USF Libraries
USF Digital Collections

Driver population factors in freeway capacity

MISSING IMAGE

Material Information

Title:
Driver population factors in freeway capacity
Physical Description:
1 online resource (70 leaves) : ill. ;
Language:
English
Creator:
Lu, Jian, 1957-
Huang, Weimin
Mierzejewski, Edward A
Florida -- Dept. of Transportation
University of South Florida -- Center for Urban Transportation Research
Publisher:
University of South Florida, Center for Urban Transportation Research
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Highway capacity -- Mathematical models -- Florida   ( lcsh )
Traffic flow -- Mathematical models -- Florida   ( lcsh )
Speed limits -- Mathematical models -- Florida   ( lcsh )
Tourism -- Florida   ( lcsh )
Genre:
bibliography   ( marcgt )
technical report   ( marcgt )
non-fiction   ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 69-70).
Statement of Responsibility:
principal authors, J. John Lu, Weimin Huang, Edward A. Mierzejewski.
General Note:
Title from cover of e-book (viewed Aug. 15, 2011).
General Note:
"Submitted to: Florida Dept. of Transportation.
General Note:
"May 1997."

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029308412
oclc - 746802702
usfldc doi - C01-00003
usfldc handle - c1.3
System ID:
SFS0032126:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ia 4500
controlfield tag 001 029308412
005 20110816122307.0
006 m d
007 cr bn|||||||||
008 110815s1997 flua obt 000 0 eng d
datafield ind1 8 ind2 024
subfield code a C01-00003
035
(OCoLC)746802702
040
FHM
c FHM
043
n-us-fl
090
HE336.H48 (ONLINE)
1 100
Lu, Jian,
d 1957-
0 245
Driver population factors in freeway capacity
h [electronic resource] /
principal authors, J. John Lu, Weimin Huang, Edward A. Mierzejewski.
260
Tampa, Fla. :
b University of South Florida, Center for Urban Transportation Research,
1997]
300
1 online resource (70 leaves) :
ill.
500
Title from cover of e-book (viewed Aug. 15, 2011).
5 FTS
"Submitted to: Florida Dept. of Transportation.
FTS
"May 1997."
FTS
504
Includes bibliographical references (leaves 69-70).
650
Highway capacity
z Florida
x Mathematical models.
Traffic flow
Florida
Mathematical models.
Speed limits
Florida
Mathematical models.
Tourism
Florida.
700
Huang, Weimin.
Mierzejewski, Edward A.
710
Florida.
Dept. of Transportation.
2
University of South Florida.
Center for Urban Transportation Research.
776
i Print version:
Lu, J. John.
t Driver population factors in freeway capacity.
Tampa, Fla. : University of South Florida, Center for Urban Transportation Research ; [Springfield, Va. : Available through the National Technical Information Service, 1997]
w (OCoLC)37319974
773
Center for Urban Transportation Research Publications [USF].
4 856
u http://digital.lib.usf.edu/?c1.3
FTS
y USF ONLINE ACCESS
951
10
SFU01:001393146;
FTS



PAGE 1

DRIVER POPULATION FACTORS IN FREEWAY CAPACITY Submitted to : F l orida Department of Transportat ion Center for Urban Transportation Research College of Engineering University of South Florida 4202 E. Fowler Avenue, CUT 100 Tampa Florida 33620-5675 Principal Authors: J. John Lu, Ph.D., P.E. Weimin Huang Edward A. Mierzejewski, Ph.D., P.E. May 1997

PAGE 2

Tf.OCNlCAL R.UORT DOCUME."'lTA1 '10N PACt I. fttpottHo 1 No. (NTIS) 3 WPI No. 0510759 4 $. RPOn Dc Driver Popul ation Factors In Freeway Capacity 5197 6. Cedi 7. NJtllot'(.) a. Pcdormifle No J John Lu. Ph .D., P.E . We imin Huang, and Edward A Mierzejewski, Ph.D .. P. E. t 1 0 v.b'klkllHo.(TRAJ$) Center for Urban Transportation Research USF College of Engineering 11. COI'IW'td Grvt No. 4202 E Fowler Avenue. CUT 100 Tampa, FL 33620 6 1 1 spomomg ..aqoncy N...,., Olld Acli:hn 1 3. Typo of Rtp0tt Olld Ptriocl C:Ow.cl FOOT Department of Tra n sportat i on Final Report 605 Suwannee Street 8/14195 5130/97 Tallahassee, FL 32399-0450 t4, $potl101'1r1Q A9MCY Codl 1!1, HotM 16. Ab.ncl The methods con t a ined in the Highway Capacity Manual are based on a traffic composition of local drivers familiar roadway characteristics. The Manual allows for t he incorporation of a driver population factor adjustment into freeway capacity calculations to reflect the influence of unfamiliar drivers. U nfortun ately the Manual offers very little guidance on the appropriate values of the factor. This project used cont i nuous count traffic data at a number of Florida freeway locations to est i mate the appropria t e values of the driver popu l at i on factor By re l ating speed-volume characteristics to measures of driver popu l ations the importance of this factor was demonstrated Estimat es were made of the factor values based on a sample of locations in Florida l 17. l(eyWds 11. Ois".ritlutiM i highway capac i ty dr i ver Rep ort available to the public t hrough the characteristics. tourist travel Nat iona l Techni ca l Information Service (NTIS) I 5285 Port Royal Road Springfield, Virgi n i a 22161 (703) 487-4650 1&. s.o.,.rii)'C .. 20. 21. No. of poget. ....... j unclassified unclass ified Form DOT F 1700.1 (8-72) Rt .prochtttion ot wmplettd page au1horiztd

PAGE 3

DISCLAIMER The opinions, findings and conclusions expressed in this interim technical report are those of the authors and not necessarily those of the State of Florida Department of Transportation. iii

PAGE 4

ACKNOWLEDGMENT S The authors express t heir s incere apprec1alion to the Florida Department of Transportation for its support of this project. I n particu l ar the assistance of the following individuals is recognized: Douglas McLeod and Kurt Eichin of the Systems Planning Office, H arshad Desai and Rick Reel of the Transportation Statistics Office, and Richard Long of the Research Office. It was through their support and assistance that this project b e came a reality i v

PAGE 5

TABLE OF CONTENTS L IST OF TABLES .............. . .. .. .. . ...... ...... ......... ....... ....................... .. ............. .. . ............ vii LIST OF FIGURES ..................................... .............................................. ........ .............. viii CHAPTER 1 : ........ .. . .. .. .. . .. .. ........ .. . .. ........ .. ................ ..... .. ...... ..... .... ................. .. .......... Background . ......... .... .. ........ ... .... ... .. .. .. . ...... ............. .. .. ........ .. . .................... .... .. . . . . I Research Problem Statement ..... .. .. ................... .. . .. .. ... ...... . .. .. .. .. ...... .. ............ .. . . ... 5 Study P e rformed ................. ........ ................................................... . . . . .. . .. .. .. . .......... .. .. 6 Scope of the Report ....... ..... .. .............................................................. ....... . . . ....... ... ... ? CHAYI'ER 2: REVIEW OF PAST STUDIES ................................................................ 9 Capacity Analysis and the H ighway Capacity Manual .... ...... ...... ... .. .. .. ................. . 9 Reviews of fw and t;w Adjustme n t Factors ........ . ............... ..... ...................... .... . .... . . 0 Past Studies of Driver Population Factors ...... .... . ............ .. .. .. .. .. . ........... ......... ... .... .. 12 Other Related Studies .. .. . .. . . .. .. ................. ................ .. .. .. .. .. ..... ...... ..... .. .. . . .. .. 15 CHAPTER 3 : PRINCIPLE USED IN THE PROJECT ................... .. ....................... 16 Definition of NonLocal Drivers ........ .. .......... .. .... .. .. .... .. ..... .. .. .. ........... ................. 16 Basic Concept . .... .. .. .. .. .......... .. .. ....... ................................ .... .. .. ... ................. ..... .. 16 Estimation of Non-L o cal Driver Population Levels ................ ............ ......................... .!? Principle .... ........ .. .. ........ .. .. ...................... ............ .... ........................................... ...... l8 Calibration of Population Adjustment Factors .. .......... .. .......... ................... ...... ....... ... 22 CHAPTER 4: RESEARCH DATA SOURCE$ ........................................................... 24 Traffic Data ....... .............. ..... .................. .. ................................. ..... .. ............ ........... .... 24 Tourist Survey . .. .. . ........... ....... .. ... .............................................................. .. .. .. .. ..... 26 Estimation of Non-Local Driver Population Levels Using Tourist Survey Data ........... 27 Estimation ofNon-Lo cal Driver Population Level s Using Traffic Data .................. ...... 30 CHAPTER 5: DEVELOPMENT OF DRIVER POPULATION ADJUSTMENT FACTOR TABLE BASED ON TOURIST SURVEY DATA .. .. .... ........ ....... .. .. .... .. ... 37 Modeling Procedure and Results .......... .. ............. .... .. .. .. ........ .................... .. ..... .. 37 More Generalized Results ................................. ........... .. .. ........... ... ............. ....... .... .... 45 Estimate of Driver Population Factors Based on Test Site 0130 .................................. 46 Discussion .. .... ...................... .. ................................................... ....... ........................ ..... .4 7 CHAPTER 6 : DEVELOPMENT OF A DRIVER POPULATION ADJUSTMENT FACTOR TABLE BASED ON TRAFFIC CHARACTERISTICS ........ ............ ........ 49 Monthly Fac t ors, Weekly Factors, and Daily Factors ................................... .. .... .. ........ .49 Month ly factor .... ........ .. . ............. ... .......................... .................. ....... .. ... ............. ... ... 49 Weekly factor . .. .. .. .. ......... ..... .. .. .. ...... .. ................. .. .. .. .... .. .. ..... .. .. .. .. ..... ..... 5! Daily factor . ................... .. . .......... ................... ....... ..... ......... ...... ....................... .. .. 52 v

PAGE 6

TABLE OF CONTEN T S ( CO NT I NUED) Correlation between monthly factor, weekly factor, and daily factor ... .. ... . .. .. ........ 53 Speed-Volume Mode l s . ..... ................ .. .. .. ... .............................................. ..... .. ......... 54 Development of an Inde x ......................... .. .... .. . .. .. .. .. .. .. ..... ....................... .. ... .... 57 Model Specificatio n s ........... ..... ..... .. .. ................ ............................... .. .................. 51 Index calibration ................. .................... .. ...... .. .. .. .. .. .. .. .. .. ................. ............. 57 F i nal index ......................................... .. ................................................ .. .. ... ........... 58 Impact ofNon-Local Driver Population Leve l s on Speed-Volume Curves ........ .. .. ...... 59 Estimate of Driver Pop u lation Factors ........................ .. ....................................... ........ 60 D . JS C USSJOn ................................................................... .. .................................................. 63 CHAPTER 7: S UMMARY, C ONCLU S I ONS, AND RECOMMEND A TI ONS . .. .. .. ............. ............. .... ................... ........ .............. ..... 64 s l1llll1l ary ................................................ .......... .......... .... . .... .... ........................ ............. 64 Conclusions ............ ............. ....... .... ........... ...... . ...... .... ..... .... ........ .............. ... . .... ....... 65 Recommendations .. .. .. .. .. ...... .. ..... .. .. .. .......................................... .. .. .. ................. 68 REFERENCES ......................... ..... ........................ ...................... ........ ............ ................... ... 6 9 vi

PAGE 7

LIST OF TABLES Table 1-1: Adjustment Factors for Driver Populations (HCM 1994) ............ ............ ... ... .4 Table 1-2: Annual Vehicle-Mile Traveled in USA and Florida ......................................... 5 Table 2 -1: Reconunended Population Adjustment Factors by Sharma ........................... l4 Table 4 -1: FOOT Traffic Count Stations and AADT ...................................................... 26 Table 4-2: TIVD and 01 Values for the Orlando Area ................................................... 29 Table 4-3: MF, WF, and OF Valu es .. ... ........................... ............................................... 34 Table S -1: Regression Analysis Results (a & bin Eq. 5-1) (1-4, WB, Orlando, Site: 0 130) and Monthly Non-Local Driver Indices (DI) (Orlando ) ................................. .40 Table S-2: Maximum Service Volumes (pcphpl) for LOS B, C, and D and Different DI Values ...................................................................... .... ...... .......................... .47 Table S-3: Driver Population Adjustment Factors for Different Levels of Service, Based on Orlando Conditions ............................................................................................ 48 Table 6-1: Correlations between Factors ..... ., ........ ..... ..................................................... 54 Table 6-2: Speed-Volume Models at the Three Sites ...................................................... 55 Table 6-3: Index Model Specifications .................... .. ......... ........................................ .... 57 Table 6: Index Model Calibration Results .................................................................... 58 Table 6-S: Maximum Service Volumes (pcphpl) for LOS B, C, and D and Different NDI Values ..................................................... ......... ............ ......... ... ................... 61 Table 6-6: Driver Population Adjustment Factors for Different Levels ofService ......... 62 vii

PAGE 8

LIST OF FIGURES Figure 1-1: Speed-Flow Characteristics for Basic Freeway Secti ons (For Ideal Conditions): (a) F our Lane Freeways, (b) Six-or-More-Lane Freeways. (HCM 1 994) .. ......................... ....... .......... ..................... .. ... ...... ................................... . ... ... 2 Figure 3-1: Impact of Non-Local Driver Population ..... ...... ..................... ...... ........... . . 19 Figure 3-2: Typical Relationship between Volume and Speed (1-4 WB Middle Lane, March 1995, Orlando) ...... ....... . ....... .. .... . .... . ...... .............. ... ..... 20 Figure 3-3: Concept of Estimating fp ...... ..... ..................... . .......... .................................. 21 Figure 3-4: Basic Principle of Developing Driver Population Adjustment Factors .. ...... 23 Figure 4 1 : Locations ofFDOT Traffic Count Stations ....................... .......................... 25 Figure 4-2: DI Values for Different Months in Orlando Area .... ... .......... ........ ..... .. ..... ... 30 Fig ure 4-3: MF Values for Different Months in Orlando Area (WB and Lane I Only) .......................................................................... ..... ............ .... . ... 32 Figure 4-4: WF Values for Different Months in Orlando Area (WB and Lane I Only) ...... ........... ... . .......... ........... ... ..... ........ ..... .... .............................. .33 Figure 4-S: OF Values for Different Months in Or l ando Area (\VB and L ane I Only) .. .......... ....... ...... ... ..... . ............. ...... ..... ..... ................ ..................... 33 Figure 5-1: Relationship between Volume and Speed (I-4 WB Lane I January 1995, Orlando, Site: 0 1 30) .. .................... ...... . .. ...... ..... ..... ... ...... ............ ........ ... ......... . . 38 Figure 5-2: Relationship between Volume and Speed (1-4 WB Lane 2, October 1995, Orlando, Site: 0130) ........ ..... ............................................. ... ... ..... .... ...................... 38 Figure 5-3: Relationship between Volume and Speed (1-4 WB Lane 3 March 1995, Orlando, Site: 0 130) ............. .... ... . ......... ... .... ................ ...................... ......... ........... 39 Figure S-4: Statistical Relationship between Parameter "b" and Monthly Non-Local Driver Index (DI) (I-4 WB, Orlando, Site: 0130) . ... ............... ..... .... ............ .41 Figure S-S: Statistical Relationship between the Average Value of Parameter "b" and Monthly Non-Local Driver Index (DI) (with 2"'-order polynomial curve) .............. .43 viii

PAGE 9

LIST OF FIGURES (CONTINUED ) Figure 5-6: Impact of Non-Local Driver Population on Average Operating Speed (1-4 WB, Orlando, Site: 0130) ... .... .... . ....... .... ... ... ....... ..... ............. .. ... ... ......... ... ...... ... .44 Figure 5-7: Impact ofNon-Local Driver Population on Average Operating Speed (1-4 Both Directions, Orlando, Site : 0130) ..... ................. ..... ..... ......... ... ..................... .45 Figure S-8: Est i mation of D river Population Adjustment Factors (1-4 Both Directions, Orlando, Site: 0130) ... ... ......... ......... .............. . .... ...................... 46 Figure 6 1: Monthly Factors of Different Months at Sites 0130 0171, and 0174 .......... 51 Figure 6-2: W eekly Factors of Different Months at Sites 0130, 0171, and 0174 ........... .52 Figure 6-3: Daily Factors of Different Months at Sites 0130, 0171, and 0174 ............... 53 Figure 6-4: Impact ofNon-local Driver Population on Average Operating Speed ........ 60 Figure 6-5: Density Line for LOS C (14.92 pclkmlln) .... . ............... .... ........ ........... ..... 61 ix

PAGE 10

CHAPTER 1: INTRODUCTION Background The capacity of a roadway facility is defmed as the maximum number of vehicles that can be accommodated by the facility during a given time period. Capacity analysis of freeways is one of the important procedures, as freeways generally carry a high proportion of an area's traffic. Freeway capacity analysis procedures have been used to evaluate level of service (LOS) of basic freeway segments, design number of freeway lanes, and estimate the maximum service flow at prevailing conditions. The Highway Capacity Manual (HCM) published by Transportation Research Board has been the guidance for capacity analyses since it was first published in 1950 (HCM 1994) The latest version of the HCM is the 1994 Update of the Third Edition (HCM 1994). The HCM covers every major aspect in highway transportation, including freeways, rural highways, urban streets, transit, bicycles, and pedestrians In the freeway section, the HCM includes basic freeway segments, weaving areas, ramps, ramp junctions, and freeway systems. The generalized analysis approach in the HCM consists of three major steps. The first step is to find the capacity of highway facilities under ideal conditions. Second, the levels of service are used to represent different operating qualities and to determine the maximum flow rates under these different levels of service. Finally, adjustment factors are applied to the ideal conditions to adjust capacity and the maximum flow rates at different levels of service to take account of capacity reductions caused by prevailing (non-ideal) roadway and traffic conditions. In the HCM, speed-volume models are used for the capacity analysis of the basic freeway segments under ideal traffic and roadway conditions. Different curves are provided for four lane freeways and six-or-more lane freeways at different free flow speeds. Figure I I presents the speed-flow curves copied from the 1994 HCM The capacity under ideal conditions i s 2,200 passenger car per hour per lane (pcphp l ) for four-lane freeways and 2,300 pcphpl for six-or-more-lane freeways (HCM 1994).

PAGE 11

Figure 1-1: Speed-Flow Characteristics for Basic Freeway Sections (For Ideal Conditions): (a) Four Laue Free w ays ( b ) Sil:-or-More-Laoe Freeways. ( HCM 1994) -" ... ... . . ... . ... . ........ ... ... .. : ....... . ...... . ........... .. ,_WPM: Ba .:, , : !300 .... : . : .... : .. lu .. ._ : i.a.:o .... : : : .. 1 .. : . . : E.vJ . .. ; .... .; .... :. . : .. . .: . . .; ..... ; .. .. :. . . .: . a4SJ . . ... : ..... : ... .. .. .... ; . ... ; .... .. . : eu-. ; ..... :. .. ; ..... ; . . :, . . ; . . .;. ... .. : .... ; ... i ,"E: ... \ . . . . OM-' : ........ ..... : . ...... . . . .... . G:,I: .. .. I u C c . . . . . ........... . .... : . . : ......... : . . : . : .... : . ;;I .. c.. : o !t : u; I' : : : : : ::!: : , -: : :...... . : .... : ........... : .. .. : .. .. : .... : .. j I ...... : . : . . , , ... .. IOU:. fLOW R.lT.E TOR 15-loi i NUTt PtRIO O (PCPHPL) (a) ,....1S-' . . ............. ; .... . . ... : I '1.0 """ "'' ;.;. Per. \300 ,c:;l:pt . \ .. 1 <( : .... :. . . I h I : . !;! I c )< Q loo !!---._ c. I L o u e : N O .: ". Ill ''' ' ' - '''' !O!A : : cw (Cii I (b) 2

PAGE 12

Six levels of service, from LOS A to LOS F, are defined for freeways. LOS A represents the best opera tions qua lity and LOS F the worst. LOS E is the situation where traffic demand just reaches maximum facility capacity As the traffic demand continues to increase beyond the capacity, the LOS will degrade into tbe stop-and-go forced flow condition of LOS F The maximum number of vehicles that could pass a freeway section in a unit time under each LOS is called the maximum service flow rate for that LOS, except LOS F at which traffic conditions are unstable. According to th e HCM, the capacity of a basic freeway segment is based on the ideal capacity that could only be observed under ideal roadway and traffic conditions. The ideal roadway and traffic conditions for basic freeway segments are described as follows (HCM 1994): Good w e ather, Good pavement conditions, No incident affecting traffic flow, Level terrain, 12-ft minimum lane widths, 6-ft minimum lateral clearance between the edge of the travel lane and the nearest roadside or median obstacle or objective influencing traffic behavior, All passenger cars in the traffic stream, and A driver populat ion dominated by regular and familiar users of the facility. Capacity reductions of a basic freeway segment would be observed under prevail.ing traffic and roadway conditions. To calculate capacity reductions due to prevailing conditions, three adjustment factors, namely adjustment factors for lane widths and lateral clearances (fw), adjustment factors for heavy vehicles (fHV), and adjustment factors for driver populations (t;,) shou ld be applied. This approach is mathematically described by the following equation: (1-1) 3

PAGE 13

where SF; = service flow rate for LOS i under prevailing roadway and traffic conditions for N Janes in one direction vph, c; =the ideal capacity, 2200 pcphpl for 4-lane freeways or 2300 pcpbpl for 6-lane freeways, (v/c}. =maximum volwnelcapacity ratio for LOS i, N = nwnber of lanes in one direction of the freeway, f w = factor to adjust for the effects of restricted Jane widths and lateral clearances, fHV =factor to adjust for the effects of heavy vehicles on the traffic stream. and = factor to adjust for the effects of recreational or unfamiliar driver populations. Among these three adjustment factors, there are specific definitions and clearly defined calculation procedures for fw and fHV in the HCM, but only a very simple table is presented for fP, as shown in Table 1-1. As noted in the 1 994 HCM, the driver population adjustment factor is said to range from 0.75 to 0.99 and is to be applied when there are significant percentages of non-commuters in the traffic stream. Unfortunately, the HCM offers little guidance on how to select appropriate values for this factor. As a result, the driver population adjustment factor is commonly ignored. Without clear instruction on selection of the driver population factor, significant bias may be introduced in capacity analysis particularly, in an area such as Florida with a significant percentage of tourism traffic volume or non-commuters. Non-local drivers or non-commuters in the traffic Table 1.1: Adjustment Fadors for Driver Populations (HCM 1994). Traffic Stream Types Population Adjustment Factors Weekday Commuters 1.00 Recreational or Other 0.75 0.99 4

PAGE 14

stream may cause capacity reductions in several ways, including perception-reaction time, car-following behavior, lane change and gap acceptance behavior, and driving speed. As the percentage of non-local drivers or non-commuters increases, these factors combine to contribute to the expected reduction in freeway capacity. Research Problem Statement Although the mileage of Interstate, Turnpike, and other fully access-controlled expressways was only 1.7% of the total highway system in Florida in 1994,24.8% of the total annual vehicle-mile-traveled took place on freeways in that year (HPMS 1994). A safe and efficient freeway system is vital for the economy of F lorida To achieve this goal, traffic engineers should have a practical tool to perfonn freeway capacity analyses. For decades, that "tool" has been the HCM Applying adjustment factors is a very important procedure in the analytical approach of the HCM. Without appropriate definitions and practical calibration methods, it is difficult to make reasonable estimates oft;, from the range of 0.75-1.00 as it appears in the HCM. This defect is especially obvious in Florida. Many freeways in Florida have good geometric desigJI, level terrain, and nonnal percentages of heavy vehicles, as shown in Table 1-2 (FHWA 1994). Such characteristics may resul t in the adjustment factors (fw and fuv ) approaching 1.0. On the other hand, Florida has many tourist attractions Table J 2: Annual Vehicle-Mile Traveled in USA and Florida Passenger Cars Single-Unit 2-Axle Freeway Typ e s Motorcycles Buse s & Other 2-Axle 6 Tire or More & (%) (%) 4-Tire Vehicles Combination Trucks (%) (%) Rural Nationwide 0.6 0.3 80.5 18.6 Interstate Florida 0.5 0.7 80.5 18.3 Urban Nationwide 0.4 0.2 91.8 7 6 Interstate florida 0.4 0.6 91.5 7.5 5

PAGE 15

and about 40 million out-of-state visitors annually (UPF 1994). That number is nearly three times of the number of residents in the State (FOTR 1995). Tourism is the biggest industry in Flori da ; thus it is almost impossible to ignore the presence of different driver populations. If the fp range of 0.75-1.00 is ap plied, combined v.itb the values of fw and fHV which are near 1.00 as mentioned earlier, t;, could be the most influential adjustment factor in capacity analyses. Study Performed In 1995 the Florida Department of Transportation (FOOT) contracted with the Center for Urban T ransportation Research (CUTR) at the University of South Florida (USF) to conduct a research project, "Driver Population Factors in Highway Capacity," to examine the driver population factors for highway facilities in Florida. The first phase of the project examined traffic data sources methodologies of estimating non-local commuters and non-local drivers, and included various experimental design protocol for both freeways and for signalized intersections. Among the methods considered for estimating non-local drivers were: Roadside interviews, o Survey data from the Florida Office of Tourism Research, o "Lights -on" type surveys o License plate recognition, o Toll plaza surveys, and Seasonal variation factors. The Phase I effort was summarized in a technical report entitled Population Factors in Highway Capacity: Interim Technical Report (Phase 1: Experimental D esign for Data Collection and Analysis)." In addition, an annotated bibliography, reflective of a comprehensive literature review and tel e phone interviews, was prepared. Following the completion of Phase I, CUTR and FOOT jointly reviewed the resources required to carry out the research protocols for different ro adway types and determined to focus on one spe c ific roadway type, i.e., basic freeway segments. Extensive efforts were 6

PAGE 16

undertaken for data collection and analyses within the constraints of the project resources. The FOOT permanent count traffic data and the F lorida Visitor Survey data were chosen because these two sources have their advantages in data quality, availability, and cost over other data sources such as manual or video count collection and license plate surveys. This project focused on basic freeway segments because they were the fundamentals of freeway capacity analyses, and as mentioned earlier, freeways played an important role in Florida. With minor modification, the methodology used in this study could be extended to other parts of freeways such as weaving areas and ramps and other highway facilities. The objectives of this study were: (I) to review available techniques related to the topic of driver population factors, (2) to develop a practically feasible procedure to estimate relative driver population levels in freeway traffic streams, (3) to develop a procedure to evaluate the impact of different non-local driver population levels on the capacity of basic freeway segments, and (4) to examine the fp values (0.75-1.00) suggested in the HCM and to present a much more detailed fP table which would correlate different driver population levels with the correspondent capacity reductions. Two different driver population adjustment factor tables were developed. One table was based on the database obtained from tourist surveys performed by the Office of Tourism Research, Bureau of Economic Analysis of the Florida Depariment of Commerce. The other table was based on the traffic database obtained from FOOT traffic count stations The main methodologies were based on the impacts of non-local driver population levels on the speed-volume curve. Capacities at different non local driver population levels were estimated by the direct-empirical method The direct-empirical method can result in the estimation of capacity which may more practically reflect real traffic situation. Scope of the Report This report surnrnari2es the study performed by CUTR and sponsored by FOOT and presents results obtained through the study. The report consists of seven chapters. Chapter 7

PAGE 17

2 reviews past studies related to the topic of driver popu l ations. Chapter 3 describes the methodo l ogies used in this project Chapter 4 discusses the data resources used in the project Chapters 5 and 6 summarize the procedures developed through this study to evaluate the impacts of non-local driver population levels on capacity reductions Chapters 5 and 6 present two different driver popu l ation adjustment tables obtained from the study and based on the data collected in the study. The results shown in Chapter 5 were bas e d on the tourist survey database, and the results shown in Chapter 6 were based on the FDOT traffic count station database. Chapter 7 discusses conclusions, summaries, and recommendations resulting from the srudy. 8

PAGE 18

CHAPTER 2: REVlEW OF PAST STUDIES Capacity Analysis and the Highway Capacity Manual The earl ies t highway capacity studies dated back to the early 1920s when a capacity analysis conunittee was set up by the Highway Research Board. In I 950, the first Highway Capacity Manual was published by the Highway Research Board, and it quickly became the standard for highway capacity analyses in the United States and many other countries (HCM 1994) In the 1960s research attention was paid toward freeway capacity analyses along with the construction of the Interstate Highway system throughout the nation (May 1990). In 1 965, the second edition of HCM was published to replace the outdated I 950 HCM (HCM 1994). The 1965 HCM included the level-of-service concept into the manual Since then, traffic has continued to grow at an even faster rate than new highway construction, and as a result, traffic operations quality had become a major concern of state and local transportation agencies and of the general public. After rwo decades of comprehensive research the third edition of HCM was published in 1985 by the Transportation Research Board The 1985 HCM was viewed as a milestone in the growing body of knowledge of highway capacity because of the extension into facilities other than highways and the refinement on the LOS concept (HCM 1994). The latest HCM is the I 994 Update of the third edition, marking another significant achievement in highway capacity research (HCM 1994). In the freeway capacity analysis sections of the 1994 HCM, fw and fHV tables were updated, and free flow speeds were used instead of design speeds. The 1994 HCM included driver population factors for freeways capacity analysis, but not for other facilities. It is possible that such factors would be equally important for unintenupted flows on arterial as w ell as intersections The HCM covered every major aspects of highway transportation, inc luding highways, transit, pedestrians, and bicycles, and it plays an even more important role today because 9

PAGE 19

transportation planners, designers, and operators are striving to maintain an efficient and safe transportation system under more and more traffic demands and limited infrastructure funding. Reviews of fw and fHvAdjustment Factors One of the most unarguable adjustment factors in the HCM is fw. and it has been in practice for decades. Nowadays fw values are usually 1.00 in practice because new highway construction normally has adequate lane widths and lateral clearances, and this could also be the reason why very few recent reviews are available for fw studies recently. In the HCM, fHV is also a well developed adjustment factor with refined calibration methods and detailed estimation. The most comprehensive ffiV study was documented in a Federal Highway Administration report (FHW A 1982). Passenger Car Equivalents (PCEs) were used to convert a traffic stream composed of a mixture of vehicle types into an equivalent traffic stream composed exclusively of passenger cars. The fHV is computed from such PCEs and the proportions of heavy vehicles in the traffic stream using the following equation according to the HCM (HCM 1994): where (2-1) E,., = passenger car equivalent for trucks/buses and recreational vehicles, respectively, and P1 P, =proportion of trucks/buses and recreational vehicles respectively. In this fHv study, vehicle classifications, headway, and speeds were collected in the field. Then headway values of different vehicle types was compared with the standard headway of passenger cars to determine the re l ative amounts of space coQSUmed by different vehicle types. This approach is described by the following equation: 10

PAGE 20

where SH PCE, = SHPCj (2-2) PCEij =passenger car equivalent of vehicle type i in conditionj, SHij = mean inter-vehicular spatial headway (measured from the vehicle type i's rear bumper to the rear bumper of the leading vehicle) for conditionj, and SHpq = mean inter-vehicular spatial headway (measured from a passenger car's rear bumper to the rear bumper of the leading vehicle) for conditionj. Another method used in the fHV study was to measure the equivalent delay. Drivers could drive at any lawful speed except when they were obstructed by slower vehicles. PCEs could be calculated from the different delay time of different vehicle types. This approach is described by the following equation: where D,Db PCE, = J ase Dbase PCEij = passenger car equivalent of vehicle type i in condition j, Dii = delay to passenger cars due to vehicle type i in condition j, and Dbase = delay to standard passenger car s due to slower passenger cars. (2-3) These two methods are very effective for fHV estimation. Unfortunately, they are not practical for estimation due to the fact that it is much more difficult to determine the driver type (driver population) than it is to determine the vehicle type. Manual surveys and video taping could provide accurate information about vehicle type at a low cost and sometimes even at a fast rate if image processing techniques are used. But these types of observations offer little information about drivers. The license plate survey is not a reliable source in detecting driver populations because this method could not identify 11

PAGE 21

rental cars, most of which are driven by visitors. Rental cars could register in different counties. It is also difficult to read customized license plates which are becoming more and more popular in Florida. It is very difficult to record accurate driver information without stopping traffic and doing conventional roadside interviews. The conventional roadside interview is the only method to obtain direct information about driver populations. This kind of survey is very costly, and in some locations such as freeways, it is impractical to implement. The complexity and difficulty in driver population estimation might b e the reason why very limited research has been done. Past Studies of Driver Population Factors The Transportation Research Information Services (TRIS) is the Transportation Research Board's bibliographic database; it is the most comprehensive and current source for transportation information retrievals in the nation. The TRIS database contains document abstracts describing the published literature of research on highway, transit, highway safety, railroad, maritime and air transportation. In this study, the TRIS was searched to find past studies related to driver populations. Only two articles were found from TRIS with the topic keyword of"driver population factor." Both of the articles were written by Sharma. One article (Sharma 1987) was based on his other article, "Road Classification According to Driver Population" in the Transportation Research Record No. I 090 (Sharma 1986). The contents of his study will be discussed later in this section. Telephone conversations with several members of the Transportation Research Board Highway Capacity Committee confmned that very linle ex ists in the way of documented studies of driver population factor Some earlier studies attempted to evaluate the impact of non-local drivers or non commuters on freeway traffic capacity. The literature search undertaken as part of this project found very little previous research to specifically quantify the magnitude of the driver population factor. It appears that some of the early interest in the driver population factor can be traced to a number of traffic engineers working in traffic operations at the 12

PAGE 22

California Department of Transportation (Caltrans). In conversations with Caltrans traffic engineers, they recalled that a number of studies were performed in the early 1970s on California freeways and that field observations indica ted substantially lower capacity level involving high levels of recreational traffic. Based on telephone conversations performed in the CUTR's project, several members of the Highway Capacity Committee confirmed that the fp range 0.75-1.00 in the HCM was largely based on these anecdotal reports. Unfortunately, the traffic studies performed by Caltrans in early 1970s took the form of internal Caltrans working memos Because the studies were done over twenty years ago, it was impossible to locate the internal memos. In the 1980s, researchers in Europe found controversial results about the driver population factor as part of their comparisons of the HCM with European practical experience (OECD 1983). In their studies, capacity drops to 17 percent were found on a Sunday evening compared to an average week-day on a motorway near Marseille. However, such capacity variations could not be found in other sites with similar traffic and roadway conditions. The researchers also found different speed-volume patterns for peak hour and off-peak traffic, but they did not examine the impacts on highway capacity. In their report, the researchers agreed that the most significant external capacity factor was the role of driving behavior, on which considerable research would be required. More recently, Sharma (1986, 1987 and 1994) has shown considerable interest in the driver population factor. His primaty contribution is in the area of classifying roadways in terms of their traffic composition (1986). He developed a classification system that characterized roads as ranging between the two extremes of urban commuter and highly recreational. The driver population factor for urban commuter traffic would be 1.0 and for highly recreational traffic would be 0.75 (1987). As shown in Table 2-1, he identified five additional categories between those two extremes, and associated a different driver population factor with each. Sharma's study defmed both trip purpose (e.g., commuter, recr eationa l) and trip length (e g urban reg i onal and inter-regional) as the descriptors of driver populations Master traffic patterns of seasonal, daily, and hourly traffic variations 13

PAGE 23

were built in his study to categorize roadway traffic streams in some study sites in Alberta, Canada. The initial step was to group the roadway types by volume distribution Table 2-1: Recommended Population Adjustment Factors by Sharma. Traffic Stream Types Population Adjustment Factors Urban Commuters 1.00 Regional Commuters 0.95 Regional Recreational and Commuters 0.90 Inter-Regional 0.85 L ong Distance 0.85 Long Distance and Recreational 0.80 Highly Recreat i onal 0.75 to provide typical flow patterns for each group. In the grouping process, a hierarchical grouping method was used to compare the characteristics and to match them as closely as possible. AU the sites in the study were classified into seven groups based on the seasonal volume variations. His study then considered daily and hourly volume patterns, and also examined the traffic variations between weekday volumes and Sunday volumes. Trip purpose and trip length information was obtained from past origin-destination surveys in the same sites to verify the proposed master traffic patterns. The two groups of trip purpose including work business trips and recreational trips were used. Work business trips were assumed to remain consistent over the year and recreational trips were assumed to have seasonal variations. Sharma's study used traffic characteristics to determine driver population. This method could be called the indirect measurement because it utilized indicators to reflect driver populations rather than trying to detennine driver population directly from traffic flows. The indirect measurement is more practical and cost-effective compared with traditional direct measurements such as roads ide interviews and license plate surveys. But, the specific techniques used in his study were too complicated for general practitioners, and the assignment of fp values was purely judgmental based albeit logical. In his study, he 14

PAGE 24

used the fp range of0.75 to LOO as included in the HCM. The fp values of0.75 and 1.00 were assigned for highly recreational highways and commuter highways, respectively. The fP values for other highway types were scaled according to this range. Other Related Studies Traffic flow models are fundamental to capacity analyses because these models establish a theoretical base for the understanding of traffic stream characteristics of the real world. The basic findings were documented in the T raffic Flow Fundamentals (May 1990) Speed-volume models were used in the HCM for freeway capacity analyses. According to field observations, the general shape of the speed-volume data curve tended to be linear regardless of the location for LOS A toE within North America (HCM 1994). Schoen eta!. completed a Transportation Research Board project, NCHRP 3-45: Speed Flow Relationships on Basic Freeway Segments (1995). Although they were unable to address the factor of commuter vs. non-commuter traffic, they were able to provide some interesting comparative data on the re l ationships between flow rates and average vehicle speed in four different cities: San Diego, Sacramento, Seattle, and Des Moines. Although these data are unre l ated to the driver population factor, it was interesting to note significant l y different speed flow relationships among the four cities. It is believed that comparisons of commuter facilities and recreational facilities would show similar shifts in the speed-flow curves. Speed-volume curves are used to evaluate external factors that affect roadway capacities by many researchers A recent study was performed by Brilon and Ponzlet (1996) to evaluate the impacts of weather conditions and traffic mix on roadway capacities. They found that under wet pavement conditions, traffic speed was lower as compared with under dry pavement conditions. They also found that driving predominantly leisure traffic, such as Sundays or during the summer vacation season, traffic speed was lower. 15

PAGE 25

CHAPTER 3 : PRINCIPLE USED IN THE PROJECT Definition of Non-Local Drivers The driver population factor is designed to reflect the presence of non-commuters or others unfamil iar with the roadway. There are a number of factors that might affect roadway capacity, including trip purpose, driver age, and trip duration. One of the tasks attempted in this study was to focus on the traffic effect of the driver's familiarity with the road. The "driv er's familiarity with the road" indicates the driver's level of knowledge of the road in question incl uding location of signs and exits and characteristics of the road (what's on the other side of the next overpass and which is the best lane to be in to avoid bottlenecks, for example). lbis variable would be measured on a continuous scale, and ideally would be based on objective measures rather than subjective opinions. Clearly, "familiarity" with the road is an imprecise term. How to measure it is equally imprecise and subject to interpretation. It might, for example, be defined: as "out-of-state" drivers, out-of-county drivers, non -commuters, and other variants. In this report, the term "non local drivers" is used to defme drivers who are not familiar with the freeway sections. Basic Concept The direct impact of non-local drivers is the capacity reduction. If freeway capacities under different non-local driver population levels can be estimated, the impact of non local driver population can, therefore, be assessed. Thus, with known non -local driver population levels, the corresponding estimated capacity reduction could be used to develop driver population adjustment factors. As summarized by Minderhord et al (1997), practically, there are two ways to estimate roadway capacities: the direct empirical and indirect-empirical methods. The direct-empirical method is based on the estimatio n of capacity values at a specific test site using direct traffic observations from the test site. Results obtained from the direct-empirical method can more practically reflect real traffic capacity and level of service conditions. The basic variables to be observed to directly estimate roadway capacity include (I) headway/density, (2) volume, 16

PAGE 26

and (3) speed. Any two of these three variable types should be collected to estimate maximum service flows at different levels of service. Un like the direct-empirical method, the indirect-empirical method is based on guidelines or simulation models such as the Highway Capacity Manual and Highway Capacity Software To estimate capacity by the indirect-empirical method, field observations are not necessary. Traffic and roadway conditions are needed as the inputs to the models. Results obtained through guidelines or simulations may not practically reflect the real roadway traffic service performance such as capacity and leve ls of service. The indirect empirical method is usually used for planning purposes To assess the impact of non-local driver population on freeway capacity, the direct-empirical method is more adequate because the capacity reductions due to different non-local driver population levels can be directly measured. The indirect-empirical method is not able to reach such an accuracy level. In this research study, the direct-empirical method was used. Estimation of Non-Local Driver Population Levels There are two ways to measure non-local driver population levels, namely direct and indirect measurements. I t is difficult to directly measure non local driver population levels using current survey methods such as vehicle license plate surveys and roadside interviews. The result of the vehicle license plate survey is doubtful due to the fact that it is impossible to identify rental cars which are usually driven by tourists. Other limitations of license plate surveys include recognition of customized plates and identification of plates issued by different states. Roadside interviews require stopping the freeway traffic which is almost impossible, and certainly impractical. Besides, in practical applications, transpottation planners and engineers may not be able to directly estimate the percentage leve l of non-local driver population in the traffic stream. To be useful for practitioners, estimates of fp should be made from readily available data parameters. An indirect measurement would use indicators or descriptors to reflect non-local driver population levels. The indirect measurement could be a practical solution to driver population estimation. A feasible indirect measurement of non-local driver population 17

PAGE 27

levels is to use the tourist survey data from the Florida Office of Tourism Research (FOTR). The FOTR is the state s official research unit for tourism studies. The FOTR develops a monthly series of estimates of air and auto visitors. Approximately I 0,000 face-to-face interviews are conducted each year with domestic tourists (US and Canadian). The FOTR also publishes an annual Florida Visitor Srudy report containing a summary of the visitor survey and other Florida tourism-related information from a variety of sources. With the Florida tourist survey data, it is feasible to estimate the monthly numbers of out-of-state visitors in popular tourist destinations, and these numbers could be used to indirectly reflect non local driver population levels in these areas. Because such estimates are based on identified destinations, this method may not generate satisfactory estimation for areas without tourist attractions. Another feasible indirect measurement of non-local driver population levels is the analysis of traffic flow characteristics such as hourly and monthly volume variations which are strongly influenced by driver populations. Unlike the tourist survey data, this measurement could be applied for not only the major tourist destinations, but also the areas with few attractions because the estimation is based on the traffic flow itself. Another advantage of this approach is that traffic characteristics such as speed, volume and classification are monitored continuously in many areas by transportation agencies. Therefore, it is relatively easy to obtain current and accurate data. The two indirect measurements (based on tourist survey data and traffic data) were studied in the project. The driver population adjustment factors were separately estimated. Details can be seen in Chapters 5 and 6, respectively. Principle Non-local drivers may have a certain impact on roadway capacity. The mam characteristics reflecting non-local driver behavior are car-following behavior (headway), gap acceptance behavior (lane change), traffic sign recognition behavior (total reaction time) and vehicle speed. With these combined impacts included in capacity analysis a certain amount of capacity reduction is expected. This concept is shown in F igure 3-1 18

PAGE 28

where the value 6C is the capacity reduction due to no nlocal driver population. Mathematically, the non-local driver population factor can be used to adjust capacity est imatio n by the following equation. where C=fpC* (3) C = capacity under prevailing conditions including non-local driver population, c = capacity under prevailing conditions not including non-local driver population, and fp = non-local driver population adjustment fuctor Figure 3.1: Impact of Non-Local Driver Population. Speed Free A ow Speed j(' / Flow Curve (not including nonlocal driver population impact) _. Aow Curve (including driver population impact) .._. Capacity Reduction Density Line for Level of Service i Traffic Aow Rate Speed-volume models are widely used in capacity analyses. In field studies, the general speed-volume curves tend to be linear regard l ess of locations for the entire range of LOS A to E (HCM 1994). It is practically feasible to use linear models to fit the speed-volume curves in the part of stable flow. Beyond LOS E, traffic flow becomes highly unstable and not able to be estimated by regression models. For a freeway se ctio n under ideal 19

PAGE 29

roadway and traffic conditions, speeds and volumes could be represented by a single linear model for LOS A to E as shown in the following equation: where Speed= a + b x Volume a = free flow speed, km per hour per lane, b =coefficient, Speed = ideal flow speed, km per hour per lane, and Volume= ideal flow volume, passenger car per hour per lane (3-2) Figure 3-2 presents speed-volume data obtained from a FDOT traffic count station located on 1-4 west-bound near Orlando. Based on the data points shown in the figure, a linear model can be used to fit these points. The equation of the model and corresponding R' value are indicated in the figure. Figure 3.2: Typical Relationsbip between Volume and Speed (1-4 WB Middle Lane, March 1995, Orlando). 120 .-------------------, 100 c i'l 80 8. (I) g,o 60 8-40 .., t 20 > < Speed= 104 95 3.786le-3Volume R"2 = 0.378 0 200 400 600 800 1000 i 200 1400 1600 1 800 2000 Vo l ume (pcph) 20

PAGE 30

This linear model could also be applied for traffic streams under prevailing conditions under which fw is negligible for freeways in Florida beeause of good geometric designs. If the percentage of heavy vehicles is known, the volume (vph) ean be adjusted to the passenger-car volume (pcphpl) using PCEs. With other conditions given, for a particular site, the speed-volume curve is mainly affected by driver population characteristics. Actually, the parameter "a" shown in the linear model is free-flow speed which is not affected by non-local driver population levels. However, different monthly speed-volume models could be observed if there is any capacity reduction caused by different monthly driver populations as illustrated in Figure 3-3. These different speed-volume models would have different "b" values which are the results of the impacts of different driver population levels. Thus, the parameter "b" is a function of non-local driver population level. Mathematically: b = f {non-local driver population level} where f{.} is a function fonn. Figure 3-3: Concept of Estimating fp. Speed Free Flow Speed . / ./ Row Curve (not including non8 ... Non-Local Driver At"" Population Level k ......-Non-Loeal Driver Population Level m Vm Vk Vo Traffic Flow Rate 21 (3-3)

PAGE 31

In Figure 3-3, the density line represents a given LOS as defined by the HCM. According to the 1994 HCM the LOS is defined based on the traffic density. The traffic conditions on the same density line should have the same level-of -service. Because the traffic volwnes have already been transferred into the passenger car equivalents using fHV and fw is negligible, the volume should be V in the density line if there is no capacity reduction caused by the non-local driver population. However, if non-local driver population level k is observed, the resulting speed-volwne curve would intersect the density line at volwne V 0 The corresponding driver population adjustment factor, fpk can be estimated according to the following equation: (3-4) If non-local driver population level is further increased to level m, by the same concept, the corresponding driver population adjustment factor fp.. can be estimated by the following equation: f. Vm pm = -v;;(3-5) Conceptually, a group of different non-local driver population levels would result in a gro up of different speed-volume curves. For each specific speed-volume curve, a driver population adjustment factor can be estimated. Thus, for each non-local driver population level, a corresponding driver population adjustment factor can be estimated. 1his conceptual method was used in this study. Calibration of Population Adjustment Factors There are two basic issues that had to be addressed in this study. First, from existing or new traffic data (data type I which inc ludes traffic flow rate, speed, and vehicle classification), capacity reduction due to non-local driver population had to be identified using statistical methods. Consequently, the adjustment factor fP can be estimated under 22

PAGE 32

different conditions. Second, methods needed to be develo ped to esti m ate indices that represent non-lo cal driver population levels (data type II). A statistica l analys i s was conducted to relate the adjustment factor f with the estimat ion of non-local driver p population levels. A wid e range of non-loca l driver population leve l s was needed so that a reasonable calibration off can be reached. Figure 3-4 illustrates the principle involved p i n developing a driver population adjustment factor table. F igu re 3. 4 : Basic Prin ci ple of Developing Driver Population Adjustment Factors. Data Type I D T U ata type f p Estimation ,_ Estimation of Non-Local rDriver Population Levels I Statistical Analysis I Final Products (Tables) The relationships between traffic flow characteristics and driver population could be estimated either by a cross sectional study comparing conditions at a variety of sites, or by a longitudinal study at a particular site, over an extended time period. 23

PAGE 33

CHAPTER 4: RESEARCH DATA SOURCES Traffic Data The FOOT permanent count traffic stations were a valuable source of freeway traffic data in this study These monitoring sites make use of inductive loop detectors. Data are stored in roadside computers and transmitted to the FOOT central computer using a telemetry system, with the data summarized in one hour time intervals. More than I 00 monitoring sites are located throughout Florida, primarily on freeway sections and other principal arterial highways along the State Highway System. A significant number of the monitoring sites include data regarding traffic volume, average vehicle speeds, vehicle classification, and at selected weigh-in-motion sites, information about vehicle weights. The historical traffic data collected at these stations are reported on a directional basis, for each traffic lane summarized in hourly intervals. The historical traffic data collected at these stations in 1995 were the major data resource for this study. There are many count stations along freeways in Florida, however, only ten stations collected more than 200 days of speed, volume, and classification data for the year of 1995. Other sites could not supply sufficient records due to some problems such as freeway construction or equipment failure. The locations of these ten stations are presented in Figure 4-1 and Table 4-1. To evaluate the impacts of the non-local driver population on freeway capacity, only the sites which experienced high traffic volumes were selected for further analysis because high traffic volumes were essential to calibrate speed-volume models. The 1995 computerized traffic data from these ten sites was acquired from the FOOT. These original data files were named according to the site, character (speed, volume, or classification), traffic direction and date. The original database provided by FOOT had a format that could not be used directly for analysis. A computer program using SAS programming was developed to convert the original database into a readable format ready for data analysis. Meanwhile, the vehicle classification data were employed to convert 24

PAGE 34

Figure 4-1 : Locations ofFDOT Traffic Count Stations. 25

PAGE 35

T a b l e 4 1 : FDOT Traffic CouDt S tatloDs &Dd AADT. Site Freeway Coun t y AAN 0130 1-4 110,68 9 0132 I-9 5 Nassau 40.099 0134 1-95 Brevard 27,027 0171 1-95 Duval 87,917 0174 1 -95 Palm Beach 126.988 0179 1-4 Vol usia 54,300 0184 1 -75 Lee 27,449 0216 1 -295 Duval 30,754 0224 I-75 34,904 0292 1 -95 Aal!er N/A traffic volwnes to the equivalent passenger car volwnes with the passenger car equivalent fact ors s uppl ied in the Highway C a pacity M anual. To meet th e requirement of Metric Syste m the speed data ( mph) was conv erted to the M etric System uni t (kmlh). In order t o e limina t e unu s ual traffic data suc h as traffic inc idents, lane closure, worl
PAGE 36

monthly updates of their tourist swveys which were composed of air-visitor swveys and auto-visitor surveys. Approximately I 0,000 person-to-person interviews were conducted with out-of-state visitors (US and Canadian) each year by the Office of Tourism Research. These visitors must have been in the state for at least one night and no more than 180 nights to be classified as visitors. Commuters were not included in the surveys. Swveys of air travelers were conducted in airport departure lounges for I!Ommercial flights leaving Florida from thirteen major airports twice each month. Auto visitor surveys were made on 27 roads near the Florida border each month. Only out-of-state visitors were defined as tourists (vis i tors) in the swveys. On Interstate Highways (1-95, I I 0 and l-75), swveys were conducted at the freeway rest areas closest to the Florida border. Traffic was stopped on all other highways to interview visitors. Visitor characteristics such as destinations, number of persons in a travel party, rental car usage, vehicle occupancies and days of stays were recorded in the swveys. In this study, the visitor-day was defined as the total number of days spent by a visitor who had used a car in these areas during his/her stay. The definition of "the visitor who had used a car" could be either an auto visitor or an air visitor who rented a car. The visitor-day was established to better measure the impacts of visitors. It also changed the number of visitors into the number of days they spent which should be more appropriate in evaluations of these drivers presence in traffic flows The "visitor-day" is defined by the following equation: visitor-day= number of visitors x total days of stay Estimation of Non-Local Driver Population Levels Using Tourist Survey Data (4-1) The tourist survey database was used to estimate non-local driver population levels. The statewide tourist infonnation and the swvey database were used in the estimation of monthly non-local driver population levels The following equation was used to estimate the total number of visitor-days in an ar e a: 27

PAGE 37

where ITVD = TVxSVD SVTxSTP TTVD -total number of visitor-vehicle days in an area, TV total number of visitor s in Florida, SVD total number of visitor-days in an area in the survey database, SVT total number of visitors in an area, and (4 -2) STP average number of people in a travel party in Florida in the survey database In the equation, SVD is the visitor-days in the area. SVD was computed according to the annual survey database; SVT is the numbers of visitors in the area according to the annual survey data. Thus, the ratio of SVD to SVT represents the average number of days per visitor spent in the area according to the annual survey data TV is the actual total visitor numbers in Florida. The ratio of SVD to SVT is multiplied by TV to obtain the total days visitors spent in the area. The final step is to divide the total visitor-days by the average number of people in a travel party to convert the visitor days to visitor-vehicle days. Visitor-vehicle-days should be more appropriate in this study because this concept focused on vehicles rather than visitors. An assumption was made that one travel party would be in the same car. Results obtained from the tourist survey database are the estimations of absolute total number of visitor-vehicle-days. Direct use of such estimates cannot provide meaningful information on the non-local driver population level in traffic stream. As stated previously, an indirect measurement of non-local driver population l evel may be a practically useful way. What might be interesting to transportation practitioners is the relative monthly distribution of non-local driver population among the 12 months each year for a particular location. For example, if the tourist survey shows that the Orlando area attracts more tourists in March than in April, it could be reasonably assumed that the non-local driver population level in the traffic stream on 1-4 near Orlando during March would be more than in April. Thus the driver population adjustment factor used for March traffic should be smaller than for April traffic. 28

PAGE 38

In the study, the estimate of non-local driver population levels by area and by month developed from the tourist surveys was used as a proxy for direct observations of vehicles in order to indirectly est i mate the non-local driver population leve l s in the traffic stream Based on the 1995 survey database of the Office of Tourism Research, a mon thly non lo cal driver index (DI) was used to estimate the relative non-local driver population level at the test location for each month The monthly non-local driver index is defined as the ratio of monthly non-local vehicle-days over yearly average non-local vehicle days. A larger index value means more non-local drivers in the corresponding month as compared with other months. Table 4-2 shows the calculated TTVD and DI values for the Orlando area. Table 4-2: TTVD and DI Values for the Orlando Area. Month TIVD DI Air Visitor AutoVisitor Total Jan 441,722 993,538 1,435,280 Ll84 Feb 617,789 709,664 1,327,453 1.093 Mar. 776,313 791,576 1.567,889 1.293 Ap r. 803,958 438,406 1 ,242 364 1.028 May 492. 975 454,157 9 47 132 0 780 June 440,376 789,233 1 ,229,609 1.015 July 607,459 769,015 1,376,474 1.134 Aug. 545,736 546,093 1.091,829 0.900 Sep. 605,788 432,45 1 1,038,239 0.856 Oct. 461,462 553,018 1 ,014 .480 0 837 Nov. 424,476 702,781 1.127 257 0.930 Dec. 500,936 653,999 1.154.935 0 .952 Total 6,718,990 7,833,931 14,55 2. 921 The estimation of non-local driver population leve l s is based on the monthly data. The monthly estimation is appropriate in Florida because tourists have distinctive variations among different months. The monthly visitor-vehicle-day variations as shown in F igure 4 -2 confirmed this assumption and also sugg ested that the corr e spondent calibration of capacity reductions should also be based on the monthly data. 29

PAGE 39

Fi gure 4-2: D l V alu es f o r Diff e r ent Moaths in Orlando Area 1.5 1.4 1.3 1.2 1.1 -Cl 1.0 0.9 o.a 0.7 0.6 0 5 1 2 3 4 5 6 7 a 9 1 0 1 1 12 Month Estim a tion o f N o n -Local Drive r Populatio n Leve l s Usin g Traffic Da t a Another method for indirect measurement of non-local driver popula tion levels is to look at the traffic volume itself. Traffic data is rel ati vely e asier 1o obtain than the tourist data because FDOT bas the full access to traffic count stations located on Florida freeways. Traffic vol ume s vary at different locations and among different months in the same si te. It is a lso true that traffi c volum e s vary according to different hours and days of the week. As evidenced by the literature reviews, particularly the work by Sharma (1986), the volume variations have been use d to estimate traffic characteristics. It is reas
PAGE 40

In this study, some indices were established to measure the traffic variations caused by different driver population levels These ind i ces should correlate with volume variations and driver population levels, and they also should have specific definitions and clear calculation procedures so that they could be easily used by general practitioners. Three different volume indices were established i n this study, namely monthly factor (MF), weekly facto r (WF), and daily factor (OF). MF is the ratio of the monthly average daily traffic (MADn to the annual average daily traffic (AADn It is believed that a major portion of the additional traffic observed during peak seasons is made up of nonlocal drivers or tourists WF is the ratio of monthly average Sunday traffic (V """) to monthly average weekday traffic (V W
PAGE 41

where VJ-2pm DF= V7-8am (4) MF, WF, OF= monthly, weekly, and daily factors, respectively, MADT = monthly average daily traffic, AADT = annual average daily traffic, V = monthly average Sunday traffic, V,.. =monthly average weekday (Tuesday, Wednesday and Thursday) traffic, v,.,,.. = monthly average weekday afternoon non-peak hour (lpm-2pm) traffic, and V '"'"' = monthly average weekday morning peak hour (7am-8am) traffic. The detailed calculation results are shown in Table 4-3 and graphically presented in Figures 4-3, 4-4, and 4-5. In Tab le 4-3, MF and WF were based on daily directional volumes, not hour ly lane volumes. According to their definitions, higher MF, WF, and DF could be the results of the higher level of non-local driver population. Figure 4-3: MF Values for Different Months in Orlando Area (WB and Lane 1 only). 1.20.-----------------------, 1.15 1.10 1.05 1.00 0 95 0.90 0.85 32

PAGE 42

1<. :3: "-Cl Figure 4-4: WF Values for Different Months in Orlando Area (WB and Lane 1 Only). us 1 10 1.05 1.00 0.95 0.90 0 .85 0.80 0.75 1 2 3 4 5 6 7 8 9 10 1 1 Month Figure 4-S: DF Values for Different Montbs in Orlando Area (WB and Lane l Only). 1.00 0.95 0.90 0.85 0.80 0.75 0 .70 0 .6 5 0.60 1 2 3 4 5 6 7 a 9 10 Month 33 1 1 1 2 12

PAGE 43

Tabl e 4-3: MF WF, and DF Values. Site Direction Month Lane MF WF OF 130 w 1 1 0 929 8 4 0 87713 0.82116 130 w 2 1 0 99197 0.92417 0 83777 130 w 3 1 1 01999 0 92856 0 84266 130 w 4 1 I 1 01888 0.89151 0 8 4866 130 w 5 1 0 96647 0. 94017 0 80686 130 w 6 1 1 03103 0.8674 2 0 83792 13 0 w 7 1 1.06808 0.90 814 0 87267 13 0 w 8 1 I 1 .03665 0.92748 0 84886 130 w 9 1 0 96248 0 90372 0 80896 130 w 10 1 0.99061 0 91239 0 83316 13 0 w 11 1 0.98523 0 86452 0 82278 130 w 12 1 1.00435 0 84002 0 87392 13o I w 1 2 0 92984 0 87713 0 75775 130 I w 2 2 I 0.9919 7 0 92417 0 77818 130 I w 3 2 I 1.01999 0 92856 0 7 8659 130 w 4 2 1 01888 0.89151 0 78458 130 w 5 2 0 96647 0 94017 0 71334 130 w 6 2 I 1 03103 0 66742 0 76139 130 w 7 2 I 1 06808 0 90814 0 77131 130 w 8 2 I 1.03665 0.92748 0.73688 13Q w 9 2 0.96248 0 90372 0.72093 130 w 10 2 0.99061 0 91239 0 .73 894 130 w 11 2 0.98523 0.86452 0 75193 130 w 1 2 2 I 1.00435 0 84002 0 78957 130 w 1 3 I 0 92984 0 87713 0 49035 130 w 2 3 I 0 99197 0 92417 0 49444 130 w 3 3 I 1 01999 0 92856 0 53370 130 w 4 3 I 1 01888 0 89151 0 52210 130 w 5 3 I 0 96647 0 94017 0 46381 130 w 6 3 I 1 03103 0.86742 0 50709 130 w 7 3 I 1 06808 0.90 814 0 55132 130 I w 8 3 I 1 03665 0 92748 0 51942 130 w 9 3 I 0 96248 0 90372 0 .43 133 130 w 10 3 I 0.99061 0 91239 0 46665 130 w 11 3 I 0.98523 0.86452 0.49413 1 30 I w 12 3 I 1 00435 0 84 0 02 0 56250 (conttnued on n e xt page) 34

PAGE 44

Table 4-3: MF, WF, and DF V alues (continued). Site Direction Month Lane MF WF OF 171 N 1 1 0 93851 0.51433 0.71924 171 N 2 1 0 98652 0.57342 0.722 44 171 N 3 1 1 07692 0.63853 0 78575 171 N 4 1 1.05429 0.61750 0.76084 171 N 5 1 1 01168 0 55605 0.73405 171 N 6 1 1 03375 0.55655 0.75432 171 N 7 1 0.99447 0.6025 1 0 80143 1 7 1 N 8 1 1 02060 0 57265 0.73567 171 N 9 1 0 .95916 0.54995 0.70563 171 N 10 1 0 96073 0 55608 0.70205 171 N 11 1 0 97215 0.52639 0.75176 171 N 12 1 0 96407 0 54490 0 .7 8873 171 N 1 2 0 93851 0.51 433 0 56771 171 N 2 2 0 98652 0.57342 0 59595 171 N 3 2 1.07692 0 63853 0.69436 1'71 N 4 2 1 05429 0 61750 0 67175 171 N 5 2 1 01168 0 55605 0 61759 171 N 6 2 1 03375 0 55655 0 62451 171 N 7 2 0 99447 0 60251 0.68129 17 1 N 8 2 1.02060 0.57265 0.61638 171 N 9 2 0.95916 0.5499 5 0.58207 171 N 10 2 0 96073 0.55608 0 58827 174 N 1 1 1.01433 0.64691 0.59672 174 N 2 1 1.07098 0 70993 0.59860 174 N 3 1 1 07834 0.71917 0.65017 174 N 4 1 1 04351 0 69725 0.58864 174 N 5 1 0 99870 0 67247 0 57560 174 N 6 1 0 99361 0.65349 0 59452 174 N 7 1 0 95242 0 69002 0 62706 174 N 8 1 0 94794 0.72865 0.59550 174 N 9 1 0 96124 0.62162 0.55585 174 N 10 1 0 96181 0 69095 0 57744 174 N 1 2 1.01433 0 64691 0 88448 174 N 2 2 1 07098 0.70993 0.89579 174 N 3 2 1 07834 0 71917 0 91554 174 N 4 2 1 04351 0.69725 0 84163 174 N 5 2 0 99870 0 67247 0.85076 174 N 6 2 0 .99361 0 65349 0.89007 174 N 7 2 0 95242 0.690 0 2 0.94062 174 N 8 2 0 94794 0.72865 I 0 87399 174 N 9 2 0 96124 0 62162 I 0 87650 174 N 10 2 0 9618 1 0 69095 0 86368 17 4 N 1 1 3 1 01433 0.64691 1 .07 869 (conttnued on next page) 35

PAGE 45

Tabl e 4 -3: MF, WF, aod DF Valu es (eoo tioued ) Site Direction Month Lane MF WF OF 174 N 2 3 1.07098 0.709 93 1.11648 174 N 3 3 1.07834 0 71917 1.08308 174 N 4 3 1.04351 0.69725 1.01571 174 N 5 3 0.99870 0.67247 1.09741 174 N 6 3 0.99361 0 65349 1 1043 4 174 N 7 3 0.95242 0.69002 1 15262 174 N 8 3 0 94794 0 72865 1 .099<41 174 N 9 3 0 .96124 0 62162 1 06768 174 N 10 3 0 .96181 0.6909 5 1 078 74 174 s 1 1 0 95063 0.67372 0 70013 174 s 2 1 1 07071 0.74 638 0 67767 174 s 3 1 1.09665 0 76384 0 77191 174 s 8 1 0.94147 0 74025 0 84010 174 s 9 1 0.86861 0 64747 0 81804 174 s 11 1 1 05581 0. 70305 0.96001 174 s 12 1 1 01383 0.68 973 0.91887 174 s 1 2 0.95063 0 67372 0 84099 17 4 s 2 2 1.07071 0.7 4638 0 76833 174 s 3 2 1.09665 0.76384 0.7 4 860 174 s 8 2 0.94147 0 74025 0 86793 174 s 9 2 0.86861 0 64747 0 84363 174 s 11 2 1.05581 0 70305 1 12907 174 s 12 2 1.01383 0.6897 3 0.99429 174 s 1 3 0.95063 0 67372 0 82355 174 s 2 3 1 07071 0 74638 0.81322 174 s 3 3 1 09665 0 76384 0 .7 0444 174 s 8 3 0 94147 0 74025 0.69634 174 s 9 3 0.86861 0 64747 0 65268 17 4 s 11 3 1.05581 0 70305 0 90603 174 s 12 3 1.01383 ) 68973 0 84565 36

PAGE 46

CHAPTER 5: DEVELOPMENT OF DRIVER POPULATION ADJUSTMENT FACTOR TABLE BASED ON TOURIST SURVEY DATA The purpose of the research effort swnmarized in this chapter was to evaluate the impact of non-local driver population on freeway traffic capacity. Based on the results of the evaluation, a driver population adjustment factor table was de v eloped. Since the 1995 traffic database from FOOT traffic count stations included 12-month traffic data and the survey database from the Office of Tourism Research provided information to estimate monthly non-local driver indices which represented the relative non-local driver population level for each month during 1995, the basic principle employed in the research was to compare the volume-speed curve differences among the 12 months Tourist survey data and traffic count data were reviewed before further statistical analysis was performed. It was found that some count stations did not show high traffic volumes and some areas did not present good tourist survey results. In the effort to develop driver population adjustment factors based on the Office of Tourist Research data, the Orlando area was selected because good tourist survey results were available and the whole year traffic counts and high traffic volumes were included in the database. Conceptually, the data set which covers one area is enough for the study purpose if the data set covers all possible tourist seasons and all possible peak and non-peak seasons. Modeling Procedure and Results As stated in Chapter 3, a speed-volume curve under the condition of stable traffic can be described by a linear equation. Practically, a driver population adjustment factor can be applied only if the traffic is stable. The factor is meaningless if forced-flow traffic exists. Therefore, to analyze the impact of non-local driver population on traffic capacity, the linear equation can be used to represent speed-volume curves. To further confirm this assumption, three speed-volume data sets were randomly selected from the database and are shown in Figures 5-1 5-2, and 5-3. From these figures, it can be concluded that a 37

PAGE 47

-"" "8 !:L "' "" = --e & 0 "" .. -!I < Figure 51: R e lation ship b e tw een Volum e and Spee d (1-4 WB Laue 1 January 199 5 Orlando, Site: 0130 ). Speed = 96.04 4.3e 3Volume R"2 = 0.470 0 200 4 00 600 800 I 000 1200 1 400 1600 1 800 2000 120 100 80 60 r 40 r 20 Vo l ume (pcph) Fig ure 5 2 : Rela tio n ship b e tween Volum e and Spee d (1-4 WB Laue 2, Oct ober 1 99 5 Orland o, Site: 01 3 0) Speed= 104.752 9 e -3Vo 1 ume R"2 = 0.2 1 5 om 400 600 o Volume (pcp h ) 38

PAGE 48

20 < Figure S-3: Relationship between Volume and Speed (l-4 WB Lane 3, March 1995, Orlando, Site: 0130) Speed= 108 .19 3.300e-3Volume R"2 = 0.195 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Volume (pcph) model is adequate to represent the speed-volwne relationship under the condition of stable traffic flow. Mathematically, the linear model is represented by: Speed =a+ b Volwne (5-1) where "a" and "b" are parameters to be estimated by the linear regression method. The units of speed and volwne are kmlh and pcphpl, respectively. Table 5-1 presents the regression analysis results for the traffic data collected from west bound (WB) direction ofl-4 at Orlando, where many tourists from other states and countries visit Disney World and other attractions. A significant percentage of non-local driver population has been observed during heavy tourist seasons at this sites. In the same table, the corresponding 01 values defined in Chapter 4 are also presented. 01 was defined as the ratio of monthly non-local vehicle-days over yearly average non-local vehicle-days. As stated in Chapter 3, the parameter "a" represents the free-flow speed, and the param eter "b'' reflects the impact of non-local driver population if other conditions are given or fixed. In fact, for a particular lane the parameter "a" did not change significantly during 1995 i.e._, the tourist 39

PAGE 49

seasons did not affect the free-flow speed. However, according to Table 5-l, the free-flow speed (parameter "a") on Lane 3 (insi de lane) was higher than the free-flow speed on Lane 2 (middl e lane), and the free-flow speed on Lane 2 was higher than the free-flow speed on Lane I (outside lane). Usually, the inside lane is considered a high speed lane as compared with the outside lane The differences of "a" values for Lanes I, 2, and 3, therefore, were expected and reasonable. The parameter "b" is the slope of the speed vo lume curve. If other conditions are fixed, as the non-local driver population increases, the absolute value of the parameter "b" should sta tistically ge t larger although the sign is negative. In order to fmd the relationship between the parameter "b" and non-local driver population, the Orlando monthly non-local driver indices presented in Table 51 correla t ed linearly with the corresponding parameter "b" values Results are shown in Figure S-4. To simplify the description of the modeling process, traffic data o f the west bound (WB) are analyzed first in this section. The final results including the data of both directions will be presented in the following section. According to Figure S-4, Table 5.1: Regression Analysis Results (a & bin Eq. 5) (1-4, WB, Orl ando Site: 0130) and Monthly Non-Loca l D r iver Indices (DI) (Orlando) Lane I (outside) Lane 2 (middle) Lane 3 (inside) D I Month a b a b a b Jon. 96.04 .0043 105.12 0.0038 108.52 .0030 1.184 Feb 96.19 0 0045 104. 88 .0040 108.28 -0.0033 1.093 Mar. 96.69 0045 104. 95 -0 .0038 108.19 -0.0033 1.293 Apr 95.90 0036 103. 99 .() .0024 107. 66 -0. 0015 1.028 May 95 60 .0028 103.93 -0 0018 107. 25 .().0002 0.780 June 96.15 0.0044 104 10 -0 0027 107.50 -0.0015 1.015 July 95.57 -0.0046 103. 96 0032 107.13 0.0019 1.134 Aug. 95.22 .0045 103.90 0029 107.01 0.0019 0.900 Sep. 95.49 -0.0034 104.01 .0021 107.Q3 -0 .0004 0.856 Oc<. 96.44 -0.0043 104.75 -0.0029 107.18 -0.0021 0.837 Nov. 97.50 -0.0040 106.09 .0031 109.34 0.0017 0.930 Dec. 97.72 .0049 106.43 .().0036 109.75 .0023 0.952 40

PAGE 50

Figure S -4 : Statistical R e lati onshi p betw e e n Parameter b and Montbly Non-Local D r iver Index (DI) (1-4 WB, Orlando, Site: 01 3 0). 0.000 -0. 0 01 ..().002 :0 -0.00 3 <> -., E :: {;,! 0.004 0 .005 0 .006 0.007 0.7 Lane I o Lane2 Lane3 0.8 b = 2.1076e 2 042le-3 DI R"2 = 0.275 b = 2.9630e 4 3.32 0 7e-3 DI R"2 = 0.531 b = 2.9859e-3 4.9 1 0 l e-3 DI R"2 = 0 .580 0. 9 1.0 1.1 1.2 1.3 Mo n th ly !'
PAGE 51

the roadway or they were not commuters. This category of drivers might choose the middle lane too. However fewer non local drivers would choose the i nside lane with higher operating speeds, as compared with outside and middle lanes. This indicates that non-local driver population may not distribute uniformly among lanes and the impacts of non-local driver population in different lanes may be different. Although a linear equation was used in Figure 5-4, the rate of change of the parameter "b" with respect to DI may not be constant in practical cases. This rate is relatively smal ler when DI is small than when DI is large. A second-order polynomial mode l could be used to represent such a statistical relationship between "b" and 01. Meanwhile, to represent statistically the impact of non-local driver population as a whole, the monthly volume-speed curves from Lanes I, 2, and 3 were combined without distinguishing lanes. Mathematically B(i) = (5-2) and A(i) = al(i)+a2ji)+ a3(i) (5-3) where bl(i), b2(i), and b3(i) are the parameter "b" values for month i (i = I, 2, .. 12), representing Lanes I, 2, and 3, respectively, and al(i), a2(i), and a3(i) are the parameter "a" values for month i, representing Lanes I, 2, and 3, respectively. In fact, A(i) is monthly free-flow speed of the WB traffic and is not affected by non-local driver population. Therefore, an average value of A(i) (i =1, 2, ... 12) was used in this study, or 12 A= j;:;} Based on Table 4-1 the value of A was calculated; A= !03 km/hour. 42

PAGE 52

To evaluate correctly the impacts of non-local drive r population on capacity, the param eter "b" under the condition of "zero non-local driver population in the traffic stream (Dl = 0)" was estimated by using traffic data collected from 5:00 am to 8:00 am during the whole year. Pract ically, drivers during this time period are almost always comm uter s or regular users. The "b" values obtained in this way were used to generate the fining equation. As stated previously, the monthly volume-speed curve slope B(i) (i =I, 2, . 12) is a function ofDI value A second order polynomial model was used to fit the relationship between B(i) and DI values, including Dl = 0. The relationship between B(i) and correspo n ding DI values are shown in Figure 5-5. Based on Table 4-1 and curve fitting method, the monthly volume-speed curve slope of the WB directional traffic at test site 0130 could be estimated from DI values and represented by: (5-4) Figure 5-S: Statistital Relation ship between the Average Value of Parameter "b" and Monthly Non-Lotal Driver Index (DI) (with 2"-order polynomia l curve). 0.000 .J:> -0.001 -" iS -0.002 E ., - rf. -0.00 3 .... 0 .. "' -;;; -0.004 > 0 01) e -0 .005 r-< B = -7.94e-4-1.23e-3DI9.93e-4Dl'2 R"2 = 0.722 lQ -0.006 r--0.007 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Month l y NonL ocal Driver Index (DI) 43

PAGE 53

with a correlation coefficientof0.122 (R2 = 0 122) The final equation to characterize the average impact of non-local driver population on the freeway segment (all lanes) capacity can be expressed as follows: A+ B Volume (5-5) or Based on Equation (5-6), with other prevailing conditions fixed, average operating speed is a function of non-local driver population level and average lane volum e. As non-local driver population increases, the average operating speed decreases, resultin g in decreased capacity under prevailing conditions. Figure S-6 presents a group of speed-volume curves with each curve representing a particular monthly non-local driver index. nus figure, with data collected from 1-4 West Bound in the Orlando area, is based on Equation (S-6). The impact of non-local driver population can be assessed graphically by reviewing this figure. Figure 5.6: Impact of Non-Local Driver Populatio n on Avenge Operating Speed (1-4 WB, Orlando, Site: 0130). 105 01:0.0 01=.6 01=.8 -.......:: t;::; 01=1.0 01=1.2 I-.... '""- 01=1.4 r---..--..._........, 01=1.6 r----..: 01=1.8 ........... 0 01=2.0 01=2.2 Dl=2.4 80 0 200 Volume (pcph) 44

PAGE 54

More Generalized Results The same procedure described previously was applied to the East Bound lanes of 1-4 located in Orlando to obtain more general speed-volume curves from the impacts of the non-local driver population. In fact, for each direction of interstate freeway at each location, a gro up of speed-volume curves similar to those in Figure 5-6 can be obtained to represent the non -l ocal driver populatio n impacts on average lane capacity. According to the study results, the difference between the two groups of speed-volume curves of both direction s was very small and can be even neg lected In this study these curve groups resulting from each direction of l-4 in Orlando were averaged. The final equation representing speed-volume curves with the impac t of non-local driver population is presented below: Speed= 102 + ( -6.82xl0-4 -1.75xl0-3xDI-8 .99xl0-4xDI2 ) Volume (5 -7 ) This equation represents a general situation, because the data used to develop the above equation were collected from both directions of 1-4 at Orlando. Consequently, Figure 5 7 presents a group of s peed-volume curves developed from the data. !OS Figure 5-7: Impact o f Non-Local Driver Population on Average Operating Speed (1-4 Both Di rections, Orlando, Site: 0130). DI=O.O ] 100 Dl=.6 DI=.8 1 95 < 85 b '---...., /"--.._.......__, 0 ............. ---.. t'-----............. 200 400 600 800100012001400160018002000 Volume (pcph) 45 0 DI=l.O DI=1. 2 DI=l.4 Dl=1.6 DI=l.8 Dl=2.0 Dl=2.2 Dl=2.4

PAGE 55

Estimate of Driver Population Factors Based on Test Site 0130 According to 1994 HCM, freeway LOS is based on vehicle density. Under ideal conditions, average vehicle operating speed s on basic freeway sections are rel atively insensitive to traffic flow rates up to LOS E. Ho wever, under prevailing conditions, average operating speeds do decrease as traffic flow rates increase. This is true particularly when a significant population of non-local drivers exists in the traffic stream. To quantify the impacts of non-local driver population on freeway capacity, average s ervice flow rates at LOS B, C, and 0 were assessed in this study. Based on the definition and method of defining LOS on the speed-volume curve presented in 1994 HCM a line starting from the origin and crossing volume-speed curves can be used to fmd corresponding vo l umes with given level of service Figure 5-8 presents the volume-speed curves based on Figure 5-7 and one density lines representing LOS C. The density lines representing LOS B and 0 can also be plotted in the same figure with the same way. Because these curves were obtained from the data collected at the given test site, the differe nces between these curves can be related to different monthly non -l ocal driver 105 Figure 5-8: Estimation o f Driver Population Adjustment Factors (1-4 Both Directions, Orlando, Site: 0130). .......; :--.. ...., --......._..........., ...... ...., ................. r-.... 'IS N (:i4 l\1' :-----...., .........., 80 0 200 400 600 800 1000 1200 1400 1600 1800 200 . . ...... ...... 0 Volume (pcph) 46 01=0.0 01=.6 01=.8 01=1.0 01=1.2 01=1.4 01=1.6 01=1.8 01=2 0 01=2.2 01=2.4 LOSC

PAGE 56

indices. The corresponding maximum service volumes along this density line for different DI values can be estimated. According to these service volumes, the corr esponding driver population factors can be calculated. Table 5-2 presents the corresponding maximum serv ice volumes for LOS B C, and 0 and different 01 valu es. According to the definition discussed in Chapter 3, fp values for different LOS and 01 values can be ealculated and are presented in Table 5-3. Table S-2: Maximum Service Volumes (pcphpl) for LOS B, C, and D and Different Dl Values. DI LEVEL OF SERVICE (LOS) LOSB LOSC LOSO 0.0 1002 1496 1990 0.6 988 1468 1940 0.8 982 1455 1915 1.0 976 1440 1892 1.2 969 1426 1865 1.4 962 1409 1837 1.6 953 1392 1808 1.8 945 1374 1777 2.0 936 1355 1746 2.2 926 1336 1714 2.4 916 1315 1680 Discussion The fp table (Table 5-3) can be used by transportation practitioners to replace the fp table shown in Basic Freeway Chapter of the HCM (also shown in Table 1-1 in this report) for the purpose of driver population adjustment. To use Table 5-3 DI values based on tourist surve y for major areas should be obtained. The annual tourist survey database is available to public A procedure can be developed to pre-process the database to obtain 01 values for major areas in Florida 47

PAGE 57

Table S-3: Driver Population Adjustment Factors for Different Levels of Service, Based on Orlando Conditions DI LEVEL OF SERVICE (LOS) Average LOSB LOSC LOSD 0.6 0.986 0.981 0.975 0 981 0 8 0.980 0.973 0.962 0 .972 1.0 0.974 0.963 0.951 0.963 1.2 0.967 0.953 0.937 0.952 1.4 0 960 0 942 0.923 0.942 1.6 0.951 0.930 0.909 0.930 1.8 0.943 0.918 0.893 0.918 2.0 0.934 0.906 0.877 0.906 2.2 0.924 0 893 0.861 0 .893 2.4 0.914 0.879 0.844 0.879 The results presented in this chapter are applicable to the areas with major tourist attractions. For a particular area with DI value available, Table 5-3 can be used to find the driver population adjustment factor to adjust freeway capacity in the area. If there is no DI value for the particular area, DI value for au area with similar social/economic aod tourist attraction scopes to that area could be used to adjust freeway capacity at that area. The results might not be applicable to the areas with significant through traffic but few tourist attractions because the areas with significant through traffic may not result in adequate DI estimations Only the areas with tourist traffic origins or destinations could be considered for the application of the results presented in this chapter. From Table 5-3, it can be concluded that non-local driver population may have more impacts on the traffic stream with worse LOS as comparid with better LOS In fact, as traffic density i ncre ases, the interactions between vehicles may be more sensitive. However, practically, the average fP values shown in Table 5-3 can be used without distinguishing LOS. 48

PAGE 58

CHAPTER 6: DEVELOPMENT OF A DRIVER POPULATION ADJUSTMENT FACTOR TABLE BASED ON TRAFFIC CHARACTERISTICS Infonnation on relative non-local driver population levels may be inferred from traffic characteristics such as traffic variations among different months, different days, or different hours. The basic assumption is that commuter traffic or regular local users of a freeway section are relatively consistent and that monthly, daily, and hourly variations in traffic are somewhat related to non local drivers. Based on such an assumption, the monthly factor, weekly factor, and daily factor were developed in the study to indirectly represent the relative non-local driver population levels in each month. The main benefit of using these factors is that it is unnecessary to directly identifY non-local driver population levels, a fonnidable task. Traffic data collected through traffic count stations can be used to indirectly represent non-local driver population level. As stated in Chapter 4, due to limitations associated with other traffic count stations, only three sites were selected for this study. Two primary criteria limited the sites suitable for analysis: (I) a requirement that at least 200 days of "good" data be available, and (2) since the methodology was based on calibrating localized speed-volume curves it was necessary that high traffic volumes, in the LOS C,D, and E ranges be observable. The three sites were 1-4 in Orlando (site 0130) 1-95 in West Palm Beach (site 0174), and 1-95 in Jacksonville (site 0171). Data from these sites cover almost the whole year traffic, representing typical tourist seasons. Monthly Factors, Weekly Factors, and Daily Factors Monthly factor Ideally, if there were no non-local drivers in the traffic stream, monthly average daily traffic for each month would show modest variations, reflective of school closings and family vacation schedules. However, since non-local driver population levels in different 49

PAGE 59

months are quite different, monthly average daily traffic is also quite different Conceptually, the following equation can be used to describe this difference: where MADTi = Vi + Ui (6-1) MADTi =monthly average daily traffic for month i (i = I, 2, . 12) Vi= proportion of MADTi for regular users or commuters for month i (i = I, 2, ... 12) Ui = proportion of MADTi for non-local driver population for month i (i = I, 2, ... 12) To analyze the relative non-local driver population levels in different months, Vi could be assumed constant across the 12 months, or Vi = V (i=l,2, ... ,12) (6-2) where V is a constant. Annual average daily traffic (AAD1) can be calculated by the following equation: 12 12 12 12 AADT= AIMADTi = -h
PAGE 60

nonlocal driver population leveL Figure 6-1 shows the monthly factors for eacb month at each of the three sites. Figure 6-1: Monthly Factors of Different Monlbs at Site s 0130, 0171, and 0174. 0 -<.> >. .c c 0 ::E 1.20 1.15 1.10 1.05 1.00 0.95 0.90 0.85 0.80 0.75 0 / I v t I 2 3 4 Site 0130 V/ ......... / v f"._ Site 0:11'\. / Site 0174 5 6 7 8 9 10 II 12 Month Weekl y factor Conceptually, as stated in Chapter 4, the weekly factor (WF) is defined as the ratio of monthly average Sunday traffic (MASl) over mon thl y average wee kday traffic (MAW!), or where WFl. = MASTi MAWTi (i=l,2, ... ,12) WFi =weekly factor for m on th i (i =I, 2, ... 12), MASTi = monthly averag e Sunday traffic for month i (i = I, 2, ... 12), and MA WTi monthly average weekday traffic for month i (i = I, 2, ... 12). (6-5) It was assumed that more non-local drivers in an area might result in more Sunday traffic or larger weekly factor as compared with the area with less non-local drivers. However, 51

PAGE 61

some areas s uch as O rlando m a y attrac t non-local drivers during the whole we e k including Sunday an d w e e k days in which case more non -l ocal driver s may not neces s arily result in a larger weekl y f acto r. F igure 6 2 presents weekly factors for each month at the three sites. Figure 6-2 : Weekly Factors of Diff erent Months at Sites 0 130,0171, and 01 74. 0 u IZ ,., .:>< 8 1.10 1.00 0.90 0.80 0 .70 0 .60 0.50 0.40 0.30 0 D aily factor v v I 2 / ........ / 3 4 5 Site 0130 ......... ...... ....... "' / Site 0174-/ S i te 0171 6 7 8 9 1 0 I I 12 Month The daily factor (D F ) is defmed as the ratio of monthly average weekday afternoon non peak traffic (I :00 pm 2:00 pm, Monday, Tuesday, Wednesday and Thursday only) to monthly average weekday morning rushh our traffic (7:00 am 8:00 am, Monday, Tue sday, Wednesday, and Thursday only), o r v DF 1 1 2pm v 17 -Sam (i = 1 2, .. 12) (6-6) 52

PAGE 62

where DFi factor for month i (i 2, ... 12), Vi1 2pm =monthly average weekday afternoon non-peak traffic for month i ( i = I 2 ... 12), and Vi7 8am =monthly average weekday morning rush-hour traffic for month i (i = 1, 2, ... 12). Generally, morning rush-hour traffic consists of mostly commuters and regular user traffic. However during the afternoon non-peak time, the traffic has a h.igher proportion of non-local driver traffic. Higher non-local driver population levels would result in larger daily factors. The daily factor may be used as an indirect measure of non-loca l driver population level. Figure 6-3 depicts daily factors for each month at the three sites Figure 6-3: Daily Factors of Different Months at Sites 0130 0171, and 0174. 1.10 1.00 0.90 0.80 0 .70 0 .60 0.50 0.40 0.30 0 ,/ I 2 3 4 5 53 .,... ......... Sice 00174 Sice Sice bm 6 7 8 9 10 II 12 Month

PAGE 63

Correlation between monthly factor, weekly factor, and daily factor To jointly use monthly factor, weekly factor, and daily factor to indirectly indicate non local driver population levels, the three factors should not statistically correlate to each other. Correlation analyses were performed in the study to confirm that there was no statistical correlation between these factors. Table 6-1 presents R' values between monthly factor, weekly factor, and daily factor. It can be seen that the correlation coefficients between any pairs of factors were very small. This suggested all three factors were independent from each other and could be used as independent variables in regression analysis to develop an index indirectly representing non-local driver population levels. Table 6-1: Correlations Between Factors. Factors R2 MF,WF 0.0153 MF,DF 0.0059 WF,DF 0.0166 Speed-Volume Models Traffic data wer e collected from three sites including sites 0130, 0171, and 0174. As stated in C hapter 3, the speed-volume curve can be represented by a linear model, or Speed= a+ b x Volume (6-7) where the parameter "a" is free-flow speed and the parameter "b" is affected by non-local driver population level if other conditions are given. Table 6-2 presents the parameters and corresponding R' v alues obtained from the data collected from the three sites. It can be seen that some data were miss ing from the original database. This database was not sufficient to include all data collected from each month and each lane. The results presented in the rest of the Chapter were based on the data files that had sufficient volume and speed data. 54

PAGE 64

Table 6.2: Speed Volume Models at tbe Three Sites. Site Directio n Month Lane 1 Lane2 lane 3 a b R" a b R" a b R' 0130 Eas t Jan 97.00 0 0081 0.6434 102.66 -0.005 1 0 .45 03 101 73 0.0023 0.1756 Feb 96.77 -0.0070 0.5803 102.3 1 .0044 0.3996 101.57 -0 .001 9 0.12 1 5 Mar 95.64 -0. 0061 0.5320 102.1 1 -0 0045 0.3796 101.55 0.0018 0.1175 Apr 94.64 -0. 0050 0.4009 100.69 -0.0027 0 2091 101 .25 -0 0007 0 0229 May 94.12 -0 0037 0.2367 99.73 0.00 14 0.0702 101 .31 0 0002 0.0015 June 94. 30 -0.0 037 0.4090 1 01.05 -0 0033 0.2077 102 .35 0 0021 0.1212 July 95.72 -0.0054 0 5297 100.93 0 0037 0.2 4 55 102 36 -0.0026 0.1835 Aug 94.18 -0.00(2 0 .4 459 101.10 -0.0037 0 2269 102.57 -0.0026 0.1679 Sep 95.37 -0.0060 0.5017 101.19 0.0028 0.2670 102.31 -0.0014 0.0867 Oct 94 55 0 0062 0.453 1 100 .81 0028 0 1 596 102.11 -0.0013 0.0536 Nov N /A N/A N/A N/A N/A N/A NIA N/A NIA Dec N/A N/A N/A N/A N/A N / A N/A N/A N/A West Jan 96.04 -0.0043 0.4697 105 .12 0.0038 0.4 4 9 108.52 -0.00 30 0 1722 Feb 96.14 -0.0045 0.4744 104.88 -0.0040 0 4165 108.28 -0 0 0 33 0.1956 Mar 96.69 -0.0045 0.4622 104.95 -0.003 8 0.3783 108 .19 -0. 0033 0 1949 Apr 95 .9 0 -0 0036 0 .37 93 103.99 0 0024 0 .2343 1 07.66 -0 0015 0 0642 May 95 .60 -0 0028 0.3179 103.93 -0 0018 0 1727 107.25 -0 0002 0.0016 June 96 15 0 0044 0 3348 104.10 .0027 0.1991 107.50 -0 0015 0.2528 Ju l y 95. 57 -0.0046 0 3227 103.96 0 .0032 0.226 1 07.13 -0 0019 0.0749 Aug 95.22 -0.0045 0.3005 103.90 -0 0029 0.1695 107.0 1 -0.0019 0.0453 Sep 95.49 -0.0034 0 3128 104.0 1 -0 0 021 0.1638 107 .03 -0.0004 0.0039 Oct 96.44 -0.0043 0 5781 104 75 -0. 0029 0.2154 107. 18 -0 .0021 0.0812 Nov 97. 50 -0.0040 0.4278 106.09 0 003 1 0.2955 109 34 -0. 0017 0 .046 6 Dec 97.72 0049 0.4429 106 43 -0.0036 0.3206 109.75 -0 0023 0.0840 0174 North Jan 107 .06 -0.002 4 0.1234 102. 76 -0 .0 035 0.1838 92 .14 -0 0031 0.2353 Feb 107 .40 -0.0022 0.1152 102.55 -0 .0 027 0.1115 9 1 .77 -0 0027 0.1478 Mar 108 .47 -0 .0030 0.3341 103.80 -0.0041 0.4543 95 28 -0. 0052 0.5577 Apr 109.05 0 .0 029 0. 2337 104.69 0 .0 034 0.2326 96 .15 -0 0041 0 3085 ( con t mued on next page ) 55

PAGE 65

Table 6.2: Speed-Volume Models at tbe Three Sites (coDtinued). S ite D irection Month Lane 1 Lane2 Lane3 May 108.81 -0.0016 0.1126 104.86 0.0022 0.1509 96.73 -0.0030 0.2255 June 108. 45 -0.0025 0.1694 104.49 .0032 0.2188 97.19 -0.0046 0.4198 Jul y 108.59 -0.0018 0 0837 104.50 0.0025 0.1330 96 97 0 0 035 0.2749 A ug 107.72 -0.0019 0 .0545 103.9 7 -0.0025 0.0867 96 55 0 0035 0.2114 Sep 108 .60 -0.0016 0.0594 104.74 -0.0021 0.0842 96.73 -0.0027 0.1209 Oct 107.44 -0.0017 0.0462 104.00 -0. 0027 0 .0949 96.07 -0.0031 0.1344 Nov 1 06.6 1 -0.0076 0.7482 103.90 -0.0040 0.4976 114. 13 -0.0031 0.4078 Dec 1 06.39 -0.0077 0.7178 98.25 -0.0040 0 3098 114.70 -0 0032 0.3539 0174 South Jan 1 1 0.37 -0.0050 0.3661 1 10.76 -0. 0077 0.5729 96.96 -0.0057 0.3574 Feb 1 10.29 -0. 0040 0 2469 110.84 -0. 0 071 0.5920 96.48 -0.0046 0.2830 Mar 101.43 -0.0053 0.1496 1 09.45 -0. 0062 0.4019 105 97 -0.0045 0.1296 Apr 99. 67 .0067 0 .6794 109.31 -0.0059 0.6843 110.93 -0.0042 0.3075 May NIA NIA NIA NIA NIA NIA NIA N / A N/A June NIA N/A NIA N/ A NIA NIA NIA N/A N/A July NIA N/ANIA N/A N/A NIA NIA N/A N/A Aug 98.25 -0. 0050 0.3098 107 90 -0.0048 0.3220 109.09 .0028 0.11 4 1 Sep 97.94 -0. 0042 0.2576 109.10 -0. 0052 0.3226 108.85 -0. 0022 0.1105 Oct 98.22 .0041 0 .229 7 NIA N/A N/A 108.75 -0.0017 0 0532 Nov 103.77 .0070 0.4099 112.43 -0.0050 0 .1873 1 16.65 -0.0056 0.6330 Dec 105.89 -0.007 4 0.4636 1 1 0 .64 -0.0062 0 .4041 1 1 6.30 -0.0055 0 .4003 0171 North Jan 101.31 -0. 005 0.5274 1 02.53 -0.0029 0.4015 NIA NIA N/A Feb 101.62 -0.005 1 0 5197 1 02.74 0 0029 0.3769 NIA NIA N/A Mar 102.79 -0 0063 0.5785 103.78 0.0037 0.4574 NIA NIA N/A Apr 103.03 .006 0.5239 104.29 -0. 0038 0.4535 N/A N/A N/A May 102 76 -0.0056 0 5418 1 04.1 -0. 0035 0.5007 NIA N / A N/A June 101.4 -0.0 0 5 0.476 1 103 .22 -0.0031 0.4469 N/A N I A N/A Jul y 102.13 -0.005 0 4417 1 03 .86 -0. 0032 0.3976 NIA N/A N/A Aug 10 1.92 -0.0056 0.44 103 .77 -0. 0037 0.4019 N/A NIA NIA Sep 10 1 .4 1 -0. 0053 0.4322 103.79 -0. 0039 0.4193 NIA NIA N/A Oct 100. 77 -0. 00 4 6 0.3544 103.29 0 0033 0.3011 NIA N/A N/A Nov 1 02 05 0.0055 0 5377 1 03.43 0 0033 0 4 240 NIA N/A N/A Dec 102.17 0 .0050 0.4763 103.44 -0.002 9 0 3480 N/A N / A N/A South Jan 115.20 -0.0 130 0 .8846 103.40 -0. 0029 0.3470 NIA NIA N/A Feb 114 .59 -0.0127 0 8619 103.74 0.0028 0.3 215 N / A NIA NIA Mar 115.01 -0.0129 0.8374 103.60 -0.0028 0 .29 2 1 NIA NIA N/A Apr 114.30 -0.0124 0.7970 103.78 -0. 0030 0 2438 NIA N I A N/A May 1 1 2.66 -0.0108 0 .7731 103 .96 -0.0027 0 2722 N/A NIA NIA June 1 14.07 -0.0 118 0.7956 102 .91 -0. 0022 0 .1916 N/A NIA NIA July 1 14.65 0 0117 0.7451 103 .0 3 -0 0 023 0.1661 NIA NIA N/A Aug 114 .59 0.0 123 0.7594 1 02 .62 0026 0.1982 NIA NIA NIA Sep 114. 4 1 -0.0128 0 7878 103 .21 0 .0031 0 .2688 NIA N/A N/A Oct 116.85 0143 0 .7 593 102 .96 -0. 0031 0.2331 NIA NIA NIA Nov 118.85 .014 5 0 77 11 104 .10 -0.0 03 2 0.30 10 N/A N / A N /A Dec 1 11.5 7 0,0105 0.7440 1 1 03 .87 -0 .0 250 0 2109 NIA NIA N/A 56

PAGE 66

Development of an Index Model specifications The monthly factor, weekly factor, and daily factor are related to nonloca l driver population level. To indirectly represent the non-local driver population level, an index was developed in the study Several form ats of the index were evaluated in the study. Specifications of the index are presented in Table 6-3. To practically use the index, the index format should be simple and easy to be implemented. The linear combination of these factors to form the index would be a practically feasible way. Determination of the final format wo u ld depend on the parameter sign of each factor and also the R' valu e of each format Table 6-3: Index Model Specifications. Format Dependent Variable Independent Variables Index Model Formats I b MF. WF. DF b=kO+ kl x MF+k2 x WK+ k3 xDF 2 b MF. WF b=kO +kl X MF +k2 X WK 3 b MF,DF b=kO+kl xMF+k3xDF 4 b WF.DF b = kO + k2 X WK + k3 DF Index calibration The index should reflect the relative levels of non-local driver population. One reasonab l e way to calibra te the index parameters (kO, kl, k2, and k3) was to correlate the index with the parameter "b" shown in Equa tion 6 7 because th e parameter "b" reflec.ts the relative levels of non -loca l driver population in the traffic streams. A linear regression analysis was performed to calculate the parameters kO, kl, k2 and k3. The analysis was based on the data s hown in Tables 4-3 and 6-2, and results are presented in Table 6-4. Conceptually, a reasonable index format should have negative signs for the parameters associated with the factors because as the factors get larger more non-local drivers may exist in the traffic stream, resulting in smaller "b" value This means that the parameters 57

PAGE 67

(except kO) should have negative signs. Based on the sign assessment, only the format 3 presented right signs. The format 3 does not include the weekly factor. In fact, in some areas with tourist attractions, non-local tourist drivers use freeways on both Sundays and also weekdays The weekly factor may be not sensitive to non-local driver population in these areas. Therefore, the weekly factor may not need to be included in the index format. Table 6-4: Index Model Calibration Results. Format Index Model Formats I b = 0.00236-0.00580 X MF + 0.00314 X WK 0.00337 X DF 2 b = 0.00039 0.00685 X MF + 0.00375 X WK 3 b = 0.00385 0.004 72 x MF0.00370 x OF 4 b = -0.00309 + 0.00283 x WK 0.0035 3 x OF Final Index According to Table 6-4, the selected index has the following format: where b = 0.00385 0.00472 x MF0.00370 x OF b = parameter "b" shown in Equation 6-7, MF =monthly factor defined by Equation 6-4, and OF= daily factor d efined by Equation 6-6. R2 0.280 0.141 0.209 0.243 (6-8) To obtain the fmal index, the model shown in Equation 6-8 was normalized, and the normalized inde x is called non-local driver index (NDI) with the following form: b-b0 0.00385-bo-0.00472xMF-0.00370xDF (6 _9 ) NDI= bo = bo where b0 is the parameter "b" shown in Equatio n 6-7 when there is no non-local drivers in the traffic stream. Ac cording to Equation 6-9, if there is no any non-local driver population in the traffic stream, NDI should be zero, and as more non-local driver 58

PAGE 68

population exist in the traffic stream NDI gets larger The parameter bo was obtained from the data collected during 5:00 am to 8:00 am During this time period, it was assumed that traffic mostly consisted of commuters and regular users. Based on the database collec ted from all thr ee sites, it was found that bo = -0.0024. Thus, NDI can be represented by the following equation: NDI -2.604 + 1.967 X MF + 1.542 X DF (6-10) The index ND! can be used to indirectly reflect non-local driver population levels. Since the monthly factor and daily factor are totally based on traffic data collected at traffic count stations it is practically feasible to obtain N DI values at any location with reliable permanent count data. Impact of Non-Local Driver Population Levels on Speed-Volume Curves Speed-volume curves can be represented by Equation 6-7. Based on the results of traffic data collected from the three sites, it was found that the free-flow speed was I 03.57 kmlh (kilometers per hour), or a= !03.57. To assess the impacts of non-local driver population levels on the speed-volume curves, the parameter "b" at different non-local driver population level s should be calculated. According to Equation 6-9, b = b0 x (I + NDI) (6-11) where b0 = -0.0024, and NDI = non-local driver index, and NDI is equal to or greater than zero. Thus, the speed-volume equation can be written in the following form : Speed= 103.57 0.0024 (1 + NDI) x Volume (6-12) Equation 6-12 can be graphically shown by using different NDI values and Volume values F igure 6-4 presents the groups of speed-volume curves. It can be concluded from 59

PAGE 69

this figure that as NDI increases (more non-local driver population in the traffic stream), speed decreases if the same volume is to be maintained. Figure 6-4: Impact of Non-L
PAGE 70

Figure 6.5: D e nsity Line for LO S C (14.92 pclkmlln). II - 80 0 G 8 B D Volume (pcphpl) Table 6-5: Maximum Service Volumes (pcphpl) for LOS B, C, and D aud Different NDI Values NDI LEVEL OF SERVICE (LOS) LOSB L OSC LOSD 0 .0 1005 1492 1963 0.2 100 0 1482 1947 0.4 996 1472 1929 0.6 991 1 462 1 911 0.8 987 1452 1895 1.0 982 1442 1880 1.2 978 1432 1862 1.4 973 1423 18-17 1.6 969 1413 1831 1.8 965 1404 1816 2 0 961 1395 1800 2 .2 956 1386 l i87 2.4 952 1377 l'i7 1 2.6 948 1368 !'57 61 NDI=O NDI=0.2 NDI=0.4 NDI=0. 6 ND1=0. 8 NDI=l.O NDI=l.2 NDI=i.4 NDI=1.6 NDI =l.S NDI=2.0 NDI= 2 2 NDI=2.4 NDJ=2.6 LOSC

PAGE 71

In this way, fp values for different NDI values can be estimated. Table 6-6 presents the t;, table which was based ori the traffic data coUected from the traffic count stations at the three sites. Again, it is concluded from this table that non-local driver population has more impacts on traffic capacity when traffic den sity gets higher. To use the table in practical applications, the average driver population factors could be used without distinguishing the levels of service. It is interesting to note that the values of fp indicated in Table 6-6 are quite significant, ranging as low as 0.895. If this adjustment was applied to freeway capacities the computed values would be significantly affected. Table 6-6: Driver Population Adjustment Factors for Different Levels of Service. NDI LEVEL OF SERVICE (LOS) Average LOSB LOSC LOSD 0.2 0.995 0.993 0.992 0.933 0.4 0.991 0.987 0 983 0.987 0.6 0.986 0.980 0.974 0.980 0.8 0.982 0.973 0.965 0.973 1.0 0.977 0.966 0.958 0.967 1.2 0.973 0.960 0.949 0.961 1.4 0.968 0.954 0.941 0.954 1.6 0.964 0.947 0.933 0.948 1.8 0.960 0.941 0.925 0.942 2.0 0.956 0.935 0.917 0.936 2.2 0.951 0 929 0.910 0.930 2 4 0.947 0.923 0.902 0.924 2.6 0.943 0.917 0.895 0.918 However, the range of values are substantially less dramatic than these included in the HCM. As indicated earlier, based on discussions with members of the TRB Highway Capacity Committee, it is apparent that the HCM values were largely based on anecdotal accounts of CAL TRANS studies during the early 1970s. Discussions with CAL TRANS engineers suggest that the anecdotal evidence confounded driver population factors, 62

PAGE 72

vehicle classification factors, and perhaps highway grade factors to read a factor as low as 0.75. If our estimates of fp were observed, in conjunction with high levels of heavy duty vehicles and lower design standards, a combined factor of0.75 is within reason. Discussion The driver population adjustment factor table obtained from this study and presented in this chapter is totally based on traffic data collected from traffic count stations. FOOT has full accessibility to obtain traffic data from their traffic count stations. This table has more practical potentials to be used by transportation practitioners as compared with the driver population adjustment factor table presented in Chapter 5. To use this driver population adjustment factor table, the monthly factor and weekly factor for a particular location and during a particular time period should be obtained. FDOT can use the traffic data collected from the traffic count station at the particular site to calculate the monthly factor and weekly factor according to the definitions of the two factors. After obtaining the monthly factor and weekl y factor, the practitioner can use Equation 6-10 to calculate the NDI value. Finally, the ;, value can be checked by using Table 6-6. Traffic data collected from the three sites were not I 00 percent comprehensive. Because of roadway construction or some other unusual activities occurred, traffic data in some months were not included in the database obtained from FDOT. Results based on such incomplete database may not fully represent the practical situation To develop better models based on a completed database, field data collection should be better controlled, and a new database would be necessary. However, the methodologies used in this study can be fully applied to obtain better models. 63

PAGE 73

CHAPTER 7: SUMMARIES, CONCLUSIONS, AND RECOMMENDATIONS Summary This study was performed to develop a driver population factor table to be used by transportation practitioners in transportation planning and design. The main objectives were: I. Conduct literature and information search on past studies related to the subject area. The main information resources included information databases searched through computer network, technical papers reviewed in the study, expert comments obtained through telephone conversations; 2. Assess methodologies which could be used to evaluate the impacts of non-local driver population on roadway capacities; 3. Design experiments for the data collection of non-local driver population information and capacity reduction due to non-local driver population; 4. Assess study data resources and collect data to be used in the study. The data sources used in this study to develop driver population adjustment factor tables included (I) tourist survey database developed by the Office of Tourism Research, Bureau of Econom i c Analysis o f the Florida Department of Commerce and (2) traffic database developed by FOOT to include traffic data collected from FOOT's traffic count stations; 5. Develop indices used to indirectly represent non-local driver population levels The indices developed through this objective were based on the data sources from the Office ofTourism Research and FDOT; and 64

PAGE 74

6. Evaluate the impacts of non-local driver population on freeway capacity and develop driver population adjustment factor tables based on tourist survey database and traffic database This study was divided into two phases. Phase I consisted of the study objectives I to 4 Phase I results were summarized in an interim report. Phase ll focused on study objectives 5 and 6 This report summarizes the study results related to objectives 5 and 6 In this study, two main methods of developing driver population adj ustments were attempted This first way was to use data to estimate non-local driver population leve l s and correspo n ding capacity reductions caused by non-local driver population Based on the estimations of non-local driver population levels and capacity reductions, a driver population adjustment factor table was developed and presented in this report. In the second way, the concept of monthly factor, weekly factor and daily factor was developed. These factors were considered as the candidates to indirectly reflect non-local driver population levels. An index called non-local driver index (NDI) was developed to linearly combine the monthly factor and daily factor. Based on the NDI model specifications and assessment the monthly factor and daily factor were selected and included in the NDI model. Similar to the first way, capacity reductions corresponding to different NDI values were estimated. Thus a driver population adjustment factor table was developed and presented in the report Conclusions The population factor is an important adjustment factor in freeway capacity analyses. Unfortunately, the HCM provides little guidance on how to choose appropriate fp v alues The f P values recommended in the HCM need to be verified. Past studies were not abl e to supply adequate solutions to deal with this problem Unlike roadway geometric and heavy vehicle percentage, it is much more difficult to measure drive r populations than other external capacity factors such as lane widths and 65

PAGE 75

clearances (fw). and vehicle types (fi!V). Traditional direct measur e ments such as roadside interviews and license plate surveys are practically not feasible for driver population estimation There are many elements and considerations associated with driver populations, including trip purpose, trip origin and destination, familiarity, driver age etc. In Florida, the dominate element is the familiarity with highway facilities. In this study, non-local driver population levels were used to represent different driver populations in freeway traffic in Florida. Tourist survey data and traffic volume variations are the two indirect measurement s which are capable of providing reasonable driver population estimation. Tourist survey data maintained by the Office of Tourism Research could supply acceptable driver population estimation for freeway sections with adjacent tourist attractions. Monthly factors and daily factors computed according to traffic volume variations could provide reasonable informa tion about driver populations for any freeway sections in Florida. Freeway traffic data are collected continuously by the Florida Department of Transportat ion. Transportation engineers could have easy access to historical freeway traffic data. Freeway capacity reductions could be estimated from the linear speed-volume models High traffic volum es are essential in the speed-volume modeling process. It is difficult to develop speed -vo lume models under low traffic volumes. In the speed-vo lume models, mixed traffic volumes were transferred to passenger cars to take account of capacity reductions caused by heavy vehicles. Therefore, different speed-volume lines could be observed if there is any capacity reduction caused by different driver populations Density lines were used to compute actual fP values. Two driver population adjustment factor tables were developed through this study and presented in this report. The first table was obtained based on the tourist survey database. To use the table, a driver index (DI) should be obtained for a particular area. Dl values 66

PAGE 76

can be calculated from the tourist survey database maintained by the Office of Tourism Research. The Office of Tourism annually updates its tourist survey database The second table was obtained based on the data collected from FOOT traffic count stations. To use the table, the monthly factor and daily factor should be calculated from the database maintained by FOO T. According to the monthly factor and daily factor an index called non-local driver index (NDI) can be calculated. With the given NDI value, a driver population adjustment factor can be checked from the table. FOOT has full accessibility to the FDOT database, it may be more practically feasible for FOOT to use the second table as compared with the first table. Indirect measurements of non-local driver population levels such as the indices DI and NOI are practically useful to transportation practitioners. It is relatively easier for transportation practitioners to obtain the indirect measurements such as DI and NDI than the absolute percentage of non-local driver population in the traffic streams. In fact, when a transportation practitioner i s going to estimate the driver population adjustment factor for planning or design purposes, this practitioner may not be able to know the information on the absolute percentage of non-local driver population. However, this practitioner is able to know the monthly factor and daily factor for the particular location and the particular time period Therefore, the indirect measurements developed in this study will be practically applicable The models developed through this study were based on 1995 databases. As stated in previously chapters, these databases were not complete. Some data from some locations and collected in some months were not included in the databases. Consequently, the models may not be as perfect as they should be because of the incomplete databases. However, the modeling procedures developed through this study provide a practically feasible way to develop driver population adjustment factor tables. In future, if new data are collected, better models can be obtained by using the same modeling procedures. As noted in Tables 5-3 and 6-6 based on our limited data sample, t;, values as low as .85 to 9 are justified in areas with high non-local driver populations. 67

PAGE 77

Recommendations for Future Research The results obtained through this study and presented in this report can only be considered preliminary fmdings. While the results are very promi s ing, they are based on a limited sample of sites Their statistical robustness co u ld be increased by additional replications. During the course of this project, efforts have been underway that are dramatically improving the reliability of FOOT s permanent count traffic data. To reliably and efficiently use the results obtained from the study, a follow-up new research study is strongly recommended. In the new study more FDOT traffic count stations should be used to collect better and more complete longitudinal traffic data in terms ofhigb traffic volume, shorter time interval (such as IS-minute interval), consistent 12-month-data-collection period, no construction activities around the station sites, and more widely distributed locations An additional area suggested for futur e research is to examine the influence of driver population factor in more complex freeway sections particularly weaving sections. It would be expected that the influence of non familiar drivers would be more severe in comp lex weaving sections than on basic freeway sections. 68

PAGE 78

REFERENC E S Brilon W and M. Ponzlet, "Variability of Spee d-Flow R elationship s on German Autobahns," Transportation Research Record No 1555, 1997, pp 91-97 FHWA, Highway Statistics 1994, FHWA-PL-95-042, Federal Highway Administration, Washington, D C 1994 FHW A Passenger Ca r Equivalents for Rural Highways, FHW AIRD-821132, Federal H ighway Admini stration, Washington, D. C., 1982. FOTR, Florida Visitor Study, Florida Office of Tourism Research, Florida Department of Commerce, Tallahassee Florida, 1995. Highway Capacity Manual, Transportation Research Boan:l, Special Repon 209, 1994. HPMS Universe Totals, Florida Department of Trans portation, Tallahassee, Florida, 1994. May, A. D., Traffic Flaw Fundamentals, Prentice-Hall, Inc., Englewood Cliffs, New Jersey 1990. Minderhoud, M M "An Assessment of Roadway Capacity Estimation Methods," presented at the 76th Annual Meeting of Transportation Research Board, Washington, D C., 1997. NCHRP 3-45, Speed Flow Relationships for Basic Freeway SegmentS, Transportation Research Board, May, 1995. OECD, Traffic C apa c ity of Major Rout e s Road Transportation Research Organization for Economic Co-operation and Development, Paris, France, July 1983. 69

PAGE 79

Sharma, S. C. L et a!, "Driver Population Factor in the New Highway Capacity Manual," Journal a/Transportation Engineering, Vol. 113, No.5, 1987, pp. 575 579. Sharma, S. C. L., et a!, "Road Classification According to Driver Population," Transportation Research Record No. 1090, 1986 pp 61-69. Sharma, S. C. L., et al, "Seasonal Traffic Counts for a Precise Estimation o f AADT," ITE Journal Vol. 64, No.9, 1994, pp. 21-28. UPF, Florida Statistical Abstract. University Press of Florida, Gainesville, Florida, 1995. 70