Traffic Engineering

I work at ARRB Group, a company which provides research, consulting and information services to the road and transport industry. ARRB stands for Australian Road Research Board and I have worked in the Melbourne office since July 2004.

I work in the Network Operations team and also provide technical skills to the Safe Systems team. I work mostly with civil engineers but also economists and behavioural psychologists. We undertake both research and consulting projects, primarily for the Australian state road authorities, such as VicRoads, NSW Roads and Maritime Services (formally RTA NSW) and Queensland Transport and Main Roads.

The Network Operations team’s primary focus is traffic management, particularly with regards to traffic congestion, but we also carry out work in a variety of other fields, such as travel time modelling, accessibility analysis by various modes of transport and traffic forecasting for toll roads. Congestion on our roads will never be completely eliminated, particularly while traffic volumes continue to grow. We can’t always build more roads to ease congestion so the emphasis is now on traffic management to minimise the extent and durations of congested periods. Measures to manage traffic include optimising traffic signal coordination on arterial roads, re-routing traffic to less congested areas of the road network and metering freeway on-ramps to regulate traffic merging onto these freeways during peak periods.

To find out more about ARRB Group, go to

Fundamental Traffic Flow Theory

The movement of traffic is referred to as traffic flow. Traffic on arterial roads must stop at intersections to give way to opposing traffic. Hence the movement of traffic on arterial roads is referred to as interrupted flow, while traffic movement on freeways is unimpeded by intersections and is thus referred to as uninterrupted flow. Note that these references to flow describe the facility in which the traffic moves, as opposed to the quality of the flow at any given time.

The three basic parameters used to describe traffic flow states are flow rate, speed and density. For uninterrupted traffic flow, the theoretical relationships between these parameters are generally referred to as the fundamental relationships of uninterrupted traffic flow.

These parameters are generally measured by loop detectors embedded within the road pavement and are best understood by considering the movement of traffic over a fixed point on the road.

Fundamental Diagrams for Uninterrupted Traffic Flow
Source: Austroads Guide to Traffic Management Part 2: Traffic Theory
Fundamental Diagrams for Uninterrupted Traffic Flow
  • Flow rate is the number of vehicles which pass over the point in a given period and is usually expressed as vehicles per hour.

  • Speed is the mean speed of all vehicles which pass over the fixed point in the given period, usually expressed in kilometres per hour.

  • Density is a measure of the number of vehicles along a specified length of road or lane at a given point in time and is expressed in vehicles per lane per kilometre. An increase in density is reflected in a reduction in the gap acceptance between vehicles. As density is a measure of area covered, it cannot be measured from a single point on the road. However occupancy, which directly correlates with density, can be measured from a fixed point and for this reason, occupancy is commonly used as a measure of density in traffic control systems. Occupancy is measured as the proportion of the given time during which vehicles occupy a fixed point on the road. For loop detectors, this is the proportion of the time they are detecting vehicles directly above them.

The image to the right shows the fundamental diagrams for uninterrupted traffic flow. As flow increases from point A in these diagrams, so does density. The gap acceptance between vehicles decreases, as do speeds. If flow increases above the capacity flow for the road (point C on these diagrams), it will quickly become unstable. At this point, speeds will drop dramatically and the rate of flow will subsequently drop. Although the rate of flow falls past point C, density (or rather occupancy) actually increases as the fall in speed means vehicles occupy the road longer. In worst case scenarios, congestion will be so great that the traffic slows down to a standstill (point D). At this point, density is at its maximum. No more vehicles can occupy the road and the gaps between jammed vehicles are at their smallest. With traffic at a standstill, loops detect single stationary vehicles constantly above them and hence record 100 % occupancy while also recording zero speed and zero flow.

Traffic recovering from a congested period moves in the reverse direction on these diagrams from D to A or any two points in between. However, the speed vs flow curvature is less pronounced as recovery involves much lower traffic volumes than the proceeding breakdown.

In illustrating the fundamental relationships between the basic parameters of traffic flow, these theoretical graphs rather simplify the relationships between speed, flow and density. In practice, traffic moves back and forth in short intervals along these curves and in actual plots of these relationships, lines of best fit usually don’t show such symmetrical curvature.

Managed Freeways

Traffic flow theory in practice has traditionally aimed at achieving capacity flows at levels deemed achievable through theoretical analysis. However, contemporary research has determined that theoretical capacity flows cannot be sustained for periods much longer than 5 to 10 minutes before becoming unstable, leading to flow breakdown. Thus in managing peak uninterrupted flows on freeways, contemporary practice is to control the rate of flow such that it is sustained at an ‘operational’ capacity, which is a flow rate just below theoretical capacity. Operational capacity is less susceptible to becoming unstable and, as such, can be sustained for several hours rather than minutes.

An integrated managed freeway system may interface a number of real time on-road traffic management tools, such as variable speed limits, traveller information signs, a lane use management system and ramp metering to control entry flows.

Ramp Metering

Ramp metering is a means of controlling the flow of vehicles entering a freeway from one or more entry ramps. Traffic signals positioned on freeway entry ramps operate in short cycles, usually allowing just one vehicle through from each lane of the ramp during each green phase. This effectively meters the flow of traffic entering the freeway from these ramps.

Ramp signals
Source: VicRoads Freeway Ramp Signals Handbook
Ramp signals on an entry ramp of
the Monash Freeway in Melbourne

Freeway ramp signals are automatically switched on during periods of high traffic flows and can either operate in isolation or, when needed, engage with signals on other entry ramps upstream in a coordinated master/slave relationship. When ramp meters are coordinated it improves the ability to manage the mainline freeway flow by matching traffic inflows from a group of ramps to the capacity of a critical bottleneck downstream. It also has the capability of balancing the queues and wait times between ramps.

Early ramp meters were usually based on vehicles entering gaps in the freeway’s outermost merge lane. The contemporary approach considers flow in all lanes across the carriageway and the related traffic density. Intelligent ramp signals use a dynamic ramp metering system that adapts to changing traffic flows on the freeway and ramp. A range of parameters in the control system algorithm can be adjusted in real time to continually reduce or extend cycle times of ramp signals and thereby adjust ramp inflows in response to changing freeway conditions.

The major parameter in a control system algorithm of a contemporary ramp metering system is occupancy, which is a measure of density. Whereas the critical capacity leading to flow breakdown at a bottleneck can vary somewhat from day to day, particularly in adverse weather conditions, the critical occupancy at which capacity flow occurs has been found to be fairly stable, even under adverse weather conditions. Hence occupancy has been found to be more appropriate as a parameter for optimising throughput than flow rate or speed.

Continual refinement achieved through dynamic operation of a ramp signal system enables freeway flows and travel speeds to be optimised during peak periods. Dynamic operation of freeway ramp signals also has the enhanced capability to prevent flow breakdown occurring at bottlenecks due to uncontrolled demand as it can provide more effective identification of, and response to, flow breakdown caused by an unplanned incident. Furthermore, after such incidents, dynamic operation can then manage ramp inflows to facilitate faster recovery.

To learn more about ramp metering, you can download chapters of the VicRoads Freeway Ramp Signals Handbook from the VicRoads website.


Microsimulation is a modelling technique used to evaluate current, proposed or theoretical road networks in terms of their operational performance under specified traffic demands. The most widely used microsimulation software packages in Australia are Aimsun, VISSIM and Paramics. Current road networks are commonly modelled to evaluate how they would perform with higher traffic volumes as projected for future years while proposed networks can include re-designs to current road networks which are modelled to assess how they would handle future increased traffic demands. At ARRB we mostly build theoretical road network models for research purposes using either Aimsun or VISSIM. By definition, microsimulation is used for detailed modelling of small areas of road network consisting of just one or more intersections.

Once the road network has been built within the microsimulation software, traffic demand volumes are then applied to the model. These can either be entered into an origin-destination (O-D) matrix within the model or just specified for each origin, along with turning proportions at each intersection. Origins and destinations are usually around the perimeter of the model where roads enter and exit the modelled area. When a simulation is run with O-D traffic demands, vehicles take the shortest path from their origin to their specified destination.

A number or other model parameters are also set before the model is run, such as speed limits, cycle and phase times for the traffic signals, vehicle compositions (i.e. the percentages of vehicle types, such as trucks, which make up the traffic demands) and proportions of passive to aggressive drivers. Other parameters are largely in-built within the model and, although they can be modified somewhat, often through use of an API, they are fundamental to how the model simulates traffic flow. One such parameter is the car following model, which is essentially the algorithm programmed within the simulation software used to determine if or when vehicles will change lanes to overtake slower vehicles ahead. A number of parameters are used in the car following model, such as the speed differential between the leading and the following vehicles, the road geometry and the sight distance for the following vehicle.

Source: ARRB Group and TTS - Transport Simulation Systems
Screen shots of Aimsun traffic microsimulations

Simulations can be run in real-time or sped up to save time. (Simulations faster than real-time still provide accurate simulations. It’s just that computers can process them faster.) Performance indicators are usually output to a database as each simulation is run. Multiple simulations are usually run for each scenario, each with a different random seed number, and an average of each measurable performance indicator taken from all runs. Traffic operation performance measures can be output for the network as a whole or for specified routes within the network. Measured outputs include average delay, travel time, speed, flow, density, fuel consumption, fuel emissions and several other performance indicators of traffic operation.

To learn more about traffic microsimulation software, visit: