1. Overview
1. The complexity of traffic stream behavior and the difficulties in
performing experiments with real-world traffic make computer
simulation an important analysis tool in traffic engineering.
2. The physical propagation of traffic flows can be specifically
described using traffic flow models.
3. By making use of different traffic simulation models, one can
simulate large-scale real-world situations in great detail.
2. Overview
Depending on the level of detail, traffic flow models
are classified into:
ā¢ Macroscopic models view the traffic flow as a whole
ā¢ Microscopic ones gives attention to individual
vehicles and their interactions
ā¢ Mesoscopic models fall in between these two.
3. Traffic Simulation Models
1. Simulation modelling is an increasingly popular and effective tool
for analyzing a wide variety of dynamical problems those
associated with complex processes which cannot readily be
described in analytical terms.
2. Usually, these processes are characterized by the interaction of
many system components or entities whose interactions are
complex in nature.
3. Specifically, simulation models are mathematical/logical
representations of real-world systems, which take the form of
software executed on a digital computer in an experimental
fashion.
4. The most important advantage is that these models are by no
means exhaustive.
4. Need for Simulation
1. Traffic simulation models have a large variety of applications in the
required fields. Nowadays they become inevitable tools of analysis
and interpretation of real-world situations especially in Traffic
Engineering.
2. The following are some situations where these models can find
their scope.
ā¢ When mathematical or analytical treatment of a problem is found infeasible or
inadequate due to its complex nature.
ā¢ When there is some doubt in the mathematical formulation or results.
ā¢ When there is a need of an animated view of flow of vehicles to study their
behaviour.
3. It is important to note that simulation can only be used as an
auxiliary tool for evaluation and extension of results provided by
other conceptual or mathematical formulations or models.
5. Overview of Simulation
Model Characteristics
1. Traffic simulation models take into account the fundamental
traffic flow, speed, and density characteristics.
2. Integrate them with analytical techniques such as demand-
supply analysis, capacity analysis, shock-wave analysis, and
queuing analysis.
3. Models logic varies across models and different approaches
to representing traffic operations are employed.
4. Traffic assignment is embedded in several simulation models.
6. Classification of Traffic
Simulation Models
1. The models were grouped according to their areas of
application.
2. The three classification of traffic simulation models:
ā¢ Microscopic modelling
ā¢ Macroscopic modelling
ā¢ Mesoscopic modelling
7. Microscopic, Mesoscopic
and Macroscopic Simulation
There are different resolutions of simulation models:
ā¢ Macroscopic simulation considers only aggregated traffic flow volumes,
and not individual agents (vehicle, road user).
ā¢ Mesoscopic simulation is based on individual agents, whose behavior is
determined from aggregated traffic flow attributes, like the density or the
average speed. Direct reaction of agents to other agents happens only at
nodes (intersections).
ā¢ In a microscopic simulation, each individual agent reacts on their current
environment, i.e. the distances and speed differences to nearby agents.
The movement is modeled continuously in time and space. Decisions
about a change of speed and direction happen usually in small time steps
of <1 second. The overall traffic state results from the individual decisions
of the agents.
ā¢ The combination of mesoscopic simulation and microscopic simulation in
certain network sections is known as hybrid simulation.
8. Microscopic Modelling
1. Microscopic modelling based on the characteristics of various
vehicle movements such as cars, buses, motorcycles and so
on in the traffic flow.
2. Microscopic modelling aimed to collect data parameters, such
as, flow, density, speed, travel and delay time, long queues,
stops, pollution, fuel consumption and shock waves.
3. The characteristics of microscopic modelling methods were
based on car-following model, lane-changing models and
gaps of the individual drivers.
9. Car Following Models
1. These algorithms introduced a concept that a driver recognize and
follow a lead vehicle in a lower speed.
2. This can be described as a situation such as in a car platoon
without being able to change the lane.
3. Generally, the following car algorithm determined the relationship
with the lead vehicle at times.
4. It was the function of spacing, speed and acceleration at times.
5. The closer the following vehicle to the lead vehicle the more
sensitive the reaction of the following vehicle to the lead vehicle.
6. This sensitivity also increases with speed.
7. This is because if the main vehicle was driven in a lower speed,
the following vehicles will then reduce the speed.
8. This resulted the occurrence of car platoon and traffic congestion.
10. Car Following Models
ā¢ Headway (h) is the horizontal distance between two vehicles.
ā¢ Spacing (s) is the distance measured from the rear-bumper of the following vehicle to rear-
bumper of the lead vehicles pass at a certain point within time interval.
ā¢ The time headway is defined as the time difference of successive vehicles crossings at the
point.
ā¢ Microscopic modelling is based on the movement of individual vehicles and their relative time
and space.
11. Lane Changing Models
1. Gipps has proposed a development of the lane-changing
model.
2. The lane-changing model is a decision process to estimate
the driverās behaviour in making a lane change within the time
given.
3. Lane changing can be categorised into two types.
ā¢ Mandatory lane change, which occurs when a driver change lane to a
specified lane. For instance, a driver changed to the right lane when he
wanted to make a right turn at the next intersection.
ā¢ Discretionary lane change, which the driver made to the next lane to
avoid following trucks and increase speed.
12. Lane Changing Models
Zone 1 : The furthest distance from the next turning to measure how the driver:
i. Change vehicles lane and speed of driver
ii. Speed and distance of lead vehicles
iii. Speed and distance upcoming lead vehicles at the other lane
Zone 2 : The intermediate zone.
The vehicles is looking for a gap to turn into the turning lane which affects the lane-changing decision.
Zone 3 : This is the shortest distance to the next turning point.
The vehicle has started to change lane at the merging lane to the next turning. It is necessary to reduce the speed to
provide a big gap for vehicle to change the lane.
13. Gap Acceptance Models
1. Gap acceptance are used to
determine the number of vehicles
of traffic that can pass through a
conflict point.
2. A gap is defined as the time
headway in between the lead and
lag vehicles in the target lane.
3. The lead gap is the gap to the lead
vehicle in the target lane.
4. The lag gap is the gap to lag
vehicle in the target.
Lane Changing with Consideration of Lead and Lag gaps
14. Gap Acceptance Models
1. Gap acceptance models were used to determine the size of gap
that will be accepted or rejected by a driver who aimed to merge or
cross the intersection.
2. Critical gap is defined as the number of accepted gap shorter than
its equal to the number of rejected gaps.
3. The driver needs time to clear the intersection and decide.
4. The parameters of the gap acceptance model are the acceleration
rate, desired speed and speed acceptance.
5. Among these, the acceleration rate, the maximum give-way time
and visibility distance at the intersection are the most important.
6. The acceleration-rate is the ability of the vehicle to accelerate with
required safety gap.
7. The maximum give-way time is to determine when a driver starts to
get intolerant if they cannot identify the gap.
15. Macroscopic Modelling
1. Macroscopic modelling describes the intersections at a low level of
detail.
2. In macroscopic model, traffic stream is represented in an aggregate
measured in terms of characteristics like speed, flow and density.
3. Flow, speed and density are the three main characteristics of traffic.
4. The researchers have studied speed-flow-density relationships and have
attempted to develop mathematical descriptions for these curves, which
is Greenshieldās Model and Greenberg Model.
5. Greenshieldās Model is used to develop a model of uninterrupted traffic
flow.
6. It is a fairly accurate and simple modelling. There are three
characteristics of the graph
16. Speed, v versus Density, k
ā¢ Greenshiedās Model shows the
relationship between speed and density
is linear.
ā¢ When density increases, flow decreases
to an optimum when more vehicles were
on the road.
ā¢ The speed also decreases due to the
interaction of vehicles.
17. Flow, q versus Density, k
ā¢ Greenshieldās model shows that the
relationship between flow and density is a
parabolic curve.
ā¢ When flow is very low, speed is higher.
ā¢ The drivers are able to travel at a desired
speed.
ā¢ As the flow increases, speed gradually
decreases.
ā¢ The highest flow shows the transition of
non-congested to congested condition.
18. Speed, v versus Density, k
ā¢ Greenshieldās model shows the relationship between speed and density.
ā¢ When the capacity (qmax) is in the maximum flow, the density is increasing and the speed will decrease
due to the maximum number of vehicles passing to a certain point.
ā¢ The characteristics of congested unstable flow is high density and low speed.
ā¢ There is no gap for vehicles to enter.
ā¢ The characteristics of uncongested stable flow is slow density and high speed, when there are gaps for
merging lane.
19. Mesoscopic Modelling
1. Mesoscopic modelling describes the analysed transportation elements in
small groups.
2. Mesoscopic models are a combination of microscopic modelling and
macroscopic modelling.
3. In this model, platoon dispersion is stimulated. There are two methods of
mesoscopic modelling which is platoon dispersion and vehicle platoon
behaviour.
ā¢ The first method is platoon dispersion. As a platoon moves downstream from an
upstream intersection, the distance between the vehicle which may be due to the
differences in the vehicle speeds, vehicle interactions (lane changing and other
interferences), (parking, pedestrians and others). This phenomenon is called as
platoon dispersion.
ā¢ The second method is vehicle platoon behaviour is a group of vehicles, travel at a
short headway and moving at the same speed. Vehicle platoon behaviour is to predict
the arrival of vehicle platoon with time, the total arrival time.
20. Mesoscopic Modelling
1. The important considerations in modelling of platoon behaviour are the
number of vehicles in a platoon, speed of a platoon and the distribution
of speed.
2. Platoon are formed when the traffic signal turns to red and the vehicles
start to queue.
3. For an isolated traffic signal, the lenght of queue is the key performance
measures that is afftected by the vehicle arrival pattern.
4. The platoon will disperse when the traffic signal starts to change from
yellow to green, so it starts with incerasing the speed, then the vehicles
start to move.
5. Once the vehicles are released from the stop line, they will start to move
in a tight platoon position with short time headways and will start to
disperse travel further downstream.
21. Traffic Simulation Models
1. The earliest computer simulation work in highway transportation
was the intersection simulation undertaken by the Transport Road
Research Laboratory in the United Kingdom in 1951, and the first
simulation work in the United States was on the intersection and
freeway models developed at UCLA in 1953.
2. The development of simulation models has grown rapidly since
then.
3. Gibson, Van Aerde et al., May and Sabra and Stockfisch review
simulation models for intersections, arterial networks, freeways,
and freeway corridors up to mid-1995.
4. Just one year later, the menu has been increased by the
introduction of TSIS/CORSIM, WATSIM and INTEGRATION 2.0.
22. Urban Street Networks
1. Urban street traffic systems comprise:
ā¢ Intersections
ā¢ grid-based networks
ā¢ Variety of complex traffic activities and control strategies such as
parking adjacent to traffic streams, bus blockage, one-way streets,
reversible lane operations, etc.
2. Both microscopic and macroscopic models are currently
available for simulating these environments.
23. Microscopic: TRAF-NETSIM
1. NETSIM (NETwork SIMulation) is the only microscopic model available for urban street
networks.
2. NETSIM, formerly called UTCS-1, was initially released in 1971 and integrated within the TRAF
(an integrated traffic simulation system) in the early 1980s (TRAF-NETSIM).
3. Most operational conditions experienced in an urban street network environment can be
simulated.
4. This model provides a high level of detail and accuracy, and it probably is the most widely-used
traffic simulation model.
5. The TRAF-NETSIM model uses an interval-scanning simulation approach to move vehicles each
second according to car-following logic and in response to traffic control and other conditions.
6. TRAF-NETSIM uses Monte Carlo procedures to represent real-world behavior. Therefore,
individual vehicle/driver combinations, vehicle turning movements on new links, and many other
behavioral and operational decisions are all represented as random processes.
7. The latest version of TRAF-NETSIM uses an identical seed number technique to represent
identical traffic streams and reduce output variability.
24. Macroscopic: TRANSYT
1. TRANSYT (TRAffic Network StudY Tool) is a deterministic, single-time period simulation and
optimization model developed at the Transport and Road Research Laboratory (TRRL, now TRL)
in the UK.
2. There have been nine British versions of TRANSYT, and the model has been applied worldwide.
3. In the early 1980s, TRANSYT-7F (F stands for Florida) was developed at the University of
Florida.
4. The data and outputs of TRANSYT-7 were modified to create a North American version, which is
now the most commonly used version in the US.
5. TRANSYT-7F models traffic behavior in signal-controlled urban areas.
6. During an optimization, it searches for a set of fixed-time signal timings that minimize vehicle
delay.
7. This is achieved by coordinating adjacent signals so that platoons of traffic can pass without
stopping.
8. TRANSYT-7F consists of a traffic model and an optimization process.
9. The former calculates a performance-index (PI) for the network using a particular set of signal
timings, and the latter adjusts these timings and checks whether the adjustments improve the PI.
10. In TRANSYT, there is no representation of individual vehicles, and all calculations are made on
the basis of the average flow rates, turning movements and queues.
25. Macroscopic: NETFLO
1. NETFLO (NETwork traffic FLOw simulation model) can simulate the traffic
flows at two levels.
2. NETFLO I (LEVEL I) is a stochastic, event-based model. It moves each vehicle
intermittently according to events and moves each vehicle as far downstream
as possible in a single move.
3. Although NETFLO I treats each vehicle on the network as an identifiable entity,
car following and lane-changing behaviors are not modeled explicitly.
4. Therefore, NETFLO I models traffic at a lower level of detail than NETSIM.
5. NETFLO II (LEVEL II) is a deterministic, interval-based model.
6. It is essentially a modified TRANSYT, but NETFLO II has no optimization
capability.
7. In NETFLO II, the traffic stream is represented in the form of movement-
specific statistical histograms.
8. Currently NETFLO and FREFLO, a macro freeway model, have been
combined as an integrated simulation system CORFLO.
26. Mesoscopic: CONTRAM
1. CONTRAM (CONtinuous TRaffic Assignment Model) is a traffic assignment
and evaluation package that models traffic flows in urban networks.
2. This model treats a group of vehicles, (packet), as a single entity, thus, vehicles
which belong to a packet travel along the same minimum cost route and arrive
at the same time.
3. CONTRAM determines time-varying link flows and route costs, in terms of
given time-varying route inflows, in a dynamic setting, and so is entirely
different from TRANSYT and NETSIM.
4. In CONTRAM, traffic demands are expressed as O-D rates for each given time
interval.
5. These O-D rates are converted into an equivalent number of packets, which
are assigned to the network at a uniform rate for each time interval.
6. A traffic assignment equilibrium is achieved through iterations in which each
packet is removed from the network and reassigned to a new minimum path.
27. Mesoscopic: SATURN
1. SATURN (Simulation and Assignment of Traffic in Urban Road
Netwoks) is a traffic assignment model based on the incorporation of two
phases:
2. a detailed simulation phase of intersection delays
3. an assignment phase which determines the routes taken by O-D trips
4. The complete model is based on an iterative loop between the
assignment and simulation phases.
5. The simulation determines flow-delay curves based on a given set of
turning movements and feeds them to the assignment.
6. The assignment in turn uses these curves to determine route choice and
hence updates turning movements.
7. These iterations continue until the turning movements reach reasonably
stable values.
30. How is traffic simulation used?
1. Design of traffic infrastructure
2. Non-standardized transportation infrastructure
3. Advanced Traffic Management Systems
4. Infrastructure planning for vulnerable road users
5. New technologies in the automotive industry
6. Pedestrian simulation
7. Public transport planning
8. Emission calculation
31. Applications
Traffic simulations models can meet a wide range of
requirements:
ā¢ Evaluation of alternative treatments
ā¢ Testing new designs
ā¢ As an element of the design process
ā¢ Embed in other tools
ā¢ Training personnel
ā¢ Safety Analysis