Implementation of BRT lanes in Belo Horizonte, Brazil

2013

Client: BHTrans, with the services of traffic consulting firm Modelle Logistics and Engineering (a subsidiary of TectranGroup) Brief: Study the various options for implementing the BRT system in Belo Horizonte, including the impact of each option on each element of the transport system, from standard nonBRT buses to private vehicles.

The city of Belo Horizonte is gearing up to host the 2014 World Cup in Brazil, where the addition of a deluge of visitors to a population of over 2.5 million means that the issue of public transport is taking high priority. Among the plans is the implementation of a set of bus priority corridors, following the BRT models already in operation in even larger cities such as Curitiba and Bogotá.

A system redefinition, along with appropriate traffic signal reprogramming, is the perfect opportunity for the city to deploy a substantial public transport revamp that will simultaneously optimise the movement of pedestrians and other vehicles, and traffic simulation was extensively used to decide on the final proposal for the BRT system, which is now in the final stages of implementation, with the completion date forecast for late 2013.

The BRT system is already in operation in Curitiba

Design concept for BRT stations in Belo Horizonte

Method

Modelle studied the impact of each option on each element of the transport system, from standard buses not included in the corridor to private vehicles. Belo Horizonte’s BRT system will apply to only 20% of the buses that enter the study area, and the remaining 80% must use the road system alongside general traffic. No single mode of transport exists in isolation, so any changes to the operation of one mode of transport will have direct and indirect consequences for other modes, and therefore the efficiency of transport as a whole. Taking this into consideration, particular attention was paid to the operation of bus lines not included in the BRT.

The main differences between the scenarios under analysis were related to the route and extensions of the special BRT sections in the central area. The first alternative to be analysed had the greatest extension and had a big impact on some major roads. Based on the results of this initial scenario, different approaches included progressively reducing the extension of the BRT and focusing mainly on a transfer area, where all the BRT from the furthest reaches of the city would arrive and transfer to smaller standard distribution lines, thus avoiding any direct impact on the major routes considered in the first scenario.

Central Belo Horizonte, the area under study, currently serves about one million passengers using the standard system of public transport by bus, along with about 150,000 passenger cars every day. Due to the city centre’s radial road structure, 60% of those using the area under study need to use it to move between the main poles of travel, showing the importance of the central area for the city traffic as a whole.

In the study area, as in the majority of urban areas, the bus efficiency measure that has the greatest impact is traffic light optimisation. This is especially true in the city centre, where signals impact on more substantial portions of the time delays that can compromise bus travel. Thus, acting directly on signal optimisation gives significant results for travel time. However, of all the elements that influence travel time, signalling is the one that can have the most unexpected results. A key aspect of signals is their trickle-down impact on the general traffic. When we try to segregate a corridor for public transport, whether by using an exclusive lane or by some other method, the signalised intersection is the point where this segregation loses efficiency as the bus has to share it with mixed traffic. If the mixed traffic is not controlled properly, congestion will eventually reach the segregated area, often from vehicles blocking the intersection. In short, one can only guarantee priority for public transport if one is able to fully manage all modes of transport. In situations like central Belo Horizonte, where there is a high density of traffic lights, it is not easy to maintain an effective gain for public transport.

A good example, which at first glance seems to greatly favour public transport efficiency, is the imposition of a bus’s specific stage time at an intersection. Actually, this needs to be used with caution, since this imposition may in fact result in an increase in travel time of the allegedly favoured bus and also directly affect the general traffic. For each green time of a specific stage serving a movement, there is a red time counterpart, represented by the total remainder time of the signal cycle. To take one example: before, the bus shared the stage with the city traffic at an intersection with a cycle time of 120 seconds and a green time of 60 seconds.

Modelle created a special stage of 20 seconds specific to buses, turning the initial 60 seconds of red time into 100 seconds. This would increase the waiting time and reduce the opportunities for the bus to find the signal open. There is a further delay of approximately 2.5 times more than before at this intersection. Deadlines were tight on this project with only had three months to deliver a final proposal and so ease of calibration and high performance were key requirements for the modelling software. Because of its dynamic and efficient architecture, Aimsun software was the best fit for this project’s microscopic traffic simulation needs. Aimsun follows, among other parameters, the statistical distributions of driver behaviour, from the most compliant to the most aggressive, and the different types of vehicles with their peculiarities, such as speed and acceleration, which affect the overall performance of traffic, creating points of conflict and increased travel times.

In terms of methodology, the first step was to search for a global measure of effectiveness that could reflect the results achieved by each alternative. Modelle chose to measure the efficiency of the system analysing the number of people who completed their journey in the system, in all modes of transport (BRT, other buses and private vehicles). That is, how many people entered the network and crossed it to reach their desired destination in a given period of time.

Another issue was energy efficiency. We obtained results for each of the alternative scenarios, determining fuel consumption and emissions of carbon dioxide, fine particles, volatile organic compounds, etc.

Microsimulation tools are efficient at analysing the dynamic evolution of traffic congestion; by dividing the period of analysis into various intervals Modelle could evaluate the formation, dissipation, and duration of a traffic jam. The model could also compute the interference that occurred when a jam formed in one location impacts on other locations. In addition, Modelle modelled variability in drivers and /or vehicles and Aimsun’s innate characteristic of matrix manipulation not only allowed the modelling of the current traffic conditions but also to use models to forecast future demand.

Modelle was able to submit all the different scenarios of real urban traffic, in a non-deterministic approach. The model predicted the impacts of any change in the road system, from structural and physical interventions or changes in the circulation plan to more subtle changes like signal control. Because the study area is very central it exceeds the urban average in terms of characteristics such as the high concentration of signalised intersections (about 200) with more than 400 bus lines, each with their respective routes and frequency tables, and also their stops and the interference caused by the system’s entry and exit points, which were usually due to parking on a network with a total length of nearly 100km.

In order to automate the process of system modelling, Modelle created specific scripts that facilitated the process of importing information into the network; for example, timesheets, dwelling time at bus stops for each line (provided by the city), and so on, thus compressing the period of system modelling and also avoiding manual data entry errors.

Conclusions

For a study of alternatives that sought to clarify the feasibility of different settings and select the one that should be subject to executive detailing, the results could hardly be better. Modelle reached a viable proposal, with good technical indicators and broad prospects for improvement to the public transportation system. Comparing the environmental performance of the final alternative with the performance of the current situation clearly shows the success of the proposal.

The results speak for themselves: a 23% reduction in fuel consumption with a 41% drop in carbon dioxide emissions and 32% fewer inhalable particles. The average bus travel speed rose from 12 km/h to nearly 20 km/h, greatly reducing travel time and increasing the availability of vehicles to passengers.