MODEL BASED ANALYSES OF ACTIVITY BALANCING POLICIES: A STUDY USING CONVENTIONAL MODELS IN SÃO PAULO, BRAZIL.

Transcription

1 MODL BASD AALYSS OF ACVY BALACG POLCS: A SUDY USG COVOAL MODLS SÃO PAULO, BRAZL. Dense Lma Loes, M.S. C he raff ngneerng Comany of São Paulo Cty São Paulo, SP, Brazl, dlloes@uol.om.br Hugo Petrantono, D.S. Polytehn Shool - São Paulo Unversty São Paulo, SP, Brazl, hetran@us.br ABSRAC hs researh evaluated the alaton of onventonal models for the analyss of atvty balanng oles (or obs/housng balane, n a restrted sense), takng the RMSP Regão Metrooltana de São Paulo (São Paulo Metrooltan Area) as a ase study. Several models develoed for the RMSP n the ast were revewed evaluated from a theoretal ratal ont of vew. A reent set of models develoed for the Metrô/SP the Metro Comany of São Paulo Metrooltan Area, based on the latest Orgn-Destnaton (O-D) survey of 1997, was seleted as the bas tool. he models were used for smulatng the mat of four senaros of atvty balanng that were omared to the smulaton of the 1997 bas senaro. Generally, oulaton redstrbuton smulatons showed better results than emloyment redstrbuton ones. he radal senaro reahed a otental reduton of more than 33% n VHR around 9% n VKM (mrovng average seed by 38%) but requred the reloaton of 37% of the oulaton. wo thrds of the effet were obtaned wth the frst half of the reloaton effort. mloyment redstrbuton resulted n smaller effet due to the nreased tr length, delvered a better result wth half reloaton than wth the radal senaro. Other senaros of foused or ombned oles are mentoned but ould not be desrbed here due to sae onstrants. KY WORDS: Urban oles, obs-housng balane, traff ongeston.

2 1. RODUCO Poles devoted to the romoton of healther urban struture are a bakground oton aganst whh the doubtful effetveness of nvestments on road transort s set. A balaned atvty dstrbuton (manly obs housng) s an examle of obetve that should be amed at by oles that ntegrate l use transort goals tools n dealng wth urban transortaton roblems (owlan; Stewart, 1991); (Malone-Lee at al, 2001); (SMPLA, 2002). Cervero (1996) onted out that there s a strong debate on the vablty of atvty balanng oles (or obs/housng balane) as usually ours when talkng on oles drven to hange urban struture both n maro mro sales. evertheless, the otental of atvty balanng oles an be ranked as romsng based on theoretal thnkng or on lmted emral results (Bento et al., 2002). akng the ase of São Paulo Cty, atvty-balanng oles are n the oltal agenda. he 2002 Master Plan nluded the reloaton of the oulaton to the ever mentoned goals of romotng emloyment n the oorer frnge areas of the ty, n a lear oly of reoulatng the entral area (that lost oulaton even atvtes n the last deades) (SMPLA, 2002). he need of a study devoted to evaluatng the otental of these oles motvated the resent researh. he frst task of the researh was the analyss of the otons avalable for a model-based evaluaton of atvty balanng oles ther sef lmtatons /or dffultes, as summarzed n Seton 2. he seond task was the defnton of a methodology for evaluatng the otental mats of atvty balanng oles, overng senaro buldng result analyss. he seleted model eased the task of buldng extreme senaros of atvty balanng estmatng ther effets. he struture arameters of the hosen model the method used for analyss of results are outlned n Seton 3. he study gathered relevant nformaton of the mat of atvty balanng oles on several key varables related to the oeraton of road traff (as average tr tme length or seed) that are resented rtally ommented on Seton 4. Conludng, Seton 5 summarzes strengths, weaknesses of the method ts results, lus t suggests themes for further researh. 2. MODL BASD VALUAO OF ACVY BALACG POLCS As onted out by Rosenbaum Koeng (1997), ntegrated oles dealng wth l use transort goals tools over a large range of roosals. Some of them are related to the mroso sale of urban desgn (as romoton of mxed use nfll or edestran frendly develoments) but are hardly amenable to the analyses usng onventonal l use transort models, gven ther aggregated reresentaton of atvtes n satal unts (zones). n ths realm, atvty balanng s an exeton more than a rule but smulaton models are nether the only nor the referred tool for analyzng suh oles. For examle, Hy (1996) revewed four alternatve aroahes n addton to the smulaton studes, whh are: aggregate statstal analyss, dsaggregate statstal analyss, analyss of hoe models atvty-based analyss. Among onventonal smulaton models, the two man alternatves are tradtonal four ste transortaton models ntegrated l use transort satal nteraton models. Both are shortly desrbed n Huthnson (1974), ther use n evaluatng atvty-balanng oles s brefly analyzed below.

3 São Paulo Cty has a long hstory on the use of both tradtonal transort models ntegrated l use transort models. he two otons were arefully evaluated as tools for the study on assessng the otental of atvty balanng oles (Loes, 2003): - the frst transort model, usng a tradtonal struture, was develoed for the orgnal study on the Metro System, arred out n 1968 based on the smlfed 1967 O-D survey; - the frst ntegrated model was develoed n 1974, when a very smlfed (but nsghtful) verson was develoed aled usng the same lmted data avalable from the 1967 O-D survey as art of the roet alled SSRA; - the seond ntegrated model, the MU model, was albrated through the ntegraton of data from the 1977 O-D survey from munal roerty tax databases. t ombned very sohstated l use transort models of stoks flows but due to nsttutonal omutatonal roblems, the MU model was never aled n real analyses; - after the drawbak of the MU model, the Metrô/SP ket a tradton of albratng alyng tradtonal transort models after eah new O-D survey, manly for evaluatng otons n the exanson of the Metro System. hese efforts were arred out after the 1987 O-D survey after the last 1997 O-D survey; - reently, a new ntegrated model was aled to the elaboraton of the 2001 ntegrated Plan for Urban ransort, usng the SAR model, also albrated usng data from the 1997 O-D survey. Deste beng urrently used, the SAR model s not fully doumented most of ther arameters are emboded n a blak-box tye of software. he man otons seleted for further analyss were the use of the Metrô/SP tradtonal transort models or the udate of SSRA ntegrated model (Loes, 2003). 2.1 Comarng Modelng Aroahes for Analyzng Atvty Balanng Poles Frst of all, the aablty of reresentng atvty balanng oles was analyzed. hs s a feature n whh both models revealed strengths weaknesses. he advantage of tradtonal transort models was the ossblty of dretly sefyng an atvty-balanng senaro, as nut (even f extreme), evaluatng ts mat on traff oeraton. As the study ntended to evaluate the otental effet of oles, ths aroah was nterestng but would requre the defnton of a set of senaros wth varyng degrees of lausblty so as to aheve rogressve realsm n results. Anyway, the vablty of the senaros (.e., ther onssteny wth l use oles market nteratons) the seleton of oles for mlementng them would be an unanswered questons under ths aroah. he bult-n onsstene of l use atterns, gven a model sefaton, was the man advantage of alyng ntegrated l use transort models. evertheless, ths aroah would requre the sefaton of atvty balanng senaros n terms of oly alternatves (nstead of l use atterns). hs feature would not be a onern f there was a lear understng about what tools should be used (or not) by suh knd of oly. hs researh was unable to dentfy suh agreement on the tehnal rofessonal lterature. n fat, there were hnts ndatng that some oltally advsable oles (on other matters) would generate unbalaned l use atterns. he results on the alaton of the SSRA models to the RMSP, valdated through the hstoral evoluton of

4 ts urban struture, was a lear examle, showng the unhealthy attern of atvty dstrbuton suorted by the nvestment on the Metro System n São Paulo (henque, 1983). Seondly, there were ratal onerns related to data model avalablty. Usually, ntegrated models would be hardly avalable at most metrooltan areas. evertheless, ths s not the ase for São Paulo Cty the RMSP, as revously desrbed. Data needs ould also dsourage the use of ntegrated models, eseally f the l use market was to be modeled ( ths s one of the man theoretal onts n these models). evertheless, the resent study had aess to a good database on urban stoks (at least for São Paulo Cty, whh omrses two thrds of the RMSP). he endogenous foreastng of zonal atvty dstrbuton would be an emboded advantage of ntegrated models (gven the dstrbuton of bas atvtes the behavoral relatons reresented by aessblty attratveness funtons). However, ths was rrelevant to the resent study as ts goal s the smulaton of alternatve senaros for the same base year (nstead of foreastng future senaros) t ould derve atvty balanng senaros at the zonal level usng data of the base year (from the 1997 O-D survey). For other studes that amed at evaluatng trends or oles, ths feature ould be a very nterestng ont as the greater model omlexty translates tself nto smler model alaton. t s noteable that some of the semnal works on model based evaluaton of the oles analyzed n ths aer used ntegrated models (Rkaby, 1987); (Wegener, 1996). A fnal ont, s that the evaluaton of the effets on the traff oeraton, gven a dstrbuton of atvtes (ether exogenously or endogenously defned), should be very smlar as both models rely on a ommon set of roedures n smulatng mode route hoe desons. Summarzng, the onts made here that drove our methodologal desons, were: - ntegrated l use transort models are theoretally better otons for analyzng the roduton of atvty balanng senaros whle the vablty of the senaros was the man queston left oened by the analyss usng tradtonal transort models; - ntegrated models, deste beng more omlex n struture mlementaton, are easer to use (requrng nut data solely on the exogenous levels of bas atvtes, lus hanges on transort l use stoks or oles) but ths was not a relevant advantage n ths study; - atvty balanng senaros are dretly sefed as nut data when usng tradtonal transort models so that several alternatve otons (deste otentally nonsstent wth l use market nteraton) ould easly be generated based on the atvty dstrbuton of the base year. hs would reman a roblem f usng ntegrated l use transort models; - the fnal evaluaton of the mat of the atvty dstrbuton (exogenous or endogenous) on the traff oeraton s smlar n both modelng aroahes. 3. MHODOLOGY FOR A CAS SUDY OF ACVY BALACG POLCS H SÃO PAULO MROPOLA ARA BRAZL Gven the eular stuaton of the resent study, the use of the 1997 Metrô/SP tradtonal transort model the udate of the SSRA ntegrated l use transort models were onsdered as vable otons. he Metrô/SP tradtonal transort model wth a bas reresentaton of the study area

5 dvded nto 1158 zones for ersonal trs was fnally seleted. Frequenes of transt lnes a 69- zone exogenous O-D matrx for truk trs were also used for omutng exogenous flows. Maros wth the MM2 software were used for most of the smulaton stes (based on the 1997 Metrô/SP tradtonal transort model) for omarson of results between the base year the senaros smulatons. A maor smlfaton n the model was the assumton of the same road network transt serves (nludng osts of the base year) for all senaros smulated, as our man nterest was on the evaluaton of the otental effet on traff oeraton. he struture of the 1997 Metrô/SP tradtonal transort model the defnton of the atvty balanng senaros are desrbed n the followng tos. 3.1 he 1997 Metrô/SP radtonal ransort Model Followng the usual struture of tradtonal transort models, the 1997 Metrô/SP verson mlemented the four stes wth some sef otons that are desrbed here. r generaton models were albrated usng lnear regresson wth aggregated data usng three tr uroses: home-work, home-shool other. Home-work trs were searately estmated for four household nome levels (nstead of ontly as done for the other uroses). Model equatons are resented below: P = a. + a. + a. + a ( 1 ) A = b. + b. + b. + b + b. M, ( 2 ) where: P s daly total tr roduton n zone for urose, A s daly total tr attraton n zone for urose, are the oulatons of zones (n eah nome level f requred), are the total emloyment ostons of zones, are the total nomes n zones (n R$ of otober/1997), are the total numbers of automobles of zones, M s the total number of shool laes of zone (for home-shool trs only), a, a, a, a, b, b, b, b b M are albrated arameters (shown n able 1). r dstrbuton was arred out for nterzonal trs only used a stard aggregated doubly onstraned gravtatonal model for mornng eak hour trs of eah urose. t nluded an exonental medane funton based on average generalzed ost for all modes. Fxed onverson fators were used for obtanng trs n the mornng eak hour searatng ntrazonal trs. hen, model equatons aled were: ' = a. O'. b. D' f [ ] wth O ( 1 f ). O. M D = O. D D ( 3 ) =, ( )

6 1 b [ ]. D' f, b = a [ ]. O'. f a =. 1 0 ( 4 ) where: s the total number of mornng eak-hour trs between zones for urose, O' D' are the nterzonal mornng eak-hour tr roduton attraton of zone, resetvely (balaned by roduton), for urose, f = ex. s the generalzed ost, f [ ] s the medane funton, where [ ] ( ) f s the mornng eak-hour ntrazonal searaton fator for zone, O D are the mornng eak-hour tr roduton attraton of zone, for urose a b are the balanng fators of zone n the doubly onstraned model, s the albrated medane arameter for tr urose (also shown n able 1). he generalzed osts used n the tr dstrbuton albraton ( also n the smulatons) were the average of ubl rvate transort modes, alulated as: nv es e ( t + t + t + ) + ( t )/ 2 + = wth ( 5 ) F F m o F F m =.60 /( / P / h ) d..( 60 /( / P / h )) ( ) to / = ( 6 ) where: t s the n-vehle travel tme n ubl transort from to (mnutes on the smulaton network), nv es t s the wat tme n ubl transort from to (mnutes on the smulaton network), e t s the walk tme n ubl transort from to (mnutes on the smulaton network), t s the n-vehle travel tme n rvate transort from to (mnutes on the smulaton network), are the osts of travel between zones for ubl or rvate modes of transort resetvely, onverted to equvalent mnutes by the hourly wage rate, s the average fare of ubl transort from to (n R$ of otober/1997), F s the average household nome er month (n R$ of otober/1997), F P s the average number of ndvduals er household, m h s the average number of work hours er month, d s the travel dstane from zone to (n km on the smulaton network), o s the oeratng ost for the rvate transort (estmated as 0.25/km n R$ of otober/1997), t s the average auto ouaton of rvate transort (estmated as 1.5 ass/auto). o Mode slt was arred out for eah tr urose usng a smle bnomal logt, but aggregatng nome levels nto two lasses (1&2 3&4). Model equatons were: P ex( U ) ex( U ) + ex( U = ) P ex( U ) ex( U ) + ex( U = wth ( 7 ) )

7 U. = α 0 + tv. tv + te. te + taux. taux + trans. trans + ( 8 ) U. f = tv. tv + e. e d. d + tm. tm + ( 9 ) where: P P are the robabltes of hoosng ubl or rvate transort modes resetvely, U U are travel utltes for ubl rvate transort modes, tv tv are n-vehle travel tmes for ubl rvate modes (mn. on the smulaton network), te s the wat tme for ubl transort mode (n mnutes on the smulaton network), taux s the exess tme on aess transfer for ubl mode (mnutes on the smulaton network), trans s the number of transfers for ubl mode (on the smulaton network), are out of oket osts for ubl rvate modes (n R$ of otober/1997) e s a dummy for arkng avalablty on destnaton zones for rvate mode (assumng value 0 f hardly avalable or harged arkng s redomnant at the destnaton zone value 1 otherwse), d s the travel dstane for rvate model (km on the smulaton network), tm s the motorzaton rate at the home zone (n autos/1000 nhabtants), F s the average household nome n the home zone (n R$ of otober/1997), 0, tv, te, taux, trans, tv e,, d, tm α,, are albrated arameters (shown n able 2). he onventonal equlbrum method of MM2 software was aled to the tr assgnment of ths auto tr matrx omutng all the traff erformane measures. o feedbak effet due to ongeston was emboded n the smulaton roedure. A mnor hange had to be made n the mode slt routne beause the transt network of 1997 ould not be reovered, as reorted by Loes (2003). he mode slt was ket fxed for eah tr nterhange urose, onentratng on the rvate mode tr smulaton. he auto ouaton fator was also nororated n the same ste rvate vehle trs were estmated as: Q. q a = ( 10 ) a where: Q s the total automoble trs from zone to for urose (n number of vehles), a s the total trs from zone to for urose (n number of ersons), a q s the ontly mode slt auto ouaton fator for auto trs between zones for urose (obtaned from the 1997 O-D survey). Several rtal hyotheses are resent when usng the Metrô/SP tradtonal transort models (manly that the effet of atvty balanng an be evaluated usng aggregated models, that tr dserson an be redted by onventonal gravtatonal models, that ntrazonal trs the ombned mode slt auto ouaton fators remans fxed) they should be ket for the result analyss. 3.2 Buldng Atvty Balanng Senaros As revously quoted, the man advantage of usng tradtonal transort models was the ossblty of evaluatng the effets of any atvty balanng senaro that ould be translated nto a gven

8 dstrbuton of atvty n zones (even f t was not very onsstent wth the dynam of the l use market). Chosen ths oton, the man task beame the generaton of a set of senaros for evaluatng the otental effets of atvty balanng on traff oeraton. Startng wth two extreme senaros, two ntermedate ones were bult as suggested by the analyss of the roblems revous results. he frst senaros sread atvty balanng over all the study area usng unform rtera. Based on the 1997 O-D survey data, these two radal senaros were defned usng average funtonal requrement fators observed n the overall RMSP: - the P100 senaro redstrbuted oulaton (also automobles, nome shool laes) roortonally to the emloyment level of eah zone, - the 100 senaro redstrbuted emloyment (exludng emloyment n eduaton) roortonally to the oulaton level of eah zone. wo ntermedate senaros (P50 50) alyng half of the redstrbuton of the radal senaros were also defned to evaluate the lnearty (or not) of the atvty balanng effets. Due to sae onstrants, other senaros of foused or ombned oles would not be desrbed here. 3.3 Defnton of Crtera for Analyss of Results he effets of the alternatve atvty balanng senaros on traff oeraton were reresented through seleted erformane measures based on the total vehle-hours (VHR) vehleklometers (VKM) of the almost fxed total of trs. Both aggregate measures an be alulated from traff assgnment results on the smulaton network used to alulate erformane measures as: - average travel tme (the dvson of VHR by the number of total trs), - average tr length (the dvson of VKM by the number of total trs), - average travel seed (the dvson of VKM by VHR). he varaton of the average travel tme was taken as the man erformane measure for the users the hange n VHR as the maor measure for the traff systems ther relatve values are smlar, beause the total number of trs s almost fxed wth the Metrô/SP tradtonal transort model. he varaton of tr length reresents the effet of reorentaton of the satal dstrbuton of tr ends. he varaton of average seed reresents the effet of the reduton of ongeston level on the road network. Both are ntermedate effets emboded n the average travel tme. ote that the analyss of aggregated results on the overall road network of the study area uses a wde average but an be weghted aganst a measure of overall oly effort. he measure of oly effort s desrbed by the roorton of oulaton or emloyment reloated n eah atvty balanng senaro (deste the fat that a real oly ould also hle the loaton of new atvtes). 4. RSULS OF H CAS SUDY OF ACVY BALACG POLCS H SÃO PAULO MROPOLA ARA, BRAZL he atvty dstrbuton n the base ase (obtaned from the 1997 O-D survey) s shown n Fgure 1, where the emloyment surlus or deft s dslayed n terms of densty er area. Grey olors reresent emloyment surlus n ooston to the whte olor that reresents deft of emloyment.

9 As revously mentoned, the base ase s reresented by the results obtaned n the smulaton of the atual atvty dstrbuton as observed n the 1997 O-D survey. he same onvergene rtera are also used for all smulatons to favour unform qualty of results. herefore, albraton errors are otentally slt nto the results of the base ase the senaros. able 3 resents the results gathered from the general senaros. As an be seen, redstrbuton of oulaton delvers better results than redstrbuton of emloyment. Senaros P50 P100 mroved the average travel tme by 26.4% 33.7%, reloatng resetvely 18.7% 37.4% of the oulaton. Senaros resulted n mrovements n average travel tme of 12.0% 4.7%, ganed by reloatng 18.1% 36.0% of the emloyment, resetvely. hen, the mrovement n 50 was subsequently lost n 100 (a reverse effet). hese results are exlaned by the nrease n the average tr length generated by strong emloyment desonentraton, what s the ooste of the oulaton redstrbuton wth onentraton n the enter. able 4 shows a better ture: for oulaton redstrbuton, eah 1% of reloaton effort resulted n 1.6% mrovement n the frst half 0.6% n the seond half, as measured by the reduton n average travel tme. he effet of emloyment redstrbuton s nonlnear nonnreasng, s reversed when more emloyment balanng was fored to follow oulaton dserson radally. hs ndates that the otmal level of emloyment redstrbuton has to be arefully determned to avod the reverse effet. ven n the favourable range, eah 1% of emloyment reloaton generates only 0.7% of reduton n the average travel tme, as also shown n able 4. Some general remarks on the results are needed: - model arameter, eseally the tr dserson arameter of the gravtatonal dstrbuton model, have a strong nfluene even on some qualtatve results (for examle, doublng the tr dstrbuton arameter,.e. redung dserson, elmnates the reverse effet of 100); - the detaled effets of general balanng oles are dstnt, as oulaton redstrbuton mroves radal arterals erheral roads but arry on entral arterals streets, whle the ooste effet s seen on emloyment redstrbuton; - the analyzed senaros were bult through the exlusve addton of atvtes (oulaton or emloyment taken from other areas) as a growth-attraton oly, but ombned addtons of oulaton or emloyment atvty substtuton were also more romsng; - the analyzed senaros balaned atvty omoston by nome level; other senaros, for examle, showed that reloatng hgher nome grous s more effetve n a rato of 4:1 (double the effet wth half the reloaton effort); - fousng on the more unbalaned areas were almost two tmes more effetve (the elastty of reduton n average travel tme was 2,6% when redstrbutng oulaton to the São Paulo ty entre 0,9% when redstrbutng emloyment to the São Paulo east frnge). 5. COCLUSOS he man onlusons are that the otental of atvty balanng oles an be easer to assess when usng tradtonal transort models (omarng to ntegrated l use transort models). Consderng general oles dealng wth the whole RMSP - the São Paulo Metrooltan Area, the study reveals relevant results more easly aheved from oulaton redstrbuton than from

10 emloyment redstrbuton. he oulaton redstrbuton senaros dslayed a farly onsstent almost lnear attern of benefts (around 1% of reduton n average travel tme for eah 1% reloaton effort drven to atvty balanng). mloyment redstrbuton should be more arefully desgned to be effetve. Deste the need of develong further studes on the robustness of these onlusons, ether usng other modelng aroahes or through senstvty analyss of model arameters, there seems to be relevant nsghts to be ganed from exlorng addtonal senaros usng ths aroah. RFRCS Bento, A.M., M.L. Croer, A.M. Mobarak K. Vnha (2002) he mat of Urban Satal Struture on ravel Dem n the Unted States. nternal reort. he World Bank. Cervero, R. (1996) Jobs-housng balane revsted. Journal of the Ameran Plannng Assoaton, v.62, n.4, henque, M. (1983) he São Paulo metrooltan study: a ase study of the effetveness of urban analyss. Proeedngs of ato Advaned Researh nsttute Meetng on System Analyss n Urban Poly-Makng Plannng, Hy, S. (1996) Methodologes for exlorng the lnk between urban form travel behavor. ransortaton Researh D, v.1, n.2, Huthnson, B.G. (1974) Prnles of Urban ransort Systems Plannng. Srta, Washngton DC. Loes, D.L. (2003) Vabldade do Uso de Modelos Sntétos ntegrados de Uso do Solo e ransortes: estudo de alação à dade de São Paulo. MS hess, Polytehn Shool, São Paulo Unversty (n Portuguese). Malone-Lee, L.C., L.L.Sm L. Chn (2001) Plannng for a more balaned home-work relatonsh: the ase study of Sngaore. Ctes, v.18, n.1, owlan, D. M. G. Stewart (1991) Downtown oulaton growth ommutng trs: reent exerene n oronto. Journal of the Ameran Plannng Assoaton, v.57, n.2, Rkaby, P.A. (1987) Sx settlement atterns omared. nvronment Plannng B, v.14, Rosenbaum, A.S. B.. Koeng (1997) valuatng Modelng ools for Assessng L Use Poles Strateges. Reort PA420-R nvronment Proteton Ageny, USA. SMPLA (2002) Strateg Master Plan :Law of 13/09/02. Seretara Munal de Planeamento Urbano, São Paulo (n Portuguese). Wegener, M. (1996) Reduton of CO 2 emssons of transort by reorganzaton of urban atvtes. n Y. Hayash J. Roy (eds.), ransort, L Use the nvronment. Kluwer A.P.,

11 able 1: Calbrated arameters for tr generaton dstrbuton models by urose r Purose Parameter Home-Work Homeshool Other All Hhs nome 1* nome 2* nome 3* nome 4* uroses a a a a b b b b b M Soure: Metrô, * nome levels are: 1 u to R$ (5 SM); 2 - from R$ to R$ 1, (5 to 15 SM); 3 - from R$ 1, to R$ 3, (15 to 30 SM); 4 - over R$ 3, (30 SM). SM sts for monthly mnmum wage 1 SM = R$ of otober/1997. able 2: Calbrated arameters for mode slt models by tr urose r Purose Parameter Home-Work nome 1&2* nome 3&4* Home-Shool Other Puroses α tv te taux trans tv e d tm Soure: Metrô, * nome levels are: 1&2 u to R$ 1, (15 SM); 3&4 - over R$ 1, (15 SM); SM sts for monthly mnmum wage 1 SM = R$ of otober/1997.

12 able 3: Comarson between atvty-balanng senaros base ase Senaro Measure Base P50 var.% P100 var.% 50 var.% 100 var.% otal rs 739, , % 742, % 737, % 736, % otal etwork 1 11,573 11,573-11,573-11,573-11,573 - Reloaton 2-3,140, % 6,280, % 1,259, % 2,509, % otal VKM 13,715,048 13,012, % 12,524, % 13,976, % 14,464, % Average length/tr % % % % Average flow/lnk 3 1,185 1, % 1, % 1, % 1, % otal VHR 1,436,639 1,057, % 952, % 1,264, % 1,368, % ravelme/tr % % % % Seed % % % % 1 - km; 2 - he reloaton efforts are relatve ether to total oulaton of nhabtants or to total emloyment of emloyees (for all the RMSP); 3 - vehles er km of lnk; 4 - hours; 5 - km/h. able 4: lasttes of Results to Balanng ffort Reloaton ffort Senaro Base P50 P Balanng ffort (%) - 50% 100% 50% 100% Reloaton ffort (%) % 37.40% 18.10% 36.05% Average ravel me (hr) last.ravel me,%bal % % % 16.81% last.ravel me,%ff % % % 46.82% Average r Length(km) last.r Length,%Bal % -7.56% 4.16% 7.36% last.r Length,%ff % % 11.49% 20.50% Average Seed (km/h) last.seed,%bal % 13.36% 29.13% -8.88% last.seed,%ff % 35.72% 80.46% % Fgure 1: Atvty dstrbuton of São Paulo Cty aordng to the 1997 O-D survey