INTEGRATED TRANSPORT AND LAND USE POLICIES FOR DEVELOPING COUNTRIES: RELOCATION OF RESIDENCES, ROAD PRICING AND TRANSIT SUBSIDY

Transcription

1 Advaced OR ad AI Methods i Trasportatio INTEGRATED TRANSPORT AND LAND USE POLICIES FOR DEVELOPING COUNTRIES: RELOCATION OF RESIDENCES, ROAD PRICING AND TRANSIT SUBSIDY Dilum DISSANAYAKE 1 Abstract. Icompatibilities betwee urba trasport ad lad use are rapidly growig i developig coutries. Due to this mismatch, vehicle owership is icreasig dreadfully causig urba areas for various problems icludig cogestio ad air pollutio. This study attempts to ivestigate the household travel behaviour o vehicle owership, mode choice ad trip sharig aspects by developig a ested logit (NL) model as a basic step of the aalysis. The, the estimated NL model is used for itegrated policy assessmets based o relocatio of resideces, road pricig, ad reductio of trasit fares. The policy impacts are preseted as the reductios of vehicle kilometres of travel ad air pollutio cosiderig Bagkok Metropolita Regio as a case study. 1. Itroductio Icompatibilities betwee urba trasport ad lad use are rapidly growig i developig coutries. Eve though both trasport ad lad use sectors have bee gaiig variety of improvemets with the growig ecoomic coditios i developig coutries, iterrelatio betwee them has ot bee well established at the iitial stages of plaig ad decisio makig of such improvemets. Due to this mismatch of trasport ad lad use, vehicle owership ad usage i developig coutries are icreasig dreadfully causig urba areas for severe evirometal damages such as traffic cogestio ad evirometal pollutio. Siriivasa ad Ferreira [2] explaied i their recet article that traffic cogestio is maily due to lack of coordiatio betwee lad use ad trasport plaig. O the other had, ucotrollably risig lad prices i Cetral Busiess District (CBD) ecourage urba dwellers to locate their resideces i the suburbs ad as a result, suburba sprawlig is gettig more attractio i recet years. The outcome from suburba sprawlig has disadvatages ad is troublesome if a coutry caot provide adequate ifrastructures ad trasit facilities to maage daily trasport eeds betwee the suburbs ad CBD. It is clear that most of the developig coutries have iefficiet trasportatio systems ad limited 1 Trasport Operatios Research Group, Uiversity of Newcastle upo Tye, NE1 7RU, dilum.dissaayake@cl.ac.uk

2 312 D. Dissaayake highway etworks, which may ot be sufficiet eough to serve suburbs. Cosequetly, most of the households i the suburbs who daily commute to CBD use their ow vehicles rather tha travellig by iferior trasit provisios available i the system. This may fially icrease vehicle owership creatig adverse effects o the eviromet ad the society. Webster et al. [3] highlighted the importace of itegratig trasport ad lad use aspects i urba plaig approaches. I developig coutries, travel decisios amog the household members are iextricably iterrelated, ad therefore, household travel behaviour govers this research study. Most of the households i developig coutries are sigle-vehicle owers, ad therefore, the possibility of geeratig household servig trips is cosiderably high. Vehicle users, i most cases the commuters, usually take the resposibility of drivig for other household members by sharig vehicles for multipurpose/destiatio trips. By observig perso trip data i developig coutries, trip sharig ca be idetified as oe of the popular travel attractios for the vehicle owig households, for istace car ad motorcycle shared trips make up 30% of household trips [Dissaayake ad Morikawa, 1]. Therefore, aalysig tripsharig behaviour is importat, especially for ivestigatig urba policies i developig coutries. Accordigly, major travel decisios of vehicle owership, mode choice ad trip sharig are aalysed i a comprehesive modellig framework i this study. This study attempts to ivestigate iteractive trasport ad lad use policies to geerate evirometally sustaiable urba system. As the first step of the study, a ested logit (NL) model is developed to ivestigate the household travel behaviour for vehicle owership, mode choice ad trip sharig aspects. The est structure cosists of two levels where the upper level characterizes with car owig, motorcycle owig ad o vehicle owig choices, ad the lower level represets the household related mode choices. Sice trip sharig is fairly popular evet amog the vehicle owig households i developig coutries, it is cosidered as oe of the mode choice optios i the proposed NL model. I the aalysis, two traveller households are cosidered where the mai traveller is a commuter ad the travel purpose of the secod traveller of the same household ca be ay type such as work, school, recreatio or shoppig. The estimated NL model is successfully applied for policy aalysis over relocatig resideces. I additio, several policies are ivestigated by combiig relocatio policies with road pricig ad trasit fare reductios to uderstad the combied effect with the itegrated policy aspects. Impacts of the policies are expressed i the forms of cogestio reductio ad air quality improvemets cosiderig Bagkok Metropolita Regio (BMR) as a case study. With the proposed policies, vehicle kilometres of travel (VKT) ad air pollutio ca be successfully reduced to improve the mobility ad the air quality of the regio. Past ivestigatios of impact assessmet of trasport/lad-use policies were maily based o aggregate demad modellig. I geeral, aggregate demad models provide over ad uder estimates ad therefore, accuracy of the results caot be well guarateed. As a attempt to overcome this, preset research cosiders disaggregate demad modellig at household level eve though the ivestigatios related to disaggregate models are very complicated ad time cosumig. Furthermore, aalysis coducted at household level is advatageous i order to icorporate the iterrelatioships of the household members travellig such as trip chaiig ad sharig. I additio, this study applies disaggregate demad models directly i estimatig reductios of Vehicle Kilometres of travel (VKT) ad air pollutio to receive accurate estimates that ca be addressed as oe of the iovative ideas i the preset research.

3 Itegrated trasport ad lad use policies for developig coutries Study area ad data descriptio The empirical aalysis is coducted for Bagkok Metropolita Regio (BMR) i Thailad. The BMR cosists of Bagkok Metropolita Area (BMA) ad five adjacet provices of Samut Praka, Nothaburi, Pathum Thai, Nakor Pathom ad Samut Sakor. The study regio icludes 505 iteral traffic zoes. The area of the BMR is about 7760 km 2 ad the total populatio was 13.8 millio i BMR is split ito three major zoes: CBD, Ier Suburb ad Outer Suburb, ad the Ier ad the Outer Rig Roads are cosidered as zoe separatio cordos (figure 1). Total daily perso trips i BMR are about 22 millio i 2001 ad 90% of those are geerated withi the CBD ad the Ier Suburb zoes. BMR is idetified as oe of the worst metropolita areas i the world due to traffic cogestio. Especially durig morig ad eveig commute hours, the traffic situatio there is extremely severe. I BMR, the lowest travel speed occurs i the cetral area of Bagkok (CBD), which is about 5~10 km/hr. The travel speed icreases as the distace from the city cetre icreases. Outer Rig Road Ier Rig Road 1. CBD 2. Ier Suburb Zoe 3. Outer Suburb Zoe Figure 1. Study Area: Bagkok Metropolita Regio. The data, which are used i this study, were obtaied from the household travel survey that was coducted i BMR durig 1995/96. The Urba Trasport Database ad Model Developmet (UTDM) Project was resposible for the survey (1998). The survey provides a wide variety of data useful for uderstadig travel behaviour, ad the data cosists of attributes of the trips that were made o the date of the survey as well as iformatio of household members. Although there was a large amout of households i the database, 1205 households were selected for the empirical aalysis accordig to the model requiremet of two-traveller households, of the two oe traveller has to be a commuter. I the database, trips were idicated usig the zoes of origis ad destiatios with all idepedet mode (uliked) trips. Therefore, it is easy to distiguish iterrelatios amog the trips for both travellers such as trip purposes, trip patters (chaied, shared or uliked), origi ad destiatio zoes, trasfer zoes, travel times ad time of day. Geographical iformatio of the study regio was origially computerized by Mapifo Geographical Iformatio System based Arc-view software, which is helpful for easy referece ad meaigful compariso as ecessary. Furthermore, locatio-based iformatio such as trip legth is measured o the existig road etwork usig the criteria of the shortest distace betwee origi ad destiatio zoes. To help stregthe the overall database, the Bagkok Evirometal Improvemet (BEIP) Project provided a additioal database for the home iterview survey (1997).

4 314 D. Dissaayake 3. Modellig household travel behaviour The trasportatio modes i BMR at the time of data collectio were bus, rail, car, motorcycle, hired motorcycle, taxi, ad ferry. Sice the user attractio towards rail ad ferry are very low, the choice set of trasport modes for this study excludes those travel optios from the aalysis. This study ivestigates the travel decisios of two-traveller households, with oe of the travellers makig a commuter trip. The travel purpose of the secod traveller i the household ca be work, shoppig, private busiess, social or recreatio. Whe both travellers share the same vehicle, the commuter has to drop by the destiatio of the secod traveller before reachig his destiatio. The model proposed i this study has two levels (figure 2). The upper level illustrates the household choices for car owig, motorcycle owig ad o vehicle owig. The lower level represets the correspodig mode choice combiatios for two-traveller households ad it has 17 optios. I Fig. 2, C, CSH, M, MSH, B, H ad T represet car, car sharig, motorcycle, motorcycle sharig, bus, hired motorcycle ad taxi, respectively. Car owig Motorcycle owig No Vehicle owig 1 & 2 CSH 1:C 1:C 1:C 1:B 1:H 1:T 1 & 2 MSH 1:M Alt. Household mode choices Household mode choices Alt. Commuter Secod traveller Commuter Secod traveller 1 Car sharig (CSH) Car sharig (CSH) 7,14,17 Taxi (T) Taxi (T) 2 Car (C) Bus (B) 8 Motorcycle sharig (MSH) Motorcycle sharig (MSH) 3 Car (C) Hired motorcycle (H) 9 Motorcycle (M) Bus (B) 4 Car (C) Taxi (T) 10 Motorcycle (M) Hired motorcycle (H) 5,12,15 Bus (B) Bus (B) 11 Motorcycle (M) Taxi (T) 6,13,16 Hired motorcycle (H) Hired motorcycle (H) Figure 2. A Nested Logit Model of vehicle owership, mode choice ad trip sharig. 1:M 1:M 1:B 1:H 1:T 1:B 1:H 1:T 4. Model estimatio ad results The model is estimated usig Gauss software package. Most of the parameters are with reasoable sigificace ad expected sigs. Alterative specific costats for Alteratives 2, 3, 4, 9, 10, 11 ad 16 are sigificatly positive idicatig the household preferece of usig separate modes tha formig shared trips. I additio, all households with or without vehicles have a preferece for bus mode (Alterative 5, 12, 15) sice the commo alterative specific costat is positively sigificat. Coefficiets for the travel time ad the travel cost/icome are sigificatly egative as expected. The scale parameter of the upper level is estimated to be 0.49, ad it falls i the limit betwee 0 ad 1 preservig the estig behaviour of the proposed NL model. I the NL model, may dummy variables are icluded to ivestigate the behavioural tred o vehicle owership, mode choice as well as

5 Itegrated trasport ad lad use policies for developig coutries 315 trip sharig aspects. The Value of Time (VOT) is calculated usig the results ad it is foud as 30 Thai Baht/hr. The goodess of fit for the model has a high value of Model applicatio for urba policies The estimated NL model is used to ivestigate urba policies o trasport ad lad use. Policies are maily coducted by relocatig resideces i suburbs to improve the traffic situatio i CBDs. Itegrated policies are also aalysed by combiig residetial relocatio with road pricig ad trasit fare reductios. Policy scearios coducted are as follows: Sceario 1: Relocatig residece Chage residetial locatios cosiderig its coveiece for the commuter ad household members, ad the easy accessibility to their destiatios. Sceario 2: Relocatig resideces ad road pricig Chage residetial locatios together with a road pricig policy for CBD travel [car tax (80 Baht/day) ad mc tax (30 Baht/day)]. Sceario 3: Relocatig resideces, road pricig ad trasit subsidy. Chage residetial locatios with a road pricig policy for CBD travel [car tax (80 Baht/day) ad mc tax (30 Baht/day)] ad trasit subsidy (25% of bus fare reductio) Assessmet of policy related impacts Policy scearios are aalysed usig the estimated NL model to quatify the policy impacts: Reductio of vehicle kilometres of travel (VKT) Reductio of air pollutio (No x, So 2, CO, PM, HC) Household choice probabilities for each alterative are explicitly icorporated i quatifyig the policy related impacts. The method of calculatig VKT for each alterative ad the total VKT for the etire household travel are described i equatios (1) ad (2). I the Equatios (1) ad (2), N is the total umber of households i the data sample, VKT (i) is VKT by alterative i for household, T(VKT) is the total VKT, P (i) is the probability that alterative i is chose by household, d is the travel distace of household, d share is the distace of the shared ride, d 1 is the commuters travel distace, ad d 2 is the secod travellers travel distace. VKT (i) = P (i)d (1) N share 1 2 T(VKT) = P ( ) (i)d P (i)d d (2) = 1 i= 1,8 i 1, Reductio of vehicle kilometres of travel (VKT) The VKT for the base case (without policy) ad the Scearios 1 through 3 are calculated separately usig the Equatios 1 ad 2. Accordigly, policy sceario 1 based o relocatig

6 316 D. Dissaayake resideces shows a great effect i reducig private vehicle usage i the area up to 55%. Although itegrated policies i Sceario 2 ad 3 cotiue to reduce VKT i the area, the resultat effects due to road pricig ad trasit fare reductio are foud as egligible Reductio of air pollutio For the air pollutio estimatios, coutry-level emissio factors are applied rather tha usig composite emissio factor with a default speed, especially to improve the accuracy of the results. Therefore, pollutat emissio factors due to mobile source emissios that are used i this study were obtaied from the Pollutio Cotrol Departmet i Bagkok. Equatio 3 explais the procedure of estimatig air pollutio where E(AP) idicates the expected value of air pollutio emissio i milligrams, F(AP) car ad F(AP) mc are emissio factors for car travel ad motorcycle travel i mg/km. The reductios of air pollutio levels for No x, So x, CO, PM ad HC are calculated. All the pollutats are reasoably reduced with the scearios discussed i this study ad reductios are i the rage from 18% to 26%. Accordig to the results, air pollutio i the regio ca be greatly reduced up to the rage of 50% with the residece relocatio policies. For all scearios 1, 2 ad 3, show similar reductios of air pollutio, ad therefore, sceario 1 is foud as the most attractive policy sice the chages made by road pricig ad bus fare reductios make oly a small effect. E ( AP) share P (1)d 1 [ P (2) P (3)] d 2 P (11)d N = = 1 share P (8)d 1 [ P ] (9) P (11)d 2 P (3)d 1 2 [ P (4) ( P (7) P (14) P (17))]( d d ) 1 2 [ P (10) ( P (6) P (13) P (16))]( d d ) F F ( AP) ( AP) (3) 6. Coclusios Itegrated trasport ad lad use policies for developig coutries are ivestigated cosiderig, relocatig resideces, road pricig ad reducig trasit fares. The NL model is foud to be a suitable techique to model the basic domai of the household mobility over vehicle owership, mode choice ad trip sharig decisios that is later applied to

7 Itegrated trasport ad lad use policies for developig coutries 317 ivestigate the policies. Relocatio of resideces is aalysed as the basic policy i this study. Cosequetly, several policy scearios are ivestigated by itegratig relocatig resideces with road pricig policy i CBD ad trasit fare reductios. Related impacts of the policies are estimated with correspodig reductios of VKT ad air pollutio. For the calculatio of VKT ad air pollutio, household choice probabilities for each travel optio are explicitly icorporated. I additio, to improve the level of accuracy of the pollutio estimates, coutry-level emissio factors are used i the aalysis. The resultat reductios of VKT ad air pollutio are foud to be very attractive ad therefore, the regioal mobility ca be sigificatly improved with the policies cocered. Furthermore, the previous ivestigatios that have bee coducted so far i this cotext, especially for air pollutio estimatios are i aggregate basis ad some ucertaities ca be ivolved i the fidigs. The pricipal objective of this study is to apply the cocepts of disaggregate modellig directly for the estimatio of VKT ad air pollutio levels aimig to improve the reliability of the aalysis. Refereces [1] D. Dissaayake, T. Morikawa. Aalysig Telecommutig as a Urba Trasport Policy, 21 st ARRB & 11 th REAAA Coferece, Australia, [2] S. Siriivasa, J. Ferreira. Travel Behaviour at the Household Level: Uderstadig Likages with Residetial Choice, Trasportatio Research D, 7, , [3] F. V. Webster, P. H. Bly, N. J. Paulley. Urba Lad-use ad Trasport Iteractio: Policies ad Models, Gower Publishig Group, Hampshire, Eglad, 1988.