Development and Application of a Freeway Priority-Lane Model

Transcription

1 16 sring traffi impats on residential streets in Denver, possibly by se of environmental apaity stdies, involving traffi, noise, safety, and attitde srveys. 3. The atomobile-diversion strategy goals, objetives, and tehniqes ontained in this report shold be appliable to a speifi area in Denver, if potential benefits that otweigh potential detriments an be determined and if spport is evidened by all involved interests and deision-making grops. ACKNOWLEDGMENT Alan L. Canter and Robert J. Werner of DPO provided gidane in the preparation of this report. The DPO Transportation Sbommittee, haired by Philip Milstein, provided gidane and proposed sbstantive reommendations. The preparation of this report was finanially assisted by a grant from UMT A passed throgh the Denver Regional Conil of Governments to the ity and onty of Denver. The ontent of this report reflets the views of DPO, whih is responsible for the fats and the aray of the data presented here. The views expressed do not neessarily reflet the offiial views or poliy of the ity and onty of Denver. This report does not onstitte a standard, speifiation, or reglation. REFERENCES 1. Transportation mprovement Program. Federal Register, Vol. 4, No. 181, Sept. 17, B. K. Reihart. Transportation System Management: A New Approah. Paper presented at the Amerian Soiety of Civil Engineers National Convention, Nov. 1975, preprint 2595, p Trends and sses, Land Use and Physial Development in Denver. Denver Planning Offie: J an. 1976, pp Soial and Eonomi Effets of Highways. Federal Highway Administration, 1974, pp D. M. Appleyard, S. Gerson, and M. Lintell. Livable Urban Streets: Managing Ato Traffi in Neighborhoods. Federal Highway Administration, Rept. FHWA/ SES-76-3, Alan M. Voorhees and Assoiates. Ato Restrited Zones Bakgrond and Feasibility, Draft Phase Report. U.S. Department of Transportation, Rept. DOT-TSC-157, Marh 1976, pp How Some Cities Are Diverting Cars, Slowing Traffi, Ptting Pedestrians Bak on the Street. Snset, May 1976, pp The Bmpy Road to Traffi Diversion. Planning, Vol. 43, No. 4, April-May, 1977, pp R. Frasto. To Save the Street, Close t. Planning, Vol. 4, No. 7, Ag. 1974, pp Berkeley Ordered to Remove Dividers. Western TE (nstitte of Traffi Engineers), Vol. 32, No. 2, Marh-April 1978, p L. C. Orlob. Traffi Diversion for Better Neighborhoods. Traffi Engineering, Vol. 45, No. 7, Jly 1975, pp nformation on: Traffi Management Plan. Berkeley Department of Transportation, Berkeley CA, Feb Six Months' Experiene. Berkeley Traffi Management P lan. Berke ley Department of Transportation, Smmary of Evalation Rept., Ellis Commnity's Sqeaky Wheel. Ellis Commnity Newsletter, Denver, Jan. 27, Ellis Commnity's Sqeaky Wheel. Ellis Commnity Newsletter, Denver, Feb. 26, D. G. Capelle, F. Wagne1, and S. Lokwood; Alan M. Voorhees and Assoiates. Tr ansportation Systems Management. Teh. Notes, Vol. 4, l\l'o. 2, Jl y 15, J. Sidener. Reyling Streets. Department of Landsape Arhitetre, Univ. of California, Berkeley, Nov Traffi in Neighborhoods. Department of City Planning, San Franiso, Nov Pbliation of this paper sponsored by Committee on Urban Systems Operations. Development and Appliation of a Freeway Priority-Lane Model Matthys P. Cilliers, Adolf D. May, and Reed Cooper, nstitte of Transportation Stdies, University of California, Berkeley This report desribes the stats of freeway priority lanes in the United States, the development of a freeway priority-lane simlation model (FREQ6PL), and the appliation of the model to a real-life sitation. Of the five feasible types of priority lanes, normal-flow exlsive lanes that reserve one or more lanes for priority vehiles are the most prevalent. FREQ6PL an simlate one or more lanes sed exlsively by priority vehiles (bses only or vehiles of either three or more or two or more opants). Three points in time are simlated: the before sitation (no exlsive lane). the short-term after sitation (the first day of operations with no traveler demand responses), and the longer-term after sitation (3-6 months later, after spatial and modal shifts). Performane is measred by an integrated measre of effetiveness that inldes osts of travel time, fel onsmption, and vehile emissions and faility operating and maintenane osts. The model was applied to the Santa Monia Freeway in two parts: (a) to the priority t-off limit, nmber of reserved lanes, and length of the exlsive lane and (b) to different parallel arterial speeds, different levels of arterial spare apaity, and different hypothetial mode shifts. t was onlded that reserving an existing or added freeway lane on sh a freeway will at best make its performane as good as before and at worst signifiantly poorer in both the short- and longer-term sitations. n reent years the emphasis in transportation planning has shifted from long-term, apital-intensive, apaity-

2 17 inreasing projets to shorter-term, relatively lowost projets aimed at sing existing transportation failities more effiiently, by stressing energy onservation and environmental impat analyses. n September 1975, the Urban Mass Transportation Administration (UMTA) and the Federal Highway Admittistration (FHW A) issed joint reglations (1) that established plamrlng reqirements for sh prajets in rban areas. These reglations plaed heavy emphasis on transportation system management (TSM). The following major ategories of TSM ations were identified: 1. Ations to ensre the effiient se of existing road spae throgh a. Traffi-operation improvements to manage and ontrol the flow of motor vehiles, b. Preferential treatment for transit and other high-opany vehiles,. Appropriate provisions for pedestrians and biyles, d. Management and ontrol of parking, and e. Changes in work shedles, fare strtres, and atomobile tolls to rede peak-period travel and to enorage off-peak se of transportation failities and transit servies; 2. Ations to rede vehile se in ongested areas; 3. Ations to improve transit servie; and 4. Ations to inrease internal transit-management effiieny. Use of exlsive lanes on rban freeways is a TSM tehniqe that provides preferential treatment to highopany vehiles. The terms "exlsive", "priority", and "reserved" lanes are sed interhangeably in this report and refer to freeway lanes reserved for the exlsive se of vehiles with two or more opants, vehiles with three or more opants, or bses only. The nstitte of Transportation stdies (TS) at the University of California, Berkeley, has done several types of TSM researh over the past deade (. The Traffi Management Grop dealt with freeway emergeny detetion systems (!, freeway orridor operations stdies (?_, Q), priority operations (7,, traffi management of srfae sti eets (9-11), andtraf.fi management on freeways (11-13)": The researh on exlsive lanes on rban freeways desribed here ontines this work. STATUS OF FREEWAY EXCLUSNE LANES N THE UNTED STATES While exlsive lanes on rban arterials are sed worldwide, exlsive lanes on freeways are sed primarily in the United states. Figre 1 lassifies 13 sh ses in terms of the following for variables: (a) aess to and egress from the exlsive lane, (a) aess to and egress from the exlsive lane, i.e., standard righthand on- and off-ramps, both right- and left-hand onand off-ramps, or speial mmps sed only by prio1 ity vehiles; (b) the lanes reserved, i.e., the median lane in the peak fl.ow diretion, the median lane in the nonpeak diretion, the oter lane in the peak diretion, or a separnte roadway for the exlsive se of priority vehiles; () the priority t-off level, i.e., how priority vehiles are defined in terms of the nmber of opants; and (d) nmber of reserved lanes. The 13 identified ses (14-!1) in hronologial order of implementation are 1. Shirley Highway, Virginia, 1969; approah to Linoln Tnnel, New York, 197; 3. Sotheast Expressway, Boston, 1971; 4. Long sland Expressway, New York, 1971; 5. US-11, Marin Conty, California, 1972; 6. San Bernardino Bsway, Los Angeles, 1973; , Boston, 1974; 8. Moanala Freeway, Honoll, 1974; , Miami, 1975; 1. CA-28, San Franiso, 1975; 11. Banfield Freeway, Portland, Oregon, 1975; 12. Santa Monia Freeway, Los Angeles, 1976; and 13. CA-58, San Franiso Bay Area, The lear trend is for one or more of the existing freeway lanes to be reserved for priority vehiles; this is the most prevalent type. MODEL DEVELOPMENT An existing freeway priority entry-ontrol model, FREQ5CP (Q), was seleted as base model for FREQ6PL, whih was developed primarily to evalate type 1 exlsive lanes bt an also evalate speial ases of types 2 and 5. Model strtre Figre 2 shows the new model's strtre. n the following desription step nmbers refer to the nmbers in Figre 2. steps 1-5 represent inpt to the program. Freeway design featres inlde sbsetion lengths, sbsetion apaities, sbsetion speed-flow rves, position and apaities of on- and off-ramps, grades, 1'Vatr e, srfae textre, and nmber of lanes. The lane definition refers to whih strategy is being investigated in terms of position, time, and the priority toff limit. The freeway demand pattern refers to the origindestination (-D) tables and opany distribtion at eah on-ramp. -Ds may vary from time slie to time slie over the peak period. The alternate rote speeds are those speified for different setions of the alternate rote and represent the level of servie on it. The measre of effetiveness (MOE) refers to the money vales plaed on the different MOEs by the ser. This is disssed below. step 6 simlates peak-period traffi operations for the before sitation, or no exlsive lane. The reslts of the simlation, expressed in terms of the performane index (Pl), will serve as the basis of omparison for later simlations. step 7 is an option in ase the ser is interested in only the before sitation. n step 8 the strtring of the exlsive lane refers to the splitting of -D tables (by the program) into different opanies, hanges in the roadway apaities, and other maniplations neessary before the shortterm after sitation an be simlated. This is also disssed below. step 9, the short-term performane with an exlsive lane, is an effort to simlate the first day of operations before drivers have hanged their behavior; i.e., all vehiles have the same time, spae, and opany patterns as before. Performane is expressed in terms of the P. step 1 is an option in ase the ser wants to ompare only the before and short-term after sitations. n step 11, spatial shift refers to ertain nonpriority drivers diverted to alternate parallel rotes. The spatial shift algorithm is disssed later on. n steps 12-14, mode shift refers to opants of non-

3 18 Figre 1. Classifiation of lane and ramp types. Ramp Type and Priority Ct-Off Limit* Standard Right Hand Side Only RAMP TYPE Both Right and Left Uand Side Speial Ramps for Priority Vehiles Type and Nmber of Exlsive Lanes 2+ Bses 4+ J+ 2+ Only Veh's Veh'e Veh 1 s Median Lane(s) n Peak Diretion TYPE 2 Median Lane(e) n Non-Peak Diretion Otside Lane(s) n Peak Diretion n 1 2 n Otside Lane(s) n Non-Peak Diretion n Separate Roadway Lanes l 2 n nfeasible Region mprobable Region C=:J Feasible Region * e.g., 3+ Veh'e means that 311 vehiles with 3 or more opants are priority vehiles ** n all freeway lanes, ia the bondary ondition ***Here speial median rossings are reqired Figre 2. Strtre of (6) FRE6PL model. FREEWAY lmul,atloll (1) DESGN BEFORE EXCLUSVE FEATURES LANE (7) Yes (8) STRUCTUR8 EXCLUSlVE LANE (9) SHORT TERM SUJLA- TON EXCLUSVE (2) LANE 8FllllTON YES FREEWAY (3) DEMAND PATTERN SPATAL (11) SHFT RmlCT!ON ALTERNAT. (4) ROUTE SPEEDS MOE (5) MONEY VALUES (Dj PRElllCTllD MODE SHFT LONGER USER TERM.,. --SPECU'ED (l 4 ) MODE QrtATON SHFT 15 priority vehiles who shift to either bses or arpools. Mode shift is either predited from travel-time differenes between priority and nonpriority vehiles or is allated from ser-spplied mode-shift magnitdes. step 15, the longer-term after simlation, is an effort to simlate operations three to six months after implementation of the exlsive lane, after the demand responses of spatial shift and modal shift have orred. Performane ndex onsmption, vehile emissions, onstrtion osts, freeway operating osts, and freeway maintenane osts), of serving a fixed nmber of people on a freeway (with or withot an exlsive lane) for a speifi modal split. The sitation withot the exlsive lane is the base sitation, to whih the short-term and longer-term after sitations for different exlsivelane designs are ompared. Differenes in Pis represent either yearly ost redtions (or gains) or yearly ost inreases (or losses): P is defined in this stdy as osts, in dollars per year, in terms of ertain seleted MOEs (travel time, fel P = TTC + FCC+ VEC +CC+ FOC +FMC ()

4 19 Figre 3. Strtre of spatial-shift algorithm. (1) CALCULATE NP TT MATRX FROM SHORT TERM AFTER STUATON (5) RECALCULATE NP TT MATRX FOR FREEWAY CALCULATE SURFACE STREET (2) TT MATRX FROM NPUT SPEEDS A MAXMUM OF TEN TERATONS, x SPECFED FOR EACH YES (7) DVERT x % OF TYPE ll NP VEHCLES TO SURFACE STREETS YES (1,) DVERT ALL TYPE V NP VEHCLES TO SURFACE STREETS (8) START MODAL SH FT (SEE SECTON C) where TTC FCC VEC FOC FMC yearly travel time osts, yearly fel onsmption osts, yearly vehile emissions osts, yearly onstrtion osts, yearly freeway operating osts, and yearly freeway maintenane osts. The definition implies that (a) the model will estimat e the six ost elements for a given freeway demand, freeway design, and exlsive lane design; (b) the fntional variables inflening P and onsidered by the model inlde: exlsive lane type, loation of exlsive lane, time dration of exlsive-lane operations, nmber of exlsive lanes, existing modal split, priority toff limit, level of servie on the parallel srfae streets, and qality of bs servie as refleted in mode-shift sensitivity; () eah of the MOEs mst have a known dollar vale, spplied by the ser, sh as a time vale of $3./ person-hor; and (d) P expresses yearly osts for one peak period per day for the peak diretional flow only. Simlation Sbmode! The FREQ6PL simlation sbmodel performs the following series of simlations: 1. The freeway before implementation of the priority lane, 2. The priority lane in the short-term after sitation, 3. The nonpriority.lanes in the short-term after sitation (inlding lanes adjaent to the priority lane as well as general prpose lanes before the exlsive lane started and after it terminated), 4. Several iterations of t he p1 iority and nonpriority lanes (in order to predit spatial shilt and modal s hift), 5. The priority lane after spatial and modal shifts have orred, and 6. The nonpriority lanes after spatial and modal shifts have orred. n order to perform these simlations, the original freeway -D demand is transformed into a priority and a nonpriority -D. This is done by sing the speified priority toff limit and for syntheti -Ds in the following way. The first syntheti destination "delivers" the priority vehiles from the nonpriority lanes into the priority lane where the priority lane begins. The first syntheti origin then "aepts" these priority vehiles into the priority lane, and the seond syntheti destination "delivers" the priority vehiles (with destinations downstream of the priority lane end) from the priority lane into the nonpriority lanes. The seond systheti origin "aepts" these priority vehiles into the nonpriority lanes downstream of the priority lane end. The model atomatially redes the apaity of the nonpriority lanes along the length of the priority lane and makes frther adjstments for weaving into and ot of the priority lane. t also allows for different p r iority toff levels (two or more or tl ee or more opants or bses only), different speed-flow rves on differ ent priority or non priority lane sbsetions, and a different nmber of reserved lanes. Spatial Shift Figre 3 otlines the strtre of the spatial-shift algorithm. n the following disssion, step nmbers

5 2 Figre 4. Strtre of modal-split algorithm. SMULATE FREEWAY (1) NP TRAFFC JS'NG -D'S AFTER - SPATAL SHFT PERFORM MODE SHFT USNG DFFERENCES (5) BETWEEN P AND NP TT MATRCES RECALL P TT MATRX (2) FROM SHORT TERM AFTER SMULATON CALCULATE NEW -D MATRCES FOR P AND NP CALCULATE (3) P TT MATRX FOR FULL STUDY SECTON OBTAN DFFERENCES (4) BETWEEN P AND NP TRAVEL TME MATRCES (6) SMULATE P AND NP TRAFFC WTH NEW (7) -D TA1lLES EXPRESS RESULTS OF LONGER TERM AFTER (8) STUATON N TERMS OF THE P refer to nmbers in Figre 3. n step 1 the nonpriority travel-time matrix is allated for all -D pairs and all time slies from the short-term after simlation. Then in step 2 the srfae street travel-time matrix is allated for all -D pairs from the srfae street sbsetion inpt speeds. n steps 3 and 4, if type lv trips an save time for any type lv -D, all sh nonpriority vehiles are diverted to the orresponding srfae street sbsetions. n step 5, after this diversion, the whole peak period is resimlated and new nonpriority travel-time matries are allated for eah time slie. steps 6 and 7 are an inremental assignment proedre where as many as 1 inrements of type traffi are assigned to the srfae streets if they an save time. After eah assignment the nonpriority freeway traffi is resimlated. The reason for this inremental assignment is that the srfae street speeds are assmed to be onstant, whih wold make it very easy to overload the srfae streets and ase free flow on the freeway if an all-or-nothing assignment is sed. With type lv an all-or-nothing assignment an be sed, bease type lv traffi normally forms a relatively small portion of freeway demand. n step 8, after the spatial shift has been ompleted, the modal shift is predited. This is desribed below. Modal Shift Predited Modal Shift The nderlying priniple of the modal-shift algorithm is that travel-time differenes between priority and nonpriority vehiles are sed to predit modal shifts from nonpriority to priority vehiles. Modal-shift sensitivities reslting from the alibration of a mltinomial logit model are sed ( to predit the shift. Figre 4 otlines the strtre of the modal shift algorithm. n the following disssion step nmbers refer to nmbers in Figre 4. n step 1, after the spatial shift is ompleted, nonpriority traffi on the freeway is simlated by sing the new nonpriority -D matries. n steps 2-4, the short-term after sitation, the priority-lane traffi was simlated. However, priority vehiles may also travel ertain distanes in general prpose lanes before the beginning of the priority lane and after it has ended. Travel-time differenes between priority and nonpriority vehiles are therefore allated over the fll distane from an origin to a destination, inlding distanes traveled in general-prpose lanes. n steps 5 and 6 the FREQ5CP modal-shift sensitivities are sed to perform the shift from nonpriority to priority vehiles. Priority vehiles, as disssed before, an be defined as vehiles with either two or more or three or more opants or bses only. Two new sets of -Ds are obtained after the modal shift: one for priority vehiles and one for nonpriority vehiles. n steps 7 and 8 the new -D tables are sed to simlate the final longer-term after sitation on the freeway, whih again will onsist of the priority-lane traffi simlation and the non-priority-lane simlation. The reslts of the longer-term after simlations are again expressed in terms of the P and are ompared with the before sitation. Speified Modal Shift The prpose of the speified modal shift is to allow the model ser to address sh qestions as, What happens if the expeted modal shift is totally different from that predited bease of travel-time differenes only? That is, if a priority lane is implemented when bs fares have dereased and parking osts and fel osts have inreased, the expeted shift will be greater than that based on travel-time differenes alone. Too mh shift may ase the priority lane's demand to exeed its apaity, whih wold then defeat one of the prposes of the lane: providing priority vehiles with a travel-time savings. This, in fat, may ase the total osts, as expressed in the P, to inrease. What wold be an optimm modal split for a given exlsive lane design? Depending on some of the external impats, sh as home se of atomobiles after a modal shift, the P may at a given point inrease as more modal shifts take plae. Figre 5 otlines the strtre of the speified modal shift proedre. n the following disssion step nmbers refer to nmbers in Figre 5. steps 1-3 refer to the simlation of the freeway before implementation of the exlsive lane, the shortterm after simlations of both the priority lanes and the nonpriority lanes, and the simlation of the nonpriority lanes after spatial shift has taken plae. n step 4, whereas the predited modal shift desribed above made se of shift sensitivity vales, the modal shift now is allated by sing speified modalshift magnitdes. A modal-shift magnitde of.2, for example, means that 2 perent of the total existing passenger demand wold shift from nonpriority vehiles to priority vehiles. Separate shift magnitdes are speified for arpools and bses. step 5 ors after the priority and nonpriority -D tables have been hanged. The longer-term after sitation is simlated and ompared to the before sitation.

6 21 n steps 6 and 7, the ser examines the otpt from the longer-term after with the speified modal-shift magnitdes and, if so desired, deides on a new set of shift magnitdes in order to make another ompter rn. Different hypothetial modal shifts, ompatible with different stimli (e.g., reded bs fares or reded bs fares and dereased parking availability), an then be investigated for a partilar exlsive-lane design. Model Appliation The model was applied to the Santa Monia Freeway in Figre 5. Modal split optimization proedre. (l) SMULATON BEFORE PL MPLEMENTATON SHORT TERM (2 AFTER SMULATONS NP SMULATON (3) AFTER SPATAL SHFT PERFORM MODE SHFT USNG SPECFED MODE SHFT MAGNTUDES LONGER TERM (5 AFTER SMULATONS ----, FREQ6PL _ USER NTERFACE DETERM NE NEW MODE SHFT MAGNTUDES AND ( 7) MAKE NEXT RUN ' NO d YES the Los Angeles metropolitan area. Data sed inlded atal freeway design featres, opany distribtions for eah on-ramp, and -D data for a 4-h morning peak period. This peak period was divided into sixteen 15-min time slies. The Santa Monia Freeway is essentially an eight-lane faility with a 6. 7-m (22-ft) median. Constrtion, operating, and maintenane osts, respetively, were taken as $1, $ 6, and $1 / year, and the following money vales were assigned to (a) time: $ 3./ h; (b) fel: $.17/L (.65/ gal); and () vehile emissions: 2. 55/ kg (HC), $ 2/ kg (CO), and $. 46/ kg {No,) osts. Design of Experiment The experiment was designed to investigate the following primary variables in the design of a type 1 exlsive lane: (a) length of the exlsive lane, (b) priority toff limit, () nmber of reserved lanes, and {d) time dration of exlsive lane. The design of the experiment is shown in Figre 6 and is disssed below. Part 1 is an analysis of existing onditions. Before any traffi-management strategy an be designed and implemented, it is neessary to nderstand the existing onditions well. The existing onditions are also needed as a basis of omparison. The analysis of existing onditions is desribed below. Parj; 2 is the priority toff limit. Three priority toff limits are investigated: bses only, all vehiles with three or more opants, and all vehiles with two or more opants. The analysis is done for both the short and the longer term. Part 3 is the nmber of lanes. Three different lane onfigrations are investigated: one of the existing lanes reserved for vehiles of three or more opants, two lanes (one of whih is added) reserved for vehiles of two or more opants, and one added lane for vehiles of three or more opants. The analysis is done for both the short and the longer term. Part 4 is the length of the exlsive lane. Two designs are investigated: a long exlsive lane and a short exlsive lane. The analysis is one again done for both the short and the longer term. Part 5 is the time dration of exlsive lane. The ongestion pattern in terms of when ongestion starts and when it ends is investigated for all the alternatives. (6) Figre 6. Design of the experiment. Existing Conditions Priority Nmber of Length of Time Dration Ct-Off Reserved Ex lsive of Exlsive Limit Lanes Lane Lane (Rns 1,4,6) (Rns l,2,5) (Rns l,j) (Rns 1-6) Compter Rns: 1. One Long Exlsive Lane, Priority Ct-Off Limit One Added Long Exlsive Lane, Priority Ct-Off Limit = J ]. One Short Exlsive Lane, Priority Ct-Off Limit = J 4. One Long Exlsive Lane, Priority Ct-Off Limit a 2 5. Two Long Exlsive Lanes, Priority Ct-Off Limit = 2, One Long Lane Added 6. One Long Exlsive Lane, Priority Ct-Off Limit - Bses Only

7 22 Smmary of Reslts Figre 7 shows the predited performane of the different exlsive-lane designs in terms of the relative hanges in travel time, fel onsmption, vehile emissions, and Pl. By sing Figre 7, the reslts of Figre 7. Predited performane of lane designs.!! 8 8., "j , " 3 > so!! 5 4 "' 3 2 " 1 Jl " " so -==-==-: ==-:.._-=-= SO so ; 1 : -1 ::: so - Before Short Longer Before Short Longer Term Term Term Term After A(L«r After After 1 PL SS 2-2S Ct-Off 3 (Existing Lane) PL SS 2-25 Ct-Off 3 (1 Added Lane) PL's SS 2-25 Ct-Off 2 (1 Added Lane) PL SS Ct-Off 3 (Existing Lane) the model appliation an be smmarized. Travel-Time Costs Using an existing lane as a priority lane, regardless of length, has severe onseqenes in the short term and in the long term is still worse than the existing ondition. Adding a lane and then reserving either one or two lanes (with toff levels of three and two) does not reslt in drasti hanges in either the short or the longer term. Fel Consmption Costs Using an existing lane as a priority lane, regardless of the length, reslts in inreased fel onsmption in both the short and the longer term. Adding a lane and then reserving either one or two lanes (with toff levels of three and two) has virtally no effet on the fel onsmption in both the short and the longer term. Vehile Emissions Costs Using an existing lane as a priority lane, regardless of the length, reslts in inreased emissions osts in the short term, whereas in the longer term total emissions osts do not differ from those of the existing sitation. Adding a lane and then reserving either one or two lanes (with toff levels of three and two) has virtally no effet on the vehile emissions osts in both the short and the longer term. Performane ndex The shape of the P rve orresponds to the shape of that of travel time osts, whih illstrates that traveltime osts are relatively mh more important than either fel or vehile emission osts in allating P. The model appliation an be smmarized by the following two statements. Taking away an existing lane for the exlsive se of priority vehiles reslts in severe short-term onseqenes, and even in the longer term Figre 8. Design of the e xperiment for sensitivity analysis. Selet Base Case Speeds on Parallel Rotes (rns 1,7,8,9) Level of Servie on Parallt<l Rotes Spare Capaity on Parallel Rotes (rns 1,1,11) Freeway Perfomane (rns 12-16) Mode Shift Optimization Freeway Performane Correted (rns 12-16) Compter Rns : 1. PL SS 2-2S, 2S mph. Arterial Speed, nlimited spare apaity 7-9. PL SS 2-25, Arterial Speeds of, 15 and JS mph., nlimited spare apaity. 1,11. PL SS 2-2S, 2S mph. Arterial Speed, some and little spare apaity Five sets of hypothetial mode shifts.

8 23 is still worse than the existing ondition. Adding a lane and then reserving either one or two lanes (with toff levels of three and two) does not reslt in any signifiant hanges in either the short or the longer term. Sensitivity Analysis The following variables were investigated in the sensitivity analysis: different parallel arterial speeds, different levels of arterial spare apaity, and different hypothetial modal shifts. Design of Experiment Figre 8 illstrates the design of the experiment for the sensitivity analysis, whih was divided into three parts. Part 1 was the seletion of a base ase; part 2 was the investigation of the effet of the level of servie on the parallel arterials in terms of the average speed existing on the arterials and the spare apaity available on the arterials. For arterial speeds were investigated for the base ase, km/h in rn 7, 24 km/h (15 mph) in rn 8, 4 km/h (2 5 mph) in rn 1, and 56 km/h (3 5 mph) in rn 9. Also, three levels of available spare apaity on the arterials were investigated for the base ase: nlimited spare apaity on rn 1, some spare apaity on rn 1, and little spare apaity on rn 11. Part 3 was the investigation of the effet of different hypothetial modal shifts on the freeway traffi performane as refleted in the norreted and the orreted P, for the ase of no available parallel arterials. Smmary of Reslts The reslts of the sensitivity analysis are illstrated in Figres 9, 1, 11, and 12, abot whih the following omments an be made. Figre 9. Longer-term vehile distanes for different speeds. Reded Vehile-Miles (or Vehile-Kilometers) Freeway Vehile-Miles i\rter:i.ai.l Vehile-Miles (or Vehile-Kilometers) Figre 9 n Figre 9 the maximm reded freeway vehile kilometers or at an arterial speed of when no diversion takes plae and maximm mode shift reslts. The maximm predited freeway vehile-kilometer redtion is abot 3 perent. From to 24 km/h ( to 15 mph) the reded vehile-kilometers rve is relatively flat, bease very little diversion takes plae. Average speeds on the freeway are higher than 24 km/h for nearly all -D pairs in nearly all time slies. From 24 to 56 km/h (15 to 35 mph) diversion inreases rapidly, and, at an arterial speed of 56 km/h, 17 perent of the vehile kilometers traveled in the longerterm after sitation are on the arterials. This heavy diversion again reslts in improved nonpriority traffi performane and therefore virtally no mode shift. Figre 1 n Figre 1 at a -km/h arterial speed the longerterm after sitation is signifiantly better than the short-term after sitation. This improvement is a reslt of the modal shift. All elements of the P improve signifiantly over the short-term after performane. At a 24-km/h arterial speed very little diversion orred, as illstrated in Figre 1. However, the diversion that did or reslted in improved freeway performane and a 16 perent redtion in total travel time. All elements of the P show an improvement when ompared to the -km/h ase. At a 4-km/h (25-mph) arterial speed heavy diversion (13 perent of longer-term vehile kilometers) takes plae and reslts in reded travel time and vehile emissions bt inreased fel onsmption. The P is still abot 9 perent more than the before sitation. At a 56-km/h arterial speed both total travel time and vehile emissions are less than the before sitation, while fel onsmption shows an 8 perent inrease over the before sitation. The net effet is that the P is abot eqal to what it was in the before sitation. Figre 11 n Figre 11 the ase of little spare apaity on the arterials does not represent a realisti longer-term after sitation, simply bease many vehiles will divert bak to the freeway bease of the low speeds (ased by the diverting traffi) on the arterials. t does, however, illstrate learly that total osts may be inreased drastially by ongestion ased by the diverting vehiles. The reason why the fel osts do not hange in Figre 11 is that a fel marginal ost fator of 1 was sed for all three levels of ongestion. The signifiane of the shape of the ost rves in Figre 11 does not lie in the atal magnitdes of the ost inreases bt in the fat that the pper bondary ase (little spare apaity) gives drastially different reslts than the lower bondary ase (nlimited spare apaity). Using marginal ost fators of 1 (or assming nlimited spare apaity on parallel arterials) wold therefore definitely nderestimate the total osts (mph) (kph) Average Speed on Arterials Figre 12 n Figre 12, the more extensive the modal shift, the better the freeway traffi performane, as illstrated

9 24 Figre 1. Performane for different speeds ,,. 7 \ bo \. \. 6 \. \ "' 5 4 3!i 2..,, "' ro p 15 "' "' "' ::. "' 1-1 \ \ \ \ \ \ Travel Time Costs Fel Consmption Costs -+- Vehile Emission Costs Performane ndex., ' " /' ' -2-3 Before Short Term After _ ,.-2,2,.- s 3o,_3_s,,_.<mp h) Average Speed on Arterials Loneer Term After 6 (kph) Figre 11. Performane for different apaities.., g..,, so Before Short Term After ---- Travel. Time Costs / "',. - - Fel Consmption Costs -+- Vehile Emission Costs --.Perfrr-.n_n ndex Unlimited Spare Capaity Some Spare Capaity Longer Term After Little Spare Capaity by the norreted P. This ontines ntil rn 16, when the priority lane beomes ongested. Only after sbstantial speified modal shift (4.5 perent to arpools and 6.4 perent to bses) does the norreted P beome less than the before P. The longer-term after traffi simlation for this ase provides the following information: The maximm volmeto-apaity (V/C) ratio in the priority lane is. 52 and ors in time slie 4. The nonpriority lanes are ongested from time slie 2 to time slie 1 (ompared to ongestion in the before sitation from time slie 3 to time slie 9). The predited modal shift reslts in a shift of 2 perent to arpools and.6 perent to bses, whih is abot 25 perent of the shift reqired to break even. Shifts of.192 to arpools and.256 to bses reslt in a ongested priority lane, whih is obviosly something that will not or. Priority vehiles will not se the priority lane if they annot save time by doing so. The orreted P does not differ mh from the norreted P in the predited modal-shift range, in spite of the rather nfavorable data sed: All vehiles left at home will be sed on an 8-km (5-mile) trip. The orreted P shows a minimm at rn 15, whih is explained as follows: As more modal shift ors, the freeway benefits beome relatively smaller and external osts relatively larger. At a shift of 9.6 perent to arpools and 12.8 perent to bses, the freeway gains eqal the external osts (primarily the bs and arpool time penalties). Frther modal shift provides greater osts than gains. The reason why the orreted P never beomes less than the before-sitation P is primarily that the priority lane annot prode enogh time savings to offset the time penalties for bs and arpool speified for these rns.

10 25 Figre 12. Modal split optimization B lli GP : Carpool B = Bs Correted P -- Unorreted P e e "' U H H. H. "' bo.j: H O "< "' <( "', <( """' "' "' N- o> "' N-.-<.-< "'N "'"' OH "'"'.-<N "' U"' "' "' Hypothetial Longer Term After CONCLUSONS The reslts of the researh are smmarized by the following three general onlsions. 1. A type 1 exlsive lane on a ongested freeway is expeted to ompare nfavorably with the before sitation in both the short-term and the longer-term after sitations, onsidering total travel time, fel onsmption, and vehile emissions. 2. A type 1 exlsive lane on a relatively lmongested freeway is expeted to perform as well as or slightly worse than the before sitation in both the short-term and longer-term after sitations, onsidering total travel time, fel onsmption, and vehile emissions. 3. There may be some operating environments signifiantly different from the Santa Monia environment in terms of opany distribtion, level of bs servie, modal-shift propensity, and parallel arterials. f a type 1 exlsive lane is onsidered in sh an environment, it is reommended that an in-depth analysis be ndertaken from the speifi type 1 exlsive-lane design before deiding to implement it. REFERENCES 1. Code of Federal Reglations: Title 23-Highways. Part 45, Offie of the Federal Register, rev. April 1, D. W. Jones, W. L. Garrison, and A. D. May. Managing the Ftre Evoltion of the Urban Transportation System-nterim Smmary Report. U.S. Department of Transportation, Projet DOT-OS- 5237, Ot. 1976, 62 pp. 3. M. Sakashita and A. D. May. Development and Evalation of nident Detetion Algorithms for Eletroni Detetor Systems on Freeways. TRB, Transportation Researh Reord 533, 1975, pp A. D. May. Optimal Design and Operation of Freeway nident Detetion-Servie Systems. U.S. Department of ''ransportation, Final Rept. DOT- FH-883, Marh 1975, 48 pp. NTS: PB / 5L. 5. M. Orthlieb and A. D. May. Freeway Operations Stdy: Phase V Report 74-5-Freeway Corridor Control strategies. nstitte of Transportation and Traffi Engineering, Univ. of California, Berkeley, Ag NTS: PB-25 47/2L. 6. A. D. May. Srfae street Conseqenes of Freeway Control. Pro., 28th Annal California Transportation and Pbli Works Conferene, Los Angeles, Marh R. D. Minister, L. P. Lew, K. Ovaii, and A. D. May. A Compter Simlation Model for Evalating Priority Operations on Freeways. nstitte of Transportation and Traffi Engineering, Univ. of California, Berkeley, Jne 1972; HRB, Highway Researh Reord 461, 1973, pp K. Ovaii, A. D. May, R. F. Teal, and J. K. Ray. Simlation of Freeway Priority strategies (FREQ3 CP). Fede1 al Highway Adm inistration, Final Rept. FHWA/ PL-77-1, Jne pp. NTS: PB / BOL. 9. A. D. May. Appliation of Simlation Models in Traffi Management Operations. Pro., Winter Simlation Conferene, Saramento, CA, De. 19, 1975, 14 pp. 1. A. D. May, T. J. Clasen, and A. J. Krger. Development and Appliation of Traffi Management Models-An Exetive Smmary. U.S. Department of Transportation, Projet OOT-OS- 5237, Ot. 1976, 26 pp. 11. A. J. Krger and A. D. May. Frther Analysis and Evalation of Seleted mpats of Traffi Management strategies on Freeways. U.S. Department of Transportation, Projet OOT-OS-5237, Sept P. P. Jovanis and A. D. May. Frther Analysis and Evalation of Seleted mpats of Traffi Management strategies on Srfae streets. U.S. Department of Transportation, Projet DOT-OS- 5237, Sept M. P. Cilliers and A. D. May. Evalation and Appliation of Freeway Corridor Models. U.S.

11 26 Department of Transportation Projet DOT-OS- 5237, Sept Priority Tehniqes for High-Opany Vehiles: State of the Art overview. Tehnology Sharing Program, U.S. Department of Transportation, Nov NTS: PB /4L. 15. H. S. Levinson and others. Bs Use of Highways: State of the Art. NCHRP Rept. 143, State of California Bsiness and Transportation Ageny. Preferential Lanes to High Opany Vehiles-Final Report to the California State Legislatre. California Department of Transportation, Saramento, De State of California Bsiness and Transportation Ageny. Bsways and Related High Opany Vehile Failities and Programs. California Department of Transportation, Saramento, Feb Pbliation of this paper sponsored by Committee on Freeway Opera tions. Safety Considerations in the Use of On-Street Parking Jak B. Hmphreys, Department of Civil Engineering, and Donald J. Wheeler, Department of Statistis, University of Tennessee, Knoxville Pal C. Box, Pal Box and Assoiates, Skokie, llinois T. Dary Sllivan, Department of Traffi Engineering, Knoxville, Tennessee The researh was intended to examine relations among parking on figrations (angle, parallel, or no parking), parking density, traffi flow, street width, pedestrian ativity, and highway safety. The variables fond in this researh to be assoiated with aident rates inlde (a) fntional lassifiation of streets, (b) parking se, and () abtting land se. An important and srprising fat is that parking onfigration did not emerge as a variable that in itself was related to aident rate. nreased parking se was fond to reslt in signifiantly higher aident rates, as many as 9 spae hors per kilometer per year ( 1 5 spae hors per mile per year). Streets abtting land ses that generate high parking trnovers and pedestrian ativity have higher aident rates than those abtting lower-intensity land ses. Heavily sed parallelparkiny areas were fond to have aident rates omparable to heavily sed high-angle-parking areas. Prohibition of parking reslted in the lowest aident rates measred. Parking-related midblok aidents aonted for 49 perent of all aidents along major streets, 68 perent along olletor streets, and 72 perent along loal streets. n the early days of rban development, when densities were relatively low, motorists old often park their atomobiles on streets near their destinations. As densities have inreased, however, rb spaes have beome inadeqate and parking itself has beome a major rban land se. The ost of remaining on-street parking is high in terms of traffi ongestion and aidents. Traffi operations are now ommonly evalated as desribed by the 1965 Highway Capaity Manal (HCM) (1), whih reognizes that rb pa1 king has a signifiant effet on the apaity and servie volmes of highways. The safety aspets of parking praties, however, have not been given eqal attention in traffi engineering literatre. No widely aepted relations have been identified among parking onfigrations (diagonal, flat angle, parallel, et.), parking density, traffi flow, pedestrian ativity, and highway safety. The need for sh definitions, however, is emphasized by the large nmber of aide11ts involving rb parking. One sore (2) has estimated that abot 2 perent of all rban aidents are related to rb parking. Five primary ases were identified: 1. Vehiles parked in the roadway present obstales and serve to narrow the sable width of the roadway and to restrit the flow of traffi. Sh parking also restrits right-trn movements into and ot of side streets, driveways, and alleys. Frthermore, parked vehiles may be strk, or their presene may ase sideswipe or rear-end aidents. 2. Vehiles leaving the parked position disrpt the traffi flow and, by inreasing ongestion, lead to rearend and sideswipe ollisions. 3. Vehiles entering the parked position freqently reqire atomobiles approahing in the lane adjaent to the parking lane to slow or stop. Parking manevers are espeially hazardos bease they sally involve a baking-and-trning movement. Rear-end and sideswipe ollisions an readily reslt from this manever, 4. Drivers or bak-seat passengers getting ot of parked vehiles on the street side present an added obstale in the roadway. Not only are the door and the alighting passengers in danger of being strk, bt passing traffi may have to swerve or stop sddenly. This ases both rear-end and sideswipe ollisions. 5. The sight distane of pedestrians-many of them hildren-attempting to ross the roadway from between parked vehiles is reded, and the motorist may not see sh pedestrians in time to avoid ollision. A danger from impaired view also exists when vehiles are parked lose to intersetions and driveways. Depending on street grades and speeds, rb parking an reate a hazardos sight obstrtion if allowed on a major rote within even a hndred meters of an egress point. HCM and other traffi engineering manals state that parallel parking is the preferred arrangement for any on-street parking adjaent to traveled lanes. The angleparking alternative has sally been onsidered ndesirable from a safety and apaity standpoint. The belief that safety and apaity are ompromised in the presene of diagonal parking is based on stdies from the late 194s throgh 196s and, to a larger degree, on intitive jdgment. However, many early stdies of diagonal parking were limited in sope. n