1. Microscopic Traffic Simulation
• Key Logic
– Car-following
– Lane Change
• Other modeling issues

2. Car Following Models Follow-the-leader model started in 1950s
Currently 3 categories of models:
safe-distance models
stimulus-response models
psycho-spacing models
Discussed in Chapter 6

3. Lane Change Types • Mandatory lane change • Discretionary lane change
• Random lane change

4. Discretionary Lane Change
Increase speed
• Bypass a slow or heavy vehicle
• Avoid the lane connected to a ramp

5. Discretionary Lane Change Factors
Driver Characteristics
– Driver aggressiveness.
– Desired free flow speed.
– Normal/Maximum Acceleration/Deceleration.
– Impatience

6. Discretionary Lane Change Factors
External Stimuli
1) In the current lane:
– Relative distance between the subject vehicle and its leader.
– Relative speed of the subject vehicle and its leader.
– Current leader is making/will make a lane change.
– Current leader is making/will make a left/right turn.
– Heavy vehicles (e.g., truck, bus) downstream.
2) In the adjacent lanes:
– Relative longitudinal distance (Lead Gap) between the subject
vehicle and the candidate leader (if available).
– Relative speed between the subject vehicle and the candidate
leader
– Candidate leader is making/will make a left/right turn.
– Candidate leader is making/will make a lane change.
– Candidate leader is a heavy vehicle (e.g. truck, bus).

7. Mandatory Lane Change Vehicle make mandatory lane change in
order to position itself to reach its
destination.

8. Mandatory Lane Change Connect to the next link to maintain their path:
– changing to the proper lane before making the turning
movement at an intersection
– merging to the freeway mainline
– diverging to the deceleration lanes before exiting through offramp
• Bypass a lane blockage downstream (typically caused by incidents or special events)
• Avoid entering a restricted lane (e.g., a HOV lane, or a lane in which no truck is allowed)
• Respond to message signs (e.g., lane drop warning sings)

9. Lane Change and Critical Gap
• Gap Acceptance studies typically used to
assess capacity for merging, crossing,
turning, and for lane change
• Available gaps and their acceptance by
drivers are the two main components for
such studies

10. Example: Gaps for Ramp Merging

11. HCM Critical Gap Min gap size that would be acceptable.
Drivers would reject smaller gaps and accept larger gaps.

12. Critical Gaps for Ramp Merging

13. Lane Change and Gap Acceptance

14. Lane Change and Gap Acceptance

15. CORSIM Car-Following Model
FRESIM: Pitt model
Acknowledge Hasham

16. AIMSUN Car-Following
Based on Gipps model

17. VISSIM Car-Following
Modified from two models developed by Wiedemann (Weidemann74 and 99 models)
Psychophysical or action-point models. This family of models uses thresholds or action-points where the driver changes his/her driving behavior.
Drivers react to changes in spacing or relative speed only when these thresholds are crossed.
The thresholds and the regimes they define are presented in the relative speed/spacing diagram for a pair of lead and follower vehicles.

18. VISSIM Car-Following Steady-state criteria (sn ≈ sdesired, Δun ≈ 0).
The desired vehicle spacing is an interval (ABX ≤ s ≤ SDX) instead of a single value.

19. PARAMICS Car-Following
Psychophysical car-following model
Developed by Fritzsche
Fritzsche’s model uses the same modeling concept as the Weidemann74 car-following model.
The difference between these two models is the way thresholds are defined and calculated.

20. PARAMICS Car-Following
Where A0 is the vehicle spacing at jam density, Tr is the risky time gap (usually 0.5 s), TD is the desired time gap (with a recommended value of 1.8 s).

21. INTEGRATION Car-Following
Developed by Van Aerde

22. Car-Following Models in Micro Packages

23. Multi-Regime Approach*
• Car-following regimes
• Lane change regimes
* Zhang et. al, 1998

24. Car-Following
Normal car-following Regime
– 5th generation GM Model

25. Car-Following Uncomfortable regime – too close
When the calculated acceleration value from the GM
Model is greater than or equal to zero, and the “comfort”
spacing requirement is not satisfied, the vehicle is in the
uncomfortable regime and needs to decelerate to extend
the headway to a comfortable and safe range. In this
regime, the acceleration is determined by the following
equation

26. Car-Following Free-flow Regime
If the distance/time headway between a leader and its follower is greater than a pre-defined threshold, then the vehicle is in the free-flow regime. If the vehicle’s current speed is less than its desired free flow speed , then the vehicle will accelerate until it reaches its desired free flow speed. The vehicle maintains the speed until it
falls into another regime. The acceleration is Computed as follows:

27. Car-Following Emergency Regime
If a vehicle has a headway smaller than its pre-determined minimum distance, then it is in the emergency regime. Emergency braking is activated to avoid collision. The deceleration is calculated from the following Emergency Collision Model.

28. Car-Following Other regimes – Intersection arrival regime
• Reacts to signal, not only the leader
• Decelerates to stop at the stop-bar
– Queue discharge regime
• 1st vehicle to achieve FFS
• Followers using 1 generation GM model

29. Lane Change

30. Lane Change

31. Lane Change
Mandatory lane change regimes

32. Lane Change Frequently, drivers have to adjust their
accelerations to execute the lane changing
maneuvers (mandatory).
Lane Changing without Acceleration Change
The subject Vehicle Needs to Accelerate
The subject Vehicle Needs to Decelerate
Vehicles in the Target Lane Need to Change Speeds (cooperative drivers)
The subject vehicle stops

33. Other Modeling issues • Drivers reaction to changing traffic,
geometric and control conditions critical to
the success of simulation.
• Special car-following, lane change
scenarios arise from those conditions

34. Vehicle generation Headway distribution: – Normal distribution
– Negative exponential distribution
– Erlang distribution
– Uniform distribution
– Composite Distribution

35. Vehicle generation

36. Headway Generation: Composite Models

37. Vehicle generation Select proper distribution based on traffic volume
• More important for low traffic situation
– For high volume conditions, vehicle generation mainly only affects entry links. As soon as vehicles enter the network, car-following and lane change logic take control

38. Lane distribution • Distribution at generation
– Depending on volume
For low volume, most vehicles generated in right lane
• As volume increases, distribution across lane tends to be more equal among lanes
Trucks typically use right lane(s)
• Lane distribution on internal links should be determined by car-following and lane change logic
• Improper lane distribution at generation (especially on short entry links may produce undesired results

39. Lane distribution

40. Sources/sink • Modeling mid-block activities
– Parking lots
– Driveways
– Abutting properties
• Does not have to create more
links/intersections

41. Incident/on-street parking
Modeling disturbances to traffic
– Blocking
– Rubber-necking
– Interferences
– Reduced speed
– Special lane change behavior
– More “cooperative” drivers

42. Lane add/lane drop Reaction distance critical for lane drop
– “Warning” sign placement to let drivers “see” the situation
– Modeling similar to blockage

43. Driver behavior modeling
• Aggressiveness
• Familiarity
• Cooperation
• Anticipation

44. Driver Aggressiveness
Aggressive drivers
Follow leader closer
Change speed more rapidly (higher max. acc. dec. rates)
Shorter reaction time
Take smaller gaps for lane change
Need to map them consistently: an aggressive
driver most likely will do all of the above consistently. Random mapping is not appropriate

45. Driver Familiarity with Network
• Driver familiarity with route very critical for urban streets with multiple lanes and short link
• Driver familiarity with the network help vehicles
position themselves for lane changes
• Familiarity: know the next turns and lane
configurations ahead and prepare in advance

46. Cooperative Drivers
• A change in the percentage of cooperative driver can change flow conditions in many scenarios:
– Lane blockage
– Merging
– Moving into turning pockets

47. Driver Anticipation Related to driver familiarity with network
Related to cooperative drivers

48. Vehicle performance • Acceleration performance
• Deceleration performance
• Desired maximum speed wrt free-flow
speed/speed limit
• Modeling trucks properly is very important

49. Driver Reaction Time
Driver reaction time is a critical aspect of car-following
• Driver reaction time should be consistent with
driver aggressiveness
• Reaction times should be different for different
car-following scenarios
– Normal car-following
– Emergency braking
– Reaction to signal change/intersection arrival/departure

50. Intersection Modeling
• Driver response to signal state changes
– When signal in amber/red, vehicle slow down to
stop and does not follow the leader even if there
is one.
– First vehicle discharge from the intersection will
gain speed to the FFS with a normal acc. rate.

51. O-D • Some Programs are O-D based
• Some programs are turning movements based
• Inputs of full-OD may not be a good idea for a large network as the data is typically not available.
• Traffic assignment may be employed
• For turning movement based program, OD may be entered to eliminate unreasonable u-turns

52. Curves/Turning Speed • Many programs do not model turning paths
at intersections or on ramps. What you see
in animation may not be the result of
simulation.
• Truck turning on sharp curves may worth
modeling.
• Turning speeds (LT, RT) may or may be
modeled in many programs.

53. Signalization • Vehicles respond to signal states
• Signal timing scheme:
– Pre-timed
– Actuated
– Lane assignments very important

54. Accident generation? Most programs assume vehicles run safely
ensured by car-following and lane change
logic
• Can simulate accidents by user inputs
• Some programs are looking at automated
accident generation based on flow conditions, such as speed variance, etc.