No abstract

<p>Este trabalho relata o desenvolvimento de um sistema automático para calibração do modelo de simulação de tráfego em rodovias de pista simples TRARR, baseado em um algoritmo genético. O processo de calibração consiste em procurar valores para alguns parâmetros selecionados do simulador, de tal modo que a diferença entre as correntes de tráfego simuladas e as observadas em campo sejam minimizadas. Os parâmetros escolhidos para a calibração estão relacionados principalmente ao comportamento dos motoristas e, por isso, de difícil calibração. As diferenças entre a simulação e o fluxo real foram determinadas em termos de duas medidas de desempenho: a velocidade média dos veículos ao percorrer o trecho estudado e a proporção de veículos viajando em pelotões. O processo proposto calibra o modelo simultaneamente para cinco trechos diferentes, a fim de encontrar o conjunto de parâmetros calibrados que melhor represente uma rodovia típica brasileira. Os dados usados foram coletados emrodovias localizadas na região centro-norte do estado de São Paulo. Os resultados obtidos demonstram que o sistema proposto para a calibração funciona adequadamente e que a diferença média encontrada foi 2,75% e a diferença máxima, 4,09%.</p>

EGAMI, C. Y (2000) Rccalibration of a traffic simulation modcl for two-lane highways. São Cados, 2000. 137p. Dissertation (lviSc.)-São Carlos School ofEnginceri.ng, Universidade de São Paulo. The objective of this rcsearch is to recalibratc a two-la.nc highway simulation modcl to allow its employment within thc context of the São Paulo state highways, Brazil. Thc modcl chosen for tl1is recalibration is TRAR.R (fRAffic on Rural Roads) version 4.0, o rigin;illy devclopcd a i.n Australia by ARRB Tra.nsport Rescarch Ltd. The procedure uscd for thc model rccalibration consis ted in modifying i.nput parameters related to heavy vehicle dimensions anel performance charac teristics (power/ mass ratio, wcight and length) as wcll as to thc behavior of vchicles in platoons, in such a way that tl1e model outputs are compatiblc witl1 real traffic behavior. T hrce measures of perform;Ulce wcre choseu for rccalibratiou of tl1e model: perccnt of joumcy time spent followi.ng, average speed anel overtaking rate. Data for moelel recalibration wcre collected at a 2.8 km mral section of SP 225, a highway which COIUlccts tl1e cities of Ribeirão Preto anel Araraquara in tl1e state of São Paulo (Brazil). Thc recalibration procedure was based on comparisons of tl1e measures of perfonnance obscrveel on thc real highway witl1 those produceel by tl1e simulator's rccaJibrated ruHI original versions. At tl1e end of tl1e recalibration process, both the rccaJibrated anel tl1e originalmodcls wcre considercd capable of rcproducing thc traffic bchavior on a two-la.ne highway similar to the oncs in tl1e state of São Paulo. Furthermore, it was concluded that the recalibrated modcl is capable of producing res ults slightly bctter tl1an tl1c original one. T he rccalibrateel model was submitted to a valichttion proccdme in which a lO km scction of SP 255 was si.mulatcd anel the measures of perfom1ance obtaincd witl1 tlte modcl were compareci to ficld observations. The rcsults of thc validation procedurc indicate tbat rccalibrated versiou of TRARR is capable of simulating traffic operations on mraJ two-lane highways in thc statc of São Paulo. The results of tl1e rcsearch aJso suggest tl1at thc modcl results may be improved witlt fiue-tuni.ng of some of tl1c parameters that were not modified during the recalibration proccdure.

Resumo: A meta deste trabalho foi descrever, de maneira prática, como aplicar uma adaptação do método do HCM2000 para análise da capacidade e do nível de serviço em rodovias de pista simples brasileiras. A adaptação proposta resulta de duas pesquisas que, basicamente, replicaram os procedimentos usados no desenvolvimento do método original do HCM, substituindo os dados obtidos em rodovias dos EUA e Canadá por dados coletados em dez trechos de rodovias brasileiras. Dessa forma, são apresentadas relações entre a taxa de fluxo, a velocidade média de viagem e a porcentagem de tempo seguindo em pelotões, além de tabelas com fatores de ajuste para condições não básicas. Podem ser avaliados segmentos longos e homogêneos, numa análise em sentido duplo ou em sentido úni-co, além de rampas específicas. Os resultados obtidos indicam que as estimativas dos níveis de serviço usando-se a adaptação do méto-do do HCM2000 tendem a refletir melhor as reais condições operacionais das vias brasileiras de pista simples.Palavras-chave: nível de serviço, capacidade, rodovia de pista simples, simulação de tráfego. Abstract:The goal of this study was to describe, in a practical way, how to apply an adaptation of HCM2000 method for capacity and level of service analysis on two-lane rural highways. The proposed adaptation is the result of two research projects that basically replicated the procedures used for developing the original HCM method, while replacing the data collected on highways in the U.S. and Canada for data from ten sections of Brazilian highways. Thus, relationships between flow rate, average travel speed and percent timespent-following are presented, as well as tables containing adjustment factors for situations differing from base conditions. It is possible to estimate the level of service of long homogeneous segments, considering one-way or two-way flows, as well as specific upgrades. The results indicate that the level of service estimates obtained with the use of the HCM2000 adaptation tend to reflect better the actual highway conditions.

Planners can apply a relatively simple and user-friendly tool to control the expansion of urban areas in developing countries. This tool, presented here, is based on a traditional accessibility index, which may be taken as a measure of urban development. The entire process of construction and analysis of the networks based on the accessibility index was carried out using a geographic information system for transportation (GIS-T). The tool has been tested in São Carlos, a medium-sized Brazilian city, in which the effects of some proposed real estate developments on the accessibility patterns have been studied. The results of this application show that the tool is suitable for selecting the best site for a new development among many alternatives, promoting more efficient use of the urban area.