = N Annual-Ops /365/2*1.5, [Eq. 3-21] where N Annual-Ops s the total annual number of operatons n the network, whch s twce the number of flghts by defnton. We multply the average daly total flghts by a factor of 1.5 to smulate the traffc n hgh season. Fgure 3-4 shows the seasonal GA pattern. Fgure 3-4. Monthly GA Flghts 3-10

40 SATS Flght Demand Model We made the followng assumptons n the smulaton: The selectons of arcraft categores are ndependent, meanng we can conduct the Monte Carlo smulaton ndependently and separately for each arcraft category. The dstrbuton of orgnatng arport, destnaton arport, and tme-of-day are ndependent among each other, meanng we can generate the orgnatng arport, destnaton arport, and the tme k ndependently and separately. The schematc of the smulaton s as follows: Repeat 1, 2, 3, and 4 for all arcraft categores 1. Compute the cumulatve probablty dstrbuton functon of the orgnatng arports O(), based on the forecast annual tnerant operatons. 2. Compute the cumulatve probablty dstrbuton functon, for each orgnatng arport, of the destnaton arport D (), based on the traffc resultng from the gravty model. 3. Compute the total number of flghts N Daly based on Equaton Repeat N Daly tmes for steps a, b, c, and d. a. Generate orgnatng arport accordng to O(). b. Generate destnaton arport accordng to D (). c. Generate tme accordng to tme-of-day dstrbuton functon T(k). d. Put the generated GA flght schedule n the approprate avoncs category accordng to the probabltes. The generaton of a random varable x {1, 2,, N} accordng to any cumulatve probablty dstrbuton functon F s done by followng two steps: 1. Generate random varable U, whch s unformly dstrbuted n [0,1]. Most general-purpose programmng languages such as C have bult-n functons for ths task. 2. x s the smallest number that F(x) U. Accordng to Table 4-7 of GAATA, by hours of flght plan n a 1998 survey, 8.98 percent of sngle-engnes, percent of mult-engnes, and 93.1 percent of et engnes are IFR, respectvely. Further, for IFR mult-engne hours, 50.7 percent s pston, whle s the rest s turbo-prop. For the Monte Carlo smulaton of 3-11

41 the future default GA traffc, we assumed the number of flghts of each category follows the same probablty of flght hours reported above, except for 100 percent IFR probablty for et engnes. Wth no any drect nformaton about the probabltes of flghts themselves, we beleve ths assumpton s a good one f the flght hours for each flght are the same for each engne category regardless of the avoncs equpage. Model users can modfy those probabltes. For addtonal SATS traffc case studes, the GA schedule s the sum of the default schedule and the schedule from the addtonal SATS. The Monte Carlo smulaton of the addtonal SATS s generated by usng the same gravty model parameters as n the default case. For the SATS smulaton n the report, whle we stll keep the same pston mult-engne probablty under IFR, we assume the IFR probabltes for all engne categores s 100 percent. We make ths assumpton because the addtonal SATS wll be used manly for transportaton, whch should be wholly IFR to mantan flght relablty. Agan, model users can select ther probablty parameters. 3-12

42 Chapter 4 LMINET-SATS Ths chapter explans how we developed a companon utlty for LMI s queueng network model of the U.S. Natonal Arspace System (NAS) to model ar traffc generated by SATS operatons. SATS operatons wll use arports that are now unused or underused for ar travel, and SATS lght arplanes wll use arspace that s now lttle used for transport. AIRSPACE FOR PISTON-DRIVEN SATS AIRPLANES Arlnes Developng LMINET-SATS and understandng ts results requre an understandng of the arspace that SATS arcraft can use. We assume that turboet and turboprop SATS arcraft wll operate at alttudes typcal of GA arcraft of the correspondng type. LMINET tracks those ts operatons. LMINET-SATS must deal wth sngle and mult-engne, pston-drven SATS arcraft. We assume that pston-sats arcraft wll not be pressurzed. Unpressurzed SATS arplanes wll use arspace below FL 120 (12,000 feet MSL), the alttude at whch FAA regulatons requre pressurzaton or oxygen equpment. 1 SATS operators may not be the only users of ths arspace. In ths secton, we consder compettors to SATS for arspace below FL 120. Presently, arlnes rarely use arspace below FL 120 except for arrvals and departures. Turboet arcraft avod ths arspace for reasons of fuel economy. Fgure 4-1 shows how rapdly optmum fuel burn per dstance flown ncreases, and the the true arspeed (TAS) that yelds the optmum burn decreases, when a regonal et transport, the Canadar CL600, operates at lower alttudes. 1 Whle personal oxygen supply systems for small arcraft are avalable, ther use s nconvenent, and t ntroduces complex safety ssues. We do not consder ths opton for SATS. 4-1

43 Fgure 4-1. Fuel Burn and Assocated Arspeed For Canadar CL600 Regonal Jet TAS for mnmum fuel per nm TAS Fuel flow Fuel flow, kg/nm Alttude, ft Fgure 4-2 shows that turboprop arcraft do not experence the same degradaton of fuel economy wth decreasng alttude. Fgure 4-2. Fuel Burn and Assocated Arspeed For Embraer E120 Turboprop TAS for mnumum fuel burn per nm, kt TAS Fuel Flow Fuel Flow, kg/nm 50 Embraer Brasla wth PW 118 engnes Alttude, ft Fgure 4-3 shows that fuel burn for a turboprop lke the Embraer E120 does not ncrease rapdly wth TAS. 4-2

44 LMINET-SATS Fgure 4-3. Varaton of Fuel Burn wth TAS at FL 180, Embraer E Fuel 220kt TAS, kg/nm Alttude, ft Specal Use Arspace In vew of these facts, arlnes flyng turboprop arcraft have the opton of operatng at alttudes well below FL 100 wthout payng a great penalty n fuel economy, even when speed s not reduced. There are reasons other than economy for arlnes to avod lower alttudes, such as ncreased turbulence and ncreased possbltes of encounterng adverse weather requrng detours. Nevertheless, we beleve that as the natonal arspace (NAS) becomes saturated, operators of turboprop equpment may decde to use alttudes where pston SATS arplanes fly. If the choce becomes a 30-mnute delay for FL 180 and an mmedate departure on FL 60, many turboprop arlner operators wll choose the lower alttude. Sgnfcant parts of the arspace over the contguous Unted States (CONUS) are reserved for specal uses. Ths specal use arspace (SUA) comprses mltary operatons areas (MOA), restrcted areas (RA), warnng areas (WA), and prohbted areas (PAs). SATS traffc must respect these. Nevertheless, the specal characterstcs of SATS arplanes may consderably reduce the effect of SUA on ther operatons. Sngle-engne and propeller-drven mult-engne SATS arplanes often can operate ether below or above SUA. For example, there are 71 MOAs on IFR Enroute Low-Alttude Charts L-17 and L-18, whch cover the south coast of the Unted States from west of Houston to east of Jacksonvlle. Ths regon has many mltary nstallatons, and thus many MOAs and RAs. Of the 71 MOAs, only four block all reasonable alttudes for propellerdrven SATS traffc. 4-3

45 Many SUAs do not operate all the tme. For example, none of the four alttuderestrctve MOAs mentoned n the precedng paragraph operates contnuously, and two of these never operate before 5 p.m. local tme. Of 52 RAs on Charts L-17 and L-18, only 12 operate contnuously and affect all propeller-sats alttudes. Moreover, many SUAs have relatvely small areas. For example, restrcted area R-3803A, one of the 12 contnuously-operatng SUAs on Charts L-17 and L-18 that obstructs all reasonable propeller-sats alttudes, can be enclosed n a rectangle roughly 10 nm by 5 nm. SATS operatons wll begn several years after 2001, when SUA may be revsed and ts area reduced. Also, the self-separatng technology envsaged for SATS may nteract automatcally wth SUA controllers to reduce to a mnmum the effect of SUAs on SATS operatons. Nevertheless, n certan locatons, great crcle routes between SATS arports do cross contnuously operatng SUAs that affect all alttudes. The SUA near Edwards AFB, east of Los Angeles and that near Holloman AFB, are examples. On balance, we beleve that t s reasonable to neglect SUA n takng a frst look at SATS traffc. A more detaled study, takng nto account specfc SATS separaton technologes and the characterstcs of specfc SUAs, s desrable for more refned dscusson. As a check on the reasonableness of neglectng SUA ntally, the mplementaton of LMINET-SATS used n ths study checks traffc through SUA near Edwards and Holloman Ar Force Bases. Mountans Mountans would obstruct pston SATS traffc over sgnfcant portons of the western Unted States. Sngle-engne SATS arplanes cannot make flghts on great crcle routes between certan SATS arports n these regons. We have not adusted SATS traectores for mountans for two reasons. Frst, n many cases passes allow SATS traffc to operate wth modest ncreases n dstance. Second, the cases where terran nterferes wth great-crcle operatons are between lghtly populated areas and consttute relatvely small fractons of SATS operatons. SATS AND ATM STAFFING If SATS operatons can be done under Vsual Flght Rules (VFR), the effect of SATS on ar traffc management wll be reduced. SATS arplanes are lght, so n addton to the VFR requrement, crosswnd lmtatons should be consdered. To get a prelmnary ndcaton of the fracton of the tme that weather, ncludng surface wnds, wll allow SATS VFR operatons, we consdered a trp from an arport n the New York area, MMU, to BED near Boston. 4-4

46 LMINET-SATS Our crosswnd lmtaton was 15 knots. Although we know of no FAA or manufacturers restrcton on crosswnd operatons, lght arplane makers typcally demonstrate operatons n no more than 15 knot crosswnds. Personal experence suggests that such wnds pose a farly sgnfcant challenge for a relatvely nexperenced plot. MMU has two runways, as does BED. Usng U. S. Weather Servce (OASIS) data and takng EWR for MMU and BOS data for BED, we found that, for the calendar years 1981 through 1995, 90.7 percent of the tme a SATS plot would fnd MMU n VMC wth acceptable crosswnds, and also would fnd BED n that state an hour later. Ths example suggests that VFR SATS operatons, whle possble a large fracton of the tme, probably are not suffcently often avalable to sut the needs of busness travelers. Mssng or reschedulng a meetng one tme out of ten s probably not acceptable to most busness people. Thus, SATS operatons supportng busness travel must be able to operate n IMC. It also follows that busness-related SATS actvty wll requre ATM staffng capable of supportng t durng IMC. The burdens that SATS operatons mpose on the ATM system wll vary wth ther tracks. An SATS flght departng IFR from an uncontrolled arport, flyng a track that never enters arspace owned by a Termnal Radar Approach Control (TRACON), and landng at another uncontrolled arport presumably would requre servces only from the low-alttude sectors through whch the track passed. If the track dd pass through arspace controlled by a TRACON, that TRACON would servce the flght. For example, a recent IFR flght of a Cessna 150 from Leesburg, Vrgna, (JYO) to Bradley Internatonal Arport, Wndsor Locks, Connectcut, (BDL) receved servces from seven dfferent TRACONs and was controlled only brefly by Washngton ARTCC, even though t never transted Class B arspace. Lght arplanes regularly operate under IFR. In addton to the analyss of ths traffc descrbed n Chapter 3, we examned ETMS data for Aprl 8, 1996, for ndcatons of the nature of present lght-arcraft IFR flghts to gan more nformaton about the knds of trps that SATS arcraft mght take. To have data for lght arcraft only we consdered IFR trps by Cessna 150, 152, 172, 177, 180, 182, 185, and 195 arplanes, together wth IFR trps by Pper Cherokee (PA28) arplanes. We found ETMS records for 1,035 IFR flghts between 0600 EDT and 2200 EDT on that date. Some of these had ntersectons, rather than arports, as destnatons. We gnored these because they gve no ndcatons of ntercty traffc. We also gnored round-robn flghts wth dentcal orgn and destnaton arports for the same reason. Certan ETMS records appeared garbled, (e.g., an arport dentfer wth 9 characters) and we gnored these as well. Ths left 889 flghts. Fgure 4-4 shows the dstrbuton of the dstances of these flghts. 4-5

47 Fgure 4-4. Dstrbuton of Dstances of 889 IFR Flghts of Lght Arcraft Probablty densty Dstance The probablty densty of Fgure 4-4 has a mode near 35 nm, and another near 85 nm. It s lkely that the frst mode largely represents tranng flghts. The rapd decrease n frequency of flghts beyond about 100 nm may reflect lght arplanes plots preferences for flght legs takng no longer than 1 to 2 hours. The very small numbers of flghts for dstances greater than 400 nm s consstent wth confrontng IFR endurance requrements wth lght arcraft fuel capactes. Meetng the requrement for suffcent fuel to fly to the destnaton, thence to an alternate, and land wth 45 mnutes of fuel remanng, generally would lmt lght arcraft to legs of about that length. Favorable wnds, lght loads, and plot endurance may, of course, enable longer flghts. DEFINITION AND OPERATION OF LMINET-SATS LMINET-SATS provdes an addton to LMINET that tracks SATS operatons, and ther nteractons wth LMINET sectors and TRACONs. It provdes 2,865 arports. Input to LMINET-SATS s a lst of SATS flghts between SATS arports, gvng orgn, destnaton, and startng tme for each. Outputs are the number of SATS flghts n each LMINET geographc sector, and n each LMINET arport s TRACON, epoch by epoch. These traffc data may be added to LMINET traffc data, to analyze SATS effects on the NAS. Ths secton gves the consderatons that led to the defnton of LMINET-SATS, descrbes ts operatons, and gves some example outputs. 4-6

48 LMINET-SATS Consderatons for Addng a SATS Underlayer to LMINET Generally, ARTCC controllers do not handle IFR traffc below FL 100 and wthn 30 nm of a Class A arport. Rather, controllers n the arport s TRACON wll drect that traffc. In some parts of the NAS, sgnfcant fractons of the arspace are wthn 30 nm of a Class A arport. Fgure 4-5 shows the regons wthn 30 nm of the 64 LMINET arports. Fgure 4-5. Thrty nm Zones around LMINET Arports The FAA s low-alttude sectors clearly recognze the presence of TRACON arspace. TRACON arspace for Alberquerque (ABQ), Phoenx (PHX), El Paso (ELP), Chsum (CME), and Tucson (TUS) s clearly vsble n Fgure 4-6, whch dagrams the low-alttude sectors for the Albuquerque ARTCC. Fgure 4-6. Low-Alttude Sectors for Albuquerque Ar Route Traffc Control Center 4-7

49 Users can defne any partton of the NAS by a collecton of ponts, and use the result as en-route sectors n LMINET. All LMINET studes to date have consdered NAS operatons many years n the future. Geographcal sectors, rather than the present FAA sectors, were used n these studes because the studes were specfcally drected not to constran operatons by the present arway and sector structures. Sector results from these studes hghlght geographc regons of heavy traffc, when traffc operates on optmal routes. Fgure 4-7 shows a plan vew of basc geographc enroute sectors used n recent LMINET studes. In most of these studes, heavly traveled basc sectors were vewed as dvded nto as many as nne subsectors. Fgure 4-7. Geographc Enroute Sectors Because fully developed SATS wll happen more than a decade n the future, we beleve that the SATS study also should use geographcal sectors for ar carrer traffc, at least ntally. Ths wll gve consstency wth prevous studes, and wll avod constranng SATS operatons by the arways and enroute sectors of today s NAS. Lght-arcraft SATS operatons under IFR are lkely to affect TRACONs, n addton to ARTCC sectors, because they typcally wll operate below FL 100. Accordngly, we ntend to use the enroute sector structure of Fgure 4-8 for lghtarcraft SATS operatons. 4-8

50 LMINET-SATS Fgure 4-8. Enroute Sector Structure For Lght-Arcraft SATS Operatons If the effects of SATS on workloads n the present NAS are desred, we wll use the FAA s hgh-alttude sectors, shown n Fgure 4-9, for ar carrer traffc and turbne SATS traffc. Fgure 4-6 s an example of the FAA s low-alttude sectors that we wll use for low-alttude SATS traffc. Fgure 4-9. FAA Hgh-Alttude Sectors FAA sectors are closely, and sometmes qute narrowly, lnked to present J- and V- ar route structures. To use the FAA sectors effectvely, we wll take as LMINET ar carrer and SATS turbne traectores not the wnd routes used n prevous studes, but representatve ETMS traectores. Fgure 4-10 shows an example. 4-9

51 Fgure Example Traectory Lght SATS traffc, however, appears lkely to relate dfferently to the ar route structure. Lght SATS trps wll be short, usually not more than 300 nm, and they wll take place below FL 120. Presently, we are nclned to send lght SATS traffc on great crcle routes, perhaps avodng TRACON arspace enroute. Under IFR, a lght-arcraft SATS flght wll mpose load on the sectors through whch t passes, ether geographc or FAA, and also on the TRACONs of each LMINET arport to whch the flght gets as close as 30 nm. In some areas, notably the md-atlantc coastal regon, flghts may be handled entrely by TRACONs. We wll track SATS lght-arcraft traffc separately from other TRACON traffc so that analyses may consder ths traffc to be handled by separate TRACON statons. Followng s a bref descrpton of arports n LMINET-SATS, the extenson of LMINET to model SATS operatons. LMINET-SATS has two arport classes, maor and SATS. The maor arports are the 64 LMINET arports. The SATS arports make up a set of about 800 other arports. The SATS arports generally wll have at least one hard-surfaced runway not less than 2,000 feet long. Maor arports capacty models are the present LMINET capacty models. Ar carrer operatons at these arports are modeled as n the present LMINET. Because the purpose of our study s to model SATS as relef to maor arports, SATS operatons wth pston arcraft do not nvolve maor arports. (Of course, SATS-to-maor arport and maor-arport-to-sats operatons n lght arcraft are possble, and they could be modeled f such operatons dd not volate the lght SATS sprt.) 4-10